Reinke, J. Schneider-Mergener, A. Radbruch, P. Walden, andH.D. Volk. ...... mixture. Error propagation was calculated using standard formula, assuming the.
Cellular Microarrays For Interrogation Of Cell-Cell And CellSubstrate Interactions
Dissertation
Presented in Partial Fulfillment of the Requirements for the Degree Doctor of Philosophy in the Graduate School of The Ohio State University
By Chaofang Yue, B.S. Graduate program in Chemical Engineering
The Ohio State University 2010
Dissertation Committee: Michael Paulaitis, Advisor Stuart Cooper Jeffrey Chalmers John Sheridan
Copyright by Chaofang Yue 2010
ABSTRACT The interaction of cell surface receptors with their counterparts on the other cell or on an extracellular substrate, and subsequent signal transduction triggered/induced by these interactions, crucially dictate the survival, development, and function of the cell. In recent years, cellular microarray technology, in which intact living cells are captured onto pre-defined planar array patterns by specific interactions, has emerged to meet the fast growing needs for interrogating cell-cell and cell-substrate interactions in a high-throughput and cost-effective manner. However, this technology is still in its infancy; issues with reproducibility, specificity, and sensitivity need to be tackled, and systematic quantification methods need to be established before cellular microarrays are used for broader studies and applications.
This dissertation presents the development of cellular microarrays with applications of fundamental and therapeutic importance in infection and immunity. A novel major histocompatibility complex (MHC)-Ig dimer cellular microarray was developed that uniquely preserves the structural integrity of labile MHC-Ig molecules and optimizes the avidity of MHC-Ig binding. The assay was shown to significantly outperform
ii
conventional MHC cellular microarrays, and highly-competitive with other immune-monitoring assays in the sensitive and quantitative detection of antigen-specific T cells present at low frequencies in heterogeneous populations. With the correlation observed between peptide loading to MHC-Ig in solution and cell capture on microarray spots, the MHC-Ig cellular microarray assay was adapted to quantify the kinetics of peptide loading. Fitting of the measurements to a mechanistic model inferred unique characteristics of the binding reaction as well as revealed optimal loading conditions. The MHC-Ig cellular microarray was further applied to quantitatively detect the presence of T cells in peripheral blood for a number of antigenic peptides derived from influenza A virus; the obtained antigen specificity profiles agreed well with measurements by standard assays conducted in parallel.
To facilitate personalized cancer therapy with immunoliposomes (ILPs), a combinatorial antibody microarray was developed, in which spots were composed of mixtures of antibodies specific for cancerous cell surface markers. Levels of cell capture on the spots indicate cell targeting efficiencies of the corresponding antibodies when surface-conjugated on ILPs. The cellular microarray was successfully applied to screen for the optimal combination of antibodies for each individual that produces the highest cell targeting efficiency; compared with spots of single antibodies, spots of these optimal combinations of antibodies increased cell binding by as much as 80%.
iii
We further proposed a graphic method to qualitatively represent the observed synergies among the antibody mixtures.
Aside from developing the microarrays based on a commercial substrate, we developed a self-assembled monolayer (SAM)-based substrate as a general platform for protein microarrays. Neutron scattering characterization were conducted, indicating the highly ordered multi-component multi-layer structure of the substrate; preliminary cell binding experiments were performed, showing the great promises of this substrate for specific and uniform cell capture. In addition, a method of immobilizing functionalized magnetic nanoparticles on planar surface to form nanoparticle microarray was reported, which significantly reduced the cost for functional characterization of the particles and further demonstrated its potential as a new small-scale cell sorter.
iv
DEDICATION Dedicated to my parents
v
ACKNOWLEDGEMENTS
I would like to express my most sincere gratitude to my advisor, Professor Michael E. Paulaitis, who has been my role model as a research scholar, supervisor, and senior friend. It is my great honor to be one of his Ph.D. students. I will always remember the education and training he has given me and will continue to require myself with the high standards.
I am very grateful to my co-advisors Professor Jonathan Schneck and Professor Mathias Oelke at Johns Hopkins University Department of Pathology. Jon has provided me with great freedom and resources to learn all my immunological skills and conducting sets of experiments in his lab. Mathias is the one who always gives me great support, encouragement and suggestions for my research. The time in Hopkins has been such a valuable and memorable experience for me.
Thank Dr. David Vanderah in National Institute of Standards and Technology (NIST) for hands-on training and continuous guidance on my work with self-assembled monolayers. Thank Professor Mathias Losche, Dr. Duncan McGillivray, and Dr. Frank Heinrich in NIST Center for Neutron Research for help and data analyses of neutron scattering experiments. Also, I would like to thank Dr. Arfaan Rampersaud and Ms. Kristie Melnik for providing magnetic nanoparticles and helpful discussions in the nanoparticle microarray project.
vi
I would like to thank Professor Stuart Cooper and Professor Jeff Chalmers for continuous serving on my qualify exam, candidacy exam, and dissertation committees, and Professor Andre Palmer for serving on my candidacy exam committee. Their suggestions and comments are valuable to my work. I also would like to thank Professor James Lee for helpful suggestions on my combinatorial antibody microarray work, and Professor James Rathman for help with some statistical analyses.
I would like to thank my group members. Particularly, Ms. Megan Balog and Nicole Guzman for discussion and assistance in some experimental work, and Hamsa Priya Mohana Sundaram, Sowmithri Utiremerur, and Ghalib Bello for comments and help with data analyses. I also would like to thank members of Jon and Mathias’ labs for help with experiments, especially Dr. Christian Schutz, Ophelia Rogers, Aaron Selya, Dr. Tonya Webb, Dr. Alessia Zoso, Dr. Yu Li and Jessica Lee. I particularly would like to thank Aaron Risinger from Perkin Elmer for his support on the Piezorray instrument. I appreciate his patient and clear guidance on the instrument use and maintenance, and respect his dedication to work and focus on the needs of customers.
Finally, I would like to thank my parents for their unconditional everlasting love and support. No words can describe their love to me and no words can express my grateful feelings for them. Thank my husband for his constant support and encouragement. I also want to thank my grandfather who passed away in Dec. 2007. He had been so proud of me and for me to accomplish my Ph.D. study was always one of his biggest wishes.
vii
Chaofang Yue Feb 6th, 2010 at OSU
viii
VITA
September 1982……………………..Born – Luoyang, P.R. China June 2004……………………………B.S. Biochemical Engineering, Zhejiang Univeristy, Hangzhou, P.R. China September 2004 – February 2010…..Graduate Research Associate, Lowrie Department of Chemical and Biomolecular Engineering The Ohio State University
FIELDS OF STUDY Major Field: Chemical Engineering
ix
TABLE OF CONTENTS ABSTRACT...................................................................................................................ii DEDICATION............................................................................................................... v ACKNOWLEDGEMENTS..........................................................................................vi VITA.............................................................................................................................ix TABLE OF CONTENTS...............................................................................................x LIST OF TABLES......................................................................................................xiii LIST OF FIGURES ....................................................................................................xiv CHAPTER 1 .................................................................................................................. 1 INTRODUCTION ......................................................................................................... 1 1.1. Cell-cell and cell-substrate interactions .....................................................1 1.2. Cellular microarrays...................................................................................2 1.2.1. Cellular microarrays in infection and immunity................................3 1.2.2. Cellular microarrays in stem cell research.........................................4 1.2.3. Cellular microarrays in gene function studies ...................................4 1.2.4. Cellular microarrays in drug discovery..............................................5 1.3. Scope of current work ................................................................................6 1.4. Protein microarrays ....................................................................................7 1.4.1. Substrates for protein microarrays ...................................................10 1.4.2. Fabrication of protein microarrays...................................................13 References.............................................................................................................15 CHAPTER 2 ................................................................................................................ 24 OVERVIEW OF MHC CELLULAR MICROARRAY TECHNOLOGY.................. 24 2.1. Molecular recognition events of T cell-mediated adaptive immune response ……………………………………………………………………………..24 2.2. MHC multimers and peptide loading...........................................................28 2.3. Technologies for studying T-cell antigen specificities ................................30 2.4. Current MHC cellular microarray technologies ..........................................33 References.............................................................................................................36 CHAPTER 3 ................................................................................................................ 42 AN HLA A2-IG BASED CELLULAR MICROARRAY METHOD FOR DETECTION OF HUMAN ANTIGEN-SPECIFIC T CELLS ...................................42 Abstract .................................................................................................................42 3.1. Introduction..............................................................................................43 3.2. Materials and Methods.............................................................................46 3.3. Results......................................................................................................53 3.3.1. Overview of the anti-Ig,λ antibody-based microarray platform ......53 3.3.2. Quantifying ability of the cellular microarray assay........................56 3.3.3. Sensitivity of the cellular microarray assay .....................................57 x
3.3.4. Profiling of influenza A-associated T cell epitopes.........................61 3.4. Discussion ................................................................................................63 Supplementary information ..................................................................................76 References.............................................................................................................78 CHAPTER 4 ................................................................................................................ 82 CELLULAR MICROARRAY ASSAY FOR QUANTIFYING PEPTIDE LOADING IN MHC-IG DIMER....................................................................................................82 4.1. Introduction..............................................................................................82 4.2. Materials and Methods.............................................................................87 4.3. Results......................................................................................................90 4.3.1. Microarray assays probing peptide loading efficiency of HLA A2-Ig ……………………………………………………………………..90 4.3.2. Correlation of cell binding with cell-surface bound dimers ............92 4.3.3. Kinetic model of peptide loading to HLA A2-Ig dimer ..................94 4.4. Discussion and Conclusions ....................................................................98 Supplemental information...................................................................................109 References...........................................................................................................110 CHAPTER 5 .............................................................................................................. 114 CELLULAR MICROARRAY EVALUATION OF HUMAN CTL RESPONSE FOR INFLUENZA-A VIRUS............................................................................................114 5.1. Introduction................................................................................................114 5.2. Materials and Methods...............................................................................116 5.3. Results and Discussion ..............................................................................120 5.3.1. Fluorescence polarization measurement of peptide binding affinity to HLA A2-Ig dimer ........................................................................................120 5.3.2. Cellular microarray assay for detection of antigen-specific CTLs from PBMCs.........................................................................................................122 5.3.3. Evaluation of the breadth of CTL specificities to influenza Aassociated epitopes.......................................................................................124 5.4. Conclusions................................................................................................127 References...........................................................................................................134 CHAPTER 6 .............................................................................................................. 136 COMBINATORIAL ANTIBODY MICROARRAY FOR PERSONALIZED FORMULATION OF IMMUNOLIPOSOMES ........................................................136 6.1. Introduction............................................................................................136 6.2. Materials and methods ...........................................................................139 6.3. Results....................................................................................................141 6.3.1. Cell capture as a function of antibody and cell concentrations .....141 6.3.2. Personalized cell surface antigen expression profiles....................142 6.3.3. Combinatorial antibody microarrays for optimal targeting efficiency 142 6.3.4. Possible graphic representation and prediction of synergies .........145 6.4. Discussion and conclusions ...................................................................146 Supplementary information ................................................................................155
xi
References...........................................................................................................156 CHAPTER 7 .............................................................................................................. 158 A MIXED TWO-COMPONENT SELF-ASSEMBLED MONOLAYER-BASED PROTEIN MICROARRAY PLATFORM ................................................................158 7.1. Introduction............................................................................................158 7.2. Preparation of the platform ....................................................................160 7.3. Structural characterization by neutron scattering ..................................161 7.4. Cellular microarrays built on the platform.............................................162 References...........................................................................................................165 REFERNCES............................................................................................................. 167 APPENDIX A............................................................................................................ 182 ANTI-DEXTRAN ANTIBODY MICROARRAYS FOR CHARACTERIZATION OF FUNCTIONALIZED MAGNETIC NANOPARTICLES...................................182
xii
LIST OF TABLES Table 5.1 Comparison of assays on detection of antigen-specific CTLs ..................132 Table 5.2 Peptide specificity (No. positive donors / No. tested)...............................133 Table 6.1 Normalized mean spot intensity and binding synergy for Raji, 697, and MEC-1........................................................................................................................153 Table 7.1 Neutron scattering characterization...........................................................161
xiii
LIST OF FIGURES Figure 2.1 Schematic diagram showing MHC-Ig fusion protein (left) and artificial antigen presenting cell (aAPC, right)...........................................................................36 Figure 3.1 Schematic depiction of the cellular microarray assay in which soluble peptide-loaded HLA A2-Ig dimers are incubated with T cells in solution..................68 Figure 3.2 Effect of peptide-dissolving reagent dimethyl sulfoxide (DMSO) on antiHLA and anti-Ig antibodies binding to spots printed with dimer molecules...............69 Figure 3.3 Comparison of cellular microarray versus FACS in detection and determination of the frequency of antigen-specific CTLs in aAPC-enriched cell populations...................................................................................................................70 Figure 3.4 Quantified spot fluorescence intensity of cell suspensions containing different frequencies of CMVpp65-specific CTLs incubated with CMVpp65-loaded dimer versus unloaded dimer .......................................................................................72 Figure 3.5 Examination of cell capture across spot replicates (10 x 10 for each subarray) as a function of antigen-specific CTL frequency.........................................73 Figure 3.6 Detection of CTLs that are specific to M1.58, NA.75, and PA.46 from CD8 cells of donor A8, by HLA A2-Ig based cellular microarrays and FACS. .........75 SP. Figure 3.1 Effect of print additives on human recombinant IFNγ (rhIFNγ) detection.......................................................................................................................76 SP. Figure 3.2 FACS detection and determination of the frequency of antigenspecific CTLs enriched and expanded with aAPCs.....................................................77 Figure 4.1 The loading efficiency of HLA-A2-Ig with CMVpp65 is strongly timedependent…………………………………………………………………………....101 Figure 4.2 Quantified mean spot fluorescence intensities for images in Figure 4.1…………………………………………………………………………….……..102 Figure 4.3 Loading of HLA A2-Ig dimer with CMVpp65 peptide at 37ºC………..104 Figure 4.4 Representative FACS analyses of CMVpp65-specific CTLs by dimer staining for dimers loaded at 4ºC for different time periods......................................105
xiv
Figure 4.5 Plot of mean spot intensity (calibrated against anti-CD3 spots) versus number of PE molecules bound per cell on CMVpp65-specific CTLs, using dimer loaded with CMVpp65 peptide for varying time periods at 4ºC and 24ºC................106 Figure 4.6 Fitting of fluorescence polarization measurement of peptide p5 binding to HLA A2-Ig at 4ºC and 24ºC. .....................................................................................107 Figure 4.7 Fitting of cellular microarray measurement of peptide CMVpp65 binding to HLA A2-Ig at 4ºC and 24ºC. .................................................................................108 Figure 5.1 HLA A2-Ig dimer competition curves obtained by incubating a constant concentration of FITC labeled peptide ......................................................................129 Figure 5.2 Frequency of antigen-specific CTLs measured in different assays: cellular microarray, FACS, and ELISpot................................................................................130 Figure 5.3 Comparison of antigen-specific CTL frequencies measured in cellular microarray and in FACS. ...........................................................................................133 Figure 6.1 Dependence of Raji cell binding on cell concentration at antibody solution concentration of 0.5 mg/ml ........................................................................................148 Figure 6.2 Surface antigen expression profiles of Jurkat cells, Raji cells, and four BCLL patients...............................................................................................................149 Figure 6.3 Cell binding on combinatorial antibody microarrays for five B-CLL patients .......................................................................................................................150 Figure 6.4 Graphic representation of synergistic networks ......................................154 SP. Figure 6.1 The dependence of cell binding on antibody concentration and cell concentration in cellular microarray assays ...............................................................155 Figure 7.1 Demonstration of the multi-component, multi-layer structure of the platform......................................................................................................................160 Figure 7.2 Comparison of cell binding on gold-coated SAM-based substrate with commercial substrate Nexterion slide H ....................................................................164 Figure A.1 Immobilization of anti-PE conjugated magnetic nanoparticles on an antidextran spot................................................................................................................183 Figure A.2 Application of the nanoparticle microarray for ligand conjugation titer. ....................................................................................................................................184 Figure A.3 Potential application of the array as small scale cell sorter ....................185
xv
CHAPTER 1
INTRODUCTION 1.1. Cell-cell and cell-substrate interactions All cells, especially mammalian cells, express a variety of surface markers or receptors on their surfaces that vary from cell type to cell type. The interaction of these surface markers/receptors on one cell with their counterparts on the other cell or on an extracellular substrate, and subsequent signal transduction induced by these interactions, crucially dictate the survival, development, and function of the cell. Prototypes of cell-cell and cell-substrate interactions are antibody and antigen, T-cell receptor (TCR) and peptide-major histocompatibility complex (pMHC), lectin and carbohydrate, or integrin and extracellular matrix components such as collagen and fibronectin.
Studying cell-cell and cell-substrate interactions is of great importance for understanding cellular functions, responses and behaviors, which has had tremendous
1
impacts on a range of scientific disciplines such as immunotherapy, tissue engineering, etc.
1.2. Cellular microarrays In recent years, cellular microarray technology has emerged to meet the growing needs for interrogating intact living cells and characterizing cellular responses in a miniaturized and high-throughput manner. In cellular microarrays, probe molecules, designed to bind specific targets on cell surfaces such as proteins and glycans, are immobilized in pre-defined array patterns on planar surfaces. Binding leads to the immobilization of living cells on specific spots, which can then be detected by visual inspection, light microscopy, fluorescence or other methods. In some cases, behavioral responses of bound cells to immobilized probe molecules can be further characterized.
Collectively, applications of cellular microarrays fall into four major fields (Fernandes et al., 2009): infection and immunity, stem cell growth and differentiation, gene function silence or expression, and drug discovery.
1.2.1. Cellular microarrays in infection and immunity
The extreme importance of immunity in disease control and therapies and its overwhelmingly complex nature (Braga-Neto and Marques, 2006a) necessitate a comprehensive and high-throughput approach for its analyses. Based on the type of
2
printed probe molecules, there have been four variants of cellular microarrays in infection and immunity: antibodies, lectins, carbohydrates, and pMHC complexes (Maynard et al., 2007).
Antibody microarrays screen for cell surface antigens for immunophenotyping of cancerous cells such as leukemia and lymphoma (Belov et al., 2003; Belov et al., 2006); resulting surface antigen profiles can establish diagnosis standards, and newly identified antigen(s) can be used as target(s) in cancer immunotherapy. Lectins, proteins that recognize and bind to specific carbohydrate structural epitopes/moieties, have been immobilized to develop lectin microarrays to identify cell surface glycan signatures, which can be used for interrogation of glycosylation-involved processes such as immune cell differentiation and pathogen-host interactions (Tao et al., 2008; Tateno et al., 2007; Zheng et al., 2005). Carbohydrate microarrays have been developed to study carbohydrate-mediated adhesion of bacteria or mammalian cells, which have potential use in pathogen detection and screening of anti-adhesion therapeutics (Disney and Seeberger, 2004; Nimrichter et al., 2004). In pMHC complex microarrays, multimeric pMHC complexes, streptavidin-based tetramers (Altman et al., 1996) or immunoglobulin (Ig)-based dimers (Greten et al., 1998), are printed in predetermined spatial patterns; T cells are then contacted with the substrate, and subsequently captured onto specific spots, as determined by their antigen specificities, through the interaction of T cell receptors (TCRs) on the T-cell surface with pMHCs immobilized on the substrate. This assay has shown great potential as a fast and cost-effective method for characterizing T-cell antigen specificities in heterogeneous T-cell populations; in addition, T-cell responses upon activation by
3
specific pMHC complexes, such as cytokine secretion and calcium release, can be detected or monitored (Chen et al., 2005; Deviren et al., 2007; Soen et al., 2003; Stone et al., 2005).
1.2.2. Cellular microarrays in stem cell research
Due to their unique characteristics such as unlimited self-renewal capacity and differentiation into functionally distinct mature cells, stem cells have been gaining extensive research interest, in the efforts to develop novel regenerative medicines for devastating diseases. Microarray technology offers a high-throughput platform for studying the complex yet poorly understood underlying mechanisms that dictate the survival, proliferation, and differentiation of stem cells. Specifically, microarrays immobilized with extracellular matrix (ECM) molecules (Flaim et al., 2005), synthetic polymers (Anderson et al., 2004), solely or in combination with growth factors (Soen et al., 2006), have been developed for probing the effects of cellular microenvironments on stem cell-cell and cell-substrate interactions.
1.2.3. Cellular microarrays in gene function studies
Understanding gene functions has been a research focus in this post-genomic era. Cell microarray technology offers an approach for systematic and high-throughput in vivo analysis of gene function in mammalian cells. Complementary DNAs (cDNAs) or DNAs for over-expression studies of genes (Ziauddin and Sabatini, 2001) or RNA interference (RNAi) for loss-of-function studies of genes (Silva et al., 2004; Vanhecke and Janitz, 2004) are directly printed onto microarray slides, and cells are incubated with printed slides and then locally transfected on the spots. Transfection
4
leads to expression or inhibition of gene products or alteration of cellular phenotypes, which can be detected accordingly using a variety of functional assays. In a recent study, genes were delivered into cells by infection with printed lentiviruses, the use of which broaden the types of mammalian cells that can be transfected and enable more robust transfection (Bailey et al., 2006).
1.2.4. Cellular microarrays in drug discovery Cellular microarray technology has shown promises in the process of drug discovery in mainly two specific aspects: small molecule screening and drug cytotoxicity screening. One of the first small molecule cellular microarray systems was established by impregnating small molecules in spots of a biodegradable polymer for slow diffusive release of the chemical compounds; cells were seeded on top of the spots and effects of each compound on cell viability were evaluated (Bailey et al., 2004). A new platform for screening for G-protein-coupled receptor agonists, which constitute approximately 30% of currently approved drugs, were realized by assembling a precasted cell-agarose gel on top of a polystyrene sheet that was printed with as many as 260,000 compounds; calcium release indicating agonist effect was measured by preloaded calcium-sensitive fluorescence dye (Gopalakrishnan et al., 2003).
1.3. Scope of current work The development of cellular microarrays for studies in the field of infection and immunity is of our great interest in this work. Most cellular microarrays in this field are essentially extension of protein microarrays, since antibodies, lectins, and pMHC
5
complexes are all proteins. For the rest of this chapter, a detailed review of protein microarray technology will be provided, with an emphasis on the substrates and fabrication of protein microarrays. Chapter 2 introduces in detail the molecular interaction between pMHC and TCR, current methods for studying the specificities of TCRs, and previously-developed pMHC cellular microarrays. Chapter 3 presents the development of a novel pMHC-Ig dimer cellular microarray for quantitative characterization of antigen-specific T cells, which outperforms conventional spotted pMHC cellular microarrays and has several distinct advantages over the gold-standard technology flow-activated cell sorting (FACS). In Chapter 4, we demonstrate the application of the developed pMHC-Ig dimer cellular microarray to studying peptide loading into MHC-Ig dimer. A mechanistic model describing the peptide loading process was proposed, and optimal loading condition was accordingly derived. In Chapter 5, the pMHC-Ig dimer cellular microarray is applied to quantitative profiling of T cells specific to influenza A virus in the complex heterogeneous population of donor peripheral blood mononuclear cells (PBMCs). The cellular microarray assays obtained highly-correlated results with FACS. Independent from previous chapters, Chapter 6 presents the development of a combinatorial antibody microarray that has the great potential for fast screening of the optimal combinations of antibodies for use in individualized cancer immunotherapy. We further developed a universal selfassembled monolayer (SAM)-based protein microarray substrate platform, and conducted structural characterization and performance evaluation of the platform, as summarized in Chapter 7. Lastly, we present in Appendix A the development of a nanoparticle-immobilized microarray that can be used for functional characterization of (magnetic) nanoparticles and has the potential to be a small-scale cell sorter.
6
1.4. Protein microarrays Modern protein microarray technology is based on DNA microarray technology that has been applied for gene expression analysis, gene mapping and the sequencing of entire genomes (Choudhuri, 2004; http://www.affymetrix.com/). DNA microarrays enable simultaneous monitoring of the RNA expression of thousands of genes in a single assay (Lockhart et al., 1996; Schena et al., 1995). However, most genes function through their protein products, and the extreme diversity and large dynamic range of protein functions (Feilner et al., 2004) produce a strong incentive to study biological processes at the level of proteins, which has driven the development of protein microarray technology (Macbeath and Schreiber, 2000; Zhu et al., 2001).
Protein microarrays are made in much the same way as DNA microarrays (Macbeath and Schreiber, 2000; Service, 2001; Utz, 2005). Glass or polymeric surfaces are modified with an organic layer that is used for protein immobilization, and probe molecules designed to bind specific target proteins are printed onto this planar surface in a predetermined pattern of spatially distinct spots that are on the order of several hundred microns up to a few millimeters in diameter. The probe molecules can be other proteins, such as avidin or streptavidin, antibodies or aptamers, antibody-like molecules, etc. Fluorescent markers or other methods of detection are used to reveal which locations or spots on the microarray captured proteins. Since the identity of the specific protein-protein complex that is formed on each spot is known, detection of binding at a particular spot identifies the captured target protein.
7
One of the first protein microarray applications was to analyze the proteome of yeast (Zhu et al., 2001). Since then, protein microarrays have been used in high throughput screening for protein biomarker discovery, protein characterization, protein-protein interaction mapping, pathogen/toxin detection, and assay development and validation (Braga-Neto and Marques, 2006b; Eisenstein, 2006; Winssinger et al., 2002). The stability of immunoglobulins and the considerable amount of information that is available on the properties of immunoglobulins have led to the development of a wide selection of antibody microarrays, including antibody microarrays for profiling protein expression of cancer antigens (Bartling et al., 2005; Gao et al., 2005; Knezevic et al., 2001). Monoclonal and recombinant antibodies provide a source of probe molecules offering essentially unlimited possibilities for novel genetic constructs. As mentioned earlier, the development of cellular microarrays is a more recent extension of protein microarray technology (Angres, 2005; Castel et al., 2006; Chen and Davis, 2006; Diaz-Mochon et al., 2007). Although still in its infancy, this technology has shown the potential for high-throughput cell sorting, the characterization of surface antigen repertoires of leukaemia cells (Belov et al., 2001; Belov et al., 2005; Belov et al., 2006; Kato et al., 2007), and the characterization of Tcell repertoires based on their antigen specificities (Deviren et al., 2007; Soen et al., 2003). Co-printing stimulatory antibodies and anti-cytokine antibodies with the probe molecules on these microarrays enables the simultaneous detection and screening of cellular functional responses, such as cytokine production (Chen et al., 2005; Stone et al., 2005).
8
By far the greatest challenge facing the development of protein or cellular microarray technology is how to apply the technology to obtain quantitative high-throughput assays. In contrast to DNA microarrays, development of these microarrays is hindered by the molecular complexity of the proteins used as probe molecules (Feilner et al., 2004; Kenyon et al., 2002). Proteins are polypeptide chains composed of 20 different amino acids that fold into unique three-dimensional structures specific to their molecular biological function. These macromolecular structures are sensitive to their environment, and are prone to unfold and break down under chemical, physical, or mechanical stresses. As noted above, antibodies alone provide a source of probe molecules that is essentially unlimited in number compared to the number of genes in the human genome. The quantitative application of protein or cellular microarrays is challenging in large part because of the large variability in the molecular biological/physicochemical nature of proteins, exacerbated by the lack of control over many protein-specific chemical, physical, and biological processes that are associated with chip fabrication.
For example, each protein must be immobilized on the microarray surface in such a way that its native state is preserved and binding site accessible (Nishioka et al., 2004). One major reason for poor quantitative results is that printing protein solutions in nanoliter volumes produces rapid evaporation that can lead to dramatic changes in protein concentration, ionic strength, protein hydration, and pH (Wu and Grainger, 2006). The result is protein denaturation or aggregation, and ultimately, an assay governed by protein-protein interactions not necessarily specific to bio-recognition.
9
Another common occurrence that produces poor spot-to-spot reproducibility is the formation of so-called “coffee rings” – i.e., non-uniform, ring-like protein concentration profiles across individual spots printed on a microarray (Deegan et al., 1997). This phenomenon is attributed to the transport of proteins molecules along the air/water interface to the perimeter of the spots, which only partially wet the solid substrate. The effect becomes more pronounced as the spot size decreases. This transport process competes with the reaction kinetics for immobilization of the proteins on the substrate (Deng and Zhu, 2006). Thus, optimizing the immobilization chemistry concurrently with the solution conditions for printing nanoliter volumes of protein solution is a major consideration in chip fabrication (Ajikumar et al., 2007; Liu et al., 2007; Schroeder et al., 2007).
Proteins also tend to adsorb nonspecifically to solid substrates, leading to reduced sensitivity and a lower signal-to-noise ratio for the quantitative detection of specific binding. The choice of substrate that will minimize nonspecific binding is, therefore, another important consideration (Zhu and Snyder, 2003).
1.4.1. Substrates for protein microarrays
One of the key steps in protein or cellular microarray development is the choice of an appropriate solid support/substrate that meets two key requirements (Angenendt et al., 2002). First, the substrate must provide high binding capacity. Second, it must provide a non-denaturing environment to prevent proteins that are printed on them from
10
unfolding, while preserving their functionality and the accessibility of their binding sites.
Currently, most protein microarrays are manufactured using modified glass microscopic slides as the substrate. These microarray surfaces fall into two major categories (Guilleaume et al., 2005): non-gel-coated two-dimensional surfaces and gel-coated three-dimensional surfaces. Non-gel coated surfaces are surfaces modified with different chemicals, such as poly-L-lysine (Ge, 2000; Snapyan et al., 2003), epoxy (Angenendt et al., 2003), or aldehyde (Macbeath and Schreiber, 2000); while gel-coated surfaces are coated with polyacrylamide (Arenkov et al., 2000), agarose (Afanassiev et al., 2000), nitrocellulose, etc. The production of substrates for protein microarrays has been commercialized (Eisenstein, 2006), and various products have been compared in terms of the spot characteristics, limit of detection (LOD), and signal-to-noise (S/N) ratio (Angenendt et al., 2002; Guilleaume et al., 2005). These comparisons reveal that there is no single substrate suitable for every application. Thus, a considerable amount of time and effort is required to identify the optimal substrate for a particular application, taking into account, for example, the proteins to be immobilized, the target species, and the fluorescence dyes to be used for detection. Davis and coworkers compared glass slides coated with polyacrylamide, aldehyde, poly-L-Lysine and nitrocellulose, and found that polyacrylamide-coated slides worked best for their specific cellular microarray application (Chen et al., 2005; Soen et al., 2003). Stern and coworkers evaluated polystyrene, permanox, and LabTek II CC2 slides obtained from Nalge as solid supports and found that simple adsorption to
11
these slides to be straightforward and reproducible in their cellular microarray applications (Stone et al., 2005).
On commercial microarray substrates, such as Nexterion slide H, proteins are immobilized onto the surface by either physical adsorption or covalent bonding in a random orientation. Since specific protein-protein interactions, such as antibodyantigen binding, will only occur when the proteins can adopt appropriate molecular orientations to one another, there have been considerate efforts in developing novel substrates that can immobilize proteins in an oriented or directed fashion. For example, surfaces coated with avidin (Rowe et al., 1999) and nickel (Zhu et al., 2001) have been designed to immobilize biotinylated and Hisx6-tagged proteins, respectively, in a specific orientation through direct coupling of these substituent pairs. Surfaces modified with polyelectrolyte multilayers have been used to control the orientation of immobilized proteins (Diederich and Losche, 1996). Mixed selfassembled monolayers (SAMs) of biotinylated alkyl thiols formed on gold-coated surfaces can also be applied to control the orientation of immobilized proteins (Bieri et al., 1999; Silin et al., 2006; Spinke et al., 1993). The SAM is formed, in this case, by attaching one end to the surface through the reaction of the thiol group with gold, while the biotinylated end binds to streptavidin, which can then couple to other biotinylated proteins (secondary capture proteins). In addition, the surface density of the secondary capture proteins can be controlled by adjusting the molar ratio of biotinylated and non-biotinylated alkyl thiols (Nelson et al., 2001).
12
1.4.2. Fabrication of protein microarrays
The fabrication of protein microarrays involves the deposition of proteins onto substrates in predetermined patterns or microarrays of spots. A number of techniques have been developed for depositing proteins onto solid substrates, including electrospray deposition (Morozov and Morozova, 1999), microstamping (Martin et al., 1998), microdispensing (Laurell et al., 1999), and soft-landing of mass-selected ions (Zheng et al., 2003), etc. However, printing is the most commonly used technique. Manual printing is done by simply pipetting protein samples onto a substrate. Sample sizes are typically on the order of 0.1 µl, which produce spots approximately 1 mm in diameter (Deviren et al., 2007; Kato et al., 2007). The need for miniaturization and automation has led to the development of robotic printing techniques. Both contact and non-contact robotic printers have been commercialized for depositing picoliter to nanoliter volumes of protein samples onto protein microarray substrates (Zhu and Snyder, 2003). Contact printers use solid or split pins to deposit samples to a surface. The pins are first immersed into the wells containing the samples of interest in order to coat or fill the pins with the samples. The pins are then positioned over the solid substrate, and the samples are transferred to the substrate by briefly contacting the substrate with the pin. Non-contact printers typically use piezoelectric dispensers consisting of a piezoelectric element that forms a collar around a borosilicate glass capillary. Upon application of a voltage, the collar compresses the glass capillary that has been pre-filled with a sample, which ejects a single droplet of the sample several hundred picoliters in volume onto the substrate. The applied voltage controls the volume of the droplet dispensed from the capillary. In addition, the number of drops
13
(typically 1-300) to be deposited on each spot can also be programmed in noncontact printers.
The choice of printer depends largely on the application; both printers have advantages and disadvantages (Mukhopadhyay, 2006). With contact printers, problems can arise due to the contact between the pins and the substrate. For example, the pins can damage soft surfaces, resulting in large background signals. With noncontact printers, proteins tend to stick non-specifically to the borosilicate glass dispensers, resulting in the loss of samples. The addition of a non-functional protein, such as bovine serum albumin (BSA), can help minimize this problem (Delehanty and Ligler, 2003). The small sample volumes that are being transferred in both contact printing and non-contact printing can cause rapid evaporation of the samples and an increase of sample viscosity during printing, which leads to high spot-to-spot variability. Rapid evaporation is more of an issue with contact printing than noncontact printing, since the volume of sample being dispensed is better controlled in non-contact printing.
Besides the type of printer, there are two other factors during the fabrication stage that affect the performance of protein microarrays. One is spot morphology. Ring structure of spots is commonly observed in protein microarrays. Prior studies on simple, dilute aqueous solutions suggested that the formation of ring structure on a solid surface as a drop dries is caused by a capillary flow (Deegan et al., 1997). As spot evaporation proceeds, pinning of the contact line of the drying drop drives a capillary flow toward the periphery (to maintain the original radius of the spot), transporting the solute from
14
the center to the periphery, and thus leaving a characteristic ring pattern upon drying. However, this cannot explain the formation of ring structure when proteins are hydrated in an environment with controlled humidity. Recent studies showed that the transport of proteins accumulated at the air/water interface to the periphery of spots is the reason causing the formation of ring structure in protein microarrays. They demonstrated experimentally that the ring structure can be eliminated by using a small amount of surfactants such as Trition X-100 as print additive, or by incorporating “facile surface reaction” (such as Cu2+ and 6xHis tag) for protein immobilization (Deng and Zhu, 2006). The other factor that needs to be taken into account is print additives, to improve spot morphology as discussed above or to maintain protein stability. Different studies recommend different additives, such as glycerol (10% (Ho et al., 2006), 40% (Macbeath and Schreiber, 2000)), PEG 200 (Lee and Kim, 2002), trehalose (Nishioka et al., 2004), and polyvinyl alcohol (Wu and Grainger, 2006), and it appeared that the choice of appropriate additive depends on the application system. This is because that the mechanisms responsible for protein instability during printing are not totally understood, and how each additive interacts with proteins to improve protein stability is not clear either.
References Afanassiev, V., V. Hanemann, andS. Wölfl. 2000. Preparation of DNA and protein micro arrays on glass slides coated with an agarose film. Nucleic Acids Res 28:e66-e66.
