Metabolomics as a Potential New Approach for Investigating Human

Article pubs.acs.org/jpr

Metabolomics as a Potential New Approach for Investigating Human Reproductive Disorders Frédérique Courant,* Jean-Philippe Antignac, Fabrice Monteau, and Bruno Le Bizec LUNAM Université, Oniris, Laboratoire d’Etude des Résidus et Contaminants dans les Aliments (LABERCA), USC INRA 1329, BP 50707, F-44307 Nantes Cedex 3, France ABSTRACT: Metabolomics has been emerging for several years as a global chemical phenotyping approach offering fascinating descriptive capabilities for addressing life complexity. It facilitates the understanding of the mechanisms of biological and biochemical processes in complex systems and promises new insights into specific research questions. The objective of this study was to use for the first time a metabolomic approach based on liquid chromatography high resolution mass spectrometry for characterizing an alteration of the testicular function, namely impaired semen quality. Metabolomic fingerprints were generated from serum samples collected from Danish young men presenting low, intermediate, or high sperm concentrations. Serum metabolic profiles were found to be significantly different among the three groups of volunteers. The developed methodology permitted to correlate the studied clinical parameter (i.e., sperm concentration) with the metabolite profiles generated. Peptides related to the Protein Complement C3f were identified as putative markers associated with this clinical parameter. The biological interpretation and further robustness linked to this observation remain to be further investigated, in particular to address the inter- and intraindividual variabilities. KEYWORDS: metabolomics, mass spectrometry, reproductive disorders resulting from transcriptional and translational changes.9,10 It enables the differential assessment of the levels of a broad range of endogenous and exogenous molecules and has been shown to have a great impact on the investigation of physiological status, diagnosis of diseases, discovery of biomarkers, and identification of perturbed pathways due to disease or treatment.11 It may provide advantages that classical diagnostic approaches do not have, based on the discovery of a suite of clinically relevant biomarkers whose levels are simultaneously affected by the disease. This approach may also bring new insights to disease etiology and, particularly, infertility.12 From a methodological point of view, metabolomics deals with the comprehensive analysis of metabolites present in a biological sample by the combined use of a fingerprint technology and multivariate statistical analyses. The most widely used technique for metabolomic purposes remains nuclear magnetic resonance (NMR), mainly for historical reasons.9,10 However, MS-based methods have recently proved to be valuable for such studies, especially thanks to recent technological advances, and furthermore present some unequaled advantages over NMR in terms of sensitivity.13−15 MS fingerprinting can be performed without any chromatographic separation step prior to analysis. If this approach may give correct results for particular applications16,17 (relatively high

1. INTRODUCTION Secular trends regarding human semen quality are a matter of concern which has been extensively discussed since the analysis published by Carlsen and co-workers.1 This review and metaanalysis of internationally published data related to semen quality among healthy men suggested the existence of a significant decline in sperm concentration and total sperm count over a period of 50 years. Retrospective analyses, performed by several researchers to study trends in their countries on the basis of their own laboratory records, have been undertaken and reported heterogeneous findings.2,3 Some confirmed a decreasing trend in semen quality, while others did not. Such diverging results from these different studies are probably partially due to their designs, which are predominantly retrospective and may be subject to bias from various factors, such as participant selection, interlaboratory differences, and duration of abstinence prior to delivery of semen samples.4 The reported decline in semen quality is also of great interest as a potential indicator of similar adverse trends for other male disorders as well.5−7 Semen quality has, in some countries, reached a level where a considerable fraction of young men are at risk of fertility problems.8 In recent years, the concept of biological fingerprint has emerged and comprehensive “omics” approaches (transcriptomics, proteomics, and metabolomics) have become a new way of addressing life complexity. Metabolomics corresponds to the study of small molecules as ultimate cellular signaling events © XXXX American Chemical Society

Received: March 6, 2013

A

dx.doi.org/10.1021/pr400204q | J. Proteome Res. XXXX, XXX, XXX−XXX

Journal of Proteome Research

Article

Table 1. Spermatozoa Concentrations and Physical Appearance [Median (Maximum−Minimum)] of the Men Included in the Study first set of samples

mean spermatozoa conc (106/mL) age (years) height (m) weight (kg) BMI (kg/m2) testis size (mL)a a

second set of samples

low sperm conc

intermediate sperm conc

high sperm conc

low sperm conc

intermediate sperm conc

high sperm conc

n = 15

n = 15

n = 15

n = 33

n = 31

n = 33

6 (0.3−17)

54 (45−70)

133 (102−167)

11 (0.3−19)

