Pericentric Heterochromatin Generated by HP1 Protein Interaction-defective Histone Methyltransferase Suv39h1*

Background: Histone H3 lysine 9 trimethylation (H3K9me3) and heterochromatin protein HP1 accumulations are hallmarks of heterochromatin. Results: Pericentric accumulation of histone methyltransferase, Suv39h, and Suv39h-mediated H3K9me3 occurs without Suv39h-HP1 binding and HP1 accumulation. Conclusion: The functional relationship between Suv39h and HP1 for pericentric heterochromatin formation is clarified. Significance: The Suv39h-mediated heterochromatin formation can be further elucidated from this model. Pericentric regions form epigenetically organized silent heterochromatin structures that accumulate histone H3 lysine 9 trimethylation (H3K9me3) and HP1. At pericentric regions, Suv39h is the major enzyme that generates H3K9me3. Suv39h also interacts directly with HP1, a methylated H3K9-binding protein. However, it is not well characterized how HP1 interaction is important for Suv39h accumulation and Suv39h-mediated H3K9me3 formation at the pericentromere. To address this, we introduced the HP1 binding-defective N-terminally truncated mouse Suv39h1 (ΔN) into Suv39h-deficient embryonic stem cells. Interestingly, pericentric accumulation of ΔN and ΔN-mediated H3K9me3 was observed to recover, but HP1 accumulation was only marginally restored. ΔN also rescued DNA methyltransferase Dnmt3a and -3b accumulation and DNA methylation of the pericentromere. In contrast, other pericentric heterochromatin features, such as ATRX protein association and H4K20me3, were not recovered. Finally, derepressed major satellite repeats were partially silenced by ΔN expression. These findings clearly showed that the Suv39h-HP1 binding is dispensable for pericentric H3K9me3 and DNA methylation, but this interaction and HP1 recruitment/accumulation seem to be crucial for complete formation of heterochromatin.

Pericentric regions form epigenetically organized silent heterochromatin structures that accumulate histone H3 lysine 9 trimethylation (H3K9me3) and HP1. At pericentric regions, Suv39h is the major enzyme that generates H3K9me3. Suv39h also interacts directly with HP1, a methylated H3K9-binding protein. However, it is not well characterized how HP1 interaction is important for Suv39h accumulation and Suv39h-mediated H3K9me3 formation at the pericentromere. To address this, we introduced the HP1 binding-defective N-terminally truncated mouse Suv39h1 (⌬N) into Suv39h-deficient embryonic stem cells. Interestingly, pericentric accumulation of ⌬N and ⌬N-mediated H3K9me3 was observed to recover, but HP1 accumulation was only marginally restored. ⌬N also rescued DNA methyltransferase Dnmt3a and -3b accumulation and DNA methylation of the pericentromere. In contrast, other pericentric heterochromatin features, such as ATRX protein association and H4K20me3, were not recovered. Finally, derepressed major satellite repeats were partially silenced by ⌬N expression. These findings clearly showed that the Suv39h-HP1 binding is dispensable for pericentric H3K9me3 and DNA methylation, but this interaction and HP1 recruitment/accumulation seem to be crucial for complete formation of heterochromatin.
Chromatin exists in two forms, euchromatin and heterochromatin (1). Euchromatin is the loosely packed form of chromatin that is rich in gene concentration and often undergoes active transcription. In contrast, heterochromatin is tightly packed and is in the transcriptionally repressed state. The pericentromere is a heterochromatic domain that provides a struc-tural scaffold for centromere formation and plays a crucial role in genome stability (2). In mice, pericentric heterochromatin consists of AT-rich sequences of extremely long tandem arrays of major satellite repeats (3). Therefore, the fluorochrome DAPI, which preferentially intercalates with A/T-rich repeat sequences, can show mouse pericentromere heterochromatin as a DAPI-dense domain.
In addition to the repetitive sequences, the pericentric heterochromatin has a distinct combination of epigenetic marks such as histone H3 lysine 9 trimethylation (H3K9me3), H4K20me3, and DNA methylation (1,4). Suv39h/KMT1A is the principal enzyme for the H3K9me3 of the pericentromere heterochromatin in mammals, which is evolutionarily conserved. Because Suv39h also indirectly regulates H4K20me3 and DNA methylation, these epigenetic marks are lost from or are in low concentration in the pericentromere in Suv39h-deficient cells (5,6). Thus, Suv39h is one of the master regulators of epigenetically organized heterochromatin.
