H3K27me1 is essential for MMP-9-dependent H3N-terminal tail proteolysis during osteoclastogenesis

H3K27me1 is essential for MMP-9-dependent H3N-terminal tail proteolysis during osteoclastogenesis Background: MMP-9 plays a direct role in the activation of pro-osteoclastogenic genes by cleaving histone H3N- terminal tail (H3NT ) and altering chromatin architecture. Although H3 acetylation at K18 has been shown to stimulate MMP-9 enzymatic activity toward H3NT, nothing is known about the influence of other H3NT modifications on this epigenetic reaction. Results: We show that H3 monomethylation at lysine 27 (H3K27me1) is essential for MMP-9-dependent H3NT prote- olysis during RANKL-induced osteoclast differentiation. Through the recognition of H3K27me1 mark, MMP-9 localizes and generates H3NT proteolysis at the genes encoding osteoclast differentiation factors. By using RNAi and small molecule inhibitor approaches, we also confirmed that G9a is the major methyltransferase to catalyze H3K27me1 for MMP-9-dependent H3NT proteolysis and trigger the expression of osteoclast-specific genes. Conclusions: Our data establish new functions for G9a-mediated H3K27me1 in MMP-9-dependent H3NT proteolysis and demonstrate how histone modification can be exploited to regulate osteoclastogenic gene expression at the molecular level. Further studies are warranted to investigate the detailed mechanism by which G9a overexpression with concomitant dysregulation of osteoclastogenesis contributes to the pathogenesis of bone disorders. Keywords: MMP-9, G9a, H3 proteolysis, H3K27me1, Osteoclast differentiation Background receptor RANK on pre-osteoclast cell membrane stimu- Bone is a highly dynamic organ that is continuously lates the expression of key determinants of osteoclast remodeled by coordinated activities of two cell types, differentiation such as NF-κB, c-Fos, and NFATc1 at the osteoclasts and osteoblasts [1, 2]. Osteoblasts are mono- early stage of the process [1, 8]. These factors then initiate nucleated mesenchymal stem cells that form bone matrix, multiple signal transduction pathways to turn on down- whereas osteoclasts are bone-resorbing multinucleated stream genes and activate quiescent osteoclast precur- cells that differentiate from hematopoietic progenitors sor (OCP) cells to become mature osteoclasts [1, 8]. The of the myeloid lineage [3–5]. Osteoclast differentiation is excessive formation and activity of osteoclasts lead to induced by receptor activator of NF-κB ligand (RANKL), pathological bone diseases such as osteoporosis, rheuma- which is expressed as a membrane-bound protein in toid arthritis, and tumor bone metastases [9, 10]. osteoblasts and provides osteoclast-specific differentia - As for other eukaryotic genes, osteoclastogenic tion signals [6, 7]. The binding of RANKL to its cognate gene expression occurs in the context of chromatin, where DNA is wound around histone proteins to form chromatin structure [11, 12]. An essential step for *Correspondence: woojinan@usc.edu understanding gene regulatory pathways at key dif- Kyunghwan Kim and Yonghwan Shin contributed equally to this work ferentiation time points,  therefore, should  lie in  char- Department of Biochemistry and Molecular Medicine, Norris acterizing the enzymes responsible for reorganizing Comprehensive Cancer Center, University of Southern California, Los Angeles, CA 90089, USA and potentiating particular chromatin domains. Not Full list of author information is available at the end of the article © The Author(s) 2018. This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creat iveco mmons .org/licen ses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creat iveco mmons .org/ publi cdoma in/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Kim et al. Epigenetics & Chromatin (2018) 11:23 Page 2 of 15 much progress has been made in supporting this idea, osteoclastogenic gene expression and MMP-9 transacti- but there is some indirect evidence functionally link- vation potential in OCP-induced cells. ing chromatin modification to osteoclastogenic gene In this study, we demonstrate that H3K27 monometh- transcription. For example, the repressive histone mark ylation (H3K27me1) is necessary for the localization H3K27me3 is removed from the master osteoclasto- and function of MMP-9 at osteoclastogenic genes as a genic gene NFATc1 [13], while it is deposited into the protease catalyzing H3NT proteolysis. The observed anti-osteoclastogenic gene IRF8 [14]. Histone deacety- H3K27me1 is attributed to G9a and stabilizes the inter- lase inhibitors have antagonistic effects on osteoclast action of MMP-9 with nucleosomes. Supporting these differentiation through the inactivation of NF-κB sign - data, selective inhibition of G9a-mediated H3K27me1 aling pathways [6, 15], suggesting a potential role for resulted in complete abrogation of RANKL-induced histone acetylation in osteoclastogenic gene expression. osteoclast formation and gene transcription. Therefore, Nonetheless, which factors are mainly responsible for G9a is a previously unrecognized regulator of osteoclast establishing transcriptionally competent chromatin differentiation with a unique function relevant to MMP- states and exactly how altered chromatin states trig- 9-dependent H3NT proteolysis. ger osteoclastogenic gene transcription remain poorly understood. Another large gap in our understanding of Results osteoclastogenic transcription program is the identifi - H3K27me1 is necessary for MMP‑9‑dependent H3NT cation of chromatin factors that have the potential to proteolysis of nucleosome substrates link specific aspects of chromatin function to osteoclast We have recently demonstrated that p300/CBP-medi- differentiation processes. ated H3K18ac neutralizes the charged lysine residue at Matrix metalloproteinases (MMPs) are a large family the primary cleavage site (P1) of MMP-9 and amplifies of extracellular enzymes, which function to remodel the MMP-9 enzymatic activity toward both free and nucleo- pericellular environment, primarily through the cleav- somal H3 substrates [20]. Since H3NTs are also subject age of extracellular matrix proteins [16–18]. MMP-9, to other types of modifications such as methylation and of special interest here, is a member of the MMP family phosphorylation, investigating their possible effects on and contains several conserved domains such as propep- MMP-9-dependent H3NT proteolysis should be a logi- 2+ tide domain, catalytic domain with Z n -binding site, cal extension of our study (Fig.  1a). Toward this end, we and hemopexin-like domain [17]. Like other MMP fam- prepared recombinant histone octamers containing H3 ily members, MMP-9 is first synthesized as an inactive/ analogs that are mono-, di-, or tri-methylated (me1, me2, latent 92-kDa proenzyme and subsequently converted to or me3) at K4, K9, K27, or K36 and phosphorylated (p) at a fully active 82-kDa form by removal of the N-terminal T3, S10, or S28. These histone octamers were then used propeptide domain [19]. Regarding MMP-9 functions, to reconstitute differentially methylated or phosphoryl - MMP-9 has been characterized as a major endopepti- ated nucleosome arrays on a tandem DNA array contain- dase with an ability to degrade extracellular matrix and ing seven copies of a 207-bp 601 nucleosome positioning stimulate osteoclastogenesis [19]. Unexpectedly, how- sequence. In our initial H3NT proteolysis assays with his- ever, our recent study revealed the nuclear function of tone octamers containing methylated H3 as substrates, MMP-9 as a protease that cleaves the histone H3N-ter- an H3 C-terminal antibody detected a faster-migrating minal tail (H3NT) during osteoclast differentiation [20]. band representing H3NT-cleaved product in all reactions In strong support of these observations, our subcellular (Fig.  1c, Oct). It was apparent in these experiments that localization analysis by Western blotting, gelatin zymog- cleavage levels of methylated H3 were comparable with raphy, and immunofluorescence clearly demonstrated the those of unmodified H3. Similarly, phosphorylation of H3 nuclear translocation and accumulation of MMP-9 dur- at T3, S10, or S28 showed no effects on MMP-9-depend - ing RANKL-induced formation of mature osteoclasts. ent H3NT proteolysis in identical assays (Fig.  1b, Oct). A functional role for the observed H3NT proteolysis in When we extended in  vitro cleavage assays to reconsti- osteoclastogenic gene transcription was evident from our tuted nucleosome arrays, MMP-9 was unable to prote- analysis of MMP-9-depleted OCP cells that had unde- olyze H3NT in nucleosome array substrates carrying tectable levels of H3NT proteolysis and OCP cell dif- H3K4me1/me2/me3, H3K9me1/me2/me3, H3K27me2/ ferentiation [20]. Our report also showed that MMP-9 me3, or H3K36me1/me2/me3 (Fig.  1c, Nuc). H3T3p, enzymatic activity toward H3NT is significantly aug - H3S10p, and H3S28p also had no effect on MMP-9-de - mented by H3K18 acetylation (H3K18ac) and that p300/ pendent H3NT proteolysis in our assays (Fig. 1b, Nuc). In CBP is responsible for H3K18ac observed in OCP cells sharp contrast, however, MMP-9 generated a reproduc- [20]. This is an important observation, meaning that ible H3NT proteolysis within H3K27me1 nucleosome p300/CBP-mediated H3K18ac has a functional role in arrays (Fig. 1c, Nuc). The observed effects are specific for Kim et al. Epigenetics & Chromatin (2018) 11:23 Page 3 of 15 Fig. 1 Requirement of H3K27me1 for MMP-9-dependent H3NT proteolysis. a Amino acid sequence of N-terminal tail (NT ) of histone H3. Amino acids that can be methylated (me) or phosphorylated (p) are indicated. b In vitro H3NT cleavage assays were performed using recombinant histone octamer or reconstituted nucleosome array substrates unmodified or phosphorylated at H3T3p, H3S10p, or H3S28p. The extent of H3NT proteolysis was analyzed by Western blot with H3 C-terminal tail (CT ) antibody. c As for b but using free histone octamers or nucleosome arrays methylated at H3K4, H3K9, H3K27, or H3K36 nucleosome substrates, since H3K27me1 did not show a osteoclasts in α-minimal essential medium (α-MEM) corresponding effect on free H3NT cleavage by MMP-9 following RANKL treatment for 0, 1, 3, or 5  days. In in our assays (Fig. 1c, Oct). Based on these data, we con- agreement with our published data, we detected a fast- cluded that H3K27me1 is required for MMP-9 to cleave migrating H3 band representing H3NT proteolysis and H3NT in a nucleosome context. an increase in MMP-9 expression after RANKL treat- ment (Fig.  2a). The loss of H3NT proteolysis following G9a catalyzes H3K27me1 and facilitates H3NT proteolysis MMP-9 knockdown is also consistent with our previ- during osteoclastogenesis ous demonstration of MMP-9-dependent H3NT prote- We next wanted to examine whether H3K27me1 is simi- olysis during osteoclast differentiation (Fig.  2b) [20]. In larly required for H3NT proteolysis during osteoclas- parallel experiments in which changes in H3K27me1 togenesis and, if so, which histone methyltransferase were analyzed over the same time period, a progressive (HMT) is responsible for H3K27me1. In our assay sys- increase in overall H3K27me1 levels was also evident tem, osteoclast precursor (OCP) cells are synchro- (Fig.  2a, b), suggesting its contribution to H3NT cleav- nously differentiated into TRAP-positive multinuclear age process. To confirm the significance of H3K27me1 (See figure on next page.) Fig. 2 Dependence of osteoclastogenic H3NT proteolysis and H3K27me1 on G9a. a Chromatin was extracted from primary OCP cells treated with RANKL for 0, 1, 3, and 5 days and analyzed by Western blotting with H2A, H2B, H3, and H4 CT antibodies (left panel). OCP-induced cells were also fixed with formaldehyde, stained for TRAP (tartrate-resistant acid phosphatase), and photographed under a light microscope (10×) (middle panel). TRAP-positive cells containing three or more nuclei were counted as osteoclasts at the indicated days (right panel). b As for a but using OCP-induced cells depleted of MMP-9. c As for a but using OCP-induced cells depleted of G9a. d As for a but using OCP-induced cells depleted of both MMP-9 and G9a Kim et al. Epigenetics & Chromatin (2018) 11:23 Page 4 of 15 0 1 3 5 Time (days) 0 1 3 5 Time (days) 0 1 3 5 Time (days) 0 1 3 5 Time (days) TRAP(+)M NCs/we ll TRAP(+)M NCs/we ll TRAP(+)M NCs/we ll TRAP(+)M NCs/we ll Kim et al. Epigenetics & Chromatin (2018) 11:23 Page 5 of 15 with respect to H3NT proteolysis more directly, we also EZH2 would affect H3K27me1 and H3NT proteolysis by transfected OCP-induced cells with plasmids express- using two inhibitors, BIX01294 and UNC1999. Of note, ing FLAG wild-type or K27-mutated H3 and prepared BIX01294 has an inhibitory property for the methyltrans- mononucleosomes as summarized in Additional file  1: ferase activity of G9a [26], whereas UNC1999 inhibits Fig. S1. Mononucleosomes containing ectopic H3 were the enzymatic activities of both EZH1 and EZH2 [27]. In then purified by immunoprecipitations using anti-FLAG analyzing the effects of EZH1/2 inhibitor UNC1999, we antibody. Our examination of purified mononucleosomes detected lower levels of H3K27me2 and H3K27me3, but by Western blot detected a high level of H3NT cleavage failed to see any apparent changes in MMP-9 protease in wild-type H3 nucleosomes, but