Nutrition 25 (2009) 1085–1093 www.nutritionjrnl.com Review at the center of nutrigenomics: Comprehensive molecular understanding of dietary health effects

Martin Kussmann, Ph.D.* and Michael Affolter, Ph.D.

Functional Group, Department of BioAnalytical Sciences, Nestle´Research Center, Lausanne, Switzerland Manuscript received May 29, 2009; accepted May 31, 2009.

Abstract Apart from the air we breathe, food is the only physical matter we take into our body during our life. Nutrition exhibits therefore the most important life-long environmental impact on human health. Food components interact with our body at system, organ, cellular, and molecular levels. These dietary com- ponents come in complex mixtures, in which not only the presence and concentrations of a single com- pound but also interactions of multiple compounds determine ingredient bioavailability and bioefficacy. Modern nutritional and health research focuses on promoting health, preventing or delay- ing the onset of disease, and optimizing performance. Deciphering the molecular interplay between food and health requires therefore holistic approaches because nutritional improvement of certain health aspects must not be compromised by deterioration of others. In other words, in nutrition, we have to get everything right. Proteomics is a central platform in nutrigenomics that describes how our expresses itself as a response to diet. Nutrigenetics deals with our genetic predisposition and susceptibility toward diet and helps stratify subject cohorts and discern responders from non-responders. Epigenetics represent DNA sequence-unrelated biochemical modifications of DNA itself and DNA-binding proteins and appears to provide a format for life-long or even transgeneration imprinting of metabolism. Proteomics in nutrition can identify and quantify bioactive proteins and peptides and addresses questions of nutritional bioefficacy. In this review, we focus on these latter as- pects, update the reader on technologic developments, and review major applications. Ó 2009 Published by Elsevier Inc.

Keywords: Proteomics; Nutrition; Health; Biomarker; Nutrigenomics; Nutrigenetics

Introduction Proteomics is a central platform in nutrigenomics that describes how our genome expresses itself as a response to Food components interact with our body at system, organ, diet [4]. Nutrigenetics deals with our genetic predisposition cellular, and molecular levels. These dietary components and susceptibility toward diet [5] and helps stratify subject come in complex mixtures, in which not only the presence cohorts and discern responders from non-responders [6]. Epi- and concentrations of a single compound but also interactions genetics represent DNA sequence-unrelated biochemical of multiple compounds determine ingredient bioavailability modifications of DNA itself and DNA-binding proteins and and bioefficacy [1]. appears to provide a format for life-long or even transgener- Modern nutritional and health research focuses on pro- ation imprinting of metabolism [7,8]. Proteomics in nutrition moting health, preventing or delaying the onset of disease, can identify and quantify bioactive proteins and peptides and and optimizing performance [2]. Deciphering the molecular addresses questions of nutritional bioefficacy [9,10]. interplay between food and health requires therefore holistic The interplay between nutrition and health has been approaches because nutritional improvement of certain known for centuries: the Greek doctor Hippocrates (fourth health aspects must not be compromised by deterioration of century B.C.) can be seen as the father of ‘‘functional others [3]. food,’’ because he recommended using food as medicine and vice versa. Another example of such long-term experi- *Corresponding author. Tel.: þ41-21-785-9572; fax: þ41-21-785-9486. ence is the record of traditional Chinese medicine: Sun Si- E-mail address: [email protected] (M. Kussmann). Miao, a famous doctor of the Tang dynasty (seventh century

0899-9007/09/$ – see front matter Ó 2009 Published by Elsevier Inc. doi:10.1016/j.nut.2009.05.022 1086 M. Kussmann and M. Affolter / Nutrition 25 (2009) 1085–1093

