Complex disease, gender and epigenetics

Abstract

Gender differences in susceptibility to complex disease such as asthma, diabetes, lupus, autism and major depression, among numerous other disorders, represent one of the hallmarks of non‐Mendelian biology. It has been generally accepted that endocrinological differences are involved in the sexual dimorphism of complex disease; however, specific molecular mechanisms of such hormonal effects have not been elucidated yet. This paper will review evidence that sex hormone action may be mediated via gene‐specific epigenetic modifications of DNA and histones. The epigenetic modifications can explain sex effects at DNA sequence polymorphisms and haplotypes identified in gender‐stratified genetic linkage and association studies. Hormone‐induced DNA methylation and histone modification changes at specific gene regulatory regions may increase or reduce the risk of a disease. The epigenetic interpretation of sexual dimorphism fits well into the epigenetic theory of complex disease, which argues for the primary pathogenic role of inherited and/or acquired epigenetic misregulation rather than DNA sequence variation. The new experimental strategies, especially the high throughput microarray‐based epigenetic profiling, can be used for testing the epigenetic hypothesis of gender effects in complex diseases.

• Chromatin
• DNA methylation
• epigenetics
• epigenomic profiling
• histones
• hormone
• microarrays
• sexual dimorphism

Simple and complex genetic diseases

In the past 20 years, there has been a major effort into the development of experimental and computational strategies for the identification of molecular causes of human complex disease. Unlike simple Mendelian disorders, for which the cloning of disease genes has become a routine procedure, identification of the molecular genetic basis of complex disease represents a major challenge to human biologists. Although the genes for some rare early onset and familial cases of complex diseases (e.g. colon cancer, breast cancer and Alzheimer's disease) have been identified, the overwhelming proportion of non‐Mendelian pathology remains unexplained. Research in complex non‐Mendelian diseases is reaching an unprecedented level of sophistication and scale: from building massive databases of DNA sequence variants (single nucleotide polymorphisms (SNPs), haplotype maps (HapMap)) to screening thousands of polymorphisms in thousands of affected individuals and controls. Why is it that complex diseases are so resistant to the discovery of the etiological factors? In an attempt to address this question, it is important to note that complex diseases, in comparison to simple ones, exhibit numerous epidemiological, clinical and molecular differences. Complex diseases as a rule are common (more than 1 case per 1000 individuals), with sporadic cases dominating over the familial ones, while simple ones are rare (less than 0.1%) and predominantly familial. Phenotypic differences (discordance) in identical twins have been one of the hallmarks of complex non‐Mendelian disease. Concordance of monozygotic twins reaches only ∼15% in breast cancer, 20% in ulcerative colitis, 25%–30% in multiple sclerosis, 25%–45% in diabetes, 50% in schizophrenia and 40%–70% for Alzheimer's disease 1. Complex diseases usually exhibit a significantly later age at onset, while the overwhelming majority (∼90%) of Mendelian diseases have an onset before puberty and only 1% after 40 years of age 2. Major psychosis, asthma, multiple sclerosis, inflammatory bowel disease, rheumatoid arthritis and numerous other complex diseases exhibit major fluctuations in disease severity, and in some cases age‐dependent decline of symptoms and even full recovery from a disease. Epidemiological studies have revealed that quite often the risk of developing complex disease in offspring depends on the sex of affected parent. For example, asthma, bipolar disorder and epilepsy are more often transmitted from the affected mother, while type 1 diabetes seems to be more often transmitted from the affected father. Finally, one of the most common and well known non‐Mendelian features of complex disease is differential susceptibility to a disease in males and females, or sexual dimorphism 2. This article will review epidemiological and genetic findings of gender effects in complex disease. The other objectives are: 1) to introduce the concept of epigenetics and demonstrate that gender effects are consistent with the epigenetic interpretation of complex disease; and 2) to describe experimental strategies for the identification of the molecular substrate of gender effects, which may contribute to uncovering the fundamental molecular mechanisms of complex disease.

Key messages

• Sexual dimorphism in complex diseases presents with an uneven disease frequency in men and women: although both genders can be affected, one is more susceptible.

• It can be hypothesized that differential susceptibility to complex disease in males and females is mediated by sex hormone‐induced differences in the epigenetic regulation of genes.