the observed proteoly- activity toward H3NT (Additional file  5: Fig. S5). Mean- sis was impaired in H3K27-mutated nucleosomes (Addi- while, treatment of OCP-induced cells with 1.5  µM G9a tional file  2: Fig. S2). These data indicate strongly that inhibitor BIX01294 led to a pronounced inhibition of H3NT proteolysis in osteoclast differentiation pathway is H3NT proteolysis and osteoclast development (Fig.  3c). dependent on H3K27me1. Consistent with our published data [20], a significant To further investigate the role of H3K27me1 in the for- decrease in H3NT proteolysis was also observed upon mation of mature osteoclasts, it was important to identify treatment with 10  nM MMP-9 inhibitor I. Moreover, a the HMT mediating the observed H3K27me1. For this failure to generate higher levels of H3NT proteolysis and objective, we suppressed the expression of three HMTs, osteoclastogenesis upon treatment with both G9a inhibi- EZH1, EZH2, and G9a, that were shown to catalyze tor BIX01294 and MMP-9 inhibitor I, compared to the H3K27me1 [21–25] (Fig.  2c and Additional file  3: Fig. levels generated by their individual treatment, confirmed S3). When OCP cells were depleted of EZH1 or EZH2, again the dual requirement of G9a and MMP-9 for signal no obvious changes in H3K27me1 and H3NT proteolysis transduction pathways operating in osteoclast differen - were detected in our Western blot analysis of chromatin tiation (Fig. 3b, d). fraction from RANKL-induced OCP cells (Additional file  4: Fig. S4). Because knockdown of EZH1 and EZH2 G9a‑mediated H3K27me1 is crucial for MMP‑9 recruitment reduced levels of H3K27me2 and H3K27me3 (Additional and function at target genes file  4: Fig. S4), these results also indicate the indispen- Having established the requirement of G9a-mediated sable role of H3K27me1 in osteoclastogenic H3NT pro- H3K27me1 for MMP-9-dependent H3NT proteolysis teolysis. On the contrary, specific knockdown of G9a in and proficient osteoclast differentiation, we next sought OCP-induced cells efficiently blocked H3K27me1 and to explore whether G9a is also directly involved in the almost completely abrogated H3NT proteolysis in our expression of MMP-9 target genes. We recently devel- assays (Fig.  2c). Since G9a knockdown also decreased oped a technique called ChIP of acetylated chromatin the average number of mature osteoclasts (Fig.  2c), (ChIPac) (schematized in Additional file  6: Fig. S6) and these results confirm the functional contribution of demonstrated that H3NT proteolysis is associated with G9a to osteoclast formation. Considering the possibility RANKL-induced activation of genes necessary for osteo- that G9a could promote osteoclast differentiation inde - clast differentiation [20]. In this new method, we made pendently of MMP-9, we also checked whether double use of methylene blue to cross-link chromatin and ace- knockdown of G9a and MMP-9 exhibits more severe tic anhydride to completely acetylate all lysine residues in defects in osteoclastogenesis. Interestingly, however, the fragmented chromatin. H3K14ac-specific antibody was simultaneous knockdown of G9a and MMP-9 attenuated then used to selectively precipitate intact H3NT-contain- osteoclast differentiation in similar level as that observed ing chromatin, and a reduced PCR or sequencing signal in individual knockdown of G9a and MMP-9 (Fig.  2d). intensity relative to the control ChIPac reactions using an These data point to the dependence of pro-osteoclasto - H3CT antibody is indicative of osteoclastogenic H3NT genic function of MMP-9 on G9a-mediated H3K27me1 proteolysis. During the process of osteoclast formation, and constitute a powerful argument that both MMP-9 MMP-9 was shown to generate H3NT cleavage in pro- and G9a are essential for efficient osteoclastogenesis. moter, coding region, or both for target gene transcrip- In an attempt to support our knockdown data, we also tion in a gene-specific manner [20]. Thus, we examined tested whether chemical inhibition of G9a, EZH1, and the localization of G9a at Nfatc1, Lif, and Xpr1 genes (See figure on next page.) Fig. 3 Abolishment of osteoclastogenic H3NT proteolysis and H3K27me1 by G9a inhibitor. OCP cells were treated with DMSO control (a), MMP-9 inhibitor (b), G9a inhibitor (c), or MMP-9 + G9a inhibitors (d) and cultured for 0, 1, 3, and 5 days with RANKL. Chromatin was isolated for Western blot analysis to assess H3NT proteolysis (left panel). Cells were also TRAP-stained middle panel) and counted (right panel) as in Fig. 2 Kim et al. Epigenetics & Chromatin (2018) 11:23 Page 6 of 15 0 1 3 5 Time (days) 0 1 3 5 Time (days) 0 1 3 5 Time (days) 0 1 3 5 Time (days) TRAP(+)M NCs/we ll TRAP(+)M NCs/we ll TRAP(+)M NCs/we ll TRAP(+)M NCs/we ll Kim et al. Epigenetics & Chromatin (2018) 11:23 Page 7 of 15 representing each H3NT-cleaved group in OCP-induced accumulation of MMP-9 in the P, CR, and both regions cells by ChIPac-qPCR, as described recently [20]. Two of Nfatc1, Lif, and Xpr1 genes, respectively (Fig.  4a). sets of primers were used to detect MMP-9, G9a, and The localization patterns of MMP-9 were similar to H3K27me1 in the promoter (P) and coding region (CR) those observed for H3K27me1 in these target genes, by qPCR. implicating H3K27me1 as the major recruitment sig- Consistent with our recently published data [20], nal for MMP-9. H3K27me1 levels were reduced at the RANKL treatment of OCP cells resulted in a rapid target genes after G9a knockdown, and such changes Fig. 4 Impaired H3NT proteolysis and osteoclastogenic gene transcription in G9a-depleted OCP cells. a Control, MMP-9-, G9a-, or MMP-9 + G9a-depleted OCP cells were cultured with RANKL to induce osteoclastogenesis for 3 days, and ChIPac was performed using H3K14ac (H3NT ), H3CT, and H3K27me1 antibodies. H3NT cleavage levels were determined at the promoter (P) and coding regions (CR) of Nfatc1 (P-cleaved), Lif (CR-cleaved), and Xpr1 (P + CR-cleaved) genes by qPCR with primers used in our previous study [20] and are listed in “Methods” section. b RT-qPCR assays were performed to determine fold changes in Nfatc1, Lif, and Xpr1 expression in 3-day RANKL-induced OCP cells depleted of MMP-9 and/or G9a Kim et al. Epigenetics & Chromatin (2018) 11:23 Page 8 of 15 diminished the measured levels of P and CR occupied by G9a localization nor H3K27me1 at the target genes was MMP-9 (Fig. 4a). No detectable effects of MMP-9 knock - affected upon treatment with MMP-9 inhibitor (Fig. 5a). down on G9a occupancy at the target genes indicate that Given the demonstrated reliance of MMP-9 on G9a- MMP-9 is dispensable for G9a recruitment and function. mediated H3K27me1 for its target gene localization, we To further confirm the results, the ChIPac-qPCR assays also examined the role of G9a with respect to RANKL- were repeated using OCP-induced cells treated with induced expression of osteoclast-specific genes. Our MMP-9 and G9a inhibitors. As was observed with G9a RT-qPCR analysis showed that G9a depletion in OCP- knockdown, OCP-induced cells treated with G9a inhibi- induced cells caused three- to sevenfold decreases in tor show significantly lower levels of H3K27me1 and mRNA levels of Nfatc1, Lif, and Xpr1 genes (Fig.  4b). MMP-9 at the target genes (Fig.  5a). Expectedly, neither Congruent with these data, treatment of OCP cells with Fig. 5 Impaired H3NT proteolysis and osteoclastogenic gene transcription in G9a inhibitor-treated OCP cells. a ChIPac-qPCR assays were conducted to assess H3NT proteolysis levels in 3-day OCP-induced cells treated with MMP-9 and/or G9a inhibitors. b RT-qPCR to quantitate Nfatc1, Lif, and Xpr1 transcript levels in 3-day OCP-induced cells treated with MMP-9 and/or G9a inhibitors Kim et al. Epigenetics & Chromatin (2018) 11:23 Page 9 of 15 G9a inhibitor also led to impaired expression of the tar- MMP-9 retained strong affinity for H3K27me1 nucleo - get genes upon RANKL-induced differentiation (Fig.  5b). some, whereas no apparent interaction was observed In addition, when OCP cells were depleted of both G9a with the remainder (amino acids 448–730) of the protein and MMP-9 or treated with G9a and MMP-9 inhibitors (Fig.  6a, d). Also, in similar binding experiments using together, target gene transcription was repressed, but the three N-terminal subregions, MMP-9 amino acids 384– level of repression was similar to that detected in knock- 447 directly interacted with H3K27me1 nucleosome, down or inhibition of G9a or MMP-9, again pointing but MMP-9 amino acids 112–212 and 213–383 failed to to G9a-mediated H3K27me1 as an essential epigenetic show any detectable interaction under the same condi- mark for MMP-9 recruitment and function (Figs. 4b, 5b). tions (Fig. 6e). Together, these data support the direct function of G9a In order to gain more support for the in vitro binding in regulating MMP-9 target gene pathways necessary for results above, OCP cells were transfected with expres- proficient osteoclast differentiation. sion vectors for FLAG-H3 wild type or K27R mutant. After treating OCP cells with RANKL for 3  days, sol- MMP‑9 specifically binds to H3K27me1 nucleosomes uble chromatin was prepared from cell nuclei and The results of the above experiments argue persuasively digested with micrococcal nuclease to yield mainly that H3K27me1 is indispensable for the stable localiza- mononucleosomes. Mononucleosomes containing tion and function of MMP-9 at target genes. However, it ectopic H3 wild type or K27R mutant were then spe- is not clear whether the observed effects of H3K27me1 cifically immunoprecipitated with anti-FLAG anti - reflect its role as a docking site to facilitate the recruit - body. When we analyzed the association of endogenous ment of MMP-9 to target genes. To check this possibility, MMP-9 with these purified nucleosomes, we found that H3NT unmodified or K27me1 peptides corresponding to MMP-9 bound avidly to wild-type H3 nucleosomes, but amino acids 1–21 or 21–44 were immobilized on strepta- bound minimally to K27R-mutated H3 nucleosomes vidin-coated wells and monitored the binding for MMP- (Fig.  6f ). To identify amino acid residues critical for 9. Our results showed that MMP-9 strongly binds to the MMP-9-H3NTK27me1 interaction, we next gener- H3NT K27me1 peptides, whereas other H3NT peptides ated a structural model of the MMP-9-H3NTK27me1 displayed only weak interaction with MMP-9 (Fig.  6b). complex based on existing crystal structures using To further confirm these results, we reconstituted nucle - the program Cluspro 2.0 [28–30] (Fig.  6g). The model osomes containing H3 unmodified or K27me1 on a identified E402 of MMP-9 to contact anionic residues 2+ 601 nucleosome positioning sequence and checked the in H3NTK27me1. Moreover, the Z n -binding site of binding of MMP-9. In agreement with the interactions MMP-9 was suggested to interact with H3NTK27me1. observed with H3NT peptides, we detected a remark- Consistent with this model, in vitro binding assays test- able binding preference of MMP-9 for the immobilized ing the E402A substitution showed a diminished bind- nucleosome containing H3K27me1 over the nucleo- ing of MMP-9 to H3NTK27me1, supporting a role some containing H3 unmodified (Fig.  6c). Additionally, for electrostatic contacts involving this residue in the in mapping the interaction region of mature MMP-9, we interaction (Fig.  6h). In contrast, the H411A substitu- found that N-terminal region (amino acids 112–447) of tion was inconspicuous, which leaves the role of the (See figure on next page.) Fig. 6 MMP-9 binding to H3K27me1 nucleosomes. a Schematic depiction of the domain structure of MMP-9. b Peptide pull-down assays with biotinylated H3 1–21 and 21–44 peptides and recombinant His-MMP-9 were analyzed by Western blotting with anti-His antibody. H3 peptides were unmodified, K18ac or K27me1 as indicated. Lane 1 represents 10% of the input MMP-9. c Nucleosomes were reconstituted on a 207-bp 601 nucleosome positioning sequence using unmodified or H3K27me1 histone octamers and immobilized on streptavidin beads. His-MMP-9 was incubated with immobilized nucleosomes, and its binding to nucleosomes was analyzed by Western blotting with anti-His antibody. Lane 1 contains 10% of the input MMP-9. d H3K27me1 nucleosomes were incubated with immobilized MMP-9 N-terminal (amino acids 112–447) and C-terminal (amino acids 448–730) domains. After extensive washing, the binding of H3K27me1 nucleosomes to MMP-9 domains was determined by Western blotting with anti-H3 antibody. Input corresponds to 10% of H3K27me1 nucleosomes used in the binding reactions. e After incubation with H3K27me1 nucleosomes, the binding of MMP-9N-terminal subregions to nucleosomes was determined by Western blotting with anti-His antibody. Input lanes 1–3 represent 10% of MMP-9 fragments used in the binding reactions. f OCP-induced cells were transfected with FLAG-H3 wild type ( WT ) or K27R mutant (K27R), and mononucleosomes were prepared by micrococcal nuclease digestion as summarized in Figure S3. Mononucleosomes containing ectopic H3 were immunoprecipitated from total mononucleosomes with FLAG antibody and analyzed by Western blotting with anti-MMP-9 antibody. g Model of the MMP-9-H3K27me1 interaction. PDB entries 4h3x (mMMP-9) and 3avr (H3.1) were used in docking simulations using the program Cluspro 2.0 [28–30]. Simulations were run with non-methylated