The G9a/CHCHD2/Sirt1 regulatory module acts on RNase H1 to control R-loop formation at rDNA sites

Le Li College of Life Sciences Wuhan University Yequn Wu College of Life Sciences Wuhan University Kui Dai College of Life Sciences Wuhan University Qing Wang Wuhan University Shiqi Ye College of Life Sciences Wuhan University Qipeng Shi College of Life Sciences Wuhan University Zhenfei Chen College of Life Sciences Wuhan University Yi-Chun Huang College of Life Sciences Wuhan University Weiwei Zhao Wuhan University Lijia Li (  ljli@whu.edu.cn ) College of Life Sciences Wuhan University


Introduction 26
The major function of the nucleolus is to transcribe ribosomal RNA (rRNA) and to assemble 27 ribosome subunits; this process must be tightly regulated to achieve proper cellular proliferation and 28 growth (Boisvert et al., 2007). The rRNA transcription abundance controls ribosome biogenesis and thus influences protein synthesis capacity, which regulates the cell growth and division rate in 30 response to cellular stimuli (Mayer and Grummt, 2006). The rDNA gene is a region with highly 31 active transcription, and the R-loop formation is a natural and frequent event during rRNA washing buffer (50 mM Tris-HCl pH 7.5, 10 mM EDTA, 50 mM NaCl; 100 mM NaCl; 150 mM 170 NaCl), and then eluted with 50 mM Tris-HCl (pH 7.5), 10 mM EDTA, 50 mM NaCl, 1%SDS and 171 20 µg proteinase K for 60 min at 55°C. Immunoprecipitated DNA was analyzed by quantitative 172 PCR using the primers listed in Supplementary Table 3. DNA in the immunoprecipitates was 173 compared with input DNA, and the difference between untreated and RNase H-treated samples is 174 presented as DRIP signals. 175 lysed with 1 mL lysis buffer (1 mM Tris-HCl pH 7.5, 1% SDS, 0.2 mM EDTA, 0.1 mM PMSF, 0.1 179 mM DTT, 0.1% Protease Inhibitor Cocktail (P8340-1ML, Sigma)). The genomic DNA dissolved in 180 lysis buffer was sonicated to 500-750 bp by ultrasonic fragmentation, of which 40 µl was used as a 181 positive control, the remaining was divided into two and added into equal volumes of incubation 182 buffer (consistent with the composition of the incubation buffer used in the DRIP assay), which 183 were blocked with rProtein A Sepharose Fast Flow and Normal Rabbit Serum (BMS0090, Abbkine, 184 Wuhan, China) to remove non-specific antibodies. After centrifugation, the supernatant was 185 incubated overnight at 4°C with antibody and then bound with protein A for 3 h. IgG-Rb (A7016, 186 Beyotime, Shanghai, China) was used as a negative control for mock immunoprecipitation. The 187 precipitate after centrifugation was washed three times with a gradient of 1 mL washing buffer 188 (consistent with the composition of the incubation buffer used in the DRIP assay), and then eluted 189 with 60℃ preheated elution buffer (20 mM Tris-HCl pH 7.5, 50 mM NaCl, 5 mM EDTA, 0.2 mM 190 PMSF, 0.2 mM DTT, 1% SDS). The resulting eluate was incubated with 200 mM NaCl and 20 µg 191 of proteinase K at 55°C for 6 h for decrosslinking followed by 20 µg of RNase A for 30 min at 37 °C. 192 The DNA was then precipitated according to the DNA purification procedure, and the precipitate 193 was subjected to quantitative PCR using the primers shown in Supplementary Table 3 Western blot analysis 216 Total proteins extracted from the treated cells using extraction buffer (100 mM Tris-HCl pH 7.5, 50 217 mM NaCl, 5 mM EDTA ,1 mM PMSF and 1 mM DTT) were separated by electrophoresis in a SDS-218 page gel. Then the proteins were transferred to PVDF membranes, blocked by 5% milk at room 219 temperature for 2 h and incubated overnight at 4°C together with antibodies. The immunoreactive 220 bands were observed by chemiluminescence after binding of the corresponding secondary antibody. 221 The secondary antibodies were the horseradish peroxidase (HRP) labeled goat anti-mouse IgG 222 (A0126, Beyotime, Shanghai, China,) and the HRP labeled goat anti-rabbit IgG (A3327, Beyotime, 223 Shanghai, China). Immunoreactivity was determined using the ECL method (K-12045-D50, 224 advansta, California, USA) according to the manufacturer's instructions (Zhou et al., 2021). 225

