Figures
Abstract
Background
Sex is an important factor in the prevalence, incidence, progression, and response to treatment of many medical conditions, including autoimmune and cardiovascular diseases and psychiatric conditions. Identification of molecular differences between typical males and females can provide a valuable basis for exploring conditions differentially affected by sex.
Methodology/Principal Findings
Using multiplexed immunoassays, we analyzed 174 serum molecules in 9 independent cohorts of typical individuals, comprising 196 males and 196 females. Sex differences in analyte levels were quantified using a meta-analysis approach and put into biological context using k-means to generate clusters of analytes with distinct biological functions. Natural sex differences were established in these analyte groups and these were applied to illustrate sexually dimorphic analyte expression in a cohort of 22 males and 22 females with Asperger syndrome. Reproducible sex differences were found in the levels of 77 analytes in serum of typical controls, and these comprised clusters of molecules enriched with distinct biological functions. Analytes involved in fatty acid oxidation/hormone regulation, immune cell growth and activation, and cell death were found at higher levels in females, and analytes involved in immune cell chemotaxis and other indistinct functions were higher in males. Comparison of these naturally occurring sex differences against a cohort of people with Asperger syndrome indicated that a cluster of analytes that had functions related to fatty acid oxidation/hormone regulation was associated with sex and the occurrence of this condition.
Conclusions/Significance
Sex-specific molecular differences were detected in serum of typical controls and these were reproducible across independent cohorts. This study extends current knowledge of sex differences in biological functions involved in metabolism and immune function. Deviations from typical sex differences were found in a cluster of molecules in Asperger syndrome. These findings illustrate the importance of investigating the influence of sex on medical conditions.
Citation: Ramsey JM, Schwarz E, Guest PC, van Beveren NJM, Leweke FM, Rothermundt M, et al. (2012) Molecular Sex Differences in Human Serum. PLoS ONE 7(12): e51504. https://doi.org/10.1371/journal.pone.0051504
Editor: Rizwan Hasan Khan, Aligarh Muslim University, India
Received: July 6, 2012; Accepted: November 1, 2012; Published: December 12, 2012
Copyright: © 2012 Ramsey et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: This study was supported by the Psynova Neurotech Inc; the European Union FP7 SchizDX research programme (grant reference 223427); the Dutch Fund for Economic Structure Reinforcement (FES) under grant agreement number 0908 (the “NeuroBasic PharmaPhenomics project”); and the AutismSpeaks grant (#6009). SBC, BA, and LR were supported by the MRC and the Wellcome Trust during the period of this work. JMR was supported by the Edmonton Churchill Scholarship (Canada), the Natural Sciences and Engineering Research Council of Canada (NSERC) Postgraduate Scholarship, and the Cambridge Commonwealth Trust for the duration of the study. The funders had no role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: ES, PCG and SB are consultants for Myriad Genetics Inc. The other authors have no potential conflict of interest.
Introduction
Sexual dimorphism in underlying processes in medical conditions are numerous and diverse, occurring in diseases ranging from autoimmune and cardiovascular conditions to neurological conditions [1]–[3]. Parameters such as disease prevalence, incidence, age at onset, progression, mortality, and treatment response can show sex differences [4]. The key to some of these differences may be in sex-dependent regulation of biological pathways. Such differences have been investigated in the context of immune response, and elevated immune activation in females has been linked to the significantly increased susceptibility of women to multiple sclerosis (MS) and other autoimmune diseases such as rheumatoid arthritis, Grave’s disease, and lupus erythematosus [1]. Several possible avenues have been explored to explain such differences in autoimmune diseases. For example, one study showed that typical female mice had detectable levels of immunoglobulin G auto-antibodies that were absent in male mice [5]. Similarly, higher blood immunoglobulin levels and CD4/CD8 T-cell ratios have been found in typical women, along with lower natural killer cell and antibody-dependent cell-mediated cytotoxicity [1]. Sex dimorphic expression of particular cytokines, such as transforming growth factor (TGF)-β1 and interleukin (IL)- 4, has also been implicated in osteoarthritis in mice [6].
Physiological and molecular mechanisms causing sex dimorphisms have also been investigated in the context of cardiovascular diseases. Myocardial infarctions in women result in a higher mortality rate and poorer prognosis compared to men [7]. Reports have shown that sex differences in myocardial function appear during physiological stress [7]. Sex differences in the stress responses of rodent cells have been observed along with sexually dimorphic gene expression of stress-related genes both before and after application of stress [8]. This study showed intrinsic sex differences in cell response to stressors such as ethanol and influenza A virus even without exposure to sex hormones [8]. Basal hypothalamic-pituitary-adrenal (HPA) axis function was also increased in females [9]. It has been suggested that the reproductive and HPA axes work together with the immune system to maintain homeostasis [9]. Stress and immune markers are likely to show sex specific expression and response to stimuli, potentially creating different susceptibilities to autoimmune, cardiovascular, and other diseases in which stress and immune responses play a role in disease vulnerability.
