Mass spectrometry in nutrition: Understanding dietary health effects at the molecular level

Authors


Abstract

In modern nutrition research, mass spectrometry has developed into a tool to assess health, sensory as well as quality and safety aspects of food. In this review, we focus on health-related benefits of food components and, accordingly, on biomarkers of exposure (bioavailability) and bioefficacy. Current nutrition research focuses on unraveling the link between dietary patterns, individual foods or food constituents and the physiological effects at cellular, tissue and whole body level after acute and chronic uptake. The bioavailability of bioactive food constituents as well as dose-effect correlations are key information to understand the impact of food on defined health outcomes. Both strongly depend on appropriate analytical tools to identify and quantify minute amounts of individual compounds in highly complex matrices—food or biological fluids—and to monitor molecular changes in the body in a highly specific and sensitive manner. Based on these requirements, mass spectrometry has become the analytical method of choice with broad applications throughout all areas of nutrition research. The current review focuses on selected areas of application: protein and peptide as well as nutrient and metabolite analysis. © 2007 Wiley Periodicals, Inc., Mass Spec Rev 26:727–750, 2007

I. INTRODUCTION

Modern mass spectrometry is an indispensable analytical platform and a widely applied tool in modern nutrition research. The technology is employed to address health, sensory as well as quality and safety aspects of food. In nutrition research, mass spectrometry is implicated in three health-related application areas, that is, the discovery and measurement of biomarkers for pre-disposition, exposure and efficacy. Since pre-disposition covers a broad area, which would go beyond the scope of this review, we focus on the latter two.

While traditional nutrition research has been aimed at providing nutrients to nourish populations and preventing specific nutrient deficiencies, for example, vitamin deficiencies, it more recently attempts to explore health related aspects of individual bioactive components. Modern nutrition research targets health promotion, disease prevention, performance improvement, and risk assessment. The concept of developing functional food requires the understanding of the disease etiology and, on that basis, the mechanisms of prevention followed by the identification of potentially active food constituents and the demonstration of their bioavailability and efficacy. Biomarkers have been proven as being an important tool to understand the latter. When combined with biomarkers of susceptibility, they should help to understand and explain the responsiveness of the human to dietary components, whole food or individual compounds, and ultimately reveal the relationship between nutrition and health (Fig. 1).

Figure 1.

Classes of biomarkers to describe the pathway from exposure to health/disease outcome.

Biomarker development and measurement highly depend on appropriate analytical tools to identify and quantify often minute amounts of bioactive compounds in food or biological specimens and to monitor molecular changes in the body in a highly specific and sensitive manner. Based on these requirements, mass spectrometry has become the analytical method of choice, offering a broad spectrum of instrumentation for biomarker discovery, characterization and quantification.

Markers for bioavailability and bioefficacy can be addressed in a hypothesis-driven, targeted fashion or at global scale without any pre-assumption. Nutrition research is taking advantage of proteomics (global protein analysis) and metabolomics (global metabolite analysis) to discover biologically active food components, to assess their bioavailability at different organ sites as well as their site specific, and eventually overall, health benefits. Independent of the context of application, proteomics represents the only platform that delivers not only markers for disposition and efficacy but also targets of intervention. Proteomic developments, applications and potential in the field of nutrition have been specifically addressed in another review (Kussmann, Affolter, & Fay, 2005).

The asset of metabolomics is the quantitative, non-invasive analysis of easily accessible human body fluids (urine, blood, saliva, tears). Metabolomics has gained a strong impact on the field of nutrition (Whitfield, German, & Noble, 2004; Kochhar et al., 2006; Wang et al., 2006). While global metabolite profiling in nutrition has been addressed in one of our previous papers (Kussmann, Raymond, & Affolter, 2006), we focus here on targeted nutrient and metabolite analysis and on stable-isotope approaches to elucidate bioavailability and efficacy of nutrients belonging to specific compound classes.

Beyond the scope of the present article, mass spectrometers are also utilized to address other aspects of food and nutrition research than health. Practically all modern mass spectrometric techniques have fed into analytical platforms for nutrition research. The instruments are for example used to investigate the pleasure aspects of food, that is, its sensory qualities. Flavor is the sum of taste and olfaction. LC-MS/MS techniques typically serve to characterize taste molecules, whereas GC-MS instruments analyze the volatile molecules that make up for olfaction of food and drinks (Fay & Brevard, 2005). Safety and quality assessment employ a large range of mass spectrometric instrumentation from isotope ratio MS (IR-MS) for authenticity checks to hyphenated MS techniques for ensuring quality, safety and compliance of food products.

II. TECHNOLOGY REVIEW

A. Mass Spectrometry of Nutrients and Metabolites

Mass spectrometry coupled to gas chromatography (GC-MS) has proved to be a useful tool to analyze volatile fractions of food and drinks (Maeztu et al., 2001; Blank, 2002), but also to quantify non-volatile nutrients like amino acids (Simpson et al., 1996; Chen et al., 1998; Kugler et al., 2006), fatty acids (Corso et al., 1995; Fedrigo, Pivato, & Traldi, 1999; Blanco-Gomis et al., 2001) and sugars (Fox, 1999; Sanz, Sanz, & Martinez-Castro, 2004).

Today, LC-MS based techniques find broader applications in nutrition research than GC-MS methods as they provide means to investigate nutrient stability, absorption, metabolism and metabolite excretion. Hence, LC-MS approaches can bridge the gap between the content of a particular ingredient in a food matter and its biological end point (Yeboah & Konishi, 2003; German & Watkins, 2004).

Two strategies can be followed for nutrient and metabolite analysis in biological systems:

  • (1)The non-targeted conception, which includes fingerprinting of samples and evaluating the unidentified signals with statistical methods (e.g., principal component analysis, linear discriminant analysis, discriminant function analysis etc.) or
  • (2)The targeted conception, which measures only selected compounds with pre-determined properties such as a given mass or structure as described later.

Two major identification techniques for nutrient metabolites are currently applied, where tandem mass spectrometry enables an elucidation and confirmation of metabolite structures, while accurate mass measurements in combination with isotope pattern evaluations provide elemental composition of the metabolite. Furthermore, hydrogen/deuterium exchange experiments are combined with mass spectrometry (Ma, Chowdhury, & Alton, 2006) to distinguish between isomeric structures of analytes.

Quantification of nutrients and metabolites is mostly achieved using the selected reaction monitoring technique on quadrupole-based platforms. Quadrupole analyzers provide the best linear and dynamic range for biological samples, but application of high resolution with a ToF or Orbitrap analyzer can also be desired to separate the signal of interest from interferences thereby enhancing detection limits.

In this chapter, we focus on the analysis of vitamins and phytochemicals. Apart from adapting MS parameters to these analytes, there is a wide variety of different sample preparation techniques available for the clean-up of vitamins and phytochemicals including deproteinization, solid phase extraction, liquid–liquid extraction, filtration, and dialysis.

Independent of these purification procedures, all modern applications utilize isotope dilution for quantification (De Bievre, 1993; Beylot, 1994; Thienpont, Stockl, & de Leenheer, 1995). Stable-isotope labeled internal standards offer five advantages:

  • (1)As these internal standards are chemically identical and physically similar to the target analytes, maximized correction for errors introduced during sample preparation, storage and detection steps can be achieved.
  • (2)Distinct and parallel detection of unlabeled, endogenous analytes and labeled administered/injected analytes becomes feasible, resulting in minimized uncontrolled, endogenous interferences (e.g., absorption of the studied compound from unknown sources).
  • (3)The fate of different forms of the same compound (free, acetylated, glucuronidated, sulfated) can be specifically monitored provided they are labeled to different degrees.
  • (4)It is possible to study the effect of food structure on the absorption if the compared food types (e.g., solid food or drink) contain the nutrient in differently labeled forms.
  • (5)The fate of nutrients introduced in different modes to the human body (e.g., absorption via lumen or injected into circulation) can be followed distinctly.

To detect stable isotope labeled compounds at unusually small quantities or with minimal labeling (with only one or two labels), isotope ratio mass spectrometry (IR-MS) can be used (Pont et al., 1997). This type of mass spectrometry is based on special instrumentation and is dedicated to the very accurate determination of isotope ratios (mostly 2H/1H, 13C/12C, 15N/14N, 18O/16O) in different biological samples (Lichtfouse, 2000). IR-MS can be considered rather as a complementary than competitor method to the “conventional MS methods.” For example, using GC- or LC-MS, the measurable range of different isotopomers ranges between 0.1% and 100 % while with IR-MS, the isotopic ratio determination complementarily covers the range from 0.001% to 0.1% (Brenna et al., 1997; Sessions, 2006). Stable isotope abundances are expressed as the ratio of the two isotopes of interest in the sample compared to the same ratio in an international reference standard, using the “delta” (δ) notation. Since the differences in ratios between the sample and standard are usually very small, they are expressed as parts per thousand (‰) or per million deviation from the standard.