H3. For context, H3K27 is shown monomethylated. h Nucleosome binding assays were conducted as in e, except that His-MMP-9 amino acids 384–447 carrying E402A mutation were used Kim et al. Epigenetics & Chromatin (2018) 11:23 Page 10 of 15 2+ levels. Together, these results constitute a powerful Zn -binding site ambiguous (Fig.  6h). As judged by argument that, although other factors and signals may crystal structures of methyltransferases and demethy- be involved, the initial recruitment of MMP-9 at tar- lases [28, 30–32], the differentiation of H3K27 meth - get genes is dependent on the recognition of H3NT- ylation states by MMP-9 likely arises from steric and K27me1 through the N-terminal domain of MMP-9. electrostatic differences between different methylation Kim et al. Epigenetics & Chromatin (2018) 11:23 Page 11 of 15 In the present study, we focused on potential roles of H3 methylation and phosphorylation in MMP-9-de- pendent H3NT proteolysis. Unexpectedly, our in  vitro cleavage assays with nucleosome arrays containing dif- ferentially modified H3 revealed that H3K27me1 is nec - essary for H3NT proteolytic activity of MMP-9. Our observation that MMP-9 can efficiently cleave free H3 substrates regardless of their H3K27me1 states is sup- portive of the idea that H3K27me1 specifically enhances MMP-9 enzymatic activity toward H3NT in a nucleo- some context. We extended these in vitro findings in sub - sequent cellular studies revealing that MMP-9 protease activity toward H3NT is dependent on G9a-mediated H3K27me1 during RANKL-induced osteoclast forma- tion. Notably, data presented here also indicate that G9a levels are highly elevated in response to RANKL treat- ment, leading to a sharp increase in H3K27me1. More- over, the level of G9a-mediated H3K27me1 correlates Fig. 7 Working model. Our previous study revealed that MMP-9 directly with the extent of MMP-9-dependent H3NT catalyzes H3NT proteolysis for osteoclastogenic gene activation proteolysis and osteoclastogenesis. This finding under - and that CBP/p300-mediated H3K18ac facilitates MMP-9 protease scores the importance of H3K27me1 in accurate target- activity. Here, we depict that G9a-mediated H3K27me1 is necessary ing of MMP-9 within defined chromatin regions, and for MMP-9-dependent H3NT proteolysis in chromatin and proficient osteoclast differentiation. Without G9a-mediated H3K27me1, MMP-9 taken together with our recently described H3K18ac- cannot localize at osteoclastogenic genes and mediate H3NT augmented MMP-9 clipping activity, it also suggests that proteolysis. In the presence of G9a-mediated H3K27me1, MMP-9 these epigenetic alterations in H3NT directly influence gets recruited to target loci, facilitating H3NT proteolysis, leading MMP-9-dependent expression of pro-osteoclastogenic to RANKL-induced gene transcription, and promoting osteoclast genes (Fig. 7). differentiation Even though EZH1 and EZH2 were shown to medi- ate H3K27me1 in  vitro assays and in certain cell types [21–24], our knockdown data indicate that they do not Discussion participate in generating H3K27me1 and that G9a is There has been a growing interest in understanding how mainly responsible for catalyzing this osteoclastogenic epigenetic processes regulate gene transcription at vari- histone mark. Another argument in favor of G9a to act ous stages of osteoclast differentiation, activation, and as the major H3K27me1 methyltransferase comes from survival. MMP-9 is highly expressed in OCP cells and is the experiments showing that treating OCP cells with known to play a key role in RANKL-induced osteoclast G9a inhibitors, but not with EZH1/EZH2 inhibitors, differentiation. The general view of MMP-9 function in leads to near complete loss of H3K27me1 and H3NT osteoclastogenesis is that MMP-9 digests structural com- proteolysis. This finding is consistent with a recent report ponents of extracellular matrix and cellular surface facili- that G9a inhibitor BIX01294 reduced RANKL-induced tating cell migration and adhesion [16]. However, this osteoclast formation from RAW 264.7 cells [33]. In this common idea has been changed recently by our report study, H3K9me was assumed to play a causal role for G9a documenting that, beside regulating extracellular matrix stimulation of osteoclast differentiation, and BIX01294 remodeling, MMP-9 moves into the nucleus and gener- treatment was considered to have inhibitory effects on ates active transcription states of osteoclastogenic genes H3K9me process. However, our data provide strong by proteolytically cleaving H3NT at promoter and cod- evidence for the significance of H3K27me1, rather than ing regions [20]. Also, our study provided the first evi - H3K9me, for the osteoclastogenic function of G9a. Also, dence that p300/CBP-mediated H3K18ac is required for G9a inhibitor-enhanced osteoclastogenesis intrinsically MMP-9 to mediate H3NT proteolysis and active expres- depends upon MMP-9 function as an H3NT protease, sion of genes encoding positive regulators of osteoclas- since knockdown of G9a in MMP-9-depleted OCP cells togenesis [20]. Since other histone modifications can also failed to generate more attenuation of osteoclastogenesis. affect gene expression, one obvious gap in our under - In this regard, H3K27me1 should be recognized as an standing of MMP-9-regulated transcription mechanism epigenetic mark reflecting the initiation and progression lies in the identification of additional histone marks that of osteoclastogenic process. may be involved in MMP-9 activity toward H3NT. Kim et al. Epigenetics & Chromatin (2018) 11:23 Page 12 of 15 Further analysis of G9a-depleted OCP cells showed activity could provide an effective treatment for bone loss that G9a has a direct impact on the expression of osteo- diseases. clastogenic genes and revealed an intriguing new tran- scription pathway regulating osteoclast differentiation. Conclusions After identifying osteoclast gene transcription being Results presented here describe a role for G9a in MMP- up-regulated by G9a, we also realized that these genes 9-dependent H3NT proteolysis and gene transcription were enriched by H3K27me1, MMP-9, and cleaved by catalyzing H3K27me1 during RANKL-induced osteo- H3 species in a G9a-dependent manner. These results clast differentiation. demonstrate the direct action of G9a on transcriptional These results establish a direct functional link between program required for osteoclastogenesis and support G9a and MMP-9 in the context of pro-osteoclastogenic the concept that G9a exerts a coactivation function in transcriptional programs. Our model suggests that G9a- MMP-9-driven transactivation. Our ChIPac analyses mediated H3K27me1 serves as an essential mark for showed that G9a generates H3K27me1 in promoter MMP-9 recruitment and proteolytic activity at genes or coding regions or both regions of osteoclastogenic encoding positive regulators of osteoclast differentiation, genes and that H3NT proteolysis patterns match well thus keeping them active (Fig. 7). with H3K27me1 states. However, high levels of tran- scription were accomplished in all these cases. This Methods observation suggests a function for G9a-mediated Plasmid construction and materials H3K27me1 at the initiation or elongation step of tran- Core histones and MMP-9 proteins were expressed in scription in a gene-specific manner. It is also con - Escherichia coli Rosetta 2 (DE3) pLysS cells (Novagen) ceivable that the levels of H3NT proteolysis are not and purified from inclusion bodies as described recently proportional to transcription rates and H3NT prote- [20]. To generate mutant H3 and MMP-9 expression olysis targeted to certain regions is sufficient to induce vectors, H3 and MMP-9 cDNAs were mutated by the gene transcription. QuikChange II site-directed mutagenesis kit (Agilent Toward understanding how H3K27me1 can facilitate Technologies) before the construction. Further details MMP-9-dependent H3NT proteolysis, we demonstrate of plasmid constructions are available upon request. that H3K27me1 is essential for MMP-9 binding to nucle- G9a inhibitor BIX01294 is from Santa Cruz Biotech, and osomal H3NTs. Only when H3K27me1 was introduced EZH1/2 inhibitor UNC1999 and MMP-9 Inhibitor I are into nucleosomes did it contribute to MMP-9 proteolytic from Sigma. Antibodies used in this study are as follows: activity. Thus, MMP-9 appears to utilize some specific H2A, H2B, H3, H4, and EZH2 antibodies from Abcam; structural features to recognize H3K27me1 nucleosomes H3K27me1 and EZH1 antibodies from Millipore; G9a, in osteoclastogenic target genes (Fig. 7). Reciprocally, the actin, and FLAG antibodies from Sigma; His antibody results also suggest that H3K27me1 may serve principally from Novagen; and MMP-9 antibody from Santa Cruz as a mark for MMP-9 recruitment to target genes, with Biotech. no additional role in MMP-9-dependent H3NT prote- olysis. These observations are reminiscent of cathepsin In vitro H3NT cleavage assays L, of which capability to catalyze H3NT proteolysis was Recombinant histone octamers and nucleosome arrays enhanced by H3K27me2 [34]. Nonetheless, no detect- containing unmodified, methylated, or phosphorylated able effects of H3K27me2 in our assays suggest the pres - H3 were prepared following the procedure described [20, ence of two different regulatory mechanisms involving 36]. MMP-9 was incubated with 1 µg of histone octamer H3K27me1 for MMP-9 and H3K27me2 for cathepsin L. or 2  µg of nucleosome arrays, and H3NT cleavage was u Th s, in addition to its known contribution to gene tran - determined by Western blotting with H3 C-terminal scription [21, 35], our results underscore the significance antibody [20]. of H3K27me1 in helping MMP-9 target specific genes to mediate H3NT proteolysis. Defining the molecular basis Osteoclast differentiation and H3NT cleavage analysis for H3K27me1 effects on MMP-9 activity is beyond the Osteoclast precursor (OCP) cells were prepared as scope of this first report but will be of interest for how recently described [20]. To generate osteoclasts, OCP epigenetic signals may alter intrinsic MMP-9 proper- cells were cultured in the presence of 30  ng/ml mac- ties in OCP cells in response to RANKL stimulation. rophage colony-stimulating factor (M-CSF) and 50  ng/ Based on the functional connection uncovered between ml receptor activator of nuclear factor kappaB ligand G9a-mediated H3K27me1 and MMP-9-dependent (RANKL). On days 0, 1, 3, and 5, the cells were fixed H3NT proteolysis, we speculate that inventing strate- with formaldehyde and stained for tartrate-resistant gies to block  osteoclastogenic G9a methyltransferase acid phosphatase (TRAP) using an acid phosphatase Kim et al. Epigenetics & Chromatin (2018) 11:23 Page 13 of 15 leukocyte kit (Sigma). TRAP-positive multinucleated 5′-GGT GGG TTC CAC TGA AAG AA-3′, 5′-GGT TCC cells containing three or more nuclei were counted TCT GAC CAA AAG CA-3′; mRNA: 5′-AGG AGC GTG as osteoclasts under a light microscope. In certain TCC AAC ATA GG-3′, 5′-CCA CGA GAT GTT TCC AGG instances, media were supplemented with G9a inhibi- AT-3′). tor BIX01294 (1.5  µM), EZH1/2 inhibitor UNC1999 (2 µM), and MMP-9 inhibitor I (10 nM) to evaluate their H3 tail peptide and nucleosome binding assays effects on OCP cell differentiation. To determine the For H3NT peptide binding assays, biotinylated forms of levels of H3NT proteolysis, nuclei were isolated from H3NT peptides unmodified, acetylated at K18, or mono - OCP-induced cells in buffer A (10  mM HEPES, pH 7.4, methylated at K27 (EZBiolab Inc) (2  μg) were immobi- 10  mM KCl, 1.5  mM MgCl2, 0.34  M sucrose, 10% glyc- lized on streptavidin–agarose beads. After washing with erol, 1 mM DTT, 5 mM β-glycerophosphate, 10 mM NaF, BC250/0.1% Nonidet P-40, His-MMP-9 was incubated protease inhibitors, and 0.2% Triton X-100) and chro- with H3NT peptides-bound beads in BC200/0.1% Noni- matin was extracted in buffer B (3  mM EDTA, 0.2  mM det P-40 for 3  h at room temperature. After extensive EGTA, 1  mM DTT, 5  mM β-glycerophosphate, 10  mM washing with BC200/0.1% NP-40, MMP-9 interaction NaF, and protease inhibitors). Western blot analysis was was analyzed by Western blotting with anti-His antibody. performed using H3 C-terminal antibody as previously For nucleosome binding assay, H3 unmodified/H3K18ac/ described [20]. H3K27me1 nucleosomes were reconstituted by mixing recombinant histone octamers and biotinylated 207- RNA interference, RT‑qPCR, and ChIPac‑qPCR bp 601 nucleosome positioning sequence templates at a Lentiviral particles were generated in HEK-293T cells ratio of 1:1.2 (w/w) and salt gradient dialysis and purified by co-transfecting plasmids encoding VSV-G, NL-BH, by sedimentation in a 5–30% (vol/vol) glycerol gradient and pLKO.1-shRNA (Addgene) for MMP-9 (5′-GAG as described previously [36]. Nucleosomes (1  μg) were GCA TAC TTG TAC CGC TAT-3) or G9a (5′-AGA CAT immobilized on streptavidin–agarose beads (Novagen) TTC TCC ATC AGA GAC-3′). OCP cells were transduced and incubated with His-MMP-9 proteins for 16 h on ice. with these viruses or 10  nM MMP-9-INI (Santa Cruz) After washing with BC250/0.1% Nonidet P-40, bound for 3  days prior to differentiation. Total RNA was iso - MMP-9 proteins were detected by Western blotting. lated from OCP-induced cells using the Qiagen RNeasy kit (Qiagen, Valencia, CA) and reverse-transcribed