Promoter ChIP-chip analysis in mouse testis reveals Y chromosome occupancy by HSF2

Abstract

The mammalian Y chromosome is essential for spermatogenesis, which is characterized by sperm cell differentiation and chromatin condensation for acquisition of correct shape of the sperm. Deletions of the male-specific region of the mouse Y chromosome long arm (MSYq), harboring multiple copies of a few genes, lead to sperm head defects and impaired fertility. Using chromatin immunoprecipitation on promoter microarray (ChIP-chip) on mouse testis, we found a striking in vivo MSYq occupancy by heat shock factor 2 (HSF2), a transcription factor involved in spermatogenesis. HSF2 was also found to regulate the transcription of MSYq resident genes, whose transcriptional regulation has been unknown. Importantly, disruption of Hsf2 caused a similar phenotype as the 2/3 deletion of MSYq, i.e., altered expression of the multicopy genes and increased mild sperm head abnormalities. Consequently, aberrant levels of chromatin packing proteins and more frequent DNA fragmentation were detected, implying that HSF2 is required for correct chromatin organization in the sperm. Our findings define a physiological role for HSF2 in the regulation of MSYq resident genes and the quality of sperm.

The mammalian Y chromosome is essential for spermatogenesis and sex determination, and contains mainly heterochromatin and only a few genes (1). It is the smallest of all chromosomes and consists mostly of a male-specific region, in addition to short pseudoautosomal regions (PAR), which are homologous to the regions of the X chromosome required for sex chromosome pairing (1). In the male-specific region of the mouse Y chromosome long arm (MSYq), a few genes exist in hundreds of copies (2). These genes are expressed predominantly in testis, and their multicopy nature is suggested to be a defense mechanism against degeneration in a non-recombining environment (2). To date, the transcriptional regulation of the multicopy MSYq resident genes is unknown.

Heat shock factor 2 (HSF2) is a transcription factor involved in mammalian spermatogenesis (3–6). HSF2 belongs to a transcription factor family, the members of which were originally found to regulate the heat shock response and later also revealed to orchestrate development (7). In addition to spermatogenesis, the only other developmental process where HSF2 is known to be active is corticogenesis (5, 6, 8). Although HSF2 exists in many tissues, it is most abundantly expressed in testis (9). During spermatogenesis, HSF2 is expressed in a stage-specific manner in the nuclei of early pachytene spermatocytes and postmeiotically in round spermatids (3, 4). Disruption of Hsf2 causes reduced size of testis and epididymis, altered morphology of the seminiferous tubules displaying extensive vacuolization, and a low number of differentiating spermatids due to elevated apoptosis at the pachytene stage (5, 6). In addition, the synaptonemal complex is disorganized in Hsf2^−/− pachytene spermatocytes (5). However, no correlation between HSF2 and expression of the classical HSF targets, Hsps, has been found (3–6), and the HSF2 target genes in spermatogenesis have remained obscure.

High-resolution chromatin immunoprecipitation on microarray (ChIP-chip) screens have successfully been used for identifying direct target genes for many transcription factors (10). For example, ≈3% of the genomic loci were found to be targets for HSF in Saccharomyces cerevisiae and Drosophila exposed to heat stress (11, 12). In mammals, however, the existence of three differently expressed HSFs (HSF1, HSF2, and HSF4) requires a strategy to investigate each HSF in a tissue-specific manner. Here, we chose to dissect the specific role for HSF2 in spermatogenesis and to map the in vivo targets for HSF2, by using mouse testis in a promoter ChIP-chip screen. We identified a multitude of target genes, and analysis of their chromosomal distribution led to an interesting discovery that the Y chromosome is predominantly occupied by HSF2 in testis. Accordingly, HSF2 regulates the transcription of the Y-chromosomal genes critical for sperm differentiation. Functional analyses of HSF2-deficient mice revealed increased sperm head anomalies, showing striking similarity to those in MSYq deletion mutants. Moreover, sperm lacking HSF2 displayed altered chromatin packing protein levels and more frequent DNA fragmentation, implying that HSF2 is required for correct chromatin organization during spermatogenesis.

Results

Global Mapping of Target Genes for HSF2 in Spermatogenesis.

