Proteomics and Systems Biology: Current and Future Applications in ...

2 downloads 0 Views 788KB Size Report
proteomics in nutrition research, and discuss the challenges for future applications of systems biology approaches in the nutritional sciences. Adv. Nutr.
REVIEW

Proteomics and Systems Biology: Current and Future Applications in the Nutritional Sciences1 J. Bernadette Moore2* and Mark E. Weeks3 2

Nutritional Sciences Division, Faculty of Health and Medical Sciences, University of Surrey, Guildford, Surrey, GU2 7XH, UK; and 3Veterinary Laboratories Agency, New Haw, KT15 3NB, UK

ABSTRACT

In the last decade, advances in genomics, proteomics, and metabolomics have yielded large-scale datasets that have driven an interest in global analyses, with the objective of understanding biological systems as a whole. Systems biology integrates computational modeling and experimental biology to predict and characterize the dynamic properties of biological systems, which are viewed as complex signaling networks. Whereas the systems analysis of disease-perturbed networks holds promise for identification of drug targets for therapy, equally the identified critical network nodes may be targeted through nutritional intervention in either a preventative or therapeutic fashion. As such, in the context of the nutritional sciences, it is envisioned that systems analysis of normal and nutrient-perturbed signaling networks in combination with knowledge of underlying genetic polymorphisms will lead to a future in which the health of individuals will be improved through predictive and preventative nutrition. Although high-throughput transcriptomic microarray data were initially most readily available and amenable to systems analysis, recent technological and methodological advances in MS have contributed to a linear increase in proteomic investigations. It is now commonplace for combined proteomic technologies to generate complex, multi-faceted datasets, and these will be the keystone of future systems biology research. This review will define systems biology, outline current proteomic methodologies, highlight successful applications of proteomics in nutrition research, and discuss the challenges for future applications of systems biology approaches in the nutritional sciences. Adv. Nutr. 2: 355–364, 2011.

Introduction The past decade has heralded tremendous technological and scientific advances that have made initial forays into systems biology possible. Innovations, particularly in sequencing and microarray technologies and more recently MS, have made possible the generation of comprehensive molecular datasets from a variety of biological systems (cells, tissues, biological fluids). However, with the generation of these datasets has emerged our realization that even extensive knowledge of the existing components of a system does not necessarily lead to understanding how a system functions or behaves. Complex systems are now recognized to demonstrate “emergent” behavior with function arising from the totality of system interactions, which are often nonlinear and stochastic. With the advent of systems biology, the view of signaling pathways as linear cascades funneling signals from the cell membrane to the nucleus has evolved to the concept of signaling networks, which are highly 1 Author disclosures: J. B. Moore and M. E. Weeks, no conflicts of interest. * To whom correspondence should be addressed. E-mail: [email protected].

ã2011 American Society for Nutrition. Adv. Nutr. 2: 355–364, 2011; doi:10.3945/an.111.000554.

interconnected, involve cross-talk across multiple pathways, and have both feed-forward and feed-back loops. Systems biology and the computational modeling of biological networks aims to understand both a system’s structure and functional dynamics such that system properties, like robustness, can be understood and that system behaviors in response to perturbation can be predicted (1). The complexity of the relationship between nutrition and health means that nutritional sciences research is, in many ways, ideal for the application of systems biology approaches. Ultimately, nutrients are consumed in the context of a complex dietary background, absorbed by a gut that can vary dramatically in terms of its microbiome, and metabolized in the context of a polymorphic genome with a plethora of individual, variant, nutrient-gene interactions. As reviewed in detail by de Graaf et al. (2), mathematical modeling has traditionally been applied in the nutritional sciences; examples include metabolic flux analysis (3), compartmental models (4), and whole body models of energy metabolism (5). Although predictive network modeling in response to nutrient perturbations and/or genetic polymorphisms has yet to be fully 355

utilized in molecular nutrition research, the increasing application of the tools of genomics, proteomics, and metabolomics to nutrition-related research questions means that datasets of sufficient depth and complexity now exist for systems biology computational approaches.

Systems biology With acknowledged roots in general systems theory (6) and cybernetics (7), systems biology has emerged in the past decade as a discipline that aims ultimately to understand and predict the behavior of biological systems as a whole. Biological systems, whether a signaling network, a cell, an organ, or an organism, in this field are viewed as a network of interacting elements (genes, proteins, metabolites) from which coherent function emerges. The modeling or reconstruction of biological networks allows computational simulations to be run that lead to predictive hypotheses on how a given network may behave (8). In its grandest vision, the application of systems biology approaches to the study of disease-perturbed networks will, through the identification of therapeutic drug targets, foster a future of personalized medicine (9). Current applications of systems biology are focused on characterizing the underlying network structure and dynamics of molecular interactions. Networks are distinguished as signaling, regulatory, or metabolic, which coordinate the cellular response to perturbation, alter transcription or translation, or convert reactants to products, respectively (10). Computational analyses of networks can be categorized as either qualitative or quantitative. Qualitative or structural network analysis focuses on network topology or connectivity, characterizing static properties derived from mathematical graph theory, including the paths from inputs to outputs, the total number of paths, reachability, redundancy, and cross-talk. Quantitative analyses aim to measure and model precise kinetic parameters of the components of a network while also using the properties of network connectivity. As recently applied to folate metabolism (11,12), most quantitative models use ordinary differential equations to link reactant and product concentrations through reaction rate constants. Recent developments in dynamic network modeling have focused on modeling the noise or stochastic nature of biological reactions (13); however, given the difficulty of acquiring precise kinetic parameters and the fact that most “omic” datasets are semiquantitative at best, most quantitative analyses to date have been of small-scale networks. Qualitative approaches have been applied to a variety of biological networks and several lines of data now suggest that network connectivity alone can yield accurate prediction of network dynamics (14–16). Driven in part by the needs of industrial metabolic engineering, there are now genome-scale reconstructions of metabolic networks for over 40 organisms (17). Metabolic networks are built initially through a combination of database mining and literature curation, then modeled mathematically, most typically using constraint-based models (18), and iteratively validated by testing the in silico model predictions in “wet” laboratory experiments. The utility of these metabolic reconstructions has most recently been 356 Moore and Weeks

