Molecular Human Reproduction Vol.7, No.10 pp. 903–911, 2001

Sperm nuclear matrix association of the PRM1→PRM2→ TNP2 domain is independent of Alu methylation

Carl Schmid1, Henry H.Q.Heng2,3, Carol Rubin1, Christine J.Ye4 and Stephen A.Krawetz2,5,6

1Section of Molecular and Cellular Biology, University of California, Davis, Davis, CA 95616, 2Center for Molecular Medicine and Genetics, 3Department of Pathology and the Karmanos Cancer Institute, Wayne State University School of Medicine 4SeeDNA Biotech Inc., Windsor, Ontario, Canada, 5Department of Obstetrics and Gynecology and the Institute for Scientific Computing, Wayne State University, Detroit, MI 48201, USA 6To whom correspondence should be addressed at: Department of Obstetrics & Gynecology, Center for Molecular Medicine & Genetics, Institute for Scientific Computing, Wayne State University School of Medicine, 253 C.S.Mott Center, 275 E. Hancock, Detroit, MI 48201, USA. E-mail: [email protected]

Genes or multigenic chromosomal regions are organized by the nuclear matrix into a series of functionally discrete genic domains. Biophysical analysis of the human 16p13.13 region has shown that the PRM1→PRM2→TNP2 containing multigenic locus is bounded by two sperm nuclear matrix attachment regions (MAR). This domain exists in a transcriptionally readied or potentiated (i.e. open) state when associated with the nuclear matrix. The MAR-bounded PRM1→PRM2→TNP2 locus is nestled in an Alu repetitive element dense region. Fluorescence in-situ hybridization, analysis of sperm nuclear matrix/halo preparations showed that the PRM1→PRM2→TNP2 domain specifically localizes to the sperm nuclear matrix. This raised the question of whether nuclear matrix association and expression in this locus is mediated by Alu methylation. The methylation status of the various Alu elements contained within the human PRM1→PRM2→TNP2 locus was therefore assayed. The seven Alu elements tested, including those associated with the matrix attachment regions within the PRM1→PRM2→TNP2 locus, were fully methylated in sperm DNA. Conversely, these same Alu repeats were hypomethylated within the erythroleukaemic cell line, K562, which does not express any of the from this domain. This study shows that Alu methylation status is independent of attachment of PRM1→PRM2→TNP2 locus to the nuclear matrix and that Alu methylation does not play a leading role in the regulation of this domain.

Key words: Alu repeat/methylation/protamine/nuclear matrix/spermatozoa

Introduction matrix attachment regions (MAR) (Ward and Coffey, 1991; The human sperm genome is at least six times more condensed Bode et al., 2000). The segments that remain associated with than in the somatic cell (Balhorn, 1982). This highly condensed the nuclear matrix are usually biologically active, while the chromatin state is achieved by replacing ~85% of the inactive regions segregate as independent looped domains. with the smaller, more basic protamine contributing Thus, the nuclear matrix provides a means for organizing during the terminal phase of spermiogenesis (Prigent et al., chromatin into discrete domains. To date, a universal MAR 1996). The PRM1→PRM2→TNP2 protamine containing consensus sequence has not been identified, although a series multigenic locus encodes the suite of proteins that package of MAR-related motifs have been identified (Singh et al., sperm DNA into this highly condensed state. This locus 1997). The regions of attachment to the nuclear matrix are exists as a single, coordinately expressed chromatin domain often found in close association to various repetitive elements. (Choudhary et al., 1995; Stewart et al., 1999) bounded by two Alu elements are reiterated at a similar frequency as MAR, male-haploid specific nuclear matrix attachment sites (Kramer and this has led to the suggestion that regulation of attachment and Krawetz, 1996). These attachment sites are thought to be may be mediated by the interaction of repetitive elements with central to the potentiative process that opens this segment of the nuclear matrix (Boulikas, 1993). the genome to permit transcription (Kramer et al., 1997). The multigenic PRM1→PRM2→TNP2 locus is embedded The proteinaceous network of the nuclear matrix binds within a repetitive element-rich segment of the chromatin at sequence-specific regions of attachment, termed (Kramer et al., 1998a). The majority of the repetitive elements

