PMC:6172693 / 5417-33348 JSONTXT

{"project":"2_test","denotations":[{"id":"30257210-27723750-20058690","span":{"begin":382,"end":386},"obj":"27723750"},{"id":"30257210-27723750-20058692","span":{"begin":606,"end":610},"obj":"27723750"},{"id":"30257210-28650317-20058694","span":{"begin":872,"end":876},"obj":"28650317"},{"id":"30257210-29480802-20058696","span":{"begin":2403,"end":2407},"obj":"29480802"},{"id":"30257210-29480802-20058699","span":{"begin":3658,"end":3662},"obj":"29480802"},{"id":"30257210-27723750-20058700","span":{"begin":5008,"end":5012},"obj":"27723750"},{"id":"30257210-27723750-20058702","span":{"begin":5409,"end":5413},"obj":"27723750"},{"id":"30257210-27067600-20058703","span":{"begin":6172,"end":6176},"obj":"27067600"},{"id":"30257210-27067600-20058704","span":{"begin":8486,"end":8490},"obj":"27067600"},{"id":"30257210-22847107-20058705","span":{"begin":9147,"end":9151},"obj":"22847107"},{"id":"30257210-28035797-20058706","span":{"begin":9191,"end":9195},"obj":"28035797"},{"id":"30257210-27686526-20058707","span":{"begin":9270,"end":9274},"obj":"27686526"},{"id":"30257210-27471034-20058709","span":{"begin":10555,"end":10559},"obj":"27471034"},{"id":"30257210-29054115-20058710","span":{"begin":10578,"end":10582},"obj":"29054115"},{"id":"30257210-25688442-20058711","span":{"begin":11142,"end":11146},"obj":"25688442"},{"id":"30257210-30013063-20058712","span":{"begin":11256,"end":11260},"obj":"30013063"},{"id":"30257210-27471034-20058713","span":{"begin":11910,"end":11914},"obj":"27471034"},{"id":"30257210-29054115-20058714","span":{"begin":11933,"end":11937},"obj":"29054115"},{"id":"30257210-23940115-20058715","span":{"begin":15092,"end":15096},"obj":"23940115"},{"id":"30257210-23485013-20058716","span":{"begin":15113,"end":15117},"obj":"23485013"},{"id":"30257210-12887903-20058717","span":{"begin":15133,"end":15137},"obj":"12887903"},{"id":"30257210-17984964-20058718","span":{"begin":20175,"end":20179},"obj":"17984964"},{"id":"30257210-27723750-20058719","span":{"begin":21806,"end":21810},"obj":"27723750"},{"id":"30257210-27723750-20058720","span":{"begin":23224,"end":23228},"obj":"27723750"},{"id":"30257210-27723750-20058721","span":{"begin":23438,"end":23442},"obj":"27723750"},{"id":"30257210-27723750-20058722","span":{"begin":23897,"end":23901},"obj":"27723750"},{"id":"30257210-27723750-20058723","span":{"begin":24293,"end":24297},"obj":"27723750"},{"id":"30257210-19407812-20058724","span":{"begin":25635,"end":25639},"obj":"19407812"},{"id":"30257210-28650317-20058725","span":{"begin":27170,"end":27174},"obj":"28650317"},{"id":"30257210-29480802-20058726","span":{"begin":27192,"end":27196},"obj":"29480802"},{"id":"30257210-27067600-20058784","span":{"begin":8486,"end":8490},"obj":"27067600"}],"text":"Results\n\nFactors Influencing Ser-ADPr of Histone Peptides\nTo analyze the substrate properties that influence Ser-ADPr, we investigated sequence features that may affect the efficiency of in vitro histone peptide ADPr reactions. Our previous proteomics data provided a short consensus motif for in vivo Ser-ADPr with either Lys or Arg N-terminal to the target Ser (Leidecker et al., 2016, Bonfiglio et al., 2017b). Based on these observations, we incubated PARP1 and HPF1 with a variety of histone peptides, each containing an Lys-Ser (KS) motif known to be the modification site in vivo (Leidecker et al., 2016). Similar to what we reported before (Bonfiglio et al., 2017b), we observed that two different histone H3 peptides as well as H2A and H4 peptides were modified by the HPF1/PARP1 complex in vitro (Figure 1A). The Ser-ADPr glycosylhydrolase ARH3 (Fontana et al., 2017) was able to efficiently remove the ADP-ribose on all of the analyzed peptides (Figure 1A). We also compared the efficiency of H3 peptide 1–20 modification to that of the H3/H4 tetramer and the whole nucleosome. As shown in Figure S1A, peptide modification is not dramatically lower, especially considering the additional ADPr sites on the histone proteins and that this H3 peptide is mostly mono-ADPr in vitro (Bonfiglio et al., 2017b). These experiments establish that KS motifs in a variety of histone peptides can be modified efficiently and reversibly, demonstrating the utility of the histone peptide as a tractable in vitro assay for histone Ser-ADPr.