A.D.), stated that, ‘‘When a person is sick, the doctor should multiple-stage fragmentation for ion traps; high selectivity first regulate the patient’s diet and lifestyle.’’ for triple-Q; high sensitivity and speed for ToF; and very The new era of nutritional research translates this rather high mass accuracy and resolution for orbitrap and FT-ICR. empirical knowledge to evidence-based molecular science, Current top-end proteomic machines are orbitrap [15] and because food components interact with our body at system, FT-ICR instruments [16], which rely on frequency readout organ, cellular, and molecular levels [11]. Modern nutritional of oscillating ions rather than ToF- or scanning-based analysis. and health research focuses on promoting health, preventing The major remaining analytical challenge is not mass ac- or delaying the onset of disease, and optimizing performance curacy (today down to subparts per million), mass resolution [11]. (today up to several hundred thousand), or absolute sensitiv- Dietary components come in complex mixtures, in which ity (today down to a picomolar range), but the dynamic range not only the presence and concentrations of a single com- of protein concentrations (e.g., estimated 1012 in human pound but also interactions of multiple compounds influence blood) [17]. Current MS-based proteomic platforms can de- food compound bioavailability and bioefficacy [1]. Hence, liver a dynamic range of 104. This means that the remaining, the necessity of developing and applying comprehensive as such inaccessible, low-abundant proteome has to be ad- analytical methods to reveal bioactive ingredients and their dressed by depletion of the most abundant proteins (e.g., action becomes evident [12]. by the commercially available multiple affinity removal sys- Proteomics is a central platform in elucidating these tem that specifically removes the top 7 or even 14 plasma pro- molecular events in nutrition: it can identify and quantify bio- teins) [18] or by selective enrichment of low-abundant active proteins and peptides and addresses questions of nutri- proteins (e.g., by the immobilized metal affinity chromatog- tional bioefficacy [9]. In this article, we focus on these latter raphy or titanium dioxide techniques for phosphoproteins aspects, update the reader on technologic developments, and [19] or lectins [20] or the cell-surface capture technique for review major applications: we summarize mass spectrometry glycoproteins [21]). All these biochemical depletion and (MS)-rooted proteomic techniques for protein identification enrichment resins and columns have matured a great deal and quantification and go through a selection of nutritional and come now in robust formats. intervention and bioefficacy studies assessed by proteomic After depletion and/or enrichment, usually further prese- means. paration measurements are taken at the protein or peptide Proteins are the key actors in virtually all biological pro- level, based on two-dimensional (2D) gels or on liquid chro- cesses in the human body—they are the ‘‘molecular robots’’ matography (LC), or on hybrid approaches (Gel-LC). Fig- that do all the work. Hence, because we want to gain a more ure 1 summarizes these proteomic workflows. comprehensive understanding of this machinery and further Gel-based protein separation methods have the advantage develop the concept of nutritional , proteo- of physically preserving the protein context and generating mics is at the center of this concerted action. real protein images. However, they have limited dynamic range, bias toward the more easily soluble proteins, and Proteomics technology a low degree of automation with, in consequence, low throughput. The most advanced method for 2D protein sepa- Protein identification ration is differential imaging gel electrophoresis [22], which relies on multiplexed staining and coprocessing of one con- Any proteomic study, be it in a nutritional or other frame- trol plus a maximum of two case samples. Protein spots work, commences with a protein survey of what can be have then to be detected, excised, digested with trypsin, ‘‘seen’’ in a given sample and condition. Identifying proteins and amended to LC-MS/MS. at a large scale and with high throughput is a mass spectro- Complementary to the gels and in view of an increasing de- metric business. Figure 1 outlines proteomic workflows for mand for throughput and speed, (multi)dimensional LC the ‘‘discovery mode.’’ setups have been coupled online to MS analysis, with simple Mass spectrometers can identify proteins and peptides by reversed-phase columns and combined strong cation ex- determination of their exact masses and generating informa- change-reversed phase systems being the most frequently ap- tion on the amino acid sequences. Today, the main ionization plied. These workflows run under the terms MudPIT methods deployed are electrospray [13] and matrix-assisted la- (multidimensional protein identification technology) or shot- ser desorption [14], which can put large, fragile biomolecules gun proteomics [23]. One major difference compared with gel such as proteins and peptides rapidly and gently into gas phase approaches is that the protein context is physically sacrificed and ionize them while preserving their integrity. These ion by upstream tryptic digestion of the protein mixture and sub- sources come in various combinations with different mass an- sequent separation and analysis at the peptide level. The pro- alyzers that separate the ions by mass over charge. The most tein context is then reconstructed in silico by reassigning the popular analyzers in proteomics are ion traps, triple-quadru- peptide identification to the same parent protein. pole (triple-Q), time-of-flight (ToF) tubes, orbitrap, and Four- With enrichment and/or depletion, gel- and/or LC-based ier-transform ion cyclotron resonance (FT-ICR) cells, with further preseparation and electrospray ionization–Q/Q-ToF/ their specific advantages, which are: high sensitivity and ion traps/FT-ICR (online workflow) or matrix-assisted laser M. Kussmann and M. Affolter / Nutrition 25 (2009) 1085–1093 1087

Fig. 1. Discovery workflow in mass spectrometry–based proteomics. Gel-LC, gel-LC-based separation; LC, liquid chromatography. desorption ionization–ToF/ToF (offline workflow) based ‘‘heavy’’ version (e.g., 13C, 2H, or 15N label). These tagging MS, powerful platforms are available for fast and compre- techniques can be executed at the protein (e.g., aniline-ben- hensive assessment of proteomes from body fluids, tissues, zoic acid labeling (AniBAL) [25]) or peptide (e.g., isotope cells, or organelles. coded affinity tag (ICAT) [26], isobaric tag for relative and absolute quantitation (iTRAQ), [27], tandem mass tag Protein quantification (TMT) [28]) level and can be introduced chemically into the sample (e.g., ICAT, iTRAQ, TMT, AniBAL) or metabol- After a protein survey, the first and foremost question ically by feeding cells or even small animals (mice, rats) with asked in proteomic studies comparing multiple conditions isotopically labeled essential amino acids stable isotope la- is: Which proteins are differentially expressed? Today’s beling of amino acids in cell culture (SILAC) [29]. The quan- means for protein quantification are gel or MS based [24]. tification readout can be obtained at the MS (ICAT, SILAC, As mentioned under PROTEIN IDENTIFICATION, the gold stan- AniBAL) or MS/MS (iTRAQ, TMT) level. dard for 2D-rooted proteomics is quantification by differen- Although it is, on the one hand, preferable to introduce tial imaging gel electrophoresis, i.e., by differential staining a label as upstream as possible in the workflow to maximize of the separated protein spots and image analysis. coprocessing of case(s) and control(s) and minimize bias Figure 2 summarizes the principal strategies and steps for (e.g., achieved in the cases of the metabolic SILAC and the gel-free global protein quantification, i.e., stable isotope chemical AniBAL methods), it is, on the other hand, advan- rooted or label free, metabolic or chemical labeling, and tageous to not label at all to maximize sample integrity and providing relative or absolute quantitative information. compare samples directly as they are. Therefore, ‘‘label- Alternatively to differential imaging gel electrophoresis free’’ approaches (Fig. 2) have been developed that deploy and compatible with online shotgun LC-MS/MS workflows, spectral counting of peptide assignments for semiquantitative stable isotopes can be introduced into the conditions in ques- analysis or compare the peak intensities of the very same pep- tion, meaning that reagents tagging amino acid side chains at tide by overlaying LC-MS runs of control and case samples a protein or peptide level can introduce a differential isotopic [30,31]. signature so that the conditions can be quantitatively All previously described methodologies provide informa- compared at the MS level; e.g., the ‘‘control’’ sample can tion on relative changes in protein abundance. Especially in be labeled with the ‘‘light,’’ unlabeled form of the reagent, nutrition it is desirable to also generate information on abso- whereas the ‘‘case’’ sample can be derivatized with the lute amounts of proteins present in a given sample. This 1088 M. Kussmann and M. Affolter / Nutrition 25 (2009) 1085–1093

