SIRT7 is a histone desuccinylase that functionally links to chromatin compaction and genome stability

Although SIRT7 is a member of sirtuin family proteins that are described as NAD+-dependent class III histone deacetylases, the intrinsic enzymatic activity of this sirtuin protein remains to be investigated and the cellular function of SIRT7 remains to be explored. Here we report that SIRT7 is an NAD+-dependent histone desuccinylase. We show that SIRT7 is recruited to DNA double-strand breaks (DSBs) in a PARP1-dependent manner and catalyses desuccinylation of H3K122 therein, thereby promoting chromatin condensation and DSB repair. We demonstrate that depletion of SIRT7 impairs chromatin compaction during DNA-damage response and sensitizes cells to genotoxic stresses. Our study indicates SIRT7 is a histone desuccinylase, providing a molecular basis for the understanding of epigenetic regulation by this sirtuin protein. Our experiments reveal that SIRT7-catalysed H3K122 desuccinylation is critically implemented in DNA-damage response and cell survival, providing a mechanistic insight into the cellular function of SIRT7.

other lines. In this context, the study will benefit if some of the claims made in knockdown cells will be shown in SIRT7 KO cells.
-Although the mass spectrometry assays seem solid, the biochemistry to support these results is not convincing. For instance, most of the westerns presented exhibit modest changes in the putative succinylation sites. It is well established that histone modifications are difficult to quantify in westerns, given intrinsic variability in signal. All these westerns should be performed at least in triplicates, and densitometry should be performed in order to provide a quantification of these experiments. As it stands, the data does not look convincing. Particularly, Fig. 1D, it seems the line for the Pan-succ antibody has a bubble exactly for the Sirt7wt and SIRT7S11A extracts. Similarly, their newly generated antibodies should be further characterized. In Fig. 1F, the authors show that the presence of succ. peptides in competition assays completely eliminated the signal in the westerns, yet they did not show other peptides (with different modification, or even H3 unmodified peptides) as negative controls. The changes claimed in Fig. 1G are modest. Without quantification and replicates, these results are inconclusive. The same could be said for all the blots in Fig. 2C-D, and those in Fig. 4H-I.
-In this context, the western blot in Fig. 1H is worrisome. The authors claim that only NAM treatment increased H3K122succ levels, yet it seems that also TSA and Sodium butyrate increased its levels, a result that goes against their hypothesis. Of concern, treatment with TSA should majorly increase pan-acetylation, yet the western showed only a modest effect (despite the authors claiming "marked increase"). Such results raise concerns on the solidity of the whole western blot.
-The IF shown in Fig. 1E is not convincing. Has this antibody been tested in IF assays before.
The authors should include controls to verify the antibody truly recognize histone succinylation.
-Loading controls and quantification should be included in the dot blot assays in Fig. 2B.
-The data in Fig. 3, indicating recruitment of SIRT7 to sites of DSBs is compelling.
-The experiments shown in Fig.4, demonstrating a role for SIRT7 in both HR and NEHJ DNA repair seem robust. It will be reassuring to see the original FACs plots for at least some of these experiments.
-However, in this context, how long after ISceI transfection was the experiment in figure 4J done? It is very difficult to assess any kinetics in this setting since this system is based on cutrepair cycles in the presence of the ISceI enzyme. -Few grammatical errors should be corrected. For instance, on page 4 "in last decade" should be "in the last decade". On page 20, "a residue could not be succinylated" should be "a residue that could not be succinylated".
Reviewer #2 (Remarks to the Author): In this study, Shi et al report a new enzymatic activity of the histone/protein deacetylase SIRT7, de-succinylation of histone H3 at Lysine K122. They show that SIRT7 associates with chromatin at DNA double-strand breaks (DSBs), where its desuccinylation of H3K122 is important for chromatin compaction and DSB repair. The identification of new enzymatic activities and substrates of SIRT7 is important for the SIRT7 field, and the identification of the functions and regulatory enzymes of new histone modifications such as succinylation are important for the chromatin field. Overall the data are clear and the experiments are well described.
Major concerns: 1. In Figure 4H and I, SIRT7 RNAi seems to increase baseline γ-H2AX but does not increase γ-H2AX upon DNA damage (in fact, in 4i, IR-induced γ-H2AX is decreased by SIRT7 RNAi). By contrast, the effects on H2K122 succinylation occur only in response to DNA damage. This suggests that the effects on DNA damage response might not be directly due to the desuccinylation by SIRT7. This seems problematic for the author's model.