using the iScript cDNA synthesis kit (Bio-Rad) and PerfeCta Nucleosome purification and analysis SYBR Green FastMix (Quanta Biosciences). ChIPac- OCP cells were transfected with expression vectors for qPCR assays were performed using chromatin that was H3 wild type or K27R mutant containing a C-terminal fixed with 10  µM methylene blue and acetylated with FLAG tag. After 3-day RANKL treatment, cells were har- 20  mM acetic anhydride as described [20]. H3K14ac, vested and lysed with buffer A (20  mM HEPES, pH 7.4, H3CT, and H3K27me1 antibodies were used to immuno- 10  mM KCl, 1.5  mM MgCl2, 0.34  M sucrose, 10% glyc- precipitate cross-linked chromatin. The immunoprecipi - erol, 1  mM dithiothreitol, and protease inhibitor cock- tated protein–DNA complexes were recovered, washed, tail) containing 0.2% Triton X-100. Nuclei were pelleted and incubated overnight at 65  °C to reverse the cross- by centrifugation at 1000g, resuspended in buffer A con - linking. DNA fragments were purified and analyzed with taining 2  mM CaCl , and digested with 0.6 U micrococ- the primers that amplify the promoter (P) and coding cal nuclease (Sigma) at 37 °C for 20 min. Digested nuclei regions (CR) of Nfatc1 (P-cleaved), Lif (CR-cleaved), and were collected and incubated in nuclear extraction buffer Xpr1 (P + CR-cleaved) genes. The sequences of primers (20  mM HEPES, pH 7.4, 420  mM NaCl, 1.5  mM M gCl , used for qPCR are as follows: Nfatc1 (P: 5′-GAA GTG 0.2  mM EGTA, and protease inhibitor cocktail) for 1  h GTA GCC CAC GTG AT-3′, 5′-TCT TGG CAC CAC ATA and centrifuged to remove nuclear debris. After adjust- AAC CA-3′; CR: 5′-GGG TCA GTG TGA CCG AAG AT-3′, ing the salt concentration of the extract to 150 mM NaCl, 5′-GGA AGT CAG AAG TGG GTG GA-3′; mRNA: 5′-CTC ectopic H3-containing nucleosomes were isolated by GAA AGA CAG CAC TGG AGCAT-3′, 5′-CGG CTG CCT immunoprecipitation using anti-FLAG M2 agarose beads TCC GTC TCA TAG-3′), Lif (P: 5′-CTC TGG CTG TCC in washing buffer (20 mM HEPES, pH 7.8, 300 mM NaCl, TGG AAC TC-3′, 5′-CCA GGA CCA GGT GAA ACA CT-3′; 1.5 mM MgCl , 0.2 mM EGTA, 10% glycerol, 0.2% Triton CR: 5′-ATC TTG TGG CTT TGC CAA CT-3′, 5′-AGT X-100, and protease inhibitor cocktail). Levels of H3NT CCT TGC CTG TCT TTC CA-3′; mRNA: 5′-TAC TGC proteolysis of bead-bound nucleosomes were analyzed by TGC TGG TTC TGC AC-3′, 5′-TGA GCT GTG CCA GTT Western blotting with anti-FLAG antibody. The purified GAT TC-3′), and Xpr1 (P: 5′-AGG ACC TTC GGA AGA nucleosomes were also subject to Western blotting with GCA GT-3′, 5′-CAG CAA GCA GCT CAT AAC CA-3′; CR: anti-MMP-9 antibody. Kim et al. Epigenetics & Chromatin (2018) 11:23 Page 14 of 15 Consent for publication Statistical analysis Not applicable. All quantitative data are presented as mean ± SD. Statis- tical analyses of datasets were performed with Student’s Ethics approval and consent to participate Not applicable. two-tailed t test or two-way ANOVA followed by Bonfer- roni’s comparison test. GraphPad Prism (GraphPad Soft- Funding ware Inc.) was used for all analyses. A P value < 0.05 was This work was supported by NIH Grant CA201561 awarded to W.A. The study was also funded by Pilot Project Grants from Keck School of Medicine of USC. considered statistically significant. This work was supported in part by the National Research Foundation of Korea (NRF) Grant 2017R1C1B2008017. Additional files Publisher’s Note Additional file 1: Fig. S1. Workflow of the purification method used for Springer Nature remains neutral with regard to jurisdictional claims in pub- isolation of ectopic H3 nucleosomes. lished maps and institutional affiliations. Additional file 2: Fig. S2. Abolishment of osteoclastogenic H3NT proteolysis by H3K27R mutation. Mononucleosomes containing ectopic Received: 6 February 2018 Accepted: 21 May 2018 H3 were purified from OCP-induced cells expressing H3 wild type or K27R mutant with C-terminal FLAG tag as summarized in Additional file 1: Fig. S1 and analyzed by Western blotting with anti-FLAG antibody. Additional file 3: Fig. S3. Validation of specific knockdown of EZH1 and References EZH2. OCP cells were transduced with lentiviral shRNAs targeting EZH1 (a) 1. Matsuo K, Irie N. Osteoclast-osteoblast communication. Arch Biochem and EZH2 (b), and knockdown efficiency and specificity were determined Biophys. 2008;473:201–9. by Western blot. 2. Nakahama K. Cellular communications in bone homeostasis and repair. Additional file 4: Fig. S4. Determination of the effects of EZH1/EZH2 Cell Mol Life Sci. 2010;67:4001–9. knockdown on osteoclastogenic H3NT proteolysis. a Chromatin was puri- 3. Ash P, Loutit JF, Townsend KMS. Osteoclasts derived from haematopoietic fied from mock-depleted, OCP-induced cells, and Western blot analysis stem cells. Nature. 1980;283:669. for H3NT proteolysis was performed as described in Fig. 2. b As for (a) but 4. Ikeda K, Takeshita S. Factors and mechanisms involved in the coupling using chromatin from EZH1-depleted, OCP-induced cells. c As for (a) but from bone resorption to formation: how osteoclasts talk to osteoblasts. J using chromatin from EZH2-depleted, OCP-induced cells. Bone Metab. 2014;21:163–7. 5. Raggatt LJ, Partridge NC. Cellular and molecular mechanisms of bone Additional file 5: Fig. S5. Determination of the effects of EZH1/EZH2 remodeling. J Biol Chem. 2010;285:25103–8. inhibitors on osteoclastogenic H3NT proteolysis. Chromatin was extracted 6. Nakashima T, Takayanagi H. New regulation mechanisms of osteoclast from OCP-induced cells after treating with an EZH1/EZH2 inhibitor and differentiation. Ann N Y Acad Sci. 2011;1240:E13–8. subject to Western blotting with H3CT antibody. 7. Yasuda H, Shima N, Nakagawa N, Yamaguchi K, Kinosaki M, Mochizuki S, Additional file 6: Fig. S6. Schematic representation of ChIPac-qPCR assay. Tomoyasu A, Yano K, Goto M, Murakami A, et al. Osteoclast differentia- tion factor is a ligand for osteoprotegerin/osteoclastogenesis-inhibitory factor and is identical to TRANCE/RANKL. Proc Natl Acad Sci USA. 1998;95:3597–602. Abbreviations 8. Teitelbaum SL, Ross FP. Genetic regulation of osteoclast development H3NT: histone H3N-terminal tail; H3K27me1: H3 monomethylation at lysine and function. Nat Rev Genet. 2003;4:638–49. 27; RANK: receptor activator of NF-κB ligand; MMPs: matrix metalloprotein- 9. Baron R. Arming the osteoclast. Nat Med. 2004;10:458–60. ases; OCP: osteoclast precursor; ChIPac: ChIP of acetylated chromatin; TRAP: 10. Ishida N, Hayashi K, Hoshijima M, Ogawa T, Koga S, Miyatake Y, Kumegawa tartrate-resistant acid phosphatase. M, Kimura T, Takeya T. Large scale gene expression analysis of osteo- clastogenesis in vitro and elucidation of NFAT2 as a key regulator. J Biol Authors’ contributions Chem. 2002;277:41147–56. KK and WA conceived and designed the study. TU offered advice on mutation 11. Tetsuro Y, Jun H, Hiroyuki A, Sakae T. Recent advance in epigenetics— experiment design. KK, YS, and JK performed the experiments with contribu- application to the regulation of osteoclast differentiation. Curr Rheumatol tions of TU and WA. YS and WA analyzed data and wrote the manuscript. All Rev. 2012;8:103–8. authors read and approved the final manuscript. 12. Vrtacnik P, Marc J, Ostanek B. Epigenetic mechanisms in bone. Clin Chem Lab Med. 2014;52:589–608. Author details 13. Yasui T, Hirose J, Tsutsumi S, Nakamura K, Aburatani H, Tanaka S. Epi- Department of Biochemistry and Molecular Medicine, Norris Comprehensive genetic regulation of osteoclast differentiation: possible involvement Cancer Center, University of Southern California, Los Angeles, CA 90089, USA. of Jmjd3 in the histone demethylation of Nfatc1. J Bone Miner Res. Department of Physiology and Neuroscience, Zilkha Neurogenetic Institute, 2011;26:2665–71. University of Southern California, Los Angeles, CA 90089, USA. Depart- 14. Fang C, Qiao Y, Mun SH, Lee MJ, Murata K, Bae S, Zhao B, Park-Min ment of Biology, College of Natural Sciences, Chungbuk National University, KH, Ivashkiv LB. Cutting edge: EZH2 promotes osteoclastogenesis Cheongju, Chungbuk 361-763, Republic of Korea. by epigenetic silencing of the negative regulator IRF8. J Immunol. 2016;196:4452–6. Acknowledgements 15. Cantley MD, Fairlie DP, Bartold PM, Rainsford KD, Le GT, Lucke AJ, Holding Not applicable. CA, Haynes DR. Inhibitors of histone deacetylases in class I and class II suppress human osteoclasts in vitro. J Cell Physiol. 2011;226:3233–41. Competing interests 16. Birkedal-Hansen H. Proteolytic remodeling of extracellular matrix. Curr The authors declare that they have no competing interests. Opin Cell Biol. 1995;7:728–35. 17. Papazafiropoulou A, Tentolouris N. Matrix metalloproteinases and cardio - Availability of data and materials vascular diseases. Hippokratia. 2009;13:76–82. All key data supporting the findings of this study are available within the 18. Vargova V, Pytliak M, Mechirova V. Matrix metalloproteinases. Exs. paper. Additional data and materials are available from the corresponding 2012;103:1–33. author upon request. Kim et al. Epigenetics & Chromatin (2018) 11:23 Page 15 of 15 19. Bergers G, Brekken R, McMahon G, Vu TH, Itoh T, Tamaki K, Tanzawa K, probe of the Lysine Methyltransferases EZH2 and EZH1. ACS Chem Biol. Thorpe P, Itohara S, Werb Z, Hanahan D. Matrix metalloproteinase-9 2013;8:1324–34. triggers the angiogenic switch during carcinogenesis. Nat Cell Biol. 28. Antoni C, Vera L, Devel L, Catalani MP, Czarny B, Cassar-Lajeunesse E, 2000;2:737–44. Nuti E, Rossello A, Dive V, Stura EA. Crystallization of bi-functional ligand 20. Kim K, Punj V, Kim JM, Lee S, Ulmer TS, Lu W, Rice JC, An W. MMP-9 facili- protein complexes. J Struct Biol. 2013;182:246–54. tates selective proteolysis of the histone H3 tail at genes necessary for 29. Kozakov D, Hall DR, Xia B, Porter KA, Padhorny D, Yueh C, Beglov D, proficient osteoclastogenesis. Genes Dev. 2016;30:208–19. Vajda S. The ClusPro web server for protein–protein docking. Nat Protoc. 21. Ferrari KJ, Scelfo A, Jammula S, Cuomo A, Barozzi I, Stutzer A, Fischle W, 2017;12:255. Bonaldi T, Pasini D. Polycomb-dependent H3K27me1 and H3K27me2 reg- 30. Sengoku T, Yokoyama S. Structural basis for histone H3 Lys 27 demethyla- ulate active transcription and enhancer fidelity. Mol Cell. 2014;53:49–62. tion by UTX/KDM6A. Genes Dev. 2011;25:2266–77. 22. McCabe MT, Graves AP, Ganji G, Diaz E, Halsey WS, Jiang Y, Smitheman 31. Del Rizzo PA, Trievel RC. Molecular basis for substrate recognition by KN, Ott HM, Pappalardi MB, Allen KE, et al. Mutation of A677 in histone lysine methyltransferases and demethylases. Biochim Biophys Acta. methyltransferase EZH2 in human B-cell lymphoma promotes hypertri- 2014;1839:1404–15. methylation of histone H3 on lysine 27 (H3K27). Proc Natl Acad Sci USA. 32. Kaniskan HU, Jin J. Recent progress in developing selective inhibitors of 2012;109:2989–94. protein methyltransferases. Curr Opin Chem Biol. 2017;39:100–8. 23. Sarma K, Margueron R, Ivanov A, Pirrotta V, Reinberg D. Ezh2 requires 33. Tsuda H, Zhao N, Imai K, Ochiai K, Yang P, Suzuki N. BIX01294 suppresses PHF1 to efficiently catalyze H3 lysine 27 trimethylation in vivo. Mol Cell osteoclast differentiation on mouse macrophage-like Raw264.7 cells. Biol. 2008;28:2718–31. Bosn J Basic Med Sci. 2013;13:271–5. 24. Shen X, Liu Y, Hsu YJ, Fujiwara Y, Kim J, Mao X, Yuan GC, Orkin SH. EZH1 34. Duncan EM, Muratore-Schroeder TL, Cook RG, Garcia BA, Shabanowitz J, mediates methylation on histone H3 lysine 27 and complements EZH2 Hunt DF, Allis CD. Cathepsin L proteolytically processes histone H3 during in maintaining stem cell identity and executing pluripotency. Mol Cell. mouse embryonic stem cell differentiation. Cell. 2008;135:284–94. 2008;32:491–502. 35. Steiner LA, Schulz VP, Maksimova Y, Wong C, Gallagher PG. Patterns of 25. Wu H, Chen X, Xiong J, Li Y, Li H, Ding X, Liu S, Chen S, Gao S, Zhu B. histone H3 lysine 27 monomethylation and erythroid cell type-specific Histone methyltransferase G9a contributes to H3K27 methylation in vivo. gene expression. J Biol Chem. 2011;286:39457–65. Cell Res. 2011;21:365–7. 36. Kim JM, Kim K, Punj V, Liang G, Ulmer TS, Lu W, An W. Linker histone H1.2 26. Kubicek S, O’Sullivan RJ, August EM, Hickey ER, Zhang Q, Teodoro ML, Rea establishes chromatin compaction and gene silencing through recogni- S, Mechtler K, Kowalski JA, Homon CA, et al. Reversal of H3K9me2 by a tion of H3K27me3. Sci Rep. 2015;5:16714. small-molecule inhibitor for the G9a histone methyltransferase. Mol Cell. 2007;25:473–81. 27. Konze KD, Ma A, Li F, Barsyte-Lovejoy D, Parton T, Macnevin CJ, Liu F, Gao C, Huang XP, Kuznetsova E, et al. An orally bioavailable chemical Ready to submit your research ? Choose BMC and benefit from: fast, convenient online submission thorough peer review by experienced researchers in your field rapid publication on acceptance support for research data, including large and complex data types • gold Open Access which fosters wider collaboration and increased citations maximum visibility for your research: over 100M website views per year At BMC, research is always in progress. Learn more biomedcentral.com/submissions http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Epigenetics & Chromatin Springer Journals

H3K27me1 is essential for MMP-9-dependent H3N-terminal tail proteolysis during osteoclastogenesis

Free
15 pages

Loading next page...