\nFigure 1 Modifiers of Serine-ADP-Ribosylation of Histone Peptides\n(A) Autoradiogram showing ADPr, and subsequent ARH3-mediated glycohydrolysis of H3 1–20aa, H3 27–45aa, H2A 1–17aa, and H4 1–23aa peptides. Coomassie staining of the SDS-PAGE is included and represents the loading control.\n(B) Autoradiogram showing PARP1/2 + HPF1-mediated ADPr of H3 peptide with Lys9 substituted by Ala and Arg, and Ser10 substituted by Ala. Coomassie staining of the SDS-PAGE is included.\n(C) 293T cells were transfected with the same amount of empty vector (EV) or plasmid expressing WT, K9A, K9R, or S10A FLAG-tagged histone H3 protein and treated for 10 min with H2O2. Inputs (A) and FLAG-IPs (B) were analyzed by western blotting.\nNext, we opted to focus on H3 Ser10 (H3S10) ADPr, because this site was previously shown to be the primary ADPr site on H3 in vivo (Palazzo et al., 2018). We investigated how alterations of the key KS residues affect the modification profile of the H3 histone peptide in vitro. Based on our previous finding that both PARP1 and PARP2 can modify this H3 peptide in the presence of HPF1 (Bonfiglio et al., 2017b), we examined both PARPs with variations on the KS motif. Substitution of Ser10 with alanine (Ala) led to a complete loss of the modification (Figure 1B), as we have previously shown (Bonfiglio et al., 2017b). Changing the neighboring Lys residue into Arg or Ala had varying effects on histone Ser-ADPr. The H3 peptide containing the K9R mutation was still modified, albeit to a lesser extent than wild-type (WT) peptide. In contrast, the H3K9A mutation strongly (but not completely) inhibited histone H3 Ser-ADPr (Figure 1B), highlighting the importance of a basic residue preceding the Ser. Both PARP1 and PARP2 modified the peptide panel with similar profiles, although PARP1 catalyzed the reactions more efficiently under the conditions used.\nWe further confirmed the importance of the consensus KS motif for Ser-ADPr in vivo. We transfected 293T cells with FLAG-tagged histone H3 WT, K9A, K9R, K9Q, or S10A mutant H3 and assessed ADPr efficiency, as described previously (Palazzo et al., 2018). DNA damage was induced by the treatment with 2 mM hydrogen peroxide (H2O2), followed by FLAG-immunoprecipitation (FLAG-IP). Western blotting was performed using a pan-ADPr reagent that recognizes all forms of cellular ADPr (Figures 1C and S1B). The ADPr patterns obtained were similar to those observed in our in vitro reactions. To note, by using a specific anti-H3K9ac antibody, we show that the KS motif is also important for K9 acetylation in vivo (Figure 1C, FLAG-IP).\nThese data extend our previous findings that the KS and RS motifs are preferred targets for Ser-ADPr and exclude the possibility that Lys rather than Ser is the modification target.\n\nDiscovery of Tyrosine as a Target Residue for ADPr\nADPr of Ser led us to question whether a hydroxyl group is sufficient and necessary to target an amino acid for ADPr when adjacent to Lys. We therefore decided to substitute H3S10 with threonine (Thr) and tyrosine (Tyr), the two other residues that contain hydroxyl groups, and additionally Glu and Asp as further controls. Not only were we unable to detect ADPr on Glu and Asp but also on Thr residues (Figure 2A). This suggests that although chemically similar to Ser, the additional methyl group on Thr interferes with the ADPr reaction mediated by PARP1/HPF1. In fact, in none of our previous proteomic analyses (Leidecker et al., 2016, Bonfiglio et al., 2017b) were we able to detect Thr-ADPr. Conversely, we identified a reproducible modification of Tyr when we introduced this amino acid instead of Ser10 (Figure 2A). Because Tyr has not previously been described as a substrate for ADPr, we sought mass spectrometric evidence for Tyr-ADPr. Although we could not detect Tyr-ADPr in our histone proteomics data (Leidecker et al., 2016), we confidently identified Tyr-ADPr of HPF1 in an in vitro reaction containing PARP1 (Figures 2B and S2B). We could also identify Ser97 in HPF1 as another site modified in this reaction (Figure 2C). These data suggested that PARP1 was the enzyme responsible for HPF1 Tyr-ADPr modification. To follow up on this point, we modified recombinant HPF1 using a panel of different PARPs and radioactively labeled NAD. We could observe a low but reproducible modification by PARP1 and possibly by PARP2 (Figures 2D, S2A, and S2E). This modification is at least partly dependent on the assembly of the PARP1/HPF1 complex, because the modification of the HPF1 R239A mutant protein (previously shown to be deficient in interacting with PARP; see Gibbs-Seymour et al., 2016) was significantly reduced (Figure 2E).