More rarely, highly specialized MS techniques, for example, accelerator mass spectrometry (Vogel et al., 1995; Vogel & Turteltaub, 1998; Jackson, Weaver, & Elmore, 2001) are deployed to monitor the fate of nutrients, especially micronutrients in the body over a longer period of time, including their bioconversion and tissue distribution (Dueker et al., 2000; Ho et al., 2004; Vuong et al., 2004). This ultra-sensitive analytical technique measures the amount of naturally rare, radioactive nuclides such as 3H, 14C, 41Ca, etc. The advantage of measuring naturally rare radioactive nuclides is that the background of such atoms in biological systems is low and thus very small concentration changes can be detected providing very good sensitivity and dynamic range. According to this, low but physiologically relevant doses of micronutrients can be administered and investigated. These doses cause minimal radiation exposures below safety thresholds. In addition, the nutrient metabolites can be traced for prolonged periods of time. Major disadvantages of AMS are the often problematic purchase of radio-labeled compounds and that the measured information is not compound specific since AMS pools all different metabolites during the combustion step and therefore abandons selectivity.

B. Mass Spectrometry of Proteins and Peptides

The following features have rendered modern mass spectrometers as powerful tools for protein identification: tandem mass spectrometry (MS/MS) for peptide sequencing (Biemann, 1993); high mass accuracy (to date: low to sub-ppm range) both in MS and MS/MS mode (Kaiser, Anderson, & Bruce, 2005); high resolving power up to several 100'000 (Bogdanov & Smith, 2005).

Since the introduction of tandem mass spectrometry and of the combination of mass analyzers based on different principles, hybrid instruments (such as the Q-ToF) further expanded the number of mass spectrometric platforms. Quadrupole (Q) analyzers, if aligned in a sequential fashion, enable the acquisition of diagnostic fragment ions and mass losses (Cox et al., 2004). Their high specificity is particularly useful for high-throughput analysis of post-translational modifications (PTMs), but their performance is compromised by limited sensitivity (Mann & Jensen, 2003). The Paul-trap allows for sensitive MS/MS and multiple-stage (MSn) experiments (Cooks et al., 1991; Cox et al., 1992). Time-of-flight (ToF) analyzers offer high speed and sensitivity. These three major types of mass analyzers are often combined to tandem and hybrid mass spectrometers and are commonly utilized for protein and peptide analysis.

A different way of analyzing peptide and protein ions is the so-called Fourier-transform ion cyclotron resonance (FT-ICR) technique (Buchanan & Hettich, 1993; McIver, Li, & Hunter, 1994): ions are analyzed by their mass-dependent frequency resulting in unsurpassed mass accuracy (sub-ppm) and resolution (>several 100'000). FT-ICR opens the way for so-called top-down protein analysis (intact proteins are characterized directly, without previous proteolytic or chemical processing) (Zubarev et al., 2000; Ge et al., 2002) and for micro-heterogeneity analysis of complex glycoprotein mixtures (Nagy et al., 2004). FT-ICR-MS is particularly powerful for peptide sequencing when combined with electron-capture dissociation (ECD), which generates clean c- and z-ion series (Roepstorff & Fohlman, 1984) not observed on other instruments.

To reduce the fragmentation complexity of peptides, a novel technique called electron-transfer dissociation (ETD) was introduced, which can be combined with conventional Paul traps (Syka et al., 2004; Hartmer & Lubeck, 2005) and hence by-passes the necessity of an investment into an expensive ECD-FT-ICR system. Moreover, the recently developed Orbitrap mass spectrometer (Hu et al., 2005; Yates et al., 2006) also determines peptide and protein molecular masses by measuring their frequencies and offers mass accuracy comparable to FT-ICR combined with a resolution power around 80'000.

Quantitative proteomics follow four main avenues: (a) differential protein labeling with dyes, two-dimensional gel electrophoresis (2DE) and relative quantification by imaging (Quadroni & James, 1999; Görg et al., 2000); (b) differential isotope-coded labeling (ICL), chromatography and relative quantification by mass spectrometry (Gygi et al., 1999; Han et al., 2001; Zhang et al., 2001); (c) label-free relative peptide and protein quantification by direct comparison of LC-MS/MS data and display of three-dimensional peptide maps (m/z, retention time and peak intensity) (Higgs et al., 2005); and (d) absolute protein quantification by spiking “proteotypic” peptides representative of each protein of interest as internal standards (Gerber et al., 2003). In view of this multitude of quantitative proteomic approaches, research groups have embarked on systematic comparisons of such techniques with real-life samples. Wu et al. (2006) compared DIGE, cICAT (cleavable isotope-coded affinity tag; quantification at MS level) and iTRAQ (multiplexed isobaric tagging; quantification at MS/MS level) for differential proteomics. These protein chemistries were combined with either 2DE- or LC-MALDI-ToF-ToF analysis. The sensitivity of the methods was assessed at peptide level: iTRAQ was found to be superior to ICAT, which in turn equaled DIGE in performance. The limited overlap of identified proteins suggested the complementarity of these methods. A. Heck's group has benchmarked DIGE and SILAC (stable-isotope labeling of amino acids in cell culture) against each other (Kolkman et al., 2005). Their assessment revealed good correlation between the two techniques, which both delivered reproducible quantification results. Finally, a recent article by Mann et al. reports on status and performance of SILAC / FTMS-based quantitative proteomics (de Godoy et al., 2006). The paper presents hard facts and figures of what can be achieved today in terms of proteome coverage and protein LOD/LOQ at global scale and also suggests areas of methods and instrument improvement.

III. MASS SPECTROMETRY IN NUTRITION

A. Nutritionally Bioactive Proteins and Peptides

1. Milk

Milk is a rich source of functional peptides and proteins beneficial for human health. A review of bioactive peptides and proteins present in milk and dairy products has recently been published by Pihlanto and Korhonen (2003).

Traditionally, milk and derived fractions have been extensively analyzed in terms of composition, physicochemical properties and biological functions and mass spectrometry is contributing to different areas of milk and dairy research (reviewed by Alomirah, Alli, & Konishi, 2000 and Ferranti, 2004). These studies include identification of milk protein variants and glycoforms, detection of milk adulteration (falsification of milk with non-dairy ingredients) and identification of peptides in dairy products. Moreover, mass spectrometry has become an indispensable technique for the quality assessment of milk- and dairy based products (reviewed in Guy & Fenaille, 2006).

Milk, colostrum, and other fractions from different species were investigated by several groups leading to cataloguing the protein complement in these samples (Miranda et al., 2004; D'Auria et al., 2005) including low-abundance proteins (Yamada et al., 2002). The profiling of human colostrum revealed, after immuno-depletion of high-abundant milk proteins, a list of 151 low abundant proteins, 83 of which have not been previously reported in human colostrum or milk (Palmer et al., 2006).

The milk fat globule membrane (MFGM) is a particularly rich source of bioactive peptides and proteins (Charlwood et al., 2002; Fortunato et al., 2003). Reinhardt and Lippolis (2006) analyzed the composition of bovine MFGM and identified 120 proteins, 71% of which were membrane associated. This lipid fraction of human milk consists mainly of triglycerides enveloped in tri-laminar units of biological cell membranes. MFGM is derived from the apical region of the mammary gland epithelial cells and is budded off around the milk lipids, the latter being secreted by the mammary gland cells. It constitutes a specific component of milk, produced via a specific cellular biosynthetic process that provides distinctive functional attributes to this source of nutrition. MFGM is considered to be similar to any other eukaryotic cell membrane and accounts for 2–4% of the total human milk protein content. MFGM may therefore contain—in addition to molecules previously described to be associated with this membrane (mucins, lactadherin, adipophilin, CD 36, and butyrophilin; Mather, 2000)—other factors to date thought to be exclusively found in cellular membranes. Despite the large body of knowledge about its unusual biochemical structure, little is known about the physiological function of MFGM for the nursing infant. As such, it bears great potential for the identification of new proteins in milk and the exploitation of these proteins for dairy product development (Cavaletto et al., 2002; Cavaletto, Giuffrida, & Conti, 2004).