© European Society of Human Reproduction and Embryology 903 C.Schmid et al. are members of the Alu family. These have even been localized segment with the sperm nuclear matrix and for gene-specific within the MAR-containing regions of this locus, raising the expression during . issue of their functional significance. Alu repeats constitute approximately one-third of the possible methylation sites within the human genome (Schmid, Materials and methods 1991). They are members of the Short Interspersed Element Fibre FISH (SINE) family and contain an internal promoter which directs DNA fibres were prepared using lymphocytes isolated from human their transcription by RNA polymerase III (Pol III) that is blood and treated as previously described (Heng et al., 1992). Cosmid repressed by methylation (Li et al., 2000; and references probes 440-e5 and 345-f2 (Kramer et al., 1998a) were biotinylated therein). In Alu repeats, these sites are almost fully methylated with dATP using the BRL BioNick labelling kit. hP3.1 (U15422 in DNA from somatic tissues. However, a subset of Alu coordinates 1–40 573 bp) (Nelson and Krawetz, 1994) was labelled elements is completely unmethylated in DNA from male germ by digoxigenin (DIG) and FISH analysis was performed essentially line tissues such as mature spermatozoa, total testis and as previously described (Heng and Tsui, 1994). seminoma (Hellmann-Blumberg et al., 1993; Kohanek et al, FISH of sperm nuclear halos 1993; Rubin et al., 1994). Similarly, Alu repeats are hypo- methylated in K562 cells to approximately the same degree Sperm nuclei were prepared fresh from spermatozoa obtained from normal healthy donors and FISH was carried out essentially as as in male germ cells. Pol III-directed Alu transcripts are described (Yaron et al., 1998). Sperm heads were separated from present at peculiarly high levels in these cells (Li et al., 2000). their tails by dounce homogenization, and their nuclei were isolated The exact composition of this subset has not been fully defined. by centrifugation through a sucrose gradient. Sperm nuclear halos Alu elements that are evolutionarily young tend to be more were then prepared and FISH was performed using the fluorescently hypomethylated than older, more divergent Alu elements. labelled cosmids 440-e5, hP 3.1 and 345-f2 as probes. The location Moreover, certain Alu members are completely methylated of the various segments with respect to the nuclear matrix was derived whereas others are completely unmethylated in sperm DNA from ~100 different observations. Over-extended fibres were excluded (Kohanek et al., 1993). from this analysis. The functional significance, if any, of gross hypomethylation is unknown. Attractive possibilities for roles in spermatogenesis Methylation of the Alu and B1 repeat families include regulating genes that are expressed only in the germ To examine the general methylation status of human Alu and mouse line and packaging sperm nucleoprotamine, or in the early B1 repeats, two different strategies were employed. Consensus Alu restriction fragments were isolated by digesting human DNA with a embryo, signalling imprinted gene expression. Considering the combination of HaeIII and HinfI, then size selecting the product by first of these possibilities, DNA methylation generally silences agarose gel electrophoresis (Schmid, 1991). Following their isolation gene expression through cis effects of chromatin structure by phenol extraction and ethanol precipitation, these size-selected (Bestor, 1998; Jones et al., 1998; Nan et al., 1998). Alu products were divided into identical 5 µg aliquots and digested methylation may equivalently silence expression of any with one of several restriction enzymes as recommended by the neighbouring genes. In the male germ line, Alu hypo- manufacturer, to 10-fold over digestion. Mouse DNA was initially methylation may be associated with the derepression of genes digested with BsmAI, then isolated by phenol extraction and ethanol that are expressed only during spermatogenesis. However, as precipitation, and then divided into two equal 10 µg aliquots for shown in K562 cells, chromatin structure, not Alu hypo- digestion with either HpaII or MspI. methylation, is now believed to be the primary regulator of Genomic digests from the above were resolved by 1.2% agarose gel ϩ Alu transcription (Li et al., 2000). electrophoresis, then blotted onto Hybond N (Amersham Pharmacia Biotech). HinfI, RsaI and BstUI digests of pUC DNA served as The human PRM1→PRM2→TNP2 locus is unusually rich markers. Human Alu repeats were identified using hybridization probe in Alu repeats and the expression of these genes is restricted #51 (Rubin et al., 1994) and the corresponding mouse B1 repeats to a specific stage in the male germ line. Various structural were identified using the [Γ-32P]-terminally labelled oligonucleotide studies have shown that the domain remains in a potentiated probe AGTGAGTTCCAGGACAGCCAG (Quentin, 1989). In brief, (i.e. open) chromatin state (Kramer et al., 1998b, 2000) from the blots were prehybridized and then hybridized at 52°C to the probe the pachytene spermatocyte stage to the fully differentiated in hybridization solution (5ϫSSPE: 5ϫDenhardt’s, with 0.5 µg/ml mature spermatozoon. In addition, the domain is attached to salmon sperm DNA). Following hybridization, the blots were washed the sperm nuclear matrix by MAR (Kramer and Krawetz, 1996). several times at room temperature in 5ϫSSPE containing 0.5% This begs the question: does repetitive element methylation sodium dodecyl sulphate (SDS). A final 52°C 5 min wash in fresh mediate binding to the nuclear matrix and thus allow the solution was then employed to remove non-specifically bound probe potentiated chromatin state? We have directly addressed this (Rubin et al., 1994). issue using fluorescence in-situ hybridization (FISH) to define Assaying methylation within the protamine cluster the region of human chromosome 16p13.13 that is bound to DNA from normal humans (100 µg) or fertile 80 day adult transgenic the nuclear matrix. In addition, we have examined the methyl- mice (50 µg) that faithfully recapitulate the expression of all members ation of several of the Alu elements throughout this region of the human PRM1→PRM2→TNP2 locus (Choudhary et al., 1995; including those that encompass the nuclear matrix attachment Stewart et al., 1999) were first digested with one of several enzymes sites in DNA from different tissues. The results presented in or a combination of two enzymes. DNA subjected to double digestion this study clearly show that Alu germ line hypomethylation is was first digested with one of the two restriction enzymes, then not required for mediating the interaction of this chromosomal