Deep-tissue multi-photon fluorescence lifetime microscopy for intravital imaging of protein-protein interactions Fruhwirth G.O.1*, Matthews D. R.1, Brock A.1, Keppler M.1, Vojnovic B.1,2, Ng T.1, Ameer-Beg S.1 1 The Richard Dimbleby Department of Cancer Studies, King’s College London, London, UK 2 The Gray Institute for Radiation Oncology and Biology, University of Oxford, Oxford, UK ABSTRACT Fluorescent lifetime imaging microscopy (FLIM) has proven to be a valuable tool in beating the Rayleigh criterion for light microscopy by measuring Förster resonance energy transfer (FRET) between two fluorophores. Applying multiphoton FLIM, we previously showed in a human breast cancer cell line that recycling of a membrane receptorgreen fluorescent protein fusion is enhanced concomitantly with the formation of a receptor:protein kinase C α complex in the endosomal compartment. We have extended this established technique to probe direct protein-protein interactions also in vivo. Therefore, we used various expressible fluorescent tags fused to membrane receptor molecules in order to generate stable two-colour breast carcinoma cell lines via controlled retroviral infection. We used these cell lines for establishing a xenograft tumour model in immune-compromised Nude mice. Using this animal model in conjunction with scanning Ti:Sapphire laser-based two-photon excitation, we established deep-tissue multiphoton FLIM in vivo. For the first time, this novel technique enables us to directly assess donor fluorescence lifetime changes in vivo and we show the application of this method for intravital imaging of direct protein-protein interactions. Keywords:
Deep-tissue imaging; fluorescence lifetime microscopy; FRET; intravital microscopy; breast cancer;
1. INTRODUCTION Multi-photon excitation has been extensively used for deep-tissue imaging in vivo1, 2, but its use has so far been predominantly restricted to intensity-based fluorescence imaging applications with some notable exceptions3-5. Intensity based fluorescence measurements have several limitations, e.g. in the quantification of the intensity signals, the spatial resolution (wavelength of used light), or applications dependent on repeated imaging (photo-bleaching, photo-toxicity). In contrast, fluorescence lifetime measurements have the advantage that they are independent of fluorescence intensities while still being very sensitive to environmental effects (e.g. pH, oxygen concentration). More importantly, FLIM measurements allow for the determination of molecular interactions between fluorophores and thus, improve the effective spatial resolution to the low nanometre scale6-9. Fluorescence resonance energy transfer (FRET), the phenomenon of dipole-dipole coupling followed by non-radiative energy transfer between an excited donor fluorophore and an acceptor molecule, is strongly dependent on the distance (1/r6) of these molecules (usually 1-10 nm). By combining multi-photon excitation with time-correlated single photon counting for the measurement of fluorescence lifetimes, we aimed at developing a novel imaging technique that enabled us to measure fluorescence lifetimes in vivo and thus allowed for the assessment of more parameters on a much improved scale as compared to state-of-the-art intensity-based deep-tissue imaging techniques. The particular advantage of multiphoton imaging is conferred by the use of intense near-infrared (NIR) light to induce non-linear absorption in a probe fluorophore. The intensity dependence of the non-linear absorption confines the excitation to the focal plane of the imaging lens. This inherent sectioning capability allows the collection of 3dimensional data without the use of a confocal aperture, since fluorescence is generated only in the focal volume allowing simplified detection systems to be employed10. Since the NIR is inherently scattered to a lesser degree than the visible (Rayleigh scattering ∝ λ-4) and linear absorption of the NIR is minimal for most biological applications, thick (< 500 μm) samples may be imaged11. Photobleaching and photodamage are also greatly reduced due to the confinement of excitation to the focal volume12. Current technological limitations restrict the application of in vivo MPM to readily accessible sites such as the skin13, the eye14 and to animal models such as the skinfold2 or cranial15 window chamber. Application of the technique to window *
[email protected]; phone: +44(0)2078476221 Multiphoton Microscopy in the Biomedical Sciences IX, edited by Ammasi Periasamy, Peter T. C. So, Proc. of SPIE Vol. 7183, 71830L · © 2009 SPIE · CCC code: 1605-7422/09/$18 · doi: 10.1117/12.817129
Proc. of SPIE Vol. 7183 71830L-1
chamber models of tumour-bearing rodents shows particular promise as an adjunct to established techniques such as intra-vital microscopy and conventional histology. The advantage lies in the three-dimensional resolving power of MPM at depth within the model for determination of structure and function. In addition, experiments to follow tumour development, angiogenesis, treatment and vascular remodelling, as a result, are possible with multiple imaging timepoints without compromising the model. One of the recognised problems with a skin-fold window chamber model for tumors is that the model is restricted to growth predominantly in the plane of the window and is not truly 3-dimensional beyond a few hundred microns restricting the total tumor mass and the volume that can be probed with the microscope. Geometrical constraints also make high-NA imaging somewhat impractical. For this application, we use a tumour skin flip model for short duration imaging in vivo16, 17. In this case, the tumor is grown without artificial compression, enabling formation of fully 3-dimensional architecture. In this paper, we describe the methodology for development of a practical protein-protein FRET experiment in vivo using multiphoton fluorescence lifetime imaging.