\nFigure 2 Discovery of Tyrosine as a Target Residue for ADPr\n(A) Autoradiogram showing ADPr of H3 peptide (1–20aa) with Ser10 substituted by Ala, Thr, Tyr, Glu, and Asp, alongside Lys9 substituted by Arg and Ala. Coomassie staining of the SDS-PAGE is included.\n(B) High-resolution ETD fragmentation spectrum of an HPF1 peptide modified by ADP-ribose on tyrosine 238. The chemical structure of ADP-ribose is depicted (see also Figure S2B). ∗1, peaks corresponding to co-isolated species in their original charge state. Multiple species in charge states 2–5 passed through the quadrupole and could not be completely deconvoluted.\n(C) High-resolution ETD fragmentation spectrum of an HPF1 peptide modified by ADP-ribose on serine 97. The chemical structure of ADP-ribose is depicted.\n(D) Autoradiogram showing a panel of PARPs incubated with HPF1 protein. Reaction with mono(ADP-ribosyl)ating PARP1 E988Q (EQ) mutant enhances detection of the HPF1 ADPr. Coomassie staining of the SDS-PAGE is included.\n(E) Autoradiogram showing PARP1 E988Q-mediated ADPr of HPF1 WT, HPF1 R239A, and GST-HPF1 proteins. Coomassie staining of the SDS-PAGE is included.\n(F) 293T cells were transfected with the same amount of EV or plasmid expressing WT, S97A, or Y238A FLAG-tagged HPF1 protein and left untreated or treated for 10 or 120 min with H2O2. Inputs and FLAG-IPs were analyzed by western blotting. CMV, cytomegalovirus.\n(G) 293T parental or PARP1 KO cells were transfected with the same amount of EV or plasmid expressing WT FLAG-tagged HPF1 protein and left untreated or treated for 10 or 120 min with H2O2. Inputs and FLAG-IPs were analyzed by western blotting.\nTo confirm the ADPr of HPF1 in vivo, we overexpressed and immunoprecipitated FLAG-tagged HPF1 WT, S97A, and Y238A mutant proteins from 293T cells, as was described above for histone H3. We observed that HPF1 is significantly modified in cells even in undamaged conditions (Figure 2F). We did not detect a major effect of the S97A mutation on the modification of HPF1. However, mutation of the Tyr238 site to Ala had a profound effect on the HPF1 ADPr signal (Figure 2F). This defect may be at least partly due to a reduced ability of the Y238A mutant to interact with PARP1 and to stimulate ADPr (Gibbs-Seymour et al., 2016). To further prove that HPF1 ADPr is dependent on PARP1, we performed FLAG-IP in PARP1 knockout (KO) 293T cells. As can be seen in Figure 2G, HPF1 ADPr is largely missing in PARP1 KO cells. It is likely that the remaining HPF1 modification is due to PARP2 activity. ADPribosylation of Tyr238 is not essential for the global HPF1-dependent ADPr of histones because the non-modifiable Y238F HPF1 mutant supports this activity both in cells and in vitro (Figures S3A and S3B).\nWhile it appears that there may be multiple ADPr sites on HPF1, we were able to confirm the ADPr of Y238 on HPF1 in cell extracts by ADPr mapping through reprocessing (Matic et al., 2012) of a published dataset (Bilan et al., 2017) (Figure S2D). Reanalysis of a large-scale ADPr dataset (Martello et al., 2016) revealed four additional high-certainty Tyr-ADPr target proteins (Figures S2C and S2E–S2G). Although the type of mass spectrometric analysis used to generate the latter dataset is suboptimal (for additional information about the inadequacies of the higher-energy collisional dissociation [HCD] technology for ADPr site mapping, please refer to Bonfiglio et al., 2017a), our discovery of a Tyr-ADPr diagnostic peak (Figure S2C) enhances the confidence of Tyr-ADPr site mapping.