CHD5 is Required for Spermiogenesis and Chromatin Condensation

Abstract

Haploid spermatids undergo extensive cellular, molecular and morphological changes to form spermatozoa during spermiogenesis. Abnormalities in these steps can lead to serious male fertility problems, from oligospermia to complete azoospermia. CHD5 is a chromatin-remodeling nuclear protein expressed almost exclusively in the brain and testis. Male Chd5 knockout (KO) mice have deregulated spermatogenesis, characterized by immature sloughing of spermatids, spermiation failure, disorganization of the spermatogenic cycle and abnormal head morphology in elongating spermatids. This results in the inappropriate placement and juxtaposition of germ cell types within the epithelium. Sperm that did enter the epididymis displayed irregular shaped sperm heads, and retained cytoplasmic components. These sperm also stained positively for acidic aniline, indicating improper removal of histones and lack of proper chromatin condensation. Electron microscopy showed that spermatids in the seminiferous tubules of Chd5 KO mice have extensive nuclear deformation, with irregular shaped heads of elongated spermatids, and lack the progression of chromatin condensation in an anterior-to-posterior direction. However, the mRNA expression levels of other important genes controlling spermatogenesis were not affected. Chd5 KO mice also showed decreased H4 hyperacetylation beginning at stage IX, step 9, which is vital for the histone-transition protein replacement in spermiogenesis. Our data indicate that CHD5 is required for normal spermiogenesis, especially for spermatid chromatin condensation.

1. Introduction

The process of spermatogenesis has been divided into stages (I to XII, in mice) that occur in waves as germ cells progress from spermatogonia to spermatocytes, and finally into spermatids in the seminiferous epithelium (Hess and Renato de Franca, 2008). Spermiogenesis is the final phase of spermatogenesis. It describes the maturation of spermatids into mature, motile spermatozoa, and is broken down into steps (1–16). Spermiogenesis is a critical postmeiotic developmental event. During this phase, haploid spermatids undergo extensive cellular, molecular and morphological changes to form spermatozoa. Haploid, round spermatids differentiate into species-specific shaped spermatozoa, with dramatic morphological changes, including elongation and condensation of the nucleus, and formation of the flagellum (Zhang et al., 2009). Some essential genes for spermiogenesis or spermiation have been identified, but the underlying mechanisms remain largely unknown.

Spermiogenesis and spermiation are composed of several finely tuned events: gene regulation in spermatid development, a localization shift from basal to the apical epithelium, reabsorption of the ectoplasmic specialization interaction with Sertoli cells, and the final release of mature spermatozoa (Cheng and Mruk, 2011). Any abnormality in these steps can lead to male infertility (Beardsley et al., 2006). Among these abnormalities, defects in chromatin condensation are a very common cause of male infertility in men (Hammoud et al., 2009; Roest et al., 1996; Zhao et al., 2004).

Alteration of chromatin structure is an important mechanism for regulating DNA transcription (Thompson et al., 2008), and members of the chromodomain–helicase–DNA binding (CHD) family are considered to be important regulators of chromatin structure. (Yap and Zhou, 2011). CHD5 is a recently identified member of the CHD family (Thompson et al., 2003), and it is expressed almost exclusively in the nervous system and testis (Kolla et al., 2012; Potts et al., 2011). The CHD proteins are characterized by tandem chromodomains N-terminal to their catalytic Snf2 helicase domain. There are currently 9 known members of the CHD gene family in humans (CHD1–CHD9) (Hall and Georgel, 2007; Marfella and Imbalzano, 2007), two of which (CHD3 and CHD4) have been shown to participate in multiprotein complexes responsible for chromatin remodeling (Hong et al., 2005).

The nucleosome remodeling and histone deacetylation (NuRD) complex (originally known as Mi2 in Drosophila) is a large complex that consists of at least seven proteins, but the key functional protein in these complexes is either Mi-2α or Mi-2β (now known as CHD3 and CHD4, respectively). In vitro, these enzymes are able to catalyze structural changes, reposition nucleosomes on a DNA template, transfer histone octamers to bare DNA, and replace histones with histone variants such as transitional protein and protamines (Thompson et al., 2008). In vivo, these activities are crucial for transcription, replication, repair, and recombination of the eukaryotic genome (Hall and Georgel, 2007). There is increasing evidence that CHD-NuRD complexes can have a profound effect on chromatin structure and gene expression, so they are likely to play important roles in regulating development, cell cycle control, and spermatogenesis (Bajpai et al., 2010; Kunert and Brehm, 2009). Based on the fact that CHD5 expression is restricted to the testis and nervous system, we wanted to know if this protein plays a role in the chromatin condensation during the spermiogenesis in mammals.