• The epidemiological and molecular evidence of gender effects in various complex diseases warrants dedicated molecular studies that would uncover underlying epigenetic mechanisms and genomic sites that are the primary targets of sex hormone action.

Gender and complex disease: epidemiological and molecular genetic findings

Sexual dimorphism in complex diseases presents with an uneven disease frequency in men and women: although both genders can be affected, one is more susceptible. Multiple sclerosis, rheumatoid arthritis, Crohn's disease, panic disorder, structural heart disease and hyperthyroidism are more common in females, while males are more often affected with autism, Hirschsprung's disease, ulcerative colitis, Parkinson's disease, alcoholism, allergies and asthma (especially at young age) 3 (Figure 1). In psychiatric diseases such as Alzheimer's disease, schizophrenia, alcoholism, and mood and anxiety disorders, psychopathology exhibits a number of differences between the sexes in rates of illness as well as the course of illness 4. It is important to note that sex effects in complex diseases should be differentiated from those in single‐gene disorders where sex chromosome‐linked genes are known to be the cause of sexual dimorphism. In simple Mendelian disorders sexual dimorphism is clear‐cut and usually affects only one of two sexes. For example, only males (excluding some rare cases of e.g. skewed X inactivation) are affected with hemophilia and Duchenne's muscular dystrophy, but only females can have Rett's syndrome (again, with some rare exceptions).

Figure 1 Complex non‐Mendelian diseases displaying differential susceptibility to complex diseases (based on epidemiological data obtained from Harrison's ‘Principles of Internal Medicine’ 15th edition, 2001) 111.

Sex hormones have been the usual ‘culprit’ in the explanation of gender effects in various morphological and physiological differences between males and females 5. Differences in endocrine mediated development and maintenance throughout the lives of males and females are the second factor next to the sex chromosomes that define the sexes and are likely to drive a dynamic divergence of epigenetic patterns between the sexes. From brain studies it is known that differential effects of androgens and estrogens contribute to neural development, affecting programmed cell death, cellular migration, synaptogenesis and axonal migration, and formation of sexually distinct neuronal circuits 6, 7. Other studies have shown that androgens are directly related to neurite arborization, while estrogens are responsible for synapse formation and initiation of cellular communication 8. Differences in the circulating hormone concentrations between sexes orchestrate the divergence of behavior and cognitive development in males through two related processes known as masculization and defeminization mediated by the estrogen receptor α and β, respectively 9. Additionally, gonadal hormones are implicated in sex differences observed in neural protection to trauma or ischemic brain injury, with clinical reports showing a higher recovery rate in females than in males, and progesterone and estrogen treatments demonstrating a neuroprotective effect 5. Numerous other disease‐associated traits are differentially affected by the sex hormones. For example, in obesity, which is a risk factor for coronary artery disease, diabetes, osteoarthritis, and some metabolic disorders 10, 11, hormones have been implicated in the control of adipose tissue proliferation and estrogen and testosterone replacement therapy in older women and men resulting in a reduction of obesity 11. Estrogen replacement also attenuates Alzheimer's disease (AD) symptoms in postmenopausal women 12. Additionally, AD‐affected men have significantly lower levels of testosterone in the brain compared to controls 13. Animal models of disease offer a valuable resource for studying the hormone‐associated aspects of susceptibility to complex disease. Female nonobese diabetic (NOD) mice are more often affected by type I diabetes than the male mice of the same strain 14. Neonatal sex hormone manipulation demonstrated that disease phenotype in castrated males and ovariectomized females increased and decreased, respectively 15.