A considerable effort of milk proteomics has focused on the elucidation of post-translational modifications such as glycosylation and phosphorylation (Kjeldsen et al., 2003; Holland, Deeth, & Alewood, 2004, 2005), of protein alterations as means to characterize milk (Galvani, Hamdan, & Righetti, 2001; Chen et al., 2004), and of covalent complexes between milk proteins (e.g., caseins and β-lactoglobulin) (Henry et al., 2002). Casein micelles, for example, formed by interaction of milk caseins and calcium phosphate, maintain a super-saturated calcium phosphate concentration in milk, providing the newborn with sufficient calcium phosphate for the mineralization of the rapidly growing calcified tissues. The phosphorylation state of caseins plays an important role in the interaction with calcium phosphate and thereby in the organization of the micelles (Sorensen et al., 2003). Other trace elements associated with milk proteins, such as iron in lactoferrin (Bratter et al., 1998), are important constituents to provide the newborn with essential nutrients. Inductively coupled plasma (ICP) MS has the analytical potential for “element-tagged” proteomics of milk resulting in quantitative information on multi-element distribution patterns in different milk sources (Sanz-Medel et al., 2003).

Specific milk fractions have been shown to alleviate various health disorders like inflammation (Goldman et al., 1986), hypercholesterolemia (Sharpe, Gamble, & Sharpe, 1994) and osteoarthritis (Colker et al., 2002), however, without addressing molecular mechanisms or identifying ingredients responsible for action. More recently, milk peptides and hydrolysates moved into the focus of studies of bioactive compounds. Peptides with opioid, antihypertensive, antithrombotic, immunomodulating, and metal-binding activities have been described in a recent review (Severin & Xia, 2005).

Bioactivities of peptides encrypted in major milk proteins are latent until released and activated, for example, during gastrointestinal digestion or food processing. Bioactive peptides can be produced in vivo following intake of milk proteins. Moreover, the proteolytic system of bacterial species used in fermentation (e.g., yogurt, cheese) can contribute to the liberation of bioactive peptides or precursors thereof (Meisel, 2004). A wide range of proteins contains concealed functional units that can be liberated to generate novel bioactivities. Autelitano et al. (2006) term this “hidden” subset of peptides, residing within the proteome, the “cryptome” and it is suggested to represent a vast array of cryptic peptides, or “crypteins,” with manifold bioactivities that can be liberated from the parent protein via proteolytic cleavage.

Hernandez-Ledesma et al. (2005) evaluated the angiotensin-converting enzyme (ACE)-inhibitory activity and antioxidant properties of a commercial fermented milk by LC-MS/MS analysis of peptides from the most active fractions. These peptides were released by either milk fermentation with Lactobacilli or Streptococci species or simulated gastrointestinal digestion (Hernandez-Ledesma et al., 2004) and originated mainly from milk casein proteins (Fig. 2). ACE-inhibitory peptides can also be found in cheese manufactured with different starter cultures but sample preparation prior to ESI-MS/MS analysis is laborious (Gomez-Ruiz, Ramos, & Recio, 2002).

Figure 2.

Identification of biologically active peptides produced by milk fermentation and simulated gastrointestinal digestion (Reproduced from Hernandez-Ledesma et al., 2004, with permission from Elsevier, copyright 2004).

Purification and characterization of novel peptide antibiotics from human milk was described by Liepke et al. (2001). Digestion of human milk by infants was simulated by using pepsin under acidic conditions to generate peptides with antimicrobial activity. LC fractionation followed by MALDI-MS analysis allowed the identification of novel casein- and lactoferrin-derived fragments, which inhibited growth of bacteria and yeasts.

Mass spectrometry was also used to trace the pattern of breakdown and formation of potential bioactive peptides during casein proteolysis of human milk. The action of a plasmin-like enzyme acting on specific lysine residues is the primary step in casein degradation. Endopeptidase- and/or exopeptidase-mediated cleavage of the oligopeptides produces a multiplicity of short peptides differing by one or more amino acid residues (Ferranti et al., 2004).

In order to be bioavailable and bioefficaceous, peptides must be absorbed in the intestine in an active form. Whereas di- and tri-peptides are readily absorbed in the gut, it is not clear whether larger bioactive peptides are absorbed from the intestine and reach the target organ(s). Most of the claimed physiological properties of casein-derived bioactive peptides have been carried out in vitro or in animal model systems and the hypothesized properties remain to be proved in humans (Severin & Xia, 2005).

Breast-fed newborns have been shown to experience a lower incidence of gastro-intestinal infections and inflammatory, respiratory, and allergic diseases. This finding has been attributed to a diversity of protective factors in breast milk (Lonnerdal, 2003). One specific biological activity in mother's milk was reported to be the one of soluble CD14 (sCD14) (Labeta et al., 2000; Vidal et al., 2001). The study of this protein indicated a sentinel role for sCD14 during bacterial colonization of the gut and suggested sCD14 to be involved in modulating local innate and adaptive immune responses, thus controlling homeostasis in the neonatal intestine. A recent study revealed an interaction between soluble Toll-like receptor 2 (sTLR2) and sCD14 in plasma and milk, suggesting the existence of a novel innate immune mechanism regulating microbially induced TLR triggering (LeBouder et al., 2003). A particular fraction of human milk, generated by a special chromatography based on restricted access material (RAM), was characterized by 2D LC-MS/MS in order to elucidate the protein composition and to discover novel molecules that potentially interact with sCD14 (Panchaud, Kussmann, & Affolter, 2005). In a different immunological context, namely allergy against cow's milk, Natale et al. (2004) characterized milk allergens by 2DE immunoblotting and mass spectrometry. The serum from 20 milk-allergic subjects was searched for major cow's milk allergens followed by MALDI-ToF-MS identification of the proteins.

Lactosylated proteins of infant formula powders were investigated and this resulted in the identification of α-lactalbumin with five lactosylated peptides (Marvin et al., 2002). These may serve as protein markers to detect chemical modification induced by milk processing and/or storage.

2. Soy

Soybeans are a source of complete protein, containing high amounts of all essential amino acids. Foods rich in soy protein are often labeled with a health claim indicating the potential lowering of heart disease risk (Erickson, 2005). The bioactive compounds, however, are generally not the soy proteins per se but rather the phytochemicals (e.g., isoflavones, saponins) present in the soy protein isolate (Fang, Yu, & Badger, 2004). One study described by Kim et al. (2006) compares different 2D-gel brain protein profiles from rats, obtained after administration of grape seed extract (rich in proanthocynidin) which are similar to polyphenol-enriched preparations from blueberry and soy.

Nevertheless, dietary soy protein preparations used in clinical studies are often poorly characterized and standardized, a fact that motivated Gianazza et al. (2003) to investigate soy protein isolates prepared for clinical studies in Europe and the United States by 2DE and MALDI-ToF-MS. A predominance of breakdown products of the 7S globulin and mainly intact 11S globulin subunits was found in soy concentrates used in European studies. By contrast, soy isolates administered in the United States showed none of the major components corresponding to 7S globulin subunits; only some of the light chains of 11S were intact, and heavy chains of 11S also were fragmented.

In a dietary intervention study published by Junghans et al. (2004), transcriptomic and proteomic analysis of pig liver tissue revealed significant differences after exposure to a soy protein diet compared a casein diet both at gene and protein expression level. Both Omic approaches led in part to the identification of the same diet-associated expressed molecules (plasminogen, trypsin, phospholipase A2, glutathione-S-transferase alpha, retinal binding protein) or at least molecules belonging to the same metabolic pathways (protein and amino acid metabolism, oxidative stress response, lipid metabolism).

A proteomic strategy was used to study the protein profiles of livers of rainbow trout that had been fed two diets containing different proportions of plant ingredients (Martin et al., 2003). The diet with a greater proportion of soybean meal increased protein consumption and protein synthesis rates. Approximately 800 liver proteins were quantitatively analyzed and 33 were found to be differentially expressed between the diets.

Jeong et al. (2003) have characterized lunasin, a novel and promising chemopreventive peptide from soybean by mass spectrometric peptide mapping, partial purification, bioactivity measurement of the resulting fractions and by protein expression in the developing seed. The identity of lunasin in the soybean seed extracts was established by Western blot analysis and mass spectrometric peptide mapping.

Protein profiles of wild and cultivated genotypes of soybean seeds using proteomic tools were compared by Natarajan et al. (2006). The total number of storage protein spots detected in wild and cultivated genotypes was 44 and 34, respectively. However, extraction of soybean seed proteins is challenging and requires optimized procedures to obtain consistent results for 2DE separation and MS detection (Natarajan et al., 2005).