An increased understanding of sex dimorphisms in biological regulation may also help to elucidate potential differences in treatment response. For example, growth hormone, which has been used to treat a variety of irregularities associated with cardiovascular, immune, metabolic, psychological and other biological functions, is less effective in women than men [10]. Sex differences in growth hormone regulation were found to be responsible for these differences. The actions of growth hormone, whose release is continuous in females and episodic in males, were suppressed in cells from females [10]. This suppression has been hypothesized to be due to suppressed signalling through cytokines and the Jak2/Stat5B pathway that usually activates growth hormone [10].
Studies measuring sex differences at the molecular level have so far been limited to the investigation of only a few molecules and specific disease and inflammation processes [1], [11]–[14]. The aim of the present study was to elucidate such differences at a systematic level through measurement of 174 serum molecules in a large cohort of typical individuals. The investigated molecules included cytokines, chemokines, hormones, growth factors, angiogenesis and central nervous system-related analytes, as well as other serum proteins important in disease (Myriad RBM website. Available: http://myriadrbm.com/. Accessed 2012 May 18). The applied multiplexed immunoassay platform has been used previously to explore molecular changes in cancer and autoimmune, cardiovascular, gastrointestinal, neurological, and various other diseases, many of which show sex differences [15]–[20]. In addition, we attempted to relate the discovered sex differences to those measured in samples from participants with Asperger syndrome. This condition shows a particularly pronounced sex dimorphism as the prevalence is 4–10-fold higher in males [21].
Materials and Methods
Clinical Samples
Protocols for the study were approved by ethical committees at all involved university hospitals (see below) and carried out in accordance with the Declaration of Helsinki. Informed written consent was obtained from all participants. Individuals with a family history of mental illness or medical conditions like hypertension, type II diabetes, cardiovascular, or autoimmune diseases were excluded from the study. Pregnant females were also excluded from the study. A total of nine cohorts of typical individuals were used in this investigation. Cohort 1 was from the Autism Research Centre, University of Cambridge, Department of Psychiatry, UK; cohorts 2 and 3 were from University of Cologne Department of Psychiatry, Germany; cohorts 4 to 6 were from the University of Muenster, Germany; cohorts 7 and 8 were from the University of Magdeburg, Germany; and cohort 9 was from Erasmus Medical Centre in the Netherlands. In addition, a separate cohort of adults with Asperger syndrome (Cohort 10) was obtained from the Autism Research Centre, University of Cambridge, Department of Psychiatry, UK. Diagnosis of Asperger syndrome by psychiatrists was based on the Structured Clinical Interview for the Diagnostic and Statistic Manual-IV-Text Review Disorders and the DSM-IV-TR. Biological sex was used to classify males and females. Females were not excluded for contraceptive use and were at variable times in the menstrual cycle. Blood was collected in the morning after overnight fasting, serum prepared and 181–247 analytes measured using multiplexed immunoassay analyses in a CLIA-certified laboratory at Myriad-RBM (Austin, TX, USA), as described previously [18].
Data Pre-processing
The statistical programming software R was used to match controls, replace missing analyte values, remove outliers, and to perform meta-analysis. For each of the nine cohorts, typical males and females were matched according to age, BMI, waist circumference, smoking, and cannabis consumption. Demographic details including available metadata used for matching in all nine cohorts are shown in Table 1. Values outside the linear range of the assay system were either replaced by half the minimum value or by double the maximum value measured for the respective analyte. Analytes with more than 70% missing values were eliminated from the dataset. Outlier removal was performed for each analyte separately and values more than four standard deviations from the overall mean were excluded.
Meta-analysis
Meta-analysis was carried out using the non-parametric Cliff’s delta as a measure of effect size [22]–[25]. Cliff’s delta estimated the probability that the level of an analyte was higher in males than in females versus the reverse probability. Cliff’s delta was calculated in all cohorts to quantify the difference in molecular levels between males and females, and the pooled effect was determined using a random effects meta-analysis. This approach was chosen after finding significant heterogeneity between cohorts as assessed by a standard χ2 test [26]. Random effects meta-analysis accounts for this heterogeneity allowing cohort-specific average effects to vary as part of a common distribution. This approach has been used recently to detect novel loci in genome wide association studies for various disorders [27], [28]. Other uses have ranged from assessing the effect of intentional weight loss on depressive symptoms to examining the relationship between physical activity and risk of colon adenoma [29], [30]. For the determination of the pooled effect size, a minimum of three cohorts was required for each analyte.