To identify novel target genes for HSF2 in spermatogenesis, we cross-linked chromatin from three wild-type (WT) mouse testes and sonicated it into fragments of 100–500 bp. The quality of DNA was controlled before the immunoprecipitation and showed no signs of degradation [supporting information (SI) Fig. S1]. The DNA amplified from the HSF2 immunoprecipitation samples was labeled and hybridized against the total input DNA samples, on a first-generation 1.5-kb promoter tiling array from NimbleGen Systems, covering ≈26,000 promoters of the mouse genome. After hybridization and scanning, HSF2 hybridization signals were divided by the input signals to provide a value for enrichment for each oligonucleotide probe covering the promoters, on the three replicate arrays. Further, the target promoters were separately ranked in the replicates according to the average log₂-ratios of all probes covering each promoter. The complete data set is available at the Gene Expression Omnibus (GEO) database (http://www.ncbi.nlm.nih.gov/geo/, GSE9289). To identify the HSF2 target population, we used R/Bioconductor package RankProd (13) to define the significance value (P) for each promoter (see Materials and Methods). The data were filtered with P value <0.005, which resulted in identification of 546 target promoters for HSF2 (Table S1). In addition to RankProd, we used a nonarbitrary analysis method previously used for promoter ChIP-chip data by Squazzo et al. (14) (data not shown), and both methods revealed similarly ranked HSF2 target gene lists.

Validation of HSF2 Binding to Target Genes in Mouse Testis.

For verification of the target genes identified in the screen, we selected six putative HSF2 target promoters, distributed across the ranking from 1 to 546 (Fig. 1A). Three of these genes, Ssty2, Sly, and Slx/Xmr (hereafter called Slx), are located on the sex chromosomes, whereas Speer4a, Hsc70, and Ftmt reside on autosomal chromosomes. In Fig. 1A, the localization of the probes is indicated on each promoter, defining the peak of enrichment and thereby the putative region for HSF2 binding, in proximity to the transcription start sites. HSFs bind to heat shock elements (HSEs) in their target promoters, and an HSE consists of inverted repeats of the NGAAN motif (15), where guanidines are the most conserved nucleotides (16). A potential HSE was indeed found at the site of HSF2 enrichment on all six promoters (Fig. 1A). The HSF2 binding to the target promoters was validated in a standard in vivo ChIP assay by using a different HSF2 antibody than in the original ChIP-chip screen (Fig. 1B). In addition, three promoters; Hsp25.1, Spata 2, and Fyn, were verified as nontargets for HSF2 (Fig. 1B), based on their low ranking in the screen (data not shown). These results prove that the screen was reliable and successfully performed.

Fig. 1.

In vivo HSF2 binding to novel target genes in testis. (A) Visualization of the HSF2-binding profile on six selected target promoters, using the SignalMap software (NimbleGen Systems): Ssty2, Sly, Slx, Speer4a, Hsc70, and Ftmt. The localization of 15 probes per promoter is indicated as bars above the promoters, determining the HSF2 enrichment. One representative promoter is displayed for the multicopy genes. A putative heat shock element (HSE) is indicated below each promoter. Asterisk indicates key nucleotides required for HSF binding; arrows indicate primers used in the ChIP assay. Log2, log₂ ratio of HSF2 enrichment, indicated as difference in positions of the bar in the log₂ scale; Chr., chromosome; +1, transcription start site (note that the genes are transcribed in different directions). (B) Verification of the ChIP-chip screen using a standard ChIP assay of WT testis extracts. Analysis of HSF2-binding (HSF2) on the six selected promoters, in addition to three nontarget promoters for HSF2: Hsp25.1, Fyn, and Spata 2. (C) Target genes are occupied by HSF2 only in testis. ChIP analysis of HSF2-binding (HSF2) on six target promoters in WT testis, brain, muscle, and kidney. Nonspecific antibody (NS) was used as a negative control, and acetylated histone 4 antibody (AcH4) was used as an indicator of transcriptionally active promoters. Input represents 1% of the total material used in the ChIP assay.