dramatically demonstrated by the in silico prediction of novel antimicrobial targets that were experimentally verified (19). In this case, indispensible enzymatic reactions in Escherichia coli and Staphylococcus aureus were predicted by flux balance analysis using existing metabolic reconstructions, a small molecule library was screened for inhibitors to the enzymes in question, and the efficacy of the predicted inhibitors was measured in cell-based enzymatic assays. These data are particularly exciting given the current pervasiveness of antibiotic resistant strains of pathogenic bacteria. While the majority of species studied to date have been microorganisms of relevance to industrial applications or human disease, importantly, in 2007, there were 2 independent reconstructions of the human metabolic network (20,21). These models both pointed out the significant gaps in our knowledge and yielded fascinating novel insights, including predicting alternative drug targets from coupled reaction sets and suggesting greater fragility in the network than that predicted based on theories of network robustness. As pointed out by the authors (20), these models are particularly relevant to the nutritional sciences where nutrigenomic, proteomic, or metabolomic datasets can be mapped to these networks and predictively analyzed. Indeed, most recently, Zelezniak et al. (22) explored these networks in the context of gene expression datasets from human skeletal muscle in type 2 diabetes (T2D)4. Using both reconstructions of the human metabolic network and a previously developed algorithm for identifying significantly regulated metabolites (23), in combination with promoter analysis of the associated enzyme genes, they identified candidate metabolic biomarkers and transcriptional regulatory nodes associated with T2D. Their approach detected both predictable (PPAR family) and less obvious [cAMP-response element binding protein (CREB) and nuclear respiratory factor 1 (NRF1) families] transcription factors and perhaps most interestingly, this work identified both NAD+/NADH and ATP/ ADP as the top-ranking reporter metabolites. Other recent applications of systems biology of relevance to the nutritional sciences are network models of human disease (24), human metabolic diseases (25), and other models related to T2D (26,27). Using the Online Mendelian Inheritance in Man database (28) to source disease genes and genetic disorders, Goh et al. (24) created both a human disease network where diseases were linked if they shared a gene that had mutations in both disorders and a disease gene network where genes were linked if they associated with the same disorder. The surprising finding of this research was that the vast majority of disease genes (78%) were nonessential genes that were peripheral rather than central “hubs” in the network and less likely to be housekeeping genes expressed in all tissues. Although the probable evolutionary explanation that mutations in essential genes are likely to result in early lethality seems 4

Abbreviations used: 2DGE, 2-dimensional gel electrophoresis; CAD, coronary artery disease; CE-MS, capillary electrophoresis–MS; IBD, inflammatory bowel disease; iTRAQ, isobaric tags for relative and absolute quantification; LC-MS, liquid chromatography-MS; MALDI, matrix assisted laser desorption ionization; MS/MS, tandem MS; PBMC, peripheral blood mononuclear cell; SILAC, stable isotope-labeling by amino acids in cell culture; T1D, type 1 diabetes; T2D, type 2 diabetes; UPR, unfolded protein response.

intuitive retrospectively, until this work, researchers had hypothesized that disease genes would code for highly connected “hub” proteins. The human disease network also showed metabolic disorders to have less genetic heterogeneity and therefore be less connected than cancer and neurological disorders, which prompted the development and interrogation of a metabolic-specific disease network (25). Here, metabolic disorders were modeled as hubs that were connected if the disease-associated mutated enzymes were linked to either correlated or adjacent metabolic reactions. Accepting the interrelationships between the flux rates of consecutive metabolic reactions, where a defect in one reaction will change the flux of the ensuing reactions in the pathways, the authors essentially tested the hypothesis that in having one metabolic disorder, a patient is more likely to have or develop a secondary disorder that is linked to shared metabolites and correlated metabolic reactions rather than by shared genes. After establishing the topology of the human metabolic disease network, in addition to using gene coexpression and flux coupling analyses to prove the functional links between linked reactions and therefore linked metabolic diseases, the authors calculated a comorbidity index for all the metabolic diseases by using Medicare records of the hospital visits by elderly patients over a 3-y period (13 million patients and 32 million hospital visits). The data convincingly demonstrate that the metabolic disorders linked in their network are 3–7 times more likely to be comorbid. Taking a different approach, Jesmin et al. (26) examined the interrelationships between T2D, hypertension, obesity, and reactive oxygen species by building a gene regulatory network from the literature base and protein-protein interaction databases. Although somewhat limited in that this research does not go on to do any predictive modeling or experimental hypothesis testing, this work clearly illustrates the maturity of open resource tools and databases available for network modeling. Last, the work of Farres et al. (27) is noteworthy in terms of taking a systems approach to analyze the relationship between a ketogenic diet and T2D. Somewhat unusually, the authors used a commercial database and mapping software; again, this work might be criticized for being limited to hypothesis generation. Nonetheless, although there have been a couple of examples in mice (29,30), it is noteworthy for being the only paper we are aware of to date in humans to analyze the molecular interactions associated with both a dietary approach and a disease using systems biology methodologies. This and the aforementioned research are all current examples of the application and utility of systems biology to research questions of interest to the nutritional sciences. There are inherent mathematical challenges associated with modeling signaling networks (10) and constraint-based approaches, which work well for metabolic networks that can be studied at steady state, are less useful for modeling the dynamic phenomena in signal transduction networks. However, current research is tackling both the mathematical algorithms and the generation of open source software with which to implement the models (14,31). The power of predictive computational modeling of signaling networks in

particular to identify potential targets for drug (or nutrient) intervention has recently been illustrated (32). We expect that the real potential for nutritional systems biology applications will lie in predictive network modeling of the signaling response to nutrient perturbation, with models that ultimately will incorporate information regarding genetic polymorphisms that influence gene-nutrient interactions.

Proteomics Early on in the articulation of systems biology as a field, proteomics was recognized as an essential discipline for the accurate building of network models (33). While still far from how routine microarray analysis of transcriptome data has become, the past 10 y has seen a linear increase in the number of proteomic publications in PubMed (Fig. 1). Although the number of systems biology publications has also increased linearly, the absolute number of systems biology manuscripts is currently still one-third of proteomics. Interestingly, whereas the number of systems biology papers related to nutrition research has remained constant at 3–4% of the total, the percent of proteomics papers related to nutrition research is somewhat lower at 2–3% of the total (Fig. 1). Advances in MS, first in the development of ionization techniques such as matrix assisted laser desorption ionization (MALDI) (34) and electrospray ionization (35) and secondly in the ability to conduct tandem MS (MS/MS) to fragment peptides by collision-induced dissociation (36), paved the way for the development of proteomics as a field. MS-based proteomics is now the only method used for the systemic characterization of proteins from identification, quantification, and characterization of either post-translational modifications or protein interactions (37). The identification of proteins by MS can be achieved by the fragmentation of peptides or intact proteins using either a bottom-up or a top-down approach. Either approach is amenable to automation, although most high-throughput global proteomic analytical programs have favored the bottom-up approach (38) primarily because, compared to proteins, peptides are relatively easy to handle and their physiochemical properties are more uniform (39). However, the analysis of complex biological matrices by MS is highly dependent on off-line separation technologies such as 2-dimensional gel electrophoresis (2DGE) (40) or HPLC that simplify such samples prior to mass analysis; and HPLC is also the standard front end on-line separation methodology for many liquid chromatography-MS (LC-MS) based instrumentation platforms (41). Although the merits of 2DGE have been debated (42) relative to multidimensional LC-MS, or “shotgun” proteomics, each approach has its advantages and disadvantages and both separation strategies are widely used. Whereas the success of bottom-up proteomic approaches has primarily been driven by the successful application of collision-induced dissociation fragmentation to peptide MS and it has had a central role in shotgun proteomics (43,44), instrument-based technological advances more recently have allowed alternate fragmentation technologies such as electron capture dissociation to be revisited as Proteomics and systems biology in the nutritional sciences 357