All determined p-values were adjusted for the false discovery rate according to the method of Benjamini and Hochberg [31]. Adjusted p-values (q-values) of less than 0.05 were considered to indicate statistical significance.
K-means Clustering and Principal Component Analysis
To assign molecules to biological pathways, we first employed a clustering approach to group analytes with similar concentration patterns. For this purpose, k-means clustering was performed on standardized analyte levels measured in the largest cohort (cohort 7). To avoid the influence of outlying observations on the clustering, such values were replaced with a uniform random number between the minimum and maximum analyte values. The number of clusters was pre-specified to 6, but different qualitatively similar results were obtained by using other cluster numbers.
The main biological functions for the analytes in each cluster of related molecules were identified using Ingenuity Pathway Knowledgebase (IPA) software with all measured molecules as the reference dataset (Ingenuity website. Available: http://www.ingenuity.com/. Accessed 2012 April 18). Main biological functions were determined by identifying overrepresented functions of the molecules in the cluster. These biological functions were verified using the Database for Annotation, Visualization and Integrated Discovery (DAVID ) version 6.7 (DAVID website. Available: http://david.abcc.ncifcrf.gov/. Accessed 2012 Oct 16) [32], [33]. The IPA software was also used to create networks for clustered analytes, taking into consideration direct and indirect relationships between proteins.
The molecular clusters and their relationships to results from the meta-analysis of sex differences were visualized using principal component analysis (PCA; SIMCA-P+ Version 12.0). Principal components were used to create a single composite variable summarizing the information of all molecules in each cluster. For this purpose, we used the first principal component of standardized analyte concentrations in each cluster and determined sex differences. These were combined across cohorts using a random effects meta-analysis. Since principal components are indifferent with respect to sign, this was determined using the average sign of the individual standardized molecular concentrations within each cluster. The average sign was calculated from the molecules with the highest principal component loadings. Subsequently, ANOVA was used on all principal component scores to quantify sex-by-condition interactions in groups of analytes. P-values were adjusted for the false discovery rate with the Benjamini and Hochberg method [31].
Results
Meta-analysis
After data pre-processing, a pooled Cliff’s delta was calculated for each of the 174 serum molecules across the nine cohorts of typical individuals. Of these, a total of 77 analytes were present at significantly different concentrations between male and female participants after adjusting for the false discovery rate (q <0.05, Table 2 and 3). A positive Cliff’s delta value indicates that a given analyte was higher in males more often, and a negative Cliff’s delta value indicates the reverse. Table 2 shows that 40 molecules had higher concentrations in females and Table 3 shows that 37 molecules were more frequently higher in male participants. Most data for the displayed analytes had no missing values and all but five had less than 15% missing values.
Clusters and Molecular Sex Differences
Functional assignment of analytes was performed by clustering molecular data from cohort 7 and determining the significantly enriched biological functions of the molecules in each cluster (Table 2 and 3). The full list of clustered molecules and their assignment to the main biological function as determined by IPA software can be found in Table S1. Apart from one, all clusters were assigned a distinct biological function.
Molecular clusters and their relationships to sex differences were visualized using PCA (Figure 1A). This figure reflects the difference in molecular profiles between clusters resulting from the applied clustering procedure. Figure 1B shows the same plot with identified molecular sex differences determined by the meta-analysis. The clustered groups with distinct biological functions mainly showed consistent sex differences in analyte levels.
A) Assignment of analyte clusters to biological functions; B) Sex differences as determined by meta-analysis. Colouring indicates significance (q<0.05). The grey area indicates analytes that show no significant sex differences. Plots were generated using data from cohort 7, in which 167 analytes were measured in 162 typical controls (81 females, 81 males).
We then assessed sex differences in each of the clustered groups of analytes combined across all cohorts using a composite value. This value summarized the sex differences observed in a given group of molecules (Figure 2) and mirrored the changes seen for individual analytes (Figure 1). The cluster associated with energy production, fatty acid metabolism, and hormone levels showed the most significant sex differences for groups with a distinct biological function. This included molecules such as growth hormone, sex hormone binding globulin, trefoil factor 3, leptin, apolipoprotein AI, adiponectin and thyroxine binding globulin. These all showed higher levels in females compared to males with ratios ranging from 1.4 to 10.0 (Table 2). The molecules which were more abundant in males were associated mainly with immune cell chemotaxis, although the ratiometric differences were smaller than those observed for catabolism of fatty acids and regulation of hormone levels in females (Table 3).
Values for typical individuals were pooled across nine cohorts; values for Asperger syndrome participants were calculated from cohort 10. The x-axis shows the difference between composite values that reflect the average molecular levels in males and females. Horizontal bars indicate 95% confidence intervals of the difference between sexes.