purified by phenol extraction and ethanol precipitation prior to 904 PRM1→PRM2→TNP2 domain in spermatozoa digestion with the second restriction enzyme. A minimum 2-fold extended in a single direction away from the nuclear matrix. excess of restriction enzyme was always employed and digestion was From the position of the hP3.1 probe, it was clear that the continued for at least 4 h using the manufacturer’s buffers and PRM1→PRM2→TNP2 domain localized to the nuclear matrix incubation temperatures. Following digestion, the DNA samples were core region. Analysis of the PRM1→PRM2→TNP2 domain again isolated then divided into equal aliquots prior to assessing DNA using this probe produced a signal within the nuclear matrix methylation by restriction analysis. In most cases, either lambda DNA in ~70% of the samples examined. This reflects the tight or plasmid DNA was mixed with the predigested sample to provide an internal control for methylation assessment by restriction digestion. binding of the hP3.1 region to the nuclear matrix as mediated At least one additional DNA sample that did not contain exogenous by the two boundary MAR of the protamine locus (coordinates lambda or plasmid DNA was also examined to assure the correct 8818–9760 and 32 586–33 536 bp) (Kramer and Krawetz, assignment of all hybridizing bands. 1996). Both 440-e5 and 345-f2 were localized to the outside The digests were separated by electrophoresis on 1% agarose gels of the nuclear matrix, relative to hP3.1, at a frequency that using Pst1 digests of lambda DNA as markers. The digests were then was constantly Ͼ50%. blotted onto Hybond Nϩ. The hybridization probes were released as Of note, Ͼ50% of the region encompassing the restriction fragments by digestion of subclones from U15422 (Nelson PRM1→PRM2→TNP2 domain that is associated with the and Krawetz, 1994). Subclone 7h5 (coordinates 34 822–35 430 bp) nuclear matrix is comprised of repetitive sequence element and 3g5 (coordinates 32 802–33 372 bp) were released with HindIII– DNA. At least 64% of the repetitive DNA contained within EcoRI, clone 5g4 (coordinates 34 008–34 325 bp) was released with this region are the various members of the Alu family (Kramer either PvuII or SstI, and clone 4c11 (coordinates 9109–9548 bp) was released with PvuII. These fragments were labelled with [α-32P]dATP et al., 1998). This raised the possibility that attachment of the → → by nick translation. The blots were hybridized to the probe at 60°C PRM1 PRM2 TNP2 domain to the sperm nuclear matrix in hybridization solution (5ϫSSC, 5ϫDenhardt’s and 0.5 µg/ml of may be mediated by the repetitive elements and perhaps by salmon sperm DNA). When the hybridization probe contained an Alu their methylation status. sequence fragment, the hybridization cocktail was supplemented with 25 ng/ml of sheared human DNA. Following overnight hybridization, Global levels of SINE methylation in human and mouse the filter was washed several times at 60°C for prolonged periods in tissues 0.5ϫSSC with 0.5% SDS. When the level of background hybridization The methylation status of individual human Alu elements in was significant, the washing stringency was increased by raising the DNA from spermatozoa, human K562 cells and transgenic ϫ temperature to 65°C or by including a final wash in 0.25 SSC. mouse testis bearing the human PRM1→PRM2→TNP2 locus (Choudhary et al., 1995; Stewart et al., 1999) was surveyed. Results The results are summarized in Figure 3, and Table I. The K562 somatic cell line that predominately contains hypo- Association of the PRM1→PRM2→TNP2 domain with the methylated Alu elements provided an ideal control. sperm nuclear matrix Human Alu repeats isolated as a consensus HaeIII–HinfI DNA fibres from fixed human lymphocytes were stretched fragment (Schmid, 1991) were tested for methylation by onto a glass slide. FISH analysis of these fibres was performed digestion with methylation sensitive enzymes (Figure 3A). In using the fluorescently labelled cosmids 440-e5, hP3.1 and contrast to that in spleen DNA (lanes 6 and 7), Alu elements 345-f2 as probes that encompass the PRM1→PRM2→TNP2 in sperm and K562 cell DNA were extensively digested with domain. As shown in Figure 1, detection of the hybridization HpaII (lanes 1, 2, 11, 12). The younger Alu repeats were signals showed that the relative order of the cosmids from visualized by carefully selecting the hybridization stringency. the 5Ј to 3Ј direction was 440-e5, hP3.1 and 345-f2. Further, These essentially contained the intact HpaII sites as shown by cosmid 345-f2 was separated from hP3.1 by ജ10 kb. the nearly complete digestion of these same DNA by MspI, Together this series of probes spanned a contiguous segment an isoschizomer which is insensitive to methylation (lanes 3, of ~120 kb of human chromosome 16p13.13. One would expect 8, 13). Similar results were obtained by cleavage with the that this segment should contain several sperm chromatin loops, methylation-sensitive enzyme SmaI and its methylation insens- since the average sperm loop has been estimated to be ~27 kb itive isoschizomer XmaI (compare lanes 4, 5; 9, 10; 14, 15). in size (Barone et al., 1994). Accordingly, these cosmids were The longer SmaI–XmaI site appears divergent when compared well suited as FISH probes to determine their relative level of to the HpaII–MspI site since digestion is not as complete with association with the nuclear matrix. this pair of enzymes. The internal control of lambda DNA DNA halos were prepared by extracting sperm nuclei with was not informative for the completeness of the SmaI and a high ionic strength reducing buffer (Barone et al., 1994). XmaI digest in this experiment. While we have not excluded This displaced the histones and from the chromatin, the possibility of sequence polymorphisms, we suspect that leaving the DNA attached at discrete points to the nuclear the XmaI digests may be incomplete and we have therefore matrix. As shown by FISH analysis in Figure 2, cosmids 440- relied upon the HpaII and MspI data for determining methyl- e5, hP3.1 and 345-f2 hybridized to discrete segments of the ation at this site in these samples. Of note, the Alu repeats in brightly stained sperm nuclear matrix core and the diffusely spermatozoa and K562 cell DNA have comparable levels of stained halo, i.e. loop. As illustrated in the middle panel, when hypomethylation (Figure 3A). the loop, i.e. halo, was splayed onto the slide it took on the The shorter mouse B1 repeat presents only a single appearance of a tear drop. The nuclear matrix core was at consensus HpaII–MspI site conducive to methylation the base of this structure, while the bulbous loop, i.e. halo, assessment (Quentin, 1989). To examine the methylation of 905 C.Schmid et al.