Luciferase reporter assays 226
The promoter of the target gene was constructed into the pGL3 plasmid with the firefly luciferase 227 gene, and the plasmid phRL-TK with the renilla luciferase gene was used as a control to co-transfect 228 into cells with the reporter gene. The total protein was obtained by lysing the cells with the lysis 229 solution in the dual fluorescence assay kit (E1910, Promega, Madison, USA). The firefly 230 fluorescence signal was first generated when the Luciferase Assay Reagent II was added through 231 automatic sample injection system, after quantifying the intensity of firefly fluorescence. The 232 Stop&Glo Reagent was added to the same sample to quench the above reaction and simultaneously 233 as means ±SD (Farr and Roman, 1992;Sherf et al., 1996). 239

GST-Pull down 240
The proteins CHCHD2-1 with a His tag, CHCHD2-2 with a His tag, G9a with a GST tag and Sirt1 241 with a GST tag were expressed in Escherichia coli (BL21) and purified using the His tag protein 242 purification kit (P2226, Beyotime, Shanghai, China) or the GST tag protein purification kit (P2262, 243 Beyotime, Shanghai, China). The GST-Pull down assays were carried out according to the method 244 reported by Einarson et al. (Einarson et al., 2007). The proteins carrying those two tags were 245 incubated together with equal amounts of pulldown binding buffer (50 mM Tris-HCl pH 8.0, 250 246 mM NaCl, 1 mM EDTA, 1% NP-40, 10 mM MgCl2, 0.2 mM PMSF and 0.2 mM DTT) and 50 μL 247 of BeyoGold TM GST-tag Purification resin (rinsing three times in pulldown binding buffer) for 2 h 248 at 4°C with end-over-end mixing. Centrifuge the samples at 13,000 rpm for 10 s at 4°C in a 249 microcentrifuge and wash the beads 6 times with 1 mL of ice-cold washing buffer (50 mM Tris-HCl 250 pH 8.0, 300 mM NaCl, 1 mM EDTA, 1% NP-40,10 mM MgCl2, 0.2 mM PMSF and 0.2 mM DTT). 251 Discard the washes, and then detect with antibodies specific for the tag and the target protein by 252 western blot analysis. 253 Co-IP assay The data and error bars were calculated from three independent experiments. The data in this 294 manuscript were analyzed for significant differences between the experimental groups and control 295 groups using the t-test which was performed using the Microsoft Excel (2019). "Two tails" was used S9.6. We conducted a real-time quantitative PCR at the rDNA locus ( Figure 1A), of which the 18S 307 rRNA-coding region (amplicon H4, H4-) and 28S rRNA-coding region (amplicon H8) are the hot 308 spot for R-loop formation, and we found that RNase H1 regulates the formation of R-loops at the 309 well-characterized hybrid-forming site. Analysis of the DRIP-qPCR signal from RNase H1-depleted 310 cells revealed a significant 2~fold increase at amplicon H4/H4-and a 1.45~fold increase at amplicon 311 H8 in R-loops compared with those in siNC cells ( Figure 1B). As expected, RNase H1-312 overexpressed cells showed a significant decrease in RNA-DNA hybrids compared with those in 313 control cells (Figure 1 C). The specificity of the DRIP-qPCR approach was controlled by vitro 314 treatment with RNase H, which compromised the DRIP signal (data not shown). Further dissection 315 of the role that RNase H1 plays revealed that RNase H1 activity has been linked to the removal of 316 R-loops in human rDNA. 317 Our previous studies suggested that loss of H3K9 dimethylation (H3K9me2) triggered the R-318 loop accumulation at the rDNA locus, which further led to the multilobed nucleoli, implying that 319 In order to understand the mechanism underlying G9a-mediated regulation of R-loop formation, 329 by blast analysis of the G9a Interaction Protein Database (Rolland et al., 2014), we found that an 330 oxidative stress-related protein CHCHD2 (MNRR1) might be associated with G9a and involved in 331 regulating R-loop formation. Thus, we examined the R-loop levels at the rDNA locus in the stable 332 shCHCHD2 HeLa cells, and the DRIP-qPCR results confirmed that knockdown of CHCHD2 333 repressed R-loop accumulation. By contrast, over-expression of CHCHD2-isoform2 increased R-334 loop accumulation more than 2~fold compared with the control. Over-expression of CHCHD2-335 isoform1 is less effective than CHCHD2-isoform2 in promoting R-loop formation at the rDNA site, 336 especially at the amplicon H4-( Figure 1E). We further observed decreased R-loop levels at the 337 amplicon H4-/H8 of rDNA locus when CHCHD2 was knocked down in the stable shG9a HeLa cells 338 ( Figure 1D). As a control for specificity, we pretreated the extracted nucleic acids with RNase H 339 enzyme in vitro to degrade existing RNA-DNA hybrids. Altogether, these results suggest that low 340 expression of G9a or high expression of CHCHD2 lead to R-loop enrichment at the rDNA locus, 341 and CHCHD2 functions at the downstream of G9a. 342 Considering that RNase H1, G9a and CHCHD2 are involved in the formation of RNA-DNA 343 hybrids at the rDNA site, we focused our research on the relationship between RNase H1, G9a and 344 CHCHD2. Endogenous RNase H1 ChIP showed that the recruitment of RNase H1 at the rDNA 345 amplicon H4-was dependent on G9a regulation ( Figure 1F). Compared with the shcon cells, the 346 shG9a cells showed a significant decrease in RNase H1 recruitment. Meanwhile, the cells that 347 ectopically expressed G9a WT showed a marked increase of RNase H1 occupancy; on the contrary, 348 G9a ΔSET failed to promote RNase H1 recruitment. Interestingly, CHCHD2 also participated in the 349 recruitment of RNase H1 at the rDNA amplicon H4-( Figure 1G). Knockdown of CHCHD2 boosted 350 the recruitment of RNase H1 compared with the control, but over-expression of CHCHD2-isoform2 351 reduced RNase H1 occupancy more than 2~fold at the amplicon H4-. Over-expression of CHCHD2-352 isoform1 had no effect on attenuating recruitment of RNase H1. Together with the DRIP results, 353 S9.6 ChIP was used to detect the R-loop levels and further supported the idea that G9a and CHCHD2 354 are involved in R-loop formation by mediating the recruitment of RNase H1 at the rDNA locus.   