58 (45−72)

167 (100−392)

19.2 (18.8−23.3) 1.81 (1.74−1.97) 70.1 (59.4−89.9) 20.1 (19.0−28.3) 18 (11−27)

19.8 (18.4−21.2) 1.82 (1.69−1.92) 73.4 (55.0−93.1) 22.8 (19.2−28.6) 25 (13−32)

19.3 (18.7−22.2) 1.83 (1.74−1.91) 76.9 (62.6−113.3) 22.7 (19.6-31.1) 27 (17−30)

19.0 (18.2−23.0) 1.81 (1.69−1.95) 72.5 (53.6−116.8) 22.1 (18.3−28.9) 22 (15−30)

19.0 (18.3−24.6) 1.81 (1.64−2.03) 73.0 (63.0−100.0) 23.0 (18.4−31.9) 22 (13−26)

19.0 (18.3−24.4) 1.81 (1.70−1.90) 74.2 (61.2−95.9) 22.4 (18.6-31.2) 25 (18−35)

Mean of two testes.

the men also had a venous blood sample drawn. Serum was separated and kept frozen at −20 °C until analysis.

concentration range and/or low matrix complexity), it does not appear adapted for trace analysis in very complex biological matrices. In this last case, an appropriate and optimized chromatographic separation is mandatory in order to reduce the ion suppression phenomenon and improve the temporal separation of isomers. Separation methods include GC,18−20 LC,21,22 and capillary electrophoresis.23 Several ionization/ desorption processes can be used, such as FAB,24 MALDI,25,26 or electrospray.27 The analysis of ions is generally achieved by means of quadrupole,28 ion trap,16,29 time-of-flight (TOF),26 Fourier transform ion cyclotron resonance,30 and the recently introduced LTQ-Orbitrap31,32 analyzers. Considering the above, it appears that there are different alternatives in terms of separation, ionization, and acquisition techniques. Consequently, there is no consensual strategy emerging from the scientific community regarding MS-based metabolomics, and a wide heterogeneity of analytical strategies and workflows exists today. The objective of this study was to characterize, for the first time through a metabolomic approach, an alteration of the human testicular function, particularly impaired semen quality. One of the most important objectives was to investigate whether an MS-based metabolomic approach could highlight some significant differences between fingerprints acquired from serum samples collected from healthy men presenting various levels of sperm concentrations. The final purpose was then to point out some particular biomarker candidates expected to characterize impaired semen quality and possibly bring new insights into this disease. Results and challenges emerging from this proof-of-concept study will be described and discussed.

2.2. Sample Preparation

Serum samples (100 μL) were filtered on PALL centrifugal devices (NANOSEP OMEGA, cut off at 10 kDa) at 10000 rpm for 30 min in order to remove molecules with molecular weight higher than 10 kDa. 60 μL of the filtrate was mixed with 20 μL of an internal standard solution. Finally, samples were frozen (−20 °C) until subsequent LC-HRMS analysis. 2.3. LC-HRMS Fingerprinting

Sample fingerprinting was performed on an Agilent 1200 HPLC system including an autosampler and a binary pump, coupled to either a Finnigan LTQ-Orbitrap hybrid mass spectrometer (Thermo Fisher Scientific, Bremen, Germany) or an Agilent Q-TOF 6530 (Agilent, Palo Alto, CA, USA). Chromatographic separation was performed on a HypersilGold column (100 mm × 2.1 mm × 1.9 μm particle size, Thermo Fisher Scientific) connected without splitting to the electrospray interface operating in positive ion mode, as already described for the LTQ-Orbitrap.34 Mass spectra were recorded from 65 Th up to 1000 Th at a resolution of 30000 (fwhm at m/z 400). For the Q-TOF, the same chromatographic separation was performed. The HPLC column was connected without splitting to the electrospray interface operating in positive or negative ion modes. The drying gas (nitrogen) temperature was set to 325 °C, drying gas flow to 5 L/min, nebulizer pressure to 45 psi, and capillary voltage to 4000 V. Fragmentor voltage was set to 100 V. Mass spectra were recorded from 100 Th up to 1000 Th. 2.4. Data Processing