Figure 1. Extended cosmid fibre fluorescent in-situ hybridization map of the PRM1→PRM2→TNP2 region of human chromosome 16p13.13. Human lymphocyte nuclei were isolated, then DNA was released and stretched onto glass slides. The resulting DNA fibres were then probed with the three contiguous cosmids, 440-e5 (yellow), hP3.1 (red) and 345-f2 (green). As shown in the example, the PRM1→PRM2→TNP2 locus-containing hP3.1 probe hybridized to the centre of the fibre relative to the other two probes. The 440-e5 hybridized to the region 5Ј and 345-f2 probe hybridized to the region 3Ј, relative to hP3.1.

Figure 2. Fluorescent in-situ hybridization analysis of human sperm nuclear halos using the three contiguous cosmids. Human sperm nuclei were isolated then placed onto glass slides. The nuclei were then extracted with 2 mol/l NaCl to release the loop DNA and to visualize the nuclear matrix DNA. The resulting halo preparation was hybridized to the fluorescently labelled 440-e5, hP3.1 and 345-f2 cosmid probes. As indicated by the arrows, the left panel shows that the hybridization signal of 440-e5 is outside of the signal produced by hP3.1. The middle panel shows that hP3.1 is primarily localized to the nuclear matrix. As indicated by the arrows, the right panel shows that the hybridization signal of 345-f2 is outside of the signal produced by hP3.1.