Figure 1 Growth in proteomic and systems biology publications in last decade. (A) The number of proteomics publications has risen rapidly in the last decade. The percentage of proteomics publications related to nutritional sciences research has remained constant at 2–3% of the total. (B) The number of systems biology publications has risen rapidly in the last decade. The percentage of these related to nutritional sciences research has remained constant at 3–4% of the total. These data were generated by performing a Pubmed [All Fields] search for either “proteomics”/ “systems biology,” or [“proteomics”/”systems biology” AND (nutrition OR obesity OR diabetes OR “cardiovascular disease”)] with the requisite publication dates.

alternate and complementary fragmentation technologies (45,46). This, in association with the development of highresolution mass analyzers such as the Orbitrap (47,48) and high resolution time-of-flight mass spectrometers, has driven the intact molecular mass measurement of proteins and supported an increased interest in top-down proteomic analysis (49). Most recently, advances in the identification of posttranslational modifications have been driven by the coupling of ion wave technology to MALDI-MS platforms, which allows the differentiation of identical mass peptide species by virtue of their 3-dimensional structure (50,51). With ongoing developments in specificity, sensitivity, fragmentation chemistry, and a variety of platforms and strategies, MS is now capable of protein analysis on a global scale. Quantitative proteomics Numerous strategies are being developed for the quantitative analysis of peptide, protein, and enzyme activity to 358 Moore and Weeks

fundamentally explore a wide range of biological questions as part of a systems biology approach. Biomarker discovery, in particular, is very dependent on the identification of accurate quantitative differences in proteins or peptides between normal and diseased samples. Although targeted approaches to MS-based protein quantification have recently emerged (52), our focus here is on global strategies that are most often applied to questions regarding the relative quantitation of proteins in different samples rather than questions of absolute amounts of protein in a sample. Untargeted or global quantitation strategies have evolved to cover 2 basic conceptual approaches, either the addition of a mass differentiated label to proteins/peptides of interest (differential mass tagging or isotopic labeling) or the quantitation of proteins/peptides in different sample sets by comparative analysis of spectral features (label free). A typical quantitative MS-based proteomics experimental workflow is outlined in Figure 2 showing the places where optional labeling may occur (Fig. 2). Quantitative isotope labels can be introduced metabolically, by chemical derivatization, or enzymatically (53). Stable isotope labeling by amino acids in cell culture (SILAC) is an in vivo metabolic labeling method that uses heavy and light versions of essential amino acids in growth media of metabolically active cell models. Recently, SILAC has been applied to labeling of primary cells (54) as well as intact organisms, including mice (55) and most recently drosophila (56). Examples of chemical derivatization techniques for quantitative proteomics include: isotope-coded affinity tags, which are based on the modification of cysteine thiol groups with iodoacetamide tags (57), and isobaric tags for relative and absolute quantification (iTRAQ), which are used to label newly formed N termini at endogenous proteolytic cleavage sites after blocking lysine amino groups by guanidation (58). There has also been the recent commercialization of tandem mass tags that adopt similar chemistry to iTRAQ and have been successfully used to quantitate differences in human cerebrospinal fluid samples after immunodepletion (59). This is by no means an exhaustive analysis of available labeling technologies and additional techniques such as the incorporation of 16O/18O either during or after enzymatic digestion for differential proteolytic labeling (60) and cleavable isobaric labeled affinity tags (61) have been reported. Although labeling technologies are useful for defining the relative abundance of peptides and proteins across limited sample groups, their effectiveness is questionable in high throughput experiments where multiple samples are a consideration. In particular, clinical studies where large sample numbers may be collected are not amenable to labeling technologies. Significant effort has therefore been directed toward the development of statistically robust, label-free, quantitative LC-MS and LC-MS/MS methodologies (62,63). Such methods can be based on spectral counting where the number of tandem mass spectra obtained for each protein acts as a surrogate for protein abundance in a mixture (64) or spectral peak intensity (65). As mentioned, in addition to these global quantitation strategies for proteomics, targeted quantitative strategies have emerged, and once assays have been

Figure 2 Typical MS-based workflow for quantitative proteomics. Depending on the approach, labeling can be at one of several points in the experiment as indicated by the dashed arrow or, alternatively, a labelfree route can be followed. Total protein is isolated in vivo or in vitro from the system under study. Metabolic labeling (e.g. SILAC) labels proteins in vivo; alternatively, following protein isolation, labeling may be by chemical derivatization or done enzymatically prior to protein separation. After protein separation, in a bottoms-up proteomic strategy, proteins are digested to peptides, which again may be labeled (e.g. isotope-coded affinity tags, iTRAQ, or tandem mass tags approaches). The complexity of the sample may be reduced by fractionation (HPLC), often in several dimensions prior to MS or MS/MS analysis. The final, and often most time-consuming step, is data analysis.

optimized, these approaches are amenable to multiplexing and high sample throughput (52). In this methodology, specific proteolytic peptides are quantified by either selective reaction monitoring or multiple reaction monitoring in the case of multiplexing. This technique relies on the use of stable isotope labeled internal standards and requires a triple quadrupole mass spectrometer where, even within a very complex sample matrix, a predefined precursor ion and one of its fragments may be selected by the 2 mass filters of the instrument and monitored over time for precise quantification (66). Analytical challenges Although the recent rapid development of MS-based proteomic technologies has made possible both the highly mass accurate identification and the relative quantification

of increasingly large data sets, the associated bioinformatics or data analysis of these proteomic data sets, remains a considerable bottle neck for most researchers. The sheer depth, complexity, and size of data sets now produced preclude any form of manual analysis. Beyond the large volume of data produced in a typical MS experiment, proprietary data formats and software by the different instrument manufacturers make it difficult for scientists to directly analyze their data or necessitate accepting a “black box” in proprietary data algorithms. On the other hand, use of available open source tools can be problematic for researchers without programming experience. Bioinformatics in proteomics is an advancing field and computational algorithms, database repositories, and a wealth of software platforms continue to be developed (67,68). In conjunction with these Proteomics and systems biology in the nutritional sciences 359

developments, there is a concerted drive to rationalize data formats across different vendor platforms and initial developments of eXtensible Markup Language-based common file formats for MS-based proteomics workflows have recently coalesced into a unified format called mzML (69). Nonetheless, currently data analysis is by far the most time-consuming step in a proteomics experiment. Indeed, given the considerable numbers of commercial and open source software tools available, each using different algorithms and often yielding conflicting results, software selection alone remains a specific and ongoing concern for many researchers with either complex labeled or label free proteomic data sets.

