Introduction

Sperm cells are, arguably, the most differentiated cells, and spermatogenesis may very well be one of the most complex biological processes undergone by any organism. This is because spermatogenesis features not only the germ-cell-specific processes of meiotic division, but also processes necessary for the characteristic properties of sperm: shape (chromatin remodeling) and mobility (flagellum development). Spermatogenesis, which incorporates juxtacrine, paracrine, and endocrine factor information, evokes thousands of genes and proteins, many of which are testis specific. Although the processes by which spermatogenesis occurs may vary between taxa, one study of Drosophila (Dorus et al. 2006) demonstrates that certain critical molecular pathways are conserved from invertebrates to mammals.

The process of spermatogenesis occurs in multiple structures of the male reproductive system, including the epithelial Sertoli cells of the seminiferous tubule, and can be partitioned into three major categories: spermatocytogenesis, spermatogenesis, and spermiogenesis. In spermatocytogenesis, through mitotic and meiotic I division, haploid spermatocytes are formed; in spermatogenesis, secondary spermatocytes are formed, which then enter meiosis II to produce haploid spermatids; in spermiogenesis, the flagellum of spermatids grows, mitochondria gather and form an axomeme, and spermatid DNA undergoes repackaging and condensation. Nuclear condensation occurs by replacement of somatic histones with intermediate proteins, which themselves are exchanged for protamines (Green et al. 1994; Kistler et al. 1996; Meistrich 1989). Protamines are a diverse family of arginine-rich late-state spermatid proteins found in both plants and animals, and are essential for normal sperm function. In mammals, there are two protamines: P1 and P2. P2 is a zinc finger protein as are protamine-like proteins in cuttlefish (Martin-Ponthieu et al. 1994).

Sperm is now known to be responsible not only for the delivery of genetic, but also epigenetic information, to the next generation (Kimmins and Sassone-Corsi 2005; Martins and Krawetz 2007; Rousseaux et al. 2005, 2008). Via these modifications, heritable changes in gene expression or cellular phenotype can occur without mutation to the underlying DNA sequence (Goldberg et al. 2007). Epigenetic modifications can take place as a result of histone modifications, DNA methylation, and chromatin remodeling. In plants and animals, small noncoding RNAs have been found to guide epigenetic modifications to underlying DNA via RNAi pathways (reviewed in Tanzer et al. 2010). Recent advances in our understanding of non-coding RNAs have clearly linked them to male-mediated transgenerational epigenetic inheritance (reviewed by He et al. 2009; Cuzin and Rassoulzadegan 2010). MicroRNAs are a class of non-coding RNAs, ~22 nt in length, found in plants, animals, viruses, and algae (Papaioannou and Nef 2010). In Xenopodines frogs (Silurana + Xenopus), microRNAs are often coded by sequences located from intronic regions of the protein-coding gene that they regulate; they can also be transcribed from intergenic regions, in which case, their transcription is not directly coupled with the gene(s) they regulate (Tang and Maxwell 2008). Although often associated with knock-down, as opposed to knock-out, microRNA can also be involved in the upregulation of the genes they affect (Vasudevan et al. 2007). Although it is now clear that RNAs, including microRNA (Grandjean and Rassoulzadegan 2009; Dadoune 2009), are involved in sperm development, many aspects of the fundamental biology have yet to be elucidated. However, it is clear that somehow small RNAs allow for the silencing of mobile elements, the imprinting of paternal genes, and DNA compaction (reviewed in Zamudio et al. 2008). Epigenetic-mediated abnormalities have also been linked to male fertility (Emery and Carrell 2006), as errors in genomic imprinting, RNA profiles, and abnormal chromatin packaging are now viewed as major contributors to subfertility and/or sterility. One such example is that mice unable to produce small RNA properly produce abnormally elongated spermatids and are consequentially infertile (Maatouk et al. 2008), while misexpression of Dicer, and thereby misexpression of microRNA, in testicular cells also results in infertility (Papaioannou et al. 2009; Papaioannou and Nef 2010).