15
Ajikumar, P.K., J. Kiat, Y.C. Tang, J.Y. Lee, G. Stephanopoulos, andH.P. Too. 2007. Carboxyl-terminated dendrimer-coated bioactive interface for protein microarray: High-sensitivity detection of antigen in complex biological samples. Langmuir 23(10):5670-5677. Altman, J.D., P.A.H. Moss, P.J.R. Goulder, D.H. Barouch, M.G. McHeyzerWilliams, J.I. Bell, A.J. McMichael, andM.M. Davis. 1996. Phenotypic analysis of antigen-specific T lymphocytes. Science 274(5284):94-96. Anderson, D.G., S. Levenberg, andR. Langer. 2004. Nanoliter-scale synthesis of arrayed biomaterials and application to human embryonic stem cells. Nat Biotechnol 22(7):863-866. Angenendt, P., J. Glokler, D. Murphy, H. Lehrach, andD.J. Cahill. 2002. Toward optimized antibody microarrays: a comparison of current microarray support materials. Analytical Biochemistry 309(2):253-260. Angenendt, P., J. Glokler, J. Sobek, H. Lehrach, andD.J. Cahill. 2003. Next generation of protein microarray support materials: Evaluation for protein and antibody microarray applications. Journal of Chromatography A 1009(12):97-104. Angres, B. 2005. Cell microarrays. Expert Review of Molecular Diagnostics 5(5):769779. Arenkov, P., A. Kukhtin, A. Gemmell, S. Voloshchuk, V. Chupeeva, andA. Mirzabekov. 2000. Protein microchips: Use for immunoassay and enzymatic reactions. Analytical Biochemistry 278(2):123-131. Bailey, S.N., S.M. Ali, A.E. Carpenter, C.O. Higgins, andD.M. Sabatini. 2006. Microarrays of lentiviruses for gene function screens in immortalized and primary cells. Nature Methods 3(2):117-122. Bailey, S.N., D.M. Sabatini, andB.R. Stockwell. 2004. Microarrays of small molecules embedded in biodegradable polymers for use in mammalian cellbased screens. Proc Natl Acad Sci U S A 101(46):16144-16149. Bartling, B., H.S. Hofmann, T. Boettger, G. Hansen, S. Burdach, R.E. Silber, andA. Simm. 2005. Comparative application of antibody and gene array for expression profiling in human squamous cell lung carcinoma. Lung Cancer 49(2):145-154.
16
Belov, L., O. de la Vega, C.G. dos Remedios, S.P. Mulligan, andR.I. Christopherson. 2001. Immunophenotyping of leukemias using a cluster of differentiation antibody microarray. Cancer Research 61(11):4483-4489. Belov, L., P. Huang, N. Barber, S.P. Mulligan, andR.I. Christopherson. 2003. Identification of repertoires of surface antigens on leukemias using an antibody microarray. Proteomics 3(11):2147-2154. Belov, L., P. Huang, J.S. Chrisp, S.P. Mulligan, andR.I. Christopherson. 2005. Screening microarrays of novel monoclonal antibodies for binding to T-, Band myeloid leukaemia cells. Journal of Immunological Methods 305(1):10-19. Belov, L., S.P. Mulligan, N. Barber, A. Woolfson, M. Scott, K. Stoner, J.S. Chrisp, W.A. Sewell, K.F. Bradstock, L. Bendall, D.S. Pascovici, M. Thomas, W. Erber, P. Huang, M. Sartor, G.A.R. Young, J.S. Wiley, S. Juneja, W.G. Wierda, A.R. Green, M.J. Keating, andR.I. Christopherson. 2006. Analysis of human leukaemias and lymphomas using extensive immunophenotypes from an antibody microarray. British Journal of Haematology 135(2):184-197. Bieri, C., O.P. Ernst, S. Heyse, K.P. Hofmann, andH. Vogel. 1999. Micropatterned immobilization of a G protein-coupled receptor and direct detection of G protein activation. Nature Biotechnology 17(11):1105-1108. Braga-Neto, U.M., andE.T. Marques, Jr. 2006a. From functional genomics to functional immunomics: new challenges, old problems, big rewards. PLoS Comput Biol 2(7):e81. Braga-Neto, U.M., andE.T.A. Marques. 2006b. From functional genomics to functional immunomics: New challenges, old problems, big rewards. Plos Computational Biology 2(7):651-662. Castel, D., A. Pitaval, M.A. Debily, andX. Gidrol. 2006. Cell microarrays in drug discovery. Drug Discovery Today 11(13-14):616-622. Chen, D.S., andM.M. Davis. 2006. Molecular and functional analysis using live cell microarrays. Current Opinion in Chemical Biology 10(1):28-34. Chen, D.S., Y. Soen, T.B. Stuge, P.P. Lee, J.S. Weber, P.O. Brown, andM.M. Davis. 2005. Marked differences in human melanoma antigen-specific T cell responsiveness after vaccination using a functional microarray. Plos medicine 2(10):1018-1030.
17
Choudhuri, S. 2004. Microarrays in biology and medicine. Journal of Biochemical and Molecular Toxicology 18(4):171-179. Deegan, R.D., O. Bakajin, T.F. Dupont, G. Huber, andS.R.N.T.A. Witten. 1997. Capillary flow as the cause of ring stains from dried liquid drops. Nature 389:827-829. Delehanty, J.B., andF.S. Ligler. 2003. Method for printing functional protein microarrays. BioTechniques 34:380-385. Deng, Y., andX.-Y. Zhu. 2006. Transport at the Air/Water Interface is the Reason for Rings in Protein Microarrays. JACS 128:2768-2769. Deviren, G., K. Gupta, M.E. Paulaitis, andJ.P. Schneck. 2007. Detection of antigenspecific T cells on p/MHC microarrays journal of molecular recognition 20(1):32-38. Diaz-Mochon, J.J., G. Tourniaire, andM. Bradley. 2007. Microarray platforms for enzymatic and cell-based assays. Chemical Society Reviews 36(3):449-457. Diederich, A., andM. Losche. 1996. Novel biosensoric devices based on molecular protein hetero-multilayer films. Advances in Biophysics, Vol 34, 1997 34:205230. Disney, M.D., andP.H. Seeberger. 2004. The use of carbohydrate microarrays to study carbohydrate-cell interactions and to detect pathogens. Chem Biol 11(12):1701-1707. Eisenstein, M. 2006. Protein arrays - Growing pains. Nature 444(7121):959-964. Feilner, T., J. Kreutzberger, B. Niemann, A. Kramer, A. Possling, H. Seitz, andB. Kersten. 2004. Proteomic studies using microarrays. Current Proteomics 1:283-295. Fernandes, T.G., M.M. Diogo, D.S. Clark, J.S. Dordick, andJ.M. Cabral. 2009. Highthroughput cellular microarray platforms: applications in drug discovery, toxicology and stem cell research. Trends Biotechnol 27(6):342-349.
18
Flaim, C.J., S. Chien, andS.N. Bhatia. 2005. An extracellular matrix microarray for probing cellular differentiation. Nat Methods 2(2):119-125. Gao, W.M., R. Kuick, R.P. Orchekowski, D.E. Misek, J. Qiu, A.K. Greenberg, W.N. Rom, D.E. Brenner, G.S. Omenn, B.B. Haab, andS.M. Hanash. 2005. Distinctive serum protein profiles involving abundant proteins in lung cancer patients based upon antibody microarray analysis. Bmc Cancer 5:-. Ge, H.e.a. 2000. UPA, a universal protein array system for quantitative detection of protein-protein, protein-DNA, protein-RNA and proteinligand interactions. (2000) 28: e3. Nucleic Acids Res. 28:e3. Gopalakrishnan, S.M., R.B. Moreland, J.L. Kofron, R.J. Helfrich, E. Gubbins, J. McGowen, J.N. Masters, D. Donnelly-Roberts, J.D. Brioni, D.J. Burns, andU. Warrior. 2003. A cell-based microarrayed compound screening format for identifying agonists of G-protein-coupled receptors. Anal Biochem 321(2):192-201. Greten, T.F., J.E. Slansky, R. Kubota, S.S. Soldan, E.M. Jaffee, T.P. Leist, D.M. Pardoll, S. Jacobson, andJ.P. Schneck. 1998. Direct visualization of antigenspecific T cells: HTLV-1 Tax11-19-specific CD8(+) T cells are activated in peripheral blood and accumulate in cerebrospinal fluid from HAM/TSP patients. Proceedings of the National Academy of Sciences of the United States of America 95(13):7568-7573. Guilleaume, B., A. Buneß, C. Schmidt, F. Klimek, G. Moldenhauer, W. Huber, D. Arlt, U. Korf, S. Wiemann, andA. Poustka. 2005. Systematic comparison of surface coatings for protein microarrays. PROTEOMICS 5(18):4705-4712. Ho, S.-W., G. Jona, C.T.L. Chen, M. Johnston, andM. Snyder. 2006. Linking DNAbinding proteins to their recognition sequences by using protein microarrays. PNAS 103:9940-9945. http://www.affymetrix.com/. Kato, K., M. Toda, andH. Iwata. 2007. Antibody arrays for quantitative immunophenotyping. biomaterials 28:1289-1297. Kenyon, G.L., D.M. DeMarini, E. Fuchs, D.J. Galas, J.F. Kirsch, T.S. Leyh, W.H. Moos, G.A. Petsko, D. Ringe, G.M. Rubin, L.C. Sheahan, N.R.C.S. Comm, and
19
2002. Defining the mandate of proteomics in the post-genomics era: Workshop report. Molecular & Cellular Proteomics 1(10):763-780. Knezevic, V., C. Leethanakul, V.E. Bichsel, J.M. Worth, V.V. Prabhu, J.S. Gutkind, L.A. Liotta, P.J. Munson, E.F. Petricoin, andD.B. Krizman. 2001. Proteomic profiling of the cancer microenvironment by antibody arrays. Proteomics 1(10):1271-1278. Laurell, T., L. Wallman, andJ. Nilsson. 1999. Design and development of a silicon microfabricated flow-through dispenser for on-line picolitre sample handling. J. Micromech. Microeng 9:369-376. Lee, C.-S., andB.-G. Kim. 2002. Improvement of protein stability in protein microarrays. Biotechnology Letters 24:839-844. Liu, Y.S., C.M. Li, L. Yu, andP. Chen. 2007. Optimization of printing buffer for protein microarrays based on aldehyde-modified glass slides. Frontiers in Bioscience 12:3768-3773. Lockhart, D.J., H.L. Dong, M.C. Byrne, M.T. Follettie, M.V. Gallo, M.S. Chee, M. Mittmann, C.W. Wang, M. Kobayashi, H. Horton, andE.L. Brown. 1996. Expression monitoring by hybridization to high-density oligonucleotide arrays. Nature Biotechnology 14(13):1675-1680. Macbeath, G., andS.L. Schreiber. 2000. Printing proteins as microarrays for highthroughput function determination. science 289:1760-1763. Martin, B.D., B.P. Gaber, C.H. Patterson, andD.C. Turner. 1998. Direct Protein Microarray Fabrication Using a Hydrogel “Stamper”. Langmuir 14:3971-3975. Maynard, J.A., R. Myhre, andB. Roy. 2007. Microarrays in infection and immunity. Curr Opin Chem Biol 11(3):306-315. Morozov, V.N., andT.Y. Morozova. 1999. Electrospray Deposition as a Method for Mass Fabrication of Mono- and Multicomponent Microarrays of Biological and Biologically Active Substances. Analytical chemistry 71:3110-3117.
20
Mukhopadhyay, R. 2006. The versatility of microarrayers. Analytical chemistry 78(17):5969-5972. Nelson, K.E., L. Gamble, L.S. Jung, M.S. Boeckl, E. Naeemi, S.L. Golledge, T. Sasaki, D.G. Castner, C.T. Campbell, andP.S. Stayton. 2001. Surface characterization of mixed self-assembled monolayers designed for streptavidin immobilization. Langmuir 17(9):2807-2816. Nimrichter, L., A. Gargir, M. Gortler, R.T. Altstock, A. Shtevi, O. Weisshaus, E. Fire, N. Dotan, andR.L. Schnaar. 2004. Intact cell adhesion to glycan microarrays. Glycobiology 14(2):197-203. Nishioka, G.M., A.A. Markey, andC.K. Holloway. 2004. Protein damage in drop-ondemand printers. J. Am. Chem. Soc. 126(50):16320-16321. Rowe, C.A., L.M. Tender, M.J. Feldstein, J.P. Golden, S.B. Scruggs, B.D. MacCraith, J.J. Cras, andF.S. Ligler. 1999. Array biosensor for simultaneous identification of bacterial, viral, and protein analytes. Analytical Chemistry 71(17):38463852. Schena, M., D. Shalon, R.W. Davis, andP.O. Brown. 1995. Quantitative Monitoring of Gene-Expression Patterns with a Complementary-DNA Microarray. Science 270(5235):467-470. Schroeder, H., B. Ellinger, C.F.W. Becker, H. Waldmann, andC.M. Niemeyer. 2007. Generation of live-cell microarrays by means of DNA-directed immobilization of specific cell-surface ligands. Angewandte Chemie-International Edition 46(22):4180-4183. Service, R.F. 2001. Protein chips - Searching for recipes for protein chips. Science 294(5549):2080-2082. Silin, V.I., E.A. Karlik, K.D. Ridge, andD.J. Vanderah. 2006. Development of surface-based assays for transmembrane proteins: Selective immobilization of functional CCR5, a G protein-coupled receptor. Analytical Biochemistry 349(2):247-253. Silva, J.M., H. Mizuno, A. Brady, R. Lucito, andG.J. Hannon. 2004. RNA interference microarrays: High-throughput loss-of-function genetics in mammalian cells. Proceedings of the National Academy of Sciences of the United States of America 101(17):6548-6552.
21
Snapyan, M., M. Lecocq, L. Guevel, M.C. Arnaud, A. Ghochikyan, andV. Sakanyan. 2003. Dissecting DNA-protein and protein-protein interactions involved in bacterial transcriptional regulation by a sensitive protein array method combining a near-infrared fluorescence detection. Proteomics 3(5):647-657. Soen, Y., D.S. Chen, D.L. Kraft, M.M. Davis, andP.O. Brown. 2003. Detection and characterization of cellular immune responses using peptide-MHC microarrays. Plos biology 1(3):429-438. Soen, Y., A. Mori, T.D. Palmer, andP.O. Brown. 2006. Exploring the regulation of human neural precursor cell differentiation using arrays of signaling microenvironments. Mol Syst Biol 2:37. Spinke, J., M. Liley, H.J. Guder, L. Angermaier, andW. Knoll. 1993. Molecular Recognition at Self-Assembled Monolayers - the Construction of Multicomponent Multilayers. Langmuir 9(7):1821-1825. Stone, J.D., J. Walter E. Demkowicz, andL.J. Stern. 2005. HLA-restricted epitope identification and detection of functional T cell responses by using MHCpeptide and costimualtory microarrays PNAS 102:3744-3749. Tao, S.C., Y. Li, J. Zhou, J. Qian, R.L. Schnaar, Y. Zhang, I.J. Goldstein, H. Zhu, andJ.P. Schneck. 2008. Lectin microarrays identify cell-specific and functionally significant cell surface glycan markers. Glycobiology 18(10):761769. Tateno, H., N. Uchiyama, A. Kuno, A. Togayachi, T. Sato, H. Narimatsu, andJ. Hirabayashi. 2007. A novel strategy for mammalian cell surface glycome profiling using lectin microarray. Glycobiology 17(10):1138-1146. Utz, P.J. 2005. Protein arrays for studying blood cells and their secreted products. Immunological Reviews 204:264-282. Vanhecke, D., andM. Janitz. 2004. High-throughput gene silencing using cell arrays. Oncogene 23(51):8353-8358. Winssinger, N., S. Ficarro, P.G. Schultz, andJ.L. Harris. 2002. Profiling protein function with small molecule microarrays. Proceedings of the National Academy of Sciences of the United States of America 99(17):11139-11144.
22
Wu, P., andD.W. Grainger. 2006. Comparison of hydroxylated print additives on antibody microarray performance. Journal of proteome research 5(11):29562965. Zheng, O., Z. Takáts, T.A. Blake, B. Gologan, A.J. Guymon, J.M. Wiseman, J.C. Oliver, V.J. Davisson, andR.G. Cooks. 2003. Preparing Protein Microarrays by Soft-Landing of mass-selected ions. Science 301:1351-1354. Zheng, T., D. Peelen, andL.M. Smith. 2005. Lectin arrays for profiling cell surface carbohydrate expression. J Am Chem Soc 127(28):9982-9983. Zhu, H., M. Bilgin, R. Bangham, D. Hall, A. Casamayor, P. Bertone, N. Lan, R. Jansen, S. Bidlingmaier, T. Houfek, T. Mitchell, P. Miller, R.A. Dean, M. Gerstein, andM. Snyder. 2001. Global analysis of protein activities using proteome chips. science 293:2101-2105. Zhu, H., andM. Snyder. 2003. Protein chip technology. current opinion in chemical biology 7:55-63. Ziauddin, J., andD.M. Sabatini. 2001. Microarrays of cells expressing defined cDNAs. Nature 411(6833):107-110.
23
CHAPTER 2 OVERVIEW OF MHC CELLULAR MICROARRAY TECHNOLOGY
2.1. Molecular recognition events of T cell-mediated adaptive immune response At the core of the molecular events comprising adaptive immune response is the interaction of the T-cell receptor (TCR) on T cell with a complementary major histocompatibility complex (MHC) mediated by a pathogen-derived small peptide called epitope, presented on the surface of so-called antigen-presenting cell (APC).
There are two types of MHC molecules, class I and class II, possessing distinct molecular structures and functions. Each MHC Class I molecule consists of an α heavy chain and a non-covalently associated β2-microglobulin (β2m), and presents peptides that are derived from intracellular pathogens including viruses and some bacteria, fungi and parasites. The peptide-binding cleft of MHC class I molecules,
24
formed by the α1 and α2 domains of the α chain, has closed ends and thus only accommodates short antigenic peptides of typically 8-10 residues. In contrast, each MHC class II molecule is a heterodimer consisting of non-covalently associated α and β chains, and presents peptides that are derived from extracellular pathogens including some bacteria, fungi and parasites. The peptide-binding cleft of MHC class II molecules, formed by the α1 and β1 domains of the α and β chains respectively, has open ends and can accommodate antigenic peptides of 10-30 residues or longer. The mechanisms and pathways for antigen processing are not covered here; for further interest, please refer to (Charles A. Laneway et al., 2005).
In accordance with the two distinct types of peptide-MHC complexes, mature T cells fall into two subsets based on function as well as surface phenotype. Those expressing surface glycoprotein CD8 recognize peptide-MHC class I complexes and are critical in the elimination of infected host cells also called; those expressing CD4 recognize peptide-MHC class II complexes and are important in release cytokines to mediate the activities of other cells such as B cells. The functions of CD4 and CD8 receptors have been known to transduce signals as well as bind to MHC molecules. Besides CD4 or CD8, all T cells express CD3, a complex of small polypeptides that are noncovalently associated with TCR and is invariant from T cell to T cell; CD3 is essential for TCR expression on cell plasma membrane and signal transduction once peptide-MHC
25
complex has bound to the TCR. In this work, CD8+ T cells, also called cytotoxic T lymphocytes (CTLs), are of our major research interest.
Immune response to intracellular infections often appears to be restricted to a small fraction of all potential peptide epitopes, which is referred to as immunodominance, with responses to influenza or other viruses, such as Epstein-Barr, as notable examples (Callan et al., 1998; Chen et al., 2000; Davenport et al., 2002; Gianfrani et al., 2000; Hill et al., 1995; Tan et al., 1999). In contrast, the response to HIV or hepatitis, among other pathogens, appears very broad, and the diversity of this response seems to correlate with its effectiveness. A quantitative assessment of this diversity can facilitate the discovery of new peptide antigens (Gianfrani et al., 2000). Factors that impact the heterogeneity of the immune response and the immunodominance of certain peptides are clearly important for understanding the underlying molecular mechanisms of adaptive immune response, as well as for biotechnological applications (Chen et al., 2000).
On the molecular level of T cell-APC interactions, there are both peptide-specific signals – referred to as signal 1 - delivered through the TCR, and co-stimulatory signals – referred to as signal 2 - delivered through the interaction of accessory molecules on the T cell surface with complexes on the surface of the APC. Signal 2 is required for the activation of naïve T cells to differentiate into effector T cells, but not
26
required for the activation of effector T cells. Above all else, TCR activation depends critically on the binding affinity of the TCR with the pMHC. Surface plasmon resonance studies of pMHC/TCR interactions indicate that while the overall range of affinities of stimulatory pMHC ligands is low relative to that for antibody-antigen interactions, they need to be sufficiently high (>10-5M) to induce T cell activation (Corr et al., 1994; Davis and Chien, 1993).
Another requirement for T cell activation is TCR cross-linking. The importance of TCR cross-linking is inferred from the observation that monovalent pMHC ligands are relatively poor stimulators of T cells while divalent ligands are much more efficient stimulators (Abastado et al., 1995; Delon et al., 1998). The mechanism by which TCRs are cross-linked under natural circumstances of pMHC recognition is unclear; however, both crystallography and solution analyses of soluble pMHC/TCR complexes suggest an intrinsic capability of MHC molecules to oligomerize TCR under certain conditions. More recent evidence indicates that, in addition to TCR cross-linking, individual pMHC ligands on APCs can serially engage multiple TCRs, thereby allowing a relatively small number of the appropriate pMHC ligands on an APC to engage a large proportion of TCRs on cognate T cells (Valitutti et al., 1995a; Valitutti et al., 1995b). Serial TCR engagement presumably allows for the accumulation of signal towards a critical threshold for T cell activation.
This
threshold has been estimated to require serial engagement of 8000 TCRs in the
27
absence of costimulatory signals and roughly 1500 TCRs in the presence of costimulatory signals (Viola and Lanzavecchia, 1996).
2.2. MHC multimers and peptide loading The low binding affinity of TCR for peptide-MHC complex, which is typically 10-4~10-5M, orders of magnitude lower than typical antibody-antigen interactions (Matsui et al., 1991), was an obstacle for detecting antigen-specific T cells using soluble peptide-MHC complex monomers. To overcome this low affinity, pMHC monomers have been engineered to form soluble multimers that have much increased apparent binding affinity, i.e. avidity, to TCR compared with monomeric pMHC. There are two broad approaches for engineering multimeric peptide-MHC molecules (Greten and Schneck, 2002). One is tetrameric MHC molecules, in which biotinylated soluble peptide-MHC complexes are conjugated to streptavidin at a molar ratio of 4:1 (Altman et al., 1996). The other is dimeric MHC molecules using mouse immunoglobulin as a molecular scaffold, in which the extracellular domains of two MHC molecules are fused to the constant region of an immunoglobulin heavy chain (Figure 2.1, left) (Dalporto et al., 1993).
MHC-Ig dimer and anti-CD28 antibody have been chemically conjugated onto cell-sized magnetic beads to form artificial antigen-presenting cells (aAPCs) (Figure
28
2.1, right), which can selectively induce and expand antigen-specific T cells for adoptive T cell transfer (Durai et al., 2009; Oelke et al., 2003). In addition, MHC-Ig has been co-immobilized with apoptosis-inducing anti-Fas ligand antibody to form killer aAPCs, which can eliminate unwanted T cells in an antigen-specific fashion for the treatment of autoimmune diseases and allograft rejection (Schutz et al., 2008).
Though MHC tetramer appears to possess higher avidity than dimer, dimer technology has distinct advantages over tetramer technology. In tetramer technology, the genetically modified MHC heavy chain is expressed in Escherichia coli and must be refolded in the presence of β2m and MHC-restricted peptide of interest to obtain native peptide-MHC complexes before coupling to streptavidin. Owing to these serial steps of refolding, biotinylation, and coupling with fluorescence-labeled streptavidin, the production of a specific antigenic tetramer has been a laborious process. Recent development of UV light-sensitive conditional peptide, which complexes with HLA during HLA refolding but degrades upon controlled photo-stimulus, has shown great promise in facilitating peptide exchange for soluble HLA (Bakker et al., 2008; Toebes et al., 2006).
On the contrary, MHC-Ig fusion proteins are expressed in eukaryotic cells with intact conformations. When secreted, these MHC-Ig molecules are associated with a pool of undefined endogenously–processed peptides, thus not showing peptide specificity and
29
denoted as “unloaded” dimer. There have been, in general, passive and active loading methods for obtaining complexes of MHC-Ig with specific peptides of interest, based on whether mild MHC denaturation is involved. Passive loading involves the passive peptide exchange facilitated by incubating MHC-Ig with excess peptide in solution. In active loading, stripping of endogenous peptides has been developed for Kb-Ig dimer and Ld-Ig dimer via alkaline and mild acidic treatments respectively (Schneck et al., 2005). Thermal destabilization of soluble HLA-A2 monomer (Buchli et al., 2004) can be adapted and optimized for the peptide exchange of HLA A2-Ig (unpublished data). In order for the UV-sensitive conditional peptide technology to be applied to MHC-Ig dimers, a pre-loading step with this conditional peptide is required. Among all these methods, passive loading remains to be the most widely utilized peptide loading strategy, because of its convenience in operation and its generality for all the currently available MHC-Ig complexes (including HLA A2-Ig, H-2Db-Ig, H-2Kb-Ig and H-2Ld-Ig).
2.3. Technologies for studying T-cell antigen specificities Current methods for characterizing T-cell antigen specificities broadly fall into two categories: indirect assays and direct assays. Indirect assays, including 51Cr-labeled target-cell lysis (Gianfrani et al., 2000), enzyme-linked immunosorbent spot
30
(ELISpot) (Assarsson et al., 2008; Scheibenbogen et al., 2000) assay, and intracellular cytokine staining by flow-activated cell sorting (FACS, also called flow cytometry) (Kern et al., 1998), detect the presence of antigen-specific T cells by measuring CTL effector functions, such as cytotoxic activity or cytokine secretion following peptide antigen stimulation. In target-cell lysis assays, three to four rounds of consecutive weekly in vitro stimulations are typically required to obtain sufficient expansion of peptide-specific CTLs for measurable 51Cr release, which can consume large quantities of specific peptide and/or APCs. In addition, these assays do not permit quantitative assessments of the frequency of antigen-specific T cells; the throughput is also low, and the sensitivity is approximately 1/1000 total cells (Romero et al., 1998). In ELISpot assays, stimulated T cells are indirectly detected at the level of single cells by binding specific secreted cytokines to antibodies immobilized on 96-well plates; via enumerating spots created by these secreted cytokines, the number of cytokine-secreting cells in the overall cell population can be obtained. ELISpot assays do not require in vitro cellular proliferation as required for target-cell lysis assays, and are roughly 100 times more sensitive: 1/100,000 total cells (Romero et al., 1998). However, this assay can take as long as five days to complete and also requires large quantities of peptide, as well as antibody reagents. Detection of intracellular cytokines after T-cell in vitro stimulation with peptides by FACS has been used to detect individual cytokine-secreting cells (Kern et al., 1998). Compared with 51Cr-release
31
assay and ELISpot assay, this assay takes a much shorter time, less than one day, to complete.
With the development of MHC tetramers (Altman et al., 1996) and dimers (Greten et al., 1998), FACS staining with fluorescence-labeled MHC tetramers or alternatively with MHC-Ig dimers followed by fluorescence-labeled secondary anti-Ig antibody that binds the dimers, has become the standard assay for the direct enumeration of antigen-specific T cells. FACS detection of antigen-specific T cells has sensitivity of 1/10,000 total cells, which is higher than target cell lysis assays, but not as high as ELISpot assays (He et al., 1999; Walker and Disis, 2003), and requires minimal time for the assay. More recent advances incorporating combinations of fluorophores and/or quantum dots (QDs) into pMHC tetramers have enabled the simultaneous screening of a single cell population for up to 25 antigen specificities using FACS (Hadrup et al., 2009; Newell et al., 2009). The number of antigen specificities that can be measured in a single assay using this approach is limited by the availability of fluorophores and QDs. The initial preparation of labeled tetramers, however, can be labor-intensive and time consuming. For example, the combinatorial strategy in (Hadrup et al., 2009) requires two solutions of each pMHC tetramer labeled with different fluorophores; thus, a total of 50 labeled tetramer complexes must be prepared in order to simultaneously screen for 25 antigen specificities. The incorporation of UV light-sensitive peptides (Toebes et al., 2006) to expedite peptide
32
exchange improves the efficiency of preparation for large libraries of labeled pMHC tetramers.
2.4. Current MHC cellular microarray technologies Since each T cell is defined by a unique TCR, an array of immobilized pMHC multimer complexes that differ in the amino acid sequence of the peptide thus can serve to distinguish T cells by their antigen-specific TCRs. As introduced in Chapter 1 (section 1.2.1), pMHC tetramer/dimer cellular microarray (Chen et al., 2005; Deviren et al., 2007; Kwong et al., 2009; Soen et al., 2003; Stone et al., 2005) has shown great promise as a fast, cost-effective, and high-throughput technology for characterizing T-cell antigen specificities and cytokine secretion for heterogeneous T-cell populations.
The first MHC cellular microarray study was conducted by Davis group, in which both dimers and tetramers of MHC class I molecules as well as tetramers of MHC class II were directly printed with 2 %(v/v) glycerol as the print additive (Soen et al., 2003). Antigen-specific T cells were shown to be specifically captured onto spots printed with cognate peptide-MHC multimers. In a later study, the functionality of microarrays were extended to monitor the cytokine secretion of captured antigen-specific CTLs upon activation, by co-printing peptide-loaded multimeric 33
MHC with anti-cytokine antibodies, so that individual disease-specific cytokine secretion profiles may be established (Chen et al., 2005). The sensitivity, equivalently the lower detection limit of the microarray assays was determined to be 0.01-0.1%. Unfortunately, the commercial substrate used in both studies has discontinued. Using several different printing techniques including hand-spotting, contact manual printing and noncontact automatic printing, stern group fabricated the pMHC tetramer microarray for T cell cytokine detection by co-printing the tetramers with anti-cytokine antibodies for cytokine capture and costimulatory or adhesion molecules for stabilizing or amplifying cytokine secretion (Stone et al., 2005). It is noted that the substrate in that study was just plastic or treated glass and molecules were immobilized by simple adsorption. The sensitivity of the microarray was determined to be ~ 0.1%. Our group developed a Kb-Ig dimer microarray based on Schott Nexterion® Slide H, a polymer-based hydrogel with amine-reactive N-hydroxysuccinimide (NHS)-ester groups for protein immobilization (Deviren et al., 2007). The developed microarray could detect the presence of antigen-specific mouse CTLs at a sensitivity of 0.05%; with an anti-IgG pre-coating, the sensitivity was further improved to 0.01%. Most recently, a single-stranded DNA (ssDNA)-based pMHC tetramer microarray was developed, in which tetramers of distinct specificities were assembled onto different spot locations via the sequence-specific hybridization between the unique ssDNA on each tetramer and its complementary ssDNA printed
34
on the spots (Kwong et al., 2009). The sensitivity of the microarray was determined to be 0.1%.
Despite all previous studies, the development of MHC cellular microarrays is still in infancy that important issues need to be solved before this technology can be widely employed. First, the sensitivity of currently developed microarrays for studying human CTLs is about 0.1%, lower than FACS or ELISpot assay. One factor associated with determining sensitivity is the level of nonspecific cell binding, which was not specifically discussed in previous studies except (Kwong et al., 2009); however, the use of Jurkat cells as background cells for diluting antigen-specific cells in that study could have substantially reduced nonspecific binding, since Jurkat cells are much more homogeneous than the cell populations in normal blood. Second, although the approach of detecting secreted cytokines appears to give a higher sensitivity, caution must be taken in interpreting the antigenic repertoire from the cytokine response. Since only a few cells could be producing cytokines upon capture, the repertoire of antigen-specific cells might be significantly different than that seen by functional assays, such as cytokine release. Third, all previous studies focused on only the detection of antigen-specific CTLs. For this technology to be highly-competing with FACS, quantitative aspects of the technology, such as estimation of the frequency of antigen-specific CTLs present in sample solutions, need to be developed.
35
Figure 2.1 Schematic diagram showing MHC-Ig fusion protein (left) and artificial antigen presenting cell (aAPC). Adapted from (Oelke and Schneck, 2004).
References Abastado, J.P., Y.C. Lone, A. Casrouge, G. Boulot, andP. Kourilsky. 1995. Dimerization of Soluble Major Histocompatibility Complex Peptide Complexes Is Sufficient for Activation of T-Cell Hybridoma and Induction of Unresponsiveness. Journal of Experimental Medicine 182(2):439-447. Altman, J.D., P.A.H. Moss, P.J.R. Goulder, D.H. Barouch, M.G. McHeyzerWilliams, J.I. Bell, A.J. McMichael, andM.M. Davis. 1996. Phenotypic analysis of antigen-specific T lymphocytes. Science 274(5284):94-96.
36
Assarsson, E., H.H. Bui, J. Sidney, Q. Zhang, J. Glenn, C. Oseroff, I.N. Mbawuike, J. Alexander, M.J. Newman, H. Grey, andA. Sette. 2008. Immunomic Analysis of the Repertoire of T-Cell Specificities for Influenza A Virus in Humans. Journal of Virology 82(24):12241-12251. Bakker, A.H., R. Hoppes, C. Linnemann, M. Toebes, B. Rodenko, C.R. Berkers, S.R. Hadrup, W.J.E. van Esch, M.H.M. Heemskerk, H. Ovaa, andT.N.M. Schumacher. 2008. Conditional MHC class I ligands and peptide exchange technology for the human MHC gene products HLA-A1, -A3, -A11 and -B7. Proceedings of the National Academy of Sciences of the United States of America 105(10):3825-3830. Buchli, R., R.S. VanGundy, H.D. Hickman-Miller, C.F. Giberson, W. Bardet, andW.H. Hildebrand. 2004. Real-time measurement of in vitro peptide binding to soluble HLA-A*0201 by fluorescence polarization. Biochemistry 43(46):14852-14863. Callan, M.F.C., L. Tan, N. Annels, G.S. Ogg, J.D.K. Wilson, C.A. O'Callaghan, N. Steven, A.J. McMichael, andA.B. Rickinson. 1998. Direct visualization of antigen-specific CD8(+) T cells during the primary immune response to Epstein-Barr virus in vivo. Journal of Experimental Medicine 187(9):1395-1402. Charles A. Laneway, J., P. Travers, M. Walport, andM.J. Shlomchik. 2005. Immunobiology-the immune system in health and disease. Garland Science. Chen, D.S., Y. Soen, T.B. Stuge, P.P. Lee, J.S. Weber, P.O. Brown, andM.M. Davis. 2005. Marked differences in human melanoma antigen-specific T cell responsiveness after vaccination using a functional microarray. Plos medicine 2(10):1018-1030. Chen, W.S., L.C. Anton, J.R. Bennink, andJ.W. Yewdell. 2000. Dissecting the multifactorial causes of immunodominance in class I-restricted T cell responses to viruses. Immunity 12(1):83-93. Corr, M., A.E. Slanetz, L.F. Boyd, M.T. Jelonek, S. Khilko, B.K. Alramadi, Y.S. Kim, S.E. Maher, A.L.M. Bothwell, andD.H. Margulies. 1994. T-Cell
37
Receptor-Mhc Class-I Peptide Interactions - Affinity, Kinetics, and Specificity. Science 265(5174):946-949. Dalporto, J., T.E. Johansen, B. Catipovic, D.J. Parfiit, D. Tuveson, U. Gether, S. Kozlowski, D.T. Fearon, andJ.P. Schneck. 1993. A Soluble Divalent Class-I Major Histocompatibility Complex Molecule Inhibits Alloreactive T-Cells at Nanomolar Concentrations. Proceedings of the National Academy of Sciences of the United States of America 90(14):6671-6675. Davenport, M.P., C. Fazou, A.J. McMichael, andM.F.C. Callan. 2002. Clonal selection, clonal senescence, and clonal succession: The evolution of the T cell response to infection with a persistent virus. Journal of Immunology 168(7):3309-3317. Davis, M.M., andY.H. Chien. 1993. Topology and Affinity of T-Cell Receptor Mediated Recognition of Peptide Mhc Complexes. Current Opinion in Immunology 5(1):45-49. Delon, J., C. Gregoire, B. Malissen, S. Darche, F. Lemaitre, P. Kourilsky, J.P. Abastado, andA. Trautmann. 1998. CD8 expression allows T cell signaling by monomeric peptide-MHC complexes. Immunity 9(4):467-473. Deviren, G., K. Gupta, M.E. Paulaitis, andJ.P. Schneck. 2007. Detection of antigen-specific T cells on p/MHC microarrays journal of molecular recognition 20(1):32-38. Durai, M., C. Krueger, Z.H. Ye, L.Z. Cheng, A. Mackensen, M. Oelke, andJ. Schneck. 2009. In vivo functional efficacy of tumor-specific T cells expanded using HLA-Ig based artificial antigen presenting cells (aAPC). Cancer Immunol. Immunother. 58(2):209-220. Gianfrani, C., C. Oseroff, J. Sidney, R.W. Chesnut, andA. Sette. 2000. Human memory CTL response specific for influenza A virus is broad and multispecific. Human Immunology 61(5):438-452. Greten, T.F., andJ.P. Schneck. 2002. Development and use of multimeric major histocompatibility complex molecules. Clinical and Diagnostic Laboratory Immunology 9(2):216-220. 38
Greten, T.F., J.E. Slansky, R. Kubota, S.S. Soldan, E.M. Jaffee, T.P. Leist, D.M. Pardoll, S. Jacobson, andJ.P. Schneck. 1998. Direct visualization of antigen-specific T cells: HTLV-1 Tax11-19-specific CD8(+) T cells are activated in peripheral blood and accumulate in cerebrospinal fluid from HAM/TSP patients. Proceedings of the National Academy of Sciences of the United States of America 95(13):7568-7573. Hadrup, S.R., A.H. Bakker, C.Y.J. Shu, R.S. Andersen, J. van Veluw, P. Hombrink, E. Castermans, P.T. Straten, C. Blank, J.B. Haanen, M.H. Heemskerk, andT.N. Schumacher. 2009. Parallel detection of antigen-specific T-cell responses by multidimensional encoding of MHC multimers. Nature Methods 6(7):520-U579. He, X.S., B. Rehermann, F.X. Lopez-Labrador, J. Boisvert, R. Cheung, J. Mumm, H. Wedemeyer, M. Berenguer, T.L. Wright, M.M. Davis, andH.B. Greenberg. 1999. Quantitative analysis of hepatitis C virus-specific CD8(+) T cells in peripheral blood and liver using peptide-MHC tetramers. Proceedings of the National Academy of Sciences of the United States of America 96(10):5692-5697. Hill, A.B., S.P. Lee, J.S. Haurum, N. Murray, Q.Y. Yao, M. Rowe, N. Signoret, A.B. Rickinson, andA.J. Mcmichael. 1995. Class-I Major Histocompatibility Complex-Restricted Cytotoxic T-Lymphocytes Specific for Epstein-Barr-Virus (Ebv) Nuclear Antigens Fail to Lyse the Ebv-Transformed B-Lymphoblastoid Cell-Lines against Which They Were Raised. Journal of Experimental Medicine 181(6):2221-2228. Kern, F., I.P. Surel, C. Brock, B. Freistedt, H. Radtke, A. Scheffold, R. Blasczyk, P. Reinke, J. Schneider-Mergener, A. Radbruch, P. Walden, andH.D. Volk. 1998. T-cell epitope mapping by flow cytometry. Nature Medicine 4(8):975-978. Kwong, G.A., C.G. Radu, K. Hwang, C.J.Y. Shu, C. Ma, R.C. Koya, B. Comin-Anduix, S.R. Hadrup, R.C. Bailey, O.N. Witte, T.N. Schumacher, A. Ribas, andJ.R. Heath. 2009. Modular Nucleic Acid Assembled p/MHC Microarrays for Multiplexed Sorting of Antigen-Specific T Cells. J. Am. Chem. Soc. 131(28):9695-9703.
