Involvement of epigenetic modifiers in the pathogenesis of testicular dysgenesis and germ cell cancer

Introduction

The vast majority of human testicular tumors are derived from germ cells and develop in young adult men between 17 and 45 years of age. The incidence of testicular germ cell tumors (TGCTs) has been increasing over several decades and is the most frequently found cancer among young adult men in most countries (1–3). It manifests either as a seminoma or non-seminoma, which are very different tumors in terms of both histology and treatment regimes. Seminomas are homogeneous tumors composed of mitotically dividing germ cells, whereas non-seminomas are heterogeneous tumors that may contain varying proportions of undifferentiated embryonal carcinoma, partially differentiated somatic tissues (teratoma) and extra-embryonic elements such as choriocarcinoma and yolk sac tumors (4). Both of these very different tumor types are, however, preceded by a symptomless and pre-invasive stage, where the common precursor cell of TGCTs, the so-called carcinoma in situ (CIS) cell, can be found in situ in normally arranged seminiferous tubules.

Development of testicular cancer

It is believed that the CIS is formed in utero as studies of protein markers [reviewed in (5)], global gene expression profiles of CIS cells (6–8) and microRNAs (9) have provided evidence that CIS cells are very similar to fetal germ cells (gonocytes). It is thus very likely that a CIS cell is a gonocyte that failed to mature during development as outlined in Figure 1. Like the fetal germ cells, the CIS cells express a large range of stemness genes, including POU5F1 (OCT-3/4), NANOG, LIN28-A, DPPA4, DPPA5, KIT and TFAP2C (6–8, 10–14) and is regarded as pluripotent or even totipotent (can give rise to both embryonic and extra-embryonic cell types).

Figure 1:

Schematic representation of the relationship between events of fetal gonadal development and epigenetic patterns observed in germ cells and in the pathogenesis of TGCT.

Epigenetic marks denoted with red and green are generally associated with repressive and active chromatin, respectively. Jumonji C domain-containing proteins play important regulatory roles during fetal gonadal development by regulating histone lysine methylation marks of, e.g. the Sry promoter in somatic gonadal cells. Histone lysine methylation marks are important regulatory set points in both germ and somatic cells during normal fetal gonadal development but also in the neoplastic and malignant transformation in the pathogenesis of TGCTs. CIS, carcinoma in situ; PGC, primordial germ cells.

Further transformation of CIS cells most likely occurs during adaptation to the changed endocrine environment of the peri- and post-pubertal testis, when the cells undergo polyploidisation (15) and acquire a characteristic pattern of genomic aberrations, including the gain of chromosome 12p (16, 17), often seen as the isochromosome 12p (18).

Our current hypothesis is that developmental arrest of gonocytes is caused by disruption of the endocrine environment in the testes (niche), which is comprised of somatic cells that secrete hormones and paracrine factors that stimulate male-specific germ cell maturation into spermatogonia (19, 20). This has recently been emphasized by the identification of DMRT1 and meiosis markers in pre-pubertal CIS cells indicative of a ‘feminized’ CIS cell (21).