 
/lp/springer_journal/h3k27me1-is-essential-for-mmp-9-dependent-h3n-terminal-tail-LLChzJb0j9
Publisher
Springer Journals
Copyright
Copyright © 2018 by The Author(s)
Subject
Life Sciences; Animal Genetics and Genomics; Human Genetics; Plant Genetics and Genomics; Cell Biology; Gene Expression; Gene Function
eISSN
1756-8935
D.O.I.
10.1186/s13072-018-0193-1
Publisher site
See Article on Publisher Site

Abstract

Background: MMP-9 plays a direct role in the activation of pro-osteoclastogenic genes by cleaving histone H3N- terminal tail (H3NT ) and altering chromatin architecture. Although H3 acetylation at K18 has been shown to stimulate MMP-9 enzymatic activity toward H3NT, nothing is known about the influence of other H3NT modifications on this epigenetic reaction. Results: We show that H3 monomethylation at lysine 27 (H3K27me1) is essential for MMP-9-dependent H3NT prote- olysis during RANKL-induced osteoclast differentiation. Through the recognition of H3K27me1 mark, MMP-9 localizes and generates H3NT proteolysis at the genes encoding osteoclast differentiation factors. By using RNAi and small molecule inhibitor approaches, we also confirmed that G9a is the major methyltransferase to catalyze H3K27me1 for MMP-9-dependent H3NT proteolysis and trigger the expression of osteoclast-specific genes. Conclusions: Our data establish new functions for G9a-mediated H3K27me1 in MMP-9-dependent H3NT proteolysis and demonstrate how histone modification can be exploited to regulate osteoclastogenic gene expression at the molecular level. Further studies are warranted to investigate the detailed mechanism by which G9a overexpression with concomitant dysregulation of osteoclastogenesis contributes to the pathogenesis of bone disorders. Keywords: MMP-9, G9a, H3 proteolysis, H3K27me1, Osteoclast differentiation Background receptor RANK on pre-osteoclast cell membrane stimu- Bone is a highly dynamic organ that is continuously lates the expression of key determinants of osteoclast remodeled by coordinated activities of two cell types, differentiation such as NF-κB, c-Fos, and NFATc1 at the osteoclasts and osteoblasts [1, 2]. Osteoblasts are mono- early stage of the process [1, 8]. These factors then initiate nucleated mesenchymal stem cells that form bone matrix, multiple signal transduction pathways to turn on down- whereas osteoclasts are bone-resorbing multinucleated stream genes and activate quiescent osteoclast precur- cells that differentiate from hematopoietic progenitors sor (OCP) cells to become mature osteoclasts [1, 8]. The of the myeloid lineage [3–5]. Osteoclast differentiation is excessive formation and activity of osteoclasts lead to induced by receptor activator of NF-κB ligand (RANKL), pathological bone diseases such as osteoporosis, rheuma- which is expressed as a membrane-bound protein in toid arthritis, and tumor bone metastases [9, 10]. osteoblasts and provides osteoclast-specific differentia - As for other eukaryotic genes, osteoclastogenic tion signals [6, 7]. The binding of RANKL to its cognate gene expression occurs in the context of chromatin, where DNA is wound around histone proteins to form chromatin structure [11, 12]. An essential step for *Correspondence: woojinan@usc.edu understanding gene regulatory pathways at key dif- Kyunghwan Kim and Yonghwan Shin contributed equally to this work ferentiation time points,  therefore, should  lie in  char- Department of Biochemistry and Molecular Medicine, Norris acterizing the enzymes responsible for reorganizing Comprehensive Cancer Center, University of Southern California, Los Angeles, CA 90089, USA and potentiating particular chromatin domains. Not Full list of author information is available at the end of the article © The Author(s) 2018. This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creat iveco mmons .org/licen ses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creat iveco mmons .org/ publi cdoma in/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Kim et al. Epigenetics & Chromatin (2018) 11:23 Page 2 of 15 much progress has been made in supporting this idea, osteoclastogenic gene expression and MMP-9 transacti- but there is some indirect evidence functionally link- vation potential in OCP-induced cells. ing chromatin modification to osteoclastogenic gene In this study, we demonstrate that H3K27 monometh- transcription. For example, the repressive histone mark ylation (H3K27me1) is necessary for the localization H3K27me3 is removed from the master osteoclasto- and function of MMP-9 at osteoclastogenic genes as a genic gene NFATc1 [13], while it is deposited into the protease catalyzing H3NT proteolysis. The observed anti-osteoclastogenic gene IRF8 [14]. Histone deacety- H3K27me1 is attributed to G9a and stabilizes the inter- lase inhibitors have antagonistic effects on osteoclast action of MMP-9 with nucleosomes. Supporting these differentiation through the inactivation of NF-κB sign - data, selective inhibition of G9a-mediated H3K27me1 aling pathways [6, 15], suggesting a potential role for resulted in complete abrogation of RANKL-induced histone acetylation in osteoclastogenic gene expression. osteoclast formation and gene transcription. Therefore, Nonetheless, which factors are mainly responsible for G9a is a previously unrecognized regulator of osteoclast establishing transcriptionally competent chromatin differentiation with a unique function relevant to MMP- states and exactly how altered chromatin states trig- 9-dependent H3NT proteolysis. ger osteoclastogenic gene transcription remain poorly understood. Another large gap in our understanding of Results osteoclastogenic transcription program is the identifi - H3K27me1 is necessary for MMP‑9‑dependent H3NT cation of chromatin factors that have the potential to proteolysis of nucleosome substrates link specific aspects of chromatin function to osteoclast We have recently demonstrated that p300/CBP-medi- differentiation processes. ated H3K18ac neutralizes the charged lysine residue at Matrix metalloproteinases (MMPs) are a large family the primary cleavage site (P1) of MMP-9 and amplifies of extracellular enzymes, which function to remodel the MMP-9 enzymatic activity toward both free and nucleo- pericellular environment, primarily through the cleav- somal H3 substrates [20]. Since H3NTs are also subject age of extracellular matrix proteins [16–18]. MMP-9, to other types of modifications such as methylation and of special interest here, is a member of the MMP family phosphorylation, investigating their possible effects on and contains several conserved domains such as propep- MMP-9-dependent H3NT proteolysis should be a logi- 2+ tide domain, catalytic domain with Z n -binding site, cal extension of our study (Fig.  1a). Toward this end, we and hemopexin-like domain [17]. Like other MMP fam- prepared recombinant histone octamers containing H3 ily members, MMP-9 is first synthesized as an inactive/ analogs that are mono-, di-, or tri-methylated (me1, me2, latent 92-kDa proenzyme and subsequently converted to or me3) at K4, K9, K27, or K36 and phosphorylated (p) at a fully active 82-kDa form by removal of the N-terminal T3, S10, or S28. These histone octamers were then used propeptide domain [19]. Regarding MMP-9 functions, to reconstitute differentially methylated or phosphoryl - MMP-9 has been characterized as a major endopepti- ated nucleosome arrays on a tandem DNA array contain- dase with an ability to degrade extracellular matrix and ing seven copies of a 207-bp 601 nucleosome positioning stimulate osteoclastogenesis [19]. Unexpectedly, how- sequence. In our initial H3NT proteolysis assays with his- ever, our recent study revealed the nuclear function of tone octamers containing methylated H3 as substrates, MMP-9 as a protease that cleaves the histone H3N-ter- an H3 C-terminal antibody detected a faster-migrating minal tail (H3NT) during osteoclast differentiation [20]. band representing H3NT-cleaved product in all reactions In strong support of these observations, our subcellular (Fig.  1c, Oct). It was apparent in these experiments that localization analysis by Western blotting, gelatin zymog- cleavage levels of methylated H3 were comparable with raphy, and immunofluorescence clearly demonstrated the those of unmodified H3. Similarly, phosphorylation of H3 nuclear translocation and accumulation of MMP-9 dur- at T3, S10, or S28 showed no effects on MMP-9-depend - ing RANKL-induced formation of mature osteoclasts. ent H3NT proteolysis in identical assays (Fig.  1b, Oct). A functional role for the observed H3NT proteolysis in When we extended in  vitro cleavage assays to reconsti- osteoclastogenic gene transcription was evident from our tuted nucleosome arrays, MMP-9 was unable to prote- analysis of MMP-9-depleted OCP cells that had unde- olyze H3NT in nucleosome array substrates carrying tectable levels of H3NT proteolysis and OCP cell dif- H3K4me1/me2/me3, H3K9me1/me2/me3, H3K27me2/ ferentiation [20]. Our report also showed that MMP-9 me3, or H3K36me1/me2/me3 (Fig.  1c, Nuc). H3T3p, enzymatic activity toward H3NT is significantly aug - H3S10p, and H3S28p also had no effect on MMP-9-de - mented by H3K18 acetylation (H3K18ac) and that p300/ pendent H3NT proteolysis in our assays (Fig. 1b, Nuc). In CBP is responsible for H3K18ac observed in OCP cells sharp contrast, however, MMP-9 generated a reproduc- [20]. This is an important observation, meaning that ible H3NT proteolysis within H3K27me1 nucleosome p300/CBP-mediated H3K18ac has a functional role in arrays (Fig. 1c, Nuc). The observed effects are specific for Kim et al. Epigenetics & Chromatin (2018) 11:23 Page 3 of 15 Fig. 1 Requirement of H3K27me1 for MMP-9-dependent H3NT proteolysis. a Amino acid sequence of N-terminal tail (NT ) of histone H3. Amino acids that can be methylated (me) or phosphorylated (p) are indicated. b In vitro H3NT cleavage assays were performed using recombinant histone octamer or reconstituted nucleosome array substrates unmodified or phosphorylated at H3T3p, H3S10p, or H3S28p. The extent of H3NT proteolysis was analyzed by Western blot with H3 C-terminal tail (CT ) antibody. c As for b but using free histone octamers or nucleosome arrays methylated at H3K4, H3K9, H3K27, or H3K36 nucleosome substrates, since H3K27me1 did not show a osteoclasts in α-minimal essential medium (α-MEM) corresponding effect on free H3NT cleavage by MMP-9 following RANKL treatment for 0, 1, 3, or 5  days. In in our assays (Fig. 1c, Oct). Based on these data, we con- agreement with our published data, we detected a fast- cluded that H3K27me1 is required for MMP-9 to cleave migrating H3 band representing H3NT proteolysis and H3NT in a nucleosome context. an increase in MMP-9 expression after RANKL treat- ment (Fig.  2a). The loss of H3NT proteolysis following G9a catalyzes H3K27me1 and facilitates H3NT proteolysis MMP-9 knockdown is also consistent with our previ- during osteoclastogenesis ous demonstration of MMP-9-dependent H3NT prote- We next wanted to examine whether H3K27me1 is simi- olysis during osteoclast differentiation (Fig.  2b) [20]. In larly required for H3NT proteolysis during osteoclas- parallel experiments in which changes in H3K27me1 togenesis and, if so, which histone methyltransferase were analyzed over the same time period, a progressive (HMT) is responsible for H3K27me1. In our assay sys- increase in overall H3K27me1 levels was also evident tem, osteoclast precursor (OCP) cells are synchro- (Fig.  2a, b), suggesting its contribution to H3NT cleav- nously differentiated into TRAP-positive multinuclear age process. To confirm the significance of H3K27me1 (See figure on next page.) Fig. 2 Dependence of osteoclastogenic H3NT proteolysis and H3K27me1 on G9a. a Chromatin was extracted from primary OCP cells treated with RANKL for 0, 1, 3, and 5 days and analyzed by Western blotting with H2A, H2B, H3, and H4 CT antibodies (left panel). OCP-induced cells were also fixed with formaldehyde, stained for TRAP (tartrate-resistant acid phosphatase), and photographed under a light microscope (10×) (middle panel). TRAP-positive cells containing three or more nuclei were counted as osteoclasts at the indicated days (right panel). b As for a but using OCP-induced cells depleted of MMP-9. c As for a but using OCP-induced cells depleted of G9a. d As for a but using OCP-induced cells depleted of both MMP-9 and G9a Kim et al. Epigenetics & Chromatin (2018) 11:23 Page 4 of 15 0 1 3 5 Time (days) 0 1 3 5 Time (days) 0 1 3 5 Time (days) 0 1 3 5 Time (days) TRAP(+)M NCs/we ll TRAP(+)M NCs/we ll TRAP(+)M NCs/we ll TRAP(+)M NCs/we ll Kim et al. Epigenetics & Chromatin (2018) 11:23 Page 5 of 15 with respect to H3NT proteolysis more directly, we also EZH2 would affect H3K27me1 and H3NT proteolysis by transfected OCP-induced cells with plasmids express- using two inhibitors, BIX01294 and UNC1999. Of note, ing FLAG wild-type or K27-mutated H3 and prepared BIX01294 has an inhibitory property for the methyltrans- mononucleosomes as summarized in Additional file  1: ferase activity of G9a [26], whereas UNC1999 inhibits Fig. S1. Mononucleosomes containing ectopic H3 were the enzymatic activities of both EZH1 and EZH2 [27]. In then purified by immunoprecipitations using anti-FLAG analyzing the effects of EZH1/2 inhibitor UNC1999, we antibody. Our examination of purified mononucleosomes detected