We generated a Chd5 knockout (KO) mouse strain to investigate the role of this gene in neuroblastoma tumorigenesis (Fujita et al., 2008), but it became apparent that Chd5−/− males were not able to produce offspring, whereas Chd5+/− males and Chd5−/− females appeared to have normal reproductive function. In the present study, we demonstrate homozygous loss of Chd5 resulted in a decreased level of histone 4 hyperacetylation, which is vital for the proper replacement of histones. This further led to the failure of spermatid chromatin condensation, showing no anterior-to-posterior direction preceding the condensation of chromatin in the Chd5 −/− mouse. The failure to properly assemble the spermatid nucleus led to failure of spermiation, phagocytosis of sperm heads, and sloughing of immature spermatids, as well as abnormally high histone retention in the epididymal spermatozoa. These defects in spermiogenesis led to male infertility in the homozygous Chd5 KO mice.

2. Results

2.1. CHD5 is highly expressed in mouse brain and testis

To examine the expression of Chd5 in tissues, we carried out real-time, quantitative reverse transcriptase PCR (RT-PCR) with cDNAs made from different adult mouse tissues. The real-time RT-PCR result showed that CHD5 is most abundantly expressed in the mouse testis, more than 4 times higher than in the brain (Figure 1). There was also moderate expression in the adrenal and cauda epididymis, but very little expression in other tissues.

Fig. 1. Chd5 expression is restricted to testis and brain.

RNA from adult mouse organs was used to synthesize cDNA, which was then quantified by real-time PCR. Signal was normalized to the expression of the housekeeping gene 18s.

2.2. Gene targeting of Chd5 in mouse embryonic stem (ES) cells

To further study the function of CHD5 in vivo, we created a targeting vector to constitutionally or conditionally inactivate the Chd5 gene in mice. The map of the targeted allele and strategies are shown in Figure S1. After electroporation, 11 clones with the correct genomic Southern pattern were obtained. Among these 11 clones, 4 lost the LoxP site located in intron 11. The remaining 5 clones were karyotyped, and the best two clones (clone 1B6 had 7/10 correct metaphases with 40 chromosomes, clone 2F5 had 10/10 with 40 chromosomes) were injected into blastocysts. The chimeric founders were mated to obtain mice that had one copy of the KO allele in the germline. We used both genomic Southern and PCR to validate their genotype (Figures 2A–2C). After 7 generations of backcrossing into a 129-E, the Chd5 −/− mice were used for analysis (Chd5 KO) in comparison to wild type (WT) mice. The Prm-Cre was eliminated during the backcrossing. In the Chd −/− mice, the CHD5 protein was completely undetectable by Western blot, whereas the expression of this gene in Chd5 +/− mice was decreased by about half compared to Chd5 WT mice (Figure 2D).

Fig. 2. Chd5 gene was knocked out in mice.

(A) A typical Southern blot showing the targeted and non-targeted ES cell genomic DNA, after cutting with SpeI. (B) Using WT and KO allele specific primers we performed PCR on genomic DNA from the tail to determine the genotype of Chd5 +/+, +/− and −/− mice. The WT gDNA only resulted the WT allele band, and the KO gDNA showed only the KO allele band; the heterozygous gDNA showed both. (C) After deletion of exons 12 and 13 by Cre, DNA from the tails of WT, heterozygous and homozygous KO mice was hybridized with the 5’-probe. (D) Testis cytoplasmic and nuclear proteins were hybridized with CHD5 rabbit polyclonal antibody. There was no signal in the cytoplasmic protein, and a very strong signal in the WT nuclear protein. The intensity of the CHD5 band was decreased to about half of the WT in the +/− mouse, and completely absent in the Cd5 −/− mice. * = CHD5 protein.

2.3. Male Chd5 KO mice are infertile

Mice with both heterozygous and homozygous Chd5 KO appeared grossly normal at birth, and their growth and development were normal, except for subtle alterations in neurobehavioral development that became apparent after detailed testing, presumably related to the lack of CHD5 expression in the brain (data not shown). However, the only observable phenotype of Chd5 KO was male infertility. In one experiment, 6 of the Chd5 KO males were paired with 2 WT 129-E females each, and none of them were able to reproduce over a period of at least two months. We observed normal mounting behavior, and all the females had post-coital plugs, indicating that mating had taken place, but no litters or pups were generated by any of these matings. We also tried to mate at least 12 Chd5 KO male mice with Chd5 WT females that were generated from our own colony, which is also 129-E, but no pups or litters generated. Again, normal mounting behavior was observed, and females consistently had plugs, indicating that mating had occurred. Thus, we conclude that the chd5 KO male mice are sterile. Interestingly, the WT and heterozygous male mice had normal litter sizes (6–8 pups/litter), so haploinsufficiency of CHD5 expression apparently still is sufficient for normal spermatogenesis. Loss of Chd5 in female mice did not affect their reproduction, as we observed normal litter sizes (6–8 pups/litter) when Chd5 KO females were mated with 129-E WT male mice. There was no significant difference in the testosterone levels between WT and Chd5 KO mice, which strongly suggests that CHD5 deficiency in the central nervous system is not responsible for the abnormalities in spermatogenesis in male Chd 5 KO mice (data not shown).