Proteomic applications in nutrition Although the percentage of proteomics publications related to nutrition has remained small at 2–3% of the total (Fig. 1), the number of primary data papers rather than review papers in this area has increased in recent years. We focus here on the recent applications of proteomics either on chronic diseases with a known nutritional component involved in its etiology, such as obesity, diabetes, or cardiovascular disease, or on applications that involve a dietary intervention. Equally, whereas we choose to focus primarily on recent research in humans (Table 1), many studies using nutritional and proteomic approaches have been done in animal models (70–73). The identification of food bioactive peptides using proteomics has recently been reviewed elsewhere (74). Proteomics on human samples has primarily focused on readily available biological fluids such as plasma or serum, urine, and saliva; although readily accessible, each have their analytical challenges for proteomics. Somewhat less available, cerebrospinal fluid has also been used primarily for research targeting neurological disease (75,76). An initial project of the Human Proteome Organization was the Human Plasma Proteome Project, which has greatly benefited the community in publishing clear guidelines on sample collection, handling, and processing (77). Guidelines have also been published for cerebrospinal fluid (78). Although a Human Proteome Organization Human Kidney and Urine Proteome Project has been initiated, standardized protocols for urine have yet to be published in full (79). Whereas the dynamic range of proteins and the depletion of the most abundant proteins in plasma has been the technical and costly challenge associated with plasma proteomics (80), the high salt levels, necessity for concentration, and high observed intra- as well as inter-individual variation are challenges unique to urine proteomics (81,82). Advances in capillary electrophoresis MS (CE-MS; 83) have led to a number of publications focusing on urine proteomics for the identification of biomarkers associated with diabetes (84) and coronary artery disease (CAD) (85,86). These studies have focused on characterizing panels of polypeptide biomarkers verified in an independent group of patients as a validation set. The 2 independent studies examining the urinary proteomes of patients with CAD demonstrated considerable overlap 360 Moore and Weeks

(50%) in urinary polypeptide profiles from different patient groups, with both studies identifying a panel of urinary peptides capable of predicting CAD and the most abundant upregulated peptides being fragments of collagen a-1 (85,86). Interestingly, fragments of collagen a-1 were found to be downregulated in urine from diabetic patients relative to healthy controls and decreased more in patients with T2D compared to type 1 diabetes (T1D) (84). This later study both validated a set of markers for distinguishing diabetic patients and identified markers capable of differentiating T2D and T1D, although an acknowledged weakness of this study was the variation in the clinical characteristics of each sample group. Additional research has focused on the salivary (87) and plasma (88) proteomes of T2D patients. Using label-free quantitation and multidimensional LC-MS/MS with significant fractionation, Rao et al. (87) identified a notable 487 salivary proteins, of which 65 were differentially expressed in T2D. A panel of these proteins was independently confirmed by either Western blotting or ELISA and, importantly, 3 proteins showed relative increases from prediabetes (patients with impaired fasting glucose or impaired glucose tolerance) to frank T2D. One of these, a-1-antitrypsin, was also identified in the recent diabetes urine proteome analyses as being increased in diabetic patients relative to controls but was not found to be specific to T2D in comparison to T1D (84). Fewer proteins were identified in the plasma proteome study of T2D, which used a 2DGE and MALDI-MS approach; however, the galectin-1 protein found and confirmed by ELISA to be significantly upregulated was also found to be induced by glucose in skeletal muscle cells (88). In summary, these studies demonstrate the usefulness of proteomic analysis of biological fluids for identifying novel candidate biomarkers and molecular mechanisms of pathogenesis in chronic disease. The technological advances and body of work generated to date clearly indicate that future studies will be able to monitor such markers in response to nutritional interventions. A limited number of studies have applied proteomic analyses to biopsy samples from patients with nutrition-related chronic disease. Notable examples include protein profiling of the intestinal epithelium in inflammatory bowel disease (IBD; 89) and adipose tissue (90,91) and skeletal muscle in obesity and T2D (92). This includes one investigation that examined changes in the adipocyte proteome in response to a nutritional intervention, specifically energy restriction in obese participants (91). In the context of IBD, samples were taken from patients undergoing surgical resection for Crohn’s disease, ulcerative colitis, or colonic cancer and analyzed by 2DGE and MALDI-MS. In spite of limited sample numbers, this study yielded several notable findings, including the induction of Rho-GDP inhibitor in both Crohn’s and colitis and a host of signal transduction and energy metabolismrelated proteins altered in inflamed, compared to uninflamed, epithelium from patients with ulcerative colitis (89). The recent proteomic analysis of skeletal muscle from lean, obese, and T2D patients illustrates the power of label-free MS/MS approaches to which the authors applied rigorous statistical

Table 1. Recent applications of proteomics in research related to human nutrition Application Skeletal muscle proteomics in obesity and T2D Urinary proteomics of T1D versus T2D Plasma proteomics after folic acid supplementation Plasma proteomics of T2D Adipocyte proteomics after energy restriction Salivary proteomics of T2D Urinary proteomics in atherosclerosis Adipose proteomics in obesity Serum proteomics after fish oil supplementation PBMC proteomics after flaxseed supplementation PBMC proteome after soyisoflavone supplementation Intestinal epithelial cell proteomics in IBD

Proteomic approach

Reference

1DGE and LC-MS/MS

(92)

CE- and LC-MS/MS

(84)

2DGE, LC-MS/MS

(97)

2DGE, MALDI-MS 2DGE, MALDI-MS

(88) (91)

LC-MS/MS CE-MS/MS

(87) (85,86)

2DGE, MALDI-MS/MS 2DGE, MALDI-MS, LC-MS/MS 2DGE, MALDI-MS

(90) (95)