It is interesting to note that gender effects have also been detected in genetic linkage and association studies. There is now an increasing list of molecular genetic findings indicating that genetic markers on autosomes also can be gender dependent (Table I). For example, genetic linkage studies in autism revealed that 17q11 at markers D17S1294 and D17S798 reached genome‐wide significance with the linkage stratification on the basis of gender (P = 0.008) 16. Another group performed a search for susceptibility loci stratified for the sex of the affected proband using a number of markers on chromosomes 2q, 7q, 9p, 15q and 16p 17. Male‐female and female‐female pairs showed evidence for linkage to chromosome 15q markers (D15S117–D15S125, P = 0.0011), while male pairs only demonstrated significance on 16p (D16S407–D16S497, P = 0.026) with a trend towards significance observed on 7q (D7S480–D7S530, P = 0.075) 17. In major depression studies, a genome‐wide linkage study in a sample of 110 Utah pedigrees with a history of major depression identified significant logarithm of odd scores (LOD) signifying linkage at 12q22‐q23.2 markers specific only to males (D12S1300, LOD = 4.6, P = 0.00003; D12S1706 LOD = 6.1, P = 0.0000007) 18. Another genome scan revealed a number of putative loci of linkage to major depression females only, with the most prominent linkage on 2q33‐q35 displaying LOD scores of 6.3 and 6.9 at D2S2321 and D2S2208, respectively 19. A haplotype analysis was performed on the TRAX/DISC1 region previously associated with schizophrenia in a Scottish population 20. A case‐control association study identified significant association of DISC1 with bipolar disorder in women (P = 0.00026) 20. Another study investigating 28 SNPs including TRAX, DISC1, and DISC2 in a Finnish cohort of 458 individuals determined a significant undertransmission of the HEP3 haplotype to affected females (P = 0.00024), suggesting that this haplotype may confer a protective effect against the disease in women 21. Finally, a recent quantitative trait loci [QTL] analysis of various other complex diseases found that of the 17 QTLs investigated, 11 were significantly sexually dimorphic (P<0.05) 22. Among the sexually dimorphic traits were triglycerides, high‐density lipoprotein cholesterol, forced expiratory volume, all anthropometric traits and cortisol levels 22.

Table I. Some loci and markers exhibiting gender‐specific evidence for association (A) and linkage (B) in complex diseases and traits. Not all significant findings from each reference are reported. A. Association studies:

Disease or trait Gene	Polymorphism (allele, genotype, haplotype)	Sex	Statistic	Reference
Age‐related maculopathy
DRD2	141C Ins/Del	Female	P<0.05	112
DRD3	MscI Gly/Gly	Female	P<0.05	112
Alzheimer's disease
TFAM	rs1937 G/G	Female	P<0.05	113
COMT	RS4680, RS4680, RS737865	Female	P<0.05	114
Asthma
Cox‐2	765 C/C	Female	P = 0.01	115
Atopy
Angiotensinogen gene	235 Met/Thr	Female	P = 0.04	116
Angiotensinogen gene	‐6 A/G	Female	P = 0.016	116
Autism
SLC6A4	rs140700	Male	P<0.05	117
Behcet's disease (uveitis)
CCL2	−2518 GA/AA	Female	P = 0.02	118
CCR5	‐403 AA	Male	P = 0.005	118
CCL2	−2076 AA/AA or AA/AT	Male	P = 0.02	118
Bipolar disorder
GPR78	rs1282	Female	P = 0.038	119
MTHFR	677 CT and TT	Female	P = 0.023	120
Hepatic Lipase (HL)	250 G/A	Male	P = 0.025	121
CHD metabolic syndrome
LRPAP	LRPAP1_1b	Female	P = 0.0003	122
PRCP	PRCP_1	Male	P = 0.039	122
LDLR	LDLR_3b	Male	P = 0.046	122
Multiple sclerosis
IFNG	325 A	Male	P = 0.044	123
IFNG	761C/A	Male	P = 0.050	123
Obsessive compulsive disorder
DRD2 with COMT	A2A2 with NlaIII	Male	P = 0.035	124
Osteoporosis
MTHFR	677 TT	Female	P<0.05	125
APOE	Haplotype CGTC	Male	P = 0.001	126
APOE	Haplotype GGTT	Male	P = 0.002	126
APOE	Haplotype GATC	Male	P = 0.006	126
Parkinson's disease
DJ1	rs161799	Female	P = 0.03	127
DJ1	rs226253	Female	P = 0.002	127
Schizophrenia
DISC1	Hep3 Haplotype	Female	P = 0.00024	21
GPR78	rs1282	Female	P = 0.015	119
PLP1	rs475827	Male	P = 0.0294	128
Type I diabetes
AR	CAG Repeat	Female	P = 0.03	129
Longevity
APOE	Allele E2, E3, E4	Female	P<0.03	130
APOC‐I	Hpa I RFLP	Female	P<0.05	130
ACE	Homozygous ALU insertion	Female	P<0.05	130
Myocardial infarction
FXIII‐A	Leu34	Male	P<0.001	131
ERα	397 C/C	Male	P = 0.003	132
CCR2	64I Val/Ile	Female	P<0.01	133
Whole blood serotonin levels
Chromosome 10	D10S1677	Female	P = 0.0005	134
Chromosome 17	ITGB3_33	Male	P = 0.0003	134