B. Biomarkers to Describe the Relationship of Nutrition and Health

Diet is implicated in a myriad of diseases involving events in the first weeks of life (folate and neural tube defects) to chronic and age-related illnesses, for example, cancer and cardiovascular disease. Thus, at least 35% of human cancers in developed countries are thought to have a dietary component with the potential to increase or decrease the disease risk (http://www.wcrf.org/research/fnatpoc.lasso). Many diet-related pathological conditions derive from an unbalanced or inadequate diet. Large proportions of the world population suffer from severe deficiencies of major nutrient groups; other populations are affected by diseases due to an overconsumption of nutrients especially certain fats and carbohydrates.

Since macronutrient and mineral requirements are generally well established, modern nutrition research places special emphasis on minor dietary constituents, vitamins and trace elements, phytochemicals (carotenoids, flavonoids, indoles, isothiocyanates, etc.), zoochemicals (conjugated linoleic and n-3 fatty acids, etc.), fungochemicals and bacteriochemicals (formed during food fermentations and by the gut microflora). The current review focuses on organic bioactive food constituents, especially on vitamins, phytonutrients and bioactive peptides.

Vitamins are widely assumed to be well characterized concerning efficacy and daily requirements. However, there is increasing evidence that their role in the body exceeds classical vitamin functions. Current research aims at defining optimal levels beyond the prevention of vitamin deficiencies. This necessitates an understanding of the mechanisms of action, which depend on the compound itself (dose and bioavailability, interactions with the food matrix and drugs, etc.), endogenous factors (nutritional status, genetic predisposition, age, gender, etc.) and environmental parameters affecting the overall predisposition. A good example is vitamin E, which was discovered in 1922 by Evans and Bishop. Vitamin E is a major lipid-soluble antioxidant in vitro and as such might protect body lipids, especially membrane lipids, phospholipids, and lipoproteins from oxidation in vivo (Naguib et al., 2003). The latter is currently strongly debated. Apart from the low doses of vitamin E compared to endogenous antioxidants, alpha-tocopherol in particular, is efficiently recycled (reduced after oxidation) in vivo indicating that it might exert functionalities exceeding the antioxidant role. Indeed, more recent studies show that vitamin E is a potent enzyme inhibitor (PKC, 5-lipoxygenase, etc.) and activator (protein phosphatase 2A and diacylglycerol kinase). Furthermore, it plays a role in transcriptional gene regulation (e.g., scavenger receptors, alpha-TTP, alpha-tropomyosin), cell proliferation, platelet aggregation, and monocyte adhesion (Azzi, 2004; Nogala-Kalucka et al., 2004; Pfluger et al., 2004; Traber, 2004). This broad functionality drives research efforts to further investigate the role of vitamin E in the prevention of cardiovascular diseases, cancer, diabetes and Alzheimer's disease (Lonn et al., 2005; Morris et al., 2005; Hino et al., 2006).

Phytonutrients cover a broad range of chemically and physiologically diverse, low-molecular weight, secondary plant constituents (Hasler & Blumberg, 1999; King & Young, 1999). Recognizing their importance in the human diet, many phytochemicals are also designated as phytonutrients. The currently most studied groups are summarized in Table 1 (Rowland, 1999). Previous research has focused on the antioxidant properties of phytochemicals as being their principal mechanism of action. However, tissue concentration of phytonutrients compared to that of endogenous antioxidants suggest that many of the phytochemicals exert their activity indirectly, as “low-dose toxins,” by activating adaptive cellular response pathways to oxidative stress and environmental toxin exposure rather than acting as direct radical scavengers. In consequence, their physiological dose range is relatively narrow and bioavailability becomes a major consideration. Sensitive, selective and comprehensive analytical methods are key to phytonutrient research, not only because absorption is very compound specific but also because they undergo extensive metabolism in vivo and the metabolites are often the biologically active components at tissue site. Eventually, human studies are required to evaluate their safety and efficacy as well as their absorption, distribution, metabolism, excretion and mechanisms of action. Human trials commonly generate large number of samples, and thus medium and high throughput methods need to be developed.

Table 1. Selected vitamins and non-nutrient phytochemicals, their sources, biological activities and most commonly applied MS techniquesThumbnail image of

The third group of minor dietary constituents encompasses bioactive peptides, mainly derived from milk and released by digestive enzymes in the gastrointestinal tract. These peptides may directly influence behavioral, gastrointestinal, hormonal, immunological, and neurological function. Current applications include casein-derived phosphopeptides, used as both dietary and pharmaceutical supplements. Based on their antimicrobial properties, bioactive peptides could also be applied to improve microbial safety of food products.

When assessing the relationship between diet and health, a distinction must be made between causality and correlation, between mechanism of action and mode of action. A causal relationship between diet and health can be defined as one in which the chain of events, which link the ingestion of a given food with a defined biological endpoint is established and the mechanism of action is known. In a correlative relationship a health/disease endpoint appears to be related in some way to a given food but the nature and extent of that relationship is not established. Most of our current knowledge concerning diet and health only allow correlations.

Assigning dietary health benefit or disease risk to a particular dietary pattern is difficult for a number of reasons. These include the complexity of diet, the co-occurrence of many nutrients in foods and also the possible interactions between diet and the genetic background of the individual as well as environmental factors. An additional problem is the difficulty of accurately measuring dietary intakes in epidemiological studies. Traditionally this has been approached by food frequency questionnaires, food diaries, and 24 hr recalls. The information obtained is then combined with food composition tables and databases to calculate intake of specific nutrients. Food composition databases build largely on MS data and are now being established for minor dietary compounds, vitamins and phytochemicals, for example, by the National Nutrient Databank System (NDBS) (http://www.ars.usda.gov/ba/bhnrc/ndl). However, the level of minor plant constituents is affected by seasonal and agronomic factors, the plant variety, age and part of the plant, food preparation conditions, etc. The resulting variability can lead to serious problems of interpretating epidemiologic studies or human intervention trials if dietary intake data and food composition tables are taken to estimate the exposure to a given food component (Riboli, Slimani, & Kaaks, 1996). A more accurate measure of exposure to a dietary bioactive compound can be obtained from biomarkers of exposure if these have been sufficiently validated.

Biomarkers are observable properties of an organism that indicate variation in cellular or biochemical components, structure, or function, which can be measured in biological systems or samples (Bearer, 1998). In simple terms: a biomarker is a measurable change related to a phenotype. Biomarkers can be used to estimate prior exposure, to identify changes within an organism, and to assess underlying susceptibility of an organism. A valid nutritional biomarker could also be considered a key measure linking a specific exposure of a dietary compound to a health outcome and thus offers great potential to understand the relationship between diet and health (Fig. 1).

Biomarkers help to understand the nutrient absorption, transport and metabolism within an organism to produce an effective dose at target tissue and to understand the interactions at the cellular and molecular levels leading to defined endpoints. Biomarkers of susceptibility consider host, environmental and lifestyle factors, but especially genetic pre-disposition.

The above-mentioned processes can be envisioned as a continuum that links, exposure, dose, and effect. As shown in Figure 1, biological steps along the pathway (exposure, internal dose, biologically effective dose, early biological effect, altered structure and/or function, and clinical disease) can potentially be observed and quantified using biomarkers. Markers of internal dose are direct measures of a dietary compound or its metabolites at systemic level, for example, in body fluids. More specific, but often difficult to obtain are biomarkers that describe the dose of a dietary compound or metabolite at target tissues. Markers of biologically effective dose assess the interaction of a food constituent with their molecular targets, and markers of early biological effect assess the molecular sequence of target compound-cell interactions (Strickland, 2002). Biomarkers of altered structure and/or function are useful for assessing morphological and/or functional changes following compound–cell interactions, which are strongly indicative for disease outcomes while markers of susceptibility may influence the magnitude of each step in the pathway.

Biomarkers useful for disease prevention and nutritional/therapeutic intervention may appear anywhere along the pathway. Earlier markers have the greatest potential to avert disease; later markers are most closely related to the disease. Biomarkers should be responsive, specific and applicable. The problems of applying biomarkers to nutrition research are less related to accuracy and sensitivity of the analytical techniques than to a lack of validation in terms of (i) understanding the parameters related to the bioavailability and the mode of action of bioactive food components; (ii) an application in large population studies. Furthermore, most micronutrients at normal dietary dose levels are only weakly biologically active in the short term and have multiple targets. Depending on their dose, they can be both beneficial and deleterious. This poses particular problems in determining the net effect of food and their constituents. Biomarkers should be able to contribute to this debate by allowing the measurement of exposure and by acting as an indicator either of a deleterious or disease preventive and health promoting effect prior to the final outcome. However, in most cases, a battery of biomarkers needs be employed to comprehensively describe the entire continuum, from exposure via effect to end point.