2DGE, MALDI-MS

(93)

2DGE, MALDI-MS

(89)

(94)

analyses to draw their conclusions of disrupted mitochondrial, cytoskeletal, proteasome, and chaperone proteins in insulin-resistant muscle (92). The work of Boden et al. (90) also found structural and stress/unfolded protein response (UPR) proteins altered in subcutaneous adipose tissue from insulin-resistant obese participants and their approach shows how successful proteomics can be at hypothesis generating. Subsequent to identifying 3 UPR proteins by proteomics, the authors then tested and confirmed by Western blotting other UPR proteins as differentially expressed between lean and obese participants. Lastly, the research examining the effect of energy restriction on the adipose proteome of obese patients is important in demonstrating that proteomic approaches are sensitive enough to detect changes induced in vivo from a nutritional intervention (91). Although all of these studies were predictably limited to a very few number of participants, they can nonetheless offer fascinating insight into in vivo tissue proteome changes during chronic disease in humans. Other research has investigated proteome dynamics in response to human nutritional intervention studies. These included experiments examining the peripheral blood mononuclear cell (PBMC) proteome response to 8 wk of dietary soy isoflavone supplementation in one study (93) and 1 wk of supplementation with flaxseed in another (94). Additional research has examined changes in the serum proteome after 6 wk of fish oil supplementation (95) and most recently the plasma proteome response to 12 wk of folic acid supplementation has been assessed (96). In this later study, a remarkable 62 of 300 reproducibly identified proteins in a 2DGE and LCMS/MS approach were found modulated by folate supplementation, predominantly proteins involved in immune function; however, these changes have yet to be independently verified.

The aforementioned research and that done in animals and in vitro systems illustrates that proteomic approaches are being successfully applied to a variety of research questions in the nutritional sciences. Ultimately, these datasets should be combined with transcriptomic and metabolomics data for rigorous nutritional systems analyses.

Conclusions In the last 10 y, technological and methodological developments in genomics, proteomics, and metabolomics have led to the generation of an increasing number of large-scale datasets in all aspects of biology. The complexity of these datasets has prompted the use of computational biology and application of systems or network theory with the objective of understanding biological systems as a whole. The ultimate goal of a systems biology approach is to characterize and predict the dynamic properties of the biological network that is under scrutiny. Whereas microarray data initially was most amenable for systems modeling, recent and ongoing advances in MS-based quantitative proteomics are yielding a growing number of datasets appropriate for systems biology applications. Indeed, given the established lack of correlation between mRNA and protein expression levels (97), proteomic datasets are clearly essential for building network models with accurate predictive power. The number of publications applying proteomic and systems approaches in the context of research questions relevant to the nutritional sciences is expanding. The proteomic research done to date has focused on both nutrition-related chronic diseases and dietary intervention studies. This body of work has illustrated the potential of proteomic investigations to yield candidate biomarkers and novel molecular mechanisms in an unbiased fashion. Although systems research has yielded disease-specific networks and indeed a network map of all human diseases, it has primarily focused on modeling metabolic networks with genome-scale models now reconstructed for humans and many microorganisms. We envisage the real potential for nutritional systems biology applications will undoubtedly lie in predictive network modeling of the cellular signaling response to nutrient perturbation, ultimately with models that incorporate information regarding genetic polymorphisms that influence gene-nutrient interactions.

Acknowledgments We thank Nicolas Spanos for help with figure artwork. J.B.M. analyzed data; J.B.M. and M.E.W. wrote the paper. J.B.M. had primary responsibility for final content. All authors read and approved the final manuscript.

Literature Cited 1. Kitano H. Systems biology: a brief overview. Science. 2002;295:1662–4. 2. de Graaf AA, Freidig AP, De Roos B, Jamshidi N, Heinemann M, Rullmann JA, Hall KD, Adiels M, van Ommen B. Nutritional systems biology modeling: from molecular mechanisms to physiology. PLOS Comput Biol. 2009;5:e1000554.

Proteomics and systems biology in the nutritional sciences 361

3. Srivastava S, Chan C. Application of metabolic flux analysis to identify the mechanisms of free fatty acid toxicity to human hepatoma cell line. Biotechnol Bioeng. 2008;99:399–410. 4. King JC, Shames DM, Lowe NM, Woodhouse LR, Sutherland B, Abrams SA, Turnlund JR, Jackson MJ. Effect of acute zinc depletion on zinc homeostasis and plasma zinc kinetics in men. Am J Clin Nutr. 2001;74:116–24. 5. Hall KD. Computational model of in vivo human energy metabolism during semistarvation and refeeding. Am J Physiol Endocrinol Metab. 2006;291:E23–37. 6. Bertalanffy LV. General system theory; foundations, development, applications. New York: G. Braziller; 1969. 7. Wiener N. Cybernetics. New York: J. Wiley; 1948. 8. Kitano H. Computational systems biology. Nature. 2002;420:206–10. 9. Hood L, Heath JR, Phelps ME, Lin B. Systems biology and new technologies enable predictive and preventative medicine. Science. 2004; 306:640–3. 10. Hyduke DR, Palsson BO. Towards genome-scale signalling-network reconstructions. Nat Rev Genet. 2010;11:297–307. 11. Reed MC, Nijhout HF, Neuhouser ML, Gregory JF III, Shane B, James SJ, Boynton A, Ulrich CM. A mathematical model gives insights into nutritional and genetic aspects of folate-mediated one-carbon metabolism. J Nutr. 2006;136:2653–61. 12. Ulrich CM, Neuhouser M, Liu AY, Boynton A, Gregory JF III, Shane B, James SJ, Reed MC, Nijhout HF. Mathematical modeling of folate metabolism: predicted effects of genetic polymorphisms on mechanisms and biomarkers relevant to carcinogenesis. Cancer Epidemiol Biomarkers Prev. 2008;17:1822–31. 13. Wilkinson DJ. Stochastic modelling for quantitative description of heterogeneous biological systems. Nat Rev Genet. 2009;10:122–33. 14. Ruths D, Muller M, Tseng JT, Nakhleh L, Ram PT. The signaling petri net-based simulator: a non-parametric strategy for characterizing the dynamics of cell-specific signaling networks. PLOS Comput Biol. 2008;4:e1000005. 15. Taylor RC, Acquaah-Mensah G, Singhal M, Malhotra D, Biswal S. Network inference algorithms elucidate Nrf2 regulation of mouse lung oxidative stress. PLOS Comput Biol. 2008;4:e1000166. 16. Komurov K, Ram PT. Patterns of human gene expression variance show strong associations with signaling network hierarchy. BMC Syst Biol. 2010;4:154. 17. Oberhardt MA, Palsson BO, Papin JA. Applications of genome-scale metabolic reconstructions. Mol Syst Biol. 2009;5:320. 18. Becker SA, Feist AM, Mo ML, Hannum G, Palsson BO, Herrgard MJ. Quantitative prediction of cellular metabolism with constraint-based models: the COBRA Toolbox. Nat Protoc. 2007;2:727–38. 19. Shen Y, Liu J, Estiu G, Isin B, Ahn YY, Lee DS, Barabasi AL, Kapatral V, Wiest O, et al. Blueprint for antimicrobial hit discovery targeting metabolic networks. Proc Natl Acad Sci USA. 2010;107:1082–7. 20. Duarte NC, Becker SA, Jamshidi N, Thiele I, Mo ML, Vo TD, Srivas R, Palsson BO. Global reconstruction of the human metabolic network based on genomic and bibliomic data. Proc Natl Acad Sci USA. 2007;104:1777–82. 21. Ma H, Sorokin A, Mazein A, Selkov A, Selkov E, Demin O, Goryanin I. The Edinburgh human metabolic network reconstruction and its functional analysis. Mol Syst Biol. 2007;3:135. 22. Zelezniak A, Pers TH, Soares S, Patti ME, Patil KR. Metabolic network topology reveals transcriptional regulatory signatures of type 2 diabetes. PLOS Comput Biol. 2010;6:e1000729. 23. Patil KR, Nielsen J. Uncovering transcriptional regulation of metabolism by using metabolic network topology. Proc Natl Acad Sci USA. 2005;102:2685–9. 24. Goh KI, Cusick ME, Valle D, Childs B, Vidal M, Barabasi AL. The human disease network. Proc Natl Acad Sci USA. 2007;104:8685–90. 25. Lee DS, Park J, Kay KA, Christakis NA, Oltvai ZN, Barabasi AL. The implications of human metabolic network topology for disease comorbidity. Proc Natl Acad Sci USA. 2008;105:9880–5. 26. Jesmin J, Rashid MS, Jamil H, Hontecillas R, Bassaganya-Riera J. Gene regulatory network reveals oxidative stress as the underlying molecular