Metabolic Chemical Chemical Internal Label free Labeling modification of modification of standard strategies proteins peptides

Cells/ tissues

Proteins

Peptides

Mass spec

Heavy label Light label No label

Fig. 2. Strategies for relative and absolute protein quantification in mass spectrometry–based proteomics. comes into play in the context of bioavailability and bioeffi- product of each of the approximately 20 500 human genes, cacy studies: the bases of proven ingredient bioavailability enabling rapid and systematic verification of candidate bio- and bioefficacy are absolute values of its amounts in the markers and laying a quantitative foundation for comprehen- original food matrix and in the relevant body fluids or tissues. sive human proteome studies. These assays could serve as an Therefore, proteins and peptides must be absolutely quanti- excellent toolset in future nutritional studies to better under- fied from ingredient and biomarker perspectives. Such abso- stand and correlate the effects of diet and food compounds on lute quantitative information can be obtained by spiking protein expression and regulation in humans. defined amounts of stable isotope-labeled peptides or entire proteins into the sample of interest and comparing the corre- Data processing sponding mass signals of the sample peptides with those of these internal standards [32]. The targeted, multiplexed pep- Apart from the ‘‘wet laboratory’’ equipment to generate tide-level version of such a strategy is called AQUA (absolute large proteomic datasets, it takes sophisticated software to quantification) [33], the protein level variant is described as acquire, store, retrieve, process, validate, and interpret these QConCat (artificial, expressed proteins consisting of stable data and to eventually transform them into useful biological isotope-coded peptides representing the proteins to be quan- information. tified and coprocessed with the sample) [34] or PSAQ The best case scenario in terms of identification would be (protein standard absolute quantification, i.e., spiking of the to use only the mass information of a peptide as a unique sig- labeled protein of interest and coprocessing with the sample) nature. Such approaches have been described by Zubarev [35]. The strategy applied at a proteomic scale with determi- et al. [41] and later on by Conrads et al. [42] as the accurate nation and synthesis of labeled unique peptide identifiers for mass tag approach. In this technique, identification is based virtually all proteins to be analyzed is known under the con- only on the peptide mass and high-resolution instruments cept of proteotypic peptides [36]. The labeled proteotypic are needed to provide subpart-per-million mass accuracy peptide standard and its unlabeled, natural counterpart are (0.1 ppm). But even with such accuracy, high levels of con- typically monitored, identified, and quantified by a targeted fidence in protein identifications can be achieved only in MS/MS acquisition mode called selected-reaction or small eukaryotic systems (e.g., yeast). multiple-reaction monitoring (Fig. 3) [37,38]: quantitative Proteins can furthermore be identified with good through- analysis of selected ion transitions specific for each peptide put [43] and high sensitivity [44] based on the set of mea- enable, e.g., validation of clinical biomarkers in plasma sured proteolytic peptide masses. This process is known as [39]. Independent of the variations of the same quantification peptide mass fingerprinting. The experimental mass profile principle, these methods can be understood as highly sensi- is matched against those generated in silico from the protein tive, multiplexed ‘‘MS-based enzyme-linked immunosorbent sequences in the database using the same enzyme cleavage assays’’ that do not depend on three-dimensional structure- sites. The proteins are then ranked according to the number based recognition of protein epitopes. An initiative (Human of peptide masses matching their sequence within a certain Proteome Detection and Quantification Project [40]) has mass error tolerance. been recently proposed, based on this strategy, to develop In contrast, MS/MS provides access to sequence data, a complete suite of assays, e.g., two peptides from the protein which enables more confident peptide identifications. In an M. Kussmann and M. Affolter / Nutrition 25 (2009) 1085–1093 1089

Fig. 3. Targeted workflow in mass spectrometry–based proteomics with relative and absolute quantification options. IS, internal standard; MRM, multiple- reaction monitoring; Q, quadrupole; SRM, selected-reaction monitoring. MS/MS experiment, a precursor ion with a known mass is protein sequence) is created. The search is done against the selected from the previous MS scan and isolated for further target and the composite database. Assuming that no correct collision to produce daughter ions with unique signature. peptides are found in the target and decoy entities and that in- This process is described as peptide mass sequencing as correct assignments from target or decoy sequences are opposed to peptide mass fingerprinting. Identification of pro- equally likely, one can estimate the total number of false pos- teins using MS/MS data is currently performed using three itives. different approaches: 1) peptide sequence tagging [45],2) Apart from the progress in proteomics data processing, cross-correlation method [46], and 3) probability-based software platforms enabling cross-correlation with other matching [47]. ‘‘’’ sources (e.g., SBEAMS, Genedata Expressionist, Although peptide and protein identification and database Rosetta Elucidator, etc.) and supporting pathway interpreta- search programs such as Mascot [47] and Sequest [46] are tion (e.g., Ingenuity Pathway Analysis [http://www. well established, new software infrastructures for data ingenuity.com products/pathways_analysis.html], BioBase processing and validation have been built such as the Explain module [http://www.biobase-international.com/ SBEAMS architecture (http://www.sbeams.org) housing pages/index], AffyAnnotator [52]) are maturing rapidly. the Trans-Proteomic Pipeline and microarray modules that cover gene and protein expressions. The PeptideProphet Proteomics in nutrition—major applications [48] and ProteinProphet [49] modules in the Trans-Proteomic Pipeline are based on a robust and accurate statistical model Our group and others have contributed to the introduction to assess the validity of peptide identifications made by MS/ and adaptation of proteomics to the field of nutrition and MS and database search. The idea behind such resources is to health [53]: applications were summarized under topics provide the researcher with means to assess the quality of the such as nutritional intervention [1], elucidation of immune- data in a dataset-dependent manner and to control the tradeoff related gut disorders [54], characterization of functional in- between false positives (specificity) and false negatives gredients such as probiotics or milk and soy proteins [10], (sensitivity) [50]. The second strategy to elucidate the or the investigation of perturbed energy metabolism as in di- false-positive/false-negative tradeoff relies on a database abetes and obesity [55]. Moreover, numerous articles on nu- search using a target-decoy database [51]: first, an appropri- tritional intervention studies [56,57] and mechanistic ate ‘‘target’’ protein sequence database is generated and then elucidation of nutrient action [58] were published from our a ‘‘decoy’’ database preserving the general composition of side and others. The following citations focus on knowledge the target database while minimizing the number of peptide building in nutritionally relevant biological pathways and on sequences in common (generally done by reversing the target dietary intervention. 1090 M. Kussmann and M. Affolter / Nutrition 25 (2009) 1085–1093