906 PRM1→PRM2→TNP2 domain in spermatozoa this site, transgenic mouse DNA was first digested at a Alu methylation of the PRM1→PRM2→TNP2 locus MAR B1-consensus BsmAI site and then digested with either HpaII boundary regions or MspI (Figure 3B). No appreciable digestion of the B1 The methylation status of two clusters of Alu repeats mapping repeats by HpaII was detected in either liver or testis DNA. 5Ј and 3Ј to the PRM1→PRM2→TNP2 gene cluster encom- This is indicative of virtually complete methylation of this site passing the MAR boundary regions was examined (Figure 4). in both tissues. The repetitive element clusters were released by digesting DNA with one or more restriction enzymes prior to assessing the methylation status of the parent fragment. This was accomplished by digesting the parent fragment with a series of restriction enzymes that included HpaII, which does not cleave a methylated site (Figure 4). The sizes of parent fragments generated by digestion with either NsiI–HpaI (or in some cases, StyI–HpaI), PstI, PvuII–SacI, or BstYI are shown in Table II. The positions of the hybridization probes used to detect these fragments by Southern analysis are shown in Figure 4. All of the fragments detected by Southern analysis for the normal human and transgenic tissues bearing the human PRM1→PRM2→TNP2 locus correspond to the known sequence of this gene cluster. The PvuII–SacI parent fragment mapping immediately 3Ј to the protamine gene cluster and encompassing the 3Ј MAR includes the testable BstUI, HpaII and HhaI restriction sites within and around the Alu repeats (Figure 4, Table II). These sites were not cleaved in DNA from human spermatozoa or placenta or from mouse transgenic testis and liver (Figure 5). Complete digestion of these DNA (Figure 5) with MspI, the methylation-insensitive isoschizomer of HpaII, demonstrated the conservation of the HpaII site in these samples. The lambda control demonstrated complete digestion (data not shown). Accordingly, the sites queried were fully methylated in these four different tissues. Alu repeats mapping to the adjacent downstream BstYI fragment (Table I) also appeared to be fully methylated in Figure 3. Methylation status of human and mouse tissues and cultured cells. (A) DNA from human cultured K562 cells (lanes human spermatozoa and testis DNA at the testable HhaI, HpaII 1–5), from spleen (lanes 6–10) and from spermatozoa (lanes and SmaI sites (Figure 6). Complete digestion of the BstYI 11–15) was restricted with HaeIII and HinfI to release the Alu parent fragment in sperm DNA with MspI and XmaI showed repeats. This fragment was further digested with the methylation- the conservation of the HpaII and SmaI sites (Figure 6). In sensitive enzymes, HpaII (H) and SmaI (S), in parallel with their marked contrast to the other tissues, this BstYI fragment in methylation-insensitive isoschizomers, MspI (M) and XmaI(X). U ϭ undigested HaeIII-HinfI fragment. The DNA was then transgenic testis was essentially digested to completion by observed by Southern analysis using Alu hybridization probe 51. HpaII and extensively digested by SmaI and HhaI (Figure 6). (B) DNA from mouse liver or testis was digested at the mouse- Thus, the corresponding sites are hypomethylated in transgenic specific Short Interspersed Element (SINE) B1 consensus with testis in contrast to their complete and identical patterns of BsmAI. The mouse SINE B1 fragments were then released by digestion with either HpaII (H) or MspI (M). The DNA was then methylation in DNA from human testis, spermatozoa and other observed by Southern analysis using a mouse B1-specific tissues examined (Figure 4). The endogenous and transgenic oligonucleotide probe. loci showed no differences in expression (Stewart et al., 1999).

Table I. Global and protamine locus Alu methylation status: methylation overview

Cell type Global Alu methylation Protamine gene cluster Alu methylation

K562 Hypomethylated Unmethylated or very low methylation at all tested sites Human spermatozoa Hypomethylated Methylated Mouse containing the human protamine transgene Not applicable* Methylated at sites upstream of PRM1. Partially methylated at sites proximally downstream of TNP2. Unmethylated at sites distally downstream of TNP2 Human spleen Hypermethylated Methylated

The Alu elements assayed (*) do not correspond to the endogenous murine Short Interspersed Element (SINE) B1. The mouse 7SL SINE B1 are hypermethylated in normal and transgenic mouse testis. 907 C.Schmid et al.

Figure 4. Methylation map of the human PRM1→PRM2→TNP2 locus in transgenic mouse testis and K562 cells. Map coordinates correspond to GenBank HSU15422. Positions of the Alu repeats surrounding the matrix attachment region (MAR) are shown as arrows indicating their orientation. Individual restriction sites tested for methylation were: BstUI (B), HhaI(H),HpaII/MspI(M),SmaI (S) and XmaI(X).HhaI, HpaII and SmaI are all methylation-sensitive enzymes. The identities of the cloned probes that were used in the various hybridization assays are indicated by the boxed regions. The methylation status of the restriction sites queried is indicated as methylated (m), unmethylated (u), partially methylated (P), and not tested (nt). All other tissues tested were methylated at this locus.