2.4. Abnormal sperm in the Chd5 KO mouse epididymis

To investigate the cause of the Chd5 KO male infertility, we examined the epididymis of male mice. In contrast to the WT mice, Chd5 KO mice either had no sperm in the epididymis, or they had morphologically abnormal sperm, with decreased sperm density and abnormal presence of sloughed germ cells (Figures 3A, 3B). We examined the epididymis of 15 Chd5 KO male mice, and the impact on spermatogenesis was variable. In some mice (6/15, 40%), no sperm were observed in the epididymis (obstruction of the efferent ductules and epididymis was excluded by serial section and injection of dye into the rete testis). We identified complete failure of spermiogenesis in the testes of these mice (Figures 3C, 3D). Also, 3/15 (20%) of Chd5 KO mice had a greatly reduced total number of cauda epididymal sperm, and all the sperm were morphologically abnormal and immotile. In 6/15 (40%), the number of sperm in the cauda epididymis was comparable to WT mice, but all sperm were morphologically abnormal (98% in KO vs 5% in WT, and see below). Furthermore, we recovered live sperm from these animals (one epididymis each, as the other was used for histology), and all sperm were immotile, with no flagellar movement. Normal, crescent-shaped sperm heads were never observed in the epididymis of Chd5 KO mice. Thus, in addition to variable sperm counts, the uniformity of morphological abnormalities and complete immotility account for the infertility of chd5 KO mice.

Fig. 3. Epididymal sperm are missing in Chd5−/− (KO) mice.

(A) WT (+/+) epididymis showing normal concentration of sperm in the lumen (arrows). Bar=25µm. (B) KO epididymis lacks sperm and numerous round germ cells (arrows) are present in the lumen. The microvilli and lumens show increased PAS staining. (C) Wild-type testis showing Stage VIII to IX transition, with steps 8 and 9 spermatids. All step 16 elongated spermatids have been released into the lumen. (D) Chd5 −/− testis showing failure of spermiation with step 16 elongating spermatids remaining in the epithelium (arrows), among steps 8–9 spermatids. Bar=25µm.

2.5. Chd5 KO mice showed abnormalities in spermatogenesis beginning in stage IX, step 9

Compared to the crescent shape of normal mature sperm, the Chd5 KO male mice produced sperm with an irregular head and incomplete removal of cytoplasm (Figure 4). This abnormality was not observed in Chd5 heterozygous mice, which again indicates that haploinsufficiency does not affect CHD5 function in murine spermatogenesis. In order to exclude an indirect effect of CHD5 deletion on testicular development, we examined day P35 mice that had just finished the first wave of spermatogenesis. Light microscopic analysis of Chd5 KO mice revealed that spermatogenesis was normal up to step 9 of spermiogenesis (Figure 5). Meiotic division appeared to be normal and round spermatids appeared to form normal acrosomes. However, there was sloughing of round and elongated spermatids (Figure 5), resulting in a decreased number of germ cells in the seminiferous epithelium. Numerous abnormal heads were observed in the elongating spermatids, as well as a mixing of spermatid steps, suggesting that spermatid differentiation was disrupted. Electron microscopic analysis showed abnormal nuclear morphology of differentiating spermatids began in stage IX, step 9 spermatids. Abnormal condensation of chromatin, resulting in the presence of fibrillar and highly condensed chromatin within the same nucleus, was associated with abnormal indentation of the acroplaxome and attached acrosome (Figure 6).

Fig. 4. Chd5 deletion affects the chromatin histone replacement and the spermatid nucleus head formation.

(A), Showing the morphological Comparison of WT and Chd5 KO sperm. The sperm head of the WT mouse has a crescent shape, whereas sperm from the Chd5 KO mouse have an irregular head, with surrounding cytoplasm. (B) Acidic Aniline staining of cauda epididymal sperm. The WT sperm head remain unstained, whereas half of the Chd5 KO sperm heads were densely stained. (C) Total histone 3 was abundantly detected in Chd5 KO sperm but was virtually undetectable in WT mouse sperm.

Fig. 5. Spermatogenesis in Chd5−/− (KO) mice.