Identification of Molecular Differences in Asperger Syndrome Associated with Sex
We next applied the sex differences identified in typical controls to investigate the potential association of these with Asperger syndrome. Figure 2 shows the overlay of the naturally occurring sex differences found for each biological function with those observed in a cohort of 44 individuals with Asperger syndrome. Most sex differences were consistent with those observed in typical participants, further validating the result of the meta-analysis. However, a disease-sex interaction approaching significance was observed for molecules associated with fatty acid metabolism and hormone production (P = 0.022; q = 0.132) (Figure 2). This indicates that females with Asperger syndrome showed lower levels of these molecules than would be expected from the meta-analysis of typical controls.
Figure 3 shows the two top networks from Ingenuity Pathway Knowledgebase software for the group of molecules associated with fatty acid metabolism and hormone function. Molecules are coloured according to typical sex differences and circled where significant interactions were found for Asperger syndrome and sex in [34]. Stem cell factor (SCF) and receptor for advanced glycosylation end products (RAGE) in one network, and growth hormone (GH) in the other network showed significant sex-specific alterations in Asperger syndrome. The software showed that these two networks, containing most of the molecules in the cluster, were partially overlapping.
Top networks of the cluster of molecules associated with energy production, fatty acid metabolism, and hormone levels from Ingenuity Pathway Knowledgebase software. Individual molecules are coloured according to significant sex difference in controls. Molecules with significant sex-disease interactions from [34] are circled. A2M (alpha 2 macroglobulin), ADIPOQ (adiponectin), RAGE (receptor for advanced glycosylation end products), ANGPT2 (angiopoietin 2), ARNT (aryl hydrocarbon receptor nuclear translocator), CRP (C-reactive protein), ENA-78 (epithelial derived neutrophil activating protein 78), EDNRB (endothelin receptor type B), EPO (erythropoietin), ERK (extracellular-signal-regulated kinase), FCER1A (Fc fragment of IgE, high affinity I, receptor for alpha polypeptide), GAB1 (GRB2-associated binding protein 1), GCLC (glutamate-cysteine ligase, catalytic subunit), IL17F (interleukin-17F), Jnk (c-Jun N-terminal kinase), SCF (stem cell factor), LEP (leptin), LHB (luteinizing hormone beta polypeptide), Mapk (mitogen-activated protein kinase), NFkB (complex) (nuclear factor of kappa light polypeptide gene enhancer in B-cells), NOX3 (NADPH oxidase 3), P38 MAPK (P38 mitogen-activated protein kinase), PRKAA1 (protein kinase, AMP-activated, alpha 1 catalytic subunit), PRKAA2 (protein kinase, AMP-activated, alpha 2 catalytic subunit), PSMD4 (proteasome (prosome, macropain) 26S subunit, non-ATPase, 4), SMPD2 (sphingomyelin phosphodiesterase 2, neutral membrane (neutral sphingomyelinase)), TFF3 (trefoil factor 3), AGT (angiotensinogen), APOA1 (apolipoprotein AI), COL18A1 (collagen, type XVIII, alpha 1), CXCL11 (chemokine (C-X-C motif) ligand 11), BLC (B lymphocyte chemoattractant), MIG (monokine induced by gamma interferon), Fcgr2 (Fc gamma R2), FDX1 (ferredoxin 1), FDXR (ferredoxin reductase), FSH (follicle stimulation hormone), GCLC (glutamate-cysteine ligase, catalytic subunit), GH (growth hormone), GHR (growth hormone receptor), SGOT (serum glutamic oxaloacetic transaminase), KIM-1 (kidney injury molecule 1), HSD3B2 (hydroxy-delta-5-steroid dehydrogenase, 3 beta- and steroid delta-isomerase 2), IL6 (interleukin-6), ITK (IL2-inducible T-cell kinase), LDLR (low density lipoprotein receptor), LPL (lipoprotein lipase), LSS (lanosterol synthase (2,3-oxidosqualene-lanosterol cyclase)), MSMO1 (methylsterol monooxygenase 1), NLRP12 (NLR family, pyrin domain containing 12), NPPB (brain natriuretic peptide), NR1H4 (nuclear receptor subfamily 1, group H, member 4), OSMR (oncostatin M receptor), PI3K (complex) (phosphoinositide-3-kinase), SCARB1 (scavenger receptor class B, member 1), TBG (thyroxine binding globulin), SST (somatostatin), Stat5a/b (signal transducer and activator of transcription a/b), TCR (T-cell receptor), TNF-alpha (tumor necrosis factor-alpha), TNFRSF11B (tumor necrosis factor receptor superfamily, member 11b), VWF (von Willebrand factor).