difference in Alu member methylation (Figure 4). Alu repeats Table II. Positions of parent fragments used in methylation analysis residing within the NsiI–HpaI and PstI parent fragments (Table Restriction fragments Sequence coordinates of parent II) were fully methylated in human sperm DNA and in mouse fragments (bp) transgenic liver DNA whereas K562 cell DNA was not methylated at these sites (data not shown). When tested in NsiI (or StyI)–HpaI 7892–9441 PstI–PstI 8436–9870 transgenic testis DNA (Figure 7), these same sites appeared PvuII–SacI 32 105–34 325 to be fully methylated within the NsiI–HpaI parent fragment. BstYI–BstYI 33 915–35 354 BstUI, HhaI, HpaII and SmaI did not digest this fragment The parent fragments were recleaved with methylation-sensitive enzymes to appreciably, as indicated by the arrows, whereas the control create the methylation map of the PRM1→PRM2→TNP2 locus (Figure 4). exogenous plasmid contained within the sample showed com- Sequence coordinates are from Nelson and Krawetz (1994). plete digestion (Figure 7). Further, both MspI and XmaI completely digested the NsiI–HpaI parent fragment, verifying Additional hybridization experiments using probe 5g4 indicated that these restriction sites were conserved within the transgenes. that the boundary between these two different Alu methylation Thus, all sites tested within this restriction fragment appear to patterns resides near the two HpaII sites mapping within this be fully methylated in transgenic testis (Figure 4). In contrast same hybridization probe (data not shown). In comparison, to the NsiI–HpaI fragment, digestion of the parent PstI fragment very slight digestion of the BstYI parent fragment by HpaII, with HhaI, HpaII and SmaI approached two-thirds completion HhaI and SmaI was detectable in transgenic mouse liver (Figure 7). Complete digestion of the exogenous plasmid DNA (Figure 6). However, the extent of hypomethylation is control DNA in these same samples indicated that the incom- qualitatively far greater in the transgenic testis than in trans- plete digestion of the parent fragment must be a consequence genic liver (Figure 6). This probably reflects the status in of partial methylation of the targeted sites. Of particular note the predominant cell type where spermatids and hepatocytes is the principal HpaII product band, which has exactly the comprise Ͼ70% of all cells in their respective tissue. In same length as the SmaI–XmaI fragment (Figure 4). Thus, comparison, the BstYI parent fragment from K562 cell DNA while the HpaII site contained within the SmaI site is essentially was almost completely digested by HhaI and SmaI (Figure 6). unmethylated, the HpaII site positioned upstream of the 4c11 Compared to the MspI site, HpaII digestion of this same hybridization probe must be fully methylated (Figures 4 and 7). fragment appeared to be incomplete (Figure 6). This suggests that while the HpaII–MspI site that is located within the SmaI site is completely unmethylated, other HpaII–MspI sites are Discussion partially methylated (Figure 6). Matrix attachment regions can function as boundaries for The region encompassing the 5Ј MAR revealed a marked genic domains (Bode et al., 2000) and can play key roles in the 908 PRM1→PRM2→TNP2 domain in spermatozoa

Figure 5. Methylation status of the PvuII–SacI fragment 3Ј to the PRM1→PRM2→TNP2 gene locus in human sperm and placental DNA and transgenic mouse liver and testis DNA. DNA samples were digested with PvuII and SacI, then subjected to Southern analysis using the 3Ј matrix attachment region (MAR) region probe 3g5. Placental DNA was unmethylated at the HpaII/MspI site nearest the SacI site bounded by region 5g4. All other sites tested were methylated. U ϭ undigested PvuII–SacI fragment; B ϭ BstUI; Hh ϭ HhaI; Hp ϭ HpaII; M ϭ MspI.

Figure 6. Methylation status of the BstYI fragment 3Ј to the PRM1→PRM2→TNP2 gene cluster in human sperm and testis DNA compared with transgenic mouse liver and testis DNA. DNA samples were cleaved with BstYI then subjected to Southern analysis using the 3Ј matrix attachment region (MAR) region probe 7h5. Human spermatozoa and testis DNA were fully methylated, whereas K562, transgenic liver and transgenic testis all displayed varying degrees of hypomethylation at one or more sites. U ϭ undigested BstYI fragment; B ϭ BstUI; Hh ϭ HhaI; H ϭ HpaII; M ϭ MspI; S ϭ SmaI; X ϭ XmaI; – ϭ lambda DNA only. potentiation, i.e. the opening of discrete regions of chromatin in mutations within the 3Ј MAR can cause loss of expression the human genome to permit transcription (Krawetz et al., and disregulation of various members of this multigenic domain 1999). The PRM1→PRM2→TNP2 multigenic locus that our (Kramer et al., 1997). As shown by FISH analysis, the laboratory has been using as a model system to unravel this PRM1→PRM2→TNP2 domain is firmly attached to the sperm mechanism (Kramer et al., 2000) is bounded at its ends by nuclear matrix along its entire 40 kb length. In contrast, two regions that specifically attach to the sperm nuclear matrix the regions abutting the PRM1→PRM2→TNP2 domain were (Kramer and Krawetz, 1996). It has been demonstrated that splayed towards the loop. Interestingly, the segments contained 909 C.Schmid et al.