In KO mice, spermatogenesis appears normal through the round spermatids. However, elongated spermatids become abnormal and often fail to be released, resulting in epididymal lumens being devoid of sperm. Seminiferous epithelium in WT (+/+) (A–D) and Cdh5−/− mice (E–L). Bar=10µm for all photos. (A) WT stage I showing step 1 round (S1) and step 13 elongated spermatids (S13). (B) WT stage VI showing step 6 round (S6) and step 15 elongated spermatids (S15). (C) WT stage IX showing step 9 early elongating spermatids (S9). (D) WT stage XII showing spermatocytes in meiosis II division (MII) and step 12 elongated spermatids (S12). (E) KO stage I showing normal step 1 round (S1) and abnormal step 13 elongated spermatid heads (S13a). (F) KO stage X showing normal step 10 (S10) with abnormal spermatid heads (S10a). Note the indentation in the S10a nucleus. Abnormal of step 15 spermatids (S15a) are present with step 16 spermatids remaining in the epithelium after failure of spermiation (S16a). (G) KO stage X showing abnormal step 10 spermatid heads (S10a) and the abnormal presence of a step 16 spermatid head (S16a). (H) KO stage XII showing normal meiosis II (MII) and mostly abnormal step 12 spermatid heads (S12a), along with step 10 spermatids (S10a) that are out of proper stage. (I) KO stage II showing normal step 2 round (S2) and abnormal step 14 elongated spermatid heads (S14a). Sloughing of germ cells is noted by the areas of missing round spermatids (*). (J) KO stage IX showing some normal step 9 spermatid heads (S9), numerous abnormal step 9 heads (S9a), and highly condensed nuclei of step 16 spermatids that failed to spermiate (S16a). (K) KO stage IX showing normal step 9 spermatids, but also abnormal step 16 elongated spermatid heads (S16a). A large vacuole (*) suggests germ cells have been sloughed into the lumen. (L) KO stage X showing abnormal step 10 spermatid heads (S10a) and abnormal presence of step 16 spermatids (S16a).

Fig. 6. Abnormal formation of spermatid head.

(A) WT testis. Two elongated spermatids with dense chromatin condensation in their nuclei (N). The acrosome (Acr) is attached to the plasmalemma and the ectoplasmic specializations (Es) and a thin acroplaxome (Ax) separates the acrosome from the nuclear lamina. A portion of the proximal centriole (C) is seen attached to the basal plate of the nucleus. Bar=0.5 µm.

(B) Chd5−/− testis. Two elongating spermatids with less dense, fibrillar (f) chromatin in their nuclei (N). The ectoplasmic specialization (Es) appears normal in its attachment to the acrosome (Acr). The acroplaxome (Ax) appears normal on the outer edge of the nucleus, but thicker within the abnormal indentation (In). Bar=0.5 µm.

(C) Chd5−/− testis. An elongating spermatid with a nucleus (N) of abnormal shape, containing fibrillar (f) and condensed (d) chromatin. The ectoplasmic specialization (Es) appears normal in its attachment to the acrosome (Acr), except within the highly indented area, where there appears to be a break in the acrosomal/Es structure (*). The boxed area is the marginal ring with a groove that limits the tip of the acrosome. The perinuclear ring (arrowhead) appears normal but the manchette microtubules do not appear to bind normally to the ring (arrow). Bar=0.5 µm.

(D) Chd5−/− testis. An abnormal elongating spermatid with an abnormal nuclear (N) shape and heavy indentation (In). The acroplaxome (Ax) appears to extend (arrowheads) beyond the leading edge (small arrows) of the acrosome (Acr). The marginal ring with microtubular manchette is absent, but the proximal centriole (C) is present, as well as the annulus (An). Bars=0.5µm.

(E) Chd5−/− testis. An abnormal elongating spermatid with fibrillar (f) chromatin extending away from the condensed chromatin (d). The nuclear envelope (Ne) bulges beyond the chromatin. Bars=0.5µm.

(F) Chd5−/− testis. An abnormal elongating spermatid with an abnormal nuclear (N) shape and heavy indentation. The acroplaxome (Ax) and acrosome (Acr) are overgrown within the indentations, but the outer edge of the acrosome is attached to the ectoplasmic specialization (Es). Bars=0.5µm.

(G) Chd5−/− testis. An abnormal elongating spermatid with normal nuclear (N) shape, but abnormal extension of the ectoplasmic specialization (Es) beyond the border of the acrosome (Acr), overlying the perinuclear ring (Pr). The acrosome border is attached to a normal acrosomal plaque (Pq) and stops at the perinuclear groove (Gr). The centriole (C) appears to be growing abnormally and is located an excessive distance from the base of the nucleus (double arrow). Bars=0.5µm.

2.6. Chd5 KO mice showed a failure of spermiation

Spermatogenesis in mammals is finely tuned so that only certain steps of spermatid maturation will be identified in a section of a testicular tubule. Step 16 spermatids mature to form spermatozoa that are released into the lumen in stage VIII. In the Chd5 KO testes, most step 16 spermatids showed a failure of normal spermiation, appearing in subsequent stages, as Sertoli cells phagocytized these abnormal cells (Figure 5), or sloughing into the lumen.