Discussion
This is the first large-scale study investigating molecular sex differences in serum of typical individuals. The present findings provide insight into biological pathways with specific differences in male and female participants, significantly extending current knowledge of sex-specific molecular profiles. These naturally occurring differences provide a baseline for comparison against diseases and may help to uncover pathways involved in disease-related sex dimorphisms. A particular strength of the present study was the multiplexed investigation of a large number of molecules covering multiple biological pathways. This allowed a comprehensive assessment of changes in these pathways at the time of sampling and circumvents problems associated with combining results from single molecular assays across studies.
The most significant finding was a higher level of molecules associated with oxidation of fatty acids and hormone function in female participants. It is well known that lipid metabolism differs between males and females, and multiple studies have suggested that these differences are not only a consequence of phenotypical differences such as percentage of body fat, but also related to sex dimorphisms in metabolism itself [35]. In this context, the glycerol rate of appearance, which has been used as an indicator of whole body lipolytic rate, has been found to be higher in women. It has been suggested that the associated increased release of fatty acids can be advantageous under conditions of elevated energy demand, but may also be related to the increased susceptibility of women to develop fatty acid liver disease [35], [36]. It is important to note that participants analyzed in the present study were not matched for percentage of body fat. However, since this is a characteristic that can be expected to differ between males and females in general, it may be an important factor associated with disease or treatment-related sex dimorphisms itself.
The molecules found to be associated with higher fatty acid metabolism in females included adiponectin, leptin and apolipoprotein AI. In contrast, molecules involved in lipid transport, including most of the measured apolipoproteins, showed inconsistent sex differences. To the best of our knowledge, this is the first report showing different effects of sex on lipid metabolism and transport. Elevated lipid metabolic rate coupled with few changes in the levels of many of the apolipoproteins involved in lipid transport in females may reflect intrinsic sex differences in the handling of lipids, including metabolism, transport and storage. For example, a study investigating basal, postabsorptive very low density lipoprotein – triglyceride (VLDL-TG) kinetics identified higher female secretion rates as well as differences in production in clearance [35].
A second important finding was that higher levels of fatty acid metabolism and hormone regulation-related molecules coincided with elevated levels of immune cell chemotaxis proteins in males. This is interesting since molecules related to immune cell growth and activation showed the opposite behaviour and were increased in females.
It has been established that humoral and cell-mediated immune responses are generally more active and robust in females compared to males and that inflammation and production of inflammatory markers show sex differences [1], [37]–[39]. Polymorphonuclear leukocytes (PMNs) in females were found to have vigorous phagocytic responses after anaesthesia, surgery, introduction of lipopolysaccharide, and acute ethanol intoxication [37]. Women also have higher amounts of serum antibodies, CD4+ T cells and CD4/CD8 T cell ratios in blood, along with higher cytokine expression during infection and stronger T cell humoral immune responses [40]. The possibility that such differences could lead to disease-related sex dimorphisms is exemplified by evidence that dysregulation of the IL1 agonist/antagonist system may cause greater severity of chronic fatigue syndrome [41]. Also, females have a poorer prognosis in cases of chronic inflammatory conditions such as cystic fibrosis and chronic pulmonary obstructive disease with greater morbidity and complications [38]. Immune reactivity has also been linked to the higher prevalence of autoimmune diseases in females [1], [41].
Recruitment of immune cells to sites of infection and inflammation is necessary to coordinate immune response and is important in disease, as the type of chemokine produced in a specific condition influences inflammation [42]. Though females generally show higher production of cytokines upon infection, we found that analytes related to immune cell chemotaxis and cell signalling were present at higher levels in males. These analytes included several growth factors, monocyte chemotactic proteins and macrophage inflammatory proteins. Lower levels of these molecules in females are consistent with reported effects of estrogen down-regulating chemokine production. Previous studies have shown that estrogen treatment leads to decreased mRNA transcription of chemoattractant proteins in macrophages and increased production of cytokines from immature dendritic cells [41], [43]. Monocytes and macrophages also showed reduced expression of pro-inflammatory cytokines due to the effects of estrogen on receptor CD16 [43]. Overall, the higher levels of many chemotactic factors found in males in this study may be due to a lack of suppressant effects of estrogen at basal levels. This could have important effects, especially in the initial response to infection, autoimmunity, and diseases with links to specific chemokine effects. For example, autoimmune disease severity in females has been linked to fluctuating estrogen levels and the type of cytokine environment that they induce [40].