2. Expression of either the H3K122E succinyl mimic or the H3K122R non-succinylated mutant reduces NHEJ, HR, and DNA damage sensitivity (Fig. 5C, 6D). But in the MN assays ( Figure 5A) versus the DNA repair proteins reported in this study. How do the authors account for this discrepancy? Did they detect any of the proteins from the previous study? Given the frequency with which the Kus and Parp are found in Flag-co-IPs, there is concern the interactions are not physiologic. The data should be confirmed by co-IP of the endogenous proteins. 4. The validation of the H3K122-Succ antibody by peptide competition does not rule out that the antibody might detect the peptide independent of succinylation. The authors should show competition (or lack of it) by the non-succinylated peptide.
5. To really prove that SIRT7 directly desuccinylates H3K122-succ in vitro, the authors should show some mass spec analysis in addition to westerns. 6. Did the authors test whether SIRT7 deacetylates H3K122Ac? 7. The introductory statement (and other comments in the introduction): "the intrinsic enzymatic activity of this sirtuin protein remains to be investigated and the cellular function of SIRT7 remains to be explored." incorrectly implies that nothing is known about SIRT7 enzymatic activity or cellular function.
Additional issues: 1. In the quantification of succinylation in Figure 1C, the error bars are very large and it is difficult to get a sense of the fold change in succinylation. Can the data be plotted on a more meaningful scale, and with comparisons to sites that were not significantly altered and negative controls.
2. The authors do not indicate why the SIRT7 S111A mutation (and the double S111A/H187A mutation) was studied, and they should provide a brief explanation in the text for what these mutations were intended to test. Mutation of the corresponding residue in other sirtuins was shown to reduce deacetylase activity (Min et al Cell 2001), and previous studies showed SIRT7 S111A impairs rDNA transcriptional activation by SIRT7 (Ford et al, Genes Dev 2006). The authors should discuss their observations that the mutation appears not to affect H2K122 desuccinulation or H3K18 deacetylation.
3. The deacetylase activity of SIRT7 on H3K18 has been shown to be more efficient on nucleosomal histones than free histones from calf thymus (Gil et al, NAR 2013). It is difficult compare the data for de-succinylation of H3K122 in these contexts in Figure 2 quantitatively (more SIRT7 was used in the nucleosome assays), but it seems that activity may be greater on free histones. A more quantitative comparison of the activities on nucleosomes versus CTHs should be straightforward, and could have implications for the contexts in which the desuccinylation actually occurs. 4. It would be nice to see if SIRT7 co-localizes with γ-H2AX at global DNA breaks induced by IR or other DSB agents? 5. Demonstration of Parp-1 dependence of SIRT7 relocalization to DSBs would be more convincing with specific Parp-1 inactivation by RNAi or knockout.
6. The authors should discuss how they envision chromatin compaction in response to desuccinylation might promote DNA repair? This is opposite of the model that decreased compaction can create accessibility for repair factors.
Reviewer #3 (Remarks to the Author): In this manuscript, the authors reported sirt7 as a desuccinylase using in vitro enzymatic assay and validate this activity in cell lines by over expression and knocking down experiments. The authors further demonstrated that sirt7 was associated with DNA repair proteins by IP/MS experiments. Interestingly, sirt7 was recruited to DNA double strand breaks (DSB) by PARP1 and successful repair of DSB was dependent on desuccinylation of H3K122 by sirt7. This is a nice story with significant novelty. The authors discovered that sirt7, one of the least studied sirtuin family members, had new activity-desuccinylase activity. They also provided functional studies and suggested that histone succinylation played an important role in DNA repair activity. Generally, the authors had proven their points by a series of in vitro and in cell line experiments. However, I suggest the authors to improve some functional studies before publication.
Major points: 1. Desuccinylase activity of sirt7: In the in vitro enzymatic assay, sirt7 was shown to have desuccinylase activity toward H3K122succ peptide while sirt5 does not. Considering the well-studied desuccinylase activity of sirt5, this should be clarified. Is it possible that sirt5 and sirt7 having specificities toward different succinylation sites? Is H3K122 the only site that sirt7 has activity in vitro? The authors are suggested to study more substrates to strengthen their point.