2. EXPERIMENTAL SECTION

Following their acquisition with a given analytical tool, metabolomic fingerprints were deconvoluted to allow the conversion of the three-dimensional raw data (m/z, retention time, ion current) to time- and mass-aligned chromatographic peaks with associated peak areas. Xcalibur (Thermo Fisher Scientific) and MassHunter (Agilent) software programs were used to convert the original data files (*.raw or *.d, respectively) to more exchangeable formats (*.cdf or *.mzdata, respectively). Data processing was then performed using the open-source XCMS software.35 XCMS parameters for the R language were implemented in an automated script. The CentWave algorithm was used for peak picking. For fingerprints acquired through Orbitrap technology, the parameters were the following: ppm = 10, prefilter = c(7,10000), snthresh = 6, peakwidth = c(6,20), mzdiff = 0.01, integrate = 1, noise =

2.1. Biological Samples

142 serum samples were collected at Rigshospitalet (Copenhagen) from Danish young men presenting different semen qualities (Table 1). The samples were divided into two sets of samples (n = 45 and n = 97). Samples were consecutively selected according to whether the men had low (>0−20 mill/ mL), intermediate (45−75 mill/mL), or high sperm concentrations (>100 mill/mL). The men were included from an ongoing study of semen quality of young (median 19 years old) Danish men from the general population. The study has been ongoing since 1996, and details of the study concept have previously been published.8,33 Each participating man delivered one semen sample to the study. At the time of semen delivery, B



Article

fingerprints collected from 15 men having high sperm concentrations and 15 men having low sperm concentrations). The second was constituted of 1660 columns and 30 rows. Our results first demonstrated that positive ionization led to more detected signals than the negative mode. This may be due to redundancy in the data corresponding to adducts and multiply charged ions detected for each metabolite, which are more easily generated in positive ionization mode. The 30 metabolic fingerprints acquired in positive ionization mode were subjected to an OPLS-DA analysis, and a good discrimination between patients presenting low or high sperm concentrations was observed (Figure 1a). The performance

10000. The same parameters were used for fingerprints acquired through Q-TOF technology, except for the prefilter, which was set to c(6,10), and the noise, which was set to 10. 2.5. Data Analysis

Multivariate statistics36,37 were performed using SIMCA-P+ software (Version 12, Umetrics AB, Sweden). To validate the statistical models, the CV-ANOVA was calculated.38 p-values below 0.05 were considered as significant. OPLS was performed with spermatozoa concentrations as the Y-variable (classes or real parameter) on preprocessed, log10[1 + x]-transformed and Pareto-scaled data.39 2.6. Metabolite Identification

Compound identification was performed on the basis of an internal data bank created for LC-ESI-HRMS acquisitions and using ACD/Laboratories software.34 2.7. Quality Control Criteria

2.7.1. Mass Accuracy. A reference solution containing a mixture of five compounds (methyltestosterone, stanozolol, medroxyprogesterone acetate, triamcinolone, and ponasterone A) diluted at a concentration corresponding to 1 ng/μL in ethanol was prepared. Twenty microliters of this solution was added to each sample before injection. It allowed assessment of some acceptability criteria for each injected sequence or highlighting any signal changes in the system (changes in peak shape, retention time, ...). Moreover, when operating under high resolution conditions, mass accuracy is the determining parameter and requires careful assessment for the determination of elemental composition. In the present case, no significant mass accuracy drift was observed during the whole study, this parameter being on average 2 ppm lower than the expected exact mass in positive mode and from 2 ppm above the exact expected mass in negative mode on both instruments used in this study. 2.7.2. Measurement Repeatability. In metabolomics, the goal of absolute quantification, as performed for targeted analysis with isotopic dilution, is not reachable considering the large diversity of metabolites in terms of physicochemical properties, chemical structures, and concentrations. Consequently, a “semi-quantification” is performed considering that we are primarily interested in relative changes and not in an accurate quantification of each metabolite. Nevertheless, some precautions are taken through the use of several “quality controls” to ensure that the instrument is as stable as possible all along the sequence of acquisition and that the observed differences (relative changes) are not spurious. Moreover, such quality controls allow the characterization and control of the total variance associated with the MS instrument. In this study, the signal repeatability was evaluated by injecting the same sample in 20 replicates onto both instruments. The absolute repeatability (mean relative standard deviation) was found to be around 25% for the majority of the MS signals constituting the metabolic fingerprints, which was considered acceptable.

Figure 1. T-predicted score plot (OPLS-DA) observed for 30 metabolic fingerprints acquired on a LC-Q-TOF instrument (first data set) and generated after positive electrospray ionization (a) or negative electrospray ionization (b). Black and white circles stand for patients presenting low or high sperm concentrations, respectively.