MicroRNAs are not alone in small RNAs being attributed to fertility. Recent studies on PIWI-interacting RNAs (piRNAs) show that this class of small RNAs likely participates in both male and female fertility (reviewed in Klattenhoff and Theurkauf 2008; Brennecke et al. 2008). For this article, however, we do not investigate the role of piRNAs in male fertility because (1) piRNAs, obtained from the pachytene stage of sperm development and beyond, e.g., beginning at ~14 days post-birth in mice, have unknown function (Aravin et al. 2006; reviewed in Aravin et al. 2007b); (2) publicly available data of verified vertebrate piRNAs are composed solely of pachytene piRNAs (RNAdb, Pang et al. 2005); and (3) piRNA conservation is low, even across genetic distances spanning mouse and rat (Aravin et al. 2006).

In Xenopus, interspecies hybrid males are sterile, though females are not (Kobel and Du Pasquier 1975; Kobel et al. 1981; Picker 1985; Picker and de Villiers 1989; Picker et al. 1996; Fischer et al. 2000). This pattern of hybrid sterility is unusual, as Xenopus males are the homogametic sex (ZZ). According to Haldane’s Rule (Haldane 1922), instances of reduced fertility and/or sterility in hybrids are predicted to occur in the heterogametic sex. However, Xenopus defy this general rule, as feminized genetic males become fertile while masculinized genetic females become sterile (Malone and Michalak 2008). Therefore, hybrid sterility in this clade has been attributed to dysfunctional spermatogenesis driven by faster male evolution. According to the faster male evolution hypothesis, hybrid males are more likely to be subfertile or sterile because of (i) the sensitive nature of spermatogenesis, and/or (ii) stronger selection on male-specific genes (Wu and Davis 1993; Wu et al. 1996). The well-documented sterility information of male Xenopus hybrids provide biologists with the unique opportunity to identify conserved microRNAs that are likely to be involved in male fertility; we can use comparisons between sterile hybrids and fertile parental individuals to identify microRNAs that are differentially expressed in hybrids. This information can then be relayed to expression profiles for microRNAs in mammalian testes. Owing to the conserved nature of microRNA, it is likely that homologs in Xenopus and mammalian testes are important for male fertility in all tetrapods; it is also likely that the subset of this homologous microRNAs found to be misexpressed in comparisons between sterile and fertile frogs is critical for tetrapod spermatogenesis. Therefore, the identification of these microRNAs could prove to be beneficial for understanding better the epigenetic control of microRNAs in male fertility.

Materials and Methods

Total RNA was obtained from the ground testes of freshly euthanized X. laevis, X. muelleri and X. laevis:X. muelleri hybrids using Ambion RNA extraction kits (IACUC protocol number A08.002). To obtain <40 nt RNAs, the samples were fractionated using an Ambion flashPAGE fractionator. Libraries of small RNA cDNAs were constructed and sequenced using ABI’s SOLiD sequencing next-generation technology (outsourced at the University of Oklahoma Medical Sciences core facility). All the samples were run on a 35-nucleotide array, and a GeneSifter (Geospiza 2011) pipeline was used to identify and quantify components of the sample, including microRNAs (employing miRBase v. 15) and piRNA (employing RNAdb; Pang et al. 2005). In addition, all reads were mapped against the Silurana tropicalis reference genome (Hellsten et al. 2010), the only anuran genome available thus far. Once each sample’s microRNA expression profile was characterized, pairwise analyses between X. laevis:hybrid and X. muelleri:hybrid were conducted to identify the most differentially expressed microRNA. These likelihood ratio tests were conducted on log-transformed data that had been normalized against the total number of reads per sample. A list of the top 20 misexpressed microRNAs was compiled for each comparison, and the lists were then condensed to reflect microRNAs common to both. This revised list was cross-referenced to microRNAs known from mouse testes (Mus musculus) (Mishima et al. 2008).