Figures 3C and 3D). 417 When we precisely controlled the expression of G9a WT in shcon HeLa cells or stable shG9a HeLa 418 cells, a marked increase in RNase H1 protein level was observed along with the gradient up-419 regulation of G9a WT ( Figure 3B). In order to explore the relationship between G9a enzyme 420 activity and RNase H1 expression, we tested the RNase H1 mRNA level after 48 h of treatment with 421 10 μM BIX or 10 μM BRD, and found that the mRNA level was down-regulated (Supplementary 422 Figure 2C). Interestingly, when using western blot to detect the RNase H1 protein level after 48 h 423 of treatment with different concentrations of BIX or BRD, we found that the RNase H1 protein 424 showed a concentration-independent decrease (Supplementary Figure 3A). RNase H1, we transfected the stable G9a knockdown HeLa cells with G9a WT or G9a ΔSET 429 expression plasmids. We found that G9a ΔSET group did not show increased RNase H1 expression 430 whereas the G9a WT group increased its expression ( Figure 3C), indicating that G9a-mediated up-431 regulation of RNase H1 expression is dependent on its HMTase activity. Taken together, these 432 results suggest that G9a positively regulates expression of RNase H1 in a SET-dependent manner. 433 Our results also showed that knockdown of CHCHD2 led to an increase in RNase H1 434 transcription (Supplementary Figure 2F). In addition, knockdown of CHCHD2 in the stable shG9a 435 cells also caused an increase in RNase H1 transcription (Supplementary Figure 2E). Consistent 436 with the RT-qPCR, the western blot results showed that RNase H1 levels were increased after 437 knockdown of CHCHD2 ( Figure 3D) whereas RNase H1 expression was repressed by CHCHD2 438 overexpression ( Figure 3E). When the expression of CHCHD2 was restored in the stable 439 shCHCHD2 HeLa cells, the RNase H1 protein returned to normal levels ( Figure 3F). Specially, the 440 CHCHD2-isoform1 showed no effect on the expression of RNase H1. After overexpressing 441 CHCHD2-isoform2 in shcon HeLa cells, stable shG9a HeLa cells and G9a WT rescued shG9a HeLa 442 cells, the western blot results indicated that overexpression of CHCHD2-isoform2 could further 443 reduce the RNase H1 protein levels ( Figure 3G). These results show that knockdown of G9a or and CHCHD2, we conducted a luciferase reporter assay using RNase H1-promoter-luc reporter 447 system. We first cloned three fragments with different RNase H1 promoter lengths and engineered 448 these RNase H1 promoter fragments into pGL3 basic luciferase reporter vectors ( Figure 4A). Then, 449 we selected the pGL3-RH1-pro2-luc with the highest promoter activity for subsequent luciferase 450 assay ( Figure 4B). Consistent with RT-qPCR and western blot results, RNase H1 transcription was 451 stimulated by depletion of CHCHD2 and repressed by overexpression of CHCHD2 ( Figure 4C). 452 The inhibitory effect of CHCHD2-isoform1 is not as obvious as that of CHCHD2-isoform2. The co-localization of CHCHD2 and G9a in the HeLa cell nuclei was confirmed by using 473 immunofluorescence staining with the CHCHD2 monoclonal antibody (66302-1-Ig) and the G9a 474 polyclonal antibody (ab183889) ( Figure 5A). When using the His antibody or the CHCHD2 was consistent with the size of input MBP-CHCHD2-His, revealing that both CHCHD2-isoform1 477 and CHCHD2-isoform2 could interact specifically with G9a in vitro ( Figure 5B). After G9a was 478 immunoprecipitated from HeLa cells with the G9a polyclonal antibody, CHCHD2 was detected in 479 the precipitate at the same position as the input, showing that CHCHD2 could interact with G9a in 480 vivo ( Figure 5C).We constructed yeast two-hybrid system bait and prey vectors to confirm the 481 importance of G9a HMTase activity domain for direct interaction with CHCHD2. Interestingly, the 482 direct interaction between CHCHD2-isoform2 and G9a showed a clear SET domain dependency 483 (Supplementary Figure 4A). However, G9a ΔSET and CHCHD2-isoform1 still have a certain 484 weak interaction compared with interaction between G9a WT and CHCHD2-isoform1. In the GST-485 pull-down system, we verified that CHCHD2-isoform1 and CHCHD2-isoform2 could directly 486 interact with G9a depended on the G9a SET domain. Once the SET domain was destroyed, this 487 interaction collapsed (Supplementary Figure 4B).  Figure 5) and a GST pulldown assay ( Figure 5E) further verified that 493 CHCHD2 was the interacting protein of Sirt1. Then, we purified and incubated the recombinant 494 MBP-CHCHD2-His protein and the GST-Sirt1 protein to construct a deacetylation reaction system 495 in vitro. A specific band was displayed using the Anti-Acetylated-Lysine antibody at a position that 496 was consistent with the size of input MBP-CHCHD2-His in the GST empty protein reaction 497 products, suggesting that CHCHD2 in the prokaryotic expression system could be acetylated. 498 Comparing with the GST empty protein, the addition of GST-Sirt1 could significantly reduce the 499 lysine acetylation level of MBP-CHCHD2-His, indicating that Sirt1 directly deacetylated CHCHD2 500 ( Figure 5F). To further confirm that CHCHD2 is the target of Sirt1-induced deacetylation in vivo, 501 we used the stable shSirt1 HeLa cell to perform lysine acetylation immunoprecipitation. 502 Knockdown of Sirt1 increased the basal acetylation level of endogenous CHCHD2, exogenous 503 CHCHD2-isoform1 and CHCHD2-isoform2. The acetylation could also be retrieved to the normal 504 level by re-introduction of Sirt1 into the stable shSirt1 HeLa cells ( Figure 5G). Taken together, 505 these results suggest that CHCHD2 interacts with G9a and is deacetylated by Sirt1.