2. MATERIALS AND METHODS 2.1.
Chemicals CXCL12 was from Peprotech (London, UK), Mowiol was from ICN (CA, US), and all standard chemicals were either from Sigma-Aldrich (Gillingham, UK) or VWR (Lutterworth, UK). 2.2.
Plasmids, cell culture, and generation of stable cell lines Human CXCR4 was fused to the N-terminus of EGFP or TagRFP by subcloning its coding sequence between the HindIII and EcoR1 sites of either pEGFP-N1 (Clontech, Saint-Germain-en-Laye, France) or pTagRFP-N1 (Evrogen, Moscow, Russia). The sequences of these C-terminal fusion proteins were subcloned into the retroviral expression vectors pLPCX or pLHCX (Clontech, Saint-Germain-en-Laye, France) and the constructs were confirmed by sequence analysis. Mammary carcinoma cell lines (MTLn3E) stably expressing human CXCR4-EGFP (3E.X4G) were obtained after retroviral infection, selection with puromycine (1 µg/mL), and fluorescence-activated cell sorting (FACS) in order to obtain single clones. A cell line expressing both fluorescent proteins (3E.X4G.X4T) was generated via sequential retroviral infection, selection and FACS-based single clone selection. One single clone of each type was selected and used for all further experiments. All mammary carcinoma cell lines were cultured in αMEM supplemented with 5% foetal bovine serum, penicillin/streptomycin (100 IU), and L-glutamine in an atmosphere containing 5% CO2 (v/v) either in the presence or the absence of the respective selection antibiotic. 2.3.
Establishment of a xenograft tumour and animal preparation for in vivo imaging 3E.X4G or 3E.X4G.X4T cells were trypsinised, washed, adjusted to a cell density of 2·107 cells/mL in PBS, and 6 10 cells were injected into the mammary fat pad of five to six weeks-old female Balb/c Nude mice (Charles River, Margate, UK). After establishment of tumours, their growth was monitored and when they reached sizes of around 1 cm3, the mice were prepared for imaging. To this end, the animals were anaesthetised with isoflurane in pure oxygen (max. 4% v/v) and placed onto a bespoke microscope stage insert (Fig.1) that had been pre-warmed to 37°C. Surgery was necessary in order to enable the tumour to be flipped-out onto an area accessible to the light path of the microscope. This was achieved by making a small crescent-shaped incision into the skin on the abdomen of the mouse while avoiding cutting through any major blood vessels. After removal of small amounts of connective tissue and fat, the tumour could be flipped out onto a glass cover slip of the bespoke stage insert. Subsequently to restraining the tumour and the mouse, the stage insert was placed onto a thermo-controlled purpose-built fluorescence lifetime microscope as described below. Immediately after imaging, all animals were culled and if desired, tissue was removed and immediately immersed in OCT and frozen in liquid nitrogen. Sections (5µm) were cut and stored at -80°C until analysis. All animal experiments were approved by the Home Office (UK Government) and an Ethical Review Panel and all requirements for humane animal treatment dictated by UK legislation and the above committee were met. 2.4. Fixing of cells and tissue sections for multi-photon lifetime microscopy Cells were stimulated with recombinant human CXCL12 at a final concentration of 100nM for 20 min at 37°C or treated with the vehicle solution, then fixed immediately (4% PFA, 15 min) and permeabilised with a 0.2% v/v Triton X-100 solution. Tissue sections were thawed, rinsed with PBS and fixed (4% PFA, 20 min). Cells and tissue sections were then treated with NaBH4 (1 mg/mL) and mounted in Mowiol containing Dabco as an antifade.