Differences in cytokines and inflammatory processes have also been linked to alterations in lipid metabolism, providing further support for the co-occurrence of such differences in the present study. Adiponectin has been found previously to be elevated in females and this sex difference decreased linearly with the stage of diabetes progression. The inflammatory molecules C-reactive protein (CRP) and IL1 receptor antagonist (RA) were also increased in pre-diabetic and diabetic women, and these increased even further as the disease progressed, relative to the findings in males [44]. Adiponectin has also been found to decrease endothelial cell CRP synthesis and secretion [45]. Adiponectin and CRP were both found to be present at higher concentrations in females in our study, and were associated with the fatty acid metabolism analyte cluster. Such links between lipid metabolism and inflammatory markers have interesting implications in sex dimorphisms in metabolic syndrome, cardiovascular disease and other related disorders [46].
To evaluate sex dimorphisms deviating from the variation seen in normal controls, we investigated a cohort of participants with Asperger syndrome, which is characterised by a particularly pronounced sex difference in incidence. This analysis led to identification of an interaction in a clustered group of molecules enriched in fatty acid catabolism and hormone regulation functions, indicating a deviation from normal sex variation that affected an entire system of molecules. Interactions for individual molecules within the cluster have previously been shown in Asperger syndrome in [34]. Investigation into other molecules part of the detected networks, which have not been associated with Asperger syndrome as yet, has potential to help explain the sex-specific effects in this cluster of molecules and its biological implications for this condition. It is worthwhile to note that the cytokine alterations that were previously observed at the individual analyte level in the same cohort were not apparent at the pathway level [34]. This may be due to the separation of inflammatory molecules into multiple distinct sets combined with a lack of change seen in other related molecules in the cluster. These results indicate the importance of this set of related molecules, associated most strongly with fatty acid oxidation and hormone function, in the molecular basis of Asperger syndrome. This is particularly interesting since Asperger syndrome has been associated with higher free testosterone levels in female participants and testosterone is known to be a strong regulator of lipid metabolism [47]–[49].
Conclusions
In summary, we have established that important sex differences exist in human serum using a multiplex immunoassay system. Specifically, we found sex differences in molecules related to metabolism and immune cell function, which may be explored more deeply in the context of disorders associated with sex effects. We have also shown that a cluster of molecules associated with fatty acid metabolism and hormone levels exhibits sex dimorphic differences in a cohort of adults with Asperger syndrome. Such differences indicate that Asperger syndrome, and potentially other autism spectrum conditions, may develop differently in males and females according to sex-specific molecular pathways [21], [50]. These findings suggest that there is considerable scope for further studies of the effect of sex on disease susceptibility and development.
Supporting Information
Table S1.
Clusters of molecules. Assignment of all molecules in cohort 7 using k-means clustering.
https://doi.org/10.1371/journal.pone.0051504.s001
(XLSX)
Acknowledgments
We want to thank Anke Dudeck, Jeanette Schadow, Dr. Wolfgang Jordan, Dr. Bernd Hahndorf, Dr. Florian Kästner, Dr. Anya Pedersen, Dr. Ansgar Siegmund, Dr. Katja Kölkebeck, Torsten Schoenborn, Dr. Christoph W. Gerth, Dr. Christian Mauss, Dr. Brit M. Nolden, Dr. M. A. Neatby and Erin Ingudomnukul for their participation in sample characterization and collection. Thanks to all members of the Bahn Laboratory and to Pietro Liò for discussions, help, and encouragement. Most of all, thanks to all volunteers for their donation of samples used in this study.
Author Contributions
Analyzed the data: JMR ES. Contributed reagents/materials/analysis tools: SB. Wrote the paper: JMR ES PCG. Initiated and conceptualized the study: ES SB. Designed the study: ES SB. Coordinated patient recruitment: NJMVB FML MR BB JS LR SBC.
References
- 1. Pelfrey CM (2001) Sexual dimorphism in autoimmunity?: a focus on Th1/Th2 cytokines and multiple sclerosis. Clin Appl Immunol Rev 1: 331–345.
- 2. Miller VM (2012) Family matters: sexual dimorphism in cardiovascular disease. Lancet 379: 873–875 .
- 3. Czlonkowska A, Ciesielska A, Gromadzka G (2006) Gender Differences in Neurological Disease. Endocrine 29: 243–256.
- 4. Tingen CM, Kim AM, Wu P-H, Woodruff TK (2010) Sex and sensitivity: the continued need for sex-based biomedical research and implementation. Women’s Health 6: 511–516 .
- 5. Verthelyi D (2001) Sex hormones as immunomodulators in health and disease. Int Immunopharmacol 1: 983–993.
- 6. Mahr S, Menard J, Krenn V, Muller B (2003) Sexual dimorphism in the osteoarthritis of STR/ort mice may be linked to articular cytokines. Ann Rheum Dis 62: 1234–1237 .
- 7. Denton K, Baylis C (2007) Physiological and molecular mechanisms governing sexual dimorphism of kidney, cardiac, and vascular function. Am J Physiol Regul Integr Comp Physiol 292: R697–R699 .