362 Moore and Weeks

27.

28.

29.

30.

31.

32.

33.

34.

35. 36.

37.

38.

39.

40.

41.

42.

43. 44. 45.

46.

47. 48.

mechanism of type 2 diabetes and hypertension. BMC Med Genomics. 2010;3:45. Farres J, Pujol A, Coma M, Ruiz JL, Naval J, Mas JM, Molins A, Fondevila J, Aloy P. Revealing the molecular relationship between type 2 diabetes and the metabolic changes induced by a very-low-carbohydrate low-fat ketogenic diet. Nutr Metab (Lond). 2010;7:88. Hamosh A, Scott AF, Amberger JS, Bocchini CA, McKusick VA. Online Mendelian Inheritance in Man (OMIM), a knowledgebase of human genes and genetic disorders. Nucleic Acids Res. 2005;33:D514–7. Hong SE, Heo HS, Kim DH, Kim MS, Kim CH, Lee J, Yoo MA, Yu BP, Leeuwenburgh C, et al. Revealing system-level correlations between aging and calorie restriction using a mouse transcriptome. Age (Dordr). 2010;32:15–30. Nookaew I, Gabrielsson BG, Holmang A, Sandberg AS, Nielsen J. Identifying molecular effects of diet through systems biology: influence of herring diet on sterol metabolism and protein turnover in mice. PLoS ONE. 2010;5:e12361. Gevorgyan A, Bushell ME, Avignone-Rossa C, Kierzek AM. SurreyFBA: a command line tool and graphics user interface for constraint-based modeling of genome-scale metabolic reaction networks. Bioinformatics. 2011;27:433–4. Li F, Thiele I, Jamshidi N, Palsson BO. Identification of potential pathway mediation targets in Toll-like receptor signaling. PLOS Comput Biol. 2009;5:e1000292. Weston AD, Hood L. Systems biology, proteomics, and the future of health care: toward predictive, preventative, and personalized medicine. J Proteome Res. 2004;3:179–96. Tanaka K, Waki H, Ido Y, Akita S, Yoshida Y, Yoshida T. Protein and polymer analyses up to m/z 100 000 by laser ionization time-of flight mass spectrometry. Rapid Commun Mass Spectrom. 1988;2:151–3. Yamashita M, Fenn JB. Electrospray ion source. Another variation on the free-jet theme. J Phys Chem. 1984;88:4451–9. Stephenson JL Jr, McLuckey SA. Simplification of product ion spectra derived from multiply charged parent ions via ion/ion chemistry. Anal Chem. 1998;70:3533–44. Gstaiger M, Aebersold R. Applying mass spectrometry-based proteomics to genetics, genomics and network biology. Nat Rev Genet. 2009;10:617–27. Huang TY, McLuckey SA. Top-down protein characterization facilitated by ion/ion reactions on a quadrupole/time of flight platform. Proteomics. 2010;10:3577–88. Mohr J, Swart R, Samonig M, Bohm G, Huber CG. High-efficiency nano- and micro-HPLC–high-resolution Orbitrap-MS platform for top-down proteomics. Proteomics. 2010;10:3598–609. Wittmann-Liebold B, Graack HR, Pohl T. Two-dimensional gel electrophoresis as tool for proteomics studies in combination with protein identification by mass spectrometry. Proteomics. 2006;6:4688–703. Fournier ML, Gilmore JM, Martin-Brown SA, Washburn MP. Multidimensional separations-based shotgun proteomics. Chem Rev. 2007; 107:3654–86. Gygi SP, Corthals GL, Zhang Y, Rochon Y, Aebersold R. Evaluation of two-dimensional gel electrophoresis-based proteome analysis technology. Proc Natl Acad Sci USA. 2000;97:9390–5. Domon B, Aebersold R. Mass spectrometry and protein analysis. Science. 2006;312:212–7. Yates JR, Ruse CI, Nakorchevsky A. Proteomics by mass spectrometry: approaches, advances, and applications. Annu Rev Biomed Eng. 2009;11:49–79. Zubarev RA, Kelleher NL, McLafferty FW. Electron capture dissociation of multiply charged protein cations. A nonergodic process. J Am Chem Soc. 1998;120:3265–6. McAlister GC, Phanstiel D, Good DM, Berggren WT, Coon JJ. Implementation of electron-transfer dissociation on a hybrid linear ion traporbitrap mass spectrometer. Anal Chem. 2007;79:3525–34. Hardman M, Makarov AA. Interfacing the orbitrap mass analyzer to an electrospray ion source. Anal Chem. 2003;75:1699–705. Hu Q, Noll RJ, Li H, Makarov A, Hardman M, Graham Cooks R. The Orbitrap: a new mass spectrometer. J Mass Spectrom. 2005;40:430–43.