Proc. of SPIE Vol. 7183 71830L-2
B
xy scanner Beam size Intensity
Condenser clamp
N
Tunable pulsed NIR laser (700-1000 nm) set to 890nm
Time
Collection lens Dichroic
A -- 1"mirror n
\LJPMT PMT
Fig.1
IR heater
Heater pa.-
Microscope setup for DT-FLIM. (A) Schematic drawing of the light path and the main components of the microscope. A pulsed Ti:sapphire laser tuned to 890 nm (pulse width=250 fs, pulse interval= 13 ns) is used for two-photon excitation of EGFP and for the generation of a second harmonic signal from highly ordered structures. Two photomultiplier tubes (PMT) equipped with appropriate filter sets (see section 2.5) for the separation of the SH and the GFP channels are used for data collection. The PMTs are connected to a TCSPC board and bespoke software to determine the fluorescence lifetimes. (B) Drawing of the modifications of the microscope necessary to accommodate for keeping an anaesthetised animal on the microscope stage. Temperature control is of outstanding importance under anaesthesia and achieved by a heater pad incorporated into the bespoke stage insert and an IR-heater mounted to the condenser of the microscope. Both heating devices are controlled via thermo-couples reporting on rectal and surface temperature of the anaesthetised animal, respectively.
2.5.
Time-resolved fluorescence lifetime microscopy All FLIM measurements of fixed samples were undertaken on an inverted Nikon TE2000 fluorescence microscope equipped with a 40x/1.3NA Nikon Plan-Fluor oil objective lens (Nikon, Kingston upon Thames, UK) on a modified multi-photon microscopy system (Fig.1) which has been described previously3, 18. Each area selected for imaging was consecutively imaged in wide-field epi-fluorescence mode and two-photon mode for FLIM measurements. In the wide-field epi-fluorescence mode EGFP or TagRFP images were acquired using the FITC (Nikon B-2 E/C cube) or the Cy3 channel (Nikon Cy3 HYQ cube). Time-resolved detection was afforded by the addition of a non-descanned detection channel with a fast photomultiplier and SPC700 time-correlated single-photon counting electronics (TCSPC, Becker and Hickl GmbH, Berlin, Germany). A 510 ± 10 nm band-pass filter (Chroma Technology, Rockingham, VT), was used in the detector channel (Coherent, Santa Clara, CA). Laser power was adjusted to give average photon counting rates of the order 104– 105 photons·s-1 (0.0001–0.001 photons/excitation event) and with peak rates approaching 106 photons·s-1, below the maximum counting rate afforded by the TCSPC electronics to avoid pulse pile-up. Acquisition times of the order of 300 s at low excitation power were used to achieve sufficient photon statistics for fitting, while avoiding either pulse pile-up or significant photo-bleaching. For excitation of EGFP a Ti:sapphire laser (Coherent, Santa Clara, CA) was tuned to 890 nm and its power output was adjusted to ~0.5 W. For deep-tissue two-photon FLIM measurements several modifications to the system described above were required. For the detection of second harmonics generated by collagen upon excitation with the Ti:sapphire laser, a second detector channel was fitted and equipped with a 435 ± 15nm (Chroma Technology, Rockingham, VT) in the detection pathway. Furthermore, a 20x/0.5NA Nikon Fluor air objective lens (Nikon, Kingston upon Thames, UK) with
Proc. of SPIE Vol. 7183 71830L-3
a working distance of 2.1 mm was used for imaging. In addition, a telescope was fitted into the excitation light path in order to optimize the beam diameter at the back of the objective lens for maximum resolution while maintaining a high power output (0.5-1 W). For in vivo animal work, additional equipment for online temperature and anaesthesia control and animal monitoring was fitted. All time-resolved FLIM data were analysed by a bespoke analysis program as described previously19, 20.