Testicular dysgenesis syndrome

Observations of an increasing trend in the incidence of TGCTs and congenital reproductive tract malformations at the same time as a downward trend in semen quality and testosterone levels have led to the proposal that these conditions may represent a syndrome of disorders, a testicular dysgenesis syndrome (TDS) (22). Most of these disorders share common risk factors and are risk factors for each other, and all involve a potential disturbance in the development of the testes during fetal life. The aetiology of TDS is unclear, but the apparent rapid increase in male reproductive health problems during a few generations suggests that changes in lifestyle and/or environmental factors are more likely causes than genetic factors. Nevertheless, testicular cancer is known to be one of the most genetically predisposed cancers. Having an affected father or brother causes approximately a four- and nine-times increased risk, respectively (23), which is greater than most other common cancers. Genome-wide association studies have indicated that genetics can account for approximately 1/4 of the cases (24–29), and risk alleles include genes like KITLG (SCF), DMRT1, SPRY4, and BAK1, which have all been described as being involved in fetal gonadal and germ-cell development. Recently, pathway-based analyses of genome-wide association data have revealed that genes specifically involved in sex determination are associated with testicular cancer (30) and that many of the TGCT risk alleles also are risk alleles for TDS (31). Hence, the close relationship between sex determination, TDS and testicular cancer is genetically evident and is also reflected both histologically and clinically. However, among the risk alleles associated with TGCT, none of the jumonji C domain-containing genes reached significance. Histologically, TGCTs often coincide with dysgenetic tubules presenting spermatogenic arrest, micro-calcifications, poorly differentiated Sertoli cells, Sertoli-cell-only patterns, etc. (32, 33). Clinically, men with low sperm quality or reduced fertility have a three-fold increased risk of developing TGCTs (34, 35). Much evidence therefore points toward a close similarity in the aetiology of TDS and TGCT.

Epigenetics in the development of testicular cancer

The epigenetic state of CIS cells shows high similarity to the epigenetic patterns found in fetal germ cells (36). CIS cells contain very low levels of DNA methylation when detected by immunohistochemical staining for 5-methylcytosine (5mC) (37), and apparently this unmethylated state is actively maintained by expression of APOBEC1 and base excision repair proteins MBD4, APEX1 and PARP1 (38). Another possible demethylation pathway, the conversion of 5mC to hydroxyl-methyl-cytosine (5hmC), was not found active in CIS cells as both 5hmC and the TET proteins, which convert 5mC to 5hmC, were not detected in CIS cells (38). Histone modifications H3K9me2 (Histone H3 dimethyl Lys9) and H3K27me3, both associated with a restrictive chromatin structure, are absent or found only at low levels in CIS cells, but H3K4me1, H3K4me2/3, H3K9ac, H3K27ac and the histone variant H2A.Z, which all are modifications associated with active and permissive chromatin structures, are abundantly present (36). Moreover, CIS cells show high levels of H4/H2AR3me2 along with expression of BLIMP1/PRMT5 (39), which are thought to repress HOX genes and somatic differentiation programs (40). Another possible way of regulating HOX genes is through repression by the polycomb group proteins as seen in human embryonic stem cells (41). This regulation is, however, dependent on H3K27me3 marks (42, 43), which as mentioned above are absent or found only at low levels in CIS cells. Moreover, the known H3K27me3 demethylases, UTX (KDM6A) and JMJD3 (KDM6B), which regulate HOX gene expression in human embryonic teratocarcinoma NT2/D1 cells, are indispensable for normal gonadal development in Caenorhabditis elegans (44) and regulate the fragmentation of spermatogonial cysts in mice (45), are not expressed in CIS cells (36). It therefore seems more likely that BLIMP1/PRMT5 controls repression of HOX genes and somatic differentiation in CIS cells. Taken together, the emerging picture of the epigenetic profile of CIS cells is an open and permissive chromatin structure that resembles the profile found in fetal germ cells.

Malignant transformation of the neoplastic CIS cells results in either a seminoma or a non-seminoma, and each tumor type develops a distinct epigenetic profile. In general, the non-seminoma, embryonal carcinoma, gains high levels of 5mC (and concurrent upregulation of DNMTs) but retains permissive histone modifications, whereas the seminoma maintains low levels of 5mC but loses the permissive histone modifications (36, 37, 46). In accordance with this, most seminomas also have high levels of H4/H2AR3me2, whereas low levels are found in non-seminomas (39). In summary, adult totipotent CIS cells retain a fetal germ-cell-like epigenetic profile that, after malignant transformation, evolves into distinct signatures reflecting the tumor types.