Proteomic investigations of the enteric nervous system, tein-90 but exerted no quantifiable effect on normal prostate the nervous system of the gastrointestinal tract [59], have epithelial cells. the potential to deliver insights into gut functionality. For ex- The Daniel group also investigated the consequences of ample, it is recognized that early life events (such as neona- nutrient deficiencies by force-feeding rats with a zinc-defi- tal–maternal separation) predispose adults to develop cient diet and analyzing the hepatic , proteome, visceral pain and enhanced colonic motility in response to and lipidome. By the combined ‘‘omics’’ analysis prime met- acute stress. To better understand the molecular basis for abolic pathways of hepatic glucose and lipid metabolism and these functional gut disorders induced by environmental their changes in zinc deficiency could be identified that cause stress, our group has established a proteomic catalog of the liver lipid accumulation and hepatic inflammation [58]. In the rat intestine [60] and assessed stress effects on intestinal pro- same context, Fong et al. [69] showed that alleviation of zinc tein expression [61]. Barcelo´-Batllori et al. [62] investigated deficiency by zinc supplementation resulted in an 80% reduc- implications of cytokine-induced proteins in human intestinal tion of cyclo-oxygenase–2 mRNA, a key enzyme involved in epithelial cells that are related to the irritable bowel syndrome inflammation. [63]. The motivation behind this study is that cytokine-regu- Altered protein expression levels of fructose-induced fatty lated proteins in intestinal epithelial cells have been associ- liver in hamsters have been studied by Zhang et al. [70]. High ated with the pathogenesis of inflammatory bowel disease fructose consumption is associated with the development of [59]. Hang et al. [64] studied by a gel-based approach the fatty liver and dyslipidemia. Matrix-assisted laser desorption molecular mechanism of necrotizing enterocolitis, a serious ionization–MS-based proteomic analysis of the liver tissue gastrointestinal inflammatory disease, which frequently oc- from those hamsters revealed a number of proteins whose curs in preterm neonates who do no adapt to enteral nutrition. expression levels were altered more than two-fold. The iden- The functional assignment of the differentially expressed tified proteins have been grouped into categories such as fatty proteins revealed that important cellular functions, such as acid metabolism, cholesterol and triacylglycerol metabolism, the heat shock response, protein processing, and purine, molecular chaperones, enzymes in fructose catabolism, and nitrogen, and energy metabolism, were involved in the early proteins with housekeeping functions. progression of necrotizing enterocolitis [64]. These nutritional intervention studies looked at gene/pro- Five studies have employed proteomics, as such or com- tein abundance changes in response to a nutritional interven- bined with gene expression analysis, to address biomarkers tion. However, several food components may not only alter for protection against cancer. Breikers et al. [57] identified gene and protein expression but also target post-translational 30 proteins differentially expressed in the colonic mucosa modifications [71]. Diet-induced protein modifications can of healthy mice with increased vegetable intake. Six proteins ideally be assessed by proteomic techniques. For instance, identified with altered expression levels could be associated the protein phosphorylation status of the extracellular sig- with a protective role in colorectal cancer. The second study nal-regulated protein kinase (ERK) protein changes after ex- integrated DNA microarrays with proteomics to investigate posure to diallyl disulfide, a compound present in processed the effects of nutrients with suggested anticolorectal cancer garlic, an effect resulting in cell cycle arrest [72]. Another ex- properties and to develop a colon–epithelial cell line–based ample is the modification of thiol groups in the cytoplasmic screening assay for such nutrients [65]. Tan et al. [66] protein Keap1 [73]. This alteration of the protein redox status assessed the sodium butyrate effects on growth inhibition affects its binding to the protein Nrf2, which acts as a tran- of human HT-29 cancer cells in vitro by employing a 2D- scriptional regulator. MS–based proteomic strategy. Butyrate treatment altered the expression of various proteins, in particular those of the Conclusions and outlook ubiquitin–proteasome pathway, a result suggesting that pro- teolysis could be an important mechanism by which butyrate Nutrition is a young field for proteomics compared with regulates key proteins in the control of the cell cycle, apopto- clinical [30] and medical [32] applications. The success of sis, and differentiation. Combining gene and protein expres- proteomics in nutrition and health will depend on several fac- sion profiling in colonic cancer cells, Herzog et al. [67] tors. The proteomic technologic platforms as such, indepen- identified the flavonoid flavone, present in a variety of fruits dent of their application, will benefit from further advanced and vegetables, as a potent apoptosis inducer in human can- protein/peptide separation techniques, better depletion and cer cells. Flavone displayed a broad spectrum of effects on enrichment methods, and more sensitive and specific mass gene and protein expression that related to apoptosis induc- analysis techniques. tion and cellular metabolism. Aalinkeel et al. [68] evaluated The second area of platform-related improvements is bio- the effect(s) of the flavonoid quercetin on normal and malig- informatics. The tools to assess data quality and to convert nant prostate cells and identified possible target(s) of querce- data into interpretable information are improving rapidly tin action. Their findings demonstrated that quercetin [74]. Current ‘‘gaps’’ in ‘‘omic’’ datasets and hidden regula- treatment of prostate cancer cells resulted in decreased cell tion motifs upstream of the observed gene product regulation proliferation and viability. Quercetin promoted cancer cell may be elucidated by interpretation tools able to reconstruct apoptosis by downregulating the levels of heat shock pro- pathways and regulatory networks even in the presence of M. Kussmann and M. Affolter / Nutrition 25 (2009) 1085–1093 1091 fragmentary data [52]. If these network-reconstructing and to deliver clearer readouts from ‘‘omics’’ applications. As nutri- motif-elucidating tools prove to be successful, they may tional science develops