3. RESULTS AND DISCUSSION In order to develop deep-tissue time-resolved fluorescence lifetime microscopy (DT-FLIM) by the combination of state-of-the-art deep tissue imaging with time-correlated single photon counting (TCSCP) techniques for the measurement of fluorescence lifetimes in vivo, we first needed to chose an appropriate biological system. Preferably, such a system should involve multimerization of the proteins of interest for easy detection of FRET and, at the same time, involve at least some extent of spatial separation of non-interacting and interacting species. A plasma membrane receptor that became internalised upon stimulation with its ligand served as a favourable model system for our purpose. We chose the chemokine receptor CXCR4, a G protein-coupled receptor, which is expressed at the plasma membrane in multiple cell lines in vivo21. In addition, only one natural ligand, CXCL12, has been discovered for this receptor, which yields to high specificity of the observed changes in our chosen model. Furthermore, the CXCL12-CXCR4 axis is of outstanding importance in the physiology of the mammalian life cycle, especially throughout development and in the immune system22. There are also several important pathological conditions in which this ligand-receptor axis is highly important; for instance, it serves as a co-receptor in HIV-infection23, 24, is an important player in various types of cancer 25-27, and is involved in the WHIM syndrome28. Several practical considerations led us to establish a xenograft tumour model of breast cancer in immuno-compromised mice. Among these were, for instance, the ease and time necessary for generating the model as well as the size, availability and cost of the animals, while still using a technique flexible enough to be easily transferred to other in vivo models. The chosen cell line for generating the xenograft was a rat mammary adenocarcinoma cell line (MTLn3, 29), which is a metastatic breast cancer cell line that was shown to efficiently grow solid xenograft tumours in mammals. We generated fluorescent fusion proteins consisting of CXCR4 and either monomeric green fluorescent protein alone (EGFP) or EGFP and Tag RFP, which is a monomeric red fluorescent protein variant of DsRed that was recently shown to serve as a FRET donor if paired with EGFP30. We then confirmed the function of these receptors by comparing them to the untagged wildtype receptor in cell based receptor down-regulation, chemotaxis and GPCR signalling assays (data not shown). Since there were no major differences in all of these parameters, we proceeded to the generation of stable cell lines, which were a prerequisite for setting up a reliable xenograft model. The fluorescent CXCR4 variants were transduced into MTLn3 cells by retroviral infection and subsequently, antibiotic-assisted selection and single-cell fluorescence-activated cell sorting were used for obtaining stable clones. Ligand treatment of the cell line expressing CXCR4.EGFP alone (3E.X4G) led to the expected endocytosis of the receptors, while not significantly changing the fluorescence lifetime of the receptors (Fig.2, top panel). The fluorescence lifetimes of representative cells with and without ligand treatment for 40 min before fixation were 2.13±0.04 ns and 2.11±0.05 ns, respectively (P>0.75, n=5, FRET efficiency ~0.7%). The fluorescence lifetimes obtained for EGFP were in good agreement with the values reported previously9, 31. In the case of the double-infected cell line stably expressing both fluorescent CXCR4 variants (3E.X4G.X4T) ligand treatment led to a marked decrease of the fluorescence lifetime of EGFP (donor) while the receptors were endocytosed. In the absence of the ligand the donor lifetime (1.95±0.03 ns) was only a little lower than in the control cell line (3E.X4G) suggesting only a limited amount of receptor multimerization in untreated cells. Once the receptors were exposed to its natural ligand CXCL12, the fluorescence lifetime decreased markedly upon endocytosis of the receptors (1.57±0.03 ns). This was due to ligand-induced multimerization of both types of CXCR4 molecules increasing the FRET between the green and red fluorescent variants (Fig.2, bottom panel). The effect proved to be highly significant in the chosen cell system (P