507
The DRIP analysis result showed that knockdown of Sirt1 reduced the R-loop levels whereas 508 overexpression of Sirt1 showed a significant promotion for the R-loop accumulation especially in 509 rDNA amplicon H4/H4-/H8, which was similar to that obtained from CHCHD2 ( Figures 6A). RT-qPCR and western blot results showed that loss of Sirt1 increased expression of RNase H1 523 ( Figures 6D and 6E). When we precisely controlled the expression of Sirt1 in shcon HeLa cells or 524 stable shSirt1 HeLa cells, a marked decrease in the RNase H1 protein level was observed along with 525 the gradient up-regulation of Sirt1 ( Figure 6F, Supplementary Figure 3E). We further used the 526 RNase H1-luc reporter system to examine Sirt1-mediated transcriptional regulation of RNase H1. 527 RNase H1 transcription was repressed by Sirt1 overexpression in the stable shSirt1 HeLa cell, and 528 the treatment with EX 527 could attenuate the inhibition of RNase H1 transcription triggered by 529 Sirt1 ( Figure 6G). Simultaneously, we examined whether CHCHD2 was involved in Sirt1-mediated 530 transcriptional regulation of RNase H1. When overexpressing CHCHD2-isoform2 in shcon HeLa 531 cells, stable shSirt1 HeLa cells and Sirt1 rescued shSirt1 HeLa cells, western blot results indicated 532 that CHCHD2-isoform2 could further reduce the protein level of RNase H1 ( Figure 6H), suggesting 533 that CHCHD2 could cooperate with Sirt1 to inhibit the expression of RNase H1. 534 We then tested the connection between G9a and Sirt1 in influence of RNase H1 expression. 535 The results of RNase H1-luc reporter assay in stable shG9a HeLa cells showed that G9a promoted RNase H1 transcription, but overexpression of Sirt1 further abolished the RNase H1 transcriptional 537 activation induced by G9a, and the shSirt1 group had a significant recovery in RNase H1 538 transcriptional repression (Supplementary Figure 6B). Similarly, the results of RNase H1-luc 539 reporter assay in stable shSirt1 HeLa cells showed that Sirt1 repressed RNase H1 transcription, but 540 overexpression of G9a showed slight recovery of the Sirt1-mediated transcriptional inhibition of 541 RNase H1. Knockdown of G9a further strengthen Sirt1-mediated RNase H1 transcriptional 542 repression (Supplementary Figure 6C). In addition, we performed a RNase H1-luc reporter assay 543 in stable shCHCHD2 HeLa cells to further investigate whether G9a or Sirt1 had any effect on 544 CHCHD2-mediated transcriptional repression of RNase H1. After transfected with OESirt1 or G9a 545 WT for 24 h, the shSirt1-#3 or shG9a-#1 were added into the transfection system for another 24 h. 546 The results showed that CHCHD2 indeed repressed RNase H1 transcription. Overexpression of 547