into a holistic molecular science with also shed a different light on the terms reproducibility and systems biology character, all intervention studies, including comparability of ‘‘omic’’ studies: rather than searching for those that use proteomic approaches, should be based on stan- the same transcripts/proteins/metabolites found regulated be- dardized diets and ingredients and stratified cohorts and ideally tween related studies, one may focus on the common motifs follow the double-blinded, placebo-controlled crossover design. behind these—often at first glance divergent—datasets to Proteomics will continue to play a major role in systems find congruence between them. biology, because it can not only identify and quantify the The third area of proteomic method improvement con- ‘‘molecular robots’’ that do all the work in biological sys- cerns the analytical strategy. Intelligent focusing on pro- tems, but also map the networks of their physical interactions teome subsets, be it at the level of cell organelles, protein among each other and with nutrients, drugs, and other small subclasses [75,76], or the mass spectral level (targeted pro- molecules. An impressive example of such a thematic net- teomics with proteotypic peptides [77] or selected-reaction work establishment has been given by Bantscheff et al. monitoring [78]), will yield less complex proteomes but [76] who revealed mechanisms of action of clinical kinase provide deeper insights into molecular networks. inhibitors by MS profiling of small-molecule interactions Apart from this expected progress within proteomics, the with hundreds of endogenously expressed protein kinases. technology will largely profit from its cross-correlation with gene expression analysis at the mRNA level and metabolite References profiling. In the search of the causality and ‘‘wiring’’ among these three observational levels, one must be aware of the fact [1] Kussmann M, Affolter M, Nagy K, Holst B, Fay LB. Mass spectrom- that the interrelated timing of gene and protein expression etry in nutrition: understanding dietary health effects at the molecular and metabolite generation remains to be understood [79]. level. Mass Spectrom Rev 2007;26:727–50. One possible solution to this is addressing protein turnover [2] Kussmann M, Raymond F, Affolter M. OMICS-driven biomarker discovery in nutrition and health. J Biotechnol 2006;124:758–87. at a proteomic scale, i.e., rather than taking proteomic snap- [3] Kussmann M, Fay LB. Nutrigenomics and personalized nutrition: shots, interpreting protein abundance changes as a result of science and concept. Pers Med 2008;5:447–55. the interplay between changing protein synthesis and degra- [4] Ordovas JM, Corella D. . Annu Rev Genomics dation [80]. Proteome turnover information will add value to Hum Genet 2004;5:71–118. nutritional intervention studies performed with stable iso- [5] Kaput J. Nutrigenomics research for personalized nutrition and medi- cine. Curr Opin Biotechnol 2008;19:110–20. tope-labeled amino acids, peptides, and proteins. [6] Fay LB, German JB. Personalizing foods: is genotype necessary? Curr What takes proteomics in nutritional and food research Opin Biotechnol 2008;19:121–8. beyond the pure technologic developments is that human [7] Gallou-Kabani C, Vige A, Gross MS, Junien C. Nutri-: genetic heterogeneity comes into play. Genetic susceptibility lifelong remodelling of our epigenomes by nutritional and metabolic may predispose an individual to a diet-induced disease and factors and beyond. Clin Chem Lab Med 2007;45:321–7. [8] Junien C, Gallou-Kabani C, Vige A, Gross MS. [Nutritionnal epige- this also relates to biomarker profiles in individuals [81,82]. nomics: consequences of unbalanced diets on epigenetics processes This shall be further illustrated with a nutritionally relevant of programming during lifespan and between generations]. Ann Endo- example: Siffert [81] and Holtmann et al. [83] identified crinol (Paris) 2005;66(pt 3):2S19–28. and characterized metabolically relevant single nucleotide [9] Kussmann M, Affolter M. Proteomic methods in nutrition. Curr Opin polymorphisms in G-proteins, the latter representing an im- Clin Nutr Metab Care 2006;9:575–83. [10] Kussmann M, Affolter M, Fay LB. Proteomics in nutrition and health. portant ‘‘funnel’’ of cellular signaling. These polymorphisms Comb Chem High Throughput Screen 2005;8:679–96. predispose individuals of different ethnicity to having [11] Kussmann M, Daniel H. Editorial overview. Curr Opin Biotechnol a higher risk of developing hypertension, atherosclerosis, 2008;19:63–5. metabolic syndrome, or functional dyspepsia [81,83]. [12] Kussmann M, Rezzi S, Daniel H. Profiling techniques in nutrition and Epigenetic regulation such as DNA methylation (gene si- health research. Curr Opin Biotechnol 2008;19:83–99. [13] Fenn JB. Electrospray wings for molecular elephants (Nobel lecture). lencing) and histone acetylation (chromatin structure) should Angew Chem Int Ed Engl 2003;42:3871–94. ideally be included in nutritional systems biology, because [14] Tanaka K. The origin of macromolecule ionization by laser irradiation these mechanisms strongly influence gene transcription and (Nobel lecture). Angew Chem Int Ed Engl 2003;42:3860–70. expression. Remarkably, MS-based proteomic methods have [15] Makarov A, Denisov E, Kholomeev A, Balschun W, Lange O, started to contribute in this regard: Beck et al. [84] presented Strupat K, Horning S. Performance evaluation of a hybrid linear ion trap/orbitrap mass spectrometer. Anal Chem 2006;78:2113–20. a quantitative analysis of human histone post-translational [16] Nielsen ML, Savitski MM, Zubarev RA. Improving protein identifica- modifications, whereas Bonenfant et al. [85] focused on the tion using complementary fragmentation techniques in Fourier trans- histone codes of H2A and H2B variants. form mass spectrometry. Mol Cell Proteomics 2005;4:835–45. In humans, dietary changes represent rather subtle interven- [17] Jacobs JM, Adkins JN, Qian WJ, Liu T, Shen Y, Camp DG, Smith RD. tions often resulting in many small rather than a few large molec- Utilizing human blood plasma for proteomic biomarker discovery. J Proteome Res 2005;4:1073–85. ular changes, making data interpretation most difficult. [18] Gong Y, Li X, Yang B, Ying W, Li D, Zhang Y, et al. Different immu- Improved definition of human cohorts undergoing dietary inter- noaffinity fractionation strategies to characterize the human plasma ventions through proper geno- and phenotyping can be expected proteome. J Proteome Res 2006;5:1379–87. 1092 M. Kussmann and M. Affolter / Nutrition 25 (2009) 1085–1093