B. Linkage studies:

Disease or trait Genomic Location	Marker	Sex	Statistic	Reference
Autism
15q	D15S117‐D15S125	Female	MLS = 2.62	17
17q11	D17S1294‐D17S798	Male	P = 0.008	16
7q	D7S480‐D7S530	Male	MLS = 2.55	17
16p	D16S407‐D16S497	Male	MLS = 2.48	17
Major depression
3centr	D3S2323	Female	LOD = 2.9	135
18q	D18S1270	Female	LOD = 3.5	135
2q35	D2S2208	Female	LOD = 6.87	19
2q33	D2S2321	Female	LOD = 6.3	19
1p36	D1S450	Female	LOD = 3.03	136
4q	D4S2631	Male	LOD = 2.6 (P = 0.00027)	135
15q	D15S1515	Male	LOD = 2.88 (P = 0.0001)	135
Early onset major depressive disease
12q22‐q23.2	D12S1300	Male	P = 0.00003	18
Osteoporosis
18p11	D18S53‐D18S478	Female	LOD = 2.83	137
20q13	D20S196‐D20S173	Female	LOD = 3.2	137
2p	194 cM	Male	LOD = 3.97	138
4q	90 cM	Male	LOD = 4.16	138
10q21	D10S196‐D10S537	Male	LOD = 4.42	137
Bone mineral density
3p22	76 cM	Female	LOD = 2.72	139
13q14	D13S788	Male	LOD = 3.46	140
Cortisol
10q	97 cM	Female	P = 0.024	22
Eosinophilia
10q	77 cM	Female	P = 0.0041	22
Serotonin
17q	70 cM	Female	P = 0.0049	22
Forced expiratory volume at 1 s/forced vital capacity
5q	154 cM	Male	P = 0.022	22
7p	55 cM	Male	P = 0.0025	22
High‐density lipoprotein cholesterol
9q	117 cM	Male	P = 0.0084	22
Systolic blood pressure
8q	109cM	Male	P = 0.0007	22
Lymphocyte count
16q	129 cM	Male	P = 0.029	22
% Lymphocyte count per white blood cell count
19p	43 cM	Male	P = 0.027	22
Triglycerides
21q	30 cM	Male	P = 0.048	22
Whole blood serotonin levels
6q	183 cM	Female	P<0.002	134
16p	56 cM	Male	P<0.002	134

While sex hormones have been the usual ‘culprit’ in explanations of gender effects in complex diseases based on the myriad of data associating hormonal differences with disease states and their critical involvement in human biology 5, there are no underlying mechanisms proposed as to how such hormones predispose to or protect from a disease relating to the specific molecular mechanisms of hormone action. The gender‐specific effects in genetic linkage and association studies suggest that chromosomes and individual genes can be the target of sex hormones. While such hormones cannot change DNA sequence, they can be potent modifiers of epigenetic status, which controls gene expression and various other genomic activities. It is known that hormones, including sex hormones, can control gene expression via epigenetic modifications 2. Therefore, it can be hypothesized that differential susceptibility to complex disease in males and females is mediated by sex hormone‐ induced differences in epigenetic regulation of genes. In the next section, a brief introduction to epigenetics is provided.