lower levels of H3K27me2 and H3K27me3, but by Western blot detected a high level of H3NT cleavage failed to see any apparent changes in MMP-9 protease in wild-type H3 nucleosomes, but the observed proteoly- activity toward H3NT (Additional file  5: Fig. S5). Mean- sis was impaired in H3K27-mutated nucleosomes (Addi- while, treatment of OCP-induced cells with 1.5  µM G9a tional file  2: Fig. S2). These data indicate strongly that inhibitor BIX01294 led to a pronounced inhibition of H3NT proteolysis in osteoclast differentiation pathway is H3NT proteolysis and osteoclast development (Fig.  3c). dependent on H3K27me1. Consistent with our published data [20], a significant To further investigate the role of H3K27me1 in the for- decrease in H3NT proteolysis was also observed upon mation of mature osteoclasts, it was important to identify treatment with 10  nM MMP-9 inhibitor I. Moreover, a the HMT mediating the observed H3K27me1. For this failure to generate higher levels of H3NT proteolysis and objective, we suppressed the expression of three HMTs, osteoclastogenesis upon treatment with both G9a inhibi- EZH1, EZH2, and G9a, that were shown to catalyze tor BIX01294 and MMP-9 inhibitor I, compared to the H3K27me1 [21–25] (Fig.  2c and Additional file  3: Fig. levels generated by their individual treatment, confirmed S3). When OCP cells were depleted of EZH1 or EZH2, again the dual requirement of G9a and MMP-9 for signal no obvious changes in H3K27me1 and H3NT proteolysis transduction pathways operating in osteoclast differen - were detected in our Western blot analysis of chromatin tiation (Fig. 3b, d). fraction from RANKL-induced OCP cells (Additional file  4: Fig. S4). Because knockdown of EZH1 and EZH2 G9a‑mediated H3K27me1 is crucial for MMP‑9 recruitment reduced levels of H3K27me2 and H3K27me3 (Additional and function at target genes file  4: Fig. S4), these results also indicate the indispen- Having established the requirement of G9a-mediated sable role of H3K27me1 in osteoclastogenic H3NT pro- H3K27me1 for MMP-9-dependent H3NT proteolysis teolysis. On the contrary, specific knockdown of G9a in and proficient osteoclast differentiation, we next sought OCP-induced cells efficiently blocked H3K27me1 and to explore whether G9a is also directly involved in the almost completely abrogated H3NT proteolysis in our expression of MMP-9 target genes. We recently devel- assays (Fig.  2c). Since G9a knockdown also decreased oped a technique called ChIP of acetylated chromatin the average number of mature osteoclasts (Fig.  2c), (ChIPac) (schematized in Additional file  6: Fig. S6) and these results confirm the functional contribution of demonstrated that H3NT proteolysis is associated with G9a to osteoclast formation. Considering the possibility RANKL-induced activation of genes necessary for osteo- that G9a could promote osteoclast differentiation inde - clast differentiation [20]. In this new method, we made pendently of MMP-9, we also checked whether double use of methylene blue to cross-link chromatin and ace- knockdown of G9a and MMP-9 exhibits more severe tic anhydride to completely acetylate all lysine residues in defects in osteoclastogenesis. Interestingly, however, the fragmented chromatin. H3K14ac-specific antibody was simultaneous knockdown of G9a and MMP-9 attenuated then used to selectively precipitate intact H3NT-contain- osteoclast differentiation in similar level as that observed ing chromatin, and a reduced PCR or sequencing signal in individual knockdown of G9a and MMP-9 (Fig.  2d). intensity relative to the control ChIPac reactions using an These data point to the dependence of pro-osteoclasto - H3CT antibody is indicative of osteoclastogenic H3NT genic function of MMP-9 on G9a-mediated H3K27me1 proteolysis. During the process of osteoclast formation, and constitute a powerful argument that both MMP-9 MMP-9 was shown to generate H3NT cleavage in pro- and G9a are essential for efficient osteoclastogenesis. moter, coding region, or both for target gene transcrip- In an attempt to support our knockdown data, we also tion in a gene-specific manner [20]. Thus, we examined tested whether chemical inhibition of G9a, EZH1, and the localization of G9a at Nfatc1, Lif, and Xpr1 genes (See figure on next page.) Fig. 3 Abolishment of osteoclastogenic H3NT proteolysis and H3K27me1 by G9a inhibitor. OCP cells were treated with DMSO control (a), MMP-9 inhibitor (b), G9a inhibitor (c), or MMP-9 + G9a inhibitors (d) and cultured for 0, 1, 3, and 5 days with RANKL. Chromatin was isolated for Western blot analysis to assess H3NT proteolysis (left panel). Cells were also TRAP-stained middle panel) and counted (right panel) as in Fig. 2 Kim et al. Epigenetics & Chromatin (2018) 11:23 Page 6 of 15 0 1 3 5 Time (days) 0 1 3 5 Time (days) 0 1 3 5 Time (days) 0 1 3 5 Time (days) TRAP(+)M NCs/we ll TRAP(+)M NCs/we ll TRAP(+)M NCs/we ll TRAP(+)M NCs/we ll Kim et al. Epigenetics & Chromatin (2018) 11:23 Page 7 of 15 representing each H3NT-cleaved group in OCP-induced accumulation of MMP-9 in the P, CR, and both regions cells by ChIPac-qPCR, as described recently [20]. Two of Nfatc1, Lif, and Xpr1 genes, respectively (Fig.  4a). sets of primers were used to detect MMP-9, G9a, and The localization patterns of MMP-9 were similar to H3K27me1 in the promoter (P) and coding region (CR) those observed for H3K27me1 in these target genes, by qPCR. implicating H3K27me1 as the major recruitment sig- Consistent with our recently published data [20], nal for MMP-9. H3K27me1 levels were reduced at the RANKL treatment of OCP cells resulted in a rapid target genes after G9a knockdown, and such changes Fig. 4 Impaired H3NT proteolysis and osteoclastogenic gene transcription in G9a-depleted OCP cells. a Control, MMP-9-, G9a-, or MMP-9 + G9a-depleted OCP cells were cultured with RANKL to induce osteoclastogenesis for 3 days, and ChIPac was performed using H3K14ac (H3NT ), H3CT, and H3K27me1 antibodies. H3NT cleavage levels were determined at the promoter (P) and coding regions (CR) of Nfatc1 (P-cleaved), Lif (CR-cleaved), and Xpr1 (P + CR-cleaved) genes by qPCR with primers used in our previous study [20] and are listed in “Methods” section. b RT-qPCR assays were performed to determine fold changes in Nfatc1, Lif, and Xpr1 expression in 3-day RANKL-induced OCP cells depleted of MMP-9 and/or G9a Kim et al. Epigenetics & Chromatin (2018) 11:23 Page 8 of 15 diminished the measured levels of P and CR occupied by G9a localization nor H3K27me1 at the target genes was MMP-9 (Fig. 4a). No detectable effects of MMP-9 knock - affected upon treatment with MMP-9 inhibitor (Fig. 5a). down on G9a occupancy at the target genes indicate that Given the demonstrated reliance of MMP-9 on G9a- MMP-9 is dispensable for G9a recruitment and function. mediated H3K27me1 for its target gene localization, we To further confirm the results, the ChIPac-qPCR assays also examined the role of G9a with respect to RANKL- were repeated using OCP-induced cells treated with induced expression of osteoclast-specific genes. Our MMP-9 and G9a inhibitors. As was observed with G9a RT-qPCR analysis showed that G9a depletion in OCP- knockdown, OCP-induced cells treated with G9a inhibi- induced cells caused three- to sevenfold decreases in tor show significantly lower levels of H3K27me1 and mRNA levels of Nfatc1, Lif, and Xpr1 genes (Fig.  4b). MMP-9 at the target genes (Fig.  5a). Expectedly, neither Congruent with these data, treatment of OCP cells with Fig. 5 Impaired H3NT proteolysis and osteoclastogenic gene transcription in G9a inhibitor-treated OCP cells. a ChIPac-qPCR assays were conducted to assess H3NT proteolysis levels in 3-day OCP-induced cells treated with MMP-9 and/or G9a inhibitors. b RT-qPCR to quantitate Nfatc1, Lif, and Xpr1 transcript levels in 3-day OCP-induced cells treated with MMP-9 and/or G9a inhibitors Kim et al. Epigenetics & Chromatin (2018) 11:23 Page 9 of 15 G9a inhibitor also led to impaired expression of the tar- MMP-9 retained strong affinity for H3K27me1 nucleo - get genes upon RANKL-induced differentiation (Fig.  5b). some, whereas no apparent interaction was observed In addition, when OCP cells were depleted of both G9a with the remainder (amino acids 448–730) of the protein and MMP-9 or treated with G9a and MMP-9 inhibitors (Fig.  6a, d). Also, in similar binding experiments using together, target gene transcription was repressed, but the three N-terminal subregions, MMP-9 amino acids 384– level of repression was similar to that detected in knock- 447 directly interacted with H3K27me1 nucleosome, down or inhibition of G9a or MMP-9, again pointing but MMP-9 amino acids 112–212 and 213–383 failed to to G9a-mediated H3K27me1 as an essential epigenetic show any detectable interaction under the same condi- mark for MMP-9 recruitment and function (Figs. 4b, 5b). tions (Fig. 6e). Together, these data support the direct function of G9a In order to gain more support for the in vitro binding in regulating MMP-9 target gene pathways necessary for results above, OCP cells were transfected with expres- proficient osteoclast differentiation. sion vectors for FLAG-H3 wild type or K27R mutant. After treating OCP cells with RANKL for 3  days, sol- MMP‑9 specifically binds to H3K27me1 nucleosomes uble chromatin was prepared from cell nuclei and The results of the above experiments argue persuasively digested with micrococcal nuclease to yield mainly that H3K27me1 is indispensable for the stable localiza- mononucleosomes. Mononucleosomes containing tion and function of MMP-9 at target genes. However, it ectopic H3 wild type or K27R mutant were then spe- is not clear whether the observed effects of H3K27me1 cifically immunoprecipitated with anti-FLAG anti - reflect its role as a docking site to facilitate the recruit - body. When we analyzed the association of endogenous ment of MMP-9 to target genes. To check this possibility, MMP-9 with these purified nucleosomes, we found that H3NT unmodified or K27me1 peptides corresponding to MMP-9 bound avidly to wild-type H3 nucleosomes, but amino acids 1–21 or 21–44 were immobilized on strepta- bound minimally to K27R-mutated H3 nucleosomes vidin-coated wells and monitored the binding for MMP- (Fig.  6f ). To identify amino acid residues critical for 9. Our results showed that MMP-9 strongly binds to the MMP-9-H3NTK27me1 interaction, we next gener- H3NT K27me1 peptides, whereas other H3NT peptides ated a structural model of the MMP-9-H3NTK27me1 displayed only weak interaction with MMP-9 (Fig.  6b). complex based on existing crystal structures using To further confirm these results, we reconstituted nucle - the program Cluspro 2.0 [28–30] (Fig.  6g). The model osomes containing H3 unmodified or K27me1 on a identified E402 of MMP-9 to contact anionic residues 2+ 601 nucleosome positioning sequence and checked the in H3NTK27me1. Moreover, the Z n -binding site of binding of MMP-9. In agreement with the interactions MMP-9 was suggested to interact with H3NTK27me1. observed with H3NT peptides, we detected a remark- Consistent with this model, in vitro binding assays test- able binding preference of MMP-9 for the immobilized ing the E402A substitution showed a diminished bind- nucleosome containing H3K27me1 over the nucleo- ing of MMP-9 to H3NTK27me1, supporting a role some containing H3 unmodified (Fig.  6c). Additionally, for electrostatic contacts involving this residue in the in mapping the interaction region of mature MMP-9, we interaction (Fig.  6h). In contrast, the H411A substitu- found that N-terminal region (amino acids 112–447) of tion was inconspicuous, which leaves the role of the (See figure on next page.) Fig. 6 MMP-9 binding to H3K27me1 nucleosomes. a Schematic depiction of the domain structure of MMP-9. b Peptide pull-down assays with biotinylated H3 1–21 and 21–44 peptides and recombinant His-MMP-9 were analyzed by Western blotting with anti-His antibody. H3 peptides were unmodified, K18ac or K27me1 as indicated. Lane 1 represents 10% of the input MMP-9. c Nucleosomes were reconstituted on a 207-bp 601 nucleosome positioning sequence using unmodified or H3K27me1 histone octamers and immobilized on streptavidin beads. His-MMP-9 was incubated with immobilized nucleosomes, and its binding to nucleosomes was analyzed by Western blotting with anti-His antibody. Lane 1 contains 10% of the input MMP-9. d H3K27me1 nucleosomes were incubated with immobilized MMP-9 N-terminal (amino acids 112–447) and C-terminal (amino acids 448–730) domains. After extensive washing, the binding of H3K27me1 nucleosomes to MMP-9 domains was determined by Western blotting with anti-H3 antibody. Input corresponds to 10% of H3K27me1 nucleosomes used in the binding reactions. e After incubation with H3K27me1 nucleosomes, the binding of MMP-9N-terminal subregions to nucleosomes was determined by Western blotting with anti-His antibody. Input lanes 1–3 represent 10% of MMP-9 fragments used in the binding reactions. f OCP-induced cells were transfected with FLAG-H3 wild type ( WT ) or K27R mutant (K27R), and mononucleosomes were prepared by micrococcal nuclease digestion as summarized in Figure S3. Mononucleosomes containing ectopic H3 were immunoprecipitated from total mononucleosomes with FLAG antibody and analyzed by Western blotting with anti-MMP-9 antibody. g Model of the MMP-9-H3K27me1 interaction. PDB