39
Matsui, K., J.J. Boniface, P.A. Reay, H. Schild, B. Fazekas De St Groth, andM.M. Davis. 1991. Low Affinity Interaction of Peptide-Mhc Complexes with T-Cell Receptors. Science 254(5039):1788-1791. Newell, E.W., L.O. Klein, W. Yu, andM.M. Davis. 2009. Simultaneous detection of many T-cell specificities using combinatorial tetramer staining. Nature Methods 6(7):497-U438. Oelke, M., M.V. Maus, D. Didiano, C.H. June, A. Mackensen, andJ.P. Schneck. 2003. Ex vivo induction and expansion of antigen-specific cytotoxic T cells by HLA-Ig-coated artificial antigen-presenting cells. Nature Medicine 9(5):619-624. Oelke, M., andJ.P. Schneck. 2004. HLA-Ig-based artificial antigen-presenting cells: setting the terms of engagement. Clinical Immunology 110(3):243-251. Romero, P., J.C. Cerottini, andG.A. Waanders. 1998. Novel methods to monitor antigen-specific cytotoxic T-cell responses in cancer immunotherapy. Mol. Med. Today 4(7):305-312. Scheibenbogen, C., P. Romero, L. Rivoltini, W. Herr, A. Schmittel, J.C. Cerottini, T. Woelfel, A.M.M. Eggermont, andU. Keilholz. 2000. Quantitation of antigen-reactive T cells in peripheral blood by IFN gamma-ELISPOT assay and chromium-release assay: a four-centre comparative trial. Journal of Immunological Methods 244(1-2):81-89. Schneck, J.P., J.E. Slansky, S.M. O'Herrin, andT.F. Greten. 2005. Monitoring antigen-specific T cells using MHC-Ig dimers. In Short protocols in immunology: a compendium of methods from current protocols in immunology. Coligan JE, Bierer BE, Margulies DH, Sherach EM, Strober N, editors. Wiley. 15-12 - 15-11. Schutz, C., M. Fleck, A. Mackensen, A. Zoso, D. Halbritter, J.P. Schneck, andM. Oelke. 2008. Killer artificial antigen-presenting cells: a novel strategy to delete specific T cells. Blood 111(7):3546-3552.
40
Soen, Y., D.S. Chen, D.L. Kraft, M.M. Davis, andP.O. Brown. 2003. Detection and characterization of cellular immune responses using peptide-MHC microarrays. Plos biology 1(3):429-438. Stone, J.D., J. Walter E. Demkowicz, andL.J. Stern. 2005. HLA-restricted epitope identification and detection of functional T cell responses by using MHC-peptide and costimualtory microarrays PNAS 102:3744-3749. Tan, L.C., N. Gudgeon, N.E. Annels, P. Hansasuta, C.A. O'Callaghan, S. Rowland-Jones, A.J. McMichael, A.B. Rickinson, andM.F.C. Callan. 1999. A re-evaluation of the frequency of CD8(+) T cells specific for EBV in healthy virus carriers. Journal of Immunology 162(3):1827-1835. Toebes, M., M. Coccoris, A. Bins, B. Rodenko, R. Gomez, N.J. Nieuwkoop, W. van de Kasteele, G.F. Rimmelzwaan, J.B.A.G. Haanen, H. Ovaa, andT.N.M. Schumacher. 2006. Design and use of conditional MHC class I ligands. Nature Medicine 12(2):246-251. Valitutti, S., M. Dessing, K. Aktories, H. Gallati, andA. Lanzavecchia. 1995a. Sustained Signaling Leading to T-Cell Activation Results from Prolonged T-Cell Receptor Occupancy - Role of T-Cell Actin Cytoskeleton. Journal of Experimental Medicine 181(2):577-584. Valitutti, S., S. Muller, M. Cella, E. Padovan, andA. Lanzavecchia. 1995b. Serial Triggering of Many T-Cell Receptors by a Few Peptide-Mhc Complexes. Nature 375(6527):148-151. Viola, A., andA. Lanzavecchia. 1996. T cell activation determined by T cell receptor number and tunable thresholds. Science 273(5271):104-106. Walker, E.B., andM.L.N. Disis. 2003. Monitoring immune responses in cancer patients receiving tumor vaccines. International Reviews of Immunology 22(3-4):283-319.
41
CHAPTER 3 AN HLA A2-IG BASED CELLULAR MICROARRAY METHOD FOR DETECTION OF HUMAN ANTIGEN-SPECIFIC T CELLS
Abstract We present a novel cellular microarray assay in which peptide-loaded HLA A2-Ig dimer complexes are not printed on the microarray, but incubated with cells in solution, and an antibody specific to the Ig portion of the dimer complex is printed. T cells are captured on the microarray based on their antigen specificities with the peptide-HLA A2-Ig complexes, by binding to anti-Ig antibodies on the microarray. By printing the inherently robust and stable antibody, this approach preserves the structural integrity of the labile HLA A2-Ig complex, and optimizes the flexibility of the complex to interact with T cells in solution by binding to their T-cell receptors. The specificity of cell capture, the lower detection limit, and the quantitative
42
characterization of antigen specificities in T-cell populations are determined using antigen-specific CD8 T cells stimulated in vitro, and CD8 T cells isolated from the peripheral blood of A2-positive human donors. The results demonstrate that this microarray assay has several advantages for quantitatively detecting the frequency of antigen-specific T cells in heterogeneous cell populations compared to microarray assays in which peptide-loaded major histocompatibility complexes are directly printed on the substrate.
3.1. Introduction Studying T-cell antigen specificities by identifying T cell epitopes is fundamentally important to disease control and therapies, epitope-based vaccine design (Olsen et al., 2000), and the mechanistic elucidation of immunodominance (Yewdell and Bennink, 1999). The tremendous diversity of T-cell antigen specificities, which is on the order of 107 T-cell clones in the human blood (Arstila et al., 1999), and the limitations on T-cell sample size that can be obtained clinically require the development of methods for identifying antigen specificities that are sensitive to detecting only those T cells specific for a given peptide-MHC (pMHC) complex in populations consisting largely of irrelevant cells, and high throughput to enable analysis of multiple specificities in parallel for a single sample.
43
MHC cellular microarray technology has shown promise as a fast, cost-effective, and high throughput platform for characterizing T-cell antigen specificities and cytokine secretion for heterogeneous T-cell populations (Chen et al., 2005; Deviren et al., 2007; Soen et al., 2003; Stone et al., 2005). In protein microarray assays, preserving the native protein structure and accessibility of binding sites on the protein are critical for performance. Many proteins are prone to unfold and breakdown under chemical, physical or mechanical stresses associated with printing and surface immobilization (Nishioka et al., 2004). Elements that can influence protein-surface interactions leading to unfolding and instability include surface properties of the substrate, immobilization chemistry, and printing method (Ajikumar et al., 2007; Delehanty and Ligler, 2003; Liu et al., 2007; Wu and Grainger, 2006; Zhu and Snyder, 2003). These challenges are particularly formidable for pMHC tetramer/dimer microarrays due to the intrinsically labile, multidomain pMHC structure. The relatively weak binding affinity that is characteristic of pMHC-TCR interactions in general (Davis et al., 1998) also places stringent demands on the accessibility of the complementary pMHC binding sites upon immobilization. These challenges have motivated the development of new pMHC-based cellular microarray assays. For example, a novel pMHC tetramer microarray assay has been developed (Kwong et al., 2009), in which pMHC tetramers are immobilized in a patterned microarray by printing single-stranded DNA (ssDNA). This approach circumvents directly printing the pMHC tetramers, but requires labeling pMHC tetramers of distinct antigen specificities with the
44
complementary sequences of ssDNA oligomers. This development offers a strategy for pMHC-based cellular microarray assays that has the potential to outperform conventional pMHC cellular microarray assays.
Here, we extend this idea further by incubating pMHC-Ig dimers with the T cells in solution, and subsequently capturing the T cells on the microarray by binding to an antibody specific to the immunoglobulin (Ig) portion of the dimer printed on the microarray, as illustrated in Figure 3.1. T-cell antigen specificities are distinguished by the specific peptides presented by the pMHC-Ig dimers bound to the cell surface, and the frequency of antigen-specific T cells in the population is determined by the number of cells captured on the microarray. The approach takes advantage of the high stability of immunoglobulins by printing the anti-Ig antibody instead of the labile pMHC-Ig dimer complex, and more importantly, optimizes the avidity of pMHC binding to the TCRs by preserving the flexibility of the pMHC-Ig dimer complex in solution. Much higher concentrations of soluble dimer can be obtained in solution relative to surface-bound dimers, which also enhances antigen-specific binding to the TCRs. The ability to form microclusters of TCR-bound pMHC-Ig dimers on the T-cell surface may also contribute to the efficiency of capturing T cells on the microarray surface.
45
3.2. Materials and Methods Anti-Ig,λ based HLA A2-Ig cellular microarray assays. Anti-mouse Ig,λ antibody microarrays were fabricated by printing a solution of 0.5 mg/ml polyclonal anti-mouse Ig,λ antibody (Southern Biotech, AL), and separately, a solution of 0.5 mg/ml anti-human CD3 antibody (BD Biosciences, San Diego CA) on polyacrylamide film-coated glass slides - Nexterion Slide H (Schott, Elmsford NY) using a PerkinElmer Piezorray noncontact microarrayer (PerkinElmer, Waltham MA). Bovine serum albumin (BSA, Sigma Aldrich, St Louis, MO) at 0.5 mg/ml (final concentration), which was shown to maintain regular spot morphology, produce uniform intra-spot antibody distribution, and minimize antibody loss during printing (SP. Figure 3.1), was added to the printing solutions prior to printing. Cell binding to anti-CD3 spots serves to calibrate antigen-specific cell capture on the anti-Ig,λ spots as a frequency of the total T cells in the solution. The solutions were typically printed in pre-programmed patterns of 5 x 5 subarrays of spots separated by a spot center-to-center distance of 1 mm. The individual spots consisted of about 10 nanoliters of solution, and were 500 µm in diameter. For the experiments evaluating microarray sensitivity, 10 x 10 spots were printed per subarray to facilitate statistical distribution analysis of spot intensities. Spot center-to-center distance was 400 µm and individual spots only consisted of ~ 0.33 nanoliters of solution and were 150 µm in diameter. The printed slides can be stored sealed at 4ºC for more than a month before use without measurable loss of activity. Prior to an experiment, microarray 46
slides were incubated in a humid chamber (relative humidity of 75%) for 1 hr at room temperature (24ºC) to facilitate the immobilization of arrayed proteins within the gel substrate. After wash with 0.05% (v/v) Tween 20 in 1x PBS solution (PBST) followed by PBS solution, microarray slides were blocked in blocking buffer (50 mM sodium borate containing 50 mM ethanolamine at pH 8.0) at room temperature for 1 hr. The slides were then washed again with 0.05% (v/v) PBST and PBS solutions.
Cells were first labeled with CFSE (carboxy-fluorescein diacetate, succinimidyl ester; Invitrogen, Carlsbad CA) according to manufacturer’s protocol, and then incubated with unloaded or peptide-loaded dimer at 4ºC for 45 min. For every 1 x105 cells, 1 µl of 0.5 mg/ml dimer was used. The cells were then centrifuged with 300 µl PBS solution at 700 rcf for 5 min to remove unbound dimer, and resuspended in PBS solution to a final concentration of ~1-2 x106 cells /ml. The cells were then contacted with the microarray surface by pipetting 50 µl of this suspension onto an area of 6.8 mm x 6.8 mm (to cover a 5 x 5 subarray of 500 µm spots or 10 x 10 subarray of 150 µm spots) separated by a silicone gasket (Grace Bio Labs, Bend OR).
For microarray sensitivity experiments, cell suspensions containing known frequencies of antigen-specific CTLs ranging from 10% to 0%, obtained by serially mixing CFSE-labeled aAPC-enriched antigen-specific CTLs with unlabeled CD8
47
cell-depleted autologous PBMCs, were incubated with peptide-loaded dimer molecules and applied to anti-Ig,λ microarrays. The same frequencies of cells but pre-incubated with unloaded dimer were included as controls. The frequency of undiluted aAPC-enriched antigen-specific CTLs was kept at a moderate level of 10% before fluorescence labeling and mixing with unlabeled PBMCs, so that ultimately the CFSE-labeled cells in each cell suspension contained not only antigen-specific CTLs but also a pool of CD8 cells expressing TCRs of diverse specificities. After 1 hr incubation in the dark at room temperature, the microarray slide was then dip-washed in PBS solution to remove any unbound cells prior to scanning.
Conventional spotted HLA A2-Ig cellular microarray assays. Conventional spotted dimer microarrays were fabricated by printing solutions of 0.5 mg/ml peptide-loaded dimer molecules and unloaded dimer molecules (negative control) onto Nexterion Slide H, in 5 x 5 subarrays with 1 mm spot center-to-center distance. Individual spots consisted of about 10 nanoliters of solution and were 500 µm in diameter. In some experiments, slides were incubated with a solution of goat anti-mouse IgG antibody (BD Biosciences, 100 µl of 0.5 mg/ml and spread by a 24 mm x 60 mm glass coverslip) for 1 hr at room temperature prior to printing, to improve the accessibility of peptide-dimer molecules on the substrate surface (Deviren et al., 2007). After protein immobilization and slide blocking (with blocking buffer), antigen-specific CTLs, induced and expanded by aAPC technology (Oelke et al., 2003), were first 48
labeled with CFSE and then contacted with the microarray surface at a concentration of ~ 1-2 x 106/ml. After incubation (for 1 hr at room temperature in the dark), the microarray slide was dip-washed in PBS solution before scanning.
Cell-free microarray assays. To assess the conformations of spotted dimer molecules, a cell-free antibody binding assay was conducted employing two fluorescence-labeled antibodies: PE-anti-HLA (clone W6/32, Dako North America, Carpinteria, CA) and FITC-anti-Ig,λ (Southern Biotech). W6/32 is a conformation-sensitive monoclonal antibody that reacts with the alpha-2 and alpha-3 domains of HLA class I molecules (Parham et al., 1979; Tanabe et al., 1992); studies have found that W6/32 binding to HLA can block or decrease TCR activation or proliferation (Genestier et al., 1997; Ware et al., 1995). By incubating W6/32 and anti-Ig,λ solutions onto printed dimer spots and measuring antibody binding, structural integrity of dimer molecules on the spots can be inferred. Briefly, spots were printed with 0.1 mg/ml unloaded HLA A2-Ig dimer molecules that were in just PBS solution, in 0.33% DMSO/PBS solution, or loaded with 66.7µg/ml CMVpp65 in 0.33% DMSO/PBS solution. Alternatively, spots were printed with 0.43mg/ml unloaded dimer that was in just PBS solution or loaded with 276µg/ml CMVpp65 in 1.4% DMSO/PBS solution. Each sample was printed into a subarray containing 5 x 5 spot replicates. Individual spots consist of 0.33 nanoliter of solution and are 150µm in diameter. After protein immobilization and slide blocking, printed microarray was incubated with PE-W6/32 or FITC-anti-Ig,λ, 49
both at a concentration of 20µg/ml, for 1 hr at room temperature in the dark. Slides were then washed with 0.05% PBST followed by PBS solution before scanning.
Flow cytometry measurements. To measure the frequency of antigen-specific CTLs in aAPC-enriched populations or in donor CD8 cells, flow cytometry (FACS) measurements using dimer staining were performed with a FACSCalibur flow cytometer (BD Biosciences, San Jose CA). The procedure for dimer staining is described in (Greten et al., 1998) in detail. Briefly, ~150,000 cells in 100 µl of PBS solution were first incubated with 1 µl of unloaded or peptide-loaded dimer at 4ºC for 45 min. After wash with 500 µl of PBS solution at 700 rcf for 5 min, the cells were incubated with 2 µl of 0.1 mg/ml PE-conjugated goat anti-mouse IgG1 antibody (Invitrogen, Carlsbad CA) at 4ºC for 10 min. After a second wash with 500 µl of the same buffer solution, the cells were incubated with 2 µl of 0.1 mg/ml FITC-conjugated anti-human CD8 monoclonal antibody (Sigma Aldrich, St Louis MO) at 4ºC for 10 min. The cells were then washed a third time with 500 µl of the buffer solution, and fixed in 300 µl of 4% (w/v) paraformaldehyde before FACS analysis. FACS data reduction was carried out using FlowJo software (TreeStar, San Carlos CA).
50
Image acquisition and data analysis. Microarray images were acquired using a ProScanArray fluorescence scanner (PerkinElmer, Waltham MA). The slides were scanned at 5 µm resolution. Image quantification and data analysis were performed using ScanArray Express 3.0 (PerkinElmer, Waltham MA) and JMP 6.0 (SAS, Cary NC) software. Bar graph plotting and linear data fitting were performed using Sigma plot 10.0 (Systat Software, San Jose CA).
Preparation of peptide-HLA A2-Ig complexes and cells. High purity (~95%) peptides CMVpp65 (NLVPMVATV), M1.58 (GILGFVFTL), PA.46 (FMYSDFHFI), and NA.75 (SLCPIRGWAI) were purchased from Genscript (Piscataway NJ). The peptides were reconstituted to 2 mg/ml in PBS solution containing 10%-20% (v/v) DMSO (10% for CMVpp65, PA.46 and NA.75, and 20% for M1.58). Peptide-dimer complexes were prepared by the method of passively loading (Schneck et al., 2005) with minor changes. Briefly, unloaded HLA A2-Ig dimer (0.5 mg/ml, BD Pharmingen, San Diego CA) was incubated with specific peptide of interest in 160:1 peptide: dimer molar ratio at 4ºC or 24ºC (room temperature) for up to 20 days. No separation was conducted after incubation.
Antigen-specific CTLs were induced and expanded using artificial antigen presenting cells (aAPCs), as described in detail in (Oelke et al., 2003). Briefly, CD8 T
51
lymphocytes were separated from peripheral blood of HLA-A2.1 healthy donors (phenotyped by FITC-labeled anti-HLA-A2.1 antibody using flow cytometry) using a negative CD8 isolation kit (Miltenyi Biotec, Auburn CA). The resulting population, consisting of >90% CD8 T cells (determined by staining with FITC-labeled anti-human CD8 antibody in FACS), were stimulated with peptide-loaded aAPCs in a 96-well plate for 7 to 9 days. The cells were then collected and FACS dimer staining was conducted to measure the frequency of antigen-specific cells; the rest of the cells were frozen in fetal bovine serum (FBS, Atlanta Biologicals, Lawrenceville GA) containing 10% DMSO and stored at -80ºC. Prior to use, the cells were thawed following standard procedures.
CD8-depleted autologous peripheral blood mononuclear cells (PBMCs) used in microarray sensitivity experiments were isolated from perioheral blood using a positive CD8 isolation kit (Miltenyi Biotec, Auburn CA). CD8 cells from donor A8 were isolated from perioheral blood using a negative CD8 isolation kit (Miltenyi Biotec, Auburn CA).
52
3.3. Results 3.3.1. Overview of the anti-Ig,λ antibody-based microarray platform In fabricating protein microarrays using inkjet printing technology, rapid evaporation can occur when printing nanoliter quantities of protein solutions that lead to abrupt changes in the solution conditions, exposing the protein solute to entirely different environments (Wu and Grainger, 2006). To examine the lability of HLA A2-Ig dimer to direct spotting/printing, fluorescence-labeled conformation-sensitive antibodies, anti-HLA (W6/32) and anti-Ig,λ, were employed to incubate with printed dimer spots; the binding of the two antibodies provides insights into the on-spot structural integrity of dimer molecules.
Figure 3.2 compares the binding of the antibodies to dimer molecules that were printed in solutions that differ in small concentrations of dimethyl sulfoxide (DMSO), the organic solvent for dissolving commonly-encountered low solubility peptides in MHC ligands (Buchli et al., 2004; Rodenko et al., 2006). In the presence of DMSO, which was only 0.33%(v/v) in printing solution, the mean intensity of unloaded dimer spots was only 7.6% of those printed in buffer without DMSO, indicating that most of the printed unloaded dimer molecules were not capturing W6/32 molecules. Since the unloaded dimer-DMSO solution was freshly prepared immediately before printing, this deleterious effect of DMSO on W6/32 binding is mainly printing-associated as
53
opposed to taking place in bulk solution. Consistently, spots printed with CMVpp65-loaded dimer solution that has the same DMSO concentration showed comparably low W6/32 binding: only 4.7% of unloaded dimer printed without DMSO (images shown in Figure 3.2 inset). When stained with FITC-labeled anti-Ig,λ antibody, spots printed in the presence of DMSO was about 54% of those printed without DMSO. Thus, at least the HLA alpha-2 and alpha-3 domains and the Ig part of the dimer molecules were not conformationally intact when printed in the presence of minute amount of DMSO. DMSO has been shown to preferentially bind proteins and induce protein unfolding when it exceeds a certain concentration, which varies from 40% to 70% for the four proteins studied: chymotrypsinogen, beta-lactoglobulin, Lysosome and RNase (Arakawa et al., 2007). It is likely that the actual DMSO concentration dimer molecules are exposed to upon printing significantly increases during the few seconds of droplet evaporation, to a level that damages the structural integrity of printed dimer molecules.
Pursuing alternative preparation methods, such as dialysis or liquid chromatography, to remove DMSO or other additives in the buffer solution prior to printing can become time-consuming and expensive, and require several protocols depending on the number and the chemical diversity of the peptides under consideration. This strategy is not realistic for high throughput assays, and can lead to uneven spot-to-spot reproducibility, as well as poor reproducibility from one microarray assay to another. 54
The strategy we adopt here circumvents printing the labile pHLA A2-Ig dimer altogether, but takes advantage of the unique structure of this dimer by printing instead the more robust and stable anti-Ig,λ antibody, which binds to the Ig scaffold of the dimer complex, and enables printing a single solution uniformly on each individual spot across the entire microarray. Moreover, the TCR-pHLA molecular interactions that define T-cell antigen specificities take place in solution, rather than at the cell-microarray substrate interface, which enables further control and optimization of the assay by manipulating the solution conditions.
The ability of the microarray for capturing cells by antigen specificity was first examined, using highly antigen-specific CTLs that were induced and expanded via aAPC technology (Oelke et al., 2003) against three epitopes: CMVpp65 from cytomegalovirus (CMV), and M1.58 and PA.46 from Influenza A virus. Figure 3.3a shows virtually complete segregation of CTLs based on the specific peptide in dimer complex they were incubated with: when cells were incubated with the inducing peptide-dimer complex (as conjugated on aAPCs), uniform cell binding was observed on anti-Ig,λ spots; in contrast, when cells were incubated with unloaded dimer, essentially no binding was observed (images not shown for M1.58-enriched cells and PA.46-enriched cells). Therefore, cells were captured onto spots as a consequence of the specific interaction between peptide-loaded dimer and cognate TCRs expressed on cell surface.
55
3.3.2. Quantifying ability of the cellular microarray assay An important feature of our developed cellular microarray is quantifying the frequency, beyond the detection of antigen-specific CTLs in suspension. Internal calibration spots, printed with anti-CD3 antibody, were included onto the slide for every cell population. Since spot intensity reflects the number of bound cells, normalizing the intensity of anti-Ig,λ spots against the mean intensity of anti-CD3 spots (shown in Figure 3.3a) takes into account possible differences in cell labeling and capture efficiencies, and estimates the frequency of antigen-specific CTLs in total T cell population.
Figure 3.3b shows a comparison of the frequencies of antigen-specific CTLs in aAPC-enriched cell populations measured in microarray and FACS (see SP. Figure 3.2 for FACS plots). Frequency of cells binding to unloaded dimer was subtracted before plotting for both cellular microarray assay and FACS. For CMVpp65-aAPC enriched cell population, FACS measured the frequency of CMVpp65-specific CTLs to be 71.5%, which is within the range of 67.9 ± 4.22% measured in cellular microarray. Similarly, cellular microarray determined the frequency of PA.46-specific CTLs to be 67.0 ± 4.04%, whereas FACS measured the frequency to be 69.5%, again showing the agreement between the two assays. For M1.58-aAPC enriched population, FACS measured the frequency of M1.58-specific CTLs to be 13.2%, very close to the
56
value of 13.3% indicated by the lower error bar in cellular microarray (14.6 ± 1.25%). This excellent agreement of cellular microarray with FACS demonstrates calibrating mean spot intensity against the mean intensity of anti-CD3 spots after cell binding as a simple and accurate way of quantifying the frequency of antigen-specific CTLs in cell suspension. Note that other antibodies, such as anti-human CD8, can also be employed as internal calibration spots, depending on out of what population the frequency of antigen-specific cells is of interest.
3.3.3. Sensitivity of the cellular microarray assay Strictly speaking, peptide antigenicity is just an operational term that is limited by the sensitivity of the utilized assay to detect T cell response to potential epitopes (Yewdell, 2006); thus sensitivity, or equivalently the lower detection limit, has become a critical parameter for assays studying T cell specificity. To evaluate it, binding to CMVpp65-dimer and unloaded dimer was compared for cell suspensions containing CFSE-labeled CMVpp65-specific CTLs at five different frequencies: 10% (undiluted), 1%, 0.1%, 0.01%, and 0% in unlabeled CD8 cell-depleted autologous PBMCs (Figure 3.4). Cell suspensions of 0% contained only CD8 cell-depleted autologous PBMCs that were not CFSE-labeled, so the resulting spot intensities stand for just background intensities due possibly to the fluorescence emission of endogenous fluorophores such as tryptophan of anti-Ig,λ and BSA on the spots. As cell frequency decreased from 10% to 0%, the mean spot intensity of cells incubated with 57
CMVpp65-dimer continuously decreased, indicating that the average number of captured cells on the spots decreased. Meanwhile, the mean spot intensity of cells incubated with unloaded dimer declined as cell frequency decreased from 10% to 1% but both at a level higher than 0%, indicating that there were cells bound to the spots in a nonspecific manner regardless of the peptide specificity in the dimer. Then, at both 0.1% and 0.01%, the mean spot intensity was about the same as 0% (shown as horizontal lines as well as vertical bars in Figure 3.4), indicating that there was no nonspecific binding when cell frequency was at 0.1% or lower. Across the frequencies from 10% to 0.01%, the mean spot intensities of cells incubated with CMVpp65-dimer were always higher than the mean spot intensities of cells incubated with unloaded dimer, and the intensity converged to be the same at 0%, indicating that the microarray was capable of detecting the presence of antigen-specific CTLs at a frequency of 0.01% in a complex cell population.
Note that the mean spot intensity of cells incubated with unloaded dimer at a frequency of 10% was about the same as the intensity of cells incubated with CMVpp65-dimer at 0.1%, and the intensity of cells incubated with unloaded dimer at a lower frequency of 1% was also about the same as the intensity of cells incubated with CMVpp65-dimer at 0.01%, suggesting the level of nonspecific binding was only about 1% of total binding. The observed decrease and disappearance of cell nonspecific binding to unloaded dimer as cell frequency decreased from 10% to 58
0.01% can be due to this likely proportionality between nonspecific binding and total binding, with the latter decreasing as cell frequency decreased. Because of the possible presence of nonspecifically bound cells on spots, in the future PE-labeled peptide-MHC tetramer can be employed to stain the cells bound to the spots to determine whether a captured cell is antigen-specific or not.
In examining microarray sensitivity, distribution of cell binding across the 10 x 10 spot replicates for cell suspension at each frequency was illustrated by an intensity variability chart (Figure 3.5a for M1.58-specific CTLs). The highest intensity among spots incubated with 0% M1.58-specific CTLs, which was about five standard deviations (74.77, arbitrary unit) above the mean intensity (311.12, arbitrary unit) across spot replicates, was chosen as the cutoff between intensities indicating cell binding and indicating no cell binding on spot. When the frequency of antigen-specific CTLs was high, 10%, all the 100 spot replicates had intensities above the cutoff intensity, indicating that all of them had cell(s) captured. Considering together the size of the sample (total cell number in the cell suspension), there were about 10,000 M1.58-specific CTLs in the cell suspension incubated on the subarray; thus, as many as 100 cells could be captured per spot on average. As the frequency decreased to 1%, 43 spots were above the cutoff intensity, indicating as many as 57 spots not having any cells bound. At this frequency, there were 1,000 M1.58-specific CTLs in the cell suspension; if all these cells are captured and equally distributed over 59
those 43 spots, each spot could capture 23 cells on average. As the percentage further fell to 0.1%, 10 spots had intensities above the cutoff intensity; considering that there were 100 M1.58-specific CTLs present in the cell suspension at this frequency, each spot could capture 10 cells on average. At the lowest frequency 0.01%, there were only 2 spots having intensities above the cutoff intensity. Considering there were only 10 M1.58-specific CTLs in the cell suspension, each spot could capture on average as many as 5 cells; however, since the chance for all 10 cells randomly binding to just 2 spots is really small given 100 spots available for binding (the probability is (1/50)10, assuming that each M1.58-specific CTL can bind to any spot with equal probability), it is very likely that the subarray did not capture all of the 10 M1.58-specific CTLs. That the microarray could capture from 2 to 10 M1.58-specific CTLs in 100,000 total cells further indicates the sensitivity of our developed cellular microarray to be within the range of 0.002%-0.01%. This sensitivity range is very close to the theoretical limit of detection, which is only one cell captured across the entire subarray, corresponding to 0.001% for 100,000 total cells. In addition, it indicates a cell capturing efficiency (percentage of captured target cells in total target cells in solution) of at least 20% and could be as high as 100%, by a printed area that is only 3.8% of the total cell seeding area (6.8 mm x 6.8 mm), suggesting that seeded cells are highly mobile on inert unprinted area for moving to printed area for potential capture.
60
The sum of the intensities of spots having cells bound, which corresponds to the total number of cells captured on the entire subarray, linearly decreased as the frequency of antigen-specific CTLs decreased from 10% to 0.01% (R2=0.999) (Figure 3.5b). The linear correlation between the average number of cells bound per spot and the cell frequency was previously observed (Kwong et al., 2009); however, in that study target cells, even at the lowest frequency of 0.1%, were in 7-fold excess (in number) compared with spots. Our finding indicates that the linear correlation still holds when target cells are sparse compared with spots, but in this case spots those have cell(s) bound and not have any cells bound need to be considered separately as we illustrated. For application, this linear relation can function as a standard curve for estimating the frequency of target cells in suspension via interpolation of measured spot intensities.
3.3.4. Profiling of influenza A-associated T cell epitopes Having established the quantifying ability and sensitivity of the microarray, we explored the microarray for quantitatively detecting antigen-specific CTLs present in complex cell populations: donor CD8 cells enriched from peripheral blood by magnetic bead-based separation. Figure 3.6 compares cellular microarray with FACS in the detection and estimation of the frequency of CTLs specific to three Influenza A-associated epitopes, M1.58, NA.75, and PA.46 (Gianfrani et al., 2000). For cellular microarray, spots with cell(s) bound were identified using the variability chart method presented in Figure 3.5a; the frequencies of antigen-specific CTLs were determined 61
by calibrating the intensity sum of these spots against the intensity sum of all anti-CD3 spots.
As shown in the figure, cellular microarray quantified the frequency of cells nonspecifically binding to unloaded dimer to be 0.37%, whereas FACS detected the frequency to be 0.63%. This lower frequency of cell nonspecific binding in cellular microarray can be caused by the requirement of a threshold level of bonds formed between cell surface receptors and substrate-bound ligands for cell adhesion on microarray surface (Dembo and Bell, 1987).
The frequencies of CTLs binding to M1.58- and PA.46-dimer were measured to be 3.5% and 2.5% in cellular microarray assay and 2.1% and 1.3% in FACS dimer staining, respectively. Subtracting the frequency of cells nonspecifically binding to unloaded dimer, the actual frequencies of cells that specifically bound to M1.58 and PA.46 were determined to be 3.1% and 2.1% in cellular microarray and 1.5% and 0.67% in FACS respectively, strongly indicating the presence of CTLs specific to each of these two epitopes and that the frequency of M1.58-specific CTLs was higher than the frequency of PA.46-specific CTLs in the CD8 cell suspension. The overall level of cell binding to NA.75-dimer was determined to be 0.56% in cellular microarray assay and 0.70% in FACS; subtracting unloaded dimer, the actual
62
frequency of specific binding became 0.19% in cellular microarray and 0.07% in FACS respectively, indicating the presence of a low frequency of NA.75-specific CTLs in donor CD8 cells. Notably, the magnitude of the antigen-specific CTL frequency determined in cellular microarray was approximately twice as much as that measured in FACS for both M1.58 and NA.75, and three times as much as the frequency determined in FACS for PA.46. These higher frequencies of antigen-specific CTLs detected by the microarray assay is evidently due to the irreversibility of cell capture on the microarray and the efficiency of antigen-specific CTLs coming in contact with the anti-Ig,λ spots, which leads to an over-representation of antigen-specific CTLs in the population. Taken together, our developed HLA A2-Ig based cellular microarray technology can sensitively detect the presence and readily quantify the frequency of antigen-specific CTLs, and agreed with FACS in detecting M1.58-, PA.46-, and NA.75-specific CTLs in donor peripheral CD8 cells. To our knowledge, this is the first example showing the use of the microarray to detect the presence and estimate the frequencies of CTLs specific to immune subdominant epitopes.