Epigenetics and its relevance to complex disease

Epigenetics refers to regulation of various genomic functions that are brought about by heritable, but potentially reversible changes in DNA modification and chromatin structure 23. The epigenetic information is encoded in two types of synergistically acting covalent modifications: DNA methylation, more specifically methylated cytosines (^metC), and chromatin proteins (modifications of chromatin proteins such as histone acetylation, methylation, phosphorylation) 24. In mammals, DNA methylation occurs most commonly where cytosine is directly followed by guanine, forming what is known as a CpG dinucleotide. Clusters of CpG dinucleotides are referred to as CpG islands 25. The epigenetic complexity conferred is through DNA and chromatin modifications that can directly affect the regulation of gene activity 26–32, and other important genomic functions 33, 34 including genetic recombination 35 and DNA mutability 36. Maintenance of existing DNA methylation and de novo DNA methylation is catalyzed by several types of enzymes known as DNA‐methyltransferases 37. The importance of proper maintenance of methylation throughout the genome is highlighted by the observation that a complete loss of maintenance DNA‐methyltranferase function results in death of mice in early embryogenesis 38. A large number of genes exhibit an inverse correlation between the degree of DNA methylation and the magnitude of gene expression 39, 40. Transcription factor binding affinity may be limited by the presence of methylation at the binding sites 28, 31. In addition to the ‘critical site’ effects of ^metC, the density of ^metC in a gene regulatory region also contributes to gene activity. DNA modification acts in concert with histone modifications, another epigenetic mechanism. Alterations in chromatin structure occur through acetylation, methylation and phosphorylation of various histone amino acid residues including lysine, arginine and serine 41–45. The information within the patterns of such modifications is sometimes referred to as the ‘histone code’ 24. Epigenetic modifications regulate genomic functioning not only in terms of gene expression but also in the suppression of repetitive DNA sequences 46 and the formation of architecturally functional chromatin structures such as centromeric regions 47. There has been significant progress in understanding the complementary functioning of epigenetic mechanisms, i.e. how changes in DNA modification affect chromatin conformation and vice versa 48–50. It is important to note that some studies have demonstrated that genetic differences such as single nucleotide polymorphisms (SNPs) can be predictors of differentially methylated alleles 51–53 allowing for a reinterpretation of the potential importance of some seemingly functionally irrelevant SNPs associated with various complex diseases. Sex effects linked to or associated with specific SNPs or haplotypes can be explained by hormone‐induced modifications such that a specific allele or haplotype only becomes a risk factor after some endocrinologically mediated epigenetic modification.

There is increasing evidence that putative epigenetic misregulation of genes may explain numerous epidemiological, clinical, and molecular complexities of complex disease 2. It has been argued that the heuristic value of the epigenetic model of complex disease lies in the possibility of integrating a variety of unrelated data into a new theoretical framework, providing the basis for new experimental approaches 54–58. Shifting the emphasis from DNA sequence variation and a hazardous environment as the main etiological factors to putative epigenetic misregulation provides a new perspective on the myriad of the above listed non‐Mendelian features (prevalence of sporadic cases, discordance of identical twins, parental origin effects, late age of disease onset and fluctuating course, among others) that cannot be explained by more traditional mechanisms 59. In the epigenetic model of complex non‐Mendelian disease, pathology is seen to result from a chain of unfavorable epigenetic events that begin with a primary epigenetic defect, or preepimutation, occurring in the germline during the error‐prone epigenetic‐reprogramming process 60, 61. Preepimutation increases the risk of developing disease, but unlike the deterministic DNA mutations in Mendelian disorders, a preepimutation does not necessarily indicate that the disease is inevitable. Such preepimutations may not cause any immediate clinical problems, and thus the age of onset may be delayed for a relatively long time. It may take decades until the epigenetic misregulation reaches a critical threshold beyond which the cell is no longer able to function normally. The phenotypic outcome depends on the overall effect of a series of pre‐ and postnatal impacts on such a preepimutation. Preepimutations are subject to further changes by the multidirectional effects of tissue differentiation, stochastic events, some external environmental factors (nutrition, medications, etc.) 62–64, and also the hormonal milieu. It is known that various hormones, including sex hormones, have a significant impact on gene expression, and this is achieved by changing chromatin conformation 65–68 and/or local pattern of gene methylation 69, 70. Some examples of how sex hormones can influence epigenetic regulation of genes are provided below.