H3K122suc in DSB repair:
Sirt7 was shown to be recruited to DSB sites in a Parp1 dependent way. Is H3K122suc level decreased at the DSB sites where sirt7 was recruited? If yes, is the decrease of H3K122suc Parp1 dependent? The authors only studied H3K122suc level by western blot, which may not be specific to the DSB sites. Laser microirradiation and immunofluorescent assay using anti-H3K122suc or pan anti-Ksuc antibodies should be considered.

H3K122suc vs H3K122ac:
The authors showed that desuccinylase activity of sirt7 was indispensable for DSB repair. Does sirt7 have other activities such as deacetylase in addition to desuccinylase activity? Care should be taken that mutation of H3K122 to E/R has influence not only on succinylation but also on acetylation. The authors should study whether H3K122ac level is changed during DNA damage insult and whether sirt7 has deacetylase activity in vitro and in cell lines.
Minor points: 1. In Figure1E, the authors are suggested to study immunofluorescence by over expression (OE) of Sirt7, showing sirt7 and lysine succinylation levels in the same cell, in a higher resolution picture. Figure 1D, the first panel indicates an uneven exposure for pan-succ signals, especially between sirt7wt and sirt7s111a.

In
3. In Figure 1F, the middle panel, the blot signals are too weak to demonstrate the antibody specificity clearly. 4. On page12 of manuscript, the authors described "treatment with TSA or sodium butyrate had no effect on the level of H3K122succ, although treatment with NAM, TSA, or sodium butyrate all resulted in a marked increase in the level of H3 pan-acetylation ( Figure 1H)". However, in Figure 1H, H3K122suc level was increased by either HDAC inhibitor treatment. Figure 4C, there is no labeling about which one is "NHEJ" or "HR".

In
6. "HBK20succ" is misspelled in Figure 4G. 7. In Introduction, wrong citation (#34) for sirt5 as an enzyme for malonylation. Major comments: 1. The rationale to use only MCF7 and 293T for the biochemical assays is not clear. Other lines exhibit equally "low" or "high" levels, and since the authors do not validate their experiments in more physiological conditions (in vivo), their results will be more convincing if replicated in other lines. In this context, the study will benefit if some of the claims made in knockdown cells will be shown in SIRT7 KO cells.
Authors: To comply with the reviewer's suggestions, we examined H3K122succ levels in control and SIRT7-depleted HCT116 and HeLa cells, and the results showed that knockdown of SIRT7 resulted in increase in H3K122succ levels in HCT116 and HeLa cells, with no obvious changes in H3K18ac, H3K122ac, and H2BK120succ levels ( Supplementary Fig. 4b), supporting our observations in MCF-7 cells. Similar results were obtained in SIRT7 knockout (KO) U2OS cells ( Supplementary Fig. 4c). In addition, clonogenic survival assays were also performed in control or SIRT7-depleted HCT116, HepG2 cells, or SIRT7 KO U2OS cells in response to IR. The results showed that either SIRT7 knockdown or KO in above cell lines significantly compromised cell survival upon IR treatment (Fig. 6d, 6e, and 6g), consistent with the observations that we described in previous version of the manuscript. Moreover, the recruitment of SIRT7 at DSB sites (Fig. 3) and the MNase sensitivity assays (Fig. 5) were performed in U2OS cells, and the results are in agreement with our arguments.
2. Although the mass spectrometry assays seem solid, the biochemistry to support these results is not convincing. For instance, most of the westerns presented exhibit modest changes in the putative succinylation sites. It is well established that histone modifications are difficult to quantify in westerns, given intrinsic variability in signal. All these westerns should be performed at least in triplicates, and densitometry should be performed in order to provide a quantification of these experiments. As it stands, the data does not look convincing.
Particularly, Fig. 1D, it seems the line for the Pan-succ antibody has a bubble exactly for the 3. In this context, the western blot in Fig. 1H is worrisome. The authors claim that only NAM treatment increased H3K122succ levels, yet it seems that also TSA and Sodium butyrate increased its levels, a result that goes against their hypothesis. Of concern, treatment with TSA should majorly increase pan-acetylation, yet the western showed only a modest effect (despite the authors claiming "marked increase"). Such results raise concerns on the solidity of the whole western blot.