To study whether the newly identified target genes were also occupied by HSF2 in other tissues, we analyzed the HSF2 binding in testis, brain, muscle, and kidney. The ChIP analyses showed that the target promoters were occupied by HSF2 only in testis (Fig. 1C), indicating that the promoter sequence alone is not sufficient to specify HSF2 binding. As supporting evidence, the same Hsp25 promoter sequence that contains a canonical HSE was not an HSF2 target in testis (Fig. 1B), but has earlier been shown to be bound by both HSF1 and HSF2 in mouse embryonic fibroblasts exposed to heat stress (17). Thus, the ChIP-chip screen in testis made it possible to identify a multitude of HSF2 targets in spermatogenesis, highlighting the importance of performing a global search for target genes in the physiologically relevant context.

The Y Chromosome Is Occupied by HSF2.

A possible formation of HSF2-binding clusters was studied through the chromosomal distribution of target genes by using the GeneMerge analysis tool (18). To our surprise, a profound accumulation of HSF2 targets was detected on the Y chromosome, in comparison to the total amount of promoters per chromosome on the microarray (Table 1). Interestingly, 34 HSF2 target genes residing in the Y chromosome were multiple copies of Ssty2, Sly, and similar to Ssty2 (LOC435023) (Table S1). The accumulation was evident also when the HSF2 targets were analyzed in a single copy (Table S2). The biological processes directly associated with the novel target genes were investigated with the DAVID analysis tool (19). Target genes in a single copy were selected (P value <0.0001, Table S1), including all Y-chromosomal HSF2-bound genes, and reproduction (P = 0.011) was the highest ranked biological process that was significantly enriched (Table S3). These results strongly indicate that HSF2 regulates target genes associated with male reproduction.

Table 1.

The chromosomal distribution of HSF2 target genes

Chromo-some	Population fraction*	Study fraction†	P‡	e-score§
1	1,254	29	0.313003	1
2	1,933	29	0.979376	1
3	1,067	16	0.937303	1
4	1,300	20	0.941805	1
5	1,291	29	0.370752	1
6	1,182	25	0.504441	1
7	2,566	29	0.282118	1
8	1,092	14	0.983677	1
9	1,272	25	0.653106	1
10	1,032	12	0.991928	1
11	1,652	27	0.927213	1
12	727	15	0.556808	1
13	915	14	0.912291	1
14	787	22	0.103817	1
15	830	18	0.469905	1
16	698	12	0.792606	1
17	1,051	21	0.614983	1
18	554	7	0.94583	1
19	745	12	0.856506	1
X	806	22	0.123984	1
Y	105	35	1.22E-32	2.8E-31
Mt	4	1	NA	NA
Un	3,266	83	0.033699	0.775067

NA, not applicable; Mt, mitochondrial DNA; Un, undefined chromosomal location.

*Number of genes located on a certain chromosome, of the total number of genes on the array (26,129).

^†Number of HSF2 target genes located on a certain chromosome, of the total number of HSF2 target genes analyzed in this study (546, with a P value <0.005).

^‡The probability of HSF2 occupancy on the promoters in each chromosome by random chance.

^§Bonferroni-corrected P value.

Ssty2, Sly, and similar to Ssty2 exist in multiple copies throughout the MSYq region (2). Sly displays substantial sequence homology to another multicopy gene on the X chromosome, Slx (2, 20), which was also identified as an HSF2 target (Fig. 1B, Table S1). Accordingly, HSF2 occupancy was observed on numerous copies of the Slx promoter on the X chromosome (Table S1). Promoter variants of each multicopy gene, Ssty2, Sly, and Slx, were defined by NimbleGen Systems, and ChIP-chip probes were designed for the unique promoter variants. The multiple copies of Ssty2, Sly, and Slx were differently ranked on the HSF2 target list, depending on the proximity of the ChIP-chip probes to the putative HSE (for description, see Fig. S2). We visualized the HSF2 binding at a chromosomal level by using the SignalMap software (NimbleGen Systems) (Fig. S3). Our results demonstrate HSF2 binding to multiple copies of Ssty2, Sly, and Slx, which is also supported by the ChIP assays showing a more intensive HSF2 enrichment on the multicopy genes than on the other targets (Fig. 1B).

HSF2 Functions as a Transcriptional Regulator of Ssty2, Sly, and Slx in Spermatogenesis.