3.4. Discussion This anti-Ig,λ-based HLA-Ig cellular microarray system outperforms previously developed HLA cellular microarrays, including conventional spotted HLA dimer or
63
tetramer microarrays (Chen et al., 2005; Deviren et al., 2007; Soen et al., 2003; Stone et al., 2005) and ssDNA-assisted HLA tetramer microarray (Kwong et al., 2009). First, the sensitivity of our developed assay (0.002-0.01%) is the highest, attributed to the enhanced functionality and accessibility of dimer molecules. Incubated with cells in solution, labile dimer molecules are not exposed to the fast-changing solution conditions upon printing, so the molecular functionality or structural integrity is maximally preserved. In addition, since dimer molecules are not immobilized directly via printing or indirectly via ssDNAs onto microarray surface before contacting with cells, there is no steric restriction at all for pMHC complex to assume appropriate orientations to interact with TCRs on cell surface. Second, pre-incubating cells with cognate peptide-MHC dimer complexes in solution is likely to enhance subsequent adhesion of T cells on anti-Ig,λ spots. Studies on the activation of T cells in contact with laterally mobile monomeric pMHC complexes that were presented in supported planar lipid bilayers showed that TCR-engaged pMHC complexes accumulated to form microclusters at the center of cell-surface contact area (Grakoui et al., 1999). Studies on CD2-madiated T cell adhesion to CD58 ligand in supported planar lipid bilayers showed that a laterally mobile isoform of CD58 enhanced cell adhesion strength markedly compared with a laterally immobile CD58 isoform, indicating that the accumulation of CD58 ligand into cell-bilayer contact area, realized/allowed by its lateral mobility, is required for effective cell adhesion (Chan et al., 1991). Incubating cells with pMHC-Ig molecules in solution allows maximum mobility of the molecules
64
for forming microclusters of TCR-pMHC complexes upon engagement, which could facilitate subsequent T cell adhesion on anti-Ig,λ spots; in contrast, in previously developed microarrays, surface-immobilized MHC dimer or tetramers have much reduced lateral mobility, thus not facilitating optimal cell adhesion. Third, our microarray assay first achieved the quantification of the frequency of detected target antigen-specific T cells, by printing anti-CD3 antibody as internal calibration spots onto the same slide as anti-Ig,λ spots. This enables the assay system to be used for not just detection, but also quantitative characterization of antigen-specific T cells. This methodology of including internal calibration spots is simple and straightforward, and will surely be integrated to other cellular microarrays for quantification purposes. Fourth, printing a single solution containing the anti-Ig antibody, rather than a variety of solutions individually optimized to accommodate different pMHC dimers, ensures high spot-to-spot reproducibility. Lastly, in our assay both anti-Ig,λ antibody and HLA-Ig dimer are bivalent, which together give a total valence of four. This multivalent binding gives higher avidity between microarray substrate and cells. In addition, it can enhance the cross-linking and clustering of TCRs on cell surface whereby inducing more efficient down-stream signal transductions and subsequent behavioral responses. In the future, cytokine secretion following T cell activation can be readily detected by co-printing anti-cytokine antibody with cell-capturing anti-Ig,λ antibody.
65
Compared with the current gold-standard technology FACS, this anti-Ig,λ-based HLA-Ig cellular microarray system possesses the same level of lower detection limit. It significantly reduces the cost of reagents (ng of antibody versus µg in FACS) and the size of cell samples (a third of that in FACS). In addition, for FACS staining with HLA-Ig dimer only one peptide-dimer complex can be examined at a time for a given cell population, whereas our array platform allows 14 combinations of peptide-dimer complexes and cell populations to be screened simultaneously on one single slide (this number can be readily increased by customizing the gaskets to contain smaller-size wells).
This technology offers great indications to binding interaction-based assay systems such as microarrays and microfluidics that involve probe molecules highly sensitive or labile to available fabrication methods. It sets an example that the issue can be circumvented by keeping the labile probe molecule in solution for interacting with target cells or molecules, and immobilizing a more stable, secondary probe molecule that can interact with and capture the primary probe molecule. For instance, a novel pMHC tetramer microarray assay can be readily developed by printing anti-streptavidin antibodies for capturing pMHC tetramers that are first incubated with cells in solution.
66
Taking advantage of the large number of DNA oligomers that can be synthesized, in the future, the throughput of our developed HLA A2-Ig microarray assay can be increased by incorporating ssDNAs as molecular tags to peptide-dimer complexes. Specifically, different ssDNA-pMHC dimer complexes, each conjugated with a distinct ssDNA, can be pooled together for incubation with a single suspension of cells; by contacting the cell suspension with a DNA microarray that is printed with complimentary ssDNAs, cells will be sorted onto different spots essentially based on their TCR specificities. Considering the labile nature of MHC-Ig dimer molecules, a better strategy for coupling ssDNA oligomers to dimer molecules would be to first chemically conjugate ssDNA oligomers to a secondary antibody, such as anti-Ig,λ in this study, that can bind to deliver the ssDNA oligomers to dimer molecules.
67
STEP 1. Incubate peptide-loaded pHLA A2-Ig dimers in solution with the cell population to identify antigenspecific CTL.
STEP 2. Selectively capture cells with surface-bound pHLA A2-Ig dimers by binding to anti-Ig,λ antibodies printed on the microarray.
Figure 3.1 Schematic depiction of the cellular microarray assay in which soluble peptide-loaded HLA A2-Ig dimers are incubated with T cells in solution. The antigen-specific sub-population of cells are identified by the surface-bound pHLA A2-Ig dimers and subsequently captured on the microarray by binding to printed anti-Ig,λ antibodies that recognize the immunoglobulin (Ig) portion of the dimer.
68
Figure 3.2 Effect of peptide-dissolving reagent dimethyl sulfoxide (DMSO) on anti-HLA (clone W6/32) and anti-Ig antibodies binding to spots printed with dimer molecules. PE-labeled W6/32 (20 µg/ml) was incubated with spots printed with 0.1mg/ml unloaded HLA A2-Ig dimer that was: (1) in just PBS solution, (2) in 0.33% DMSO/PBS solution, and (3) loaded with 66.7µg/ml CMVpp65 in 0.33% DMSO/PBS solution. FITC-labeled anti-Ig,λ (20 µg/ml) was incubated with spots printed with 0.43mg/ml unloaded dimer that was: (4) in just PBS solution and (5) loaded with 276 µg/ml CMVpp65 in 1.4% DMSO/PBS solution. For each antibody staining, quantified fluorescence intensities (mean and standard error across spot replicates) were normalized against the mean spot intensity of unloaded dimer in PBS solution for plotting, with the percent means shown on top of the bars. Inset images are the scanned fluorescence microarray images of sample (1) and (3), as directed by arrows.
69
a Unloaded
CMVpp65
M1.58
PA.46
Anti-CD3
Continued
Figure 3.3 Comparison of cellular microarray versus FACS in detection and determination of the frequency of antigen-specific CTLs in aAPC-enriched cell populations. (a) Representative images of cell capture on anti-Ig,λ spots and on anti-CD3 spots. From left to right: CMVpp65-enriched CTLs incubated with unloaded dimer, CMVpp65-enriched CTLs incubated with CMVpp65-(loaded) dimer, M1.58-enriched CTLs incubated with M1.58dimer, PA.46-enriched CTLs incubated with PA.46-dimer, and CMV-enriched CTLs on anti-CD3 spots. (b) Comparison of the frequency of antigen-specific CTLs determined by cellular microarray versus by FACS. For microarray, the mean spot intensity of unloaded dimer was subtracted before the calibration; error bars represent standard errors for 5 x 5 spot replicates. For FACS, the nonspecific binding determined by unloaded dimer was subtracted for each sample. Dimer loading conditions: 15 days at 4ºC for CMVpp65, 11 days at 24ºC for M1.58, and 16 days at 24ºC for PA.46. Each subarray contains 5 x 5 spots, and each individual spot has ~ 10 nl of printing solution and is 500µm in diameter. Spot-to-spot distance is about 1mm. Cell seeding condition for each subarray: 50µl of ~0.8 x 106/ml total cells.
70
Figure 3.3 continued
b
71
Figure 3.4 Quantified spot fluorescence intensity of cell suspensions containing different frequencies of CMVpp65-specific CTLs incubated with CMVpp65-loaded dimer versus unloaded dimer. A CMVpp65-aAPC enriched cell population containing ~10% CMVpp65-specific CTLs (determined by FACS), were first labeled with CFSE and incubated with CMVpp65-dimer (loaded at 24ºC for 2 days) or unloaded dimer, and then serially diluted in unlabeled autologous CD8-depleted PBMCs before contacting with printed anti-Ig,λ spots. The frequencies of CMVpp65-specific CTLs were: 10% (undiluted), 1%, 0.1%, 0.01%, and 0% (100% CD8-depleted PBMCs). Each anti-Ig,λ antibody subarray contains 10 x 10 spots. Individual spots have 1 drop of 333pl of printing solution and are 150µm in diameter. Cell seeding condition: 50µl of 2 x 106/ml total cells on an area of 6.8 x 6.8 mm2. Error bars in the graph stand for standard errors across the 10 x 10 spot replicates. Horizontal dashed line and dotted line stand for the mean intensity of 0% sample pre-incubated with CMVpp65-loaded dimer and unloaded dimer, respectively.
72
a 40000 30000
20000
10000 8000 7000
Spot intensity
6000 5000 4000 3000
2
10
43
100
2000
1000 800 700 600 500 400 300
0%
0.01%
0.10%
1%
10%
Antigen-specific CTL frequency
Continued Figure 3.5 Examination of cell capture across spot replicates (10 x 10 for each subarray) as a function of antigen-specific CTL frequency. (a) Spot intensity variability chart of cell suspensions containing CFSE-labeled M1.58-specific CTLs at frequencies of 0% (100% CD8 cell-depleted autologous PBMCs), 0.01%, 0.1%, 1%, and 10%. Individual spots were denoted as open circles and vertically aligned based on quantified fluorescence intensities. The fluorescence intensities of spots incubated with 0% M1.58-specific CTLs defined the range of background intensity. Spots with intensities that fall in this range are considered to be empty spots with no cell binding and shown in black color, whereas spots with intensities above this range are considered to have cell(s) captured and accordingly were shown in red color. For each frequency from 0.01% to 10%, the number next to the range bar stands for the number of spots containing bound cells. (b) Sum of the fluorescence intensities of spots represented in red color in (a) as a function of the frequency of antigen-specific CTLs. Experimental settings are the same as Figure 3.4, except that unloaded dimer samples were not included.
73
Figure 3.5 continued
b
74
Figure 3.6 Detection of CTLs that are specific to M1.58, NA.75, and PA.46 from CD8 cells of donor A8, by HLA A2-Ig based cellular microarrays and FACS. The level of nonspecific binding to unloaded dimer, denoted as Ul. dimer, is also included in the figure. The quantification method in microarray assay was: the sum of the intensities of spots having cell(s) bound calibrated against the sum of anti-CD3 spot intensities. Each subarray contains 5 x 5 spots, and individual spots have 32 drops of 333pl of printing solution and are 500µm in diameter. Spot-to-spot distance is about 1mm. Cell seeding condition: 50µl of ~2 x 106/ml total cells.
75
Supplementary information Mean spot intensity ± standard error
Additive
1549 ± 53.6
No additive
2191 ± 70.0
BSA, 0.5mg/ml
1322 ± 58.8
Glycerol, 2%(v/v)
200.8 ± 10.7
Sucrose, 0.5%(w/v)
208.2 ± 14.8
PEG 200, 5%(v/v)
431.1 ± 24.1
Glucose, 5%(w/v)
383.9 ± 38.1
PEG 1000, 0.5%(w/v)
184.8 ± 9.71
PEG 1000, 5%(w/v)
1690 ± 107
PVA 9000, 0.05%(w/v)
1264 ± 57.1
PVA 9000, 0.5%(w/v)
SP. Figure 3.1 Effect of print additives on human recombinant IFNγ (rhIFNγ) detection. Anti-IFNγ at a concentration of 0.5mg/ml was printed with 9 different additives, with no additive as a control. Each sample was printed at a replicate of 8 spots. Individual spots contain 333 picoliters of printing solution. A sandwich assay was conducted by contacting the array with 20ng/ml rhIFNγ followed by 20µg/ml PE-Cy7-anti-IFNγ. The mean spot intensity ± standard error for each sample across the spot replicates is shown to the left of the slide image and the identity of the spots in terms of the additives are shown to the right of the image.
76
CMVpp65-enriched CMVpp65-dimer
M1.58-enriched
PA.46-enriched
M1.58-dimer
PA.46-dimer
13.9%
71.9%
Unloaded dimer
Unloaded dimer
0.39%
0.67%
69.9%
Unloaded dimer 0.41%
SP. Figure 3.2 FACS detection and determination of the frequency of antigen-specific CTLs enriched and expanded with aAPCs. For each aAPC-enriched population, FACS plots on the top panel show cells stained with dimer loaded with the inducing peptide. That is, from left to right: CMVpp65-enriched cells stained with CMVpp65-(loaded) dimer, M1.58-enriched cells stained with M1.58- dimer, and PA.46-enriched cells stained with PA.46-dimer. Cells stained with unloaded dimer were included as negative control, as shown in the bottom panel.
77
References Ajikumar, P.K., J. Kiat, Y.C. Tang, J.Y. Lee, G. Stephanopoulos, andH.P. Too. 2007. Carboxyl-terminated dendrimer-coated bioactive interface for protein microarray: High-sensitivity detection of antigen in complex biological samples. Langmuir 23(10):5670-5677. Arakawa, T., Y. Kita, andS.N. Timasheff. 2007. Protein precipitation and denaturation by dimethyl sulfoxide. Biophysical Chemistry 131(1-3):62-70. Arstila, T.P., A. Casrouge, V. Baron, J. Even, J. Kanellopoulos, andP. Kourilsky. 1999. A direct estimate of the human alpha beta T cell receptor diversity. Science 286(5441):958-961. Buchli, R., R.S. VanGundy, H.D. Hickman-Miller, C.F. Giberson, W. Bardet, andW.H. Hildebrand. 2004. Real-time measurement of in vitro peptide binding to soluble HLA-A*0201 by fluorescence polarization. Biochemistry 43(46):14852-14863. Chan, P.Y., M.B. Lawrence, M.L. Dustin, L.M. Ferguson, D.E. Golan, andT.A. Springer. 1991. Influence of Receptor Lateral Mobility on Adhesion Strengthening between Membranes Containing Lfa-3 and Cd2. Journal of Cell Biology 115(1):245-255. Chen, D.S., Y. Soen, T.B. Stuge, P.P. Lee, J.S. Weber, P.O. Brown, andM.M. Davis. 2005. Marked differences in human melanoma antigen-specific T cell responsiveness after vaccination using a functional microarray. Plos medicine 2(10):1018-1030. Davis, M.M., J.J. Boniface, Z. Reich, D. Lyons, J. Hampl, B. Arden, andY.H. Chien. 1998. Ligand recognition by alpha beta T cell receptors. Annual Review of Immunology 16:523-+. Delehanty, J.B., andF.S. Ligler. 2003. Method for printing functional protein microarrays. BioTechniques 34:380-385.
78
Dembo, M., andG.I. Bell. 1987. The Thermodynamics of Cell-Adhesion. Current Topics in Membranes and Transport 29:71-89. Deviren, G., K. Gupta, M.E. Paulaitis, andJ.P. Schneck. 2007. Detection of antigen-specific T cells on p/MHC microarrays journal of molecular recognition 20(1):32-38. Genestier, L., R. Paillot, N. BonnefoyBerard, G. Meffre, M. Flacher, D. Fevre, Y.J. Liu, P. LeBouteiller, H. Waldmann, V.H. Engelhard, J. Banchereau, andJ.P. Revillard. 1997. Fas-independent apoptosis of activated T cells induced by antibodies to the HLA class I alpha 1 domain. Blood 90(9):3629-3639. Gianfrani, C., C. Oseroff, J. Sidney, R.W. Chesnut, andA. Sette. 2000. Human memory CTL response specific for influenza A virus is broad and multispecific. Human Immunology 61(5):438-452. Grakoui, A., S.K. Bromley, C. Sumen, M.M. Davis, A.S. Shaw, P.M. Allen, andM.L. Dustin. 1999. The immunological synapse: A molecular machine controlling T cell activation. Science 285(5425):221-227. Greten, T.F., J.E. Slansky, R. Kubota, S.S. Soldan, E.M. Jaffee, T.P. Leist, D.M. Pardoll, S. Jacobson, andJ.P. Schneck. 1998. Direct visualization of antigen-specific T cells: HTLV-1 Tax11-19-specific CD8(+) T cells are activated in peripheral blood and accumulate in cerebrospinal fluid from HAM/TSP patients. Proceedings of the National Academy of Sciences of the United States of America 95(13):7568-7573. Kwong, G.A., C.G. Radu, K. Hwang, C.J.Y. Shu, C. Ma, R.C. Koya, B. Comin-Anduix, S.R. Hadrup, R.C. Bailey, O.N. Witte, T.N. Schumacher, A. Ribas, andJ.R. Heath. 2009. Modular Nucleic Acid Assembled p/MHC Microarrays for Multiplexed Sorting of Antigen-Specific T Cells. J. Am. Chem. Soc. 131(28):9695-9703. Liu, Y.S., C.M. Li, L. Yu, andP. Chen. 2007. Optimization of printing buffer for protein microarrays based on aldehyde-modified glass slides. Frontiers in Bioscience 12:3768-3773. Nishioka, G.M., A.A. Markey, andC.K. Holloway. 2004. Protein damage in 79
drop-on-demand printers. J. Am. Chem. Soc. 126(50):16320-16321. Oelke, M., M.V. Maus, D. Didiano, C.H. June, A. Mackensen, andJ.P. Schneck. 2003. Ex vivo induction and expansion of antigen-specific cytotoxic T cells by HLA-Ig-coated artificial antigen-presenting cells. Nature Medicine 9(5):619-624. Olsen, A.W., P.R. Hansen, A. Holm, andP. Andersen. 2000. Efficient protection against Mycobacterium tuberculosis by vaccination with a single subdominant epitope from the ESAT-6 antigen. Eur. J. Immunol. 30(6):1724-1732. Parham, P., C.J. Barnstable, andW.F. Bodmer. 1979. Use of a Monoclonal Antibody (W6-32) in Structural Studies of Hla-a,B,C Antigens. Journal of Immunology 123(1):342-349. Rodenko, B., M. Toebes, S.R. Hadrup, W.J.E. van Esch, A.M. Molenaar, T.N.M. Schumacher, andH. Ovaa. 2006. Generation of peptide-MHC class I complexes through UV-mediated ligand exchange. Nature Protocols 1(3):1120-1132. Schneck, J.P., J.E. Slansky, S.M. O'Herrin, andT.F. Greten. 2005. Monitoring antigen-specific T cells using MHC-Ig dimers. In Short protocols in immunology: a compendium of methods from current protocols in immunology. Coligan JE, Bierer BE, Margulies DH, Sherach EM, Strober N, editors. Wiley. 15-12 - 15-11. Soen, Y., D.S. Chen, D.L. Kraft, M.M. Davis, andP.O. Brown. 2003. Detection and characterization of cellular immune responses using peptide-MHC microarrays. Plos biology 1(3):429-438. Stone, J.D., J. Walter E. Demkowicz, andL.J. Stern. 2005. HLA-restricted epitope identification and detection of functional T cell responses by using MHC-peptide and costimualtory microarrays PNAS 102:3744-3749. Tanabe, M., M. Sekimata, S. Ferrone, andM. Takiguchi. 1992. Structural and Functional-Analysis of Monomorphic Determinants Recognized by Monoclonal-Antibodies Reacting with the Hla Class-I Alpha-3 Domain. Journal of Immunology 148(10):3202-3209. 80
Ware, R., H. Jiang, N. Braunstein, J. Kent, E. Wiener, andC. Leonard. 1995. Human Cd8(+) T-Lymphocyte Clones Specific for T-Cell Receptor V-Beta Families Expressed on Autologous Cd4(+) T-Cells. Immunity 2(2):177-184. Wu, P., andD.W. Grainger. 2006. Comparison of hydroxylated print additives on antibody microarray performance. Journal of proteome research 5(11):2956-2965. Yewdell, J.T.W., andJ.R. Bennink. 1999. Immunodominance in major histocompatibility complex class I-restricted T lymphocyte responses. Annual Review of Immunology 17:51-88. Yewdell, J.W. 2006. Confronting complexity: Real-world immunodominance in antiviral CD8(+) T cell responses. Immunity 25(4):533-543. Zhu, H., andM. Snyder. 2003. Protein chip technology. current opinion in chemical biology 7:55-63.
81
CHAPTER 4 CELLULAR MICROARRAY ASSAY FOR QUANTIFYING PEPTIDE LOADING IN MHC-IG DIMER
4.1. Introduction As described in detail in Chapter 2, MHC-Ig dimer molecules have been widely used in flow cytometry (Greten et al., 1998), artificial antigen-presenting cell (aAPC) technologies (Oelke et al., 2003; Schutz et al., 2008), and cellular microarray technologies (Chen et al., 2005; Deviren et al., 2007). However, peptides employed in these technologies have been restricted to mainly a few immunodominant epitopes; meanwhile, the increasing importance of subdominant epitopes and the discovered broad breadth of immune response (Assarsson et al., 2008; Gianfrani et al., 2000; Oukka et al., 1996) predict the expansion of the multitude of peptides under study in these assays (Braga-Neto and Marques, 2006). At present, the major challenge has been the insufficient sensitivity of these assays compared to the extremely low
82
frequencies of CTLs specific for subdominant epitopes if present. Though a property of the entire assay system, assay sensitivity can largely depend on the peptide loading efficiency of the dimer, which is the extent of the replacement of resident endogenous peptides by the peptide of interest. In this context, it is important to study the peptide loading process of MHC-Ig dimer, in the hope to discover the optimal loading conditions for increased assay sensitivity. Furthermore, understanding the loading process could help eliminate possible bias in interpreting the data caused by different loading efficiencies across chemically distinctive peptides.
A number of assays have been developed over the years for measuring the binding of peptides to MHC class I molecules. Based on the form of MHC and/or peptide, these assays fall into three categories, cellular approach, cell-free approach, and solid phase approach. For each approach, there are both competitive binding assays and direct binding assays. Competitive binding assays focus on studying peptide affinity relative to a reference peptide and are generally qualitative, whereas direct binding assays focus on studying peptide-MHC interaction kinetics and equilibrium and are more quantitative.
Assays of cellular approach measure the binding of radio-labeled or fluorescencelabeled peptides to MHC molecules bound on intact cells. The use of cell-bound MHC eliminates the labor-intensive production and purification of soluble MHC molecules. The detection technique varies among assays in this category. In one study, the degree of radio-labeled peptide binding to MHC on intact cells was determined by radio counting of peptide-MHC complexes that were separated by
83
immunoprecipitation of cell lyses using antibodies against the MHC (Christinck et al., 1991). However in that study endogenous peptides were ignored in the kinetic binding model and subsequent data analysis, resulting in estimated rate constants not the true intrinsic values for peptide-MHC interaction. Another more recently developed cellbound MHC approach involves the mild acid treatment of MHC-expressing cell lines to elute endogenous peptides and then the measurement of the binding of fluorescently-labeled peptide to cell-bound MHC by flow cytometry (Kessler et al., 2003; vanderBurg et al., 1995). This assay has some complications such as the selection of a proper cell line that the peptide under measurement binds only the MHC allele of interest, not other co-expressed MHC alleles on the cells.
Assays of cell-free approach measure the binding of radio-labeled or fluorescencelabeled peptides to soluble peptide-free MHC molecules in solution. Compared with cell-bound MHC, the use of soluble MHC offers the flexibility of experimental conditions such as temperature and prevents possible cell physiological changes such as MHC expression level change at different experimental conditions. The employment of radio-labeling of peptides in earlier work required the separation of bound peptide from free peptide through size-based chromatographic methods such as gel filtration chromatography and spun column chromatography (Buus et al., 1995; Olsen et al., 1994; Sette et al., 1994) or through immunoprecipitation (Boyd et al., 1992), followed by radioactivity measurement of the two populations. Later use of fluorescence-labeling of peptides allowed the employment of various fluorescence measurement techniques including fluorescence polarization (Buchli et al., 2006; Buchli et al., 2004; Buchli et al., 2005; Dedier et al., 2001) and anisotropy (Binz et al.,
84
2003), and fluorescence resonance energy transfer (FRET) (Gakamsky et al., 1996; Gakamsky et al., 1999; Gakamsky et al., 2000), which can discriminate bound from free peptides in solution. Thus, fluorescence-labeling based assays could avoid the peptide separation step and monitor fluorescence change in a real-time, compared with radio-labeling based assays. While some of these assays of this cell-free approach are competitive binding assays (Buchli et al., 2006; Buchli et al., 2004; Buchli et al., 2005), most assays conducted more detailed direct kinetic and equilibrium studies and extended the analyses into thermodynamic features of the interactions such as entropy and enthalpy changes via Eyring analyses. Particularly, the series of FRET studies (Gakamsky et al., 1996; Gakamsky et al., 1999; Gakamsky et al., 2000) and subsequent fluorescence anisotropy study (Binz et al., 2003) found an allosteric mechanism as a general feature of peptide-heavy chain-β2m heterotrimer interactions, through which the association or dissociation of peptide or β2m with or from the heavy chain affects the association or dissociation of the other component (β2m or peptide) with or from the heavy chain.
However, these peptide-labeling based cellular and cell-free assays have a disadvantage: the chemical addition of fluorochorme or radio label or the substitution of amino acid residue for the addition, even at a non-anchor position in the peptide binding motif (Falk et al., 1991), could largely alter the overall binding behavior of the peptide to the MHC allele as observed (Christinck et al., 1991; Kessler et al., 2003).
85
Assays of solid phase approach measure the binding of soluble MHC molecules to peptides bound on a solid phase support. The solid support can be simply plastic surface of microtitre plates, on which synthetic peptides were coated and soluble radio-labeled MHC molecules were incubated in the plate wells for a time period followed by extensive washing to remove unbound MHC and counting of radio activity from bound MHC molecules (Bouillot et al., 1989). In more recent studies, surface plasmon resonance (SPR) technology has been exploited to evaluate the interaction of soluble peptide-free MHC molecules in flowing solution with synthetic peptides immobilized on sensor surface (Khilko et al., 1993; Khilko et al., 1995). This label-free technology allowed real-time monitor of the association and dissociation of peptide-MHC complex, from which binding rate constants and equilibrium dissociation constant KD were determined. A big drawback for this technology though, is the possible alteration of binding kinetics due to peptide immobilization on solid substrate. As shown quantitatively (Khilko et al., 1995), for all three peptides tested, immobilization of the same peptides through the substitution of different non-anchor residues with cysteine resulted in dissociation rate constants spanning over two orders of magnitude (10-6 s-1 to 10-4 s-1), indicating that the measured kinetics and affinity probably differ from their true intrinsic values.
In this chapter, we demonstrate the feasibility of the developed anti-Ig,λ based MHCIg cellular microarray assay for quantitatively evaluating the kinetics of passive loading of peptides to soluble HLA A2-Ig dimer. In the assay, dimer is loaded with peptides for different time periods and then incubated with fluorescently-labeled, antigen-specific CTLs. Following the incubation, binding of the cells onto microarray
86
spots that only capture cells with surface bound peptide-HLA complex is measured. The extent of cell binding correlates with the extent of peptide loading. The assay does not involve any peptide and MHC labeling or immobilization, so probable alteration of binding kinetics is eliminated. A model describing the kinetics of the passive loading process has been set up. We found the passive loading of peptide into HLA A2-Ig an extremely slow process, which suggests conformational changes of the formed peptide-MHC complexes for TCR recognition that can only be detected in this advantageous cellular assay, not traditional cell-free assays.
4.2. Materials and Methods Peptides and antigen-specific cytotoxic T lymphocytes (CTLs). High purity (~95%) peptide CMVpp65 (NLVPMVATV) was purchased from Genscript (Piscataway NJ), and reconstituted to 2 mg/ml in PBS solution containing 10% (v/v) DMSO before use. Healthy donors were HLA A2.1-phenotyped by flow cytometry. CMVpp65-specific CTLs were then induced and expanded using peptide-loaded artificial antigen presenting cell (aAPC) technology, as described in (Oelke et al., 2003) and “Materials and Methods” section of Chapter 3.
pMHC-Ig based cellular microarray assays. HLA A2-Ig cellular microarray assays were conducted to systematically evaluate the kinetics of peptide loading into HLA A2-Ig dimer, with CMVpp65 peptide as an example. Specifically, dimer was passively loaded with CMVpp65 for different time periods (up to 23 days), and then incubated with CMVpp65-specific CTLs followed by cell binding assay on the
87
microarray printed with anti-mouse Ig,λ. Loading was conducted at three different temperatures, 4ºC, 24ºC (room temperature) and 37ºC, to examine the effects of temperature changes on the kinetics of loading. In each assay, anti-human CD3 antibody was printed on the same slides as calibration spots quantifying the maximum level of CD8 CTL binding. Refer to “Materials and Methods” section of Chapter 3 for the experimental procedures of fabricating and processing the cellular microarrays. HLA A2-Ig dimer in 1xPBS solution at a concentration of 0.5 mg/ml was purchased from BD Pharmingen (San Diego CA). The dimer was passively loaded by adding the peptide of interest in 160-fold molar excess, and then incubating the solution at constant temperature over the time course for each experiment.
Flow-activated cell sorting (FACS) measurements. FACS measurements were performed using a FACSCalibur flow cytometer (BD Biosciences, San Jose CA) available at Ohio State Flow Cytometry Core Lab. Dimer staining was performed to measure the frequency of antigen-specific CTLs in aAPC-enriched populations. Refer to the “Materials and Methods” of Chapter 3 for the procedures of dimer staining.
FACS measurements were also conducted to quantify the binding of PE-labeled antibodies to cells using QuantiBRITETM PE calibration beads (BD Biosciences). Briefly, a lyophilized pellet of the beads was resuspended with 500 µl (microliters) of 1x PBS solution, and then run at the same instrument settings as cell samples. Four peaks, corresponding to beads conjugated with four pre-calibrated levels of PE molecules, were identified on a FL2 histogram and geometric mean intensity of PE of each bead population was quantified. A standard curve of PE intensity versus number
88
of PE per bead was then obtained. Geometric mean intensities from PE-labeled antibodies were converted to the number of molecules per cell using the standard curve.
All FACS data reduction, including dimer staining and QuantiBRITE PE beads measurement, was carried out using FlowJo software (TreeStar, San Carlos CA).
Fluorescence polarization (FP) assay. FP competitive binding assays were conducted by Nicole Guzman. FP measurements were conducted using a SpectraMax M5 reader (Molecular Devices, Sunnyvale CA). When set for fluorescence polarization, the reader detects changes in the polarization of emitted light due to the rotational speed of molecules. Small fluorescently labeled ligands rotate faster than much larger receptors. When these ligands associate with receptors, their rotational speed is reduced. The SpectraMax reader detects these changes in polarization and measures real-time binding events in solution.
FP measurements of the association of a fluorescein (FITC)-labeled peptide p5 (ALMDKVLKV, Genescript) to HLA A2-Ig dimer were carried out at 4ºC and 24ºC. Each reaction mixture included HLA A2-Ig dimer at a concentration of 200 nM and the peptide p5 at a concentration of 2 nM. To avoid evaporation and photobleaching, the reaction mixture was incubated in microcentrifuge tubes covered with aluminum foil. A 20µl solution of the reaction mixture was loaded into individual wells of a black, non-binding surface (NBS) 384-well plate (Corning). Three control groups were included: buffer only (blank), protein only and pFITC only. The plate was read
89
after different time periods until equilibrium has been reached, approximately after 3 days. Each experiment was performed in quadruplicate and the mean and standard deviation of these 4 measurements were calculated before fitting.
Image acquisition and data analysis. Microarray images were acquired using a ProScanArray fluorescence scanner (PerkinElmer, Waltham MA). The slides were scanned at 5 µm resolution. Image quantification and data analysis were performed using ScanArray Express 3.0 (PerkinElmer, Waltham MA) and JMP 6.0 (SAS, Cary NC) software. Bar graph plotting and linear data fitting were performed using Sigma plot 10.0 (Systat Software, San Jose CA). Nonlinear fitting of FP measurements and cellular microarray measurements was performed using Matlab 7.0 (The Mathworks, Inc., Natick MA), based on a cost function that was generated for minimizing the sum of error between experimental data and calculated values.
4.3. Results 4.3.1. Microarray assays probing peptide loading efficiency of HLA A2-Ig Figure 4.1 shows representative fluorescence-scanned subarray images of cell binding on spots, for cells pre-incubated with CMVpp65 that was loaded for different time periods at 4ºC and 24ºC. For each subarray, mean fluorescence intensity and standard error across the spot replicates were quantified against the mean intensity of anti-CD3 calibration spots (cell binding on anti-CD3 spots also shown in Figure 4.1). As Figure 4.2 shows, nonspecific binding of cells to unloaded dimer was at a low level of below 1% for loading at both temperatures. When dimer was loaded with
90
CMVpp65 at 4ºC for 3 days, 44% of CMVpp65-specific CTLs were captured on antiIg,λ, and the percentage was approximately 36% for loading at 24ºC, indicating that CMVpp65 was indeed binding to the dimer to form CMVpp65-HLA A2 complexes during the loading periods. As the incubation of CMVpp65 and dimer proceeded, the percentage of CMVpp65-specific CTLs that were captured onto the microarrays increased, to up to 87% (of total CTLs) for loading at 4ºC and 76% for loading at 24ºC respectively, indicating that more CMVpp65-HLA A2 complexes were formed in the loading solutions. In terms of the effect of temperature on loading, loading at 24ºC for 1 day resulted in a cell binding level of 49%, almost as high as the 52% cell binding when loading at 4ºC for 5 days, indicating that loading at 24ºC has a faster kinetics than loading at 4ºC. These results demonstrate that CMVpp65 loading to HLA A2-Ig is both time and temperature dependent.
CMVpp65 loading to HLA A2-Ig at a higher temperature, 37ºC, was also examined (Figure 4.3). Using dimer loaded with CMVpp65 at 37ºC for 1 day, 7.3% of total CTLs were consequently captured in the cellular microarray assay. Compared with loading for the same time period but at lower temperatures 4ºC and 24ºC, this much lower percentage of captured CTLs indicates that there was thermal denaturation or aggregation (or both) of labile dimer molecules at 37ºC. As loading proceeded for 3 days and further 5 days, the percentage of cells captured decreased to 0.30% and 2.2% respectively. To determine whether this loss of functionality is reversible, the loading samples were subsequently re-incubated/rescued at 4ºC for 14 days and then reexamined in cellular microarry assays (Figure 4.3). The percentage of captured CMVpp65-specific CTLs determined using rescued dimer that was loaded at 37ºC for
91
1 day dramatically increased to about 54%, indicating the thermal denaturation or aggregation was reversible; However, since this re-incubation at 4ºC for 14 days resulted in a binding level comparable to loading at 4ºC for only 5 days, not all denatured or aggregated were rescued. Those still denatured or aggregated dimers either need longer incubation at 4ºC to recover, or the denaturation or aggregation of them had developed to irreversible. For dimer loaded at 37ºC for 3 days, after rescue the percentage of captured CTLs only increased to 3.4%; for loaded at 37ºC for 5 days, the percentage even decreased to 0.76%, indicating that the loss of functionality developed from reversible to irreversible during the incubation time of up to 5 days.
4.3.2. Correlation of cell binding with cell-surface bound dimers For comparison, flow cytometry measurements using dimer staining of CMVpp65specific CTLs were conducted in parallel with cellular microarray assays. Figure 4.4 shows representative results of loading CMVpp65 at 4ºC for different time periods. Employing the same batch of CMVpp65-aAPC enriched and expanded CTLs, the detected percentage of CMVpp65-specific CTLs increased as peptide loading proceeded, from 65.1% for a loading period of 3 days, to 68.1% for 5 days, and finally to 76.2% for 23 days. This trend of increase in detected CMVpp65-specific CTLs in flow cytometry is consistent with the increase in capture of CMVpp65specific CTLs in cellular microarray assays (top graph of Figure 4.2).