Authors: To address the reviewer's concerns, we have repeated the experiments in Fig. 1h. The results showed that while H3K122succ levels increased slightly under TSA and sodium butyrate treatment, treatment with NAM was associated with a much more evident increase in H3K122succ level ( Fig. 1h and Supplementary Fig. 4d). We have optimized the experiments concerning the effect of TSA treatment on pan-acetylation and observed a robust increase in pan-acetylation upon TSA treatment ( Supplementary Fig. 4d). The effect of TSA and Sodium butyrate on H3K122succ levels might be indirect, as it is documented that various histone modifications interplay and are interdependent 1-3 . Fig. 1E is not convincing. Has this antibody been tested in IF assays before.

The IF shown in
The authors should include controls to verify the antibody truly recognize histone succinylation. 6. The data in Fig. 3, indicating recruitment of SIRT7 to sites of DSBs is compelling.
Authors: We appreciate the reviewer for this comment. 7. The experiments shown in Fig.4 demonstrating a role for SIRT7 in both HR and NEHJ DNA repair seem robust. It will be reassuring to see the original FACs plots for at least some of these experiments.
Authors: Original FACs plots for both HR and NHEJ DNA repair upon SIRT7 knockdown in Fig. 4a and Fig. 4b have been provided in Supplementary Fig. 8a-8b. 8. However, in this context, how long after ISceI transfection was the experiment in figure 4J done? It is very difficult to assess any kinetics in this setting since this system is based on cutrepair cycles in the presence of the ISceI enzyme.
Authors: In DR-GFP-U2OS cells used in our cell flow cytometry and ChIP assays, the reporter, DR-GFP, contains a GFP-encoding cDNA that has an endogenous BcgI restriction site replaced by an I-SceI restriction site, thereby rendering it non-functional. A DSB is induced by transfection of cells with a plasmid that encodes I-SceI enzyme to cut the I-SceI site. An incomplete GFP sequence is located downstream and can serve as a donor for intrachromosomal homologous recombination. Repair by HR restores a functional GFP cDNA with the original BcgI site. So after HR repair, the I-SceI restriction site was replaced by BcgI restriction site 4-6 . Therefore, there're no cut-repair cycles in this DR-GFP systems. Considering that the recruitment of SIRT7 to DSB sites is transient, we tested the recruitment of SIRT7 to DSB sites at different time points after I-SceI transfection by using ChIP assay before the experiments described in Fig.4d and 4j. Briefly, DR-GFP-U2OS cells transfected with I-SceI were collected at different time points and subjected to qChIP analysis with antibodies against SIRT7. The results showed that SIRT7 had maximal recruitment to DSB sites at 40 h after transfection in our experiment (Supplementary Fig. 9). So we chose this time point for subsequent qChIP experiments in Fig.4d and 4j.  Supplementary Fig. 12a), suggesting that H3K122E mutation resulted in a loose chromatin structure; 2) MNase experiments in control or SIRT7-depleted U2OS cells using different amount of MNase upon IR treatment showed that SIRT7 depletion significantly increased MNase sensitivity of chromatin ( Supplementary Fig. 13a-b); 3) To further confirm the effect of H3K122succ on chromatin condensation, nucleosome stability assay was performed to analyze the difference of histone-DNA interactions between H3wt-, H3K122E-, and H3K122R-stably expressing U2OS cells. The results showed that H3K122E (H3K122succ mimic) was associated with a reduced histone-DNA interaction while H3K122R (succinylation resistant mutant) was associated with a stabilized nucleosomes (Supplementary Nucleosome stability assays showed that NaCl solubility increased for γH2AX, H2AX, and H3 upon DNA damage, and SIRT7 depletion further enhanced NaCl solubility of histones at 0.25, 1, 4, and 8 h after IR and delayed their recovery ( Supplementary Fig. 13c), consistent with the results of MNase assay in Fig. 5b. These observations altogether point to a role of SIRT7mediated H3K122succ desuccinylation in chromatin compaction.
Regarding to the question in Fig. 4i and Fig. 5b, it has been reported that chromatin is hypersensitive to nuclease digestion following exposure to IR 7-9 . The extent of the increased sensitivity to nuclease digestion indicates that the decrease in nucleosome compaction in response to DSBs affects a significant fraction of the chromatin 10 . This is consistent with our observations from MNase experiments (Fig. 5b, and Supplementary Fig. 13a)  10. Assuming that SIRT7-dependent desuccinylation indeed enhances chromatin compaction, how such compaction will benefit DNA repair? One would think that increased compaction may decrease recruitment of repair factors, in turn negatively influencing DNA repair, which is the opposite to what the authors claim.