Epigenetics and sex hormones

A number of earlier studies have demonstrated the direct effects of sex hormone administration on epigenetic states. An example is the effect of estradiol administration on the methylation status of various CpG dinucleotides located in an estradiol‐mediated regulatory region for the avian vitellogenin II gene 69, 71–73. Two of these methylatable cytosines located within the estradiol‐receptor binding site are actively demethylated in response to estradiol treatment in hormone responsive tissues in a strand‐specific manner 69. This DNA demethylation persists after transcription has ended 71, 73 and has been suggested to result in a ‘memory effect’ resulting in a quicker induction of vitellogenin II mRNA synthesis in response to subsequent estradiol stimulations 69. This finding highlights the fact that endocrinological influence can mediate long‐lasting epigenetic change to levels of gene transcription.

One of the mechanisms of sex hormone action is through the molecular epigenetic signatures of particular chromosomal regions that modulate the access of transcription factors to the transcribed sequences. If DNA is tightly packed around the histone proteins, the access of transcription factors to their respective transcription factor binding sites is restricted 24, 74–76. Such compacted chromatin is supported by high density of ^metC in the DNA sequence and various types of suppressing modifications of amino‐terminal residues of histone proteins, e.g. deacetylation 75. Conversely, low density of ^metC in the DNA sequence and high levels of histone acetylation unravels the chromatin, allowing for access of DNA binding elements and transcriptional activation 24, 74–76.

Chromatin modification can be directly affected by members of the nuclear hormone receptor (NHR) superfamily, and the steroid receptor (SR) subset of NHR is of particular interest as it contains the androgen receptor (AR) and estrogen receptor (ER) 77. These sex hormone receptors can have activating or repressing transcriptional activity dependent on the presence or absence of ligand, respectively 78, 79. Steroid hormone mediated transcriptional activation or repression results from the SR recruitment of coactivator and corepressor complexes—protein complexes which associate with various epigenetic modifiers such as histone deacetylases (HDAC), histone acetyltransferases (HAT) and histone methyltransferases (HMT) 78, 80. It is these ‘coregulatory complexes’ that achieve the epigenetic modification necessary for chromatin remodeling, allowing or restricting access of transcription factors and RNA polymerase II and thus mediating the epigenetic effects of the sex hormones (Figure 2).

Figure 2 In the absence of ligand or in the presence of antagonist, steroid receptors(SR) recruit transcriptional corepressor complexes to the DNA, producing an epigenetically silenced state through histone deacetylation or methylation. In the presence of hormone, SR recruit coactivator complexes, histone‐modifying machines or ATP‐dependent chromatin remodeling machines, to acetylate histones and produce a euchromatic chromatin formation suitable for transcriptional activation.

There are two primary types of coregulatory complexes, the first being the adenosine triphosphate (ATP)‐dependant chromatin remodeling complexes (Figure 2), such as the switch/sucrose nonfermenting (SWI/SNF) complex 77. The ATP‐dependent chromatin remodeling complexes may be primarily responsible for opening of chromatin 81, 82, formation of nucleosome arrays, homologous strand pairing and DNA transcription 83. Only recently, the actions of SWI/SNF have been linked to changes in DNA methylation through a mechanism mediated by interactions with methyl CpG binding protein 2 (MeCP2) 84. SWI/SNF interacts with both the ER and AR, and mutations in a core subunit have been implicated in the development of ovarian cancer 85.

Histone‐modifying complexes constitute the second type of coregulatory complex that interact with the sex hormone receptors. Coactivators include the cyclic adenosine monophosphate (cAMP) response element‐binding protein (CBP)/p300, while the most widely studied corepressor complexes are histone‐modifying machines such as the nuclear receptor corepressor complex (NCoR) and the silencing mediator of retinoic acid and thyroid hormone receptor (SMRT) 77 (Figure 2). Histone modifications mediated by histone‐modifying machines unravel and compact chromatin through acetylation and deacetylation of lysine residues on histone H4, respectively, and alternatively through methylation of histone lysine and arginine residues 74, 86. In addition to receptor agonist binding, antagonists to the progesterone receptor (PR) and androgen receptor (AR) elicit an interaction with NCoR and SMRT, demonstrating that antagonistic actions are mediated through corepressor recruitment 77.