[19] Thingholm TE, Jensen ON, Larsen MR. Analytical strategies for phos- [41] Zubarev RA, Hakansson P, Sundqvist B. Accuracy requirements for phoproteomics. Proteomics 2009;9:1451–68. peptide characterization by monoisotopic molecular mass measure- [20] Mechref Y, Madera M, Novotny MV. Glycoprotein enrichment ments. Anal Chem 1996;68:4060–3. through lectin affinity techniques. Methods Mol Biol 2008; [42] Conrads TP, Anderson GA, Veenstra TD, Pasa Tolic L, Smith RD. 424:373–96. Utility of accurate mass tags for proteome-wide protein identification. [21] Wollscheid B, Bausch-Fluck D, Henderson C, O’Brien R, Bibel M, Anal Chem 2000;72:3349–54. Schiess R, et al. Mass-spectrometric identification and relative quanti- [43] Pappin DJ. Peptide mass fingerprinting using MALDI-TOF mass spec- fication of N-linked cell surface glycoproteins. Nat Biotechnol 2009; trometry. Methods Mol Biol 1997;64:165–73. 27:378–86. [44] Schuerenberg M, Luebbert C, Eickhoff H, Kalkum M, Lehrach H, [22] Sellers KF, Miecznikowski J, Viswanathan S, Minden JS, Eddy WF. Nordhoff E. Prestructured MALDI-MS sample supports. Anal Chem Lights, camera, action! Systematic variation in 2-D difference gel elec- 2000;72:3436–42. trophoresis images. Electrophoresis 2007;28:3324–32. [45] Mann M, Wilm M. Error-tolerant identification of peptides in sequence [23] Swanson SK, Washburn MP. The continuing evolution of shotgun databases by peptide sequence tags. Anal Chem 1994;33:4390–9. proteomics. Drug Discov Today 2005;10:719–25. [46] Eng JK, McCormack AL, Yates JR. An approach to correlate tandem [24] Moresco JJ, Dong MQ, Yates JR III. Quantitative mass spectrometry as mass spectral data of peptides with amino acid sequences in a protein a tool for nutritional proteomics. Am J Clin Nutr 2008;88:597–604. database. J Am Soc Mass Spectrom 1994;5:976–89. [25] Panchaud A, Hansson J, Affolter M, Bel RR, Piu S, Moreillon P, [47] Perkins DN, Pappin DJ, Creasy DM, Cottrell JS. Probability-based pro- Kussmann M. ANIBAL, stable isotope-based quantitative proteomics tein identification by searching sequence databases using mass by aniline and benzoic acid labeling of amino and carboxylic groups. spectrometry data. Electrophoresis 1999;20:3551–67. Mol Cell Proteomics 2008;7:800–12. [48] Keller A, Nesvizhskii AI, Kolker E, Aebersold R. Empirical statistical [26] Gygi SP, Rist B, Gerber SA, Turecek F, Gelb MH, Aebersold R. Quan- model to estimate the accuracy of peptide identifications made by MS/ titative analysis of complex protein mixtures using isotope-coded affin- MS and database search. Anal Chem 2002;74:5383–92. ity tags. Nat Biotechnol 1999;17:994–9. [49] Nesvizhskii AI, Keller A, Kolker E, Aebersold R. A statistical model [27] Ross PL, Huang YN, Marchese JN, Williamson B, Parker K, Hattan S, for identifying proteins by tandem mass spectrometry. Anal Chem et al. Multiplexed protein quantitation in Saccharomyces cerevisiae us- 2003;75:4646–58. ing amine-reactive isobaric tagging reagents. Mol Cell Proteomics [50] Urfer W, Grzegorczyk M, Jung K. Statistics for proteomics: a review of 2004;3:1154–69. tools for analyzing experimental data. Proteomics 2006;6:48–55. [28] Thompson A, Schafer J, Kuhn K, Kienle S, Schwarz J, Schmidt G, et al. [51] Elias JE, Gygi SP. Target-decoy search strategy for increased confi- Tandem mass tags: a novel quantification strategy for comparative anal- dence in large-scale protein identifications by mass spectrometry. Nat ysis of complex protein mixtures by MS/MS. Anal Chem 2003; Methods 2007;4:207–14. 