Key points to consider when assessing the use of biomarkers in studies of nutrition and health include:

  • The timing of a measurement in relation to the response (bioavailability and efficacy)

  • Disease may affect the measured level of an indirectly related biomarker

  • Biomarkers may correlate with intake but often, for example, metabolites or DNA adducts, the marker is an integration of intake, absorption, metabolism, and excretion. Frequently it is these markers that can be very informative with respect to risk of disease.

  • In the case of homeostatic regulation of nutrient levels in body fluids, biomarkers may poorly correlate with intakes.

  • Environmental factors and the genetic background of the individual may modulate the correlation between dietary intake and biomarkers.

In summary, biomarkers are indicators of molecular and cellular events in biological systems, and may allow epidemiologists and clinicians to better understand relationships between food constituents, whole food or diets and human health effects. However, based on the complexity of the issue it becomes clear that the biomarker approach will not provide us with all information needed to unravel the relationship between diet and health. A combination of methods, for example, with questionnaires, non-invasive and invasive measurements will probably prove to be most valuable and practicable.

C. Biomarkers of Exposure

1. Understanding Biomarkers of Exposure

In order for a bioactive compound to exert any activity, it must reach the target organ at a minimal concentration that determines both biological effect and mechanism of action. As discussed above, dietary intake of a bioactive compound does not necessarily reflect the dose reaching the target tissue and many intervention trials lacking validated measures of bioavailability have failed to demonstrate the anticipated results. Therefore, biomarkers of exposure are being developed and applied to determine the internal, target tissue and biologically effective dose.

For biomarker development and validation, an understanding of the factors that constrain the release of a bioactive food component from the food matrix, the extent of absorption, and their fate in the organism, is crucial. This is particularly critical for minor dietary constituents since, in comparison with that of macronutrients, the beneficial intake range of micronutrients is relatively narrow and both adverse effects at higher doses and no effect at very low doses are observed.

The basis for validation of a biomarker of exposure is an understanding of major events describing the fate of the nutrient including its absorption, metabolism, distribution and excretion and an evaluation of the major parameters affecting the latter. A general scheme of these events is depicted in Figure 3. Parameters affecting excretion may be divided into exogenous factors (such as complexity of food matrix; chemical form of the compound; structures and amounts of co-ingested compounds) and endogenous factors (including mucosal mass, intestinal transit time, rate of gastric emptying, metabolism and extent of conjugation). Based on the complexity of these factors large inter-individual and intra-individual variations in bioavailability, sometimes ranging from 0 % to 100 % of the ingested dose are observed. For more details see (Holst & Williamson, 2004b).

  • The food matrix is much more complex than drug formulations;

  • The prevalence of synergistic effects;

  • Competition with other food constituents for common metabolic enzymes and transporters;

  • The much lower concentrations of compounds in food compared to those of active components in drugs; and

  • The longer term (chronic) application of low doses of compounds in food compared with higher (acute) doses of drugs.

Figure 3.

Basic events describing the fate of nutrients: (1) Liberation, the release and dissolution of a compound to become available for absorption (bioaccessibility); (2) Absorption, the movement of a compound from the site of administration to the blood circulation; (3) Distribution, the process by which a compound diffuses or is transferred from the intravascular space (blood) to the extra-vascular space (body tissues); (4) Metabolism, the biochemical conversion or transformation of a compound into a form that is easier to eliminate; and (5) Excretion, the elimination of unchanged compound or metabolites from the body, mainly via renal, biliary, or pulmonary processes. The complex connection between the luminal content, lumen and body content results in a web of possible pathways for each compound.

Given these challenges, it is more difficult to identify appropriate biomarkers of exposure in nutritional than in pharmaceutical sciences. Furthermore, when validating a potential biomarker, all the above issues need to be considered. Consequently, only few validated biomarkers are available and accepted that relate to dietary micronutrient exposure. At the same time, there is substantial research ongoing to understand the liberation, absorption, distribution, metabolism and excretion of minor dietary constituents and, on that basis, to develop and validate biomarkers of exposure for these compounds. The following paragraph highlights the latter by detailing examples of biomarkers which are established, under development or needed.

2. MS for Discovery and Measurement

Biomarkers can be measured in a variety of specimens including urine, feces, expired air, saliva, hair, breast milk, nails (non-invasive sampling) or serum, plasma, red blood cells, white blood cells, bone marrow, and different tissues (invasive sampling). Although these matrices represent diverse analytical problems (e.g., high anticoagulant, protein and salt content of blood; high creatinine and organic acid content of urine, difficult homogenization of tissues) the key factors in determining the choice of sample analysis are the chemical properties of the target analytes. The two major groups of vitamins illustrate this well (Table 1).

Water-soluble vitamins are relatively polar compounds: for instance, vitamin B9 (folate, folic acid) is a molecule with both protonation and deprotonation sites. Accordingly, mass spectrometric analysis of vitamin B9 in erythrocytes (Clifford et al., 2005; Smith et al., 2006), whole blood (Owens, Holstege, & Clifford, 2005), serum, plasma (Kok et al., 2004; Nelson et al., 2005), meat, cereals, bread, vegetables (Freisleben, Schieberle, & Rychlik, 2003a,b) etc. is typically performed using ESI both in positive and negative mode. More specifically, RP-HPLC-ESI-MS is the method of choice to investigate absorption and metabolism of folates. To study the impact of food texture on folate absorption, differently labeled (13C and deuterium) folic acid-containing matrices (capsule and drink) were orally administered and the stable isotope labeled analytes were analyzed in plasma (Kok et al., 2004). Commercially available stable-isotope labeled homologues were used as internal standards for quantitation. It was concluded that folates are absorbed similarly from the capsules and from the drink.

Absolute bioavailability is commonly studied in pharmaceutical research and involves a comparison of differently labeled compound or metabolite levels in plasma or urine after oral versus intravenous administration. Absolute bioavailability of folate was studied using differently labeled folates by determination of the urinary excretion ratios of the isotopomers (Finglas et al., 2002). The difference between these two indicates the efficiency of absorption in the lumen, the extent of metabolism in the lumen and liver and the time-shift from ingestion to appearance in the plasma.

Fat-soluble vitamins such as vitamin E (tocopherols and tocotrienols with a weakly acidic chroman ring and a phytil side chain) are analyzed by both APCI-MS as well as ESI-MS (Table 1). It is often the chromatographic separation that determines the ionization. Vitamin E is a mixture of different vitamers including beta- and gamma-isomer pairs, which show different absorption, metabolism, and biological activity and thus their distinct quantification is required. In general, tandem mass spectrometry often can identify isomers if they are injected in a pure form but distinct quantification of isomers in mixtures is usually not feasible: preliminary chromatographic separation is necessary. While most separation tasks can be addressed by ESI-MS compatible RP-HPLC, separation of certain, mainly lipophilic nutrients requires normal-phase liquid chromatography (Schuep & Rettenmaier, 1994; Ball, 1998), the latter being compatible only with APCI.

Absorption, distribution and metabolism of the antioxidant vitamin E have been studied by both LC-APCI-MS and LC-ESI-MS. Leonard et al. (2004) investigated the effect of food matrix on the bioavailability of vitamin E. Encapsulated d9-tocopheryl-acetate was orally ingested with and without wheat cereal and blood samples were drawn after different time intervals. Labeled free alpha-tocopherol was quantified in the plasma samples by comparing the analyte peak areas to the peak area of the D6-tocopherol internal standard. The authors concluded that the presence of cereal matrix, including fats, increases alpha-tocopherol levels in plasma and that vitamin E can be absorbed more easily from fortified food products than from the encapsulated form in the absence of food.

Metabolism of vitamin E depending on the oxidative stress level has been investigated monitoring the major tocopherol metabolites, carboxyethylhydroxychromans (CEHCs) as biomarkers for exposure. Deuterated alpha-tocopheryl-acetate was ingested by smokers and non-smokers and blood samples were collected. The data suggested that alpha tocopherol metabolism in smokers who are subject to increased oxidative stress show significantly lower levels of CEHCs than non-smokers (Bruno et al., 2005). In another study racemic and stereochemically pure (RRR) tocopheryl-acetates labeled to different degrees were orally administered. By measuring the CEHC metabolites in plasma it was shown that racemic forms of tocopherols underlie increased catabolism and excretion when compared to the natural occurring RRR-stereoisomer (Traber, Elsner, & Brigelius-Flohe, 1998). The administration of differently labeled alpha and gamma vitamers in a similar experimental design revealed that while alpha-tocopherol is maintained in the plasma, gamma-tocopherol is rapidly metabolized to its CEHC metabolite and excreted (Leonard et al., 2005).