Authors: It has been reported that chromatin condensation is a transient but integral part of DNA damage response, and that condensed chromatin enhances upstream signaling, thus promotes DNA repair 11 , whereas persistent condensation inhibits downstream repair and recovery 11 , which is consistent with our results in Fig. 5c that both persistent succinylation (H3K122E) and persistent non-succinylation (H3K122R) impaired HR and NHEJ repair efficiency. In addition, it has been proposed that a repressive chromatin domain is required for inhibiting local transcription, compacting local chromatin structure, and rewriting the local epigenetic landscape during the priming state of DNA repair 16 . Thus, linking SIRT7-mediated transient H3K122succ desuccinylation at DSB site to early stage of DDR is consistent with these studies.
11. In Fig. 5C, why do the control "H3wt" cells exhibit 15% GFP cells, while in the previous   Fig. 5c and Supplementary Fig. 14a-b). Figure 6 B-D: the DNA damage sensitivity assays show a very modest effect (please note the scale is set up to enhance the difference. Survival assays should be plotted in log scale). P-values should be indicated to determine whether the difference claimed is significant.

In
Can these experiments be repeated in other lines?
Authors: Survival curves from MTS assays have been plotted in log scale and p-values have been added (Figures to reviewer). Considering that MTS assays were performed at 72 h after VP16 or IR treatment, a time might be too short to show the effect of IR and VP16 on cell survival, we thus used clonogenic survival assays to determine the effect of IR on cells survival as reported in other studies16-19, in which control or SIRT7-depleted MCF-7, U2OS cells and U2OS cells stably expressing H3wt, H3K122E or H3K122R were treated with or without IR at the indicated doses and then subjected to clonogenic survival assays after 10 days (Fig. 6b-c,   Fig. 6f). Similar experiments were also repeated in HCT116, HepG2 and SIRT7 knockout U2OS cells (Fig. 6d-e, Fig. 6g).
Minor concerns: 1. Few grammatical errors should be corrected. For instance, on page 4 "in last decade" should be "in the last decade". On page 20, "a residue could not be succinylated" should be "a residue that could not be succinylated".
Authors: We appreciate the reviewer and have corrected the writings.

RE: NCOMMS-16-00869
Title: SIRT7 Is a Histone Desuccinylase that Functionally Links to Chromatin Compaction and Genome Stability Response to reviewers' comments-Reviewer #2: Major concerns: 1. In Figure 4H and I, SIRT7 RNAi seems to increase baseline γ-H2AX but does not increase γ-H2AX upon DNA damage (in fact, in 4i, IR-induced γ-H2AX is decreased by SIRT7 RNAi).
By contrast, the effects on H2K122 succinylation occur only in response to DNA damage. This suggests that the effects on DNA damage response might not be directly due to the desuccinylation by SIRT7. This seems problematic for the author's model.
Authors: γH2AX was originally identified as an early event after the direct formation of DNA double-strand breaks (DSBs) by ionizing radiation 20 and now the generation of γH2AX is also believed to occur in association with secondarily formed DSBs by endogenous genotoxin or cellular processing such as reactive oxygen species (ROS) produced during normal cell metabolism, DNA replication and repair at the site of the initial damage, including DNA adducts, crosslinks, and UV-induced photolesions 21 . It is estimated that in normal human cells approximately 1% of single-strand lesions are converted to approximately 50 endogenous DSBs per cell per cell cycle 22 . Thus, SIRT7 might play a role in response to these damages, a possible reason why SIRT7 RNAi seems to increase baseline γH2AX (Fig. 4h-i and Supplementary Fig. 10a-c). Since the occurrence of endogenous DSB is much lower than that of the DSBs induced by the DNA damage reagents or IR treatment, the changes of desuccinylation of H3K122succ catalyzed by SIRT7 might be too weak to be detected during endogenous DSB repair, a possible explanation for the deviation between the changes of H3K122 succinylation and γH2AX during endogenous DSB repair. In responding to the reviewer's concerns, all of the experiments in Fig. 4h and Fig. 4i have been repeated at least in triplicate, and the results showed that SIRT7 knockdown was associated with γH2AX level increment in response to CPT and IR treatment ( Fig. 4h and Supplementary Fig. 10a-c).