Epigenetic modifiers

Epigenetic regulation is a dynamic process where modifications like methyl-groups are not only added to DNA and histones but are also actively removed by demethylases with different specificity. Three classes of histone demethylases have been discovered [reviewed by Klose et al. (47)]. The first class is represented by PADI4 (Peptidylarginine deiminase 4), which converts methylarginine to citrulline and thus cannot strictly be categorized as a demethylase. The founding member of the second class of enzymes is LSD1 (Lysine specific demethylase 1), which removes methyl groups from mono- and dimethylated H3K4 and H3K9 in an oxidative manner and with flavin as a cofactor. The third and largest group of demethylases is constituted by jumonji C (JmjC) domain-containing enzymes, which specifically demethylate mono-, di-, and tri-methylated histone lysines in a reaction dependent on iron and α-ketoglutarate. Thirty proteins with an evolutionarily conserved JmjC domain have been identified. So far, 18 jumonji demethylases have documented demethylase activity toward H3K4, H3K9, H3K27, H3K36 or H4K20 [reviewed by Franci et al. (48)]. JmjC proteins have been shown in mice to be necessary for neural tube formation (49), normal heart development (50) and proper testes formation and function (51–53), and their deregulation has been found in numerous tumors, including colon, lung, hematological, pancreatic and breast cancer (54).

Jumonji and testicular development

Fetal and postnatal testicular development is an intricate process where precise epigenetic regulation and derived gene expression cues are pivotal for normal testes function in adulthood (see above). Several studies of JmjC proteins have revealed roles related to germ-cell development, sex determination, fertility and metabolism (51–53, 55).

JMJD1A is necessary for normal spermatogenesis in mice

In 2007, Okada et al. reported that knockout (KO) of Jmjd1A (also known as Kdm3a, Jhmd2a or testis-specific gene A) in C57BL/6 mice led to infertility and small testes (51). Infertility was caused by the failure of spermatids to elongate due to impaired chromatin condensation and could be related to a decreased expression of two testis-specific basic proteins; transition nuclear protein 1 (TNP1) and protamine 1 (PRM1) in round spermatids of KO mice. In wild-type mice, both JMJD1A protein and transcript were found to be moderately expressed in spermatocytes and highly expressed in round spermatids, and chromatin immunoprecipitation (ChIP) analysis revealed that JMJD1A is recruited to the core promoter regions of Tnp1 and Prm1. Genome-wide H3K9 methylation levels were unaltered, but higher levels were found at the promoters of Tnp1 and Prm1 in KO vs. wild type. Probably as a consequence of impaired chromatin condensation, elongating spermatids showed abnormal sperm head morphology and resulted in most spermatozoa being immotile. Surprisingly, there were no differences in hormone levels between wild-type and KO mice.

Liu et al. (52) created another Jmjd1a KO mouse model, also in a C57BL/6 background, that in concordance with the previous findings by Okada et al. (51) exhibited small testes, blocked spermatid elongation and infertility. However, in contrast to Okada et al., Liu et al. found that JMJD1A deficiency caused a general increase in H3K9me2 and H3K9me1 levels when assayed by Western blot analysis of whole testes lysates. Moreover, extensive germ-cell apoptosis was reported in spermatocytes and in round and elongating spermatids and resulted in a 1000-fold reduction in the amount of spermatozoa found in the epididymis. Transcriptional regulation of Tnp1 and Prm1 is mainly facilitated by a cAMP-response element modulator (CREM) and coactivator ACT (56–58). Liu et al. found decreased levels of Act mRNA but not Crem mRNA in Jmjd1a-deficient mice. However, there was a decreased recruitment of CREM to the promoters of Tnp1 and Odf1. Jmjd1a deficiency caused increased H3K9me2/me1 levels at both genes and a decreased expression of CREM and ACT target genes, including Tnp1, Tnp2, Prm1, Prm2, Odf1 and Gsg3. These results demonstrate that JMJD1A, by regulating repressive histone marks H3K9me2 and H3K9me1, is necessary for expression of Act, recruitment of CREM and expression of their target genes.