There are numerous molecular mechanisms through which AR and ER coactivator/corepressor‐binding modulate chromatin structure and influence epigenetic marks; however, the distribution and expression of the genes encoding the coregulatory factors have been shown to vary between the sexes. Gender differences in the expression of steroid receptor co‐activator‐1 (SCR‐1), CBP/p300, NCoR, and SMRT were observed in various tissues in male and female rats 87. Furthermore, regulation of transcript levels can be influenced by the differences in circulating hormones between males and females, as estradiol differentially influenced SCR‐1 and SMRT transcript levels in the anterior pituitary of male and female rats 87. Therefore, despite being ubiquitously expressed, the gender‐ and tissue‐specific coregulator transcript differences have been suggested to result in variability in the level of response to hormone 87, a finding that could explain observations of tissue‐specific differences in ER‐mediated response to ligand.

AR and ER differentially interact with coregulatory complexes through binding to different types of α‐helix LXXLL motifs on the complexes 88, 89. Through these interactions, recent studies have suggested that different types of ligand binding to the sex hormone receptors or conformational changes over time may influence the ability to recruit various cofactors, as in the case of AR and ER where changes in coactivator‐binding have been linked to prostate and breast cancer development, respectively 90, 91. In a similar way, differences in AR and ER interactions with coactivator LXXLL motifs 88 may result in the recruitment of different epigenetic modifying coregulatory complexes between the sexes.

Finally, various studies have demonstrated that tissue‐specific distribution of the sex hormone receptors varies between the sexes 92 as do the target genes for transcriptional regulation, and thus the epigenetic modifications mediated through the sex hormone receptors will likely vary by gene and tissue between the sexes. A study investigating the transcriptional regulation of the prolactin gene tested the effects of estradiol administration in GH4 cells 93. Estradiol was found to have a stimulating effect on histone H4 acetylation in the promoter region of the prolactin gene 93. In agreement with other studies, such histone modifications are implicated as necessary factors in gene transcription via the formation of a euchromatic state of the promoter region 94. Various forms of histone methylation are associated with estradiol binding to the estrogen receptor (ER), affecting the transcription at the target promoter in vitro77, 95 and further highlighting that sex hormone‐regulated gene transcription is associated with the formation of an epigenetic mark. In rat, estrogen treatment was shown to increase DNA methylation on the prolactin gene in pituitary and liver tissues with subsequent reduction in mRNA amounts, suggesting a direct effect on transcription in a tissue‐specific manner 96. Prolactin administration caused demethylation of DNA in rat liver and kidney in both mature and immature rats 97. Prenatal exposure to diethylstilbestrol, a synthetic form of estrogen that has been implicated in development of cancers in humans and mice in later life, permanently alters the DNA methylation pattern of the oncogene, c‐fos, in mice, implicating an epigenetic mechanism for hormone‐induced carcinogenesis 98. Testosterone and estradiol administration have been shown to affect amyloid precursor protein (APP) levels in AD model mice 99. DNA methylation levels at the APP promoter exhibit a sex difference in mice and are generally higher in females 99. Increased levels of estrogens in transgenic breast cancer model mice lead to epigenetic modifications in cancer relevant genes involved in cell cycle, cell proliferation and apoptosis 100. Finally, investigation of the effects of estrogen on Japanese medaka fish demonstrated that DNA methylation varied between tissues in a gender‐specific manner 101.

Molecular strategies for identification of gender specific epigenetic differences

The epidemiological and molecular evidence of gender effects in various complex diseases warrant dedicated molecular studies that would uncover underlying epigenetic mechanisms and genomic sites that are the primary targets of sex hormone action. As mentioned above, evidence for linkage or association of a particular genomic region could be a marker for differential epigenetic modification of that region, therefore transforming those loci in Table I into candidate targets for gender‐specific epigenetic evaluation. Site‐specific methods for the epigenetic investigation of candidate loci exist, such as in sodium bisulfite modification‐based mapping of methylated cytosines, or chromatin immunoprecipitation assays for histone modification studies. Sodium bisulfite modification is considered the gold standard in DNA methylation measurement and functions through selectively deaminating unmethylated cytosines to uracil while methylated cytosines remain unchanged. After polymerase chain reaction (PCR) amplification, the relative proportions of DNA methylation can be determined at each CpG dinucleotide through a variety of methods including sequencing of clones 102, 103, pyrosequencing 80, MALDI mass spectrometry 104, 105, and SNaPshot 106, amongst others. Chromatin immunoprecipitation is performed with antibodies for modified histone residues, such as H4‐K9 methylation for example, followed by real time PCR amplification of a candidate region in order to quantify levels of histone protein modification 107. Use of these methods would allow for the locus‐specific determination of epigenetic differences in populations stratified for sex; however, while these candidate loci may represent targets of epigenetic modification through the actions of sex hormones, the overwhelming majority of sex hormone targets in the genome remain unknown. For this reason, a comprehensive epigenetic interrogation of the entire genome would be optimal for identifying the yet unknown targets of sex hormone‐mediated epigenetic modification that may be associated with sex effects in disease.