49. Macek B, Waanders LF, Olsen JV, Mann M. Top-down protein sequencing and MS3 on a hybrid linear quadrupole ion trap-orbitrap mass spectrometer. Mol Cell Proteomics. 2006;5:949–58. 50. Thalassinos K, Grabenauer M, Slade SE, Hilton GR, Bowers MT, Scrivens JH. Characterization of phosphorylated peptides using traveling wave-based and drift cell ion mobility mass spectrometry. Anal Chem. 2009;81:248–54. 51. Shvartsburg AA, Creese AJ, Smith RD, Cooper HJ. Separation of peptide isomers with variant modified sites by high-resolution differential ion mobility spectrometry. Anal Chem. 2010;82:8327–34. 52. Pan S, Aebersold R, Chen R, Rush J, Goodlett DR, McIntosh MW, Zhang J, Brentnall TA. Mass spectrometry based targeted protein quantification: methods and applications. J Proteome Res. 2009;8:787–97. 53. Ong SE, Mann M. Mass spectrometry-based proteomics turns quantitative. Nat Chem Biol. 2005;1:252–62. 54. Spellman DS, Deinhardt K, Darie CC, Chao MV, Neubert TA. Stable isotopic labeling by amino acids in cultured primary neurons: application to brain-derived neurotrophic factor-dependent phosphotyrosineassociated signaling. Mol Cell Proteomics. 2008;7:1067–76. 55. Kruger M, Moser M, Ussar S, Thievessen I, Luber CA, Forner F, Schmidt S, Zanivan S, Fassler R, et al. SILAC mouse for quantitative proteomics uncovers kindlin-3 as an essential factor for red blood cell function. Cell. 2008;134:353–64. 56. Sury MD, Chen JX, Selbach M. The SILAC fly allows for accurate protein quantification in vivo. Mol Cell Proteomics. 2010;9:2173–83. 57. Gygi SP, Rist B, Gerber SA, Turecek F, Gelb MH, Aebersold R. Quantitative analysis of complex protein mixtures using isotope-coded affinity tags. Nat Biotechnol. 1999;17:994–9. 58. Ross PL, Huang YN, Marchese JN, Williamson B, Parker K, Hattan S, Khainovski N, Pillai S, Dey S, et al. Multiplexed protein quantitation in Saccharomyces cerevisiae using amine-reactive isobaric tagging reagents. Mol Cell Proteomics. 2004;3:1154–69. 59. Dayon L, Hainard A, Licker V, Turck N, Kuhn K, Hochstrasser DF, Burkhard PR, Sanchez JC. Relative quantification of proteins in human cerebrospinal fluids by MS/MS using 6-plex isobaric tags. Anal Chem. 2008;80:2921–31. 60. Ye X, Luke B, Andresson T, Blonder J. 18O stable isotope labeling in MS-based proteomics. Brief Funct Genomics Proteomics. 2009;8: 136–44. 61. Li S, Zeng D. CILAT: a new reagent for quantitative proteomics. Chem Commun (Camb). 2007;2181–3. 62. Levin Y, Schwarz E, Wang L, Leweke FM, Bahn S. Label-free LCMS/MS quantitative proteomics for large-scale biomarker discovery in complex samples. J Sep Sci. 2007;30:2198–203. 63. Clough T, Key M, Ott I, Ragg S, Schadow G, Vitek O. Protein quantification in label-free LC-MS experiments. J Proteome Res. 2009;8: 5275–84. 64. Liu H, Sadygov RG, Yates JR III. A model for random sampling and estimation of relative protein abundance in shotgun proteomics. Anal Chem. 2004;76:4193–201. 65. Wang W, Zhou H, Lin H, Roy S, Shaler TA, Hill LR, Norton S, Kumar P, Anderle M, et al. Quantification of proteins and metabolites by mass spectrometry without isotopic labeling or spiked standards. Anal Chem. 2003;75:4818–26. 66. Lange V, Picotti P, Domon B, Aebersold R. Selected reaction monitoring for quantitative proteomics: a tutorial. Mol Syst Biol. 2008;4:222. 67. Deutsch EW, Lam H, Aebersold R. Data analysis and bioinformatics tools for tandem mass spectrometry in proteomics. Physiol Genomics. 2008;33:18–25. 68. Matthiesen R, Azevedo L, Amorim A, Carvalho AS. Discussion on common data analysis strategies used in MS-based proteomics. Proteomics. 2011;11:604–19. 69. Martens L, Chambers M, Sturm M, Kessner D, Levander F, Shofstahl J, Tang WH, Rompp A, Neumann S, et al. mzML–a community standard for mass spectrometry data. Mol Cell Proteomics. 2011;10:R110 000133. 70. Schmid GM, Converset V, Walter N, Sennitt MV, Leung KY, Byers H, Ward M, Hochstrasser DF, Cawthorne MA, et al. Effect of high-fat

71.

72.

73.

74.

75.

76.

77.

78.

79.

80.

81. 82.

83.

84.

85.

86.

87.