- 8. Penaloza C, Estevez B, Orlanski S, Sikorska M, Walker R, et al. (2009) Sex of the cell dictates its response: differential gene expression and sensitivity to cell death inducing stress in male and female cells. FASEB J 23: 1869–1879 .
- 9. Vamvakopoulos NV (1995) Sexual dimorphism of stress response and immune/inflammatory reaction: the corticotropin releasing hormone perspective. Mediators Inflammation 4: 163–174.
- 10. Thangavel C, Shapiro BH (2007) A molecular basis for the sexually dimorphic response to growth hormone. Endocrinology 148: 2894–2903 .
- 11. Thorand B, Baumert J, Kolb H, Meisinger C, Chambless L, et al. (2007) Sex differences in the prediction of type 2 diabetes by inflammatory markers: results from the MONICA/KORA Augsburg case-cohort study, 1984–2002. Diabetes Care 30: 854–860 .
- 12. Beasley LE, Koster A, Newman AB, Javaid MK, Ferrucci L, et al. (2009) Inflammation and race and gender differences in computerized tomography-measured adipose depots. Obesity 17: 1062–1069 .
- 13. Wang M, Zhang W, Crisostomo P, Markel T, Meldrum KK, et al. (2007) Sex differences in endothelial STAT3 mediate sex differences in myocardial inflammation. Am J Physiol Endocrinol Metab 293: E872–E877 .
- 14. Da Silva J, Larbre J, Seed M, Cutolo M, Villaggio B, et al. (1994) Sex differences in inflammation induced cartilage damage in rodents. The influence of sex steroids. J Rheumatol 21: 330–337.
- 15. Delaleu N, Immervoll H, Cornelius J, Jonsson R (2008) Biomarker profiles in serum and saliva of experimental Sjögren’s syndrome: associations with specific autoimmune manifestations. Arthritis Res Ther 10: R22 .
- 16. Umstead TM, Lu C-jung K, Freeman WM, Myers JL, Clark JB, et al. (2012) The kinetics of cardiopulmonary bypass: a dual-platform proteomics study of plasma biomarkers in pediatric patients undergoing cardiopulmonary bypass. Artif Organs 36: E1–E20 .
- 17. Kang K, Schmahl J, Lee J-M, Garcia K, Patil K, et al. (2012) Mouse ghrelin-O-acyltransferase (GOAT) plays a critical role in bile acid reabsorption. FASEB J 26: 259–271 .
- 18. Schwarz E, Guest PC, Rahmoune H, Harris LW, Wang L, et al. (2012) Identification of a biological signature for schizophrenia in serum. Mol Psychiatry 17: 494–502 .
- 19. Bertenshaw GP, Yip P, Seshaiah P, Zhao J, Chen T-H, et al. (2008) Multianalyte profiling of serum antigens and autoimmune and infectious disease molecules to identify biomarkers dysregulated in epithelial ovarian cancer. Cancer Epidemiol Biomarkers Prev 17: 2872–2881 .
- 20. Ghobrial IM, Munshi NC, Harris BN, Shi P, Porter NM, et al. (2011) A phase I safety study of enzastaurin plus bortezomib in the treatment of relapsed or refractory multiple myeloma. Am J Hematol 86: 573–578 .
- 21. Baron-Cohen S, Lombardo MV, Auyeung B, Ashwin E, Chakrabarti B, et al. (2011) Why are autism spectrum conditions more prevalent in males? PLoS Biol 9: e1001081 .
- 22.
Cliff N (1996) Alternatives to mean comparisons. In: Ordinal methods for behavioral data analysis. Mahwah: Lawrence Erlbaum Associates, Inc. 123–157.
- 23. Feng D, Cliff N (2004) Monte Carlo evaluation of ordinal d with improved confidence interval. J Mod App Stat Meth 3: 322–332.
- 24.
Kromrey JD, Hogarty KY, Ferron JM, Hines CV, Hess MR, et al.. (2005) Robustness in meta-analysis: An empirical comparison of point and interval estimates of standardized mean differences and Cliff’s delta. In: Joint Statistical Meetings. Minneapolis, MN.
- 25.
Hess MR, Kromrey JD (2004) Robust confidence intervals for effect sizes: A comparative study of Cohen’s d and Cliff's delta under non-normality and heterogeneous variances. In: Annual Meeting of the American Education Research Association. San Diego, CA.
- 26.
Sutton A (2000) Methods for meta-analysis in medical research. Chichester: John Wiley & Sons Ltd. 317 p.
- 27. McMahon FJ, Akula N, Schulze TG, Muglia P, Tozzi F, et al. (2010) Meta-analysis of genome-wide association data identifies a risk locus for major mood disorders on 3p21.1. Nat Gen 42: 128–131 .