The advent of microarray technologies that interrogate a large number of DNA/RNA fragments in a highly parallel fashion has also opened new opportunities for epigenetic studies 55, 57, 58, 108. The microarray‐based large‐scale DNA modification analysis is based on interrogation of the unmethylated (or hypermethylated) fraction of genomic DNA on the microarray containing oligonucleotides that represent the genomic regions of interest. Enrichment of the unmethylated fraction of genomic DNA is performed through a number of steps that include digestion with methylation‐sensitive restriction enzymes. Such enzymes cut the unmethylated fraction of genomic DNA into short fragments, whereas the methylated DNA sequences remain undigested. In the next step, the restriction enzyme‐treated DNA is ligated to DNA adaptors and subjected to polymerase chain reaction (PCR). PCR primers complementary to the adaptor sequences preferentially amplify the shorter DNA fragments, namely the unmethylated DNA fraction. Fluorescently labeled amplicons are then hybridized to the microarrays. The lower the degree of DNA methylation at some specific site of the gene of interest, the stronger the hybridization signal seen on the array. Additionally, antibodies specific to methylated cytosines have been developed and applied in accordance with microarray technology to interrogate DNA methylation within the genome 109. In addition to DNA methylation, histone residue modification can be investigated using chromatin immunoprecipitation on microarray chips (ChIP‐on‐chip) 108.

As an example of the microarray‐based approaches, a recent analysis adds weight to the hypothesis that some molecular causes of gender differences may manifest in different profiles of epigenetic markers between the sexes. As a part of our ongoing study on epigenomics of human brain we performed a pilot analysis of epigenomic profiles of postmortem brain samples stratified for gender. The unmethylated fraction of DNA extracted from the prefrontal cortex of postmortem brains was interrogated on the 12K CpG island microarray 58, 61. The group of males (n = 20) was compared to the group of females (n = 7). DNA methylation differences in males and females (P<0.05) identified by the microarray between the sexes are shown in Figure 3. While this pilot data is not corrected for multiple testing, the identified regions might serve as a starting point for the in‐depth epigenetic studies of sex differences, mapping of gender‐dependent epigenetic effects across the entire genome, and analysis of the relevance of such sites to various complex diseases.

Figure 3 Loci identified as significantly different in DNA methylation between male and female postmortem brain samples. DNA methylation profiles were measured by human CpG island microarrays in a common reference design. Each microarray interrogates 12,912 loci whose locations on the chromosomes are shown in white in the plot. Data were logarithmically transformed and then normalized, by the print‐tip loess method, to remove nonbiological effects. Vertically: human chromosomes 1 to 22. The t‐statistic of the male group versus female group was calculated for each locus and its P‐value was also obtained. Loci with P‐values below 0.05 are painted black in the plot.

Experimental techniques applied to investigate longitudinal changes in gender‐specific epigenetic patterns will be of particular interest. Also, studies on epigenetic effects of hormones may allow for the identification of critical preclinical epigenetic states that could be used in early diagnostics as well as in designing new epigenetic therapies 110. It is our hope that through an acceleration of molecular biology technologies of epigenetic investigation and a combination of animal and human studies, the role of epigenetic factors in complex disease—both mediated through sex hormones and other mechanisms—will soon be elucidated.

Acknowledgements

This research has been supported by the Ontario Mental Health Foundation, Canadian Institutes for Health and Research, National Institute of Mental Health (R01 MH074127‐01), as well as NARSAD and the Stanley Foundation. Zachary Kaminsky is a CIHR Doctoral Fellow. Special thanks to Sigrid Ziegler for her critical reading of the manuscript.