2.7. Chd5 KO mice have abnormally high histone retention

Abnormal histone retention in the spermatozoa may cause defects in spermiogenesis and male infertility. Therefore, we examined the epididymal sperm from Chd5 WT and Chd5 KO mice. Acidic aniline was used to determine the chromatin compaction of sperm nuclei. All the sperm from Chd5 WT mice were morphologically normal, and did not stain with acidic aniline. In contrast, only 21% of the sperm from Chd5 KO mice were not stained, and 38% of the sperm heads from Chd5 KO mice were strongly stained with acidic aniline (Figure 4B). Western analysis for H3 total protein of the epididymal sperm chromatin also showed substantial retention of H3 protein in the Chd5 KO mice, whereas the Chd5 WT mice showed virtually no signal (Figure 4C and S2). Although the H4 hyperacetylation level was dramatically decreased in the testes of Chd5 KO mice (Figure 7), histone methylation (H3K4me3, H3K27me3) was not changed (Figure S2).

Fig. 7. H4 hyperacetylation not upregulated in elongating spermatids of Chd5 KO mice.

The hyperacetylation of H4 starts from stage IV, step 4, starting from the acrosomic granule. At stage VII–VIII, the signal spreads to the whole RS body but still keeps a cap on the spermatid. WT (A–C) and Chd5 KO (D–F) mouse testis slides stained with H4 hyperacetylation antibody (penta). In stage IV–VIII, round spermatid were faintly stained in both the WT and the Chd5 KO mice. After the spermatids entered the elongation phase, the staining became very dense (in A–C: IX, XI and XII) with XI staining most densely. In Chd5 KO, the staining intensity of H4 hyperacetylation after step 9 and beyond does not increase as dramatically as in the WT (D–F). Compare the stage X (in B vs E) and XII (in C and F). (G) We loaded 30 µg of whole testis protein on a 4–12% NuPAGE (Invitrogen) with MES running buffer (Invitrogen). H4 hyperacetylation in the WT is 5–10 times higher than those in the Chd5 KO mice. (H), Densitometry analysis of the H4 hyperacetylation signals in WT and Chd5 KO testes.

2.8. Histone H4 hyperacetylation was decreased in elongating spermatids of Chd5 KO mice

Because the Chd5 KO mice exhibited higher retention of histone, we investigated whether the modification of histones during the elongation steps was altered using an H4 hyperacetylation antibody. Core histone hyperacetylation precedes histone–>transitional protein–>protamine exchange in haploid spermatids and prepares an open chromatin status for the replacement of histones (Sonnack et al., 2002). We found that the H4 hyperacetylation level was comparable in Chd5 KO mice to WT mice, from step 4–8 (Figure 7A and 7D). From step 4 the round spermatids were stained slightly, concentrated in the developing acrosome. However, from step 9, the elongating spermatid in WT mice became very densely stained, reaching its peak in step 11 (Figure 7B and 7C). However, in Chd5 KO mice the staining in steps 9–12 showed greatly reduced intensity of staining compared to WT spermatids, and no significant difference compared to the level in step 8 (Figures 7E, 7F).

2.9. Chd5 KO did not alter the expression pattern of other spermatogenesis-related genes

We evaluated the expression of a number of other important spermatogenesis-related genes in the Chd5 KO mice. These include: acrosome biogenesis–Agfg1, GOPC; flagella formation–AKAP4, Teket2; nuclear condensation–Tnp2, Prm2; cytoplasm removal–spem1; later spermatid genes–Crem, TPAP; spermiation–RAR-alpha SOX8, WASL; cholesterol metabolism–ABCA1, ABCA3; hormone receptors–AR, FSHR, LHCGR, RXR-beta; ectoplasmic specialization complex–Arpc1a, Arpc1b, Actr3, Eps8; and CHD5 itself (Cheng and Mruk, 2011; Chung et al., 2005; D'Souza et al., 2009; Kashiwabara et al., 2002; Lie et al., 2009; McLachlan et al., 1996; O'Bryan et al., 2008; Ouvrier et al., 2009; Yan, 2009; Yan et al., 2010). No significant differences were seen in the expression levels of these genes between Chd5 WT and Chd5 KO mice (except for CHD5) (Figure 8).

Fig. 8. Expression of genes related to normal spermatogenesis in WT and KO mice testes.

There were no significant differences in gene expression levels between Chd5 WT and KO mice. Q-PCR of genes important in spermatogenesis was measured by real time-PCR. CHD5 was added to show the result of KO.

3. Discussion

CHD5 belongs to the nine-member CHD family (CHD1-CHD9), all of which have motifs that belong to the ATP-dependent chromatin-remodeling protein SNF2 superfamily (Hall and Georgel, 2007; Marfella and Imbalzano, 2007). Similar to other CHD family members, CHD5 has two N-terminal chromodomains and two SNF2-like ATPase central domains that bear the helicase function. Furthermore, CHD5, and its closest homologs CHD3 and CHD4, are the only members that have two PHD domains near the N-terminus, suggesting CHD5 belongs to subfamily II (Hall and Georgel, 2007; Marfella and Imbalzano, 2007; Thompson et al., 2003). Members of the CHD family play critical roles in organizing chromatin structure and in chromatin-based transcriptional regulation of genes (Hall and Georgel, 2007; Marfella and Imbalzano, 2007). In this report, we describe the consequences of constitutionally knocking out mouse Chd5, a CHD family member that is preferentially expressed in the brain and testis (Kolla et al., 2012; Potts et al., 2011). We focused on the role of CHD5 in the testis, as male mice with homozygous Chd5 KO were sterile.