This suggests that the level of chromatin compaction doesn't correlate with the effects on repair/DNA damage sensitivity. How do the authors account for this? Authors: Our data showed the recruitment of SIRT7 to DNA damage sites is a transient process, and that SIRT7 mediated transient H3K122succ desuccinylation at DSB site in the early stage of DDR promotes DNA repair by promoting chromatin compaction. Though little is known about the mechanisms of chromatin condensation during DNA repair, it is believed that chromatin condensation is a transient but integral part of the DNA damage response; condensed chromatin enhances upstream signaling, whereas persistent condensation inhibits downstream repair and recovery 11 . In Fig. 5c and 6f, stably expressed H3K122E and H3K122R mimic persistent succinylation and persistent non-succinylation of H3K122, respectively, but they could not mimic the dynamics of SIRT7-mediated transient H3K122succ desuccinylation.
Thus, the expression of H3K122E succinyl mimic as well as H3K122R succinylation-resistent mutant reduces NHEJ, HR, and increases DNA damage sensitivity (Fig. 5c and Fig. 6f), by contrast to the results of the MNsae assays ( Fig. 5a and Supplementary Fig. 12a) in which H3K122E increases chromatin accessibility whereas H3K122R decreases its accessibility. versus the DNA repair proteins reported in this study. How do the authors account for this discrepancy? Did they detect any of the proteins from the previous study? Given the frequency with which the Kus and Parp are found in Flag-co-IPs, there is concern the interactions are not physiologic. The data should be confirmed by co-IP of the endogenous proteins.

To really prove that SIRT7 directly desuccinylates H3K122-succ in vitro, the authors should
show some mass spec analysis in addition to westerns.
Authors: in vitro assays with calf thymus histones and purified SIRT7wt in the presence of NAD + were analyzed by mass spec. The results show that SIRT7 had strong H3K122succ desuccinylase activity, whereas the level of H2BK120succ, H3K122ac, and H3K18ac had no detectable change ( Supplementary Fig. 6a-6c).

Did the authors test whether SIRT7 deacetylates H3K122Ac?
Authors: The deacetylation activity for H3K122ac by SIRT7 was determined by following methods: 1) SIRT7 was overexpressed in HEK293T cells or knocked down in MCF-7, HCT116, HeLa cells or knockout in U2OS cells. Histones were extracted and H3K122ac were analyzed by western blotting. The results showed that the levels of H3K122ac were unaffected in these cells, regardless of overexpression or knockdown of SIRT7 ( Fig. 1g and Supplementary Fig. 4a-4c); 2) in vitro desuccinylation assays using SIRT7wt/H187Y and calf thymus histones or mononucleosomes were analyzed for the H3K122ac levels by western blotting and mass spec. The results indicated that SIRT7wt showed no H3K122ac deacetylation activity toward calf thymus histones and mononucleosomes (Fig. 2c-d and Supplementary Fig. 5 and 6).
7. The introductory statement (and other comments in the introduction): "the intrinsic enzymatic activity of this sirtuin protein remains to be investigated and the cellular function of SIRT7 remains to be explored." incorrectly implies that nothing is known about SIRT7 enzymatic activity or cellular function.
Authors: We have modified the relevant statements.
Additional issues: 1. In the quantification of succinylation in Figure 1C, the error bars are very large and it is difficult to get a sense of the fold change in succinylation. Can the data be plotted on a more meaningful scale, and with comparisons to sites that were not significantly altered and negative controls.
Authors: According to the reviewer's suggestion, the mass spec data have been plotted on the fold change of control/knockdown of the experiments in triplicate (Fig.1c), with comparisons to sites that were not significantly altered.
2. The authors do not indicate why the SIRT7 S111A mutation (and the double S111A/H187A mutation) was studied, and they should provide a brief explanation in the text for what these 3. The deacetylase activity of SIRT7 on H3K18 has been shown to be more efficient on nucleosomal histones than free histones from calf thymus (Gil et al, NAR 2013). It is difficult compare the data for de-succinylation of H3K122 in these contexts in Figure 2 quantitatively (more SIRT7 was used in the nucleosome assays), but it seems that activity may be greater on free histones. A more quantitative comparison of the activities on nucleosomes versus CTHs should be straightforward, and could have implications for the contexts in which the desuccinylation actually occurs.