diet on the expression of proteins in muscle, adipose tissues, and liver of C57BL/6 mice. Proteomics. 2004;4:2270–82. Tosco A, Siciliano RA, Cacace G, Mazzeo MF, Capone R, Malorni A, Leone A, Marzullo L. Dietary effects of copper and iron deficiency on rat intestine: a differential display proteome analysis. J Proteome Res. 2005;4:1781–8. Beattie JH, Gordon MJ, Rucklidge GJ, Reid MD, Duncan GJ, Horgan GW, Cho YE, Kwun IS. Aorta protein networks in marginal and acute zinc deficiency. Proteomics. 2008;8:2126–35. Starkey JM, Zhao Y, Sadygov RG, Haidacher SJ, Lejeune WS, Dey N, Luxon BA, Kane MA, Napoli JL, et al. Altered retinoic acid metabolism in diabetic mouse kidney identified by O isotopic labeling and 2D mass spectrometry. PLoS ONE. 2010;5:e11095. Kussmann M, Panchaud A, Affolter M. Proteomics in nutrition: status quo and outlook for biomarkers and bioactives. J Proteome Res. 2010; 9:4876–87. Hong Z, Shi M, Chung KA, Quinn JF, Peskind ER, Galasko D, Jankovic J, Zabetian CP, Leverenz JB, et al. DJ-1 and alpha-synuclein in human cerebrospinal fluid as biomarkers of Parkinson’s disease. Brain. 2010; 133:713–26. Schutzer SE, Angel TE, Liu T, Schepmoes AA, Clauss TR, Adkins JN, Camp DG, Holland BK, Bergquist J, et al. Distinct cerebrospinal fluid proteomes differentiate post-treatment lyme disease from chronic fatigue syndrome. PLoS ONE. 2011;6:e17287. Omenn GS, States DJ, Adamski M, Blackwell TW, Menon R, Hermjakob H, Apweiler R, Haab BB, Simpson RJ, et al. Overview of the HUPO Plasma Proteome Project: results from the pilot phase with 35 collaborating laboratories and multiple analytical groups, generating a core dataset of 3020 proteins and a publicly-available database. Proteomics. 2005;5: 3226–45. Teunissen CE, Petzold A, Bennett JL, Berven FS, Brundin L, Comabella M, Franciotta D, Frederiksen JL, Fleming JO, et al. A consensus protocol for the standardization of cerebrospinal fluid collection and biobanking. Neurology. 2009;73:1914–22. Yamamoto T, Langham RG, Ronco P, Knepper MA, Thongboonkerd V. Towards standard protocols and guidelines for urine proteomics: a report on the Human Kidney and Urine Proteome Project (HKUPP) symposium and workshop, 6 October 2007, Seoul, Korea and 1 November 2007, San Francisco, CA, USA. Proteomics. 2008;8:2156–9. Echan LA, Tang HY, Ali-Khan N, Lee K, Speicher DW. Depletion of multiple high-abundance proteins improves protein profiling capacities of human serum and plasma. Proteomics. 2005;5:3292–303. Thongboonkerd V. Practical points in urinary proteomics. J Proteome Res. 2007;6:3881–90. Weeks ME, Hariharan D, Petronijevic L, Radon TP, Whiteman HJ, Kocher HM, Timms JF, Lemoine NR, Crnogorac-Jurcevic T. Analysis of the urine proteome in patients with pancreatic ductal adenocarcinoma. Proteomics Clin Appl. 2008;2:1047–57. Mischak H, Coon JJ, Novak J, Weissinger EM, Schanstra JP, Dominiczak AF. Capillary electrophoresis-mass spectrometry as a powerful tool in biomarker discovery and clinical diagnosis: an update of recent developments. Mass Spectrom Rev. 2009;28:703–24. Maahs DM, Siwy J, Argiles A, Cerna M, Delles C, Dominiczak AF, Gayrard N, Iphofer A, Jansch L, et al. Urinary collagen fragments are significantly altered in diabetes: a link to pathophysiology. PLoS ONE. 2010;5:piie13051. Zimmerli LU, Schiffer E, Zurbig P, Good DM, Kellmann M, Mouls L, Pitt AR, Coon JJ, Schmieder RE, et al. Urinary proteomic biomarkers in coronary artery disease. Mol Cell Proteomics. 2008;7:290–8. von Zur Muhlen C, Schiffer E, Zuerbig P, Kellmann M, Brasse M, Meert N, Vanholder RC, Dominiczak AF, Chen YC, et al. Evaluation of urine proteome pattern analysis for its potential to reflect coronary artery atherosclerosis in symptomatic patients. J Proteome Res. 2009;8: 335–45. Rao PV, Reddy AP, Lu X, Dasari S, Krishnaprasad A, Biggs E, Roberts CT, Nagalla SR. Proteomic identification of salivary biomarkers of type2 diabetes. J Proteome Res. 2009;8:239–45.

Proteomics and systems biology in the nutritional sciences 363

88. Liu X, Feng Q, Chen Y, Zuo J, Gupta N, Chang Y, Fang F. Proteomicsbased identification of differentially-expressed proteins including galectin-1 in the blood plasma of type 2 diabetic patients. J Proteome Res. 2009;8:1255–62. 89. Shkoda A, Werner T, Daniel H, Gunckel M, Rogler G, Haller D. Differential protein expression profile in the intestinal epithelium from patients with inflammatory bowel disease. J Proteome Res. 2007;6: 1114–25. 90. Boden G, Duan X, Homko C, Molina EJ, Song W, Perez O, Cheung P, Merali S. Increase in endoplasmic reticulum stress-related proteins and genes in adipose tissue of obese, insulin-resistant individuals. Diabetes. 2008;57:2438–44. 91. Bouwman FG, Claessens M, van Baak MA, Noben JP, Wang P, Saris WH, Mariman EC. The physiologic effects of caloric restriction are reflected in the in vivo adipocyte-enriched proteome of overweight/obese subjects. J Proteome Res. 2009;8:5532–40. 92. Hwang H, Bowen BP, Lefort N, Flynn CR, De Filippis EA, Roberts C, Smoke CC, Meyer C, Hojlund K, et al. Proteomics analysis of human skeletal muscle reveals novel abnormalities in obesity and type 2 diabetes. Diabetes. 2010;59:33–42.

364 Moore and Weeks

93. Fuchs D, Vafeiadou K, Hall WL, Daniel H, Williams CM, Schroot JH, Wenzel U. Proteomic biomarkers of peripheral blood mononuclear cells obtained from postmenopausal women undergoing an intervention with soy isoflavones. Am J Clin Nutr. 2007;86:1369–75. 94. Fuchs D, Piller R, Linseisen J, Daniel H, Wenzel U. The human peripheral blood mononuclear cell proteome responds to a dietary flaxseed-intervention and proteins identified suggest a protective effect in atherosclerosis. Proteomics. 2007;7:3278–88. 95. de Roos B, Geelen A, Ross K, Rucklidge G, Reid M, Duncan G, Caslake M, Horgan G, Brouwer IA. Identification of potential serum biomarkers of inflammation and lipid modulation that are altered by fish oil supplementation in healthy volunteers. Proteomics. 2008;8:1965–74. 96. Duthie SJ, Horgan G, de Roos B, Rucklidge G, Reid M, Duncan G, Pirie L, Basten GP, Powers HJ. Blood folate status and expression of proteins involved in immune function, inflammation, and coagulation: biochemical and proteomic changes in the plasma of humans in response to long-term synthetic folic acid supplementation. J Proteome Res. 2010;9:1941–50. 97. Gygi SP, Rochon Y, Franza BR, Aebersold R. Correlation between protein and mRNA abundance in yeast. Mol Cell Biol. 1999;19:1720–30.