In mice, the cycle of the seminiferous epithelium is composed of 12 stages (I–XII), and each stage contains a defined collection of cell types, which are classified by the morphology of the developing spermatids (21) (Fig. 2A). To characterize the spatiotemporal relationship between expression of Hsf2 and the multicopy genes, we isolated stages IX–XI, XII–I, II-VI, and VII–VIII from WT seminiferous epithelium as described by Kotaja et al. (21). Hsf2 mRNA was most abundant in stages XII–I and II–VI, whereas Ssty2, Sly, and Slx mRNAs were found at high levels in stages II–VI and VII–VIII (Fig. 2A). These stages contain round spermatids (21), and our findings are in accordance with those of previous studies showing that Ssty, Sly, and Slx transcripts are present only during spermatogenesis, predominantly in round spermatids (2, 20, 22, 23). Importantly, the Hsf2 expression coincided with transcription of the multicopy genes in stages containing round spermatids (II–VIII) (Fig. 2A). In addition, Hsf2 mRNA was found in earlier stages of the epithelial cycle (XII–I) (Fig. 2A), which probably is a consequence of its expression in the pachytene spermatocytes (3, 4). Next, using WT and Hsf2^−/− testes, we investigated whether HSF2 is required for the transcription of the multicopy genes. Due to enhanced apoptosis, the Hsf2^−/− testes contain fewer spermatids (5) and the stage patterns are more diffuse than in the WT; therefore the mRNA expression was studied in whole testis instead of stages and normalized to round spermatid-specific Acrv1/SP-10 (24). Interestingly, Ssty2 and Sly mRNAs were significantly reduced, whereas Slx mRNA was increased in the Hsf2 knockout (Fig. 2B). Taken together, our results imply that HSF2 is a stage-specific transcriptional regulator of the multicopy genes in spermatogenesis.

Fig. 2.

HSF2 functions as a transcriptional regulator of multicopy gene expression in spermatogenesis. (A) RT-PCR analysis of Hsf2, Ssty2, Sly, and Slx expression in the indicated stages of WT seminiferous epithelial cycle was performed. Relative quantities of mRNA were normalized to Gapdh, which was evenly expressed throughout the seminiferous epithelial cycle. A schematic presentation of the 12 stages (I–XII) in the mouse seminiferous epithelial cycle is shown below. Each stage is defined by a specific collection of cell types, which are classified by the morphology of the developing spermatids (21). (B) RT-PCR analysis of gene expression in whole WT (Hsf2 WT) and Hsf2 knockout (Hsf2 KO) testes. Relative quantities of mRNA were normalized to Acrv1/SP-10. All PCRs were in duplicates using samples derived from at least three biological repeats. Error bars denote standard deviations (±SD). The relative expression was calculated from the Hsf2 WT sample, which was arbitrarily set to 1.

Hsf2^−/− Mice Display Increased Sperm Head Abnormalities, Impaired Protamine Expression, and Prominent DNA Fragmentation.

In mice, a 2/3 deletion of the MSYq region (XY^RIIIqdel) causes lowered transcript levels of Ssty2 and Sly in testis, but increases Slx levels (2, 25). The changes in expression of the multicopy genes are strikingly similar to those detected in the Hsf2^−/− testis (Fig. 2B). Moreover, XY^RIIIqdel mutant mice display abnormalities in the sperm heads (23, 25, 26), which has been interpreted to reflect incorrect chromatin organization in the nucleus (2, 25). We examined the Hsf2^−/− sperm morphology by assessing the degree of sperm abnormalities in hematoxylin-stained sperm smears from WT and Hsf2^−/− males. Interestingly, mild but consistent defects were observed in Hsf2^−/− sperm heads, manifested by flattened heads with a less hydrodynamic structure (Fig. 3A). To define the severity of the head distortion, we calculated the percentage of normal, slightly abnormal and grossly abnormal sperm. A marked increase in slightly abnormal heads was found in the Hsf2^−/− sperm, in comparison to the WT, whereas gross sperm head defects occurred equally in both WT and knockout males (Fig. 3B). Our results show that in addition to the similar alterations in Ssty2, Sly, and Slx expression, the Hsf2^−/− head abnormalities were equivalent to the flattened sperm head phenotype of the XY^RIIIqdel mutant (23, 25, 26).