3. RESULTS AND DISCUSSION 3.1. Patients Presenting Different Semen Qualities Exhibited Different Metabolic Profiles

characteristics of this multivariate model from a descriptive and predictive point of view were as follows: R2(X) = 0.488, R2(Y) = 0.995, and Q2(Y) = 0.718. This model was statistically validated (p-value of CV ANOVA = 0.0016). Similarly, an OPLS-DA analysis was performed on the data set acquired in negative ionization mode, and the same classification was obtained (Figure 1b, R2(X) = 0.452, R2(Y) = 0.998, and Q2(Y)

Two distinct data sets were generated initially from the first sample set corresponding to metabolic fingerprints acquired on the LC-Q-TOF instrument after the positive and negative ionization modes, respectively. The first one was made up of 5041 columns ([m/zi - rti]) and 30 rows (metabolic C



Article

= 0.727, CV ANOVA p-value = 0.0012). These results demonstrated the existence of clear different serum biological signatures from patients presenting different sperm concentrations. 3.2. Metabolic Profiles Were Confirmed To Be Linked with the Sperm Concentration Parameter

In a second step, a model no longer based on the class membership of the samples, that is low or high sperm concentrations, but based on the multivariate correlation between the LC/MS data from the first sample set and the clinical parameter values was built. To this end, new data sets were generated by adding to the former ones the fingerprints collected from patients presenting intermediary sperm concentrations. This resulted in two data sets constituted of 5041 columns and 45 rows and 1660 columns and 45 rows, in positive and negative ionization mode, respectively. These new sets were subjected to quantitative OPLS analyses. The characteristics of the two new models were as follow: R2(X) = 0.451, R2(Y) = 0.987, and Q2(Y) = 0.673 (CV ANOVA p-value = 7.7 × 10−6) and R2(X) = 0.357, R2(Y) = 0.954, and Q2(Y) = 0.503 (CV ANOVA p-value = 6.7 × 10−4) for data sets obtained after positive and negative electrospray, respectively. Most of the variance related to the sperm concentration values (quantitative Y-variable) seemed to be explained by these two models. Figure 2 shows the good correlation found between the known sperm concentrations and the predicted ones. This result demonstrates that the metabolic differences observed between patients presenting low and high sperm concentrations appear directly linked to this specific clinical parameter. 3.3. The Primer Results Were Confirmed with an Independent Sample Set

In order to confirm these observations, a second and independent set of samples was analyzed. For this second part of the study, metabolomic fingerprints were acquired also on a second independent LC-LTQ-Orbitrap system operating in positive ionization mode. The obtained data set was constituted of 1922 columns and 97 rows. When subjected to OPLS-DA analysis, the same classification previously observed between patients presenting low or high sperm concentrations was obtained (Figure 3, R2(X) = 0.608, R2(Y) = 0.954, Q2(Y) = 0.725, and CV-ANOVA p-value = 3.7 × 10−12). Moreover, the correlation between the generated metabolic fingerprints and the clinically determined sperm concentration values was confirmed (Figure 3, R2(X) = 0.564, R2(Y) = 0.773, Q2(Y) = 0.5, and CV-ANOVA p-value = 1.24 × 10−10). The fact that patients presenting different sperm concentrations present different metabolic signatures in relation to this clinical parameter was then demonstrated twice on two independent data sets.

Figure 2. Y-predicted score plot (OPLS) observed for 45 metabolic fingerprints acquired on a LC-Q-TOF instrument (first data set) and generated after positive electrospray ionization (a) or negative electrospray ionization (b). Black dots and white circles stand for patients presenting low or high sperm concentrations, respectively. White squares stand for patients presenting intermediary sperm concentrations.

biological interpretation remains tricky, as they can be related to a lot of metabolic pathways. During this study the correct statistical classification of individuals according to their sperm concentration on the basis of metabolic fingerprints collected on serum samples has been demonstrated twice on independent sample sets, so that the observed outcomes cannot be spurious. However, the number of common signals (in bold in Table 2) involved in this classification between the two sets of samples remains relatively low. They were found down-regulated in patients presenting low sperm concentrations (Figure 4). They were structurally elucidated through the use of multidimensional mass spectrometry (Figure 5) and corresponded to small peptides which all seem to be related to the same protein, putatively a fragment of Protein Complement C3f, involved in innate immunity. These peptides were already highlighted in an ovarian cancer study,40 but the biological interpretation of this finding remains unclear at this stage. Thus,

3.4. Which Metabolites Are Associated with Low Sperm Concentrations?

As demonstrated earlier, the existence of a metabolic signature associated with sperm concentration was found and confirmed. Nevertheless, after a deeper investigation of these complex chemical phenotypes, we found that the signature of impaired semen qualityeven if statistically significantappeared quite discrete: few metabolites were evidenced (Table 2), which presented, in addition, relatively small fold changes (from 0.4 to 1.7) between subjects. Among the differential MS signals highlighted in the sample sets studied, mostly amino acids and carboxylic acids were identified, compounds for which a D