To determine putative gene targets of microRNAs shared between Xenopus and Mus, the EMBL-EBI Microcosm v5 (Enright 2011) database was searched using the Xenopus microRNA name. If no results were returned, the precursor and mature sequences were retrieved from MirBase v16 (Griffiths-Jones et al. 2008). These sequences were then entered into microRNAminer (Artzi et al. 2008), and Mus homologs were identified. The homologs were then entered into Microcosm. If no targets were identified, then the microRNA target was considered “unknown.” For those microRNAs targets of which had been identified, the Universal Protein Resource (UniProt; Jain et al. 2009; The UniProt Consortium 2011), database was used to identify proteins known to be expressed in testes.

PiRNAs, which are almost entirely gametically expressed were not included in analyses as piRNAs, currently available in the publically accessible RNAdb (Pang et al. 2005), are from pachytene populations. Unlike prepachytene piRNAs, which frequently map to transposable elements (Aravin et al. 2007a) and have been touted as “defenders of the germline,” these latter expressed piRNAs are not TE associated, and presently remain as an unknown function (reviewed in Aravin et al. 2007b).

Results and Discussion

SOLiD sequencing of the small cDNA library from testes of recovered 21,738,343, 18,991,070, and 15,585,086 reads for X. laevis, X. muelleri, and hybrid samples, respectively. Only a minority of the reads mapped to the known rRNAs, snRNAs, microRNAs, piRNA, intron/exon or other parts of the reference genome (Table 1). Despite overall low mapping of reads to reference sequences, the low number of rRNA reads clearly demonstrates that the small RNA libraries were prepared from high-quality small RNAs; because of the sheer abundance of rRNA in total RNA samples; a high number of rRNA hits would indicate that the original sample was degraded. This was not the case.

Table 1 Reads recovered from small cDNA libraries for X. laevis, X. muelleri, and their interspecies hybrid
Full size table

Global microRNA expression was reduced in the hybrid, and this massive misexpression makes the hybrid cluster outside of the group containing both X. laevis and X. muelleri in the expression profile (Fig. 1). Expression patterns almost universally placed X. laevis as having the highest expression, followed by X. muelleri, with hybrid expression consistently the lowest. This contrasts with microarray gene expression data in these taxa (Malone et al., 2007) where not only was the bulk of differentially expressed genes the highest between X. muelleri:hybrid, but that most of the expression differences between X. laevis:hybrid revealed overexpression in hybrids. It is, however, consistent with the results of an earlier microarray study of microRNA expression in these taxa (Michalak and Malone 2008) which recovered underexpression in hybrids relative to X. laevis (X. muelleri was not included).

Fig. 1
figure 1

Global expression for X. laevis, X. muelleri, and their interspecific hybrid. The hybrid exhibits widespread underexpression of microRNA. The degree of underexpression places the hybrid outside of the cluster containing both parental species

Full size image

Of the top 20 microRNA misexpressed in comparisons between X. laevis:hybrid and X. muelleri:hybrid (Table 2), eight were shared between lists. These eight were then cross-referenced against microRNA expressed in Mus musculus testes. Seven microRNAs commonly misexpressed in Xenopus hybrids were found to have homologs in Mus (Table 3). Microcosm results recovered a common association between these seven microRNA with genes having (1) zinc finger proteins and (2) homeobox function (Table 4). UniProt directly linked four of the 19 protein-coding genes identified by Microcosm to mammalian testes (mouse, rat, and human; see Table 5), two to closely related testes expressed protein, and two more to testes expressed associates. One of these two proteins is EED, a member of a protein complex associated with epigenetic repression of transcription of a number of genes, including Hox-c8, by methylating “Lys-9” and “Lys-27” of Histone-3. It is also thought to be part of an additional complex that recruits methyltransferase, thereby linking two epigenetic repression systems. Manger et al. (2003; reviewed in Ferguson-Smith and Reik 2003) found evidence linking EED to the maintenance of silenced alleles on paternal chromosomes. In our own Xenopus data (not published), we have recovered EED-associated microRNAs in multiple microRNA expression clusters of fertile non-hybrid frogs (X. tropicalis, X. laevis, and X. mulleri), demonstrating its involvement in multiple pathways of proper testicular cell function and/or spermatogenesis. Therefore, it makes sense to conclude that microRNA associated with this gene would be misexpressed in sterile males.