JMJD1A is necessary for expression of Sry and Sox9

In 2013, Kuroki et al. showed that KO of the same jumonji gene (Jmjd1a) in a C57BL/6 mice strain with a Y-chromosome from CBA embryonal stem cells frequently led to partial or complete sex reversal in XY animals with development of ovotestes or ovaries (53). Interestingly, this effect was greatly dependent on the genetic origin of the Y-chromosome. A CBA origin was the most sensitive, where 88% of XY JMJD1A-deficient mice displayed abnormal sex differentiation compared to 14% in mice with the Y-chromosome coming from C57BL/6 embryonic stem cells. All XY Jmjd1a^+/+ and XY Jmjd1a^+/Δ mice developed two testes. In concordance with previous studies (see above), spermiogenesis defects were observed in mice of both backgrounds. JMJD1A was dispensable for both female sex differentiation and female fertility. A very important finding was that JMJD1A deficiency reduced the expression of the protestis mammalian Y-chromosome sex-determining gene Sry to approximately 30% in the gonads of both mice strains at embryonic day 11.5 (E11.5). The group also demonstrated with a ChIP analysis that JMJD1A directly binds to regulatory regions within the Sry locus in wild type gonadal somatic cells at E11.5 and that Jmjd1a deficiency led to a significant increase in H3K9me2 levels within the Sry locus. Concordantly the levels of the activating histone mark H3K4me2 decreased in KO mice compared to wild types. It is therefore likely that JMJD1A directly activates the expression of Sry by removing suppressive H3K9me2 marks and thereby allowing the methylation of H3K4, which acts as an activator of transcription. If Sry is not expressed, the default path will lead to ovarian development. However, minor or local disturbances in the process of sex determination might lead to dysgenetic male gonadal development and could consequently be involved in lack of maturation and neoplastic transformation of germ cells. In principle, the degree of JMJD1A regulation of Sry locus could determine the degree of dysgenetic development.

The expression of the Sertoli cell marker Sox9 was also significantly reduced at E11.5 in KO mice, while at the same time the number of gonadal somatic cells were unaltered between controls and mutants. However, H3K9me2 levels on Sox9 were unchanged by JMJD1A deficiency, indicating an indirect mode of regulation.

JMJD1A facilitates transcription activation through the androgen receptor

JMJD1A contains an ⁸⁸⁵LXXLL⁸⁸⁹ sequence, which is a signature motif involved in nuclear hormone-receptor interaction (59), and the protein has been found to bind to the androgen receptor (AR) in a ligand-dependent manner in vitro. Furthermore, in vivo studies with LNCaP cells (androgen-sensitive human prostate adenocarcinoma cells) demonstrated that treatment with synthetic androgen R1881 increased the recruitment of JMJD1A to enhancers of the known AR target genes PSA (KLK3; prostate specific antigen) and NKX3-1 (NK3 homeobox 1) concomitant with a decrease in the level of H3K9me2. Knockdown of Jmjd1a led to a significant reduction in hormone-induced transcription (60). Together this indicates that JMJD1A functions as a co-activator of transcription by the androgen receptor. Clinically a link between obesity, testosterone levels and male subfertility is well known, and it is therefore relevant to note that JMJD1A deficiency has also been reported to cause obesity in male mice (55). This was manifested as increased body fat deposition and higher serum lipid content and was possibly caused by an altered expression of metabolic genes, which caused impaired β-oxidation and glycerol release in skeletal muscle. In line with JMJD1A, JMJD2A, B and C (KDM4A, B and C) have also been reported to enhance AR stability and regulate its transcription (61, 62).

Taken together, JMJD1A seems to be specifically involved in crucial steps of proper gonadal development (sex determination) and in maintaining function of the developed testis (DNA condensation and AR signalling).