Authors: To comply with the reviewer's requests, in vitro desucinylation assays were carried out with different amounts of purified FLAG-SIRT7wt and 1 μg of calf thymus histones or 1 μg of HeLa cells-derived mononucleosomes in the presence or absence of 1.0 mM NAD + and/or 10 mM NAM. The reaction mixture was analyzed by western blotting and the densities were quantified by Image J. The results showed no obvious substrates preference for SIRT7 between nucleosomes and CTHs ( Supplementary Fig. 5c). Though the deacetylase activity of SIRT7 on H3K18ac has been shown to be more efficient on nucleosomal histones than free histones from calf thymus 23 , succinyl group is different from acetyl group, and H3K122 is located on the lateral surface of the histone globular domain closing to the dyad symmetry axis.
The DNA wrapped around the globular domain might affect the substrate affinity of SIRT7, and hence its desuccinylase activity when mononucleosomes were used as substrates.
4. It would be nice to see if SIRT7 co-localizes with γ-H2AX at global DNA breaks induced by IR or other DSB agents?
Authors: We agree with the reviewer. However, despite extensive efforts, the antibodies against SIRT7 from almost every company did not work for foci stain, although the antibodies worked for stain SIRT7 in nucleolus and laser path. U2OS cells transfected with GFP-SIRT7 were also used to perform this experiment, but the cells with GFP-SIRT7 overexpression had too high background in nucleus to detect the foci.  Fig. 7a-7b).
6. The authors should discuss how they envision chromatin compaction in response to desuccinylation might promote DNA repair? This is opposite of the model that decreased compaction can create accessibility for repair factors.
Authors: As we mentioned early, although little is known about the mechanisms of chromatin condensation during DNA repair process, it has been shown that chromatin condensation is a transient but integral part of the DNA damage response, and that condensed chromatin enhances upstream signaling, thus promote DNA repair, whereas persistent condensation inhibits downstream repair and recovery 11 , which is consistent with our results in Fig. 5c that both persistent succinylation (H3K122E) and persistent non-succinylation (H3K122R) impaired HR and NHEJ repair efficiency. In addition, it is proposed that that the requirement of repressive chromatin might be related with its' roles in inhibiting local transcription, compacting the local chromatin structure, and rewriting the local epigenetic landscape, stabilizing open chromatin structures and limiting DSB mobility during the initial moments following DSB production 16  1. Desuccinylase activity of sirt7: In the in vitro enzymatic assay, sirt7 was shown to have desuccinylase activity toward H3K122succ peptide while sirt5 does not. Considering the wellstudied desuccinylase activity of sirt5, this should be clarified. Is it possible that sirt5 and sirt7 having specificities toward different succinylation sites? Is H3K122 the only site that sirt7 has activity in vitro? The authors are suggested to study more substrates to strengthen their point.
Based on the reviewer's suggestions, in vitro desucinylation assays were carried out with calf thymus histones and purified SIRT7wt in the presence of NAD + . Mass spec analysis revealed that besides H3K122succ, SIRT7 also desuccinylates H3K56succ and H3K79succ ( Supplementary Fig. 6). As there are no available antibodies against these two modifications, we could not verify these desuccinylase activities by western blotting at present.
2. H3K122suc in DSB repair: Sirt7 was shown to be recruited to DSB sites in a Parp1 dependent way. Is H3K122suc level decreased at the DSB sites where sirt7 was recruited? If yes, is the decrease of H3K122suc Parp1 dependent? The authors only studied H3K122suc level by western blot, which may not be specific to the DSB sites. Laser microirradiation and immunofluorescent assay using anti-H3K122suc or pan anti-Ksuc antibodies should be considered.
Authors: To comply with the reviewer's requests, laser microirradiation and immunofluorescent assays have been performed using anti-H3K122suc antibody. Briefly, we analyzed H3K122succ level at DSB sites in control and SIRT7-depleted U2OS cells. The results showed that, in control U2OS cells, H3K122succ level at DSB sites decreased 5 min after microirradiation compared to non-microirradiated regions, while in SIRT7-depleted U2OS cells, H3K122succ level at DSB sites did not change (Supplementary Fig. 11a). Furthermore, we analyzed H3K122succ level at DSB sites in U2OS cells transfected with siControl or siPARP1. The results showed that, in siControl cells, H3K122succ level at DSB sites decreased 5 min after microirradiation compared to non-microirradiated regions, while in siPARP1 cells, H3K122succ level at DSB sites had no detectable change (Supplementary Fig.   11b). These results suggest that the decrease of H3K122succ level is SIRT7-and PARP1dependent and specific to DSB sites.