entries 4h3x (mMMP-9) and 3avr (H3.1) were used in docking simulations using the program Cluspro 2.0 [28–30]. Simulations were run with non-methylated H3. For context, H3K27 is shown monomethylated. h Nucleosome binding assays were conducted as in e, except that His-MMP-9 amino acids 384–447 carrying E402A mutation were used Kim et al. Epigenetics & Chromatin (2018) 11:23 Page 10 of 15 2+ levels. Together, these results constitute a powerful Zn -binding site ambiguous (Fig.  6h). As judged by argument that, although other factors and signals may crystal structures of methyltransferases and demethy- be involved, the initial recruitment of MMP-9 at tar- lases [28, 30–32], the differentiation of H3K27 meth - get genes is dependent on the recognition of H3NT- ylation states by MMP-9 likely arises from steric and K27me1 through the N-terminal domain of MMP-9. electrostatic differences between different methylation Kim et al. Epigenetics & Chromatin (2018) 11:23 Page 11 of 15 In the present study, we focused on potential roles of H3 methylation and phosphorylation in MMP-9-de- pendent H3NT proteolysis. Unexpectedly, our in  vitro cleavage assays with nucleosome arrays containing dif- ferentially modified H3 revealed that H3K27me1 is nec - essary for H3NT proteolytic activity of MMP-9. Our observation that MMP-9 can efficiently cleave free H3 substrates regardless of their H3K27me1 states is sup- portive of the idea that H3K27me1 specifically enhances MMP-9 enzymatic activity toward H3NT in a nucleo- some context. We extended these in vitro findings in sub - sequent cellular studies revealing that MMP-9 protease activity toward H3NT is dependent on G9a-mediated H3K27me1 during RANKL-induced osteoclast forma- tion. Notably, data presented here also indicate that G9a levels are highly elevated in response to RANKL treat- ment, leading to a sharp increase in H3K27me1. More- over, the level of G9a-mediated H3K27me1 correlates Fig. 7 Working model. Our previous study revealed that MMP-9 directly with the extent of MMP-9-dependent H3NT catalyzes H3NT proteolysis for osteoclastogenic gene activation proteolysis and osteoclastogenesis. This finding under - and that CBP/p300-mediated H3K18ac facilitates MMP-9 protease scores the importance of H3K27me1 in accurate target- activity. Here, we depict that G9a-mediated H3K27me1 is necessary ing of MMP-9 within defined chromatin regions, and for MMP-9-dependent H3NT proteolysis in chromatin and proficient osteoclast differentiation. Without G9a-mediated H3K27me1, MMP-9 taken together with our recently described H3K18ac- cannot localize at osteoclastogenic genes and mediate H3NT augmented MMP-9 clipping activity, it also suggests that proteolysis. In the presence of G9a-mediated H3K27me1, MMP-9 these epigenetic alterations in H3NT directly influence gets recruited to target loci, facilitating H3NT proteolysis, leading MMP-9-dependent expression of pro-osteoclastogenic to RANKL-induced gene transcription, and promoting osteoclast genes (Fig. 7). differentiation Even though EZH1 and EZH2 were shown to medi- ate H3K27me1 in  vitro assays and in certain cell types [21–24], our knockdown data indicate that they do not Discussion participate in generating H3K27me1 and that G9a is There has been a growing interest in understanding how mainly responsible for catalyzing this osteoclastogenic epigenetic processes regulate gene transcription at vari- histone mark. Another argument in favor of G9a to act ous stages of osteoclast differentiation, activation, and as the major H3K27me1 methyltransferase comes from survival. MMP-9 is highly expressed in OCP cells and is the experiments showing that treating OCP cells with known to play a key role in RANKL-induced osteoclast G9a inhibitors, but not with EZH1/EZH2 inhibitors, differentiation. The general view of MMP-9 function in leads to near complete loss of H3K27me1 and H3NT osteoclastogenesis is that MMP-9 digests structural com- proteolysis. This finding is consistent with a recent report ponents of extracellular matrix and cellular surface facili- that G9a inhibitor BIX01294 reduced RANKL-induced tating cell migration and adhesion [16]. However, this osteoclast formation from RAW 264.7 cells [33]. In this common idea has been changed recently by our report study, H3K9me was assumed to play a causal role for G9a documenting that, beside regulating extracellular matrix stimulation of osteoclast differentiation, and BIX01294 remodeling, MMP-9 moves into the nucleus and gener- treatment was considered to have inhibitory effects on ates active transcription states of osteoclastogenic genes H3K9me process. However, our data provide strong by proteolytically cleaving H3NT at promoter and cod- evidence for the significance of H3K27me1, rather than ing regions [20]. Also, our study provided the first evi - H3K9me, for the osteoclastogenic function of G9a. Also, dence that p300/CBP-mediated H3K18ac is required for G9a inhibitor-enhanced osteoclastogenesis intrinsically MMP-9 to mediate H3NT proteolysis and active expres- depends upon MMP-9 function as an H3NT protease, sion of genes encoding positive regulators of osteoclas- since knockdown of G9a in MMP-9-depleted OCP cells togenesis [20]. Since other histone modifications can also failed to generate more attenuation of osteoclastogenesis. affect gene expression, one obvious gap in our under - In this regard, H3K27me1 should be recognized as an standing of MMP-9-regulated transcription mechanism epigenetic mark reflecting the initiation and progression lies in the identification of additional histone marks that of osteoclastogenic process. may be involved in MMP-9 activity toward H3NT. Kim et al. Epigenetics & Chromatin (2018) 11:23 Page 12 of 15 Further analysis of G9a-depleted OCP cells showed activity could provide an effective treatment for bone loss that G9a has a direct impact on the expression of osteo- diseases. clastogenic genes and revealed an intriguing new tran- scription pathway regulating osteoclast differentiation. Conclusions After identifying osteoclast gene transcription being Results presented here describe a role for G9a in MMP- up-regulated by G9a, we also realized that these genes 9-dependent H3NT proteolysis and gene transcription were enriched by H3K27me1, MMP-9, and cleaved by catalyzing H3K27me1 during RANKL-induced osteo- H3 species in a G9a-dependent manner. These results clast differentiation. demonstrate the direct action of G9a on transcriptional These results establish a direct functional link between program required for osteoclastogenesis and support G9a and MMP-9 in the context of pro-osteoclastogenic the concept that G9a exerts a coactivation function in transcriptional programs. Our model suggests that G9a- MMP-9-driven transactivation. Our ChIPac analyses mediated H3K27me1 serves as an essential mark for showed that G9a generates H3K27me1 in promoter MMP-9 recruitment and proteolytic activity at genes or coding regions or both regions of osteoclastogenic encoding positive regulators of osteoclast differentiation, genes and that H3NT proteolysis patterns match well thus keeping them active (Fig. 7). with H3K27me1 states. However, high levels of tran- scription were accomplished in all these cases. This Methods observation suggests a function for G9a-mediated Plasmid construction and materials H3K27me1 at the initiation or elongation step of tran- Core histones and MMP-9 proteins were expressed in scription in a gene-specific manner. It is also con - Escherichia coli Rosetta 2 (DE3) pLysS cells (Novagen) ceivable that the levels of H3NT proteolysis are not and purified from inclusion bodies as described recently proportional to transcription rates and H3NT prote- [20]. To generate mutant H3 and MMP-9 expression olysis targeted to certain regions is sufficient to induce vectors, H3 and MMP-9 cDNAs were mutated by the gene transcription. QuikChange II site-directed mutagenesis kit (Agilent Toward understanding how H3K27me1 can facilitate Technologies) before the construction. Further details MMP-9-dependent H3NT proteolysis, we demonstrate of plasmid constructions are available upon request. that H3K27me1 is essential for MMP-9 binding to nucle- G9a inhibitor BIX01294 is from Santa Cruz Biotech, and osomal H3NTs. Only when H3K27me1 was introduced EZH1/2 inhibitor UNC1999 and MMP-9 Inhibitor I are into nucleosomes did it contribute to MMP-9 proteolytic from Sigma. Antibodies used in this study are as follows: activity. Thus, MMP-9 appears to utilize some specific H2A, H2B, H3, H4, and EZH2 antibodies from Abcam; structural features to recognize H3K27me1 nucleosomes H3K27me1 and EZH1 antibodies from Millipore; G9a, in osteoclastogenic target genes (Fig. 7). Reciprocally, the actin, and FLAG antibodies from Sigma; His antibody results also suggest that H3K27me1 may serve principally from Novagen; and MMP-9 antibody from Santa Cruz as a mark for MMP-9 recruitment to target genes, with Biotech. no additional role in MMP-9-dependent H3NT prote- olysis. These observations are reminiscent of cathepsin In vitro H3NT cleavage assays L, of which capability to catalyze H3NT proteolysis was Recombinant histone octamers and nucleosome arrays enhanced by H3K27me2 [34]. Nonetheless, no detect- containing unmodified, methylated, or phosphorylated able effects of H3K27me2 in our assays suggest the pres - H3 were prepared following the procedure described [20, ence of two different regulatory mechanisms involving 36]. MMP-9 was incubated with 1 µg of histone octamer H3K27me1 for MMP-9 and H3K27me2 for cathepsin L. or 2  µg of nucleosome arrays, and H3NT cleavage was u Th s, in addition to its known contribution to gene tran - determined by Western blotting with H3 C-terminal scription [21, 35], our results underscore the significance antibody [20]. of H3K27me1 in helping MMP-9 target specific genes to mediate H3NT proteolysis. Defining the molecular basis Osteoclast differentiation and H3NT cleavage analysis for H3K27me1 effects on MMP-9 activity is beyond the Osteoclast precursor (OCP) cells were prepared as scope of this first report but will be of interest for how recently described [20]. To generate osteoclasts, OCP epigenetic signals may alter intrinsic MMP-9 proper- cells were cultured in the presence of 30  ng/ml mac- ties in OCP cells in response to RANKL stimulation. rophage colony-stimulating factor (M-CSF) and 50  ng/ Based on the functional connection uncovered between ml receptor activator of nuclear factor kappaB ligand G9a-mediated H3K27me1 and MMP-9-dependent (RANKL). On days 0, 1, 3, and 5, the cells were fixed H3NT proteolysis, we speculate that inventing strate- with formaldehyde and stained for tartrate-resistant gies to block  osteoclastogenic G9a methyltransferase acid phosphatase (TRAP) using an acid phosphatase Kim et al. Epigenetics & Chromatin (2018) 11:23 Page 13 of 15 leukocyte kit (Sigma). TRAP-positive multinucleated 5′-GGT GGG TTC CAC TGA AAG AA-3′, 5′-GGT TCC cells containing three or more nuclei were counted TCT GAC CAA AAG CA-3′; mRNA: 5′-AGG AGC GTG as osteoclasts under a light microscope. In certain TCC AAC ATA GG-3′, 5′-CCA CGA GAT GTT TCC AGG instances, media were supplemented with G9a inhibi- AT-3′). tor BIX01294 (1.5  µM), EZH1/2 inhibitor UNC1999 (2 µM), and MMP-9 inhibitor I (10 nM) to evaluate their H3 tail peptide and nucleosome binding assays effects on OCP cell differentiation. To determine the For H3NT peptide binding assays, biotinylated forms of levels of H3NT proteolysis, nuclei were isolated from H3NT peptides unmodified, acetylated at K18, or mono - OCP-induced cells in buffer A (10  mM HEPES, pH 7.4, methylated at K27 (EZBiolab Inc) (2  μg) were immobi- 10  mM KCl, 1.5  mM MgCl2, 0.34  M sucrose, 10% glyc- lized on streptavidin–agarose beads. After washing with erol, 1 mM DTT, 5 mM β-glycerophosphate, 10 mM NaF, BC250/0.1% Nonidet P-40, His-MMP-9 was incubated protease inhibitors, and 0.2% Triton X-100) and chro- with H3NT peptides-bound beads in BC200/0.1% Noni- matin was extracted in buffer B (3  mM EDTA, 0.2  mM det P-40 for 3  h at room temperature. After extensive EGTA, 1  mM DTT, 5  mM β-glycerophosphate, 10  mM washing with BC200/0.1% NP-40, MMP-9 interaction NaF, and protease inhibitors). Western blot analysis was was analyzed by Western blotting with anti-His antibody. performed using H3 C-terminal antibody as previously For nucleosome binding assay, H3 unmodified/H3K18ac/ described [20]. H3K27me1 nucleosomes were reconstituted by mixing recombinant histone octamers and biotinylated 207- RNA interference, RT‑qPCR, and ChIPac‑qPCR bp 601 nucleosome positioning sequence templates at a Lentiviral particles were generated in HEK-293T cells ratio of 1:1.2 (w/w) and salt gradient dialysis and purified by co-transfecting plasmids encoding VSV-G, NL-BH, by sedimentation in a 5–30% (vol/vol) glycerol gradient and pLKO.1-shRNA (Addgene) for MMP-9 (5′-GAG as described previously [36]. Nucleosomes (1  μg) were GCA TAC TTG TAC CGC TAT-3) or G9a (5′-AGA CAT immobilized on streptavidin–agarose beads (Novagen) TTC TCC ATC AGA GAC-3′). OCP cells were transduced and incubated with His-MMP-9 proteins for 16 h on ice. with these viruses or 10  nM MMP-9-INI (Santa Cruz) After washing with BC250/0.1% Nonidet P-40, bound for 3  days prior to differentiation. Total RNA was iso - MMP-9 proteins were