Other jumonji C domain-containing proteins and testis function

Knockout of Jmjd1c in mice (63) resulted in an age-dependent infertility caused by an age-dependent increase in apoptosis of germ cells. JMJD1C was found primarily in undifferentiated spermatogonia, and KO therefore probably perturbed the supply of undifferentiated spermatogonia. Global H3K9 methylation was unaffected in the KO mice, and it has been questioned whether JMJD1C actually contain H3K9 demethylase activity (64).

JMJD1C has also been reported to regulate the expression of CYP17A1, also known as 17α-hydroxylase, via SF-1-mediated transcription (65). CYP17A1 is a key enzyme in the synthesis of androgens and converts pregnenolone and progesterone to DHEA and androstenedione, respectively.

JMJD3 (KDM6B) has been reported to be involved in regulating the fragmentation of spermatogonial cysts, and down-regulation of Jmjd3 in spermatogonial stem cells have been reported to increase the number of undifferentiated spermatogonia without affecting their differentiation (45). In fetal germ cells and in adult CIS cells, JMJD3 is not expressed (Figure 2) and could play a role in securing the high proliferation rates of CIS cells (36).

Figure 2:

Immunohistochemical staining of gestational-week-21 gonads and adult testis containing CIS, with antibodies directed against OCT 3/4 (POU5F1) and JMJD3.

Arrows indicate fetal germ cells and CIS cells, respectively. JMJD3 is not found expressed in fetal gonads or CIS cells, whereas OCT 3/4 is a classical marker of both. Images are amended from Almstrup et al. (36), where staining patterns in other tissues are also reported. The scale bar represents 50 microns.

In summary, several jumonji C domain-containing proteins have been shown to play important roles at multiple levels of testicular development. JMJD1A seems particularly important for testis function, but it is unknown to what extent these enzymes can compensate each other, especially in humans, where very little is known about the functional roles of jumonji C domain-containing proteins.

Expert opinion

Multiple lines of evidence suggest a close relationship between sex determination, testicular dysgenesis and development of testicular cancer. Germ cells undergo a tightly regulated epigenetic program as they develop and consequently position epigenetic modifiers as key players. Jumonji C domain-containing proteins seem to be implicated in many of the processes that are necessary for both proper fetal testicular development and maintenance of testicular function in adult life. However, the importance of JmjC proteins in human germ-cell development remains to be established.

Outlook

It is a puzzle how histone demethylases can be specific for certain cell types, certain genomic regions and different time points during development. How can JMJD1A be recruited specifically to promoters of Sry, Tnp1 and Prm1 but not to other genes? JMJD1A has no defined DNA-binding motif, and there is no obvious consensus DNA sequence in the Tnp1 and Prm1 promoters. Can other proteins be involved? Moreover, it remains unknown to what extent the family of histone demethylases can compensate each other in the process of human germ-cell development. To describe this complex spatio-temporal relationship, massive genome-wide mapping of epigenetic marks along with its modifiers on a single cell basis is needed.

Highlights

• Testicular germ cell tumors (TGCTs) are the most common cancers among young men.

• Evidence indicates a close similarity in the aetiology of testicular dysgenesis and TGCT.

• Testicular germ cell cancer originates from a totipotent carcinoma in situ (CIS) cell.

• CIS cells resemble fetal germ cells and retain an open and permissive chromatin structure in the adult testis.

• Upon malignant transformation, CIS cells develop into two epigenetically distinct tumor types: seminomas and non-seminomas.

• Jumonji C domain-containing (JmjC) proteins specifically demethylate histone lysines.

• Dysregulation of JmjC proteins in mice and human cancer cell lines has recently been reported to cause various degrees of sex reversal, infertility and smaller testes and to influence androgen action.

• We speculate that the regulation of JmjC proteins may be central in the aetiology and pathogenesis of testicular dysgenesis, including testicular cancer.

List of abbreviations

5hmC

5-hydroxymethylcytosine

5mC

5-methylcytosine

AR

androgen receptor

ChIP

chromatin immunoprecipitation

CIS

carcinoma in situ

TDS

testicular dysgenesis syndrome

TGCT

testicular germ cell tumor