Mouse spermatogenesis was seriously affected, showing failure of spermiation, failed up-regulation of H4 hyperacetylation in elongating spermatids, partial failure in chromatin condensation, abnormal sperm head morphology, immotility of epididymal sperm, and male infertility. In the Chd5 KO mouse testis, spermatogenesis appeared normal from spermatogonia through meiotic division of the spermatocytes, an indication that androgen stimulation was adequate for spermatogenesis (O'Donnell et al., 1999). Because the most significant phenotype observed in Chd5 KO mice was male infertility, we also investigated the expression of many of the genes that are known to control each individual step of spermatogenesis in our Chd5 KO mouse. Interestingly, there was no significant difference in the expression levels of any of these genes in Chd5 KO mice compared to Chd5 WT mice. In addition, we analyzed the expression level of key hormone receptor genes, and none of them showed significant differences. Although its close homologue, CHD4, is known to regulate gene expression in a variety of tissues (Hong et al., 2005), CHD5 does not seem to play such a role for the selected spermiogenesis genes we investigated.

Examination of histone retention levels showed that Chd5 KO mice had defects in replacing histones with transition proteins and ultimately with protamines. In the progress of chromatin condensation, the lysine-rich somatic histones are gradually substituted by testis-specific histones (transition proteins, TP) and finally by protamine (Gill-Sharma et al., 2011). As a consequence, protamine-chromatin is 30 times more condensed than the histone-chromatin (Oliva, 2006). There is a strong association between male infertility, incomplete spermiation and abnormal nuclear condensation (Okada et al., 2010; Perdrix et al., 2011; Ward, 2010; Yan, 2009). During spermiogenesis, the condensation of chromatin and the organization of DNA within the nucleus determine the final shape of the nucleus (Hammadeh et al., 1999). Indeed, there are many reports describing the association between disturbances in sperm chromatin condensation, morphology and male infertility (Lin et al., 2011; Sadek et al., 2011; Steilmann et al., 2010; Talebi et al., 2011).

H4 hyperacetylation helps chromatin to stay in an ‘open’ conformation, which facilitates histone-TP-protamine replacement (Sonnack et al., 2002). We observed failure of H4 hyperacetylation in Chd5 KO mice beginning at step 9 elongating spermatids. Since the replacement of histones peaks in steps 9–13 spermatids (Lahn et al., 2002; Mali et al., 1989), we speculate that the failure in H4 hyperacetylation in the elongating spermatid hinders the complete removal of histones, which leads to disturbed chromatin condensation in the sperm head nucleus without its critical anterior-posterior direction. Without CHD5, spermatogenesis initiates normally, but in elongating spermatids, histone replacement is incomplete, which results in improper chromatin condensation. This leads to aberrant sperm head morphology in the seminiferous tubules, and subsequent phagocytosis of most elongating spermatids by Sertoli cells.

Control of sperm head shape involves several important morphological structures, including the acroplaxome, attached acrosome and associated actin/keratin filaments, the microtubular manchette, and perinuclear rings, all of which interact with the Sertoli cell’s ectoplasmic specialization (Kierszenbaum et al., 2007; Kierszenbaum and Tres, 2004; Meistrich et al., 1990; Russell et al., 1991). The link between CHD5 and these structural elements remains unclear. However, disruption of transcription in spermatids has been shown after the inhibition of histone acetylase (Xia et al., 2012). Therefore, it is possible that loss of CHD5 interferes with either the synthesis of key proteins associated with these structures, or with the positioning of nuclear lamins or nesprins along the nuclear envelops of round spermatids (Polychronidou and Grobhans, 2011; Shimi et al., 2010; Taranum et al., 2012), and thus indirectly interferes with nuclear shape formation. This link between CHD5 and structural elements in the cytoplasm could explain the dramatically impaired motility seen in the limited number of mature sperm we identified in a few of the Chd5 KO mice. Furthermore, there is abundant expression of Chd4 in the mouse testis, so it is clear that this or perhaps other Chd family members cannot substitute for the critical functional role that Chd5 plays in spermatogenesis.