detected by Western blotting. lated from OCP-induced cells using the Qiagen RNeasy kit (Qiagen, Valencia, CA) and reverse-transcribed using the iScript cDNA synthesis kit (Bio-Rad) and PerfeCta Nucleosome purification and analysis SYBR Green FastMix (Quanta Biosciences). ChIPac- OCP cells were transfected with expression vectors for qPCR assays were performed using chromatin that was H3 wild type or K27R mutant containing a C-terminal fixed with 10  µM methylene blue and acetylated with FLAG tag. After 3-day RANKL treatment, cells were har- 20  mM acetic anhydride as described [20]. H3K14ac, vested and lysed with buffer A (20  mM HEPES, pH 7.4, H3CT, and H3K27me1 antibodies were used to immuno- 10  mM KCl, 1.5  mM MgCl2, 0.34  M sucrose, 10% glyc- precipitate cross-linked chromatin. The immunoprecipi - erol, 1  mM dithiothreitol, and protease inhibitor cock- tated protein–DNA complexes were recovered, washed, tail) containing 0.2% Triton X-100. Nuclei were pelleted and incubated overnight at 65  °C to reverse the cross- by centrifugation at 1000g, resuspended in buffer A con - linking. DNA fragments were purified and analyzed with taining 2  mM CaCl , and digested with 0.6 U micrococ- the primers that amplify the promoter (P) and coding cal nuclease (Sigma) at 37 °C for 20 min. Digested nuclei regions (CR) of Nfatc1 (P-cleaved), Lif (CR-cleaved), and were collected and incubated in nuclear extraction buffer Xpr1 (P + CR-cleaved) genes. The sequences of primers (20  mM HEPES, pH 7.4, 420  mM NaCl, 1.5  mM M gCl , used for qPCR are as follows: Nfatc1 (P: 5′-GAA GTG 0.2  mM EGTA, and protease inhibitor cocktail) for 1  h GTA GCC CAC GTG AT-3′, 5′-TCT TGG CAC CAC ATA and centrifuged to remove nuclear debris. After adjust- AAC CA-3′; CR: 5′-GGG TCA GTG TGA CCG AAG AT-3′, ing the salt concentration of the extract to 150 mM NaCl, 5′-GGA AGT CAG AAG TGG GTG GA-3′; mRNA: 5′-CTC ectopic H3-containing nucleosomes were isolated by GAA AGA CAG CAC TGG AGCAT-3′, 5′-CGG CTG CCT immunoprecipitation using anti-FLAG M2 agarose beads TCC GTC TCA TAG-3′), Lif (P: 5′-CTC TGG CTG TCC in washing buffer (20 mM HEPES, pH 7.8, 300 mM NaCl, TGG AAC TC-3′, 5′-CCA GGA CCA GGT GAA ACA CT-3′; 1.5 mM MgCl , 0.2 mM EGTA, 10% glycerol, 0.2% Triton CR: 5′-ATC TTG TGG CTT TGC CAA CT-3′, 5′-AGT X-100, and protease inhibitor cocktail). Levels of H3NT CCT TGC CTG TCT TTC CA-3′; mRNA: 5′-TAC TGC proteolysis of bead-bound nucleosomes were analyzed by TGC TGG TTC TGC AC-3′, 5′-TGA GCT GTG CCA GTT Western blotting with anti-FLAG antibody. The purified GAT TC-3′), and Xpr1 (P: 5′-AGG ACC TTC GGA AGA nucleosomes were also subject to Western blotting with GCA GT-3′, 5′-CAG CAA GCA GCT CAT AAC CA-3′; CR: anti-MMP-9 antibody. Kim et al. Epigenetics & Chromatin (2018) 11:23 Page 14 of 15 Consent for publication Statistical analysis Not applicable. All quantitative data are presented as mean ± SD. Statis- tical analyses of datasets were performed with Student’s Ethics approval and consent to participate Not applicable. two-tailed t test or two-way ANOVA followed by Bonfer- roni’s comparison test. GraphPad Prism (GraphPad Soft- Funding ware Inc.) was used for all analyses. A P value < 0.05 was This work was supported by NIH Grant CA201561 awarded to W.A. The study was also funded by Pilot Project Grants from Keck School of Medicine of USC. considered statistically significant. This work was supported in part by the National Research Foundation of Korea (NRF) Grant 2017R1C1B2008017. Additional files Publisher’s Note Additional file 1: Fig. S1. Workflow of the purification method used for Springer Nature remains neutral with regard to jurisdictional claims in pub- isolation of ectopic H3 nucleosomes. lished maps and institutional affiliations. Additional file 2: Fig. S2. Abolishment of osteoclastogenic H3NT proteolysis by H3K27R mutation. Mononucleosomes containing ectopic Received: 6 February 2018 Accepted: 21 May 2018 H3 were purified from OCP-induced cells expressing H3 wild type or K27R mutant with C-terminal FLAG tag as summarized in Additional file 1: Fig. S1 and analyzed by Western blotting with anti-FLAG antibody. Additional file 3: Fig. S3. Validation of specific knockdown of EZH1 and References EZH2. OCP cells were transduced with lentiviral shRNAs targeting EZH1 (a) 1. Matsuo K, Irie N. Osteoclast-osteoblast communication. Arch Biochem and EZH2 (b), and knockdown efficiency and specificity were determined Biophys. 2008;473:201–9. by Western blot. 2. Nakahama K. Cellular communications in bone homeostasis and repair. Additional file 4: Fig. S4. Determination of the effects of EZH1/EZH2 Cell Mol Life Sci. 2010;67:4001–9. knockdown on osteoclastogenic H3NT proteolysis. a Chromatin was puri- 3. Ash P, Loutit JF, Townsend KMS. Osteoclasts derived from haematopoietic fied from mock-depleted, OCP-induced cells, and Western blot analysis stem cells. Nature. 1980;283:669. for H3NT proteolysis was performed as described in Fig. 2. b As for (a) but 4. Ikeda K, Takeshita S. Factors and mechanisms involved in the coupling using chromatin from EZH1-depleted, OCP-induced cells. c As for (a) but from bone resorption to formation: how osteoclasts talk to osteoblasts. J using chromatin from EZH2-depleted, OCP-induced cells. Bone Metab. 2014;21:163–7. 5. Raggatt LJ, Partridge NC. Cellular and molecular mechanisms of bone Additional file 5: Fig. S5. Determination of the effects of EZH1/EZH2 remodeling. J Biol Chem. 2010;285:25103–8. inhibitors on osteoclastogenic H3NT proteolysis. Chromatin was extracted 6. Nakashima T, Takayanagi H. New regulation mechanisms of osteoclast from OCP-induced cells after treating with an EZH1/EZH2 inhibitor and differentiation. Ann N Y Acad Sci. 2011;1240:E13–8. subject to Western blotting with H3CT antibody. 7. Yasuda H, Shima N, Nakagawa N, Yamaguchi K, Kinosaki M, Mochizuki S, Additional file 6: Fig. S6. Schematic representation of ChIPac-qPCR assay. Tomoyasu A, Yano K, Goto M, Murakami A, et al. Osteoclast differentia- tion factor is a ligand for osteoprotegerin/osteoclastogenesis-inhibitory factor and is identical to TRANCE/RANKL. Proc Natl Acad Sci USA. 1998;95:3597–602. Abbreviations 8. Teitelbaum SL, Ross FP. Genetic regulation of osteoclast development H3NT: histone H3N-terminal tail; H3K27me1: H3 monomethylation at lysine and function. Nat Rev Genet. 2003;4:638–49. 27; RANK: receptor activator of NF-κB ligand; MMPs: matrix metalloprotein- 9. Baron R. Arming the osteoclast. Nat Med. 2004;10:458–60. ases; OCP: osteoclast precursor; ChIPac: ChIP of acetylated chromatin; TRAP: 10. Ishida N, Hayashi K, Hoshijima M, Ogawa T, Koga S, Miyatake Y, Kumegawa tartrate-resistant acid phosphatase. M, Kimura T, Takeya T. Large scale gene expression analysis of osteo- clastogenesis in vitro and elucidation of NFAT2 as a key regulator. J Biol Authors’ contributions Chem. 2002;277:41147–56. KK and WA conceived and designed the study. TU offered advice on mutation 11. Tetsuro Y, Jun H, Hiroyuki A, Sakae T. Recent advance in epigenetics— experiment design. KK, YS, and JK performed the experiments with contribu- application to the regulation of osteoclast differentiation. Curr Rheumatol tions of TU and WA. YS and WA analyzed data and wrote the manuscript. All Rev. 2012;8:103–8. authors read and approved the final manuscript. 12. Vrtacnik P, Marc J, Ostanek B. Epigenetic mechanisms in bone. Clin Chem Lab Med. 2014;52:589–608. Author details 13. Yasui T, Hirose J, Tsutsumi S, Nakamura K, Aburatani H, Tanaka S. Epi- Department of Biochemistry and Molecular Medicine, Norris Comprehensive genetic regulation of osteoclast differentiation: possible involvement Cancer Center, University of Southern California, Los Angeles, CA 90089, USA. of Jmjd3 in the histone demethylation of Nfatc1. J Bone Miner Res. Department of Physiology and Neuroscience, Zilkha Neurogenetic Institute, 2011;26:2665–71. University of Southern California, Los Angeles, CA 90089, USA. Depart- 14. Fang C, Qiao Y, Mun SH, Lee MJ, Murata K, Bae S, Zhao B, Park-Min ment of Biology, College of Natural Sciences, Chungbuk National University, KH, Ivashkiv LB. Cutting edge: EZH2 promotes osteoclastogenesis Cheongju, Chungbuk 361-763, Republic of Korea. by epigenetic silencing of the negative regulator IRF8. J Immunol. 2016;196:4452–6. Acknowledgements 15. Cantley MD, Fairlie DP, Bartold PM, Rainsford KD, Le GT, Lucke AJ, Holding Not applicable. CA, Haynes DR. Inhibitors of histone deacetylases in class I and class II suppress human osteoclasts in vitro. J Cell Physiol. 2011;226:3233–41. Competing interests 16. Birkedal-Hansen H. Proteolytic remodeling of extracellular matrix. Curr The authors declare that they have no competing interests. Opin Cell Biol. 1995;7:728–35. 17. Papazafiropoulou A, Tentolouris N. Matrix metalloproteinases and cardio - Availability of data and materials vascular diseases. Hippokratia. 2009;13:76–82. All key data supporting the findings of this study are available within the 18. Vargova V, Pytliak M, Mechirova V. Matrix metalloproteinases. Exs. paper. Additional data and materials are available from the corresponding 2012;103:1–33. author upon request. Kim et al. Epigenetics & Chromatin (2018) 11:23 Page 15 of 15 19. Bergers G, Brekken R, McMahon G, Vu TH, Itoh T, Tamaki K, Tanzawa K, probe of the Lysine Methyltransferases EZH2 and EZH1. ACS Chem Biol. Thorpe P, Itohara S, Werb Z, Hanahan D. Matrix metalloproteinase-9 2013;8:1324–34. triggers the angiogenic switch during carcinogenesis. Nat Cell Biol. 28. Antoni C, Vera L, Devel L, Catalani MP, Czarny B, Cassar-Lajeunesse E, 2000;2:737–44. Nuti E, Rossello A, Dive V, Stura EA. Crystallization of bi-functional ligand 20. Kim K, Punj V, Kim JM, Lee S, Ulmer TS, Lu W, Rice JC, An W. MMP-9 facili- protein complexes. J Struct Biol. 2013;182:246–54. tates selective proteolysis of the histone H3 tail at genes necessary for 29. Kozakov D, Hall DR, Xia B, Porter KA, Padhorny D, Yueh C, Beglov D, proficient osteoclastogenesis. Genes Dev. 2016;30:208–19. Vajda S. The ClusPro web server for protein–protein docking. Nat Protoc. 21. Ferrari KJ, Scelfo A, Jammula S, Cuomo A, Barozzi I, Stutzer A, Fischle W, 2017;12:255. Bonaldi T, Pasini D. Polycomb-dependent H3K27me1 and H3K27me2 reg- 30. Sengoku T, Yokoyama S. Structural basis for histone H3 Lys 27 demethyla- ulate active transcription and enhancer fidelity. Mol Cell. 2014;53:49–62. tion by UTX/KDM6A. Genes Dev. 2011;25:2266–77. 22. McCabe MT, Graves AP, Ganji G, Diaz E, Halsey WS, Jiang Y, Smitheman 31. Del Rizzo PA, Trievel RC. Molecular basis for substrate recognition by KN, Ott HM, Pappalardi MB, Allen KE, et al. Mutation of A677 in histone lysine methyltransferases and demethylases. Biochim Biophys Acta. methyltransferase EZH2 in human B-cell lymphoma promotes hypertri- 2014;1839:1404–15. methylation of histone H3 on lysine 27 (H3K27). Proc Natl Acad Sci USA. 32. Kaniskan HU, Jin J. Recent progress in developing selective inhibitors of 2012;109:2989–94. protein methyltransferases. Curr Opin Chem Biol. 2017;39:100–8. 23. Sarma K, Margueron R, Ivanov A, Pirrotta V, Reinberg D. Ezh2 requires 33. Tsuda H, Zhao N, Imai K, Ochiai K, Yang P, Suzuki N. BIX01294 suppresses PHF1 to efficiently catalyze H3 lysine 27 trimethylation in vivo. Mol Cell osteoclast differentiation on mouse macrophage-like Raw264.7 cells. Biol. 2008;28:2718–31. Bosn J Basic Med Sci. 2013;13:271–5. 24. Shen X, Liu Y, Hsu YJ, Fujiwara Y, Kim J, Mao X, Yuan GC, Orkin SH. EZH1 34. Duncan EM, Muratore-Schroeder TL, Cook RG, Garcia BA, Shabanowitz J, mediates methylation on histone H3 lysine 27 and complements EZH2 Hunt DF, Allis CD. Cathepsin L proteolytically processes histone H3 during in maintaining stem cell identity and executing pluripotency. Mol Cell. mouse embryonic stem cell differentiation. Cell. 2008;135:284–94. 2008;32:491–502. 35. Steiner LA, Schulz VP, Maksimova Y, Wong C, Gallagher PG. Patterns of 25. Wu H, Chen X, Xiong J, Li Y, Li H, Ding X, Liu S, Chen S, Gao S, Zhu B. histone H3 lysine 27 monomethylation and erythroid cell type-specific Histone methyltransferase G9a contributes to H3K27 methylation in vivo. gene expression. J Biol Chem. 2011;286:39457–65. Cell Res. 2011;21:365–7. 36. Kim JM, Kim K, Punj V, Liang G, Ulmer TS, Lu W, An W. Linker histone H1.2 26. Kubicek S, O’Sullivan RJ, August EM, Hickey ER, Zhang Q, Teodoro ML, Rea establishes chromatin compaction and gene silencing through recogni- S, Mechtler K, Kowalski JA, Homon CA, et al. Reversal of H3K9me2 by a tion of H3K27me3. Sci Rep. 2015;5:16714. small-molecule inhibitor for the G9a histone methyltransferase. Mol Cell. 2007;25:473–81. 27. Konze KD, Ma A, Li F, Barsyte-Lovejoy D, Parton T, Macnevin CJ, Liu F, Gao C, Huang XP, Kuznetsova E, et al. An orally bioavailable chemical Ready to submit your research ? Choose BMC and benefit from: fast, convenient online submission thorough peer review by experienced researchers in your field rapid publication on acceptance support for research data, including large and complex data types • gold Open Access which fosters wider collaboration and increased citations maximum visibility for your research: over 100M website views per year At BMC, research is always in progress. Learn more biomedcentral.com/submissions

Journal

Epigenetics & ChromatinSpringer Journals

Published: May 28, 2018

References

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off