One of important roles of these epigenetic modifications is recruitment of different effector molecules to the specific chromatin loci (7,8). Therefore, at heterochromatin, various transcriptionally silent effector molecules are recruited by the heterochromatin-specific epigenetic marks. Heterochromatin protein 1 (HP1) is such an effector molecule that was originally discovered in Drosophila as a dominant suppressor of positioneffect variegation (9). Similar to Suv39h, the HP1 family is evolutionarily conserved, with members in fungi, plants, and animals, and it has multiple isoforms within the same species (10). The N-terminal chromodomain (CD) 2 shows a high affinity to methylated H3K9 (highest affinity for H3K9me3), causing HP1 to be tethered to heterochromatin (11,12). This recruitment system is also highly conserved in different species. Furthermore, this regulation is interdependent. For example, the HP1 * This work was supported in part by a grant-in-aid from the Ministry of Edu-This article has been withdrawn by the authors. In this work, we reported that pericentric accumulation of histone methyltransferase, Suv39h1 and Suv39h1-mediated H3K9me3 can take place in the absence of the interaction of Suv39h1 with HP1 and HP1 accumulation. In extending this work, we found that we could not reproduce our finding that there was pericentric accumulation of the HP1 interaction-defective Suv39h1 mutant (Suv39h1 1-41 deletion mutant) in Suv39h-deficient ES cells (depicted in Fig. 2, C and G); we again observed induction of pericentric H3K9me3. Although we could not repeat precisely the same experiments as the supply of the original cell lines had been exhausted, our recent results no longer support our original published conclusions. Therefore, we wish to withdraw this paper. We apologize for any inconvenience that may have resulted from its publication.
homolog Swi6 in fission yeast is also crucial for the Suv39h homolog Clr4 accumulation and Clr4-mediated H3K9 methylation to heterochromatin (13). HP1 homologs can physically interact with Suv39h homologs in different species and sequential cycles of Swi6 binding, and Clr4 recruitment/deposition of H3K9me have been proposed (14) to explain the interdependent regulation of Clr4-and Swi6-mediated silent heterochromatin formation.
A similar functional concept has been proposed for the Suv39h-and HP1-mediated heterochromatin formation in mammals (11,12). However, so far, it has not been validated experimentally much whether HP1 binding to Suv39h is crucial for Suv39h-mediated heterochromatin formation, which is what we are addressing in this study.
Protein Immunoprecipitation-48 h after transfection, HEK293 T cells were incubated in PBS containing 5 mM dimethyl dithiobispropionimidate at 4°C for 1 h. Whole cell lysate was obtained using lysis buffer (150 mM NaCl, 50 mM Tris-HCl, pH 7.5, 0.3% digitonin, 20 mM N-ethylmaleimide) after quenching for 10 min at 4°C using 150 mM glycine PBS. We then followed two protocols. 1) The FLAG-IP protocol wherein the lysate was incubated with an anti-FLAG antibody affinity gel (Sigma, A2220 -10ML) for 2 h at 4°C. The immune complex was washed with washing buffer (150 mM NaCl, 50 mM Tris-HCl, pH 7.5, 0.1% digitonin) three times, and then precipitated proteins were eluted by an excess amount of 3ϫ FLAG peptide (Sigma, F4799). 2) In the myc-IP protocol, the lysate was incubated with anti-Myc (9E10) for 2 h at 4°C. The immune complex was captured using protein G-Sepharose (GE Healthcare, 17-0618-02) and washed with washing buffer. For Western blot analysis, anti-FLAG (M2) and anti-Myc (9E10) were used as primary antibodies, and HRP-conjugated anti-mouse Ig (RKL, 18 -8817-31) was used as a secondary antibody.