- 28.
Chen DT, Jiang X, Akula N, Shugart YY, Wendland JR, et al.. (2011) Genome-wide association study meta-analysis of European and Asian-ancestry samples identifies three novel loci associated with bipolar disorder. Mol Psychiatry: 1–11. doi:10.1038/mp.2011.157.
- 29. Fabricatore A, Wadden T, Higginbotham A, Faulconbridge L, Nguyen A, et al. (2011) Intentional weight loss and changes in symptoms of depression: a systematic review and meta-analysis. Int J Obes 35: 1363–1376 .
- 30. Wolin K, Yan Y, Colditz G (2011) Physical activity and risk of colon adenoma: a meta-analysis. Br J Cancer 104: 882–885 .
- 31. Benjamini Y, Hochberg Y (1995) Controlling the false discovery rate: A practical and powerful approach to multiple testing. J R Stat Soc Ser B Stat Methodol 57: 289–300.
- 32. Huang D, Sherman B, Lempicki R (2009) Bioinformatics enrichment tools: paths toward the comprehensive functional analysis of large gene lists. Nucleic Acids Res 37: 1–13.
- 33. Huang D, Sherman B, Lempicki R (2009) Systematic and integrative analysis of large gene lists using DAVID Bioinformatics Resources. Nat Protoc 4: 44–57.
- 34. Schwarz E, Guest PC, Rahmoune H, Wang L, Levin Y, et al. (2011) Sex-specific serum biomarker patterns in adults with Asperger’s syndrome. Mol Psychiatry 16: 1213–1220 .
- 35. Mittendorfer B (2005) Sexual dimorphism in human lipid metabolism. J Nutr 135: 681–686.
- 36. McCullough A (2002) Update on nonalcoholic fatty liver disease. J Clin Gastroenterol 34: 255–262.
- 37. Spitzer J (1999) Gender differences in some host defense mechanisms. Lupus 8: 380–383.
- 38. Casimir GJ, Duchateau J (2011) Gender differences in inflammatory processes could explain poorer prognosis for males. J Clin Microbiol 49: 478–479 .
- 39. Grossman C (1989) Possible underlying mechanisms of sexual dimorphism in the immune response, fact and hypothesis. J Steroid Biochem 34: 241–251.
- 40. Nalbandian G, Kovats S (2005) Estrogen, immunity & autoimmune disease. Curr Med Chem - Immun, Endoc & Metab Agents 5: 85–91 .
- 41. Cannon JG, St. Pierre B (1997) Gender differences in host defense mechanisms. J Psychiat Res 31: 99–113.
- 42.
Luster AD, Hospital MG (2001) Chemotaxis?: Role in immune response. Encyclopedia of Life Sciences: 1–12.
- 43. Fish EN (2008) The X-files in immunity: Sex-based differences predispose immune responses. Nat Rev Immunol 8: 737–744.
- 44. Saltevo J, Kautiainen H, Vanhala M (2009) Gender differences in adiponectin and low-grade inflammation among individuals with normal glucose tolerance, prediabetes, and type 2 diabetes. Gender Med 6: 463–470 .
- 45. Devaraj S, Torok N, Dasu MR, Samols D, Jialal I (2008) Adiponectin decreases C-reactive protein synthesis and secretion from endothelial cells: evidence for an adipose tissue-vascular loop. Arterioscler Throm Vas Biol 28: 1368–1374 .
- 46. Regitz-Zagrosek V, Lehmkuhl E, Weickert M (2006) Gender differences in the metabolic syndrome and their role for cardiovascular disease. Clin Res Cardiol 95: 136–147 .
- 47.
Saad F, Gooren LJ (2011) The role of testosterone in the etiology and treatment of obesity, the metabolic syndrome, and diabetes mellitus type 2. J Obes 2011. doi:10.1155/2011/471584.
- 48. Staprans I, Rapp JH, Pan XM, Ong DL, Feingold KR (1990) Testosterone regulates metabolism of plasma chylomicrons in rats. Arterioscler Throm Vas Biol 10: 591–596 .
- 49. von Eckardstein A, Kliesch S, Nieschlag E, Chirazi A, Assmann G, et al. (1997) Suppression of endogenous testosterone in young men increases serum levels of high density lipoprotein subclass lipoprotein A-I and lipoprotein(a)*. J Clin Endocrinol Metab 82: 3367–3372.
- 50. Lai M-C, Lombardo MV, Pasco G, Ruigrok ANV, Wheelwright SJ, et al. (2011) A behavioral comparison of male and female adults with high functioning autism spectrum conditions. PLoS ONE 6: e20835 .