Table 2 The top 20 microRNA misexpressed in the testes of Xenopus comparisons and the fold change associated with each comparison, the direction of the gene regulation, and P values
Full size table
Table 3 The top 20 microRNA misexpressed in the testes of Xenopus comparisons; 8 of the 40 that were shared between the two comparisons, and seven of the eight that were shared between Xenopus and Mus
Full size table
Table 4 MicroRNAs shared between misexpressed Xenopus and Mus
Full size table
Table 5 UniProt report of microRNA and their protein or gene targets having testicular expression
Full size table

Zinc is an important structural component in different proteins associated with nucleic acid binding and/or gene regulation (reviewed in Berg 1990), alterations to zinc-stabilized structures (zinc fingers and zinc-bridges) could result in reduced stability of protein tertiary and quaternary structure (Sakai-Kato et al. 2009). Therefore, reduced Zn function is likely to result in ineffective sperm and male infertility. The association of Zn and zinc fingers and mammalian male fertility has been known for quite some time, as protamine 2 (P2) is a zinc finger protein specific to this clade, and mammalian males with decreased protamine P2 levels are often infertile (de Yebra et al. 1998). P2 is not the only zinc finger protein found in mammalian testes: others include ZPF-95 and ZPF-96, which appear to be expressed at different times during spermatogenesis (Weissig et al., 2003), and BORIS, a male-specific protein associated with epigenetic reprogramming of sperm (Loukinov et al. 2002). Zinc fingers are also important components of the male steroid hormones androgens (reviewed in Freedman 1992) which are crucial for processes ranging from sexual differentiation (reviewed in Hughes 2001; Sharpe 2006), sexual maturation (Buzek and Sanborn 1988), and spermatogenesis (Chang et al. 2004). The link between Zn and fertility of other vertebrates is not as clearly understood. However, recent demonstration of a role for Zn in fish spermatogenesis (Yamaguchi et al. 2009) suggests that Zn and zinc fingers may be ubiquitously required for vertebrate spermatogenesis to proceed normally. In this study, all six of the microRNA for which gene interactions were known were found to be associated with Zn or were zinc finger proteins; of these, three proteins have known expression in mammalian testes (see Table 5).

All homeobox genes contain a highly conserved 180-bp sequence. The homeobox codes for a 60-amino acid domain that binds specific DNA sequences and regulates transcription of other genes (Duboule 1994). One set of possible targets for homeobox genes are cellular adhesion molecules (CAMs) (Jones et al. 1992). CAMs are essential for embryonic tissues, and are intimately tied to initial boundary formation in tissues, embryonic induction and migration, tissue stabilization, and regeneration (reviewed in Edelman and Crossin 1991). Ep-CAM is a type of adhesion molecule that has been isolated in fetal and adult testes (Anderson et al. 1999), with Ep-CAM expression in adult males restricted to spermatogonia, undifferentiated germ cells present in the earliest stages of spermatogenesis. In mammals, disruption of Rhox (reproductive homeobox X-linked) genes results in reduced spermatozoa in testes and reduces motile spermatozoa in the epididymis, leading to subfertility (MacLean et al. 2005). Rhox are not alone in terms of homeodomain proteins being critical for male fertility. Sperm-1, which is expressed in germ cells following the first meiotic division and haploid spermatids, has been shown to affect fertility in mice (Pearse et al. 1997). Other known homeobox genes or proteins expressed in the male germ line include Hoxa-4, Hoxb-4, Hoxa-13, Oct, Pem, MH-3 (Hox-A4), Gtx, KIMI, LIM homeobox protein 9, EMX2, and M33 (Chromobox protein homolog 2) (Innis et al. 2002; Gupta 2005; reviewed in Wilson and Davies 2007).