Immunofluorescence Analysis-Cells cultured on chamber slides (nunc, 177437JP) were fixed with 4% paraformaldehyde for 8 min at room temperature, permeabilized with 1% Triton X-100 for 15 min, and incubated overnight with primary antibodies (4°C). Anti-mouse IgG conjugated with Alexa-568,488 Fluor (invitrogen) was used as a secondary antibody. The nuclei were counterstained with DAPI, as observed under a confocal microscope (Olympus, FV1000). Three-dimensional reconstructed image analysis was performed using the Z-stack function of analysis software (FV10 ASW). Signal intensity of the antigen per DAPI intensity was measured by ImageJ.
Southern Blot Analysis with Methylation-sensitive Restriction Enzyme-Genomic DNA was isolated and digested with methylation-sensitive restriction enzyme MaeII and analyzed on DNA blot. Major satellite repeat DNA probe was amplified from genomic DNA of Suv39 dn cell line. PCR primers used for this cloning were as follows: MajF.BamHI (5Ј-tacggatccGAC-GACTTGAAAAATGACGAAATC-3Ј) and MajR.EcoRI (5Ј-tacgaattcCATATTCCAGGTCCTTCAGTGTGC-3Ј). PCR fragment (326 bp) was inserted to pBluescript SKII(ϩ) by BamHI and EcoRI site. Major satellite probe sequence was GACGA- Northern Blot Analysis-Probe for Northern blot analysis was the same as Southern blot analysis with methylation-sensitive restriction enzyme.

N-terminal Region of Mouse Suv39h1 Was Essential for HP1
Interaction-It has been shown that the N terminus of Suv39h1 and Drosophila homolog Su(var)3-9, which is upstream of the CD, interacts with the chromoshadow domain (CSD) of HP1 in vitro (18 -20). Therefore, we assayed the interaction of wildtype (WT) mouse Suv39h1 and the N-terminal deletion (⌬1-41) mutant (named ⌬N) with mouse HP1␣, -␤, and -␥ in HEK293T cells (Fig. 1). As shown in Fig. 1B, Myc-tagged HP1␣, -␤, and -␥ were clearly co-immunoprecipitated with FLAGtagged Suv39h1 but not with ⌬N (center IP:FLAG, two panels). Anti-Myc co-immunoprecipitation experiments showed the same results (Fig. 1B, bottom IP:myc, two panels). These data demonstrate that the ⌬N mutant of Suv39h does not interact with HP1.
Pericentric Focus Formation of ⌬N and ⌬N-mediated H3K9me3 Could Be Induced in the Suv39h1/2 Double Null (dn) ES Cells, but HP1 Accumulation Continued to Remain Impaired-To address how the Suv39h-HP1 interaction is crucial for Suv39h-mediated epigenetic heterochromatin formation, we introduced WT Suv39h1 and ⌬N into Suv39h dn ES cells (6,17). It is also known that enzymatically inactive Suv39h cannot establish/maintain H3K9me3 and its own focus formation on pericentric regions in a Suv39h1/2-deficient background (12). Therefore, we introduced the same enzymatically inactive mutant of Suv39h1 (named H324L, Fig. 1A). Immunofluorescence staining analysis clearly showed that there is no accumulation of H3K9me3, HP1␣, -␤, and -␥ on pericentric DAPI-dense regions in the Suv39h dn ES cells (Fig. 2, A, B, and D) as reported previously (12). Introduction of FLAG-tagged WT Suv39h1, but not H324L, rescued H3K9me3 and HP1␣, -␤, and -␥ focus formation on DAPI-dense regions in the Suv39h dn ES cells. Interestingly, these H3K9me3 signals were also recovered by the expression of FLAG-tagged ⌬N; however, pericentric HP1␣, -␤, and -␥ signals were still rarely detectable (Fig. 2, A, B, and D). Furthermore, FLAG-tagged ⌬N in the Suv39h dn ES cells was accumulated on the DAPI-dense regions, as shown for WT Suv39h1 (Fig. 2C) (12). Western blot analysis showed that the expression of HP1␣, -␤, and -␥ was not changed in these transfected cells (Fig. 2E), suggesting that the level of HP1 protein was not affected. Accumulation of H3K9me3 and FLAG-tagged Suv39h1 (WT and ⌬N) on the pericentric regions was also validated by ChIP-quantitative PCR analysis. Fig. 2, F and G, clearly shows that H3K9me3 on the pericentric major satellite repeat regions was rescued by not only WT Suv39h1 but also ⌬N; both molecules were enriched on these loci. Furthermore, it was also confirmed by HP1␤ ChIP-quantitative PCR analysis that HP1␤ accumulation is only marginally restored by ⌬N expression (Fig. 2H). Because all these phenotypes were observed in multiple stable expressing clones for each construct, we have shown the results for only one of the clones (i.e. representative clone data). Moreover, immunofluorescence staining data (low magnification images) for only two clones are shown (Figs. 2, I and J, 3, F and G, and 5, F and G).
