Differences in gene expression that correlate with gender can be dissected into a small set of sex chromosome genes and a larger set of genes that appear to be programmed by androgens, most likely during embryonic development. PBMCs from individuals with diverse well-defined etiologies of DSD including gonadal dysgenesis, defects of androgen biosynthesis, and high levels of androgenic steroids due to 21-hydroxylase deficiency showed strikingly consistent patterns of gene expression across 146 transcripts that correlated most strongly with the genital phenotype. Expression across all 157 transcripts identified as differentially expressed between normal males and females allowed separation of PBMCs into four categories of "PBMC-sexes": chromosomal and phenotypical male, chromosomal and phenotypical female and two intersex constellations in which the chromosomal and steroid influenced transcripts were discordant. While it is possible that transcript profiling of a higher number of DSD-patients could lead to demarcation of even diagnosis-specific patient clusters, the consistency of the transcript profiles across the different diagnoses argues strongly that these four categories are relatively robust.
Animal studies on rodents underlined the importance of sex specificity of the pattern of GH secretion and GH signaling involving STAT5b for maintenance of a large part of sexual dimorphism of gene expression in the liver . Since human PBMC express the GH receptor , this mechanism could have a potential influence on the genes detected in the present study. However, in contrast to the extensive sex specific differences in rats with males showing pulsatile GH secretion and females showing more continuous GH levels, data on sex specificity of GH secretion in humans is less clear. While age and puberty status but not sex are important determinants for GH secretion during growth and puberty , there is other data showing that GH pulse amplitude but not pulse frequency differ between women and men in adulthood . However, looking at the data of the current study, the correlation between gene expression profiles and genital masculinization was independent of the individuals' sex hormone levels at the time of expression profiling. Puberty status, presence of gonads and sex hormone replacement did not influence expression across the sex-specific transcripts. Additional GSEA on transcription factor targets gene sets did not reveal sex specific enrichment of STAT5b target genes in our study (data not shown). Therefore, we have no experimental evidence that sex specific differences of GH secretion or changes of GH secretion due to age or puberty could have influenced the sex specific gene set of the PBMC of our study. However, it cannot be excluded that a considerably larger set of investigated patients would allow the identification of some sex-dimorphic genes influenced by the GH signaling pathway in PBMCs.
Genital masculinization is a direct consequence of early embryonic androgen action between the 7th and 12th week of gestation. Our findings therefore support the existence of lasting programming of sex-specific genes implemented by presence or absence of androgen during the first trimester of embryogenesis. Sex-specific gene expression patterns have been observed in other tissues [8–11, 46–48], yet the relative contribution of the sex chromosomes and androgen programming was unknown. Based on expression patterns readily apparent in PBMCs, cells not usually thought to differ significantly between males and females, our data suggest that many of the sex-specific differences in gene expression are due to the influence of presence or absence of androgens during embryonic development. Given the limited lifespan of PBMCs, the lasting changes in transcript programs due to androgens are likely embedded in progenitor stem cells of fetal leukopoesis. It cannot be excluded that some of the genes detected in our study also played functional roles in androgen programming of lasting PBMC expression patterns during embryogenesis. However, more research is needed to understand the molecular mechanisms of long term hormonal programming, e.g., by studying the effects of androgens on embryonic stem cell differentiation . Whether such programming affects the stem cells of other tissue types is unknown, but highly likely.
In a previous genome-wide assessment of transcript profiles, we have identified transcripts differentially expressed in cultured genital fibroblasts from normal males and 46,XY females with androgen insensitivity syndrome due to inactivating androgen receptor mutations . Similar to the current study we identified a set of transcripts with reproducible differences in gene expression that correlated with the degree of genital masculinization. Since subjects in those studies all had a 46,XY karyotype, the differentially expressed transcripts were almost certainly programmed during development and remained apparent despite serial passage of the fibroblasts in culture. Interestingly, the overlap was restricted to 7 genes. In both datasets FZD6 (Frizzled 6) (Figure 1G), SSX2IP, SDC1, ALKBH, SPRED1 (Sprouty-related, EVH1 domain containing 1), PYCR2 (Pyrroline-5-carboxylate reductase family, member 2) and TOMM40 (Translocase of outer mitochondrial membrane 40) were differentially expressed between the subjects with male and female genitalia. Recent work by Yang and coworkers  in which 334 male and female mice were profiled confirms our findings that a significant number of genes differ between males and females. More importantly, as we observed in genital fibroblasts and PBMCs, the degree of overlap of the sex-specific genes between tissues was limited in this study. Our work extends on that of Yang et al by demonstrating that sex steroid induced programming accounts for the majority of the differences in gene expression between phenotypic males and females. Taken together these data suggest that the sex chromosomes to a small degree and androgen signaling during development to a larger degree account for a large part of the differences in gene expression in different tissue types between males and females. Furthermore, the influences of each of these differ significantly between tissue types as evidenced by the limited overlap in the sex-specific transcripts between tissue types. Additional work will be necessary to test whether these differences in gene expression between analogous tissues between males and females underlie developmental, anatomical, and functional differences between the sexes.
Our data also provide a framework for investigating differences in diseases between the sexes and could have future implications for risk assessment and disease monitoring . PBMC expression profiling has been suggested as a diagnostic tool in that germline mutations in disease-causing genes could produce unique transcriptional signatures detectable in blood cells [50–52]. Disease-linked expression profiles are likely to be modified by biological sex [8, 9] and might need to be accounted for when developing diagnostic tools. In addition, our data raise the possibility that lasting transcriptional programs that differ between males and females could underlie differential susceptibility to a variety of diseases. For instance, we observed a striking over-representation of genes involved in growth control and cancer in PBMCs between males and females. Whether sex-specific gene expression could affect a tissue's transcriptional background and oncogenic potential or response to anti-cancer therapy is unknown, but merits consideration in light of differences in cancer incidence between the sexes and our new findings.
Our data have relevant implications for understanding the biology of DSD as well as potential future applications in its management. In Western societies, every schoolchild learns that "sex chromosomes" decide "who's a boy and who's a girl". Our data broaden the view of what determines sex and gender by revealing distinct male and female transcript profiles that correlate with either the karyotype or the androgen milieu present during embryonic development. Therefore, our data confirm the concept of long term androgen programming as established in rodent models [12–14] for the first time on the level of the human PBMC-transcriptome. Our data also reveal 2 categories of gene expression in PBMCs in individuals with DSD in which the karyotypic and androgen-programmed transcript profiles are discordant. Based on our findings, it is possible that other transcript profiles in PBMC could be identified that correlate with sex or gender specific traits outside the genitalia. Were that the case, it might be possible to specify transcript signatures that reflect sex-specific differences in tissues such as the brain and in turn mirror sex-specific behavior, sexual orientation or gender identity. If so, subsets of the sex specific genes of our study could serve as future transcriptional biomarkers in DSD outcome studies in order to develop novel diagnostic tools contributing to decisions on gender assignment in DSD children.