Phytochemicals represent a wide variety of compounds with different chemical characteristics: flavonoids and glucosinolates are rather polar while phytosterols are non-polar compounds. Major groups and representatives thereof are summarized in Table 1. The current understanding of absorption, metabolism, distribution and excretion of phytochemicals, issues developing biomarkers of exposure, gaps in knowledge and the potential of MS techniques in this area are demonstrated below for glucosinolates and flavonoids.

Glucosinolates (GLSs) are (Z)-cis-N-hydroximinosulfate esters, possessing a sulfur-linked β-D-glucopyranose moiety and a variable side chain. Glucosinolates are widespread among Cruciferous vegetables. According to their structure, glucosinolates are strongly acidic compounds, thus allowing anion-exchange solid-phase extraction for sample preparation. Their separation is often performed by ion-pair liquid chromatography followed by ESI-MS in negative mode (Mellon et al., 2002; Bennett, Mellon, & Kroon, 2004). Mellon et al. (2002) presented a comprehensive method based on negative ion ESI-MS for the analysis of glucosinolates in crude plant extracts. They compared different scanning functions (including neutral loss and precursor ion scans) for the selective detection and identification of intact glucosinolates and concluded that programmed cone voltage fragmentation (in-source CID) delivered superior results due to the better sensitivity and preserved isotope pattern of the characteristic sulfur isotope peaks (4% at 2 AMU mass shift). Alternatively, glucosinolates can be desulfonated and analyzed by RP-HPLC-APCI-MS in positive mode.

Glucosinolates are chemically and physiologically relative inert, however following cell disruption, enzymatic hydrolysis occurs by plant myrosinases, which results in the formation of a diverse group of chemically reactive and biologically active compounds. There is considerable evidence that these breakdown products -when consumed in the diet - may affect the risk of developing chronic diseases, especially cancer. One of the most promising dietary cancer-preventive compounds is the glucosinolate hydrolysis product sulforaphane, which is an isothiocyanate, mainly found in broccoli. The physiological dose window for glucosinolate hydrolysis products is narrow, and hence bioavailability is a key factor determining beneficial versus detrimental effects. While a range of compounds is derived from sulforaphane, the precursor glucosinolate may also form additional hydrolysis products than sulforaphane, for example, nitrile, and this further broadens the range of possible metabolites. While the absorption of intact glucosinolates in human has not been confirmed, most glucosinolate hydrolysis products, for example, sulforaphane readily diffuse across the membrane of the small intestinal barrier. Mainly within the small intestine and liver, GLS-HPs such as sulforaphane are subject to extensive metabolism, where the major route includes glutathione (GSH) conjugation catalyzed by glutathione-S-transferases. Using a small intestinal perfusion in humans, Petri et al. (2003) have shown that the conjugates formed will be transported into the systemic circulation, but also that a substantial amount of the conjugates are effluxed back into the lumen. They analyzed perfusion solution and the perfusate leaving the small intestine by LC-MS, but no blood or urine samples. A comprehensive analysis of broccoli glucosinolate derived compounds has not been published up to now, however such method would allow the monitoring of both, conjugation and hydrolysis derived products. As most cruciferous vegetables contain a complex profile of glucosinolates, deriving to even more complex product and metabolite profiles, the limited availability of standards poses a significant challenge to achieve quantitative analysis.

Before excretion, GSH conjugates are converted into the corresponding mercapturic acid derivatives. Although accounting for only a small proportion of glucosinolate hydrolysis products ingested, these mercapturates have been used as biomarkers of exposure, and some correlation of mercapturate excretion of individual isothiocyanates with physiological effects was observed (Kensler et al., 2005). The authors employed a cyclocondensation method followed by HPLC with UV detection to analyze isothiocyanates and metabolites in the plant samples and urine stressing again the limited application of MS in this area of research despite the advantages that MS, especially coupled to HPLC could provide. Future research should address these issues to allow identification of validated biomarkers of exposure by mass spectrometry as an inevitable tool.

Non-absorbed glucosinolates and hydrolysis products and their effluxed metabolites reach the colon where they can be metabolized by the gut microflora. Due to analytical limitations, these microbial products have not been well characterized but have the potential to be absorbed and exert systemic effects or effects at colonic level. LC-MS/MS could play a key role in identifying and quantifying these products. The bioavailability of glucosinolates and derived compounds, including parameters that affect absorption, distribution metabolism and excretion was recently reviewed (Holst & Williamson, 2004a).

Flavonoids are water-soluble polyphenolic plant constituents that are often related to dietary prevention of several diseases. They exist predominantly in their glucosylated form in plants and most food products. Accordingly, their analysis in food matrix and at the luminal site covers detection of glucosides and aglycones. Following absorption analysis of flavonoid derived compounds includes the analysis of aglycones and glucuronide-, sulfate- and methyl-conjugates. Fate and transfer of flavonoid glucosides, aglycones, conjugates and breakdown products in the human body is illustrated in the case of quercetin in Figure 4.

Figure 4.

Overview of transport, absorption and metabolism of flavonoids illustrated by Quercetin. Predominant source of Quercetin is its glucoside form, which first must be cleaved by enzymes to allow its absorption. In addition, several conjugation/deconjugation and microbial degradation processes occur to the aglycone in different segments of the lumen, and the efflux from the epithelial layer back into the luminal side further complicates the distribution of quercetin derivatives. These wide range of metabolites/breakdown products again can be absorbed and can appear in the systemic circulation. ST, sulfotransferase; UGT, UDP-glucuronosyltransferase; COMT, catecholamine-O-methyltransferase.

Flavonoids and their metabolites are commonly extracted from biological fluids using SPE and LLE procedures and analyzed by HPLC-ESI-MS in both positive- and negative-ion mode. Absorption, distribution, metabolism and excretion of numerous flavonoids have been studied by mass spectrometry. Day et al. (2001) employed RP-HPLC-ESI-MS in positive mode to show that, after consumption of quercetin glucoside-rich onions, the flavonoid-glucoside cannot be detected in the plasma in contrast to major metabolites such as glucuronidated, sulfated and methylated quercetin.

Liu et al. (2005) used LLE and RP-HPLC-ESI-MS in negative mode to study absorption of flavonoids after oral administration of a plant extract. They identified 18 flavonoids in the plasma samples (including isoflavons, isoflavonons, neoflavons, flavanons, chalcones etc.). Moreover, they confirmed that flavonoids are present in plasma in their free (aglycone) form. RP-HPLC-ESI-MS in positive mode was successfully used to detect appearance of glucuronide- and methylglucuronide-metabolites in rat brains after oral administration of epicatechin, hereby confirming that epicatechin was able to cross the blood/brain barrier and reach the brain (Abd El Mohsen et al., 2002).

Felgines et al. (2003) studied the absorption, metabolism and excretion of anthocyanins by analyzing the urine samples collected from volunteers that consumed 200 g strawberries. Based on MS they identified five anthocyanin metabolites. The total urinary anthocyanine metabolite excretion was less than 2% and two thirds of the metabolites were excreted within 4 hr. Again, only glucuronides could be detected due to the sample preparation and analytical procedure, that is, glucuronidase treatment followed by ESI-MS in positive mode. The decreasing levels of anthocyanins before reaching the colon were also studied using ileostomy patients (Kahle et al., 2006): after oral intake of polyphenol-rich apple juice and blueberries, ileostomy effluent was collected and analyzed by HPLC-ESI-MS in positive-ion mode and HPLC-DAD. The authors concluded that polyphenols originating from apple juice were quickly absorbed and metabolized. Only 0–33% of the oral dose reached the ileostomy bags after 2 hr while up to 85% of the orally applied anthocyanins were recovered in the ileostomy fluid. This confirms that absorption/metabolism of anthocyanins from blueberries is less efficient than absorption/metabolism of apple polyphenols.

There are two commonly applied approaches to study metabolism of polyphenols and vitamins. Either the glucuronide/sulfate conjugates can be cleaved and the sum of the cleavage products is determined by HPLC-ESI-MS or the major glucuronides and aglycones are determined in their intact form using HPLC-ESI-MS in positive mode. The first approach is very sensitive, but it pools the information carried by the different conjugate-types and positional isomers. The main disadvantage of the second approach is the limited potential of positive ESI to ionize the corresponding sulfated metabolites. This also explains why only the glucuronide metabolites are discussed in most applications. Biological activities of sulfate as well as glucuronidated and sulfated conjugates can differ from the ones of glucuronides and there is a strong need to develop more sensitive and comprehensive methods able to quantify the different metabolites including glucuronides, sulfates and mixed modifications separately in their intact form. Recently Roura et al. (2005) have demonstrated that this latter approach is feasible and quantified aglycones, glucuronides, sulfates simultaneously using negative ESI-MS.