In conclusion, these results show that Suv39h1⌬N, which does not interact with HP1, can be recruited to and accumulate in the pericentric regions and deposit H3K9me3. HP1 possesses binding activity with methylated H3K9; however, this activity was not sufficient for substantive HP1 pericentric accumulation without Suv39h binding.
H4K20me3 and ATRX Did Not Accumulate at Pericentric Heterochromatin in the Suv39h1/2 dn ES Cells Expressing ⌬N-In addition to H3K9me3 and HP1, other epigenetic heterochro- matin marks or molecules such as H4K20 methyltransferase Suv4-20h, H4K20me3, and ATRX are depleted from pericentric regions in Suv39h dn cells (5,17). Therefore, we analyzed pericentric deposition or accumulation of H4K20me3 and ATRX in Suv39h dn ES cells rescued with ⌬N. As shown in Fig.  3, A, D, and F, pericentric H4K20me3 signals were absent in Suv39h dn ES cells, and they were recovered by the expression of FLAG-tagged WT Suv39h1. However, ⌬N did not rescue pericentric H4K20me3 signals. ATRX is a SWI/SNF-like chromatin remodeling protein, and it localizes to pericentric heterochromatin regions (21). Part of ATRX also forms a complex with the death domain-associated protein DAXX and localizes in promyelocytic leukemia nuclear bodies (22). The ATRX ADD domain shows binding activity to the H3 peptide containing H3K9me3/K4me0 (23,24), and the heterochromatin accumulation is facilitated by association with HP1 (17) and MeCP2 (25). As shown for H4K20me3, ATRX pericentric accumulation was absent in Suv39h dn ES cells and was rescued by WT Suv39h1 expression (Fig. 3, B, D, and G). ATRX pericentric accumulation was not rescued in the Suv39h dn ES cells expressing ⌬N, but ATRX protein content was not changed (Fig. 3E). In contrast, ATRX localization at the promyelocytic leukemia nuclear bodies marked by DAXX was unaffected in Suv39h dn ES cells and additional rescued cells (Fig. 3C, yellow arrowheads).
Chromatin Compaction Did Not Significantly Change in ES Cells without Suv39h Expression-A recent study by Wang et al. (26) illustrated that the genome of the human cell line was more sensitive to micrococcal nuclease after SUV39H1 knockdown (KD), suggesting that nuclear chromatin is less compacted in SUV39H KD or depleted cells. Therefore, we also performed an micrococcal nuclease sensitivity assay. However, in contrast to the KD study, micrococcal nuclease sensitivity of the Suv39h dn ES genome (global and major satellite repeat regions) did not undergo significant changes (data not shown). We also compared the average number of DAPI-dense chromocenters per cell among WT and Suv39h dn ES cells with and without complementation by Suv39h1 or ⌬N expression. As shown in Fig. 4A, the average number of chromocenters per cell did not differ significantly. Finally, we performed three-dimensional reconstructed image analysis for DAPI-dense chromocenters for each cell type (Fig. 4B). Again, clear morphological differences could not be observed. Consistent with these results, the size distribution of chromocenters for each cell type also was not statistically different (Fig. 4C). It is generally recognized that the nucleus in ES cells is an open chromatin configuration (27). If we could further analyze differentiated WT and Suv39h dn cells, we might see less compacted chromatin phenotypes in Suv39h dn cells, especially at pericentric hetero-chromatin regions and thus validate the role of ⌬N for this issue.