Our previous studies and reports of others have shown that CHD5 is a tumor suppresser gene that controls cell growth in neuroblastomas, as well as many other adult tumors (Bagchi et al., 2007; Fujita et al., 2008; Garcia et al., 2010; Mulero-Navarro and Esteller, 2008; Wang et al., 2011; Wang et al., 2009; Wong et al., 2011). Our findings strongly suggest that CHD5 deficiency causes a failure of developmentally regulated chromatin condensation during spermatogenesis. Therefore, a failure of chromatin condensation of selected domains of DNA in other tissues may also explain why CHD5 inactivation contributes to tumorigenesis. Very recently Egan and colleagues have demonstrated a new functional role of CHD5 in neurogenesis by CHD5 knockdown in the developing neocortex which lead to blocked neuronal differentiation and accumulation of undifferentiated neural progenitors (Egan et al., 2013). Moreover, it has also been shown that the chromodomains of CHD5 may directly bind H3K27me3 which is required for neuronal differentiation (Egan et al., 2013). Chd5 is most abundantly expressed in the nervous system and the testis. In humans, high CHD5 and other chromatin remodeling factors are associated with normal spermatogenesis, whereas decreased expression of these genes closely associated with round spermatid arrest (Steilmann et al., 2010). CHD5 may play different roles in different cell types, and so the suppression of cell growth and facilitating chromatin condensation are only two aspects of this protein’s chromatin-remodeling functions (Saito et al., 2010). There are many factors at work during chromatin condensation in elongating spermatids (Forgione et al., 2010; Kurtz et al., 2009; Steilmann et al., 2010), so it would be of great interest to identify the cofactor(s) that interact with CHD5 in these haploid cells. Additionally, elucidation of the chromatin element (histone, TP, protamine) ratios and their modification status (Awe and Renkawitz-Pohl, 2010; Song et al., 2011; Vigodner, 2011) would be of great help in understanding the role CHD5 plays in spermiogenesis.

4. Materials and methods

4.1. Generation of Chd5 KO mice

To generate Chd5 KO mice, 12 kb of the Chd5 locus spanning from intron 6 to intron 18 (nt20655 to 33246, NC_000070) was cloned and manipulated to construct a targeting vector. LoxP sites were inserted at nt27008 loxP1-GTCTCTAGATGAGGCTG….and nt28140 ….CCTGCCCAGCATCTAC- loxP2, and a PGK-neo cassette, with FRT sites on both ends, was inserted into the immediate 5’-region upstream of the 2^nd LoxP site (nt 33246) (Figure S1).

After targeting, the ES clones were screened by genomic Southern blotting. A probe near intron 3 (nt15577-15621, NC_000070) was used for MfeI digestion, and a probe near intron 24 to intron 26 (nt38227-39149, NC_000070) was used for SpeI digestion. Then, 20 µg of genomic DNA was digested with either MfeI or SpeI, and separated on a 0.7% agarose gel. Transfer and hybridization were carried out with the DIG EZ Hyb kit (Roche) using the protocol provided. The expected bands of MfeI digestion are: WT 12.6 kb, targeted band 14.6 kb; the SpeI digested bands are: WT 13.3 kb, targeted band 11.3 kb (there is an SpeI site introduced by the insertion of the 5’ FRT site).

Female mice with the targeted allele were mated with male mice carrying Prm-cre (129-Tg(Prm-cre)580g/J (Jackson Lab, stock# 003328) to delete exons 12 and 13 in the male offspring. They were mated with WT 129-E mouse (Charles River) to generate constitutive Chd5+/− mice. Genomic Southern blot and PCR were used to genotype the constitutive Chd5 KO allele. MfeI digestion of the WT DNA produced a 12.6 kb band, whereas the Chd5 constitutive allele produced an 11.5 kb band. For PCR genotyping, primer intron 11 mF1 (cctggctggttgctgaatcta) and primer exon 12 mR1 (ccttccacttgatcaggtaa) amplify a band of 580 bp for the WT allele, but there is no amplification for KO allele. Primer intron 11 mF1 and primer intron 13 mR1 (gccaaagataaccctgaatg) amplify a 321 bp band for the Chd5 KO allele, whereas amplification of the WT allele yields a 1,300 bp product. Chd5 heterozygous male and female mice were mated to generate Chd5 +/+, +/−, −/− mice. The work described herein was approved by the Children’s Hospital of Philadelphia (CHOP) Institutional Animal Care and Use Committee (IACUC protocol #2009-8-776).

4.2. Southern Analysis

Genomic DNA from ES cells or mouse tail was digested with MfeI or SpeI, separated on a 0.7% agarose gel, treated with 0.5 M NaOH/1.5 M NaCl for 45 minutes and neutralized in 1M Tris(pH8.0)/1.5 M NaCl for 30 minutes twice (Zhuang et al., 2004). Transfer and hybridization were carried out with the DIG EZ Hyb kit (Roche) as per the manufacturer’s instructions.

4.3. Real-time polymerase chain reaction (PCR)

Real-time PCR primers located in exons 19 and 20 of the Chd5 gene were designed with information obtained from Origene (http://www.origene.com/qPCR/primers.aspx). PCR reactions with SYBR Green (Applied Biosystems) were carried out with an ABI 7900HT Fast Real-Time PCR System. Mouse Chd5 primer sequences were as follows: Forward sequence (in exon 19): TGCAACCATCCGTACCTCTTCC, reverse sequence (in exon 20): TCAGCACTCTGTGCCCTTCATC.