3. H3K122suc vs H3K122ac: The authors showed that desuccinylase activity of sirt7 was indispensable for DSB repair. Does sirt7 have other activities such as deacetylase in addition to desuccinylase activity? Care should be taken that mutation of H3K122 to E/R has influence not only on succinylation but also on acetylation. The authors should study whether H3K122ac level is changed during DNA damage insult and whether sirt7 has deacetylase activity in vitro and in cell lines.
Authors: To address the reviewer's concerns, the level of H3K122ac was detected upon SIRT7 overexpression in HEK293T cells or knockdown in MCF-7, HCT116, HeLa cells or knockout in U2OS cells. The results showed that the levels of H3K122ac were unaffected in these cells, regardless of overexpression or knockdown, or even knockout of SIRT7 ( Fig. 1g and Supplementary Fig. 4). In addition, in vitro desuccinylation assays were carried out using SIRT7wt/H187Y and calf thymus histones or mononucleosomes, and the H3K122ac levels were analyzed by western blotting and mass spec. The results indicated that SIRT7wt showed no detectable H3K122ac deacetylation activity toward calf thymus histones and mononucleosomes (Fig. 2b-c and Supplementary Fig. 5 and 6). Moreover, H3K122ac levels were also determined in MCF -7 and U2OS cells treated with or without CPT, VP16, or IR, the results revealed that H3K122ac levels showed no changes upon genotoxic insults (Fig. 4g-i and Supplementary Fig. 10a-c).
Minor points: 1. In Figure1E, the authors are suggested to study immunofluorescence by over expression (OE) of Sirt7, showing sirt7 and lysine succinylation levels in the same cell, in a higher resolution picture.
Authors: To follow the reviewer's suggestion, immunofluorescence assays were performed by overexpressing GFP-SIRT7 in U2OS cells and staining with pan-lysine succinylation polyclonal antibodies. The results showed that, compared to cells without overexpression, panlysine succinylation levels were decreased in U2OS cells with overexpression of GFP-SIRT7 ( Supplementary Fig. 2b) 2. In Figure 1D, the first panel indicates an uneven exposure for pan-succ signals, especially between sirt7wt and sirt7s111a.
Authors: The experiment has been repeated and figure has been replaced ( Fig. 1d and Supplementary Fig. 2a). Figure 1F, the middle panel, the blot signals are too weak to demonstrate the antibody specificity clearly.

In
Authors: The experiments in Fig. 1f have been repeated and the figures have been replaced in the revised manuscript. In addition, the specificity of anti-H3K122succ and anti-H2BK120succ antibodies were further characterized by dot blotting and western blotting competed with succinyl peptides, unmodified control peptides, or malonyl peptides. The results showed that only H3K122succ peptide blocked the binding of anti-H3K122succ and only H2BK120succ peptide blocked the binding of anti-H2BK120succ ( Supplementary Fig. 3).
4. On page 12 of manuscript, the authors described "treatment with TSA or sodium butyrate had no effect on the level of H3K122succ, although treatment with NAM, TSA, or sodium butyrate all resulted in a marked increase in the level of H3 pan-acetylation ( Figure 1H)". However, in Figure 1H, H3K122suc level was increased by either HDAC inhibitor treatment.
Authors: We found that, under certain conditions, H3K122succ levels alters, although modestly, under TSA or sodium butyrate treatment. We speculate that the effect of TSA and sodium butyrate on H3K122succ levels might be indirect, as it is well documented that various histone modifications interplay and are interdependent 1-3 . Figure 4C, there is no labeling about which one is "NHEJ" or "HR". were treated with IR and subjected to MTS assays for cell survival. Error bars represent the mean ± SD of triplicate experiments. (d) U2OS cells stably expressing H3wt, H3K122E or H3K122R were treated with IR and subjected to MTS assays for cell survival. Error bars represent mean ± SD of triplicate experiments. *p < 0.05 and **p < 0.01 (two-tailed unpaired Student's t test).