DNA Methylation of Major Satellite Repeats Was Rescued by ⌬N Expression-DNA methylation has been observed to decrease in the Suv39h-deficient cells (6). Therefore, we examined the DNA methylation status of major satellite repeats in the Suv39h dn ES cells complemented with ⌬N. Purified DNA was isolated from WT, Dnmt1 Ϫ/Ϫ , -3a Ϫ/Ϫ , -3b Ϫ/Ϫ triple knock-out (Dnmt triple KO) (28) and Suv39h dn ES cells; the Suv39h dn ES cells expressing FLAG-tagged Suv39h1, H324L, or ⌬N were digested with DNA methylation-sensitive restriction enzyme MaeII and subjected to Southern blot analysis with DNA fragments encoding major satellite repeat sequences as a probe. As shown in Fig. 5A, the major satellite repeats were heavily methylated in WT ES cells and completely demethylated in the Dnmt triple KO cells. In the Suv39h dn ES cells, the level of major satellite repeat DNA methylation decreased as reported previously (6). Interestingly, this DNA hypomethylation phenotype was completely rescued by FLAG-tagged WT Suv39h1 and ⌬N but not by H324L.
Next, we examined nuclear localization of DNA methyltransferase Dnmt3a and -3b in those ES cell lines (Fig. 5, B-D). As reported (6,29), Dnmt3a and -3b were localized at pericentric regions in WT ES cells, and this accumulation was lost in Suv39h dn ES cells. Consistent with the DNA methylation phenotype, both Dnmt3a and -3b pericentromere accumulations were rescued in the Sub39h dn ES cells expressing Suv39h1 and ⌬N but not H324L. The level of Dnmt3a and -3b content in these cell lines did not change (Fig. 5E). These data suggest that Suv39h localizes to the pericentromere and recruits Dnmt3a and -3b to induce DNA methylation of major satellite repeats.
Transcription of Major Satellite Repeats Was Partially Repressed by ⌬N-⌬N could localize to the pericentromere and induce H3K9me3 and DNA methylation. However, other pericentric heterochromatin features such as accumulation of HP1 and ATRX and the H4K20me3 formation were not rescued by ⌬N. To evaluate how these two heterochromatins established by WT Suv39h1 and ⌬N are qualitatively different, we examined the transcriptional status of the major satellite repeats (Fig.  6). Northern blot analysis showed that the major satellite repeats were derepressed in the Suv39h dn ES cells, as reported previously (6). Induction of WT Suv39h1 completely repressed this reactivation. However, this major satellite repeat transcript was only partially repressed by ⌬N. In Dnmt triple KO ES cells, these transcripts were also partially derepressed. Our results suggest that ⌬N can create a partially silent heterochromatin structure that may be DNA methylation-mediated, and the missing components shown here seem to be important for completely silent heterochromatin. FIGURE 2. Suv39h1⌬N could accumulate and deposit H3K9me3 at the pericentromere, but HP1 was generally absent from the pericentric region. Immunohistochemical staining analysis for H3K9me3 (red) and HP1␤ (green) (A), HP1␣ (red) and HP1␥ (green) (B), or FLAG-tagged Suv39h1 WT, H324L, or ⌬N (red) (C) was conducted with WT or Suv39h dn ES cells or Suv39h dn ES cells complemented with either FLAG-tagged WT Suv39h1, H324L, or ⌬N. DNA was counterstained with DAPI (blue). D, DAPI intensity and the intensity of the antigen at the same region were quantified using ImageJ based on the fluorescence imaging data obtained for A-C. The intensities for 30 cells were quantified. The value of each antigen intensity was expressed as a ratio between the DAPI intensity and the intensity of the antigen being measured. E, Western blot analysis of endogenous HP1 subtypes in the indicated cell lines. F-H, chromatin was isolated from the indicated cell lines. ChIP-quantitative PCR analysis was conducted using anti-H3K9me3 (F), anti-FLAG (G), anti-HP1␤ (H), and primers specific for major satellite repeats and the Gapdh gene. I and J, immunohistochemical staining analysis for the two independent cell lines indicated at low magnification. A combination of color and antigen was same as that observed in A and B. AUGUST 30, 2013 • VOLUME 288 • NUMBER 35

DISCUSSION
In the Suv39h dn ES cells, H3K9me3 and other repressive epigenetic molecules or marks such as HP1, ATRX, Suv4-20h, Dnmt3a and -3b, H4K20me3, and DNA methylation were depleted or reduced from pericentric DAPI-dense regions. However, major satellite repeats were derepressed. In this study, we demonstrated that the Suv39h1 mutant that did not interact with HPI (⌬N) could rescue the pericentric accumulation of itself and H3K9me3 deposition. Furthermore, Dnmt3a, -3b, and DNA methylation were recovered on the pericentric regions. However, pericentric accumulation of HP1 was severely affected, and ATRX and H4k20me3 were not recovered at all. These phenotypes are shown in Fig. 7.