Fig. 3.

HSF2-deficient mice display a significant increase in sperm head abnormalities. (A) Analysis of hematoxylin-stained sperm smears from adult WT (Hsf2 WT) and Hsf2 knockout (Hsf2 KO) males. Representative examples of different morphology are shown in the main figure, and a blow-up in the inset. (Scale bar, 10 μm.) (B) The sperm heads were classified in three categories of normal, slightly abnormal, and grossly abnormal, as previously described (23). The number of sperm in each category, obtained from Hsf2 WT (n = 4) and Hsf2 KO (n = 4) male mice, was calculated in blind. Error bars denote standard deviations (±SD).

The Hsf2^−/− sperm head anomaly (Fig. 3A) suggests defects in the chromatin condensation process, which requires replacement of histones first with transition proteins and finally with protamines at the end of spermatogenesis (27). This prompted us to analyze the amounts of transition proteins and protamines incorporated in WT and Hsf2^−/− sperm. We observed profoundly more transition protein 2 (TNP2), but substantially less protamine 1 and slightly less protamine 2 in the Hsf2 knockout epididymis (Fig. 4A). The difference is not due to sloughing of immature germ cells from the seminiferous epithelium of Hsf2^−/− mice, as demonstrated by the morphological and cytological analyses (Fig. S4). Thus, our data imply that the replacement of transition proteins with protamines is disturbed in the Hsf2 knockout mice. The changes in chromatin packing proteins levels are probably an indirect effect originating from earlier steps of spermatogenesis, given that neither TNP2 nor protamines were direct targets for HSF2 (Table S1). Even small changes in chromatin packing proteins have vital consequences and compromise the chromatin remodeling (28), by affecting the DNA status. Therefore, we assessed the DNA fragmentation in Hsf2^−/− sperm by performing a comet assay (29). More extensive DNA fragmentation was observed in the Hsf2^−/− than in WT sperm (Fig. 4B). To quantify the amount of sperm with DNA damage, we calculated the percentage of positive sperm comets. A two-fold increase in comets was found in Hsf2^−/− sperm compared with WT sperm (Fig. 4C). The analyses of the Hsf2 knockout phenotype, i.e., sperm head morphology defect together with altered chromatin packing protein levels and increased DNA damage, indicate that HSF2 is critical for sperm differentiation and for correct packing of the chromatin in the male germ cells.

Fig. 4.

Altered chromatin packing protein levels and increased DNA fragmentation in Hsf2 knockout sperm. (A) Western blot analysis of transition protein 2 (TNP2), protamine 1 (PRM1), and protamine 2 (PRM2) levels in cauda epididymis isolated from adult WT (Hsf2 WT) and Hsf2 knockout (Hsf2 KO) mice. The blots are representative examples of four biological repeats. Equal loading was assessed by β-tubulin and α-actin. (B) Analysis of DNA damage by comet assay. After separation of DNA fragments by electrophoresis, the sperm DNA from adult Hsf2 WT and Hsf2 KO males was stained with SYBR green. Representative examples of comets are shown with white arrows in the figure. (Scale bar, 10 μm.) (C) The percentages of sperm positive for DNA damage after analysis by comet assay were calculated from Hsf2 WT (n = 3) and Hsf2 KO (n = 3) mice. Error bars denote standard deviations (±SD).

Discussion

Our ChIP-chip screen on mouse testis revealed that a high number of the HSF2 targets are located on the Y chromosome (Table 1) and consist of the multicopy MSYq resident genes Ssty2, Sly, and similar to Ssty2 (Fig. S3A and Table S1). Previous transcriptome analyses of mouse MSYq deletion mutants have identified 23 down-regulated genes, of which 15 were recognized as copies of Ssty2 and 5 as copies of Sly (2). Similarly, both Ssty2 and Sly mRNA levels were lowered in the Hsf2^−/− testis (Fig. 2B). We conclude that HSF2 directly binds and regulates the transcription of the Y-chromosomal multicopy genes Ssty2 and Sly (Figs. 1B and 2B) in addition to the X-chromosomal multicopy relative Slx (Figs. 1B and 2B). Although Sly is down-regulated in Hsf2^−/− and XY^RIIIqdel mutant mice, its X-linked homologue Slx is up-regulated in both mouse models (Fig. 2B, ref. 25). However, the mechanism by which Slx transcription is affected in these mice might be different. Importantly, the expression of Hsf2 mRNA seems to coincide with transcripts of the multicopy genes in round spermatids (Fig. 2A) (2–4, 20, 22, 23), where the chromatin undergoes a substantial remodeling process (27). The MSYq resident genes have been proposed to have a role in chromatin remodeling (2, 23), indicating that HSF2 could, through direct regulation of Ssty2 and/or Sly, be involved in chromatin remodeling during spermatogenesis.