To compare quantitatively with cellular microarray assays, the percentages of CMVpp65-specific CTLs shown in Figure 4.4 were converted to percentages over CD8 CTLs by taking into account the population that is positive in CD8 expression
92
but negative in cognate TCRs (the FITC+PE- population). The conversion resulted in 73.3%, 77.1% and 87.8% CMVpp65-specific CTLs over all CD8 cells for loading for 3 days, 5 days, and 23 days respectively. By comparison, cellular microarray assays detected the percentage of CMVpp65-specific CTLs to be 25-30% lower than flow cytometry for dimer loaded for 3 days and 5 days, but detected almost exactly the same percentage as flow cytometry for dimer loaded for 23 days, at which loading has reached saturation (as will be shown in Figure 4.7). These together indicate that, without compromising the accuracy in quantitative detection when using dimer that has been loaded to saturation, microarray assay has a larger dynamic range of change in detected CMVpp65-specific CTLs than flow cytometry in response to peptide loading process, and therefore is more sensitive than flow cytometry as a technology for studying the kinetics of peptide loading.
For CMVpp65 loading at 4ºC and 24ºC, the percentage of captured antigen-specific CTLs measured in cellular microarray assay was plotted against the average number of PE molecules bound per cell on CMV-specific CTLs measured in FACS (Figure 4.5). First of all, the figure strongly indicates a linear, loading temperatureindependent correlation between the level of antigen-specific CTL binding on microarray spots and the number of CMVpp65-loaded dimer molecules bound on the surfaces of the cells. The more dimer molecules have been successfully loaded, the more loaded dimer molecules bound per cell on average, leading to more cells bound onto the spots via the interaction between anti-Ig,λ on spots and the Ig portion of dimer on cell surface. This linear relation offers the mechanistic basis for using cellular microarry assay to study peptide loading. Second, loading at both 4ºC and
93
24ºC reached saturation level (will be shown in Figure 4.7) of about 7,000 PE molecules bound per cell. According to the manufacturer, the ratio of PE to antibody in the PE-labeled anti-mouse IgG1 conjugate is about 1:1. If each conjugate simultaneously binds to two dimer molecules, there are 14,000 dimers bound per cell at the saturation level, which can correspond to 28,000 copies of TCRs on cell surface, consistent with the known number of ~ 30,000 copies (Charles A. Laneway et al., 2005). However, since it is a polyclonal antibody more than one antibody could bind to one dimer and therefore we may be over-estimating the number of dimers bound per cell at saturation level. Finally, according to a thermodynamic model of cell adhesion to substrate via specific receptor-ligand interaction, a threshold number of receptor-ligand bonds exist for a cell to overcome nonspecific repulsion for adhesion to substrate (Dembo and Bell, 1987). The x-intercept of 604 PE molecules bound per cell in Figure 4.5 suggests that there could be between 604 and 2416 TCRs per cell engaged as the threshold for the initiation of CTL adhesion on the spots. This threshold number is at the lower end of the range of the number of TCRs engaged for synapse formation (Grakoui et al., 1999).
4.3.3. Kinetic model of peptide loading to HLA A2-Ig dimer
We developed a mathematical model, as shown in Eq. 1, to describe the process of peptide loading to HLA-A2 dimer. Let p*MHC denote MHC in complex with an endogenous peptide p*. eMHC is the empty MHC following the dissociation of p* from the complex. eMHC can aggregate or unfold to become inactive MHC, iMHC, which is in a peptide-inaccessible form, as has been well documented (Binz et al.,
94
2003; Gakamsky et al., 2000). eMHC, in peptide-accessible form, can bind specific peptide of interest fp in solution to form peptide-MHC complex fpMHC. Quantities labeled as ki are the rate constants of the corresponding reactions. To simplify the model for only containing key reactions, the re-binding of dissociated endogenous peptides to eMHC and the dissociation of fp from the fpMHC complex are not considered. Ordinary differential equations (ODEs) describing the kinetics of reactions in Eq.1 were derived in supplemental information.
k1 k2 p* MHC → eMHC → fpMHC
k3 k-3
iMHC
(Eq. 1)
Prior to fitting cellular microarray measurements, cell-free FP measurements of fluorescence-labeled peptide p5 binding to HLA A2-Ig dimer at 4ºC and 24ºC were fitted versus time using the model (Figure 4.6). The best fit parameter values were determined to be: k1 = 131 day-1, k2 = 23.2 µM-1day-1, k3 = 307 µM-1day-1, k-3 = 0.923 day-1, and n = 2 for 4ºC, and k1 = 31.3 day-1, k2 = 122 µM-1day-1, k3 = 214 µM-1day-1, k-3 = 2.02 day-1, and n = 2 for 24ºC. n is the number of MHC molecules involved in the inactive form. Though n = 2 suggests aggregation (n > 1) instead of unfolding (n = 1), it is noted that this FP assay cannot distinguish aggregation from unfolding; in fact, native polyacrylamide gel electrophoresis (PAGE) experiments did not indicate the formation of aggregates for loaded dimer sample compared with unloaded dimer (results not shown). For both loading temperatures investigated, the conversion of the 95
inactive MHC to the active conformation constitutes the rate-limiting step among the reactions in the model, consistent with previous findings (Binz et al., 2003; Gakamsky et al., 2000). The k1 value for loading at 4ºC is more than 5 times higher than that for loading at 24ºC, indicating that endogenous peptides dissociate faster from the complex with HLA at a lower temperature. The k2 value at 24ºC is approximately 5fold as much as the value at 4ºC, showing specific peptide associates with empty MHC more rapidly at the elevated temperature. In contrast to the significant temperature effect on k2, k3 is slightly higher at 4ºC than at 24ºC, suggesting that empty MHC molecules aggregate slightly faster at the lower temperature. Taken together, the greater ratio of k3 over k2 at 4ºC than at 24ºC suggests that empty MHC molecules at the lower temperature 4ºC may have a stronger tendency to aggregate or unfold than to bind specific peptides for forming complexes.
Compared with peptide loading measured using FP assays, which reached equilibrium within 1 day at 24ºC, peptide loading measured using cellular microarray assays did not reach equilibrium until after 5 days. The apparently slower kinetics observed in cellular microarray assay than in FP assay suggest that, at least two isomeric complexes, one short-lived and the other long-lived are formed during peptide loading, with the short-lived complex being an intermediate that can convert to the long-lived complex. Both isomers are detected in FP assays, whereas only the long-lived isomer displays the complementary conformation for cell-surface bound antigen-specific TCRs, and thus detected in cellular microarray assays.
96
The mechanistic model is shown in Eq. 2 and ODEs for the model are derived in supplemental information. Symbols in Eq. 2 are the same as in Eq.1, except that (fpMHC)TCR-inacc stands for the short-lived fpMHC complex that is in a TCRinaccessible conformation, (fpMHC)TCR-acc stands for the long-lived fpMHC complex that is in a TCR-accessible conformation, and k4 is the rate constant for the conversion from inactive conformation to active conformation. For simplicity, the conversion of the long-lived complex to the short-lived complex is ignored.
k1 k2 k4 p* MHC → eMHC →( fpMHC )TCR −inacc →( fpMHC )TCR − acc
k3 k-3
iMHC
(Eq. 2)
The kinetic model in Eq. 2 was fit to cellular microarray measurements (normalized mean spot intensity versus the loading time), using the parameters values obtained from fitting FP measurements as initial parameter guesses (Figure 4.7). The fit parameter values are: k1 = 212 day-1, k2 = 22.6 µM-1day-1, k3 = 353 µM-1day-1, k-3 = 0.866 day-1, n = 2, and k4 = 0.215 day-1 for 4ºC, and k1 = 34.4 day-1, k2 = 122 µM-1day1
, k3 = 198 µM-1day-1, k-3 = 2.02 day-1, n = 2, and k4 = 1.54 day-1 for 24ºC. As show in
the figure, CMVpp65 loading at 4ºC reached saturation at 23 days; when the loading temperature was elevated to 24ºC, cell binding increased at a faster rate and reached saturation at about 5 days (though to a level lower than that of 4ºC). For parameters k1, k2, k3, k-3, and n, the fit values are very close to the initial guesses; this high stability of the fit supports that the reaction pathway in Eq. 1 is indeed part of Eq. 2. The value of
97
k4 is on the same magnitude but even lower than k-3, indicating that the conversion from the TCR-inaccessible conformation to the accessible conformation is another rate-limiting step for the loading process. A rise in loading temperature from 4ºC to 24ºC resulted in a 6-fold increase in k4.
4.4. Discussion and Conclusions The mechanistic model we proposed explicitly separates the components of dissociation, association, and conformational changes occurring in the complex process of peptide loading to HLA A2-Ig dimer. Previous kinetic studies on peptide association and dissociation with HLA A2 could not separate the steps; as a result, the obtained rate constants were the apparent constants for multiple steps (Binz et al., 2003; Gakamsky et al., 2000).
The cellular microarray assay we have developed is a functional assay based on the complimentary match between peptide-dimer complex and cognate T cell receptors (TCRs), and thus possesses the great advantage of selectively binding to only the TCR-recognizable conformation(s). In “peptide register shifting”, a single peptide can bind to the same MHC using different anchor residues to form conformational isomers, which can differentially activate T cells and elicit T cell responses (Bankovich et al., 2004; Rabinowitz et al., 1997). Though most peptide register shifting observations have been on MHC class II molecules owing to possibly the open-ended structure of peptide-binding groove, studies on the dissociation kinetics of peptide-MHC class I complex have frequently observed bi-phases, strongly
98
suggesting that the complex existed in at least two conformations, one short-lived and the other one long-lived (Gakamsky et al., 1996; Gakamsky et al., 1999; Gakamsky et al., 2000). An induced fit that involves conformational rearrangements of the peptidedimer complex upon peptide binding, or selective binding from a pre-existing equilibrium ensemble of configurations of the peptide-dimer complex, could also be the mechanism for slow association (Tsai et al., 1999).
Because of both peptide-independent rate-limiting steps, the overall loading kinetics of different peptides binding to HLA A2-Ig dimer appear to fall within a small range, as seen from CMVpp65, M1.58 and PA.46 (results not shown). For future immunological studies on a larger number of peptides, it is not realistic to obtain antigen-specific CTLs for each epitope for studying loading kinetics and revealing the optimal loading condition. With findings in this work, it is reasonable to use the same loading condition (i.e. loading time, temperature, etc.) for all the peptides of interest.
By fitting the proposed model to cellular microarray data, we estimated CMVpp65 loading efficiency to be 71% after 5 day incubation with HLA A2-Ig at 24ºC and 83% after 23 day incubation at 4ºC. Considering dimer molecules in approximately 280fold molar excess to TCRs (assuming 30,000 copies of TCRs on T cell surface) in the assay, the actual loading efficiency can be as low as only 1/280 of the estimated magnitudes. However, the model does not take into account the interaction between T cells and pMHC, and as cells fall and tend to accumulate to the bottom of the sample tube due to gravity, the probability for dimer encountering and interacting with surface-bound TCRs can be much reduced.
99
To conclude, this work has been the first report that systematically studied the passive loading of peptide into HLA A2-Ig dimer fusion protein. The assay method, cellular microarray, which does not depend on any peptide and MHC labeling or immobilization, serves as a new methodology for quantifying peptide loading. Results obtained in this work will largely improve the current understanding of passive loading process, especially the possible formation of multiple conformational complexes, and facilitate cellular immunological research involving the use of peptide-loaded monomeric or dimeric MHC as antigen-presenting agent. This cellular microarray technology has more applications beyond quantifying peptide loading, such as serving as a new HLA A2-Ig cellular microarray technology for the detection of antigen-specific CTLs.
100
(a) Dimer loaded with CMVpp65 at 4ºC Unloaded dimer
3 days
5 days
15 days
23 days
Anti-CD3
(b) Dimer loaded with CMVpp65 at room temperature (24ºC) Unloaded dimer
5 hours
1 day
5 days
19 days
Anti-CD3
Figure 4.1 The loading efficiency of HLA-A2-Ig with CMVpp65 is strongly time-dependent. Slides were printed identically with 0.5mg/ml polyclonal anti-mouse Ig,λ. To standardize cell binding level, anti-human CD3 was printed into separate subarrays. Each subarray contains 5 x 5 spots. Individual spots have 32 drops of 333pl of printing solution and are 500µm in diameter. Spot-to-spot distance is 1000µm. Dimers were loaded with CMVpp65 peptide for varying time periods at two temperatures: 4ºC and 24ºC (room temperature). CFSE-labeled CMVpp65-specific cells were pre-incubated with dimers and then contacted with anti-Ig,λ spots, or directly contacted with anti-CD3 spots. Cells pre-incubated with unloaded dimer were used as a control for nonspecific binding. The concentration of the cells for all samples was adjusted to be the same: about 0.5x106/ml before incubation on the microarrays, and 50µl of cell solution was applied to each subarray. Representative subarray fluorescent images are shown in for (a) 4ºC and (b) 24ºC.
101
Continued
Figure 4.2 Quantified mean spot fluorescence intensities for images in Figure 4.1. Top bar graph shows cell binding levels for dimer loaded at 4ºC and bottom graph (next page) shows cell binding levels for dimer loaded at 24ºC. “UL. dimer” Stands for unloaded dimer. Error bars standard for the standard error across 25 spot replicates of each sample.
102
Figure 4.2 Continued
103
Figure 4.3 Loading of HLA A2-Ig dimer with CMVpp65 peptide at 37ºC. Dimer was incubated with CMVpp65 sealed in a 37ºC water bath for 1 day, 3 days, and 5 days. The resulting samples were examined in a cellular microarray assay. After that, the rest of the dimers were kept at 4ºC for 14 days and then re-examined in the same manner. The mean spot intensities were plotted as the ratios against the cell binding level on anti-CD3 spots; error bars standard for standard errors across spot replicates.
104
Unloaded dimer
Dimer (loaded) with CMVpp65 for 3 days
1.46%
65.1%
Dimer with CMVpp65 for 5 days
Dimer with CMVpp65 for 23 days
68.1%
76.2%
Figure 4.4 Representative FACS analyses of CMVpp65-specific CTLs by dimer staining for dimers loaded at 4ºC for different time periods, as shown on top of each density plot. The procedure of dimer staining has been described in the methods section.
105
Figure 4.5 Plot of mean spot intensity (calibrated against anti-CD3 spots) versus number of PE molecules bound per cell on CMVpp65-specific CTLs, using dimer loaded with CMVpp65 peptide for varying time periods at 4ºC and 24ºC. The number of bound PE molecules was converted from the geometric mean intensity of PE of the CMVpp65-specific CTLs (the PE+FITC+ population) detected in flow cytometry. Briefly, BD QuantiBRITETM PE beads were analyzed by flow cytometry under the same instrument settings as cell samples, from which a linear correlation was obtained between the geometric mean intensity of PE and the number of PE molecules per bead (log IPE = 0.9986 * log (PE/bead) – 1.454, R2=1.000). PE intensities were then interpolated into the linear standard curve to obtain corresponding numbers of PE molecules bound per cell.
106
Figure 4.6 Fitting of fluorescence polarization measurement of peptide p5 binding to HLA A2-Ig at 4ºC and 24ºC. The fitted parameter values are k1 = 131 day-1, k2 = 23.2 µM-1day-1, k3 2 = 307 µM-1day-1, k-3 = 0.923 day-1, n = 2 and R = 0.99 for 4ºC, and k1 = 31.3 day-1, k2 = 122 2 µM-1day-1, k3 = 214 µM-1day-1, k-3 = 2.02 day-1, n = 2 and R = 0.99 for 24ºC.
107
Figure 4.7 Fitting of cellular microarray measurement of peptide CMVpp65 binding to HLA A2-Ig at 4ºC and 24ºC. Please refer to Figure 4.6 for initial parameter guesses. The fitted parameter values are k1 = 212 day-1, k2 = 22.6 µM-1day-1, k3 = 353 µM-1day-1, k-3 = 0.866 day-1, n = 2, k4 = 0.215 day-1 and R2= 0.96 for 4ºC, and k1 = 34.4 day-1, k2 = 122 µM-1day-1, k3 = 198 µM-1day-1, k-3 = 2.02 day-1, n = 2, k4 = 1.54 day-1 and R2= 0.91 for 24ºC.
108
Supplemental information Ordinary differential equations describing the mechanism in Eq. 1: d * [ p MHC ] = − k1[ p* MHC ] dt d [eMHC ] = k1[ p* MHC ] − k2 [eMHC ][ fp ] − k3[eMHC ]n + k−3n[iMHC ] dt d [iMHC ] = k3 [eMHC ]n − k−3n[iMHC ] dt d [ fpMHC ] = k2 [eMHC ][ fp ] dt [ fp ] = [ fp ]0 − [ fpMHC ]
Ordinary differential equations describing the mechanism in Eq. 2: d * [ p MHC ] = − k1[ p* MHC ] dt d [eMHC ] = k1[ p* MHC ] − k2 [eMHC ][ fp ] − k3[eMHC ]n + k−3n[iMHC ] dt d [iMHC ] = k3 [eMHC ]n − k−3n[iMHC ] dt d [ fpMHC ]TCR −inacc = k2 [eMHC ][ fp ] − k4 [ fpMHC ]TCR −inacc dt d [ fpMHC ]TCR − acc = k4 [ fpMHC ]TCR −inacc dt [ fp ] = [ fp ]0 − [ fpMHC ]
109
References Assarsson, E., H.H. Bui, J. Sidney, Q. Zhang, J. Glenn, C. Oseroff, I.N. Mbawuike, J. Alexander, M.J. Newman, H. Grey, andA. Sette. 2008. Immunomic Analysis of the Repertoire of T-Cell Specificities for Influenza A Virus in Humans. Journal of Virology 82(24):12241-12251. Bankovich, A.J., A.T. Girvin, A.K. Moesta, andK.C. Garcia. 2004. Peptide register shifting within the MHC groove: theory becomes reality. Molecular Immunology 40(14-15):1033-1039. Binz, A.K., R.C. Rodriguez, W.E. Biddison, andB.M. Baker. 2003. Thermodynamic and kinetic analysis of a peptide-class I MHC interaction highlights the noncovalent nature and conformational dynamics of the class I heterotrimer. Biochemistry 42(17):4954-4961. Bouillot, M., J. Choppin, F. Cornille, F. Martinon, T. Papo, E. Gomard, M.C. Fourniezaluski, andJ.P. Levy. 1989. Physical Association between Mhc ClassI Molecules and Immunogenic Peptides. Nature 339(6224):473-475. Boyd, L.F., S. Kozlowski, andD.H. Margulies. 1992. Solution Binding of an Antigenic Peptide to a Major Histocompatibility Complex Class-I Molecule and the Role of Beta-2-Microglobulin. Proceedings of the National Academy of Sciences of the United States of America 89(6):2242-2246. Braga-Neto, U.M., andE.T.A. Marques. 2006. From functional genomics to functional immunomics: New challenges, old problems, big rewards. Plos Computational Biology 2(7):651-662. Buchli, R., R.S. VanGundy, C.F. Giberson, andW.H. Hildebrand. 2006. Critical factors in the development of fluorescence polarization-based peptide binding assays: An equilibrium study monitoring specific peptide binding to soluble HLA-A*0201. Journal of Immunological Methods 314(1-2):38-53. Buchli, R., R.S. VanGundy, H.D. Hickman-Miller, C.F. Giberson, W. Bardet, andW.H. Hildebrand. 2004. Real-time measurement of in vitro peptide binding to soluble HLA-A*0201 by fluorescence polarization. Biochemistry 43(46):14852-14863.
110
Buchli, R., R.S. VanGundy, H.D. Hickman-Miller, C.F. Giberson, W. Bardet, andW.H. Hildebrand. 2005. Development and validation of a fluorescence polarization-based competitive peptide-binding assay for HLA-A*0201 - A new tool for epitope discovery. Biochemistry 44(37):12491-12507. Buus, S., A. Stryhn, K. Winther, N. Kirkby, andL.O. Pedersen. 1995. ReceptorLigand Interactions Measured by an Improved Spun Column Chromatography Technique - a High-Efficiency and High-Throughput Size Separation Method. Biochimica Et Biophysica Acta-General Subjects 1243(3):453-460. Charles A. Laneway, J., P. Travers, M. Walport, andM.J. Shlomchik. 2005. Immunobiology-the immune system in health and disease. Garland Science. Chen, D.S., Y. Soen, T.B. Stuge, P.P. Lee, J.S. Weber, P.O. Brown, andM.M. Davis. 2005. Marked differences in human melanoma antigen-specific T cell responsiveness after vaccination using a functional microarray. Plos medicine 2(10):1018-1030. Christinck, E.R., M.A. Luscher, B.H. Barber, andD.B. Williams. 1991. Peptide Binding to Class-I Mhc on Living Cells and Quantitation of Complexes Required for Ctl Lysis. Nature 352(6330):67-70. Dedier, S., S. Reinelt, R. Severine, G. Folkers, andD. Rognan. 2001. Use of fluorescence polarization to monitor MHC-peptide interactions in solution. Journal of Immunological Methods 255(1-2):57-66. Dembo, M., andG.I. Bell. 1987. The Thermodynamics of Cell-Adhesion. Current Topics in Membranes and Transport 29:71-89. Deviren, G., K. Gupta, M.E. Paulaitis, andJ.P. Schneck. 2007. Detection of antigenspecific T cells on p/MHC microarrays journal of molecular recognition 20(1):32-38. Falk, K., O. Rotzschke, S. Stevanovic, G. Jung, andH.G. Rammensee. 1991. AlleleSpecific Motifs Revealed by Sequencing of Self-Peptides Eluted from Mhc Molecules. Nature 351(6324):290-296. Gakamsky, D.M., P.J. Bjorkman, andI. Pecht. 1996. Peptide interaction with a class I major histocompatibility complex-encoded molecule: Allosteric control of the ternary complex stability. Biochemistry 35(47):14841-14848.
111
Gakamsky, D.M., L.F. Boyd, D.H. Margulies, D.M. Davis, J.L. Strominger, andI. Pecht. 1999. An allosteric mechanism controls antigen presentation by the H2K(b) complex. Biochemistry 38(37):12165-12173. Gakamsky, D.M., D.M. Davis, J.L. Strominger, andI. Pecht. 2000. Assembly and dissociation of human leukocyte antigen (HLA)-A2 studied by real-time fluorescence resonance energy transfer. Biochemistry 39(36):11163-11169. Gianfrani, C., C. Oseroff, J. Sidney, R.W. Chesnut, andA. Sette. 2000. Human memory CTL response specific for influenza A virus is broad and multispecific. Human Immunology 61(5):438-452. Grakoui, A., S.K. Bromley, C. Sumen, M.M. Davis, A.S. Shaw, P.M. Allen, andM.L. Dustin. 1999. The immunological synapse: A molecular machine controlling T cell activation. Science 285(5425):221-227. Greten, T.F., J.E. Slansky, R. Kubota, S.S. Soldan, E.M. Jaffee, T.P. Leist, D.M. Pardoll, S. Jacobson, andJ.P. Schneck. 1998. Direct visualization of antigenspecific T cells: HTLV-1 Tax11-19-specific CD8(+) T cells are activated in peripheral blood and accumulate in cerebrospinal fluid from HAM/TSP patients. Proceedings of the National Academy of Sciences of the United States of America 95(13):7568-7573. Kessler, J.H., B. Mommaas, T. Mutis, I. Huijbers, D. Vissers, W.E. Benckhuijsen, G.M.T. Schreuder, R. Offringa, E. Goulmy, C.J.M. Melief, S.H. van der Burg, andJ.W. Drijfhout. 2003. Competition-based cellular peptide binding assays for 13 prevalent HLA class I Alleles using fluorescein-labeled synthetic peptides. Human Immunology 64(2):245-255. Khilko, S.N., M. Corr, L.F. Boyd, A. Lees, J.K. Inman, andD.H. Margulies. 1993. Direct-Detection of Major Histocompatibility Complex Class-I Binding to Antigenic Peptides Using Surface-Plasmon Resonance - Peptide Immobilization and Characterization of Binding-Specificity. Journal of Biological Chemistry 268(21):15425-15434. Khilko, S.N., M.T. Jelonek, M. Corr, L.F. Boyd, A.L.M. Bothwell, andD.H. Margulies. 1995. Measuring Interactions of Mhc Class-I Molecules Using Surface-Plasmon Resonance. Journal of Immunological Methods 183(1):77-94. Oelke, M., M.V. Maus, D. Didiano, C.H. June, A. Mackensen, andJ.P. Schneck. 2003. Ex vivo induction and expansion of antigen-specific cytotoxic T cells by
112
HLA-Ig-coated artificial antigen-presenting cells. Nature Medicine 9(5):619624. Olsen, A.C., L.O. Pedersen, A.S. Hansen, M.H. Nissen, M. Olsen, P.R. Hansen, A. Holm, andS. Buus. 1994. A Quantitative Assay to Measure the Interaction between Immunogenic Peptides and Purified Class-I Major Histocompatibility Complex-Molecules. Eur. J. Immunol. 24(2):385-392. Oukka, M., J.C. Manuguerra, N. Livaditis, S. Tourdot, N. Riche, I. Vergnon, P. Cordopatis, andK. Kosmatopoulos. 1996. Protection against lethal viral infection by vaccination with nonimmunodominant peptides. Journal of Immunology 157(7):3039-3045. Rabinowitz, J.D., M.N. Liang, K. Tate, C. Lee, C. Beeson, andH.M. McConnell. 1997. Specific T cell recognition of kinetic isomers in the binding of peptide to class II major histocompatibility complex. Proceedings of the National Academy of Sciences of the United States of America 94(16):8702-8707. Schutz, C., M. Fleck, A. Mackensen, A. Zoso, D. Halbritter, J.P. Schneck, andM. Oelke. 2008. Killer artificial antigen-presenting cells: a novel strategy to delete specific T cells. Blood 111(7):3546-3552. Sette, A., J. Sidney, M.F. Delguercio, S. Southwood, J. Ruppert, C. Dahlberg, H.M. Grey, andR.T. Kubo. 1994. Peptide Binding to the Most Frequent Hla-a ClassI Alleles Measured by Quantitative Molecular-Binding Assays. Molecular Immunology 31(11):813-822. Tsai, C.J., S. Kumar, B.Y. Ma, andR. Nussinov. 1999. Folding funnels, binding funnels, and protein function. Protein Science 8(6):1181-1190. vanderBurg, S.H., E. Ras, J.W. Drijfhout, W.E. Benckhuijsen, A.J.A. Bremers, C.J.M. Melief, andW.M. Kast. 1995. An HLA class I peptide-binding assay based on competition for binding to class I molecules on intact human B cells Identification of conserved HIV-1 polymerase peptides binding to HLAA*0301. Human Immunology 44(4):189-198.
113
CHAPTER 5 CELLULAR MICROARRAY EVALUATION OF HUMAN CTL RESPONSE FOR INFLUENZA-A VIRUS
5.1. Introduction Studying the breadth of CTL immune response is of fundamental and medicinal importance. CTL immune responses for some viral pathogens appear to be restricted to only one or a few epitopes out of the large number of potential epitopes contained in complex viral antigenic proteins. This phenomenon, known as immunodominance, is an overall consequence of a variety of contributing factors, including antigen processing and transport, MHC binding affinity, TCR recognition, immunodomination and cross competition (Chen et al., 2000; Kastenmuller et al., 2007; Yewdell and Bennink, 1999; Yewdell, 2006). With the development of more sensitive assays for detecting the presence of CTLs specific for potential epitopes, CTL responses to some viral infections that were thought to be strictly immunodominant such as Influenza (flu) A virus, have been found to be broad and multi-specific (Assarsson et al., 2008; Bednarek et al., 1991; Gianfrani et al., 2000). 114
The discovered subdominant epitopes in turn have shown growing importance in the protection from viral infections (Oukka et al., 1996) and been incorporated in designing peptide-based vaccines (Assarsson et al., 2008; Olsen et al., 2000).
In Chapter 3, we have described a novel anti-Ig,λ-based HLA A2-Ig cellular microarray technology that outperforms traditional spotted dimer/tetramer microarrays. In this technology, pMHC dimer complexes are incubated with cells in sample solutions; an anti-Ig,λ antibody specific to the Ig portion of the dimer is printed, which interacts with dimer molecules bound on T cell surfaces and thus captures cells specific to the peptide-HLA A2-Ig complex onto the spots. This technology maximally preserves the structural integrity and allows complete orientational freedom of the complex when interacting with TCRs on cell surface, thus ensuring a high detection sensitivity equivalent to that of FACS, 0.01%. Incorporation of internal calibration spots such as anti-CD3 spots achieved simple and relatively accurate estimation of the frequency of detected antigen-specific CTLs.
In this chapter, we demonstrate the feasibility of this cellular microarray technology in evaluating the breadth of human immune response, by screening six healthy adult blood donors for CTLs specific for Influenza A virus-associated epitopes. As proof of concept, six previously identified Influenza A-associated epitopes were selected: M1.58, PA.46, PA.225, PB1.413, NA.75 and NA.231 (Bednarek et al., 1991; Gianfrani et al., 2000; Kasprowicz et al., 2008). Peripheral blood mononuclear cells (PBMCs), separated from peripheral blood, were directly applied to the microarrays and frequencies of detected antigen-specific CTLs were estimated using anti-CD8
115
calibration spots. CTLs specific to M1.58 were detected in all six donors, whereas for all other five epitopes at least one donor was found to be positive, showing that human immune response to Influenza A virus is indeed broad and multi-specific. Quantitative comparison of the frequencies estimated in cellular microarray with those measured in FACS showed good agreement between the two assays.
5.2. Materials and Methods Peptides and preparation of peptide-HLA A2-Ig complexes. High purity (~95%) peptides CMVpp65 (NLVPMVATV), M1.58 (GILGFVFTL), NA.75 (SLCPIRGWAI), PA.225 (SLENFRAYV), PA.46 (FMYSDFHFI), NA.231 (CVNGSCFTV) and PB1.413 (NMLSTVLGV) were purchased from Genscript (Piscataway NJ), and reconstituted to 2 mg/ml in PBS solution containing 10% (v/v) DMSO (except for M1.58 which was in 20% DMSO/PBS solution). Peptide-dimer complexes were prepared by the method of passively loading (Schneck et al., 2005) with minor changes. Briefly, unloaded HLA A2-Ig dimer (0.5 mg/ml, BD Pharmingen, San Diego CA) was incubated with specific peptide of interest in 160-fold molar excess at 24ºC (room temperature) for 5 days. No separation was conducted after incubation.
Donor peripheral blood mononuclear cells (PBMCs). This study was approved by the Institutional Ethics Committee. All donors gave written informed consent before enrolling in the study. Healthy donors were HLA A2.1-phenotyped by fluorescencelabeled anti-HLA A2 antibody using flow cytometry.
116
Blood, typically 12 ml, was drawn from HLA A2 donors and PBMCs were isolated from the collected blood by standard Ficoll separation. Briefly, tubes containing the blood were centrifuged at 1700 rpm for 10 min. The resulting bottom portion, cellular components of the blood, was transferred to a 50 ml Falcon tube and rinsed with 1x PBS solution to a volume of 35ml. Then, about 13ml Ficoll-Pague solution (GE healthcare) was carefully underlaid to the bottom of the tube and the tube was then centrifuged at 2000 rpm for 30 min with brake off. The buffy layer that appeared in the middle of the tube was carefully collected and washed with 40-45 ml PBS by centrifuging at 1700 rpm for 10min. After removing the supernatant, cell pellet was suspended in 14ml erythrocyte lysis buffer (155 mM NH4Cl, 10 mM KHCO3, 0.1 mM EDTA (pH 7.4); all chemicals from Sigma-Aldrich) and incubated on a shaker for 5 min, and then rinsed with PBS to a total volume of 50ml and centrifuged at 1200 rpm for 10 min. Discard the supernatant and the cells are ready for CFSE labeling for cellular microarray assay or dimer staining for FACS analysis. Alternatively, cryopreserved PBMCs (from donor 2F) that were characterized by IFNγ ELISpot assay were directly purchased (Cellular Technology Ltd., Shaker Hts, OH). Cells were thawed according to the provider’s protocol before microarray or FACS analysis.
Cellular microarray assays. Refer to “Materials and Methods” section of Chapter 3 for the general procedures of fabricating and processing the cellular microarrays. Instead of anti-CD3, anti-human CD8 antibody (0.5mg/ml, BD Biosciences) spots were employed as calibration spots for estimating the frequency of captured cells on anti-Ig,λ spots in total CD8 cells in the cell suspension. The solutions were printed in
117
pre-programmed patterns of 5 x 5 subarrays of spots separated by a center-to-center distance of 1 mm. The individual spots consisted of ~ 10 nanoliters of solution, and were approximately 500 µm in diameter. For each incubation with unloaded dimer or peptide-loaded dimer, a number of 400,000 PBMCs was employed, to ensure ~ 60,000 CD8 cells in the cell suspension (considering around 15% of CD8 cells in PBMCs in healthy donors).
Flow-activated cell sorting (FACS) measurements. Dimer staining using FACS was conducted in parallel with cellular microarray assays to measure the frequency of antigen-specific CTLs in isolated donor PBMCs. Refer to the “Materials and Methods” of Chapter 3 for the staining procedure. Considering the significant lower target cell frequency in PBMCs, the starting sample size for dimer staining was 106 cells.
Fluorescence polarization (FP) competitive binding assays. FP competitive binding assays were conducted by Nicole Guzman. FP measurements were obtained from a SpectraMax M5 reader (Molecular Devices, Sunnyvale CA). When set for fluorescence polarization, the reader detects changes in the polarization of emitted light due to the rotational speed of molecules. Small fluorescently labeled ligands rotate faster than much larger receptors. When these ligands associate with receptors, their rotational speed is reduced. The SpectraMax reader detects these changes in polarization and measures real time binding events in solution.
118
Peptide binding to HLA A2-Ig dimer was measured in competitive assays, in which HLA A2-Ig dimer was incubated with a constant concentration of FITC-labeled reference peptide (ALMDKVLKV) in the presence of titrated concentrations of an unlabeled flu peptide. Briefly, the assays were carried out at room temperature for an incubation period of 3 days. Based on previous kinetic studies (data not shown), this 3 day incubation period ensures that equilibrium has been reached between all components of the mixture. Each reaction mixture included HLA A2-Ig dimer at a 2x concentration of 200 nM; the stabilizing β2-microglobulin protein (Fitzgerald Industries, Acton MA) at an 8x concentration of 17,000 nM; the reference peptide, a fluorescein labeled design peptide (ALMDKVLKV, Genescript), at an 8x concentration of 16 nM; and the unlabeled competitor peptide, at one of the 13 different serially diluted 4x concentrations used (400,000 nM-0.4 nM).
Both labeled and unlabeled peptides were added to the reaction mixture before the HLA A2-Ig fusion protein, ensuring that both peptides were presented simultaneously to the receptor. To avoid peptide photobleaching, the reaction mixture was incubated within aluminum foil covered microcentrifuge tubes. After the 3 day incubation period, 20µl of the reaction mixture was loaded into individual wells of a black, nonbinding surface (NBS) 384-well plate (Corning). Three control groups were included: buffer only (blank), protein only and pFITC only. Each experiment was performed in quadruplicate and the mean and standard deviation of these 4 measurements were plotted. IC50 value, which stands for the concentration of the flu peptide yielding 50% inhibition of the binding of the FITC-labeled reference peptide to the dimer, was obtained by fitting the mean of 13 data points to a 4 parameter logistic equation:
119
y = min +
(max − min) 1 + 10 (log IC 50− x )
B
Where, max, indicates the upper plateau of the curve; min, the lower plateau and B, is the slope factor which describes the steepness of the curve transition.
Image acquisition and data analysis. Microarray images were acquired using a ProScanArray fluorescence scanner (PerkinElmer, Waltham MA). The slides were scanned at 5 µm resolution. Image quantification and data analysis were performed using ScanArray Express 3.0 (PerkinElmer, Waltham MA), ImageJ 1.42v (National Institutes of Health, Bethesda MD) and JMP 6.0 (SAS, Cary NC) software. Graph plotting and fluorescence polarization data fitting were performed using Sigma plot 10.0 (Systat Software, San Jose CA). In order to the frequencies of antigen-specific CTLs in CD8 cells, for FACS assays, measured percentage of antigen-specific CTLs in lymphocytes was divided by measured percentage of CD8 cells in lymphocytes; for ELISpot assay, measured percentage of antigen-specific CTLs in PBMCs was divided by the percentage of CD8 cells in PBMCs, which can be obtained from the measured percentages of lymphocytes in PBMCs and CD8 cells in lymphocytes in FACS.