HP1 Binding to Suv39h1 Was Dispensable for the Establishment of Suv39h1 Pericentromere Accumulation and H3K9me3
Formation-It is known that HP1 homolog Swi6 is crucial for Clr4-mediated H3K9 methylation of heterochromatin in fission yeast (13). Furthermore, it has been proposed that the physical interaction of Clr4 with Swi6 is important for the establishment/spreading of H3K9 methylation (30). A similar functional concept for the Suv39h-HP1 complex on the H3K9me-mediated heterochromatin establishment/spreading has been proposed for other species, including mammals (11,12). However, the analysis of Suv39h dn ES cells rescued with ⌬N demonstrated that HP1 binding to Suv39h was dispensable for the establishment of Suv39h1 pericentromere accumulation FIGURE 3. ATRX and H4K20me3 pericentromere accumulations were not restored in the Suv39h dn ES cells expressing Suv39h1⌬N. Immunohistochemical staining analysis for H4K20me3 (red) and HP1␤ (green) (A), H3K9me3 (red) and ATRX (green) (B), and DAXX (red) and ATRX (green) (C) were conducted with WT or Suv39h dn ES cells or Suv39h dn ES cells complemented with either FLAG-tagged WT Suv39h1, H324L, or ⌬N. DNA was counterstained with DAPI (blue). C, white and yellow arrowheads shows the DAPI-dense pericentric chromocenter and DAXX-positive promyelocytic leukemia nuclear body, respectively. D, DAPI intensity and the intensity of the antigen at the same region were quantified using ImageJ based on the fluorescence imaging data obtained for A and B. The intensities for 30 cells were quantified. The value of each antigen was expressed as a ratio between the DAPI intensity and the intensity of the antigen being measured. E, Western blot analysis for ATRX expression in the indicated cell lines. F and G, immunohistochemical staining analysis for the two independent cell lines indicated at low magnification. A combination of color and antigen was same as that observed for A and B. and its H3K9me3 formation. The level of H3K9me3 over the major satellite repeat regions in the Suv39h dn ES cells was completely recovered by ⌬N, similar to that in WT ES cells (Fig.  2F). These findings are consistent with a previous report indicating that the recruitment of SUV39H1 to heterochromatin is at least partly independent from HP1 interaction (31). However, this does not exclude the possibility that other heterochromatic regions are not fully reconstituted by ⌬N, especially at the heterochromatin-euchromatin boundaries or facultative heterochromatins in euchromatic regions. Future genomewide epigenetic analysis will clarify this issue. How is it possible then that HP1 binding to Suv39h and/or HP1 accumulation are dispensable for Suv39h-mediated H3K9me3 formation on the pericentromere? In the original transfection experiments done by Lachner et al. (12), exogenous Suv39h1 pericentromere accumulation was induced in both WT and Suv39h dn cells, but the enzymatically inactive mutant H324L only accumulated in WT and not in the Suv39h dn cells. These data suggest the possibility that H3K9me3 deposited by Suv39h itself is critical for Suv39h pericentromere accumulation. It was already known that HP1 CD showed higher affinity for H3K9 methylation (highest to H3K9me3) (11,12), and HP1 and Suv39h form a complex (32). The hypothesis of the self-enforcement cycle of HP1-Suv39h and H3K9 methylation has been applied to the mechanism of pericentric Suv39h and HP1 accumulation and H3K9me3 formation. However, recent Clr4 studies in fission yeast demonstrated that not only Swi6 CD but also Clr4 CD showed higher affinity for H3K9 methylation, and this interaction seemed to be crucial for Clr4 accumulation and H3K9me2/3 formation on heterochromatin (33). Furthermore and most recently, it has been reported that Suv39h CD also showed higher affinity to methylated H3K9 (highest to H3K9me3) in vitro (34). Therefore, our new findings and recent information strongly suggest that H3K9me3 deposited by Suv39h can directly tether its own enzyme to pericentric regions. Future studies will examine the Suv39h recruitment/ accumulation mechanism from this perspective.