Deletions in the MSYq region have emerged as the most common genetic cause for spermatogenic failures in the human population worldwide, resulting in oligo- or azoospermia (30). In mice, deficiency in MSYq transcripts deriving from mouse-specific multicopy genes causes abnormalities in the sperm heads, and MSYq deletion mutants display a range of teratozoospermia and infertility phenotypes with severities corresponding to the extent of the deletion (23, 25, 26, 31–33). The increase in flat sperm heads is specific to MSYq deletions in mice, and even minor changes in the sperm head structure can affect sperm motility and thereby fertility (26, 32, 33). Accordingly, the mild but consistent head defects found in the Hsf2^−/− sperm (Fig. 3) could have severe consequences, although some sperm retain the fertilization capacity. The nuclear status of sperm cells is determined by two major events that occur at the final stages of spermatogenesis: replacement of transition proteins with protamines and acquisition of the correct sperm head shape. It would be important to decipher whether the MSYq multicopy genes are involved in the regulation of the replacement process. The molecular and morphological changes are prerequisite for efficient packaging of the sperm chromatin and influence the chromatin stability, which have direct effects on fertility (28, 34). Altered chromatin packing protein levels were observed in the Hsf2^−/− epididymes, together with increased sperm fragmentation (Fig. 4), implying impaired sperm quality. Protamine deficiency in mature sperm from both mice and humans has been reported as a manifestation of DNA damage and fertility problems, as the protamine structure protects the genetic material from physical and chemical damage (35, 36).

In this study, we present a promoter ChIP-chip screen focusing on the HSF2 target genes in a specific mammalian tissue, i.e., mouse testis. This approach revealed a striking HSF2 occupancy on multicopy genes within the MSYq region (Fig. 5). We discovered a physiological role for HSF2 in spermatogenesis, as our results provide evidence for transcriptional regulation of the MSYq resident genes. Based on the accumulation of HSF2 on the MSYq and the phenotypic characterization of Hsf2^−/− sperm, we propose that the HSF2-mediated transcriptional regulation of MSYq multicopy genes could be crucial for proper chromatin organization and sperm quality. Our results expand the understanding of spermatogenetic anomalies and may open avenues for future research on male fertility.

Fig. 5.

A schematic presentation of HSF2 occupancy on the male-specific Y chromosome long arm (MSYq). The majority of the Y chromosome genes are located on the short arm (Yp) and were not found as HSF2 targets, whereas the MSYq mostly contains heterochromatin and repetitive sequences. All well annotated Y-chromosomal genes included in the ChIP-chip array are indicated in the figure. Note that the chromosome length and the number of HSF2 molecules are only illustrative. PAR, pseudoautosomal region.

Materials and Methods

Mice.

Male hybrid mice of the B6129SF2/J strain were used in the ChIP-chip screen. Hsf2 knockout mice were obtained by matings of heterozygous mice that have been described earlier (5), and were maintained in a C57BL/6N background. The pathogen-free mice were housed under controlled environmental conditions and fed with complete pellet chow and allowed tap water. The mice were killed by CO₂ asphyxiation. All mice were handled in accordance with the institutional animal care policies of the Åbo Akademi University (Turku, Finland). Adult (60–80 days old) mice were used for isolation of testes.

ChIP.

Testes were isolated and lysed in 4 ml of buffer, and the ChIP assay was performed as earlier described (8). Antibodies are described in SI Materials and Methods.

DNA Amplification for ChIP Experiments.

PCR analysis was performed on 1/10 of each ChIP sample using puRe Taq Ready-to-Go PCR Beads (GE Healthcare). For ChIP primer sequences see SI Materials and Methods.