one of the seven human PRM1→PRM2→TNP2 cluster Alu elements examined exhibited hypomethylation in sperm DNA. This observation argues against a requirement for Alu hypomethylation associated with the potentiation of this domain. Trasler et al. examined the methylation status of several sites within the mouse protamine gene cluster of mouse and the transition 1 (TP1) gene of mouse chromosome 1 by a similar strategy (Trasler et al., 1990). While methylation of the TP1 gene CpG repeats in the 5Ј promoter and coding regions decreased during mouse spermatocyte development, similar sites in the protamine genes became more methylated in testis, when all three genes are expressed. Thus, for some genes, the relationship between hypomethylation and gene expression is not reflected in all repeats or is mediated by other factors such as specific sequence binding proteins or chromatin structure. Results from this investigation of the methylation of seven human PRM1→ PRM2→TNP2 gene cluster Alu elements throughout and encompassing the MAR leads to a similar conclusion. There Figure 7. Methylation status of the NsiI–HpaI and PstI fragments is no simple or strict requirement of one for the other. Ј → → of transgenic testis DNA 5 to the PRM1 PRM2 TNP2 gene Examination of the transgenic model that faithfully cluster. DNA was digested with either NsiI–HpaIorPstI. Control plasmid DNA was then added to the parent fragments. The recapitulates the expression of all members of the human parent and control fragments were then digested with either BstUI PRM1→PRM2→TNP2 locus (Stewart et al., 1999) revealed (B), HhaI (Hh), HpaII (H), MspI(M),SmaI (S), XmaI(X). site specificity in the methylation of the transgene Alu repeats. Complete digestion of the control plasmid was confirmed prior to Demethylated and methylated sites are not intermingled, subjecting the remaining portion of the digest to Southern analysis Ј Ј rather Alu elements at the 5 MAR are methylated and those using the 5 matrix attachment region (MAR) probe 4c11 (Figure Ј 4). Bands resulting from transgenic testis DNA are marked with an at the 3 MAR are hypomethylated. Since the transgenes are arrow. U demarcates the undigested parent fragment. inserted as multiple tandem copies, the partial methylation of certain Alu elements implies that they are fully methylated in some copies of the transgene yet unmethylated in others. This in the abutting loop remained in a somewhat compact form, may reflect their position relative to the nuclear matrix as as they do not extend the entire length of the halo formed by only a few copies of the tandemly integrated locus are the loop DNA. This suggests that there are certain defined attached. Dyck et al. reported a similar transgene-specific exceptions to the average loop size in human spermatozoa pattern of methylation—hypomethylation in testis but not in being ~27 kb. other tissues—for human IGF-I genes expressed in mouse It has been proposed that attachment to the nuclear matrix (Dyck et al., 1999). Like the members of the PRM1→ may be mediated through a series of repetitive elements PRM2→TNP2 domain, IGF-I is naturally expressed in the (Boulikas, 1993). More than 54% of the region of human testis. The transgenic human IGF-I is partially demethylated → → chromosome 16 encompassing the PRM1 PRM2 TNP2 at two HpaII–MspI sites within the construct’s intron. These multigenic locus described in this communication is comprised experiments suggest that the IGF-I transgene may be of repetitive elements, over half of which are SINE (Kramer hypermethylated in non-testis tissue by activation of an antiviral et al., 1998a). Both MAR attachment sites contain a series of defence mechanism in response to the cytomegalovirus repetitive elements that may play a strategic role in regulating (CMV) promoter used in this construct. However, this rationale this locus or may simply reflect the repetitive nature of this does not explain why testis should be exempt from protection. region. The results of this study discussed below are in accord Further, it does not explain the methylation pattern we observe with the latter. for the transgenic PRM1→PRM2→TNP2 locus since viral Alu repeats, the dominant human SINE, are grossly hypo- sequences were not utilized in their creation. Rather a mechan- methylated in mature sperm DNA as well as in DNA from ism peculiar to testis that exempts exogenous sequences from other male germ line tissues (testis and seminoma). This global hypermethylation may be present. unusual hypomethylation pattern, as displayed by Alu repeats, may be tightly linked to genes that are specifically expressed during male germ line development. Alternatively, there is in- Acknowledgements vitro evidence for an Alu binding protein in spermatozoa that can specifically protect Alu repeats from methylation This work was supported in part by USPHS grant GM21346 (C.W.S.) → → and the Agricultural Experiment Station at the University of California (Chesnokov and Schmid, 1995). The PRM1 PRM2 TNP2 and in part by NIH grant (HD36512) to S.A.K. The authors would cluster provides an extreme example of genes that are only like to thank Dr J.A.Kramer and Mr S.Merchant for their initial expressed in the developing male germ line. However, not contributions to this study. 910 PRM1→PRM2→TNP2 domain in spermatozoa