Critical Role of Suv39h on HP1 Pericentromere Accumulation-It has been well established that Suv39h/Su(var)3-9/Clr4 is essential for HP1/Swi6 heterochromatin accumulation (12,18,35). However, this HP1 heterochromatin accumulation is induced/maintained by (at least) two distinct mechanisms, one of which is HP1 CD-mediated and the other is CSD-mediated. In case of mouse HP1␤, both the binding activity of CD to methylated H3K9 and the binding site of CSD to the PXVXLcontaining proteins are crucial for HP1␤ pericentromere accumulation (36). In other artificial experiments, the Gal4-SUV39H1 fusion tethering system showed that H3K9 methylation is not sufficient for recruitment of HP1 to chromatin, but the direct interaction of HP1 with SUV39H1 is also important (37). Furthermore, both H3K9me binding and CSD binding to PXVXL-containing proteins are important for nucleosome binding of Drosophila HP1 in vitro (20). In this case, HP1 can bind to Su(var)3-9 through the PXVXL-containing proteinbinding site, and HP1 nucleosome binding is enhanced by Su(var)3-9 loading. Because this enhancement is blocked by the FIGURE 5. DNA methylation of major satellite repeats was completely restored in the Suv39h dn ES cells expressing Suv39h1⌬N. A, genomic DNA isolated from the indicated cell lines was digested with the methylation-sensitive restriction enzyme MaeII. Southern blot analysis was conducted using a major satellite repeat probe. Immunohistochemical staining analysis for Dnmt3a (B) or Dnmt3b (C) (red) and HP1␤ (green) were conducted with WT or Suv39h dn ES cells or Suv39h dn ES cells complemented with either FLAG-tagged WT Suv39h1, H324L, or ⌬N. DNA was counterstained with DAPI (blue). D, DAPI intensity and the intensity of the antigen at the same region were quantified using ImageJ based on the fluorescence imaging data obtained for B and C. The intensities of 30 cells were quantified. The value of each antigen was expressed as a ratio between the DAPI intensity and the intensity of the antigen being measured. E, Western blot analysis of endogenous Dnmt3a and -3b in the indicated cell lines. F and G, immunohistochemical staining analysis for the two independent cell lines indicated at low magnification. A combination of color and antigen was same as that observed for B and C. disruption of the HP1-Su(var)3-9 interaction, it is proposed that the CSD-mediated HP1 heterochromatin accumulation is mediated by Su(var)3-9. Our new findings further strengthen the idea that both HP1 CD binding to methylated H3K9 and HP1 (CSD) binding to Suv39h are critical for HP1 pericentromere accumulation in mammals. However, there are some discrepancies between the ⌬N phenotypes shown here and those observed in the HP1␤ CD or CSD mutant studies (36). In the Suv39h dn cells expressing ⌬N, HP1 pericentromere accumulation was generally not restored; if the level of WT and Suv39dn were defined as 100 and 0%, respectively, only ϳ10 -20% of the WT level for HP1␣, -␤, and -␥ signals on the DAPIdense regions were detected in Suv39h dn cells expressing ⌬N (Fig. 2D). However, such a severely depleted phenotype is only induced by CD and CSD dual mutations and not by each single mutation in the HP1␤ studies (36). Thus, if H3K9me3 is restored in the Suv39h dn cells expressing ⌬N, one might expect that the HP1 pericentromere accumulation would be partially affected. Our current hypothesis for this ⌬N phenotype is as follows. Because in the Suv39h dn cells expressing ⌬N, the HP1 CSD-Suv39h module is not functional, and other HP1interacting heterochromatin molecules such as ATRX and Suv4-20h are also absent from the pericentric regions, these HP1binding molecules may also contribute to HP1 pericentromere accumulation. Therefore, in the ⌬N case, even though the HP1-H3K9me binding module is functional, several other HP1-binding heterochromatin modules are missing, which may be the reason for the severe HP1 loss of phenotype. ATRX Pericentric Heterochromatin Accumulation-In addition to HP1, ATRX pericentric accumulation is also maintained by multiple mechanisms (23)(24)(25). First, the ATRX ADD domain shows high affinity to H3K9me3 (plus Lysine 4 of same H3 is unmethylated). Second, HP1 binds to ATRX through the ATRX PXVXL motif. Finally, MeCP2 binds to ATRX through the ATRX helicase domain. Therefore, inactivation of each single module seems to have a partial or small impact on ATRX localization on pericentromere heterochromatin (23,24). In the Suv39h dn cells, all these modules are inactivated; thus 1) the H3 binding module is not functional because pericentric H3K9me3 is highly suppressed; 2) the HP1-mediated recruitment is absent because HP1 is not localized at the pericentromere, and 3) MeCP2-mediated recruitment is absent because MeCP2 is undetectable in ES cells. Although H3K9me3 formation is recovered by ⌬N, the two modules are not functional. Therefore, it is not surprising that the ATRX phenotype is not rescued at all in the Suv39h dn cells expressing ⌬N.