DNA Amplification and Microarray Hybridization for ChIP-Chip Experiments.

DNA amplification of material obtained from three biological replicates for the microarray hybridizations was performed by using a protocol from NimbleGen Systems. Whole ChIP samples and 20 ng of the input samples were used for the ligation-mediated PCR (LM-PCR). The DNA was blunted by using dNTP mix (Promega) and T4 DNA polymerase (New England Biolabs), purified and dissolved in water. Purified DNA was ligated by using T4 DNA ligase (New England Biolabs) and annealed with linkers made from HPLC-purified oligonucleotides: oligo 1, 5′-GCG GTG ACC GGG AGA TCT GAA TTC-3′ and oligo 2, 5′-GAA TTC AGA TC-3′. DNA was purified again and dissolved in water. LM-PCR was performed by using Taq DNA polymerase (New England Biolabs) and Pfu DNA polymerase (Stratagene). An aliquot of the final DNA was separated on an agarose gel for verification of the DNA fragment sizes. The experimental HSF2 amplicons were labeled with Cy5 dye, and the total input amplicons were labeled with Cy3 dye (including one dye-swap) and then cohybridized to high-density oligonucleotide tiling arrays. The HSF2 ChIP signal was compared with control input signal and the data were extracted according to standard operating procedures by NimbleGen Systems (www.nimblegen.com).

ChIP-Chip Data Analysis.

The two-channel raw data were normalized between channels with the Lowess normalization method, and ChIP-to-input log₂-ratios were produced separately from all three replicates. The target promoters were separately ranked in the three replicates according to the average log₂ ratios of all probes covering each promoter. The log₂ ratios in the replicates showed positive correlation, resulting in Pearson's correlation values between 0.34 and 0.41. The R/Bioconductor package RankProd (13) was used for determining the reliability of the bound promoters. RankProd provides an average log₂ ratio and a P value for each promoter from the separately ranked promoters of the three replicates (Table S1). The data were filtered with P value <0.005, which resulted in a list of 546 HSF2 bound promoters (Table S1).

Quantitative Real-Time RT-PCR.

Whole WT and Hsf2^−/− testes or stages IX–XI, XII–I, II—VI, and VII–VIII of WT seminiferous epithelial cycle were isolated as described by Kotaja et al. (21). The RT-PCR reactions were prepared and run as earlier described (17). Relative quantities of the target gene mRNAs were normalized against Acrv1/SP-10 or Gapdh, and the fold induction from WT samples was calculated. All reactions were in duplicates using samples derived from at least three biological repeats. For primer and probe sequences, see SI Materials and Methods.

Analysis of Sperm Head Morphology and DNA Fragmentation.

Adult WT and Hsf2^−/− male mice were killed, the caudae epididymes were isolated, and the sperm was released into PBS by using small surgical scissors. Diluted sperm smears of Hsf2^−/− (n = 4) and WT (n = 4) mice were spread onto microscope glass slides to dry. The dried slides were fixed in 4% paraformaldehyde, followed by hematoxylin staining. The glass slides were coded and randomized, and 300–500 sperm heads from each slide were analyzed by using light microscopy, in blind by three persons. The sperm heads were classified into three categories: normal, slightly abnormal, and grossly abnormal, as previously described by Touré et al. (23).

Western Blot Analysis.

Caudae epididymes from WT and Hsf2^−/− males were lysed in 3× Laemmli buffer followed by boiling, and subjected to SDS/PAGE followed by transfer to nitrocellulose membrane (Protran nitrocellulose, Schleicher & Schuell). Proteins from four biological repeats were analyzed. Antibodies are described in SI Materials and Methods. The blots were developed with an enhanced chemiluminescence method (ECL kit, Amersham Biosciences).

Analysis of Sperm DNA Fragmentation.

The single-cell gel electrophoresis, called “comet assay,” was previously described by Sakkas et al. (29). Sperm was isolated as for sperm analysis of the head morphology, and diluted sperm smears of Hsf2^−/− (n = 3) and WT (n = 3) mice were used. DNA from 300–500 sperm from each slide was analyzed, and the percentage of positive sperm comets was calculated.