5.3. Results and Discussion 5.3.1. Fluorescence polarization measurement of peptide binding affinity to HLA A2-Ig dimer Figure 5.1 shows the competitive binding curves of the six flu-associated peptides to HLA A2-Ig dimer measured fluorescence polarization assays. As shown in the legend,
120
the IC50 values of the flu peptides span over a wide range, from 312 nM for PB1.413 to 15,313 nM for NA.231. For every peptide, IC50 value measured in this study is much higher than previously reported (Gianfrani et al., 2000) (except for NA.231 that was not measured). Both experimental observations (Buchli et al., 2005) and theoretical analyses (Cheng and Prusoff, 1973) show that IC50 is not an intrinsic property of the binding reaction but increases with MHC concentration. The much higher IC50 values obtained in this study can stem from the 20-fold higher MHC concentration used in this study than in the previous study. Besides the magnitude of IC50, the rank order of the peptides in terms of IC50 values also differs from previous study (Gianfrani et al., 2000). For example, PA.46 was found to have a higher IC50 than PB1.413, M1.58, and PA.225 in the present study, but in (Gianfrani et al., 2000) the IC50 value of PA.46 was lower than all other three peptides, in fact the lowest among all the 126 peptides tested. This variation in IC50 rank order can be a result of the different assay types employed: polarization measurement of fluorescence labeling in the present study versus radio labeling and counting in (Gianfrani et al., 2000). Systematic comparisons of IC50 values measured for the same peptides in these two assay methods suggest a linear correlation between the logarithmic values of IC50 obtained in these two independent assays; however, there are a number of measurements that significantly deviated from the fitted linear line (Buchli et al., 2005). In addition, MHC molecules employed in these two studies are different: HLA-Ig dimer in the present study versus HLA monomer in (Gianfrani et al., 2000). Though differences in the intrinsic binding affinity of a given epitope to the two types of HLA molecules are not expected, possible variations in the purities of peptides and HLA molecules and their solubilization conditions in the assays can influence the
121
binding of peptides to HLA molecules, causing the non-linear correlation between the measurements.
Peptides with IC50 values measured in FP assays below 5,000 nM are considered to be high-affinity binding peptides and 5,000-50,000 nM IC50 values were considered to be medium-affinity binding peptides (Buchli et al., 2005). According to this classification, NA.231 is a medium-affinity binding peptide and all other five peptides were high-affinity binding peptides.
5.3.2. Cellular microarray assay for detection of antigen-specific CTLs from PBMCs The capability of the cellular microarray assay in detecting antigen-specific CTLs directly from donor PBMCs was examined using commercial PBMCs that were precharacterized by IFNγ ELISpot assays to be positive for CMVpp65, M1.58 and PA.46.
Figure 5.2a shows representative fluorescence-scanned images of anti-Ig,λ subarrays, each containing 5 x 5 spot replicates, after incubation with CFSE-labeled PBMCs that were pre-incubated respectively with solutions of the three peptide-dimer complexes and unloaded dimer. From the images, anti-CD8 spots were densely and uniformly covered with cells, indicating the abundance of CD8 cells in the PBMC suspension. Cells pre-incubated with unloaded or peptide-loaded dimers showed very low, yet differing binding levels on anti-Ig,λ spots: highest binding was observed for cells preincubated with CMVpp65-dimer, followed by cells pre-incubated with PA.46-dimer and cells pre-incubated with M1.58-dimer, with minimal binding level observed for cells pre-incubated with unloaded dimer. As discussed earlier in Chapter 3, this
122
nonspecific binding to unloaded dimer can be attributed to some TCRs on CD8 cell surface interacting with the endogenous peptide-MHC complex part of unloaded dimer, to a surface density level that cells are captured onto the spots. Considering that PBMCs consist of more heterogeneous cell types besides CD8 T cells, another possible source of nonspecific binding to the spots can be the cross-reactivity between the printed goat anti-Ig,λ antibodies and Fc receptors expressed on various cell types such as B cells and NK cells (Charles A. Laneway et al., 2005).
Figure 5.2b shows FACS plots of dimer staining that was conducted in parallel with cellular microarray assays. The percentage shown in each plot stands for the frequency of pMHC-specific CD8 cells (double positive population) in total lymphocytes: the frequency of cells binding to unloaded dimer was estimated to be 0.04%, and the frequencies of cells binding to dimer loaded respectively with CMVpp65, M1.58, and PA.46 were measured to be 0.87%, 0.05%, and 0.04%. Subtracting with the frequency of cells stained with unloaded dimer determines the frequency of cells binding in an antigen-specific manner: 0.83% for CMVpp65specific CTLs and 0.01% for M1.58-specific CTLs, indicating the presence of CTLs specific to the two peptides. However, the same frequency of cells binding to unloaded dimer as to PA.46-loaded dimer suggests a non-detectable level of PA.46specific CTLs by FACS.
Figure 5.2c quantitatively compares the frequencies of CTLs respectively specific to CMVpp65, M1.58, and PA.46 in total CD8 cells measured in the three assays: cellular microarray, FACS, and ELISpot. The level of unloaded dimer in each assay was
123
subtracted before plotting. While FACS detected the presence of CTLs specific to CMVpp65 and M1.58, both cellular microarray and ELISpot detected CTLs specific to all three peptides above the background of nonspecific binding to unloaded dimer. Notably, the frequencies of antigen-specific CTLs estimated in cellular microarray were higher than the other two assays; especially, the detected frequency of 1.5% PA.46-specific CTLs is unusually high for a well-recognized subdominant epitope. As discussed in Chapter 3, this is evidently due to the irreversibility of cell capture on the microarray and the efficiency of antigen-specific CTLs coming in contact with the anti-Ig,λ spots, which leads to an over-representation of antigen-specific CTLs in the population. Specifically, anti-CD8 calibration spots can only accommodate certain number of CD8 cells, which comprise only a portion of the total CD8 cells in cell suspensions; meanwhile, anti-Ig,λ spots can potentially capture all the antigen-specific CTLs present in the cell suspension. As a result, the frequency of antigen-specific CTLs, represented by comparing the cell binding levels on the two types of spots, is higher than the frequency in suspension. Considering that a typical 5 x 5 spot subarray captures approximately 1/6 of total CD8 cells in the cell suspension, the frequency of antigen-specific CTLs determined in cellular microarray assay is over-represented by 6-fold. Thus, the frequency of 1.5% for PA.46-specific CTLs could be only 0.25%, which is close to the 0.19% determined in ELISpot assay.
5.3.3. Evaluation of the breadth of CTL specificities to influenza Aassociated epitopes Having shown the capability of the cellular microarray in detecting the presence and estimating the frequency of antigen-specific CTLs directly from PBMCs, we applied
124
this assay for profiling the antigen specificities of human CTLs to the six fluassociated peptides and CMVpp65 across multiple HLA-A2 donors. Table 5.1 summarizes frequencies of antigen-specific CTLs detected in cellular microarray assay and FACS. For both assays, nonspecific binding to unloaded dimer was subtracted beforehand; for cellular microarray assay, measured frequencies were further divided by an over-representation factor of 6, to indicate the frequencies of antigen-specific CTLs in total CD8 cell populations in suspensions. Across all measurements, the two assays differed in 7 pairs: the microarray assay did not detect the presence of PA.225-specific CTLs in donor CD27 and PB1.413-specific CTLs in donor CE63, whereas FACS assay did not detect NA.231- and PB1.413-specific CTLs in CG2, NA.75- and PA.46-specific CTLs in donor AS, as well as PA.46specific CTLs in donor 2F. Since the estimated 0.007% PB1.413-specific CTLs in donor CG2 was below the established lower detection limit of 0.01% (Chapter 3), we can not consider the presence of PB1.413-specific CTLs in CG2 to be positive. Therefore, microarray and FACS assays differed in only 6 out of the total 32 measurements, in more than 80% agreement with each other. These “mis-detected” peptides are subdominant peptides to which antigen-specific CTLs, if present, are usually at very low frequencies. To resolve the discrepancy between the two assays, in the future a third, more sensitive assay such as ELISpot can be performed. For each donor, as summarized in Table 5.1, antigen-specific CTLs were detected for one to as many as five flu-derived peptides, indicating that the TCR repertoire of an individual specific to Influenza A is broad and multi-specific.
125
Figure 5.3 quantitatively compares the frequency of antigen-specific CTLs out of total CD8 cells measured in cellular microarray assay versus FACS. Besides data points that correspond to peptides determined to be negative for donors in one or both of the assays, 6 measurements fell on the upper side of the y=x line and 7 measurements fell below the line. This almost equal number of data points on each side of the y=x line indicates that compared with FACS, cellular microarray assay can estimate the frequency of antigen-specific CTLs without any apparent bias. A further statistical comparison revealed the correlation coefficient to be 0.73, showing a good agreement between the two assays.
Table 5.2 summarizes the number of positive donors over the number of tested donors for each peptide measured in cellular microarray and FACS assays as well as in the previous study by 51Cr-release assays (Gianfrani et al., 2000). CTLs specific for CMVpp65, known as the immunodominant epitope of CMV, were detected in two out of four donors tested, suggesting that the two donors negative for it had never been infected with CMV by the time the blood samples were collected. M1.58 was positive in all six tested donors, confirming its immunodominant role in influenza A infection. For all other five flu-associated peptides, one to four donors were found to be positive in cellular microarray assays, confirming that these peptides are indeed immunogenic and subdominant. Notably, in the previous study only one or two out of six donors were found positive for these subdominant epitopes (Gianfrani et al., 2000), giving rise to constantly smaller percentages of positive donors (out of total tested donors) than observed in this cellular microarray assay. Though the small size of the donor population may not be statistically representative, this difference could indicate that
126
the cellular microarray assay is a more sensitive technique for detecting antigenspecific CTLs than traditional 51Cr-release assay, which heavily depends on the efficient proliferation of antigen-specific cells under proper stimulation and culture conditions.
5.4. Conclusions In this chapter, we for the first time demonstrated the application of the anti-Ig,λ based HLA A2-Ig cellular microarray assay to the evaluation of the breadth of human CD8 CTL immune response. As proof-of-concept work, the frequencies of CTLs specific to six Influenza A-associated epitopes, including both immunodominat and subdominant epitopes, were measured in the cellular microarray assay. The results showed that the frequency of CTLs specific for each peptide varied from individual to individual, and confirmed the role of M1.58 as the immunodominant epitope and the other five peptides as subdominant epitopes.
The frequencies of antigen-specific CTLs detected in cellular microarray assays showed good correlation with measurements by standard FACS dimer staining assay. In addition, compared with FACS, reagent costs (ng of antibody versus µg in FACS) and cell sample sizes (a third of that in FACS) are significantly reduced in cellular microarray. Compared with previously developed pMHC cellular microarray technologies, since peptide-loaded dimer is not directly printed in this assay (Chen et al., 2005; Kwong et al., 2009), any likely variation in functionality across different peptide-dimer complexes due to fast changes in peptide solution conditions upon
127
printing is minimized, making the difference in cell binding truly reflecting the variation across inherent frequencies of antigen-specific CTLs in cell suspension. In addition, the most complex sample studied in previous pMHC tetramer cellular microarrays was pure CD8 cells that were separated and enriched by magnetic beads from donor peripheral blood, which is time and reagent costly, against the needs for high throughput analyses. This work is the first to demonstrate the sensitive detection of antigen-specific CTLs directly from the complex population of donor PBMCs, which significantly simplifies the sample preparation process.
With the increasing recognition of the importance of subdominant epitopes, this cellular microarray assay will for sure find prompt applications in the characterization of immune responses to medically-relevant agents and in the design of universal epitope-based vaccines.
128
Figure 5.1 HLA A2-Ig dimer competition curves obtained by incubating a constant concentration of FITC labeled peptide (ALMDKVLKV) with titrated amounts of unlabeled flu peptides. After equilibrium was reached, fluorescence polarization was measured and IC50 values were obtained by fitting the data to a four parameter logistic equation.
129
a Unloaded dimer
CMVpp65-dimer
PA.46-dimer
M1.58-dimer
Anti-CD8
Continued Figure 5.2 Frequency of antigen-specific CTLs measured in different assays. (a) Representative subarray images on HLA A2-Ig cellular microarray. CFSE-labeled PBMCs from donor 2F were incubated separately with unloaded dimer, or dimer loaded with peptides CMVpp65, M1.58, and PA.46 and then applied onto anti-Ig,λ spots, or directly applied onto anti-CD8 spots as calibration. (b) Representative FACS density plots using dimer staining. Shown plots were gated on lymphocytes and frequencies were out of total lymphocytes. (c) Comparison of the three assays, cellular microarray, FACS, and ELISpot, in measuring the frequency of antigen-specific CTLs out of total CD8 lymphocytes. For both microarray and FACS, unloaded dimer was subtracted before plotting. Error bars represent standard errors across 5 x 5 spot replicates (cellular microarray assay) or triplicate wells (ELISpot assay).
130
Figure 5.2 continued
b Unloaded dimer 0.04%
CMVpp65dimer
M1.58-dimer 0.05%
0.87%
c
131
PA.46-dimer 0.04%
Table 5.1 Comparison of assays on detection of antigen-specific CTLs
Donor
Assay
Frequency of antigen-specific CTLs (out of total CD8 cells)
No. positive peptides / No. Tested
CMVpp65
M1.58
NA.75
PA.225
PA.46
NA.231
PB1.413
CD27
Microarray FACS
4/6 5/6
0.11% 0.14%
0.50% 0.43%
0.05% 0.08%
NEG 0.14%
0.43% 0.15%
NT NT
NEG NEG
CE63
Microarray FACS
1/7 2/7
NEG NEG
0.58% 0.05%
NEG NEG
NEG NEG
NEG NEG
NEG NEG
NEG 0.04%
CG2
Microarray FACS
3/7 1/7
NEG NEG
0.30% 0.30%
NEG NEG
NEG NEG
NEG NEG
0.05% NEG
0.007% NEG
CF
Microarray FACS
5/5 4/4
NT 0.29%
0.14% 0.35%
0.05% 0.23%
0.27% NT
0.38% 0.18%
NT NT
0.08% NT
AS
Microarray FACS
5/5 3/5
NT NT
0.28% 0.35%
0.13% NEG
0.18% 0.03%
0.15% NEG
NT NT
0.42% 0.09%
2F
Microarray FACS ELISPOT
3/3 2/3 3/3
1.0% 3.7% 2.4%
0.14% 0.04% 0.28%
NT NT NT
NT NT NT
0.25% NEG 0.19%
NT NT NT
NT NT NT
NT: not tested; NEG: negative;
132
R=0.73
Figure 5.3 Comparison of antigen-specific CTL frequencies measured in cellular microarray and in FACS. For clarification purpose, the frequencies were plotted in logarithmic scale. R, Pearson correlation coefficient, was obtained from JMP analysis. Dashed line represents y = x. For transforming to logarithmic scale all zero frequencies (negative) were converted to 0.001 for plotting.
Table 5.2 Peptide specificity (No. positive donors / No. tested) CMVpp65
PB1.413
M1.58
PA.225
PA.46
NA.75
NA.231
Microarray
2/4
3/5
6/6
2/5
4/6
3/5
1/2
FACS Gianfrani et al. (Gianfrani et al., 2000)
3/5
3/4
6/6
2/4
2/6
2/5
0/2
NT
2/6
8/8
2/6
2/6
1/6
NT
133
References Assarsson, E., H.H. Bui, J. Sidney, Q. Zhang, J. Glenn, C. Oseroff, I.N. Mbawuike, J. Alexander, M.J. Newman, H. Grey, andA. Sette. 2008. Immunomic Analysis of the Repertoire of T-Cell Specificities for Influenza A Virus in Humans. Journal of Virology 82(24):12241-12251. Bednarek, M.A., S.Y. Sauma, M.C. Gammon, G. Porter, S. Tamhankar, A.R. Williamson, andH.J. Zweerink. 1991. The Minimum Peptide Epitope from the Influenza-Virus Matrix Protein - Extra and Intracellular Loading of Hla-a2. Journal of Immunology 147(12):4047-4053. Buchli, R., R.S. VanGundy, H.D. Hickman-Miller, C.F. Giberson, W. Bardet, andW.H. Hildebrand. 2005. Development and validation of a fluorescence polarization-based competitive peptide-binding assay for HLA-A*0201 - A new tool for epitope discovery. Biochemistry 44(37):12491-12507. Charles A. Laneway, J., P. Travers, M. Walport, andM.J. Shlomchik. 2005. Immunobiology-the immune system in health and disease. Garland Science. Chen, D.S., Y. Soen, T.B. Stuge, P.P. Lee, J.S. Weber, P.O. Brown, andM.M. Davis. 2005. Marked differences in human melanoma antigen-specific T cell responsiveness after vaccination using a functional microarray. Plos medicine 2(10):1018-1030. Chen, W.S., L.C. Anton, J.R. Bennink, andJ.W. Yewdell. 2000. Dissecting the multifactorial causes of immunodominance in class I-restricted T cell responses to viruses. Immunity 12(1):83-93. Cheng, Y., andW.H. Prusoff. 1973. Relationship between Inhibition Constant (K1) and Concentration of Inhibitor Which Causes 50 Per Cent Inhibition (I50) of an Enzymatic-Reaction. Biochemical Pharmacology 22(23):3099-3108. Gianfrani, C., C. Oseroff, J. Sidney, R.W. Chesnut, andA. Sette. 2000. Human memory CTL response specific for influenza A virus is broad and multispecific. Human Immunology 61(5):438-452. Kasprowicz, V., S.M. Ward, A. Turner, A. Grammatikos, B.E. Nolan, L. LewisXimenez, C. Sharp, J. Woodruff, V.M. Fleming, S. Sims, B.D. Walker, A.K. Sewell, G.M. Lauer, andP. Klenerman. 2008. Defining the directionality and
134
quality of influenza virus-specific CD8(+) T cell cross-reactivity in individuals infected with hepatitis C virus. Journal of Clinical Investigation 118(3):11431153. Kastenmuller, W., G. Gasteiger, J.H. Gronau, R. Baier, R. Ljapoci, D.H. Busch, andI. Drexler. 2007. Cross-competition of CD8(+) T cells shapes the immunodominance hierarchy during boost vaccination. Journal of Experimental Medicine 204(9):2187-2198. Kwong, G.A., C.G. Radu, K. Hwang, C.J.Y. Shu, C. Ma, R.C. Koya, B. CominAnduix, S.R. Hadrup, R.C. Bailey, O.N. Witte, T.N. Schumacher, A. Ribas, andJ.R. Heath. 2009. Modular Nucleic Acid Assembled p/MHC Microarrays for Multiplexed Sorting of Antigen-Specific T Cells. J. Am. Chem. Soc. 131(28):9695-9703. Olsen, A.W., P.R. Hansen, A. Holm, andP. Andersen. 2000. Efficient protection against Mycobacterium tuberculosis by vaccination with a single subdominant epitope from the ESAT-6 antigen. Eur. J. Immunol. 30(6):1724-1732. Oukka, M., J.C. Manuguerra, N. Livaditis, S. Tourdot, N. Riche, I. Vergnon, P. Cordopatis, andK. Kosmatopoulos. 1996. Protection against lethal viral infection by vaccination with nonimmunodominant peptides. Journal of Immunology 157(7):3039-3045. Schneck, J.P., J.E. Slansky, S.M. O'Herrin, andT.F. Greten. 2005. Monitoring antigen-specific T cells using MHC-Ig dimers. In Short protocols in immunology: a compendium of methods from current protocols in immunology. Coligan JE, Bierer BE, Margulies DH, Sherach EM, Strober N, editors. Wiley. 15-12 - 15-11. Yewdell, J.T.W., andJ.R. Bennink. 1999. Immunodominance in major histocompatibility complex class I-restricted T lymphocyte responses. Annual Review of Immunology 17:51-88. Yewdell, J.W. 2006. Confronting complexity: Real-world immunodominance in antiviral CD8(+) T cell responses. Immunity 25(4):533-543.
135
CHAPTER 6 COMBINATORIAL ANTIBODY MICROARRAY FOR PERSONALIZED FORMULATION OF IMMUNOLIPOSOMES
6.1. Introduction Cancer patients treated with chemotherapy using common anti-cancer drugs, such as doxorubicin, often suffer from dose-limiting toxicities due to the lack of drug selectivity to only malignant cells. Potential therapeutic application of gene-silencing agents, such as antisense oligonucleotides (ODNs), also requires effective and selective delivery of ODNs to cancer cell nuclei. To minimize the side effects to nonmalignant cells, these anti-cancer agents can be incorporated into liposomes (such as FDA-approved DaunoXome and Doxil®), which could be preferentially uptaken by solid tumors taking advantage of the enhanced vascular permeability and impaired lymphatic drainage in growing tumors. To further enhance site-specific delivery of the drugs, ligands, most commonly antibodies that are specific for antigens uniquely expressed or over-expressed on target cells have been conjugated to the surface of
136
liposomes. These antibody-targeted, drug-encapsulated liposomes, referred to as immunoliposomes (ILPs), specifically bind to surface antigens on target cells, whereby achieve enhanced and selective drug delivery to cancer cells and therefore higher therapeutic effects (Bendas, 2001; Mastrobattista et al., 1999).
ILPs have been largely limited to “single-targeted”, i.e., having only one antibody population on liposomal surfaces. Recently, in vitro studies on B-cell lymphoma show that targeting both CD19 and CD20 using dual-targeted, doxorubicin-carrying ILPs that have equal amount of both anti-CD19 and anti-CD20 on the same ILPs, or mixtures of single-targeted ILPs (anti-CD19 ILPs and anti-CD20 ILPs) achieved higher binding, uptake, and cell cytotoxicity than using any single-targeted ILPs (Laginha et al., 2005). In another study, paramagnetic ILPs, when conjugated with two ligands specific for two angiogenesis markers, achieved significantly elevated binding, uptake, and therapeutic effects to angiogenesis than either single-targeted ILPs or mixtures of both (Kluza et al., 2010). The advantages of dual-targeting seem to stem from the synergy among the heteromultivant ligands on the liposome surface, the underlying mechanism of which, however, has not been fully understood. Despite the great promises of dual-targeted ILPs in personalized patient treatment that compositions of the ligands on ILPs are tailored to individual patient disease profile, and the development of approach for synthesizing ILPs containing multiple ligands (Ishida et al., 1999), there lacks a high throughput methodology for systematically interrogating the formulations of multiple-targeted ILPs in terms of the possible wide ranges of antibody types and compositions, and for quantitatively characterizing the synergistic effects for mechanistic elucidation.
137
In this work, we adapted the recently developed cellular microarray technology (Chen and Davis, 2006) as a novel combinatorial antibody-based cellular microarray technology to fulfill the purposes of screening for the optimal ILP formulation for individual patients. In this technology, microarray spots are composed of pure or combinations of antibodies that are against surface antigens of cancerous cells; cell binding on one spot, indicated by the spot intensity when cells are fluorescently labeled, provides information on targeting efficiency of the corresponding antibody or combination of antibodies on that spot. Since in cellular microarray technology antibodies are surface-immobilized as on ILPs, the level of cell binding to microarray spots and the level of ILP binding to cell surfaces are determined by the same factors: the expression level of cell surface antigens (assuming antibodies in excess), the percentage of cells expressing the antigen recognized by the printed antibody, and the affinity between the antibody and antigen. This sets the theoretic basis for this cellular microarray technology in studying ILP formulation for optimal cell targeting.
B-cell chronic lymphocytic leukemia (B-CLL), the most common type of leukemia, was chose for study; to show proof-of-concept, the targeting/binding to four surface antigens (CD19, CD20, CD37, and CD52) that had been extensively immunophenotyped (Belov et al., 2003; Belov et al., 2005; Belov et al., 2006) was interrogated using the combinatorial cellular microarray technology. Results showed that some combinations/mixtures of antibodies produced higher cell binding efficiency than any of the pure antibodies the mixtures are composed of, indicating the presence of synergetic effects among the antibodies on the microarray surface.
138
The finding of synergy as a common phenomenon existing in targeting cancerous cells promises this novel combinatorial microarray technology as a platform for high throughput and systematic screening of personalized antibody formulations of ILPs in cancer therapies. We further proposed a graphic method to represent binary synergies, from which synergies among higher order mixtures can be qualitatively predicted.
6.2. Materials and methods Cells and antibodies. B-CLL patients were defined and samples were processed according to (Zhao et al., 2007). Raji, Jurkat, 697 and MEC-1 cell lines were maintained in RPMI 1640 media (Invitrogen) supplemented with 10% heatinactivated fetal bovine serum (FBS: Hyclone Laboratories, Logan UT) at 37ºC in an atmosphere of 5% CO2. Anti-CD20 (Rituximab) and anti-CD52 (Alemtuzumab) were purchased from respectively Genentech (South San Francisco CA) and Ilex Pharmaceuticals (San Antonio TX). Anti-CD19, anti-CD37, and isotype controls were purchased from BD Biosciences (San Jose CA).
Combinatorial antibody microarrays. A library of 15 antibody mixtures was prepared by mixing the four antibodies at equal concentrations at all possible combinations: pure antibodies and mixtures of two antibodies, three antibodies and all four antibodies. Total antibody concentration was kept at 0.5mg/ml. Isotype controls were included as negative controls. A volume of 10µl of each sample was loaded into a 384-well plate (Bio-Rad Laboratories) for printing onto Nexterion® slide H using a non-contact piezoelectric arrayer (Perkin Elmer, Waltham MA). Bovine serum albumin (BSA, Sigma Aldrich, St Louis, MO) at 0.5 mg/ml (final concentration) was
139
added prior to printing. Each sample was arrayed in triplicate a spot center-to-center distance of 400 µm into a subarray of 6.8 mm x 6.8 mm, and each subarray was repeated in a 2x7 pattern on one single slide with subarray center-to-center distance of 9 mm. Individual spots contained one drop of 0.33 nl of printing solution and were approximately 150 µm in diameter. The printed slides can be stored sealed at 4ºC for more than a month before use without measurable loss of activity. Refer to Chapter 3 for the procedures of protein immobilization and slide blocking.
Cells, from Raji or Jurkat cell line, or from B-CLL patients, were labeled with CFSE (carboxy-fluorescein diacetate, succinimidyl ester; Invitrogen, Carlsbad CA) according to manufacturer’s protocol. Due to size difference, Raji and Jurkat cells were spun at 1200 rpm whereas patient cells were spun at 1800 rpm. The cells were suspended in PBS at a concentration of 1.5 x 106/ml and then contacted with the microarray surface by pipetting 50 µl of this suspension onto an area of 6.8 mm x 6.8 mm (to cover a 5 x 5 subarray of 500 µm spots or 10 x 10 subarray of 150 µm spots) separated by a silicone gasket (Grace Bio Labs, Bend OR). After incubation for 1 hr at room temperature (covered with aluminum foil to protect from light), the microarray slide was carefully dip-washed in PBS solution to remove unbound cells and then imaged as detailed below.
Image acquisition and data analysis. Microarray images were acquired by ProScanArray fluorescence scanner (Perkin Elmer, Waltham MA). Slides were scanned at 5µm resolution. Image quantification and analysis were performed by using ScanArray Express 3.0 (Perkin Elmer, Waltham MA) and JMP 6.0 (SAS, Cary
140
NC) software. Data plotting was performed by Sigma plot 10.0 (Systat Software, San Jose CA). Graphic representation of synergy was performed by Microsoft Visio 2000 SR1.
Binding synergy quantification. For each cell sample, from a cultured cell line or from a B-CLL patient, all spot intensities were normalized against its respective highest spot intensity. For each antibody combination, binding synergy among the surface antigens/printed antibodies was estimated by dividing the mean spot intensity of that combination over the average (when the mixture contained equal concentration of antibodies) of the spot intensities of associated single antibodies composing that mixture. Error propagation was calculated using standard formula, assuming the absence of covariance.
6.3. Results
6.3.1. Cell capture as a function of antibody and cell concentrations
The binding of Raji cells on the spots as a function of antibody solution concentration and sample size, i.e., the concentration of cell solution since a constant volume of cell solution is incubated on the microarrays was investigated (SP. Figure 6.1). For all four antibodies, cell binding increased with cell concentration, and then either reached a plateau or continued increasing within the range of cell concentration studied, depending on the concentration of the antibody. For anti-CD19, when printed at 0.05 mg/ml and 0.1 mg/ml, cell binding saturated at a concentration 5x106 /ml, indicating
141
that all antibody binding sites on the spots had been deprived by cells present in the solution; when printed at higher concentrations 0.3 mg/ml or 0.5 mg/ml, more antibody molecules were available on the spots and thus more cells were subsequently captured. For the other three antibodies, cell binding reached saturation at 5x106 /ml for the range of antibody concentrations investigated. For all four antibodies, spots printed with 0.5 mg/ml antibody solution produced the highest cell binding across the cell concentration range, thus, cell binding assay at this antibody solution concentration is the most sensitive. The mean spot intensities versus the cell concentration of up to 1.5 x 106 /ml at 0.5 mg/ml antibody concentration were further plotted (Figure 6.1). Clearly across all four antibodies, cell binding linearly increased with cell concentration within this cell concentration range (fits shown in the figure). To work within the linear range of mean spot intensity and cell concentration, and to work with the highest microarray sensitivity, all our microarray experiments were performed at a total antibody concentration of 0.5mg/ml and a cell density of 1.5x106/ml.
6.3.2. Personalized cell surface antigen expression profiles
Figure 6.2 shows the levels of cell binding, indicated by the mean spot intensities, on spots printed with antibodies specific for the four cell surface antigens for cells from Jurkat and Raji cell lines and four B-CLL patients. For Jurkat cells, no binding was observed on anti-CD19 or anti-CD37 spots, negligible binding to anti-CD20 spots, and apparent binding to anti-CD52 spots, indicating that only CD52 was expressed on Jurkat cells. For Raji cells, high but similar levels of cell binding were observed to
142
anti-CD19 and anti-CD20 spots, with lower level of binding to anti-CD37 and the lowest level of binding observed to anti-CD52 spots. Compared with flow-activated cell sorting (FACS) immunostaining conducted for Raji and Jurkat cells (plots not shown), relative cell binding levels across the four antibodies in microarray technology correlates well with the percentages of positively-stained cells with the antibodies in FACS. The binding profiles of cells from four B-CLL patients, as also shown in Figure 6.2, displayed significant variations from individual to individual. For example, for AS1, BO2, and MM3 the lowest cell binding was observed on antiCD20 spots; in contrast, for NN4 the lowest binding occurred on anti-CD52 spots and cell binding on anti-CD20 spots was instead the highest among all antibody spots. These cell binding differences on antibody spots among patients indicate the highly variable expression levels of surface antigens among individual patients, which can lead to the inefficiency if using single-targeted ILPs to treat all the patients. Thus, personalized antibody formula for the ILPs based on individual patient targeting profiles need to be established.
6.3.3. Combinatorial antibody microarrays for optimal targeting efficiency
Figure 6.3b shows the quantitative binding profiles of cells from five B-CLL patients on the combinatorial antibody microarrays (representative microarray images shown in Figure 6.3a). First, the levels of cell binding across the spots printed with single antibody solutions (anti-CD19, anti-CD20, anti-CD37, and anti-CD52) clearly varied from patient to patient, as has been observed from other patients in Figure 6.2. Second, for all five patients the highest binding occurred on spots printed with
143
mixtures of two or three antibodies, with the composition of the mixture varied from individual to individual: anti-CD19 & 52 for patient P1, anti-CD19 & 20 & 37 for both P2 and P4, anti-CD19 & 37 & 52 for P3, and anti-CD19 & 37 for P5. Compared with the highest cell binding level on any of the spots printed with single antibodies (anti-CD37 for P1, P2, P3 and P5, and anti-CD19 for P4), spots printed with these optimal formulae of antibodies increased cell binding by 40% to 80%, highlighting the significance of using multiple antibody ligands for optimal targeting of cell surface antigens and the great promises of this combinatorial antibody microarray as a high throughput screening tool.
Figure 6.3c further shows the corresponding binding synergies among the antibody ligands on microarray spots for the five B-CLL patients calculated based on Figure 6.3b. Strikingly, for the five patients studied most antibody combinations had a synergy value larger than 1, indicating that positive synergy among surfaceimmobilized antibodies is a prevalent phenomenon among B-CLL patients. There were also some combinations having synergy values below 1, such as anti-CD20 & 37 for patients P2 and P5, indicating that the interaction between the two antibodies had negative effects on cell surface targeting. Across the patients, there were remarkable differences in synergy profiles. First, the composition of the antibody mixture having the highest synergy varied, with the combination of anti-CD19 & 20 optimal for patient P1, anti-CD19 & 52 optimal for patient P2, anti-CD19 & 37 & 52 combination optimal for patient P3, anti-CD19 & 20 & 37 combination optimal for patient P4, and anti-CD19 & 52 optimal for patient P5, respectively. Second, the synergy values for patient P3 were generally low, only around 1.5 or below, and 7 out of the 11
144
combinations had synergy values lower than all other patients. These differences across patients probably have to do with the differential expression levels, i.e. densities of these antigens on patient cells, since steric hinderance at high antibody densities has been suggested to lower ILP binding efficiency (Laginha et al., 2005).
6.3.4. Possible graphic representation and prediction of synergies
To gain further insights into the underlying synergetic mechanisms, more homogeneous cells, obtained from three B-CLL cell lines, Raji, 697 and MEC-1, were employed on the combinatorial microarrays. Resulting mean spot intensities were normalized and binding synergies were summarized in Table 6.1. A graphic network for representing the complex synergies present among the antibodies was developed (Figure 6.4). Given the synergies between all binary antibody mixtures, a graph in which undirected edges presenting the pairwise synergy (positive, negative, or neutral) between the binary antibodies, denoted as nodes in the graph, can be constructed; based on this graph, we propose that the synergies among ternary and quaternary antibody mixtures can be qualitatively predicted. Take the cell line 697 as an example. Based on Table 6.1, the positive synergy between anti-CD19 and anti-CD20, between anti-CD19 and anti-CD52, between anti-CD19 and anti-CD37, and between antiCD37 and anti-CD52, and the negative synergy between anti-CD20 and anti-CD37 were represented in Figure 6.4 (middle). The synergy between anti-CD20 and antiCD52 was considered to be nonexistent since the range of 1.1 ± 0.20 is very close to 1. Thus in the graph the triangle formed by anti-CD19, anti-CD20 and anti-CD37 consists of two positive connecting edges and one negative connecting edge;
145
assuming additivity, the overall synergy among the three antibodies would be positive since the number of positive edges is larger than negative edges, consistent with the positive synergy value in Table 6.1. As another example, the triangle formed by antiCD20, anti-CD37 and anti-CD52 consists of one positive and one negative connecting edge; thus the overall synergy among the three antibodies is approximately neutral, which again agrees with the calculation in the table. In addition, the distinct graphs across different cell lines in Figure 6.4 indicate that the synergies among the antibodies are unique for each cell line, which can be attributed to the differential expression levels of antigens on cell surfaces. Taken together, we have established a straightforward graphic method that can uniquely represent the existing complex binary synergy network, and can qualitatively predict the synergies among higher order mixtures of the antibodies in the network.
6.4. Discussion and conclusions In this work, we report a novel combinatorial antibody microarray for interrogating antibody formulae for optimal cell binding of immunoliposomes (ILPs) in targeted immunotherapy of B-CLL. The existence of positive synergy in the context of multivalent interactions has been well recognized (Kluza et al., 2010); however, most current ILP targeting strategies use only one targeting antibody and to the best of our knowledge this has been the first report for studying the targeting strategy in a systematic and high throughput manner.
146
This technology offers unique advantages over FACS in the evaluation of targeting cell surface antigens by surface-immobilized antibodies. It allows for simultaneous/inparallel analyses of multiple cell samples (as many as 14 per slide currently, and can be readily increased with smaller-size gaskets) over thousands of spots; meanwhile, its in-parallel characteristic minimizes errors in analytical conditions across the samples. Whereas in FACS, the binding of each antibody (or antibody combination) to each cell sample is analyzed separately, which not only is reagent-costly and laborintensive, but also introduces human errors across the samples.
To show proof of concept, current work only covered antibodies at equal concentrations in the mixtures; with the high throughput characteristic of this technology, in the future a larger number of monoclonal antibodies at a variety of concentration ratios can be easily included. Ultimately, we want to build a microfluidic device that with proper design the composition of the antibody mixture continuously varies within the device. In addition, this strategy of combinatorial antibody microarray can be readily applied to studying the optimal formula and synergies of ILP targeting of other cell surface antigen-involved diseases.
It is important to note that future in vivo applications of multiple-targeted ILPs need to consider other factors besides the binding efficiencies, such as the internalization efficiency of the targeted receptors. For example, in vivo cytotoxic studies of ILPs targeted against the internalizing epitope CD19 showed greater therapeutic efficacy than ILPs targeted against the noninternalizing epitope CD20 (Sapra and Allen, 2002). It is possible to design antibody formulae that aim at both binding and internalization
147
to reach optimal ILP therapeutic efficacy, by including high-binding antibodies so that more drug-carrying ILPs can be bound to cell surfaces, and meanwhile the ILPs are subsequently uptaken by cells through antibodies against internalizing epitopes and drugs were subsequently released within the cells.