Article

Table 2. List of Annotated Features Highlighted in Sample Set 1 and/or Sample Set 2 Based on Internal Database (Monoisotopic Mass, Observed Retention Time, Observed Ratio between Intensities Measured in Men, Identified Metabolites, and Chemical Taxonomy)

a

Mmi

RT (min)

ratioa

metabolites

203.116

0.87

1.70**

259.178

7.32

1.62***

131.095

0.95

0.71***

117.079

0.75

0.73*

o-acetyl carnitine hexanoylcarnitine leucine/ isoleucine valine/betaine

181.074 149.051 204.090 115.063 104.047

0.82 0.77 2.72 0.67 0.8

0.5*** 0.6*** 1.61*** 0.81* 0.51***

174.016 118.027 106.027 129.042

0.73 0.82 0.7 0.73

0.52** 0.49*** 0.70** 0.68***

942.366 245.185 619.328

7.85 6.67 7.42

0.54*** 0.44*** 0.41***

tyrosine methionine tryptophan proline 3-hydroxy isobutyric acid cis-aconitic acid succinic acid glyceric acid pyroglutamic acid HWESASLL LL ESASLL

taxonomy acyl-carnitine (lipids) acyl-carnitine (lipids) essential amino acid essential amino acid/ glycine derivatives essential amino acid essential amino acid essential amino acid amino acid organic acid carboxylic acid carboxylic acid carbohydrate glutamic acid derivative peptide peptide peptide

p-value of a Student t test: *, p < 0.05; **, p < 0.01; ***, p < 0.001.

Figure 3. (a) T-predicted score plot (OPLS-DA) observed for 66 metabolic fingerprints (second data set) generated after positive electrospray ionization. Black dots and white circles stand for patients presenting low or high sperm concentrations, respectively. (b) Ypredicted score plot (OPLS) observed for 97 metabolic fingerprints (second data set) generated after positive electrospray ionization. Black dots and white circles stand for patients presenting low or high sperm concentrations, respectively. White squares stand for patients presenting intermediary sperm concentrations.

the specificity of this protein as a marker of cancer and/or fertility remains to be better characterized. 3.5. Challenges of the Metabolomic Approach in Human

An important challenge encountered during this study has originated from the very high interindividual variability observed among the volunteers. This directly affected the statistical strength of the correlation between metabolic phenotypes and targeted clinical outcomes. Indeed, in human studies conducted at population scale, the observed changes in terms of metabolic profiles result from many different sources of variability (diet,41 diurnal variation,42,43 cultural influences,44 but also therapeutics, ethnic differences45). These multiple sources of variability could have drowned the relevant information and contributed to blur the metabolic picture to characterize. This issue represents a limitation of the metabolomic approach, particularly when the biological signature associated with the studied pathology is subtle (i.e., small fold changes). Besides, the metabolic profiles generated

Figure 4. Mean abundances of peptide HWESASLL measured in the three groups of samples collected from patients presenting low, medium, or high sperm concentrations. The 95% confidence interval is marked on each column (***, p < 0.001, Student t-test).

from the individual serum samples may not characterize comprehensively the population studied. Finally, metabolomics today appears as a new methodological approach with real potential in many fields, including environmental sciences. This integrative biological approach may, in particular, strengthen studies related to endocrine disruption and possibly relate pathologies outcomes and environmental contaminants, thus demonstrating fascinating descriptive capabilities as well as new promises to reveal meaningful biomarkers from a mechanistic point of view. E



Article

between men with poor or high sperm concentrations) were pointed out, the number and fold change amplitude of which however remain limited. Peptides related to the Protein Complement C3f were identified as putative markers. The biological interpretation and further robustness associated with this observation remain to be further investigated, in particular to address the inter- and intraindividual variabilities and the representativeness of single serum samples.

■

AUTHOR INFORMATION

Corresponding Author

*Tel: 00 33 2 40 68 78 80. Fax: 00 33 2 40 68 78 78. E-mail: [email protected]. Notes

The authors declare no competing financial interest.

■

Figure 5. Quadrupole−time of flight ion product scan for ion M942T478 (singly charged of M472T478) at a collision energy of 45 eV after positive electrospray ionization.