Sirt1 further strengthened CHCHD2-mediated RNase H1 transcriptional repression. Knockdown of 548
Sirt1 partially restored the RNase H1 transcriptional repression compared with OESirt1, strongly 549 suggesting that Sirt1 was involved in CHCHD2-induced RNase H1 transcriptional repression 550 Figure 6A). Notably, overexpression of G9a abolished the RNase H1 Supplementary Figure 3C and 3D). Thus, we conducted luciferase reporter assay using a 557 CHCHD2-luc reporter system to examine G9a-mediated transcriptional regulation of CHCHD2. We 558 cloned three fragments with different CHCHD2 promoter lengths and engineered these CHCHD2 559 promoter fragments into pGL3 basic luciferase reporter vectors ( Figure 7A). Then, we selected the 560 pGL3-CHCHD2-pro2-luc with the highest promoter activity for subsequent luciferase assay 561 ( Figure 7B). The results showed that CHCHD2 transcription was upregulated by depletion of G9a 562 and repressed by G9a overexpression in the stable shG9a HeLa cells ( Figure 7C). Simultaneously, 563

(Supplementary
G9a ΔSET did not significantly repress the expression of CHCHD2 compared with G9a WT, 564 indicating that G9a regulated CHCHD2 in a SET-dependent manner ( Figure 7C). Sirt1 functions knockdown of Sirt1 also resulted in upregulation of CHCHD2 ( Figures 6D, 6E and 6F, 567 Supplementary Figure 3E), and we used the same CHCHD2-luc reporter system to examine Sirt1-568 mediated transcriptional regulation of CHCHD2 ( Figure 7A and 7B). Consistent with the western 569 blot results, CHCHD2 transcription was stimulated by depletion of Sirt1 and repressed by Sirt1 570 overexpression ( Figure 7D). 571 G9a prevents CHCHD2 from being recruited to the promoter of the RNase H1