IR-MS was originally developed and used to follow/discover the fractionation processes in order to unveil biogeochemical changes or tracing the geographical origin of biomolecules. In spite of this, IR-MS is also used to follow the fate of nutrients in humans. Parker et al. (1999) developed a method based on extrinsically stable-isotope-labeled vitamin A (2H4-labeled retinyl acetate) and measured the post-prandial chylomicron carotenoid or retinyl ester response following a single dose of carotenoids. The compounds of interest were analyzed with HPLC and IRMS following a protocol originally developed to study β-[13C]-carotene metabolism (Parker, 1997; Parker et al., 1997). Richelle et al. (2004) showed that in addition to reducing cholesterol absorption, plant sterols reduce β-carotene and α-tocopherol bioavailability in normo-cholesterolemic men. Plant sterol esters were found to be more efficient than plant free sterols.

D. Biomarkers for Efficacy: Nutritional Intervention

Most of the currently monitored nutritional efficacy biomarkers represent single or few-molecule read-outs achieved through specific assays. Gene arrays are still the most widespread tool to globally investigate outcomes of nutritional interventions. Metabolite profiling still largely relies on NMR rather than on MS platforms. Therefore, we will review molecular efficacy measurements of dietary interventions from a more general perspective and give an outlook on how mass spectrometry should contribute in this context in the future.

The molecular mechanisms by which diet alters health and the related efficacy are to date predominantly revealed through the identification of diet-regulated genes. DNA and oligoarray formats have expanded the scale of such approaches to multiple genes within a pathway (Dhahbi et al., 1999) and further to the genomic scale (Sunde, 2001; Velazquez, 2001). In addition, the expression of candidate genes in different animal tissues in response to dietary variation (Cousins, 1999) and caloric restriction (Lee et al., 2002) has been investigated.

Apart from expressed genes, biomarkers supporting the efficacy of nutritional interventions have been mass spectrometrically searched at protein and more recently, at metabolite level. A knowledge building study to comprehend liver metabolism at protein level has been published in the form of a proteomic survey of the rat liver (Fountoulakis & Suter, 2002): two hundred seventy three gene products have been identified and roughly two thirds of them represent enzymes covering a broad range of catalytic activities. In the same view of generating molecular understanding of metabolism, Ernest et al. (2003) have taken a metabolic model system down a nutrigenetics route in order to study the effects of genetic perturbations on metabolism at systems biology level: they monitored the response of the homocysteine-folate metabolism to single gene mutations in mice by measuring serum metabolite levels and gene expression. A result of immediate relevance for future nutritional treatment was the possibility to assess dietary supplementations for suppression of adverse developmental outcomes of single gene mutations in the homocysteine-folate pathway (Carter et al., 1999). However, in more applied cases of (pre-) clinical dietary interventions, biomarkers of nutritional efficacy have been analyzed in a targeted fashion meaning that specific assays for proteins, peptides, nutrients and metabolites have been developed and employed (Branca et al., 2001). Typically, HPLC, ELISA, and RIA-based methods are utilized in such settings and for such purposes (Crews et al., 2001).

The three diseases cancer, coronary heart disease (CHD) and osteoporosis (OP) have been reviewed in terms of their amenability to beneficial dietary intervention (Branca et al., 2001). Strikingly, most of the cited molecular read-outs are punctual, that is, one or very few compounds have been monitored in response to a nutritional change. In all three cases, it has become apparent that the disease, once manifest, is difficult to be positively influenced solely by diet. Diseases caused by factors other than essential nutrient deficiency are unlikely to be alleviated by nutritional means alone (Crews et al., 2001). Markers to measure immunomodulation in human nutrition intervention studies have been reviewed by Albers et al. (2005). As it applies to the above discussed diseases cancer, CHD and OP, these markers have not (yet) been derived from Omic approaches but rather reflect targeted measurements of biomolecules or read-outs from cellular assays, typically performed in (pre-) clinical settings.

Two studies employing proteomics, either as stand-alone or integrated with gene expression analysis, and addressing biomarkers for nutritional health effects in the context of intestinal cancer deserve special attention: Breikers et al. (2006) found 30 proteins differentially expressed in the colonic mucosa of healthy mice upon increased vegetable intake. Six of these 30 proteins were identified by MALDI-ToF-MS and the alteration of the respective expression levels agree with their elsewhere-reported protective role against colorectal cancer. Figure 5 shows the 2DE display of proteins isolated from mouse colonic mucosal cells with the identified protein spots marked. The second project integrated DNA microarrays with proteomics to investigate the effects of nutrients with presumed anti-colorectal cancer properties and to develop a colon-epithelial cell line-based screening assay for such nutrients (Stierum et al., 2001). In an in vitro experiment, Tan et al. (2002) looked at sodium butyrate effects on growth inhibition of HT-29 cancer cells following a 2DE-MS-based proteomic strategy. Butyrate treatment altered the expression of various proteins in the ubiquitin-proteasome pathway; a result suggesting that proteolysis could be a “tool” for butyrate to regulate key proteins in the control of cell cycle, apoptosis and differentiation.

Figure 5.

Silver-stained 2DE display of proteins isolated from mouse colonic mucosal cells with the identified protein spots marked. Breikers et al. found 30 proteins differentially expressed upon increased vegetable intake. Six of these 30 proteins were identified by MALDI-ToF-MS and the alteration of their expression levels agree with their protective role against colorectal cancer. MLRN: Myosin regulatory light chain 2, smooth muscle isoforms; CAH-1: Carbonic anhydrase I; HMG-1: High-mobility group protein 1; PAP3: Pancreatitis-associated protein 3 (precursor); GAPDH: Glyceraldehyde-3-phosphate dehydrogenase; OSCP: ATP synthase oligomycin sensitivity conferral protein, mitochondrial (precursor). (Reproduced from Breikers et al. (2006) with permission from Wiley-CH-Verlag, copyright 2006, and from the authors).

Mass spectrometry- and NMR-based metabolic profiling is increasingly being employed to monitor dietary interventions. With the objective of establishing a quantitative molecular catalogue, a large database of lipid metabolites and their concentrations in humans is currently being constructed. Building on this knowledge, quantitative lipidomic data showed differential effects of dietary fats on cardiac and liver phospholipid turnover (Watkins et al., 2001a,b). Global metabolite profiling serves also to elucidate the relationships between amino acid levels in plasma. Correlation-based analysis of metabolomic data has been proposed to determine, which metabolites cause the biological effects of excessive amino acid intake (Noguchi, Sakai, & Kimura, 2003). Flavonoid consumption via tea drinking has been attributed to a number of potential health benefits like cancer prevention, anti-inflammatory action and cardioprotectant activity. Metabolomic methods utilizing 1H-NMR and PCA were applied to investigate the bioavailability and metabolic effects of a single dose of epicatechin administered in rats (Solanky et al., 2003b). Metabolic effects and excretion of epicatechin metabolites demonstrated both epicatechin bioavailability and bioactivity. The multiple cellular actions of isoflavone antioxidants support their protective effect in a number of experimental and human diseases such as renal and cardiovascular disease. The administration of soy isoflavones in healthy post-menopausal women under controlled environmental conditions was followed by 1H-NMR spectroscopy and chemometric analysis (Solanky et al., 2003a). The soy intervention changed plasma lipoprotein, amino acid, and carbohydrate profiles suggesting a soy-induced change in energy metabolism.

More recently, Omic approaches recruiting mass spectrometry for protein and metabolite analysis have been undertaken in an increasingly integrative manner to shed more light on mechanisms triggered by (compromised) nutrient bioavailability or (beneficial) nutritional intervention. An example for investigated consequences of nutrient deficiencies is a study, in which rats were force-fed a zinc-deficient diet and then subjected to hepatic transcriptome, proteome, and lipidome analysis. The observed changes in liver metabolites were discussed in the context of gene and protein expression regulation (tom-Dieck et al., 2005). Along the same lines, Fong et al. (2005) demonstrated that alleviation of zinc deficiency by zinc supplementation led to an 80% reduction of COX-2 mRNA, a key enzyme involved in inflammation. Moreover, two supposedly beneficial diets enriched in different long-chain poly-unsaturated fatty acids (LC-PUFAs) were tested in a rat nutritional intervention model. This transcriptomic/lipidomic study based on gene chips and gas chromatography without subsequent MS revealed steaoryl-CoA desaturase as a target enzyme for an arachidonate-enriched diet (Mutch et al., 2005). Another suggested beneficial effect of LC-PUFAs was assessed by a nutrigenomics experiment designed to understand how these lipids induce and control gene signaling involved in carcinogenesis (Anderle et al., 2004). In a combined gene and protein expression study by Herzog et al. (2004) the flavonoid flavone, present in a variety of fruits and vegetables, was identified as a potent apoptosis inducer in human colonic cancer cells. Flavone revealed a surprisingly broad spectrum of effects on gene and protein expression related to apoptosis induction and cellular metabolism. Figure 6 shows 2DE results from HT-29 cells grown in the absence or presence of flavone. Zooms into particular regions of gels from control and flavone-treated samples, respectively, reveal the protein abundance differences between the two cell growth conditions.