References organization, evolution and gene identification. J. Expt. Zoology, 282, 245–253. Balhorn, R. (1982) A model for the structure of chromatin in mammalian Kramer, J.A., McCarrey, J.R., Djakiew, D. and Krawetz, S.A. (1998b) sperm. J. Cell Biol., 93, 298–305. Differentiation: the selective potentiation of chromatin domains. Barone, J.G., De Lara, J., Cummings, K.B. and Ward, W.S. (1994) DNA Development, 125, 4749–4755. organization in human spermatozoa. J. Andrology, 15, 139–145. Kramer, J.A., McCarrey, J.R., Djakiew, D. and Krawetz, S.A. (2000) Human Bestor, T.H. (1998) Methylation meets acetylation. Nature, 393,311–312. spermatogenesis as a model to examine gene potentiation. Mol. Reprod. Bode, J., Benham, C., Knopp, A. and Mielke, C. (2000) Transcriptional Dev., 56, 254–258. augmentation: modulation of gene expression by scaffold/matrix-attached Krawetz, S.A., Kramer, J.A. and McCarrey, J.R. (1999) Reprogramming the regions (S/MAR elements). Crit. Rev. Eukaryot. Gene. Expr., 10,73–90. male gamete genome: a window to successful gene therapy. Gene, 234,1–9. Boullikas, T. (1993) Nature of DNA sequences at the attachment regions of Li, T.-H., Kim, C., Rubin, C.M. and Schmid, C.W. (2000) K562 cells implicate the nuclear matrix. J. Cell. Biochem., 52,14–22. increased chromatin accessibility in Alu transcriptional activation. Nucl. Chesnokov, I. and Schmid, C.W. (1995) Specific Alu binding protein from Acids Res., 28, 3031–3039. human sperm chromatin prevents DNA methylation. J. Biol. Chem., 270, Nan, X., Ng, H.-H., Johnson, C.A. et al. (1998) Transcriptional repression by 18539–18542. the methyl CpG binding protein MeCP2 involves a deacetylase Choudhary, S.K., Wykes, S.M., Kramer, J.A. et al. (1995) A haploid expressed complex. Nature, 393, 386–389. gene cluster exists as a single chromatin domain in human sperm. J. Biol. Nelson, J.E. and Krawetz, S.A. (1994) Characterization of a human locus in Chem., 270, 8755–8762. transition. J. Biol. Chem., 269, 31067–31073. Dyck, M.K., Ouellet, M., Gagn, M. et al. (1999) Testes-specific transgene Prigent, Y., Muller, S. and Dadoune, J.P. (1996) Immunoelectron microscopical expression in insulin-like growth factor-I transgenic mice. Mol. Reprod. distribution of histone H2B and H3 and protamines during human Dev., 54,32–42. spermiogenesis. Mol. Hum. Reprod., 2, 929–935. Hellmann-Blumberg, U., Hintz, M.F., Gatewood, J. and Schmid, C.W. (1993) Quentin, Y. (1989) Successive waves of fixation of B1 variants in rodent Developmental differences in the methylation of human Alu repeats. Mol. lineage history. J. Mol. Evol. 28, 299–305. Cell. Biol., 13, 4523–4530. Rubin, C.M., Vandevoort, C.A., Teplitz, R.L. and Schmid, C.W. (1994) Alu Heng, H.H.Q. and Tsui, L.-C. (1994) Free chromatin mapping by FISH. In repeated DNAs are differentially methylated in primate germ cells. Nucl. Choo, K.H.A. (ed.), Methods in Molecular Biology, Vol. 33, In Situ Acids Res., 22, 5121–5127. Hybridization Protocols. Humana Press, Totowa, NJ, pp. 109–122. Schmid, C.W. (1991) Human Alu subfamilies and their methylation revealed by blot hybridization. Nucl. Acids Res., 19, 5613–5617. Heng, H.H.Q., Squire, J. and Tsui, L-C. (1992) High resolution mapping of mammalian genes by in situ hybridization to free chromatin. Proc. Natl. Singh, G.B., Kramer, J.A. and Krawetz, S.A. (1997) Mathematical model to Acad. Sci. USA, 93, 9509–9513. predict regions of chromatin attachment to the nuclear matrix. Nucl. Acid Res., 25, 1419–1425. Jones, P.L., Veenstra, G.J.C., Wade, P.A. et al. (1998) Methylated DNA and Stewart, K., Kramer, J.A., Evans, M.I. and Krawetz, S.A. (1999) Temporal MeCP2 recruit histone deacetylases to repress transcription Nature Genet., expression of the transgenic human protamine gene cluster. Fertil. Steril., 19, 187–191. 71, 739–745. Kohanek, S., Renz, D. and Doerfler, W. (1993) DNA Methylation in the Alu Trasler, J.M., Hake, L.E., Johnson, P.A. et al. (1990) DNA methylation and sequences of diploid and haploid primary human cells. EMBO J., 12, demethylation events during meiotic prophase in the mouse testis. Mol. 1141–1151. Cell. Biol., 10, 1828–1834. Kramer, J.A. and Krawetz, S.A. (1996) Nuclear matrix interactions within the Ward, W.S. and Coffey, D.S. (1991) DNA packaging and organization in sperm genome. J. Biol. Chem., 271, 11619–11622. mammalian spermatozoa: comparison with somatic cells. Biol. Reprod., Kramer, J.A., Zhang, S., Yaron, Y. et al. (1997) Genetic testing for male 44, 569–574. infertility: a postulated role for mutations in sperm nuclear matrix attachment Yaron, Y., Kramer, J.A., Gyi, K. et al. (1998) Centromere sequences localize regions. Genet. Testing, 1, 125–129. to the human sperm nuclear halo. Int. J. Androl., 21,13–18. Kramer, J.A., Adams, M.D., Singh, G.B. et al. (1998a) Extended analysis of the region encompassing the PRMI→PAM2→TNP2 domain: genomic Received on April 4, 2001; accepted on August 8, 2001

911