75:1895–904. [52] Staab CA, Ceder R, Jagerbrink T, Nilsson JA, Roberg K, Jornvall H, [29] de Godoy LM, Olsen JV, de Souza GA, Li G, Mortensen P, Mann M. et al. processing of protein and transcript profiles of nor- Status of complete proteome analysis by mass spectrometry: SILAC mal and transformed cell lines indicates functional impairment of tran- labeled yeast as a model system. Genome Biol 2006;7:R50. scriptional regulators in buccal carcinoma. J Proteome Res 2007; [30] Mueller LN, Brusniak MY, Mani DR, Aebersold R. An assessment of 6:3705–17. software solutions for the analysis of mass spectrometry based quanti- [53] Fuchs D, Winkelmann I, Johnson IT, Mariman E, Wenzel U, Daniel H. tative proteomics data. J Proteome Res 2008;7:51–61. Proteomics in nutrition research: principles, technologies and applica- [31] Wong JW, Sullivan MJ, Cagney G. Computational methods for the tions. Br J Nutr 2005;94:302–14. comparative quantification of proteins in label-free LCn-MS experi- [54] Kussmann M, Blum-Sperisen S. OMICS-derived targets for inflamma- ments. Brief Bioinform 2008;9:156–65. tory gut disorders: opportunities for the development of nutrition re- [32] Brun V, Masselon C, Garin J, Dupuis A. Isotope dilution strategies for lated biomarkers. Endocr Metab Immune Disord Drug Targets 2007; absolute quantitative proteomics. J Proteomics 2009. Article in press; 7:271–87. doi:10.1016/j.jprot.2009.03.007. [55] Kussmann M, Affolter M. Proteomics and metabonomics routes to- [33] Gerber SA, Rush J, Stemman O, Kirschner MW, Gygi SP. Absolute wards obesity. In: Sorensen T, Cle´ment K, editors. Obesity—geno- quantification of proteins and phosphoproteins from cell lysates by mics and postgenomics. New York: Informa Health Care; 2008, p. tandem MS. Proc Natl Acad Sci U S A 2003;100:6940–5. 527–36. [34] Pratt JM, Simpson DM, Doherty MK, Rivers J, Gaskell SJ, Beynon RJ. [56] Fuchs D, Piller R, Linseisen J, Daniel H, Wenzel U. The human periph- Multiplexed absolute quantification for proteomics using concatenated eral blood mononuclear cell proteome responds to a dietary flaxseed-in- signature peptides encoded by QconCAT genes. Nat Protoc 2006; tervention and proteins identified suggest a protective effect in 1:1029–43. atherosclerosis. Proteomics 2007;7:3278–88. [35] Dupuis A, Hennekinne JA, Garin J, Brun V. Protein standard absolute [57] Breikers G, van Breda SG, Bouwman FG, van Herwijnen MH, Renes J, quantification (PSAQ) for improved investigation of staphylococcal Mariman EC, et al. Potential protein markers for nutritional health food poisoning outbreaks. Proteomics 2008;8:4633–6. effects on colorectal cancer in the mouse as revealed by proteomics [36] Mallick P, Schirle M, Chen SS, Flory MR, Lee H, Martin D, et al. Com- analysis. Proteomics 2006;6:2844–52. putational prediction of proteotypic peptides for quantitative proteo- [58] tom-Dieck H, Doring F, Fuchs D, Roth HP, Daniel H. Transcriptome mics. Nat Biotechnol 2007;25:125–31. and proteome analysis identifies the pathways that increase hepatic [37] Lange V, Picotti P, Domon B, Aebersold R. Selected reaction moni- lipid accumulation in zinc-deficient rats. J Nutr 2005;135:199–205. toring for quantitative proteomics: a tutorial. Mol Syst Biol 2008; [59] Hansen MB. The enteric nervous system I: organisation and classifica- 4:222. tion. Pharmacol Toxicol 2003;92:105–13. [38] Yocum AK, Chinnaiyan AM. Current affairs in quantitative targeted [60] Marvin-Guy L, Lopes LV, Affolter M, Courtet-Compondu MC, proteomics: multiple reaction monitoring–mass spectrometry. Brief Wagniere S, Bergonzelli GE, et al. Proteomics of the rat gut: analysis Funct Genomic Proteomic 2009;8:145–57. of the myenteric plexus-longitudinal muscle preparation. Proteomics [39] Schiess R, Wollscheid B, Aebersold R. Targeted proteomic strategy for 2005;5:2561–9. clinical biomarker discovery. Mol Oncol 2009;3:33–44. [61] Lopes LV, Marvin-Guy LF, Fuerholz A, Affolter M, Ramadan Z, [40] Anderson NL, Anderson NG, Pearson TW, Borchers CH, Kussmann M, et al. Maternal deprivation affects the neuromuscular Paulovich AG, Patterson SD, et al. A human proteome detection and protein profile of the rat colon in response to an acute stressor later in quantitation project. Mol Cell Proteomics 2009;8:883–6. life. J Proteomics 2008;71:80–8. M. Kussmann and M. Affolter / Nutrition 25 (2009) 1085–1093 1093