5000 Anti-CD19 Anti-CD20 Anti-CD37 Anti-CD52
Mean spot intensity
4000
3000
2000
1000
0 0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
6
Cell concentration (x 10 /ml)
Figure 6.1 Dependence of Raji cell binding on cell concentration at antibody solution concentration of 0.5 mg/ml. Cell binding on anti-CD19 spots was fitted by the solid line (y=189+1873*x, R2=0.8386), on anti-CD20 spots was fitted by the long dash line (y=59.96+2006*x, R2=0.9886), on anti-CD37 was fitted by the short dash line (y=-55.33+1574*x, R2=0.9036), and on anti-CD52 spots was fitted by the dotted line (y=90.58+208.9*x, R2=0.9225).
148
40000 Anti-CD19 Anti-CD20 Anti-CD37 Anti-CD52
Mean spot intensity
30000
20000
10000
0 Jurkat
Ragi
AS1
BO2
MM3
NN4
Figure 6.2 Surface antigen expression profiles of Jurkat cells, Raji cells, and four B-CLL patients: AS1, BO2, MM3, and NN4. Antibodies against the four surface antigens, CD19, CD20, CD37, and CD52 were printed at 0.5 mg/ml, and then contacted with the CFSElabeled cell populations at a concentration of 1.5x106 /ml. Error bars standard for the standard errors across 9 spot replicates.
149
a
P1
P2
P3
P4
P5 Continued
Figure 6.3 Cell binding on combinatorial antibody microarrays for five B-CLL patients: P1, P2, P3, P4 and P5. (a) Representative fluorescence-scanned microarray images. (b) Quantified mean spot intensities. (c) Quantified binding synergy across combinations of the antibodies for each patient. Dashed line in this figure stands for when there is no synergy (i.e. value equals 1).
150
Figure 6.3 continued
b P1
P2
50000
35000
30000
Mean spot intensity
Mean spot intensity
40000
30000
20000
25000
20000
15000
10000
10000 5000
0
0
2 7 0 0 9 7 2 7 2 2 2 7 2 2 2 D1 D2 D3 D5 & 2 & 3 & 5 & 3 & 5 & 5 & 3 & 5 & 5 & 5 & 5 ti-C ti-C ti-C ti-C 19 19 19 20 20 37 20 20 37 37 37 an an an an i-CD i-CD i-CD i-CD i-CD i-CD 9 & 9 & 9 & 0 & 0 & t t t t t t D1 D1 D1 D2 2 n n n n n n & a a a a a a i-C i-C i-C i-C 9 t t t t 1 an an an an i-CD t an
9 0 7 2 0 7 2 7 2 2 7 2 2 2 2 D1 D2 D3 D5 & 2 & 3 & 5 & 3 & 5 & 5 & 3 & 5 & 5 & 5 & 5 ti-C ti-C ti-C ti-C 19 19 19 20 20 37 20 20 37 37 37 an an an an i-CD i-CD i-CD i-CD i-CD i-CD 9 & 9 & 9 & 0 & 0 & t t nt t t t D1 D1 D1 D2 2 n n n n n & a a a a a a i-C i-C i-C i-C 9 t t t t 1 an an an an i-CD t an
P3
P4
30000
35000
30000
Mean spot intensity
Mean spot intensity
25000
20000
15000
10000
5000
25000
20000
15000
10000
5000
0
0
9 0 7 2 0 7 2 7 2 2 7 2 2 2 2 D1 D2 D3 D5 & 2 & 3 & 5 & 3 & 5 & 5 & 3 & 5 & 5 & 5 & 5 ti-C ti-C ti-C ti-C 19 19 19 20 20 37 20 20 37 37 37 an an an an i-CD i-CD i-CD i-CD i-CD i-CD 9 & 9 & 9 & 0 & 0 & t t nt t t t D1 D1 D1 D2 2 n n n n n a a a a a a i-C i-C i-C i-C 9 & t t t t 1 an an an an i-CD t an
9 0 7 2 0 7 2 7 2 2 7 2 2 2 2 D1 D2 D3 D5 & 2 & 3 & 5 & 3 & 5 & 5 & 3 & 5 & 5 & 5 & 5 ti-C ti-C ti-C ti-C 19 19 19 20 20 37 20 20 37 37 37 an an an an i-CD i-CD i-CD i-CD i-CD i-CD 9 & 9 & 9 & 0 & 0 & t t nt t t t D1 D1 D1 D2 2 n n n n n a a a a a a i-C i-C i-C i-C 9 & t t t t 1 an an an an i-CD t an
Continued
151
Figure 6.3 continued
P5 20000 18000
Mean spot intensity
16000 14000 12000 10000 8000 6000 4000 2000 0 9 0 7 2 0 7 2 7 2 2 7 2 2 2 2 D1 D2 D3 D5 & 2 & 3 & 5 & 3 & 5 & 5 & 3 & 5 & 5 & 5 & 5 ti-C ti-C ti-C ti-C 19 19 19 20 20 37 20 20 37 37 37 an an an an i-CD i-CD i-CD i-CD i-CD i-CD 9 & 9 & 9 & 0 & 0 & t t t t t t 1 1 1 2 2 an an an an an an i-CD i-CD i-CD i-CD 9 & t t t t 1 an an an an i-CD t an
c 4 P1 P2 P3 P4 P5
Binding synergy
3
2
1
0 2 2 2 2 7 7 2 2 7 2 0 & 2 9 & 3 9 & 5 0 & 3 0 & 5 7 & 5 0 & 3 0 & 5 7 & 5 7 & 5 37&5 1 3 1 2 2 19 3 3 2 2 & 0 D D D D D D & & & & 2 0 9 9 & ti-C nti-C nti-C nti-C nti-C nti-C D19 19 D2 D1 D1 a a a a a an ti-C nti-C nti-C nti-C ti-CD n a a a a an
152
Table 6.1 Normalized mean spot intensity and binding synergy for Raji, 697, and MEC-1
Raji
697
MEC-1
Samples
Normalized spot intensity
Binding synergy
Normalized spot intensity
0.88 ± 0.16
-
0.95 ± 0.031
-
anti-CD19
0.65 ± 0.085
-
0.59 ± 0.060
-
anti-CD20 anti-CD37
-
0.76 ± 0.051
-
0.22 ± 0.038
-
anti-CD52
0.57 ± 0.096 0.088 ± 0.019
-
0.34 ± 0.084
-
0.21 ± 0.038
-
anti-CD19 & 20
0.87 ± 0.18
1.1 ± 0.27
1.0 ± 0.037
1.3 ± 0.074
0.83 ± 0.092
1.2 ± 0.18
anti-CD19 & 37
0.69 ± 0.086
0.95 ± 0.17
0.95 ± 0.023
1.1 ± 0.047
0.79 ± 0.16
1.3 ± 0.27
anti-CD19 & 52
0.75 ± 0.12
1.5 ± 0.36
0.92 ± 0.029
1.4 ± 0.11
0.75 ± 0.058
1.2 ± 0.12
anti-CD20 & 37
0.75 ± 0.10
1.0 ± 0.21
0.65 ± 0.10
0.96 ± 0.16
0.70 ± 0.072
2.2 ± 0.58
anti-CD20 & 52
0.42 ± 0.083
1.2 ± 0.26
0.50 ± 0.074
1.1 ± 0.20
0.12 ± 0.029
0.38 ± 0.13
anti-CD37 & 52
0.41 ± 0.080
1.2 ± 0.31
0.73 ± 0.048
1.3 ± 0.15
0.24 ± 0.057
1.2 ± 0.30
anti-CD19 & 20 & 37
0.70 ± 0.090
1.0 ± 0.16
0.90 ± 0.017
1.2 ± 0.049
0.72 ± 0.12
1.3 ± 0.24
anti-CD19 & 20 & 52
1.0 ± 0.13
1.9 ± 0.32
0.98 ± 0.025
1.6 ± 0.098
0.99 ± 0.13
1.8 ± 0.31
anti-CD19 & 37 &52
0.90 ± 0.15
1.8 ± 0.36
0.92 ± 0.030
1.3 ± 0.081
0.55 ± 0.062
1.2 ± 0.15
anti-CD20 & 37 & 52 anti-CD19 & 20 & 37 & 52
0.32 ± 0.045
0.75 ± 0.13
0.55 ± 0.033
0.98 ± 0.088
0.26 ± 0.066
0.94 ± 0.29
0.83 ± 0.070
1.5 ± 0.19
0.94 ± 0.038
1.4 ± 0.086
0.83 ± 0.073
1.8 ± 0.23
153
Binding synergy
Normalized spot intensity
Binding synergy -
1.0 ± 0.058 0.41 ± 0.14
Raji
697
19
20
19
MEC-1
+
19
20
20
--
++
+
-
-
52
+ +
+
37
52
+
37
52
37
Figure 6.4 Graphic representation of synergistic networks of Raji (left), 697 (middle), and MEC-1 (right). Each node, with a number on it, denotes the corresponding single antibody. A undirected edge connecting two nodes denotes an existing synergy between the two antibodies represented by the node. Solid edges standard for positive synergy (larger than 1) and dashed edges standard for negative synergy (smaller than 1). Symbols “+” and “++”, and “-” and “--” are used to qualitatively represent the magnitude of the positive and negative synergy, respectively.
154
Supplementary information
18000
25000
14000
15000
10000
Anti-CD20
0.05mg/ml 0.1mg/ml 0.3mg/ml 0.5mg/ml
16000
Mean spot intensity
20000
Mean spot intensity
Anti-CD19
0.05mg/ml 0.1mg/ml 0.3mg/ml 0.5mg/ml
12000 10000 8000 6000 4000
5000
2000 0
0 0
5
10
15
0
20
5
10
15
20
Cell concentration (x 106/ml)
Cell concentration (x 106/ml)
14000 1600
10000
8000
6000
4000
Anti-CD52
0.05mg/ml 0.1mg/ml 0.3mg/ml 0.5mg/ml
1400
Mean spot intensity
Mean spot intensity
Anti-CD37
0.05mg/ml 0.1mg/ml 0.5mg/ml 0.3mg/ml
12000
1200 1000 800 600 400
2000
200
0
0 0
5
10
15
20
0
Cell concentration (x 106/ml)
5
10
15
20
Cell concentration (x 106/ml)
SP. Figure 6.1 The dependence of cell binding on antibody concentration and cell concentration in cellular microarray assays. Anti-CD19, anti-CD20, anti-CD37 and anti-CD52 at concentrations of 0.05 mg/ml, 0.1 mg/ml, 0.3 mg/ml and 0.5 mg/ml were printed. For concentrations lower than 0.5 mg/ml, BSA was supplemented as print additive to keep the total concentration to be 0.5mg/ml. Individual spots contain 0.33 nanoliters of printing solution. 50µl solutions of CFSE-labeled Raji cells at densities (concentrations) of 0.056x106 /ml, 0.1x106 /ml, 0.5x106 /ml, 1x106 /ml, 1.5x106 /ml, 5x106 /ml and 20 x106 /ml were incubated onto subarrays separated by silicone gasket. Error bars are standard error means across spot replicates (6 replicates per sample on the entire slide).
155
References Belov, L., P. Huang, N. Barber, S.P. Mulligan, andR.I. Christopherson. 2003. Identification of repertoires of surface antigens on leukemias using an antibody microarray. Proteomics 3(11):2147-2154. Belov, L., P. Huang, J.S. Chrisp, S.P. Mulligan, andR.I. Christopherson. 2005. Screening microarrays of novel monoclonal antibodies for binding to T-, Band myeloid leukaemia cells. Journal of Immunological Methods 305(1):10-19. Belov, L., S.P. Mulligan, N. Barber, A. Woolfson, M. Scott, K. Stoner, J.S. Chrisp, W.A. Sewell, K.F. Bradstock, L. Bendall, D.S. Pascovici, M. Thomas, W. Erber, P. Huang, M. Sartor, G.A.R. Young, J.S. Wiley, S. Juneja, W.G. Wierda, A.R. Green, M.J. Keating, andR.I. Christopherson. 2006. Analysis of human leukaemias and lymphomas using extensive immunophenotypes from an antibody microarray. British Journal of Haematology 135(2):184-197. Bendas, G. 2001. Immunoliposomes - A promising approach to targeting cancer therapy. Biodrugs 15(4):215-224. Chen, D.S., andM.M. Davis. 2006. Molecular and functional analysis using live cell microarrays. Current Opinion in Chemical Biology 10(1):28-34. Ishida, T., D.L. Iden, andT.M. Allen. 1999. A combinatorial approach to producing sterically stabilized (Stealth) immunoliposomal drugs. FEBS Lett 460(1):129133. Kluza, E., D.W.J. van der Schaft, P.A.I. Hautvast, W.J.M. Mulder, K.H. Mayo, A.W. Griffioen, G.J. Strijkers, andK. Nicolay. 2010. Synergistic Targeting of alpha(v)beta(3) Integrin and Galectin-1 with Heteromultivalent Paramagnetic Liposomes for Combined MR Imaging and Treatment of Angiogenesis. Nano Letters 10(1):52-58. Laginha, K., D. Mumbengegwi, andT. Allen. 2005. Liposomes targeted via two different antibodies: Assay, B-cell binding and cytotoxicity. Biochimica Et Biophysica Acta-Biomembranes 1711(1):25-32. Mastrobattista, E., G.A. Koning, andG. Storm. 1999. Immunoliposomes for the targeted delivery of antitumor drugs. Advanced Drug Delivery Reviews 40(12):103-127.
156
Sapra, P., andT.M. Allen. 2002. Internalizing antibodies are necessary for improved therapeutic efficacy of antibody-targeted liposomal drugs. Cancer Research 62(24):7190-7194. Zhao, X.B., R. Lapalombella, T. Joshi, C. Cheney, A. Gowda, M.S. Hayden-Ledbetter, P.R. Baurn, T.S. Lin, D. Jarjoura, A. Lehman, D. Kussewitt, R.J. Lee, M.A. Caligiuri, S. Tridandapani, N. Muthusamy, andJ.C. Byrd. 2007. Targeting CD37-positive lymphoid malignancies with a novel engineered small modular immunopharmaceutical. Blood 110(7):2569-2577.
157
CHAPTER 7 A MIXED TWO-COMPONENT SELFASSEMBLED MONOLAYER-BASED PROTEIN MICROARRAY PLATFORM
7.1. Introduction
As introduced in Chapter 1, there have been considerable efforts in developing novel substrates that can immobilize proteins in an oriented or directed fashion as well as maximize retention of protein bioactivities and minimize nonspecific protein adsorption. In recent years, oligoethylene oxide (OEO)-contained self-assembled monolayers (SAMs) have attracted extensive interest for studying molecular recognition at interfaces, due to their characteristics of spontaneously forming highly ordered assemblies upon chemisorption on a metal or metal oxide surface, and the protein adsorption resistance feature of the OEO group (Vanderah et al., 2004). One model system that has been successfully used in studying specific protein-protein interactions is two mixed SAMs that contain an ordinary alkylthiol and a biotinterminated alkylthiol (Bieri et al., 1999; Silin et al., 2006; Spinke et al., 1993a). A multi-layer structure can be built up based on this two-component SAMs. The
158
employment of biotin takes advantages of its well-characterized robust binding interaction with streptavidin and streptavidin’s four biotin binding sites, so that biotinylated proteins can be further immobilized via surface-bound streptavidin. The presence of the non-biotinylated SAM at proper molar ranges is necessary for two reasons: first, to minimize possible steric hinderance to streptavidin binding in the case of pure biotinylated alkylthiols (Spinke et al., 1993b); second, the biotin head groups of pure biotinylated alkylthiols on surfaces have been experimentally indicated to be poorly ordered (Nelson et al., 2001).
In this section, we report the structural characterization of a protein microarray platform that is based on self-assembled monolayers (SAMs) consisting of mixed biotinylated and non-biotinylated polyethylene oxide (PEO)-modified alkyl thiols on gold-coated surface, and its preliminary applications in cellular microarrays. The cartoon in Figure 7.1 shows the multi-component, multi-layer structure of the SAMbased protein platform. Specifically, the mixed SAMs consist of a short SAM, HS(CH2)3O(EO)3CH3 (abbreviated as C3EO3), and a longer biotinylated SAM, HS(CH2)3O(EO)6-biotin (abbreviated as C3EO6-Bt), with the employment of the EO groups for resistance of nonspecific protein adsorption (Prime and Whitesides, 1993; Vanderah et al., 2004). Streptavidin is immobilized onto the surface via the biotin end of the longer SAM molecules to form the second layer. A biotinylated protein, specific for its target in solution or on cell surface, can be captured and immobilized via another biotin-streptavidin linkage, forming a third layer on top of the streptavidin layer. Alternatively, this biotinylated protein can be a secondary capture antibody,
159
such as anti-IgG antibody, so that proteins of interest that are not easy to biotinylate can be immobilized onto the surface.
Biotin-antiIgG Streptavidin
OEO-thiol mixture Biotin
Figure 7.1 Demonstration of the multi-component, multi-layer structure of the platform.
7.2. Preparation of the platform The preparation of mixed SAMs has been described (Silin et al., 2006). Briefly, immediately after metal deposition, gold-coated glass slides or silicon wafers were immersed overnight in mixed SAMs/H2O solutions at a total SAM concentration of 0.5 mM. Multiple molar compositions of the mixed SAM solution were studied: 1 mol% (i.e., 0.495 mM of C3EO3 and 0.005 mM of C6EO6-Bt) and 20 mol% (i.e., 0.4 mM of C3EO3 and 0.1 mM of C6EO6-Bt) were prepared. Then, the samples were
160
rinsed with ethanol (Sigma Aldrich), dried under a nitrogen stream, and immersed into 5 x 10-7 M streptavidin (American Qualex) in PBS solution for 2 hr at room temperature. At the end of immersion, the samples were rinsed with distilled water (diH2O), dried under nitrogen, and then immersed in 0.02 mg/ml (1.3 x 10-7 M) biotin-anti-IgG (BD Biosciences) in PBS solution for 2 hr at room temperature. After rinsing with diH2O and dried under nitrogen, samples were ready for microarray fabrication.
7.3. Structural characterization by neutron scattering Layer-by-layer structural characterization of the developed platform were conducted using advanced neutron diffractometer/reflectometer (AND/R) at the NIST Center for Neutron Research (NCNR) (Dura et al., 2006). Thickness and water content (volume fraction) of each layer of the platform built on the SAM mixture at two molar compositions, 20 mol% C3EO6-Bt and 1 mol% C3EO6-Bt, were determined by fitting the collected neutron reflectivity measurements, as summarized in Table 7.1.
Table 7.1 Neutron scattering characterization
20 mol% C3EO6-Bt
1 mol% C3EO6-Bt
Layer
Thickness (Å)
Volume fraction
Thickness (Å)
Volume fraction
SAM
11.2
0.88
13.0
0.87
Streptavidin
40.5
0.41
38.0
0.016
Anti-IgG
53.4
0.23
NT
NT
161
Notably, the thickness of the SAM layer formed from 20 mol% C3EO6-Bt is 11.2 Å, about 2 Å thinner than the layer formed from 1 mol% C3EO6-Bt in which there were apparently fewer C3EO6-Bt molecules, the longer SAM compared with C3EO3. This “counter-intuitive” observation probably resulted from the physical compression of the SAM layer by the streptavidin and anti-IgG layers: in the absence of the two layers, the thickness of the SAM formed from 20 mol% C3EO6-Bt is 14.6 Å (not shown in the table). The streptavidin layer formed on the 20 mol% C3EO6-Bt is close to the two-dimensional crystal layer of streptavidin (Darst et al., 1991; Nelson et al., 2001). For 1 mol% C3EO6-Bt SAMs the volume fraction of streptavidin is very small, less than 0.2; as a result, the presence of an anti-IgG layer could not be detected in AND/R measurements. In addition, with the thickness and volume fraction of a layer, and the size of the single molecule known, the number of the molecules in that layer can be further estimated.
7.4. Cellular microarrays built on the platform Figure 7.2 compares the performance of an HLA A2-Ig dimer and anti-CD3 microarray that is built on this SAM-based platform with a commercial microarray substrate, Nexterion slide H. For both substrates, anti-CD3 spots were densely packed with cells, as indicated by the bright fluorescence emitted from the spots. The more uniform fluorescence from anti-CD3 spots on SAM-based substrate than on Nexterion slide H suggests that anti-CD3 molecules on SAM-based substrate are immobilized in a more uniform fashion, probably attributed to the highly ordered structure of the
162
platform as characterized; in contrast, Nextetrion slide H is a three-dimensional hydrogel in which proteins such as anti-CD3 are randomly immobilized via primary amine groups.
Cell binding to spots printed with various HLA A2-Ig dimer solutions either did not exist or at low levels. For the large spots on Nexterion slide, no cell binding was observed on spots printed with the unloaded BD dimer (dimer purchased from BD Biosciences), while some cell binding was observed on spots printed with the passively-loaded BD dimer and to a less extent on spots printed with the activelyloaded JPS dimer (dimer provided by Dr. Scheck). Greater cell binding was observed on spots printed with actively-loaded JPS dimer mixed with CMVpp65, but mainly in the periphery of the spots. For the large spots on SAM-based slide, no cell binding was observed on spots printed with either the passively-loaded BD dimer or activelyloaded JPS dimer, while some cell binding was observed on spots printed with actively-loaded JPS dimer mixed with CMVpp65. The advantage of SAM-based platform for producing uniform cell binding on spots is strongly indicated by comparing cell binding on CMVpp65-mixed actively-loaded JPS dimer spots.
To the best of our knowledge, this has been the first report showing successful cell binding in a cellular microarray that is built on a SAM-based multi-component multilayer platform.
163
Nexterion slide H
SAM-based substrate
Figure 7.2 Comparison of cell binding on gold-coated SAM-based substrate (right) with commercial substrate Nexterion slide H (left). The two substrates were printed identically with spots of two sizes: one consisting of 10 nl printing solution and having a diameter of 500 µm, and the other consisting of 90 nl printing solution and having a diameter of 1000 µm. Five samples were printed in each spot size on both substrates. For 1000 µm diameter spots, samples were printed in triplicate in columns, as marked with arrows on Nexterion slide H; from left to right, the samples were: unloaded BD dimer, CMVpp65 passively loaded BD dimer, CMVpp65 actively loaded JPS dimer, CMVpp65 actively loaded JPS dimer mixed with CMVpp65, and anti-CD3. For 500 µm diameter spots, each sample was printed in 12 spot replicates, organized in two 6 x 5 subarrays (not marked on the images). The mixed SAMs contained 1 mol% C3EO6-Bt in 99 mol% C3EO3. Prior to printing, Nexterion slide H was pre-coated with anti-IgG1, to help improve the orientation of HLA A2-Ig (Deviren et al., 2007). CMVpp65-specific CTLs, enriched and expanded from aAPC technology (Oelke et al., 2003), were labeled with CFSE and incubated on the entire subarrays at a concentration of ~ 4 x 106 /ml.
164
References Bieri, C., O.P. Ernst, S. Heyse, K.P. Hofmann, andH. Vogel. 1999. Micropatterned immobilization of a G protein-coupled receptor and direct detection of G protein activation. Nature Biotechnology 17(11):1105-1108. Darst, S.A., M. Ahlers, P.H. Meller, E.W. Kubalek, R. Blankenburg, H.O. Ribi, H. Ringsdorf, andR.D. Kornberg. 1991. 2-Dimensional Crystals of Streptavidin on Biotinylated Lipid Layers and Their Interactions with Biotinylated Macromolecules. Biophysical Journal 59(2):387-396. Deviren, G., K. Gupta, M.E. Paulaitis, andJ.P. Schneck. 2007. Detection of antigenspecific T cells on p/MHC microarrays journal of molecular recognition 20(1):32-38. Dura, J.A., D.J. Pierce, C.F. Majkrzak, N.C. Maliszewskyj, D.J. McGillivray, M. Losche, K.V. O'Donovan, M. Mihailescu, U. Perez-Salas, D.L. Worcester, andS.H. White. 2006. AND/R: Advanced neutron diffractometer/reflectometer for investigation of thin films and multilayers for the life sciences. Review of Scientific Instruments 77(7):-. Nelson, K.E., L. Gamble, L.S. Jung, M.S. Boeckl, E. Naeemi, S.L. Golledge, T. Sasaki, D.G. Castner, C.T. Campbell, andP.S. Stayton. 2001. Surface characterization of mixed self-assembled monolayers designed for streptavidin immobilization. Langmuir 17(9):2807-2816. Oelke, M., M.V. Maus, D. Didiano, C.H. June, A. Mackensen, andJ.P. Schneck. 2003. Ex vivo induction and expansion of antigen-specific cytotoxic T cells by HLA-Ig-coated artificial antigen-presenting cells. Nature Medicine 9(5):619624. Prime, K.L., andG.M. Whitesides. 1993. Adsorption of Proteins onto Surfaces Containing End-Attached Oligo(Ethylene Oxide) - a Model System Using Self-Assembled Monolayers. J. Am. Chem. Soc. 115(23):10714-10721. Silin, V.I., E.A. Karlik, K.D. Ridge, andD.J. Vanderah. 2006. Development of surface-based assays for transmembrane proteins: Selective immobilization of functional CCR5, a G protein-coupled receptor. Analytical Biochemistry 349(2):247-253.
165
Spinke, J., M. Liley, H.J. Guder, L. Angermaier, andW. Knoll. 1993a. Molecular Recognition at Self-Assembled Monolayers - the Construction of Multicomponent Multilayers. Langmuir 9(7):1821-1825. Spinke, J., M. Liley, F.J. Schmitt, H.J. Guder, L. Angermaier, andW. Knoll. 1993b. Molecular Recognition at Self-Assembled Monolayers - Optimization of Surface Functionalization. Journal of Chemical Physics 99(9):7012-7019. Vanderah, D.J., H.L. La, J. Naff, V. Silin, andK.A. Rubinson. 2004. Control of protein adsorption: Molecular level structural and spatial variables. J. Am. Chem. Soc. 126(42):13639-13641.
166
REFERNCES Abastado, J.P., Y.C. Lone, A. Casrouge, G. Boulot, andP. Kourilsky. 1995. Dimerization of Soluble Major Histocompatibility Complex Peptide Complexes Is Sufficient for Activation of T-Cell Hybridoma and Induction of Unresponsiveness. Journal of Experimental Medicine 182(2):439-447. Afanassiev, V., V. Hanemann, andS. Wölfl. 2000. Preparation of DNA and protein micro arrays on glass slides coated with an agarose film. Nucleic Acids Res 28:e66-e66. Ajikumar, P.K., J. Kiat, Y.C. Tang, J.Y. Lee, G. Stephanopoulos, andH.P. Too. 2007. Carboxyl-terminated dendrimer-coated bioactive interface for protein microarray: High-sensitivity detection of antigen in complex biological samples. Langmuir 23(10):5670-5677. Altman, J.D., P.A.H. Moss, P.J.R. Goulder, D.H. Barouch, M.G. McHeyzerWilliams, J.I. Bell, A.J. McMichael, andM.M. Davis. 1996. Phenotypic analysis of antigen-specific T lymphocytes. Science 274(5284):94-96. Anderson, D.G., S. Levenberg, andR. Langer. 2004. Nanoliter-scale synthesis of arrayed biomaterials and application to human embryonic stem cells. Nat Biotechnol 22(7):863-866. Angenendt, P., J. Glokler, D. Murphy, H. Lehrach, andD.J. Cahill. 2002. Toward optimized antibody microarrays: a comparison of current microarray support materials. Analytical Biochemistry 309(2):253-260. Angenendt, P., J. Glokler, J. Sobek, H. Lehrach, andD.J. Cahill. 2003. Next generation of protein microarray support materials: Evaluation for protein and antibody microarray applications. Journal of Chromatography A 1009(12):97-104. Angres, B. 2005. Cell microarrays. Expert Review of Molecular Diagnostics 5(5):769779. Arakawa, T., Y. Kita, andS.N. Timasheff. 2007. Protein precipitation and denaturation by dimethyl sulfoxide. Biophysical Chemistry 131(1-3):62-70. Arenkov, P., A. Kukhtin, A. Gemmell, S. Voloshchuk, V. Chupeeva, andA. Mirzabekov. 2000. Protein microchips: Use for immunoassay and enzymatic reactions. Analytical Biochemistry 278(2):123-131.
167
Arstila, T.P., A. Casrouge, V. Baron, J. Even, J. Kanellopoulos, andP. Kourilsky. 1999. A direct estimate of the human alpha beta T cell receptor diversity. Science 286(5441):958-961. Assarsson, E., H.H. Bui, J. Sidney, Q. Zhang, J. Glenn, C. Oseroff, I.N. Mbawuike, J. Alexander, M.J. Newman, H. Grey, andA. Sette. 2008. Immunomic Analysis of the Repertoire of T-Cell Specificities for Influenza A Virus in Humans. Journal of Virology 82(24):12241-12251. Bailey, S.N., S.M. Ali, A.E. Carpenter, C.O. Higgins, andD.M. Sabatini. 2006. Microarrays of lentiviruses for gene function screens in immortalized and primary cells. Nature Methods 3(2):117-122. Bailey, S.N., D.M. Sabatini, andB.R. Stockwell. 2004. Microarrays of small molecules embedded in biodegradable polymers for use in mammalian cellbased screens. Proc Natl Acad Sci U S A 101(46):16144-16149. Bakker, A.H., R. Hoppes, C. Linnemann, M. Toebes, B. Rodenko, C.R. Berkers, S.R. Hadrup, W.J.E. van Esch, M.H.M. Heemskerk, H. Ovaa, andT.N.M. Schumacher. 2008. Conditional MHC class I ligands and peptide exchange technology for the human MHC gene products HLA-A1, -A3, -A11 and -B7. Proceedings of the National Academy of Sciences of the United States of America 105(10):3825-3830. Bankovich, A.J., A.T. Girvin, A.K. Moesta, andK.C. Garcia. 2004. Peptide register shifting within the MHC groove: theory becomes reality. Molecular Immunology 40(14-15):1033-1039. Bartling, B., H.S. Hofmann, T. Boettger, G. Hansen, S. Burdach, R.E. Silber, andA. Simm. 2005. Comparative application of antibody and gene array for expression profiling in human squamous cell lung carcinoma. Lung Cancer 49(2):145-154. Bednarek, M.A., S.Y. Sauma, M.C. Gammon, G. Porter, S. Tamhankar, A.R. Williamson, andH.J. Zweerink. 1991. The Minimum Peptide Epitope from the Influenza-Virus Matrix Protein - Extra and Intracellular Loading of Hla-a2. Journal of Immunology 147(12):4047-4053. Belov, L., O. de la Vega, C.G. dos Remedios, S.P. Mulligan, andR.I. Christopherson. 2001. Immunophenotyping of leukemias using a cluster of differentiation antibody microarray. Cancer Research 61(11):4483-4489. Belov, L., P. Huang, N. Barber, S.P. Mulligan, andR.I. Christopherson. 2003. Identification of repertoires of surface antigens on leukemias using an antibody microarray. Proteomics 3(11):2147-2154. Belov, L., P. Huang, J.S. Chrisp, S.P. Mulligan, andR.I. Christopherson. 2005. Screening microarrays of novel monoclonal antibodies for binding to T-, Band myeloid leukaemia cells. Journal of Immunological Methods 305(1):10-19. 168
Belov, L., S.P. Mulligan, N. Barber, A. Woolfson, M. Scott, K. Stoner, J.S. Chrisp, W.A. Sewell, K.F. Bradstock, L. Bendall, D.S. Pascovici, M. Thomas, W. Erber, P. Huang, M. Sartor, G.A.R. Young, J.S. Wiley, S. Juneja, W.G. Wierda, A.R. Green, M.J. Keating, andR.I. Christopherson. 2006. Analysis of human leukaemias and lymphomas using extensive immunophenotypes from an antibody microarray. British Journal of Haematology 135(2):184-197. Bendas, G. 2001. Immunoliposomes - A promising approach to targeting cancer therapy. Biodrugs 15(4):215-224. Bieri, C., O.P. Ernst, S. Heyse, K.P. Hofmann, andH. Vogel. 1999. Micropatterned immobilization of a G protein-coupled receptor and direct detection of G protein activation. Nature Biotechnology 17(11):1105-1108. Binz, A.K., R.C. Rodriguez, W.E. Biddison, andB.M. Baker. 2003. Thermodynamic and kinetic analysis of a peptide-class I MHC interaction highlights the noncovalent nature and conformational dynamics of the class I heterotrimer. Biochemistry 42(17):4954-4961. Bouillot, M., J. Choppin, F. Cornille, F. Martinon, T. Papo, E. Gomard, M.C. Fourniezaluski, andJ.P. Levy. 1989. Physical Association between Mhc ClassI Molecules and Immunogenic Peptides. Nature 339(6224):473-475. Boyd, L.F., S. Kozlowski, andD.H. Margulies. 1992. Solution Binding of an Antigenic Peptide to a Major Histocompatibility Complex Class-I Molecule and the Role of Beta-2-Microglobulin. Proceedings of the National Academy of Sciences of the United States of America 89(6):2242-2246. Braga-Neto, U.M., andE.T. Marques, Jr. 2006a. From functional genomics to functional immunomics: new challenges, old problems, big rewards. PLoS Comput Biol 2(7):e81. Braga-Neto, U.M., andE.T.A. Marques. 2006b. From functional genomics to functional immunomics: New challenges, old problems, big rewards. Plos Computational Biology 2(7):651-662. Buchli, R., R.S. VanGundy, C.F. Giberson, andW.H. Hildebrand. 2006. Critical factors in the development of fluorescence polarization-based peptide binding assays: An equilibrium study monitoring specific peptide binding to soluble HLA-A*0201. Journal of Immunological Methods 314(1-2):38-53. Buchli, R., R.S. VanGundy, H.D. Hickman-Miller, C.F. Giberson, W. Bardet, andW.H. Hildebrand. 2004. Real-time measurement of in vitro peptide binding to soluble HLA-A*0201 by fluorescence polarization. Biochemistry 43(46):14852-14863. Buchli, R., R.S. VanGundy, H.D. Hickman-Miller, C.F. Giberson, W. Bardet, andW.H. Hildebrand. 2005. Development and validation of a fluorescence
169
polarization-based competitive peptide-binding assay for HLA-A*0201 - A new tool for epitope discovery. Biochemistry 44(37):12491-12507. Buus, S., A. Stryhn, K. Winther, N. Kirkby, andL.O. Pedersen. 1995. ReceptorLigand Interactions Measured by an Improved Spun Column Chromatography Technique - a High-Efficiency and High-Throughput Size Separation Method. Biochimica Et Biophysica Acta-General Subjects 1243(3):453-460. Castel, D., A. Pitaval, M.A. Debily, andX. Gidrol. 2006. Cell microarrays in drug discovery. Drug Discovery Today 11(13-14):616-622. Charles A. Laneway, J., P. Travers, M. Walport, andM.J. Shlomchik. 2005. Immunobiology-the immune system in health and disease. Garland Science. Chen, D.S., Y. Soen, T.B. Stuge, P.P. Lee, J.S. Weber, P.O. Brown, and M.M. Davis. 2005. Marked differences in human melanoma antigen-specific T cell responsiveness after vaccination using a functional microarray. Plos medicine 2(10):1018-1030. Chen, D.S., and M.M. Davis. 2006. Molecular and functional analysis using live cell microarrays. Current Opinion in Chemical Biology 10(1):28-34. Chen, W.S., L.C. Anton, J.R. Bennink, andJ.W. Yewdell. 2000. Dissecting the multifactorial causes of immunodominance in class I-restricted T cell responses to viruses. Immunity 12(1):83-93. Cheng, Y., andW.H. Prusoff. 1973. Relationship between Inhibition Constant (K1) and Concentration of Inhibitor Which Causes 50 Per Cent Inhibition (I50) of an Enzymatic-Reaction. Biochemical Pharmacology 22(23):3099-3108. Choudhuri, S. 2004. Microarrays in biology and medicine. Journal of Biochemical and Molecular Toxicology 18(4):171-179. Christinck, E.R., M.A. Luscher, B.H. Barber, andD.B. Williams. 1991. Peptide Binding to Class-I Mhc on Living Cells and Quantitation of Complexes Required for Ctl Lysis. Nature 352(6330):67-70. Callan, M.F.C., L. Tan, N. Annels, G.S. Ogg, J.D.K. Wilson, C.A. O'Callaghan, N. Steven, A.J. McMichael, andA.B. Rickinson. 1998. Direct visualization of antigen-specific CD8(+) T cells during the primary immune response to Epstein-Barr virus in vivo. Journal of Experimental Medicine 187(9):13951402. Chan, P.Y., M.B. Lawrence, M.L. Dustin, L.M. Ferguson, D.E. Golan, andT.A. Springer. 1991. Influence of Receptor Lateral Mobility on Adhesion Strengthening between Membranes Containing Lfa-3 and Cd2. Journal of Cell Biology 115(1):245-255.