572
To further elucidate the mechanisms underlying RNase H1 transcriptional regulation by G9a, Sirt1 573 and CHCHD2, we performed the ChIP analysis with RT-qPCR using corresponding stable 574 knockdown HeLa cells. We analyzed the RNase H1 promoter sequence to identify possible 575 transcription factor binding sites. The histone modification of the RNase H1 promoter region was 576 discovered in the ChIP-seq public database Cistrome Data Browser, and RT-qPCR primers for ChIP 577 analysis were designed for the H3K9ac and H3K9me2 enrichment peak positions of the RNase H1 578 promoter region (Supplementary Figure 7A). The final primers RH pro A, B and C covered the 579 distal, middle and proximal regions of the RNase H1 promoter ( Figure 8A). First, we observed 580 decreased G9a recruitment as well as decreased levels of H3K9me2 on the RNase H1 promoter in 581 stable shG9a HeLa cells whereas G9a was highly recruited to the RNase H1 promoter and H3K9me2 582 levels increased when G9a was overexpressed ( Figure 8B). Interestingly, the CHCHD2 and Sirt1 583 recruitment both increased (almost ~2.5 fold and ~3 fold) when G9a was knocked down. Then, we 584 overexpressed G9a in the stable shG9a HeLa cells and observed that CHCHD2 and Sirt1 recruitment 585 on the RNase H1 promoter was decreased significantly. The H3K9ac levels on the promoter region 586 of RNase H1 also changed with the change of Sirt1 enrichment ( Figure 8B). These results suggest 587 that high expression of G9a along with high level of H3K9me2 prevent CHCHD2 and Sirt1 from 588 accessing the RNase H1 promoter to activate RNase H1. 589 Furthermore, in the absence of Sirt1, the H3K9ac level on the RNase H1 promoter was 590 significantly upregulated, and more G9a was bound to the RNase H1 promoter, which led to 591  and methylate the promoter of the RNase H1 gene, which inhibited CHCHD2 binding. By contrast, 669 when G9a was knocked down, the decreased expression of G9a resulted in a reduction of H3K9me2 670 markers at the promoter of the RNase H1 gene, which was conducive to the recruitment of CHCHD2 671 to suppress RNase H1 expression. We also found that G9a could directly interact with CHCHD2 672 which possibly decreased free CHCHD2. Previous studies have predicted that CHCHD2 is a target 673 of Sirt1-induced deacetylation (Aras et al., 2020). Our results revealed that Sirt1 could indeed 674 interact with and deacetylate CHCHD2. We found that loss of G9a led to the recruitment of more 675 Sirt1 as well as more CHCHD2 to the RNase H1 promoter to co-suppress transcription of the RNase 676 H1 gene (Figure 9). By the contrary, loss of Sirt1 led to binding of more G9a to the RNase H1 677 promoter due to the increase of H3K9ac markers at the promoter of the RNase H1 gene. The 678 enrichment trends of factors in the G9a/CHCHD2/Sirt1 functional module were almost the same in 679 each RNase H1 promoter region. However, when the expression level of Sirt1 changed, the 680 significant effect on the recruitment level of CHCHD2 was only manifested in the RNase H1 681 proximal promoter region. The RNase H1 distal promoter region was not considered because the 682 changes in Sirt1 enrichment and H3K9me2 levels were not significant in the three groups of control 683 experiments. After knockdown of Sirt1, although G9a and corresponding H3K9me2 levels which could be because Sirt1 negatively regulated the expression of CHCHD2 (Figures 6D, 6E  686 and 6F, Figure 7D, Supplementary Figure 3E). 687 In summary, this study showed that G9a boosts the recruitment of RNase H1 and positively 688 regulates RNase H1 expression whereas CHCHD2 suppresses RNase H1 recruitment and acts as a 689 repressive transcription factor to inhibit the expression of RNase H1 to increase R-loop formation 690 at the rDNA site. CHCHD2 can form a complex with Sirt1 as the co-repressor, which binds to the 691 RNase H1 promoter under depleting of G9a. These findings provide a possible strategy to regulate 692 R-loop formation, rRNA transcription and cancer cell growth through co-targeting G9a, CHCHD2 693 and Sirt1. 694

695
Ethics approval and consent to participate Not applicable.

696
Consent for publication Not applicable.

697
Availability of data and materials All data generated or analyzed during this study 698 are included in this published article and its supplementary information files or are 699 available upon request.

700
Competing interests The authors declare that they have no competing interests.