[62] Barcelo´-Batllori S, Andre´ M, Servis C, Le´vy N, Takikawa O, of Keap1 are the sensors regulating induction of phase 2 enzymes that Michetti P, et al. Proteomic analysis of cytokine-induced proteins in hu- protect against carcinogens and oxidants. Proc Natl Acad Sci USA man intestinal epithelial cells: implications for inflammatory bowel 2002;99:11908–13. diseases. Proteomics 2002;2:551–60. [74] Panchaud A, Affolter M, Moreillon P, Kussmann M. Experimental and [63] Wood JD, Alpers DH, Andrews PL. Fundamentals of neurogastroenter- computational approaches to quantitative proteomics: status quo and ology. Gut 1999;45(suppl 2):II6–16. outlook. J Proteomics 2008;71:19–33. [64] Hang P, Sangild PT, Sit WH, Ngai HH, Xu R, Siggers JL, Wan JM. [75] Mancone C, Amicone L, Fimia GM, Bravo E, Piacentini M, Tripodi M, Temporal proteomic analysis of intestine developing necrotizing Alonzi T. Proteomic analysis of human very low-density lipoprotein by enterocolitis following enteral formula feeding to preterm pigs. J Pro- two-dimensional gel electrophoresis and MALDI-TOF/TOF. Proteo- teome Res 2009;8:72–81. mics 2007;7:143–54. [65] Stierum R, Burgemeister R, van Helvoort A, Peijnenburg A, Schutze K, [76] Bantscheff M, Eberhard D, Abraham Y, Bastuck S, Boesche M, Seidelin M, et al. Functional food ingredients against colorectal cancer. Hobson S, et al. Quantitative chemical proteomics reveals mechanisms An example project integrating , nutrition and of action of clinical ABL kinase inhibitors. Nat Biotechnol 2007; health. Nutr Metab Cardiovasc Dis 2001;11(suppl):94–8. 25:1035–44. [66] Tan S, Seow TK, Liang RCMY, Koh S, Lee CPC, Chung MCM, [77] Mallick P, Schirle M, Chen SS, Flory MR, Lee H, Martin D, et al. Com- Hooi SC. Proteome analysis of butyrate-treated human colon cancer putational prediction of proteotypic peptides for quantitative proteo- cells (HT-29). Int J Cancer 2002;98:523–31. mics. Nat Biotechnol 2007;25:125–31. [67] Herzog A, Kindermann B, Doring F, Daniel H, Wenzel U. Pleiotropic [78] Anderson L, Hunter CL. Quantitative mass spectrometric multiple re- molecular effects of the pro-apoptotic dietary constituent flavone in action monitoring assays for major plasma proteins. Mol Cell Proteo- human colon cancer cells identified by protein and mRNA expression mics 2006;5:573–88. profiling. Proteomics 2004;4:2455–64. [79] Nicholson JK, Connelly J, Lindon JC, Holmes E. Metabonomics: a plat- [68] Aalinkeel R, Bindukumar B, Reynolds JL, Sykes DE, Mahajan SD, form for studying drug toxicity and gene function. Nat Rev Drug Chadha KC, Schwartz SA. The dietary bioflavonoid, quercetin, se- Discov 2002;1:153–61. lectively induces apoptosis of prostate cancer cells by down-regulat- [80] Doherty MK, Beynon RJ. Protein turnover on the scale of the proteome. ing the expression of heat shock protein 90. Prostate 2008; Expert Rev Proteomics 2006;3:97–110. 68:1773–89. [81] Siffert W. G protein polymorphisms in hypertension, atherosclerosis, [69] Fong LY, Zhang L, Jiang Y, Farber JL. Dietary zinc modulation of and diabetes. Annu Rev Med 2005;56:17–28. COX-2 expression and lingual and esophageal carcinogenesis in rats. [82] Chen Y, Rollins J, Paigen B, Wang X. Genetic and genomic insights J Natl Cancer Inst 2005;97:40–50. into the molecular basis of atherosclerosis. Cell Metab 2007;6:164–79. [70] Zhang L, Perdomo G, Kim DH, Qu S, Ringquist S, Trucco M, [83] Holtmann G, Siffert W, Haag S, Mueller N, Langkafel M, Senf W, et al. Dong HH. Proteomic analysis of fructose-induced fatty liver in ham- G-protein beta 3 subunit 825 CC genotype is associated with unex- sters. Metabolism 2008;57:1115–24. plained (functional) dyspepsia. Gastroenterology 2004;126:971–9. [71] Davis CD, Milner J. Frontiers in nutrigenomics, proteomics, metabolo- [84] Beck HC, Nielsen EC, Matthiesen R, Jensen LH, Sehested M, Finn P, mics and cancer prevention. Mutat Res 2004;551:51–64. et al. Quantitative proteomic analysis of post-translational modifica- [72] Knowles LM, Milner JA. Diallyl disulfide induces ERK phosphoryla- tions of human histones. Mol Cell Proteomics 2006;5:1314–25. tion and alters gene expression profiles in human colon tumor cells. [85] Bonenfant D, Coulot M, Towbin H, Schindler P, van Oostrum J. Char- J Nutr 2003;133:2901–6. acterization of histone H2A and H2B variants and their post-transla- [73] Dinkova-Kostova AT, Holtzclaw WD, Cole RN, Itoh K, tional modifications by mass spectrometry. Mol Cell Proteomics Wakabayashi N, Katoh Y, et al. Direct evidence that sulfhydryl groups 2006;5:541–52.