ACKNOWLEDGMENTS This work was financially supported by the European Commission in the frame of the 7th FP (call proposal FP7ENV-2007-1, Grant Agreement No. 212844, Acronym DEER). The authors express thanks to Pf Niels Jørgensen and Pf Niels Erik Skakkebæk from Righospitalet for supplying serum samples and related clinical information used for this study.

Nevertheless, the high intra- and interindividual variability observed in samples of human origin remains a major current limitation of metabolomics, especially for predictive or diagnostic purposes. Then, controlled animal and/or in vitro/ ex vivo experimental models may still constitute a valuable help, at least as a preliminary step, by limiting the biological variability between individuals as well as minimizing possible confounding factors. Results obtained from such experimental models would permit refinement of the research hypotheses, the experimental design to be used, and/or some methodological aspects of the metabolomic worflow (hypothesis-driven research). Besides, metabolomics is still facing some limitation regarding data processing and analysis, and developments in advanced statistics are still required despite important developments made in this field in recent years, for instance the ones allowing separation of the within-subject variation from the intersubject variation in metabolite profiles using multilevel paired analysis.46−48 This kind of tool (which could be tailored to unpaired analysis) could be extremely useful in human studies interested in discrete signatures for which the possibility of separating interindividual variability caused by genetic, dietary, lifestyle, and environmental factors from related metabolic effects of exposure or pathologies could open up new perspectives. Finally, the extremely wide diversity of potential metabolites in terms of chemical structures and concentrations still makes the goal of exhaustiveness in metabolomics unrealistic. A substantial benefit is then expected from crossing complementary analytical techniques in the scope of reaching a more comprehensive metabolomic profiling.

■

REFERENCES

(1) Carlsen, E.; Giwercman, A.; Keiding, N.; et al. Evidence for decreasing quality of semen during past 50 years. BMJ 1992, 305, 609−13. (2) Jouannet, P.; Wang, C.; Eustache, F.; et al. Semen quality and male reproductive health: the controversy about human sperm concentration decline. APMIS 2001, 109, 333−344. (3) Merzenich, H.; Zeeb, H.; Blettner, M. Decreasing sperm quality: a global problem. BMC Public Health 2010, 10, 24. (4) Nordkap, L.; Joensen, U.; Jensen, M.; Jørgensen, N. Regional differences and temporal trends in male reproductive health disorders: Semen quality may be a sensitive marker of environmental exposures. Mol. Cell. Endocrinol. 2012, 355, 221−230. (5) Chia, V.; Quraishi, S.; Devesa, S.; et al. International trends in the incidence of testicular cancer, 1973−2002. Cancer Epidemiol. Biomarkers Prev. 2010, 5, 1151−1159. (6) Boisen, K.; Chellakooty, M.; Schmidt, I.; et al. Hypospadias in a cohort of 1072 Danish newborn boys: prevalence and relationship to placental weight, anthropometrical measurements at birth, and reproductive hormone levels at three months of age. J. Clin. Endocrinol. Metab. 2005, 7, 4041−4046. (7) Skakkebaek, N.; Rajpert-De Meyts, E.; Main, K. M. Testicular dysgenesis syndrome: an Increasingly common developpmental disorder with environmental aspects. Hum. Reprod. 2001, 16, 972− 978. (8) Jorgensen, N.; Joensen, U.; Jensen, T.; Jensen, M.; Almstrup, K.; Olesen, I.; et al. Human semen quality in the new millennium: a prospective cross-sectional population-based study of 4867 men. BMJ Open 2012, DOI: 10.1136/bmjopen-2012-000990. (9) Nicholson, J.; Lindon, J.; Holmes, E. ’Metabonomics’: understanding the metabolic responses of living systems to pathophysiological stimuli via multivaraiate statistical analysis of biological NMR spectroscopic data. Xenobiotica 1999, 29, 1181−1189. (10) Nicholson, J.; Connelly, J.; Lindon, J.; Holmes, E. Metabonomics: a platform for studying drug toxicity and gene function. Nat. Rev. Drug Discovery 2002, 1, 153−161. (11) Zhang, A.; Sun, H.; Wang, X. Serum metabolomics as a novel diagnostic approach for disease: a systematic review. Anal. Bioanal. Chem. 2012, 404, 1239−1245.