170
Charles A. Laneway, J., P. Travers, M. Walport, andM.J. Shlomchik. 2005. Immunobiology-the immune system in health and disease. Garland Science. Chen, D.S., Y. Soen, T.B. Stuge, P.P. Lee, J.S. Weber, P.O. Brown, andM.M. Davis. 2005. Marked differences in human melanoma antigen-specific T cell responsiveness after vaccination using a functional microarray. Plos medicine 2(10):1018-1030. Chen, W.S., L.C. Anton, J.R. Bennink, andJ.W. Yewdell. 2000. Dissecting the multifactorial causes of immunodominance in class I-restricted T cell responses to viruses. Immunity 12(1):83-93. Corr, M., A.E. Slanetz, L.F. Boyd, M.T. Jelonek, S. Khilko, B.K. Alramadi, Y.S. Kim, S.E. Maher, A.L.M. Bothwell, andD.H. Margulies. 1994. T-Cell Receptor-Mhc Class-I Peptide Interactions - Affinity, Kinetics, and Specificity. Science 265(5174):946-949. Dalporto, J., T.E. Johansen, B. Catipovic, D.J. Parfiit, D. Tuveson, U. Gether, S. Kozlowski, D.T. Fearon, andJ.P. Schneck. 1993. A Soluble Divalent Class-I Major Histocompatibility Complex Molecule Inhibits Alloreactive T-Cells at Nanomolar Concentrations. Proceedings of the National Academy of Sciences of the United States of America 90(14):6671-6675. Darst, S.A., M. Ahlers, P.H. Meller, E.W. Kubalek, R. Blankenburg, H.O. Ribi, H. Ringsdorf, andR.D. Kornberg. 1991. 2-Dimensional Crystals of Streptavidin on Biotinylated Lipid Layers and Their Interactions with Biotinylated Macromolecules. Biophysical Journal 59(2):387-396. Davenport, M.P., C. Fazou, A.J. McMichael, andM.F.C. Callan. 2002. Clonal selection, clonal senescence, and clonal succession: The evolution of the T cell response to infection with a persistent virus. Journal of Immunology 168(7):3309-3317. Davis, M.M., J.J. Boniface, Z. Reich, D. Lyons, J. Hampl, B. Arden, andY.H. Chien. 1998. Ligand recognition by alpha beta T cell receptors. Annual Review of Immunology 16:523-+. Davis, M.M., andY.H. Chien. 1993. Topology and Affinity of T-Cell Receptor Mediated Recognition of Peptide Mhc Complexes. Current Opinion in Immunology 5(1):45-49. Dedier, S., S. Reinelt, R. Severine, G. Folkers, andD. Rognan. 2001. Use of fluorescence polarization to monitor MHC-peptide interactions in solution. Journal of Immunological Methods 255(1-2):57-66. Deegan, R.D., O. Bakajin, T.F. Dupont, G. Huber, andS.R.N.T.A. Witten. 1997. Capillary flow as the cause of ring stains from dried liquid drops. Nature 389:827-829.
171
Delehanty, J.B., andF.S. Ligler. 2003. Method for printing functional protein microarrays. BioTechniques 34:380-385. Delon, J., C. Gregoire, B. Malissen, S. Darche, F. Lemaitre, P. Kourilsky, J.P. Abastado, andA. Trautmann. 1998. CD8 expression allows T cell signaling by monomeric peptide-MHC complexes. Immunity 9(4):467-473. Dembo, M., andG.I. Bell. 1987. The Thermodynamics of Cell-Adhesion. Current Topics in Membranes and Transport 29:71-89. Deng, Y., andX.-Y. Zhu. 2006. Transport at the Air/Water Interface is the Reason for Rings in Protein Microarrays. JACS 128:2768-2769. Deviren, G., K. Gupta, M.E. Paulaitis, andJ.P. Schneck. 2007. Detection of antigenspecific T cells on p/MHC microarrays journal of molecular recognition 20(1):32-38. Diaz-Mochon, J.J., G. Tourniaire, andM. Bradley. 2007. Microarray platforms for enzymatic and cell-based assays. Chemical Society Reviews 36(3):449-457. Diederich, A., andM. Losche. 1996. Novel biosensoric devices based on molecular protein hetero-multilayer films. Advances in Biophysics, Vol 34, 1997 34:205230. Disney, M.D., andP.H. Seeberger. 2004. The use of carbohydrate microarrays to study carbohydrate-cell interactions and to detect pathogens. Chem Biol 11(12):1701-1707. Dura, J.A., D.J. Pierce, C.F. Majkrzak, N.C. Maliszewskyj, D.J. McGillivray, M. Losche, K.V. O'Donovan, M. Mihailescu, U. Perez-Salas, D.L. Worcester, andS.H. White. 2006. AND/R: Advanced neutron diffractometer/reflectometer for investigation of thin films and multilayers for the life sciences. Review of Scientific Instruments 77(7):-. Durai, M., C. Krueger, Z.H. Ye, L.Z. Cheng, A. Mackensen, M. Oelke, andJ. Schneck. 2009. In vivo functional efficacy of tumor-specific T cells expanded using HLA-Ig based artificial antigen presenting cells (aAPC). Cancer Immunol. Immunother. 58(2):209-220. Eisenstein, M. 2006. Protein arrays - Growing pains. Nature 444(7121):959-964. Falk, K., O. Rotzschke, S. Stevanovic, G. Jung, andH.G. Rammensee. 1991. AlleleSpecific Motifs Revealed by Sequencing of Self-Peptides Eluted from Mhc Molecules. Nature 351(6324):290-296. Feilner, T., J. Kreutzberger, B. Niemann, A. Kramer, A. Possling, H. Seitz, andB. Kersten. 2004. Proteomic studies using microarrays. Current Proteomics 1:283-295.
172
Fernandes, T.G., M.M. Diogo, D.S. Clark, J.S. Dordick, andJ.M. Cabral. 2009. Highthroughput cellular microarray platforms: applications in drug discovery, toxicology and stem cell research. Trends Biotechnol 27(6):342-349. Flaim, C.J., S. Chien, andS.N. Bhatia. 2005. An extracellular matrix microarray for probing cellular differentiation. Nat Methods 2(2):119-125. Gakamsky, D.M., P.J. Bjorkman, andI. Pecht. 1996. Peptide interaction with a class I major histocompatibility complex-encoded molecule: Allosteric control of the ternary complex stability. Biochemistry 35(47):14841-14848. Gakamsky, D.M., L.F. Boyd, D.H. Margulies, D.M. Davis, J.L. Strominger, andI. Pecht. 1999. An allosteric mechanism controls antigen presentation by the H2K(b) complex. Biochemistry 38(37):12165-12173. Gakamsky, D.M., D.M. Davis, J.L. Strominger, andI. Pecht. 2000. Assembly and dissociation of human leukocyte antigen (HLA)-A2 studied by real-time fluorescence resonance energy transfer. Biochemistry 39(36):11163-11169. Gao, W.M., R. Kuick, R.P. Orchekowski, D.E. Misek, J. Qiu, A.K. Greenberg, W.N. Rom, D.E. Brenner, G.S. Omenn, B.B. Haab, andS.M. Hanash. 2005. Distinctive serum protein profiles involving abundant proteins in lung cancer patients based upon antibody microarray analysis. Bmc Cancer 5:-. Ge, H.e.a. 2000. UPA, a universal protein array system for quantitative detection of protein-protein, protein-DNA, protein-RNA and proteinligand interactions. (2000) 28: e3. Nucleic Acids Res. 28:e3. Genestier, L., R. Paillot, N. BonnefoyBerard, G. Meffre, M. Flacher, D. Fevre, Y.J. Liu, P. LeBouteiller, H. Waldmann, V.H. Engelhard, J. Banchereau, andJ.P. Revillard. 1997. Fas-independent apoptosis of activated T cells induced by antibodies to the HLA class I alpha 1 domain. Blood 90(9):3629-3639. Gianfrani, C., C. Oseroff, J. Sidney, R.W. Chesnut, andA. Sette. 2000. Human memory CTL response specific for influenza A virus is broad and multispecific. Human Immunology 61(5):438-452. Gopalakrishnan, S.M., R.B. Moreland, J.L. Kofron, R.J. Helfrich, E. Gubbins, J. McGowen, J.N. Masters, D. Donnelly-Roberts, J.D. Brioni, D.J. Burns, andU. Warrior. 2003. A cell-based microarrayed compound screening format for identifying agonists of G-protein-coupled receptors. Anal Biochem 321(2):192-201. Grakoui, A., S.K. Bromley, C. Sumen, M.M. Davis, A.S. Shaw, P.M. Allen, andM.L. Dustin. 1999. The immunological synapse: A molecular machine controlling T cell activation. Science 285(5425):221-227. Greten, T.F., J.E. Slansky, R. Kubota, S.S. Soldan, E.M. Jaffee, T.P. Leist, D.M. Pardoll, S. Jacobson, andJ.P. Schneck. 1998. Direct visualization of antigen-
173
specific T cells: HTLV-1 Tax11-19-specific CD8(+) T cells are activated in peripheral blood and accumulate in cerebrospinal fluid from HAM/TSP patients. Proceedings of the National Academy of Sciences of the United States of America 95(13):7568-7573. Greten, T.F., and J.P. Schneck. 2002. Development and use of multimeric major histocompatibility complex molecules. Clinical and Diagnostic Laboratory Immunology 9(2):216-220. Guilleaume, B., A. Buneß, C. Schmidt, F. Klimek, G. Moldenhauer, W. Huber, D. Arlt, U. Korf, S. Wiemann, andA. Poustka. 2005. Systematic comparison of surface coatings for protein microarrays. PROTEOMICS 5(18):4705-4712. Hadrup, S.R., A.H. Bakker, C.Y.J. Shu, R.S. Andersen, J. van Veluw, P. Hombrink, E. Castermans, P.T. Straten, C. Blank, J.B. Haanen, M.H. Heemskerk, andT.N. Schumacher. 2009. Parallel detection of antigen-specific T-cell responses by multidimensional encoding of MHC multimers. Nature Methods 6(7):520-U579. He, X.S., B. Rehermann, F.X. Lopez-Labrador, J. Boisvert, R. Cheung, J. Mumm, H. Wedemeyer, M. Berenguer, T.L. Wright, M.M. Davis, andH.B. Greenberg. 1999. Quantitative analysis of hepatitis C virus-specific CD8(+) T cells in peripheral blood and liver using peptide-MHC tetramers. Proceedings of the National Academy of Sciences of the United States of America 96(10):56925697. Hill, A.B., S.P. Lee, J.S. Haurum, N. Murray, Q.Y. Yao, M. Rowe, N. Signoret, A.B. Rickinson, andA.J. Mcmichael. 1995. Class-I Major Histocompatibility Complex-Restricted Cytotoxic T-Lymphocytes Specific for Epstein-BarrVirus (Ebv) Nuclear Antigens Fail to Lyse the Ebv-Transformed BLymphoblastoid Cell-Lines against Which They Were Raised. Journal of Experimental Medicine 181(6):2221-2228. Ho, S.-W., G. Jona, C.T.L. Chen, M. Johnston, andM. Snyder. 2006. Linking DNAbinding proteins to their recognition sequences by using protein microarrays. PNAS 103:9940-9945. http://www.affymetrix.com/. Ishida, T., D.L. Iden, andT.M. Allen. 1999. A combinatorial approach to producing sterically stabilized (Stealth) immunoliposomal drugs. FEBS Lett 460(1):129133. Kasprowicz, V., S.M. Ward, A. Turner, A. Grammatikos, B.E. Nolan, L. LewisXimenez, C. Sharp, J. Woodruff, V.M. Fleming, S. Sims, B.D. Walker, A.K. Sewell, G.M. Lauer, andP. Klenerman. 2008. Defining the directionality and quality of influenza virus-specific CD8(+) T cell cross-reactivity in individuals infected with hepatitis C virus. Journal of Clinical Investigation 118(3):11431153. 174
Kastenmuller, W., G. Gasteiger, J.H. Gronau, R. Baier, R. Ljapoci, D.H. Busch, andI. Drexler. 2007. Cross-competition of CD8(+) T cells shapes the immunodominance hierarchy during boost vaccination. Journal of Experimental Medicine 204(9):2187-2198. Kato, K., M. Toda, andH. Iwata. 2007. Antibody arrays for quantitative immunophenotyping. biomaterials 28:1289-1297. Kenyon, G.L., D.M. DeMarini, E. Fuchs, D.J. Galas, J.F. Kirsch, T.S. Leyh, W.H. Moos, G.A. Petsko, D. Ringe, G.M. Rubin, L.C. Sheahan, N.R.C.S. Comm. 2002. Defining the mandate of proteomics in the post-genomics era: Workshop report. Molecular & Cellular Proteomics 1(10):763-780. Kern, F., I.P. Surel, C. Brock, B. Freistedt, H. Radtke, A. Scheffold, R. Blasczyk, P. Reinke, J. Schneider-Mergener, A. Radbruch, P. Walden, andH.D. Volk. 1998. T-cell epitope mapping by flow cytometry. Nature Medicine 4(8):975-978. Kessler, J.H., B. Mommaas, T. Mutis, I. Huijbers, D. Vissers, W.E. Benckhuijsen, G.M.T. Schreuder, R. Offringa, E. Goulmy, C.J.M. Melief, S.H. van der Burg, andJ.W. Drijfhout. 2003. Competition-based cellular peptide binding assays for 13 prevalent HLA class I Alleles using fluorescein-labeled synthetic peptides. Human Immunology 64(2):245-255. Khilko, S.N., M. Corr, L.F. Boyd, A. Lees, J.K. Inman, andD.H. Margulies. 1993. Direct-Detection of Major Histocompatibility Complex Class-I Binding to Antigenic Peptides Using Surface-Plasmon Resonance - Peptide Immobilization and Characterization of Binding-Specificity. Journal of Biological Chemistry 268(21):15425-15434. Khilko, S.N., M.T. Jelonek, M. Corr, L.F. Boyd, A.L.M. Bothwell, andD.H. Margulies. 1995. Measuring Interactions of Mhc Class-I Molecules Using Surface-Plasmon Resonance. Journal of Immunological Methods 183(1):7794. Kluza, E., D.W.J. van der Schaft, P.A.I. Hautvast, W.J.M. Mulder, K.H. Mayo, A.W. Griffioen, G.J. Strijkers, andK. Nicolay. 2010. Synergistic Targeting of alpha(v)beta(3) Integrin and Galectin-1 with Heteromultivalent Paramagnetic Liposomes for Combined MR Imaging and Treatment of Angiogenesis. Nano Letters 10(1):52-58. Knezevic, V., C. Leethanakul, V.E. Bichsel, J.M. Worth, V.V. Prabhu, J.S. Gutkind, L.A. Liotta, P.J. Munson, E.F. Petricoin, andD.B. Krizman. 2001. Proteomic profiling of the cancer microenvironment by antibody arrays. Proteomics 1(10):1271-1278. Kwong, G.A., C.G. Radu, K. Hwang, C.J.Y. Shu, C. Ma, R.C. Koya, B. CominAnduix, S.R. Hadrup, R.C. Bailey, O.N. Witte, T.N. Schumacher, A. Ribas, andJ.R. Heath. 2009. Modular Nucleic Acid Assembled p/MHC Microarrays
175
for Multiplexed Sorting of Antigen-Specific T Cells. J. Am. Chem. Soc. 131(28):9695-9703. Laginha, K., D. Mumbengegwi, andT. Allen. 2005. Liposomes targeted via two different antibodies: Assay, B-cell binding and cytotoxicity. Biochimica Et Biophysica Acta-Biomembranes 1711(1):25-32. Laurell, T., L. Wallman, andJ. Nilsson. 1999. Design and development of a silicon microfabricated flow-through dispenser for on-line picolitre sample handling. J. Micromech. Microeng 9:369-376. Lee, C.-S., andB.-G. Kim. 2002. Improvement of protein stability in protein microarrays. Biotechnology Letters 24:839-844. Liu, Y.S., C.M. Li, L. Yu, andP. Chen. 2007. Optimization of printing buffer for protein microarrays based on aldehyde-modified glass slides. Frontiers in Bioscience 12:3768-3773. Lockhart, D.J., H.L. Dong, M.C. Byrne, M.T. Follettie, M.V. Gallo, M.S. Chee, M. Mittmann, C.W. Wang, M. Kobayashi, H. Horton, andE.L. Brown. 1996. Expression monitoring by hybridization to high-density oligonucleotide arrays. Nature Biotechnology 14(13):1675-1680. Macbeath, G., andS.L. Schreiber. 2000. Printing proteins as microarrays for highthroughput function determination. science 289:1760-1763. Martin, B.D., B.P. Gaber, C.H. Patterson, andD.C. Turner. 1998. Direct Protein Microarray Fabrication Using a Hydrogel “Stamper”. Langmuir 14:3971-3975. Mastrobattista, E., G.A. Koning, andG. Storm. 1999. Immunoliposomes for the targeted delivery of antitumor drugs. Advanced Drug Delivery Reviews 40(12):103-127. Matsui, K., J.J. Boniface, P.A. Reay, H. Schild, B. Fazekas De St Groth, andM.M. Davis. 1991. Low Affinity Interaction of Peptide-Mhc Complexes with T-Cell Receptors. Science 254(5039):1788-1791. Maynard, J.A., R. Myhre, andB. Roy. 2007. Microarrays in infection and immunity. Curr Opin Chem Biol 11(3):306-315. Morozov, V.N., andT.Y. Morozova. 1999. Electrospray Deposition as a Method for Mass Fabrication of Mono- and Multicomponent Microarrays of Biological and Biologically Active Substances. Analytical chemistry 71:3110-3117. Mukhopadhyay, R. 2006. The versatility of microarrayers. Analytical chemistry 78(17):5969-5972. Nelson, K.E., L. Gamble, L.S. Jung, M.S. Boeckl, E. Naeemi, S.L. Golledge, T. Sasaki, D.G. Castner, C.T. Campbell, andP.S. Stayton. 2001. Surface
176
characterization of mixed self-assembled monolayers designed for streptavidin immobilization. Langmuir 17(9):2807-2816. Newell, E.W., L.O. Klein, W. Yu, andM.M. Davis. 2009. Simultaneous detection of many T-cell specificities using combinatorial tetramer staining. Nature Methods 6(7):497-U438. Nimrichter, L., A. Gargir, M. Gortler, R.T. Altstock, A. Shtevi, O. Weisshaus, E. Fire, N. Dotan, andR.L. Schnaar. 2004. Intact cell adhesion to glycan microarrays. Glycobiology 14(2):197-203. Nishioka, G.M., A.A. Markey, andC.K. Holloway. 2004. Protein damage in drop-ondemand printers. J. Am. Chem. Soc. 126(50):16320-16321. Oelke, M., andJ.P. Schneck. 2004. HLA-Ig-based artificial antigen-presenting cells: setting the terms of engagement. Clinical Immunology 110(3):243-251. Oelke, M., M.V. Maus, D. Didiano, C.H. June, A. Mackensen, andJ.P. Schneck. 2003. Ex vivo induction and expansion of antigen-specific cytotoxic T cells by HLA-Ig-coated artificial antigen-presenting cells. Nature Medicine 9(5):619624. Olsen, A.C., L.O. Pedersen, A.S. Hansen, M.H. Nissen, M. Olsen, P.R. Hansen, A. Holm, andS. Buus. 1994. A Quantitative Assay to Measure the Interaction between Immunogenic Peptides and Purified Class-I Major Histocompatibility Complex-Molecules. Eur. J. Immunol. 24(2):385-392. Olsen, A.W., P.R. Hansen, A. Holm, andP. Andersen. 2000. Efficient protection against Mycobacterium tuberculosis by vaccination with a single subdominant epitope from the ESAT-6 antigen. Eur. J. Immunol. 30(6):1724-1732. Oukka, M., J.C. Manuguerra, N. Livaditis, S. Tourdot, N. Riche, I. Vergnon, P. Cordopatis, andK. Kosmatopoulos. 1996. Protection against lethal viral infection by vaccination with nonimmunodominant peptides. Journal of Immunology 157(7):3039-3045. Parham, P., C.J. Barnstable, andW.F. Bodmer. 1979. Use of a Monoclonal Antibody (W6-32) in Structural Studies of Hla-a,B,C Antigens. Journal of Immunology 123(1):342-349. Prime, K.L., andG.M. Whitesides. 1993. Adsorption of Proteins onto Surfaces Containing End-Attached Oligo(Ethylene Oxide) - a Model System Using Self-Assembled Monolayers. J. Am. Chem. Soc. 115(23):10714-10721. Rabinowitz, J.D., M.N. Liang, K. Tate, C. Lee, C. Beeson, andH.M. McConnell. 1997. Specific T cell recognition of kinetic isomers in the binding of peptide to class II major histocompatibility complex. Proceedings of the National Academy of Sciences of the United States of America 94(16):8702-8707.
177
Rodenko, B., M. Toebes, S.R. Hadrup, W.J.E. van Esch, A.M. Molenaar, T.N.M. Schumacher, andH. Ovaa. 2006. Generation of peptide-MHC class I complexes through UV-mediated ligand exchange. Nature Protocols 1(3):1120-1132. Romero, P., J.C. Cerottini, andG.A. Waanders. 1998. Novel methods to monitor antigen-specific cytotoxic T-cell responses in cancer immunotherapy. Mol. Med. Today 4(7):305-312. Rowe, C.A., L.M. Tender, M.J. Feldstein, J.P. Golden, S.B. Scruggs, B.D. MacCraith, J.J. Cras, andF.S. Ligler. 1999. Array biosensor for simultaneous identification of bacterial, viral, and protein analytes. Analytical Chemistry 71(17):38463852. Sapra, P., andT.M. Allen. 2002. Internalizing antibodies are necessary for improved therapeutic efficacy of antibody-targeted liposomal drugs. Cancer Research 62(24):7190-7194. Scheibenbogen, C., P. Romero, L. Rivoltini, W. Herr, A. Schmittel, J.C. Cerottini, T. Woelfel, A.M.M. Eggermont, andU. Keilholz. 2000. Quantitation of antigenreactive T cells in peripheral blood by IFN gamma-ELISPOT assay and chromium-release assay: a four-centre comparative trial. Journal of Immunological Methods 244(1-2):81-89. Schena, M., D. Shalon, R.W. Davis, andP.O. Brown. 1995. Quantitative Monitoring of Gene-Expression Patterns with a Complementary-DNA Microarray. Science 270(5235):467-470. Schneck, J.P., J.E. Slansky, S.M. O'Herrin, and T.F. Greten. 2005. Monitoring antigen-specific T cells using MHC-Ig dimers. In Short protocols in immunology: a compendium of methods from current protocols in immunology. Coligan JE, Bierer BE, Margulies DH, Sherach EM, Strober N, editors. Wiley. 15-12 - 15-11. Schroeder, H., B. Ellinger, C.F.W. Becker, H. Waldmann, andC.M. Niemeyer. 2007. Generation of live-cell microarrays by means of DNA-directed immobilization of specific cell-surface ligands. Angewandte Chemie-International Edition 46(22):4180-4183. Schutz, C., M. Fleck, A. Mackensen, A. Zoso, D. Halbritter, J.P. Schneck, andM. Oelke. 2008. Killer artificial antigen-presenting cells: a novel strategy to delete specific T cells. Blood 111(7):3546-3552. Service, R.F. 2001. Protein chips - Searching for recipes for protein chips. Science 294(5549):2080-2082. Sette, A., J. Sidney, M.F. Delguercio, S. Southwood, J. Ruppert, C. Dahlberg, H.M. Grey, andR.T. Kubo. 1994. Peptide Binding to the Most Frequent Hla-a Class-
178
I Alleles Measured by Quantitative Molecular-Binding Assays. Molecular Immunology 31(11):813-822. Silin, V.I., E.A. Karlik, K.D. Ridge, andD.J. Vanderah. 2006. Development of surface-based assays for transmembrane proteins: Selective immobilization of functional CCR5, a G protein-coupled receptor. Analytical Biochemistry 349(2):247-253. Silva, J.M., H. Mizuno, A. Brady, R. Lucito, andG.J. Hannon. 2004. RNA interference microarrays: High-throughput loss-of-function genetics in mammalian cells. Proceedings of the National Academy of Sciences of the United States of America 101(17):6548-6552. Snapyan, M., M. Lecocq, L. Guevel, M.C. Arnaud, A. Ghochikyan, andV. Sakanyan. 2003. Dissecting DNA-protein and protein-protein interactions involved in bacterial transcriptional regulation by a sensitive protein array method combining a near-infrared fluorescence detection. Proteomics 3(5):647-657. Soen, Y., D.S. Chen, D.L. Kraft, M.M. Davis, andP.O. Brown. 2003. Detection and characterization of cellular immune responses using peptide-MHC microarrays. Plos biology 1(3):429-438. Soen, Y., A. Mori, T.D. Palmer, andP.O. Brown. 2006. Exploring the regulation of human neural precursor cell differentiation using arrays of signaling microenvironments. Mol Syst Biol 2:37. Spinke, J., M. Liley, H.J. Guder, L. Angermaier, andW. Knoll. 1993. Molecular Recognition at Self-Assembled Monolayers - the Construction of Multicomponent Multilayers. Langmuir 9(7):1821-1825. Stone, J.D., J. Walter E. Demkowicz, andL.J. Stern. 2005. HLA-restricted epitope identification and detection of functional T cell responses by using MHCpeptide and costimualtory microarrays PNAS 102:3744-3749. Silin, V.I., E.A. Karlik, K.D. Ridge, andD.J. Vanderah. 2006. Development of surface-based assays for transmembrane proteins: Selective immobilization of functional CCR5, a G protein-coupled receptor. Analytical Biochemistry 349(2):247-253. Spinke, J., M. Liley, H.J. Guder, L. Angermaier, andW. Knoll. 1993a. Molecular Recognition at Self-Assembled Monolayers - the Construction of Multicomponent Multilayers. Langmuir 9(7):1821-1825. Tanabe, M., M. Sekimata, S. Ferrone, andM. Takiguchi. 1992. Structural and Functional-Analysis of Monomorphic Determinants Recognized by Monoclonal-Antibodies Reacting with the Hla Class-I Alpha-3 Domain. Journal of Immunology 148(10):3202-3209.
179
Tan, L.C., N. Gudgeon, N.E. Annels, P. Hansasuta, C.A. O'Callaghan, S. RowlandJones, A.J. McMichael, A.B. Rickinson, andM.F.C. Callan. 1999. A reevaluation of the frequency of CD8(+) T cells specific for EBV in healthy virus carriers. Journal of Immunology 162(3):1827-1835. Tao, S.C., Y. Li, J. Zhou, J. Qian, R.L. Schnaar, Y. Zhang, I.J. Goldstein, H. Zhu, andJ.P. Schneck. 2008. Lectin microarrays identify cell-specific and functionally significant cell surface glycan markers. Glycobiology 18(10):761769. Tateno, H., N. Uchiyama, A. Kuno, A. Togayachi, T. Sato, H. Narimatsu, andJ. Hirabayashi. 2007. A novel strategy for mammalian cell surface glycome profiling using lectin microarray. Glycobiology 17(10):1138-1146. Toebes, M., M. Coccoris, A. Bins, B. Rodenko, R. Gomez, N.J. Nieuwkoop, W. van de Kasteele, G.F. Rimmelzwaan, J.B.A.G. Haanen, H. Ovaa, andT.N.M. Schumacher. 2006. Design and use of conditional MHC class I ligands. Nature Medicine 12(2):246-251. Tsai, C.J., S. Kumar, B.Y. Ma, andR. Nussinov. 1999. Folding funnels, binding funnels, and protein function. Protein Science 8(6):1181-1190. Utz, P.J. 2005. Protein arrays for studying blood cells and their secreted products. Immunological Reviews 204:264-282. Valitutti, S., M. Dessing, K. Aktories, H. Gallati, andA. Lanzavecchia. 1995a. Sustained Signaling Leading to T-Cell Activation Results from Prolonged TCell Receptor Occupancy - Role of T-Cell Actin Cytoskeleton. Journal of Experimental Medicine 181(2):577-584. Valitutti, S., S. Muller, M. Cella, E. Padovan, andA. Lanzavecchia. 1995b. Serial Triggering of Many T-Cell Receptors by a Few Peptide-Mhc Complexes. Nature 375(6527):148-151. VanderBurg, S.H., E. Ras, J.W. Drijfhout, W.E. Benckhuijsen, A.J.A. Bremers, C.J.M. Melief, andW.M. Kast. 1995. An HLA class I peptide-binding assay based on competition for binding to class I molecules on intact human B cells - Identification of conserved HIV-1 polymerase peptides binding to HLAA*0301. Human Immunology 44(4):189-198. Vanhecke, D., andM. Janitz. 2004. High-throughput gene silencing using cell arrays. Oncogene 23(51):8353-8358. Viola, A., andA. Lanzavecchia. 1996. T cell activation determined by T cell receptor number and tunable thresholds. Science 273(5271):104-106. Walker, E.B., andM.L.N. Disis. 2003. Monitoring immune responses in cancer patients receiving tumor vaccines. International Reviews of Immunology 22(34):283-319.
180
Ware, R., H. Jiang, N. Braunstein, J. Kent, E. Wiener, andC. Leonard. 1995. Human Cd8(+) T-Lymphocyte Clones Specific for T-Cell Receptor V-Beta Families Expressed on Autologous Cd4(+) T-Cells. Immunity 2(2):177-184. Winssinger, N., S. Ficarro, P.G. Schultz, andJ.L. Harris. 2002. Profiling protein function with small molecule microarrays. Proceedings of the National Academy of Sciences of the United States of America 99(17):11139-11144. Wu, P., andD.W. Grainger. 2006. Comparison of hydroxylated print additives on antibody microarray performance. Journal of proteome research 5(11):29562965. Yewdell, J.T.W., andJ.R. Bennink. 1999. Immunodominance in major histocompatibility complex class I-restricted T lymphocyte responses. Annual Review of Immunology 17:51-88. Yewdell, J.W. 2006. Confronting complexity: Real-world immunodominance in antiviral CD8(+) T cell responses. Immunity 25(4):533-543. Zhao, X.B., R. Lapalombella, T. Joshi, C. Cheney, A. Gowda, M.S. HaydenLedbetter, P.R. Baurn, T.S. Lin, D. Jarjoura, A. Lehman, D. Kussewitt, R.J. Lee, M.A. Caligiuri, S. Tridandapani, N. Muthusamy, andJ.C. Byrd. 2007. Targeting CD37-positive lymphoid malignancies with a novel engineered small modular immunopharmaceutical. Blood 110(7):2569-2577. Zheng, O., Z. Takáts, T.A. Blake, B. Gologan, A.J. Guymon, J.M. Wiseman, J.C. Oliver, V.J. Davisson, andR.G. Cooks. 2003. Preparing Protein Microarrays by Soft-Landing of mass-selected ions. Science 301:1351-1354. Zheng, T., D. Peelen, andL.M. Smith. 2005. Lectin arrays for profiling cell surface carbohydrate expression. J Am Chem Soc 127(28):9982-9983. Zhu, H., M. Bilgin, R. Bangham, D. Hall, A. Casamayor, P. Bertone, N. Lan, R. Jansen, S. Bidlingmaier, T. Houfek, T. Mitchell, P. Miller, R.A. Dean, M. Gerstein, andM. Snyder. 2001. Global analysis of protein activities using proteome chips. science 293:2101-2105. Zhu, H., andM. Snyder. 2003. Protein chip technology. current opinion in chemical biology 7:55-63. Ziauddin, J., andD.M. Sabatini. 2001. Microarrays of cells expressing defined cDNAs. Nature 411(6833):107-110.
181
APPENDIX A ANTI-DEXTRAN ANTIBODY MICROARRAYS FOR CHARACTERIZATION OF FUNCTIONALIZED MAGNETIC NANOPARTICLES
Functionalized magnetic nanoparticles have been widely used in cell sorting, cell characterization, biomarker identification, and cancer diagnostics. However, to date there have not been established methods for validating the chemically-coupled functional molecules such as antibodies on magnetic nanoparticles. Traditional direct protein assays, such as BCA (bicinchoninic acid), Lowry, Bradford, etc., are insensitive for samples of small sizes and incompatible with solid phase particles due to possible light scattering from the particles. Indirect assays, such as cellular assays using flow-activated cell sorting (FACS), are not quantitative, labor intensive, and expensive. Here we report a novel microarray-based assay for fast and cost-effective characterization of functionalized magnetic nanoparticles.
182
The basis of this technology is the capture of magnetic nanoparticles via a printed or coated monoclonal anti-dextran antibody on Nexterion slide H surface, taking advantage of the frequently-used dextran coating of the particles. Specifically, the anti-ddextran antibody used in this study was developed against dextrans conjugated with a saturated C18 fatty acid chain. The light microscopic image in Figure A.1 shows a typical anti-dextran microarray spot that has nanoparticles uniformly immobilized on the surface.
Figure A.1 Immobilization of anti-PE conjugated magnetic nanoparticles on an anti-dextran spot. A Nexterion slide H was pre-coated with anti-dextran antibody solution (0.5 mg/ml) for 1 hr at room temperature (RT). After wash (with PBST followed by PBS), the slide was blocked in ethanolamine blocking buffer for 1 hr at RT. After wash, the slide was printed with anti-PE conjugated magnetic nanoparticles at an iron concentration of 0.17 mg/ml. Each spot only contained 0.33 nl of particle solution and was about 150 µm in diameter. The nanoparticles were analyzed by dynamic light scattering to have average hydrodynamic diameters of approximately 200 nm.
183
Figure A.2 below demonstrates the ability of the nanoparticle microarrays for detecting/estimating the amount of ligand coupled to the nanoparticles. Spots printed with uncoupled beads showed very low fluorescence, indicating the negligible levels of nonspecific binding of labeled antibodies to the immobilized particles or microarray surface. For spots printed with nanoparticles conjugated with 2.5 mg/ml neutravidin, the mean spot intensity across the spot replicates was 8283, whereas for those printed with particles conjugated with 5.0 mg/ml neutravidin the mean spot intensity was 13979, almost doubled. In the future, the inclusion of standard nanoparticle samples with known amounts of ligands would allow the direct quantification of unknown samples in terms of ligand amount. The ability of the nanoparticle microarray for detecting conjugated ligands can also be used to compare different coupling chemistries (results not shown).
Stemcell beads,uncoupled
beads,uncoupled
Miltenyi beads,HEA
beads,5.0mg/ml neutravidin
Miltenyi beads,anti-PE
beads,2.5mg/ml neutravidin
Miltenyi beads,AC133
beads,2.5mg/ml anti-biotin
0.0
0.0
5000.0 10000.0 15000.0 Mean spot intensity
5000.0 10000.0 15000.0 Mean Spot intensity
Figure A.2 Application of the nanoparticle microarray for ligand conjugation titer. Left: Commercial nanoparticle spots incubated/stained with fluorescence-labeled anti-mouse IgG antibody; right: CNW (Columbus Nanoworks Inc.) nanoparticle spots incubated/stained with first biotin-anti-PE solution (20 µg/ml) followed by PE-labeled anti-CD45 solution (50 µg/ml). Refer to Figure A.1 for microarray fabrication and development procedures. For commercial nanoparticles, iron concentration was not pre-determined, so the particles were just printed undiluted; for CNW nanoparticles, all particles were adjusted to 0.1 mg/ml iron concentration for printing. Uncoupled beads, i.e. beads not conjugated with any ligands, were included as negative control. Each sample was printed in 5 x 5 spot replicates.
184
In addition, we have demonstrated for the first time the capture of live human CD45+ cells onto neutravidin nanoparticle spots (Figure A.3). We believe that, with proper integration with magnetic properties of the nanoparticles, this nanoparticle microarray can be developed into a new generation cell sorter.
Figure A.3 Potential application of the array as small scale cell sorter. Human CD45+ cells were captured onto neutravidin-conjugated nanoparticle spots, via biotinylated anti-PE and PE-anti-CD45.
185