Of the six microRNA for which gene interactions were known, three of them (miR-30, miR-10, and miR-196) were found to be associated with nine different homeobox proteins (see Table 4), one of which has known expression in mammalian testes, two of which are closely related to proteins known to be expressed in testes, and another two who are associated with proteins known to be expressed in the testes (see Table 5). Although it is unclear how most of the microRNA and homoedomain proteins identified in this study interact, proteins closely related to Hox-A1 participate in fetal testicular development and subsequent adult fertility (Kondo et al. 1997; Hsieh-Li et al. 1995; Satokata et al. 1995; Innis et al. 2002) and proteins closely linked to Hox-C8 are involved in the epigenetic process of histone methylation (Cao and Zhang 2004; Montgomery et al. 2005).

Conclusions

The sperm of sterile hybrids is significantly larger than the sperm of either X. laevis or X. muelleri (Malone et al. 2007). This is consistent with defective chromatin packaging in the later stages of spermatogenesis, when DNA-incorporated protamines replace histones, thus allowing for hyper-condensation of sperm, resulting in its characteristic morphology. This process is epigenetically guided, and results from hyper-acetylation during spermiogenesis: sperm compaction and elongation occurs as a function of protamine incorporation and histone degradation (Green et al. 1994; Kistler et al. 1996; Meistrich 1989). Many protamine or protamine-like proteins contain zinc finger domains; therefore, perturbation to the regulation of these genes or any changes that may affect Zn affinity to these domains are likely to contribute to a disruption in chromatin packaging and could cause morphological changes in sperm. Consistent with epigenetic regulation of spermatogenesis, P2 protamine content in infertile mammalian males is not necessarily associated with any mutations at the level of DNA sequence (de Yebra et al. 1998). MicroRNA, one pathway of epigenetic regulation, is thought to be critical in the regulation of gene expression at mitotic, meiotic, and post-meiotic stages of spermatogenesis (He et al. 2009).

For this study, we identified microRNAs that are misexpressed in sterile male hybrid Xenopus and compared these to microRNA profiles of fertile mice. In doing so, seven microRNA were identified; we hypothesize these seven to having critical roles in tetrapod spermatogenesis. Of these seven microRNAs, two have already been linked to male fertility and/or testes. In mammals, microRNA-222 has known association with the KIT (Mast/stem cell growth factor receptor) receptor and its ligand KL (Gabbianelli et al. 2010) as well as the homologous C-kit and its ligand stem cell factor (SCF) (He et al. 2005). These proteins participate in gametogenesis, and the KIT-KL/C-kit–SCF complexes are involved in the survival and proliferation of spermatogonia (Yoshinaga et al. 1991; Dym et al. 1995; Dirami et al. 1996), as well as in critical interactions between spermatocytes and Sertoli cells (Vincent et al. 1998). Males with mutated Kit or c-kit genes have reduced or non-existent fertility; mice with mutated Kit have a block at the pre-meiotic stages of spermatogenesis (Kissel et al. 2000) and humans with mutated c-kit experience increased apoptosis and subsequent subfertility (Sandlow et al. 1996; Feng et al. 1999). KIT proteins are not restricted to mammals; they are also present in, and critical for, spermatogenesis in Anurans (the clade containing all frogs and toads) (Raucci and Di Fiore 2007). Although their functions remain unknown, microRNA-30b and microRNA-30c have both been found abundantly expressed in mouse testes, each occupying >2% of the entire cloned population (Mishima et al. 2008), while microRNA-30a has been found expressed in human testes (Liu et al. 2004). MicroRNA-30 is associated with homeobox proteins and Zn transport, both of which are critical for male fertility; therefore, this may explain as to why testicular expression of this microRNA is conserved across tetrapods.