4. CONCLUSION In this study, a metabolomic approach was applied to serum samples collected from men presenting different semen qualities in order to reveal from these chemical phenotypes potential biomarkers characterizing an impaired semen quality. Serum metabolic profiles from men presenting different sperm concentrations were found to be significantly different. The developed methodology also permitted relating quantitatively the studied clinical parameter (i.e., sperm concentration) to the generated metabolite profiles. These results were confirmed with an independent set of samples. Few differential metabolites (i.e., presenting statistically different abundances F



Article

(12) Deepinder, F.; Chowdary, H.; Agarwal, A. Role of metabolomic analysis of biomarkers in the management of male infertility. Expert Rev. Mol. Diagn. 2007, 7, 351−358. (13) Theodoridis, G.; Gika, H.; Wilson, I. LC-MS-based methodology for global metabolite profiling in metabonomics/metabolomics. Trends Anal. Chem. 2008, 27, 251−260. (14) Antignac, J. P.; Courant, F.; Pinel, G.; Monteau, F.; Le Bizec, B. Mass spectrometric based metabolomic in the field of chemical risk: state of the art, pitfalls, and future trends. Trends Anal. Chem. 2011, 30, 292−301. (15) Alexandre-Gouabau, M. C.; Courant, F.; Le Gall, G.; et al. Offspring metabolomic response to maternal protein restriction in a rat model of intrauterine growth restriction (IUGR). J. Proteome Res. 2011, 10, 3292−3302. (16) Madalinski, G.; Godat, E.; Alves, S.; et al. Direct introduction of biological samples into a LTQ-Orbitrap hybrid mass spectrometer as a tool for fast metabolome analysis. Anal. Chem. 2008, 80, 3291−3303. (17) McDougall, G.; Martinussen, I.; Stewart, D. Towards fruitful metabolomics: high throughput analyses of polyphenol composition in berries using direct infusion mass spectrometry. J. Chromatogr., B 2008, 871, 362−369. (18) Pasikanti, K.; Ho, P.; Chan, E. Gas chromatography/mass spectrometry in metabolic profiling of biological fluids. J. Chromatogr., B 2008, 871, 202−211. (19) Anizan, S.; Bichon, E.; Monteau, F.; et al. A new reliable sample preparation for high throughput focused steroid profiling by Gas Chromatography-Mass Spectrometry. J. Chromatogr., A 2010, 1217, 6652−6660. (20) Zhang, Q.; Wang, G.; Du, Y.; Zhu, L. A. J. GC/MS analysis of the rat urine for metabonomic research. J. Chromatogr., B 2007, 854, 20−25. (21) Idborg, H.; Zamani, L.; Edlund, P.-O.; Schuppe-Koistinen, I.; Jacobsson, S. Metabolic fingerprinting of rat urine by LC/MS Part1. Analysis by hydrophilic interaction liquid chromatography-electrospray ionization mass spectrometry. J. Chromatogr. B 2005, 828, 9−13. (22) Idborg, H.; Zamani, L.; Edlund, P.-O.; Schuppe-Koistinen, I.; Jacobsson, S. Metabolic fingerprinting of rat urine by LC/MS Part 2. Data pretreatment methods for handling of complex data. J. Chromatogr., B 2005, 828, 14−20. (23) Garcia-Villalba, R.; Leon, C.; Dinelli, G.; et al. Comparative metabolomic study of transgenic versus conventional soybean using capillary electrophoresis-time-of-flight mass spectrometry. J. Chromatogr., A 2008, 1195, 164−173. (24) Platt, J.; Uy, O.; Heller, D.; Cotter, R.; Fenselau, C. Computerbased linear regression analysis of desorption mass spectra of microorganisms. Anal. Chem. 1988, 60, 1415−1419. (25) Dave, K.; Hamilton, B.; Wallis, T.; Furness, S.; Whitelaw, M.; Gorman, J. Identification of N,Nε-dimethyl-lysine in the murine dioxin receptor using MALDI-TOF/TOF- and ESI-LTQ-Orbitrap-FT-MS. Int. J. Mass Spectrom. 2007, 268, 168−180. (26) Jarman, K.; Cebula, S.; Saenz, A.; et al. An algorithm for automated bacterial identification using matrix-assisted laser desoption/ionisation mass spectrometry. Anal. Chem. 2000, 72, 1217−1223. (27) Nordström, A.; O’Maille, G.; Qin, C.; Siuzdak, G. Nonlinear data alignment for UPLC-MS and HPLC-MS based metabolomics: quantitative analysis of endogenous and exogenous metabolites in human serum. Anal. Chem. 2006, 78, 3289−3295. (28) Anizan, S.; Bichon, E.; Di Nardo, D.; et al. Screening of androstenedione misuse in cattle by LC-MS/MS profiling of glucuronide and sulfate steroids in urine. Talanta 2011, 86, 186−194. (29) Rochfort, S.; Trenerry, V.; Imsic, M.; Panozzo, J.; Jones, R. Class targeted metabolomics: ESI ion trap screening methods for glucosinates based on MSn fragmentation. Phytochemistry 2008, 69, 1671−1679. (30) Sleighter, R.; Hatcher, P. The application of electrospray ionization coupled to ultrahigh resolution mass spectrometry for the molecular characterization of natural organic matter. J. Mass Spectrom. 2007, 42, 559−574.