4.4. Protein Analysis

Mouse testes from adult Chd5 −/− and WT mice were dissected, and the capsule was removed. Cytoplasmic and nuclear protein fractions were extracted with the NE-PER Nuclear and Cytoplasmic Extraction Reagents (Thermo Scientific), following the provided protocols. Proteins were snap frozen and kept −80°C until use. Total protein concentration was measured using the Bradford assay (BioRad), and 20–40 µg of total protein was separated on a 4–12% NuPAGE gel (Invitrogen) or 15% PAGE gel and was immunoblotted with the H4 Hyperacetylation ab (Millipore 06-946) and total H3 ab (Cell Signaling, #3680), respectively. Coomassie brilliant blue (CBB) staining of the membrane after Western or total H3 was used as a loading control.

4.5. Tissue Preparation for Light and Electron Microscopy

We anesthetized 35-day-old Chd5-deficient and wild-type male mice with CO₂, and the testes and epididymis were harvested and fixed with formalin for 24 hrs, or Bouin fixative (Polysciences Inc.) for 18hrs. These tissues were processed, and embedded in paraffin. Sections (5 µm) were stained with hematoxylin and eosin or periodic acid–Schiff for histological analysis (Dass et al., 2007).

For electron microscopy, testes were fixed with 2.5% glutaraldehyde, 2.0% paraformaldehyde in 0.1 M sodium cacodylate buffer, (pH 7.4) overnight at 4°C. After subsequent buffer washes, the samples were post-fixed in 2.0% osmium tetroxide for 1 hour at room temperature, and rinsed in dH₂O prior to en bloc staining with 2% uranyl acetate. After dehydration through a graded ethanol series, the tissues were infiltrated and embedded in EMbed-812 (Electron Microscopy Sciences, Fort Washington, PA). Thin sections were stained with uranyl acetate and lead citrate and examined with a JEOL 1010 electron microscope fitted with a Hamamatsu digital camera, and images were analyzed with AMT Advantage image capture software.

4.6. Aniline staining

Acidic aniline staining (AS) was used to determine the chromatin compaction of sperm nuclei (Hammoud et al., 2009). Briefly, sperm from the cauda epididymis were collected, and examined for motility under phase-contrast microscopy. The remainder were then spread on slides, and air-dried overnight. Specimen were fixed in 10% formalin for 30 minutes and washed with PBS, then treated with 0.2% Triton X-100 for 15 minutes. Slides were stained for 5 min using acidic aniline blue staining (5% methyl blue in 100 ml 4% acetic acid). We counted 200 spermatozoa on each slide in order to determine the percentage of stained spermatozoa. The spermatozoa were classified as either unstained or partially/completely stained.

4.7. Immunohistochemistry

Hyperacetylation of H4 was detected by immunohistochemistry in mouse testicular paraffin-embedded sections fixed in Bouin. Sections of 5 µm were mounted on Super-Frost_Plus glass. After deparaffinization, heat-induced antigen retrieval was performed at a controlled temperature in a microwave processor in 10 mM citrate buffer, pH 6.0, for 10 minutes at 97°C. Immunohistochemistry was performed using a three-step indirect process based on the labeled-(strept) avidin-biotin (LAB-SA) peroxidase complex method. The polyclonal rabbit antisera anti–Hypac-H4 (penta) (Millipore) was used as the primary antibody diluted 1:500. Immunohistochemistry was performed using the I-View DAB detection kit. After rinsing, slides were incubated with the avidin biotin complex (Vector Laboratories) for 30 min at room temperature. Slides were then rinsed and incubated with DAB (DAKO Cytomation) for 10 min at room temperature. Slides were then rinsed and counterstained with Hematoxylin (Fisher Scientific) for 30 seconds, then rinsed, dehydrated through a series of ascending concentrations of ethanol and xylene, and examined under a coverslip.

Supplementary Material

01. Gene targeting map.

A PGK-neo was inserted between the two FRT sites. The 5’-genomic Southern probe was designed at the 5’ side of this map, and is not shown in this map. The 3’-probe was designed 5’-of the endogenous SpeI restriction site.

02. Western with histone 3 antibodies on testis whole protein and cauda epididymis spermatid protein.

The total H3 antibody showed that the H3 content in the WT and Chd5 KO mice are expressed at the same level. However, although the WT spermatozoa showed almost no H3 signals, the Chd5 KO spermatozoa were strongly stained. However, assessment of H3K27me3 and H3K4me3 showed no significant difference between the Chd5 KO and WT sperm chromatin.

• Homozygous knockout of CHD5 causes infertility in male mice

• CHD5 expression is essential for normal spermiogenesis, beginning in step 9

• Absence of CHD5 expression leads to a defect in chromatin condensation

• There is a dramatic reduction of H4 hyperacetylation in elongating spermatids

• There is a failure of histone to protamine transition and abnormal spermiogenesis