Figure 6.

2D gel results from HT-29 cells grown in the absence or presence of the flavonoid flavone. Flavone, present in a variety of fruits and vegetables, was identified as a potent apoptosis inducer in human colonic cancer cells and revealed effects on apoptosis induction and cellular metabolism. Zooms into particular regions of gels from control and flavone-treated samples, respectively, reveal the protein abundance differences between the two cell growth conditions. (Reproduced from Herzog et al. (2004) with permission from Wiley-CH-Verlag, copyright 2004, and from the authors).

All intervention studies cited above have investigated abundance changes at gene, protein and/or nutrient/metabolite level in response to a nutritional intervention. However, several food components are not only recognized to alter the concentrations of such molecules but to also post-translationally modify proteins (Davis & Milner, 2004). These dietary induced protein modifications can ideally be assessed by mass spectrometric means. For instance, the protein phosphorylation status of the ERK protein changes after exposure to diallyl disulfide, a compound present in processed garlic, an effect resulting in cell cycle arrest (Knowles & Milner, 2003). Another such example is the modification of thiol groups in the cytoplasmic protein Keap1 (Dinkova-Kostova et al., 2002). This alteration of the protein redox status affects its binding to the protein Nrf2.

IV. CONCLUSIONS

Mass spectrometry has a traditional use in nutrition research to assess food authenticity, safety, flavor, and to characterize the food matter for its nutrient composition. More recently, nutrition research is focusing on unraveling the relationship between diet and physiological outcomes and how to exploit this knowledge to improve consumer health. Epidemiological evidence consistently suggests that increased fruit, vegetable, and milk consumption affords significant protection against many chronic diseases. These dietary food groups contain numerous compounds that possess health-promoting activities in addition to their complement of essential nutrients. Intensive efforts are now focusing on identification and quantification of these bioactive compounds in plant, for example, phytochemicals, and animal derived foods, for example, bioactive milk proteins and peptides. Many of these bioactive compounds represent only minor constituents in a highly complex matrix. Due to the specificity and sensitivity of mass spectrometry, especially in combination with high-performance chromatography, it has developed into the key analytical platform for identification and quantification of a complex array of health promoting components in food, including phytonutrients, proteins and peptides.

Although many foods are hypothesized to have functional properties, their impact on health and disease risk has rarely been thoroughly characterized. Recent awareness linking nutrition, altered metabolism, health and prevention of disease has dramatically increased the interest in functional foods and nutraceuticals. Functional foods are defined as products providing health benefits beyond their contribution to nutrient requirements. An emerging field concerns personalized nutrition, which is taking the functional food and neutraceutical development one step further by considering the individual predisposition and thus susceptibility.

In comparison to drug studies, the concentration of active compounds in food products is much lower. Therefore, chronic doses are likely to be the basis of any effect and the general agreement that the combined effects of many such compounds in foods, rather than the one of any single compound, provide the optimal health promoting benefits are further reasons for the lack of knowledge concerning the efficacy of dietary bioactive compounds. A logical extension of such conventional wisdom is the development and study of functional foods as compared to individual bioactive compounds. However, when studying the relationship between exposure to bioactive compounds, but even more so, exposure to whole foods on one side and health outcomes derived from complex biochemical pathways on the other side, it is unlikely that these can be described by individual biomarkers. A combination and integration of appropriate biomarkers to describe susceptibility, exposure and effect will be required to understand the link between diet and health. Mass spectrometry has a large potential in studying bioavailability of key compounds and even combinations of bioactive compounds in food and biological samples as well as in studying health outcomes. Nevertheless, it will be an intelligent combination and integration of different, complementary analytical platforms that will drive our understanding, especially concerning complex interactions and activities of combinations of bioactive compounds, such as present in functional foods. In consequence, further mass spectrometric developments need to enable this integration in terms of software design, inlet systems, etc. Bioinformatics will have to play a major role in data processing and data integration.

A widespread need for mass spectrometry applications in nutrition research supports its further distribution in food research laboratories. However, major barriers for a broad application are the high costs of instruments. The availability of instruments with different performance capabilities, from table-top to high-end devices, an increased user friendliness, and high throughput potential should be major development areas in mass spectrometry instrumentations for food and nutrition research.

To achieve even higher selectivity and thus to minimize interferences, new mass analyzers with increased resolution and reduced prices have been developed and introduced. A typical example for this trend was the introduction of Q-ToF instruments 10–15 years ago, introduction of high-resolution triple-quadrupoles 5 years ago and the present introduction of Orbitraps. Beside increased resolution, emphasis is also put on improved mass accuracy during development, for example, of Q-Tof (5–10 ppm mass errors), Orbitrap (below 5 ppm mass errors) and also FT-ICR machines (sub ppm mass errors). While sensitivity of the present mass spectrometers is outstanding, developments of ion guide and focusing devices are in progress for even better sensitivity and a simplification of sample preparation requirements. The ultimate sample preparation could consist of only sample dilution, if necessary, in order to avoid suppression effects. Most importantly, hybrid mass spectrometers with sophisticated scan functions are increasingly used as these devices combine the advantages of different mass spectrometer types into one instrument (e.g., Q-ToF, Q-Trap, Paul trap-Orbitrap, Ion Mobility-triple quadupole, etc.).

Mass spectrometry is one of the most versatile identification and quantification techniques nutritionists have today and, more importantly, it ranges among the very few analytical tools that can deliver information-rich data. Its potential as a nutritional knowledge provider is constantly growing because of both the push from instrument manufacturers and the pull of researchers aiming at transforming nutrition into a mechanism-based discipline.

V. ABBREVIATIONS

APCI

atmospheric pressure chemical ionization

BMI

body mass index

CHD

coronary heart disease

ChREBP

carbohydrate-responsive element-binding protein

COX

cyclo-oxygenase

CR

caloric restriction

CVD

cardiovascular disease

2DE

two-dimensional gel electrophoresis

DIGE

differential imaging gel electrophoresis

ELISA

enzyme-linked immuno sorbent assay

ENS

enteric nervous system

ESI

electrospray ionisation

FT-ICR

Fourrier-transform ion cyclotron resonance

GALT

galactose-1-phosphate uridyltransferase

GC

gas chromatography

GI

glycemic index

GIST

global internal standard technology

GIT

gastro-intestinal tract

(h)EC

(human) endothelial cells

(h)IEC

(human) intestinal epithelial cells

HMS

hypothesis-driven multi-stage mass spectrometry

ICAT

isotope-coded affinity tag

ICL

isotope-coded labeling

ICP

inductively coupled plasma

IFN

interferon

IL

interleukin

IMAC

immobilized-metal affinity chromatography

iNOS

inducible nitric oxide synthetase

IR-MS

isotope ratio mass spectrometer

IT

ion trap

LC

liquid chromatography

(LC)-PUFA

(long-chain) poly-unsaturated fatty acid

LDL

low-density lipoprotein

LOD

limit of detection

LOQ

limit of quantification

LOX

5-lipoxygenase

MALDI

matrix-assisted laser desorption/ionisation

MRM

multiple reaction monitoring

MS

mass spectrometry; multiple sclerosis

MS/MS, MS2

tandem mass spectrometry

MudPIT

multi-dimensional protein identification technology

NMR

nuclear magnetic resonance

OA

osteoarthritis

OP

osteoporosis

PAGE

polyacrylamide gel electrophoresis

PBMC

peripheral blood mononuclear cells

PC(A)

principal component (analysis)

pI

isolelectric point

PKU

phenylketonurea

PMF

peptide mass fingerprint

PPAR

peroxisome proliferator-activated receptor

ppm

parts per million

PTM

post-translational modification

Q

quadrupole

RP

reversed phase

RIA

radioactivity immuno assay

RXR

retinoic acid receptor

SCX

strong cation exchange

SFA

saturated fatty acid

SIM

single ion monitoring

SREBP

sterol regulatory element-binding protein

TGF

tissue growth factor

TLR

toll-like receptor

ToF

time-of-flight

UC

ulcerative colitis

UPLC

ultra performance liquid chromatography

Ancillary