(31) Dunn, W.; Broadhurst, D.; Brown, M.; et al. Metabolic profiling of serum using Ultra Performance Liquid Chromatography and the LTQ-Orbitrap mass spectrometry system. J. Chromatogr., B 2008, 871, 288−298. (32) Courant, F.; Pinel, G.; Bichon, E.; et al. Development of a metabolomic approach based on liquid chromatography high resolution mass spectrometry to screen for clenbuterol abuse in calves. Analyst 2009, 134, 1637−1646. (33) Jorgensen, N.; Carlsen, E.; Nermoen, I.; et al. East-West gradient in semen quality in the Nordic-Baltic area: a study of men from the general population in Denmark, Norway, Estonia and Finland. Hum. Reprod. 2002, 17, 2199−2208. (34) Courant, F.; Royer, A. L.; Chéreau, S.; et al. Implementation of a semi-automated strategy for the annotation of metabolomic fingerprints generated by liquid chromatography-high resolution mass spectrometry from biological samples. Analyst 2012, 137, 4958−4967. (35) Smith, C.; Want, E.; O’Maille, G.; Abagyan, R.; Siuzdak, G. XCMS: Processing mass spectrometry data for metabolite profiling using non linear peak alignment, matching, and identification. Anal. Chem. 2006, 78, 779−787. (36) Trygg, J.; Wold, S. Orthogonal projections to latent structures (O-PLS). J. Chemom. 2002, 16, 119−128. (37) Trygg, J. O2-PLS for qualitative and quantitative analysis in multivariate calibration. J. Chemom. 2002, 16, 283−293. (38) Eriksson, L.; Trygg, J.; Wold, S. CV-ANOVA for significance testing of PLS and OPLS models. J. Chemom. 2008, 22, 594−600. (39) van den Berg, R.; Hoefsloot, H.; Westerhuis, J.; Smilde, A.; van der Werf, M. Centering, scaling, and transformations: improving the biological information content of metabolomics data. BMC Genomics 2006, 7. (40) Silva, E.; Regal, P.; Atkinson, E.; Fente, C. A new approach for identifying patients with ovarian epithelial neoplasms based on highresolution mass spectrometry. Am. J. Clin. Pathol. 2010, 34, 903−909. (41) Stella, C.; Beckwith-Hall, B.; Cloarec, O.; et al. Susceptibility of human metabolic phenotypes to dietary modulation. J. Proteome Res. 2006, 5, 2780−2788. (42) Hodson, M.; Dear, G.; Roberts, A.; et al. A gender-specific discriminator in Sprague-Dawley rat urine: The deployment of a metabolic profiling strategy for biomarker discovery and identification. Anal. Biochem. 2007, 362, 182−192. (43) Lenz, E.; Wilson, I. Analytical Strategies in Metabolomics. J. Proteome Res. 2007, 6, 443−458. (44) Lenz, E.; Bright, J.; Wilson, I.; et al. Metabonomics, dietary influences and cultural differences: a 1H NMR-based study of urine samples obtained from healthy British and Swedish subjects. J. Pharm. Biomed. Anal. 2004, 36, 841−849. (45) Dumas, M.; Maibaum, E.; Teague, C.; et al. Assessment of Analytical Reproducibility of 1H NMR Spectroscopy Based Metabonomics for Large-Scale Epidemiological Research: the INTERMAP Study. Anal. Chem. 2006, 78, 2199−2208. (46) van der Greef, J.; Smilde, A. Symbiosis of chemometrics and metabolomics: past, present, and future. J. Chemom. 2005, 19, 376− 386. (47) van Velzen, E.; Westerhuis, J.; van Duynhoven, J.; et al. Multilevel data analysis of a crossover designed human nutritional intervention study. J. Proteome Res. 2008, 7, 4483−4491. (48) Van Dorsten, F.; Grün, C.; Van Velzen, E.; Jacobs, D.; Draijer, R.; Van Duynhoven, J. The metabolic fate of red wine and grape juice polyphenols in humans assessed by metabolomics. Mol. Nutr. Food Res. 2010, 54, 897−908.

G


Metabolomics as a Potential New Approach for Investigating Human

Recommend Documents