DNA Methylation of Major Satellite Repeats-⌬N could not rescue HP1 or the HP1-interacting heterochromatin molecules to pericentromere localization, but DNA methylation of major satellite repeats was recovered to the level of WT ES cells and the Suv39h dn ES cells rescued by WT Su39h1 (Fig. 5A). Two possible mechanisms of Suv39h-mediated DNA methylation have been proposed. One is the H3K9me3-mediated indirect recruitment mechanism because HP1 binds to Dnmt3b (6). The other one is the Suv39h direct mechanism because Suv39h1 can bind to Dnmt3a (38). Our data suggest the latter mechanism because WT Suv39h1 and ⌬N could rescue Dnmt3a and -3b, but not HP1 (data not shown) pericentromere recruitment in the Suv39h dn cells (Fig. 5, B-D).
Although DNA methylation of major satellite repeats in the Suv39h dn ES cells expressing ⌬N was rescued to the WT ES level, derepressed major satellite repeat were only partially suppressed. In contrast, WT Suv39h1 could completely suppress this derepression. It is known that HP1 and ATRX depleted from the pericentromeres play a role in transcriptional silencing (9, 39 -43). Our new experimental evidence further indi- HP1 also accumulated on pericentromere depending on both the Suv39h interaction and H3K9me3. Recruitment of other pericentromere proteins, such as Suv4-20h and ATRX, to the pericentric region was dependent on HP1 or HP1 and the H3K4me0/K9me3 interaction, respectively. The recruited Suv4-20h deposited H4K20me3. Suv39h also recruited Dnmt3a/b to the pericentromere and induced DNA methylation of major satellite repeats. The epigenetically organized structure was crucial for functionally silent heterochromatin and repression of major satellite repeats. B, in the Suv39h dn ES cells, HP1, Suv4-20h, ATRX, and Dnmt3a/b were depleted from the pericentromere due to deficiency of Suv39h and Suv39h-mediated H3K9me3. Because Suv4-20h and Dnmt3a were depleted from the pericentromere, H4K20me3 and DNA methylation were severely reduced. Major satellite repeats were derepressed. C, in the Suv39h dn ES cells expressing Suv39h1⌬N, ⌬N localized to the pericentromere and deposited H3K9me3. However, HP1 pericentromere accumulation was generally not recovered even though H3K9me3 was restored because the HP1-Suv39h interaction was missing. Suv4-20h, H4K20me3, and ATRX were not restored on the pericentromere likely due to a lack of HP1 pericentromere accumulation. However, ⌬N recruited Dnmt3a/b causing DNA methylation. This incomplete heterochromatin could partially silence transcription of major satellite repeats.
cates that Suv39h is the major master regulator of heterochromatin formation. Thus, it recruits multiple downstream molecules to the pericentric regions and creates a functional heterochromatin structure using several mechanisms such as a direct interaction or a H3K9me-mediated interaction.
In conclusion, we have shown that the HP1 interaction with Suv39h or HP1 pericentromere accumulation is mostly dispensable for Suv39h accumulation and H3K9me3 formation at pericentric regions. However, the mechanism of Suv39h or H3K9me3 target specificity to the pericentromere still remains unknown. We hope our new findings will provide some insights to understand this challenging and long-standing problem.