\n\nCanonical H3 Histone Marks Reduce the Efficiency of H3S10ADPr on H3 Peptide\nWe observed that removal of the positively charged Lys through the synthesis of an H3 peptide containing an Ala in position 9 instead of an Lys almost completely abolished Ser-ADPr (Figures 1B and 1C). It is known that acetylation neutralizes the positive charge of Lys residues, whereas methylation maintains the charge. Thus, the presence of this frequently modified residue in our consensus motif led us to hypothesize that modifications of the Lys preceding the Ser may have different effects on Ser-ADPr, a potential mechanism of interplay between the known histone modifications in the H3S10 environment and H3S10ADPr. Additionally, the recent evidence of PARPs conjugating ADPr to phosphorylated DNA (Talhaoui et al., 2016, Munnur and Ahel, 2017) raised the intriguing possibility of PARPs ADPr a phosphorylated peptide—H3S10ph in this case. Because these endogenous histone PTMs (histone marks) are highly dynamic in cells and organisms, an interplay is likely to have important biological consequences. By examining “marked” histone peptides in vitro, we can generate “snapshots” of this dynamic interplay.\nWe therefore set out to investigate the effect on H3S10ADPr of the histone mark environment around H3S10, which is a particularly PTM-rich and biologically important histone region (Huang et al., 2015). H3K9ac severely inhibits histone H3S10ADPr (Figure 3A), as also shown in a recent report (Liszczak et al., 2018), and reversal of Lys9 acetylation by using deacetylase enzymes (HDAC2, SIRT2) re-established this peptide as a substrate for Ser-ADPr by PARP1/HPF1 (Figure S4A). In comparison, H3K9me1 causes only a very mild reduction of H3S10ADPr levels compared to the unmodified peptide. Phosphorylation of the target residue, Ser10, completely blocked ADPr of the peptide, confirming that ADPr and phosphorylation of the Ser10 site are mutually exclusive (Figure 3A). In agreement with this, we did not find any mass spectrometric evidence for ADPr of a phosphorylated Ser. This indicates that PARP1-mediated ADPr of DNA on a phosphate group (Talhaoui et al., 2016, Munnur and Ahel, 2017) is mechanistically different from HPF1-dependent ADPr by PARP1 on protein substrates.\nFigure 3 Canonical H3 Histone Marks Reduce the Efficiency of H3S10ADPr on H3 Peptide\n(A) Autoradiogram showing PARP1/2 + HPF1-mediated ADPr of H3 peptide with WT, K9ac, K9me1, and S10ph modifications. Coomassie staining of the SDS-PAGE is included.\n(B) As in (A), except PARP1 and HPF1 only, with H3 (1–20aa) WT, K9ac, K9me1, K9me2, and K9me3 peptides. Coomassie staining of the SDS-PAGE is included.\n(C) As in (B), except with H3 (1–20aa) WT, K4ac, K4me3, K9ac, K9me1, K9me3, S10ph, K14ac, K18ac, and K18me3 peptides. Coomassie staining of the SDS-PAGE is included.\n(D) As in (B), except with H3 (21–44aa) WT, K27ac, K27me1, K27me2, K27me3, and WT peptides. Coomassie staining of the SDS-PAGE is included.\nGiven that H3K9me1 did not notably compromise S10ADPr levels, we analyzed whether dimethylations or trimethylations, both commonly observed in the histone code, would have a greater impact on the reaction. We noted a stepwise decrease in H3S10ADPr levels on H3K9me1, H3K9me2, and H3K9me3 substrates, with H3K9me3 permitting only a very modest degree of PARP1/HPF1-dependent H3S10ADPr (Figure 3B). Because H3K9me and H3S10ADPr modifications could coexist on the H3 peptide, we investigated whether the recently identified enzyme that removes Ser-ADPr, ARH3, could still access and remove H3S10ADPr in the presence of H3K9me. Our analysis showed that ARH3 was still active against H3S10ADPr, irrespective of the H3K9me marks, and could efficiently erase H3S10ADPr signals from modified H3 peptides (Figure S4B).\nTo investigate the effect of known histone marks in a wider context, we broadened the scope of our analysis of residues surrounding H3S10ADPr by testing H3K4ac, H3K4me3, H3K14ac, H3K18ac, and H3K18me3 peptides. Of these additional histone marks, only H3K14ac notably affected the subsequent addition of ADPr to H3S10 (Figure 3C).\nBecause our earlier experiments had determined that H4 1–23 and H3 27–45 peptides were suitable for PARP1/HPF1-dependent Ser-ADPr modification, we tested both for crosstalk between nearby acetylation and methylation modifications with H4S1ADPr and H3S28ADPr. We found that the modification of the KS motif at S28 has effects similar to those seen for Ser10 (Figure 3D). Alternatively, H3K36me1, H3K36me2, or H3K36me3 did not reduce the H3S28ADPr modification signals, while the H3K36ac had only a modest effect (Figure S4C). We also found that none of the H4R3me2, H4K5ac, or H4K8ac marks had a discernible effect on H4S1ADPr levels compared to the unmodified peptide (Figure S4C).\n\nSer-ADPr on H3S10 Prevents the Efficient Incorporation of H3K9 Acetylation and H3S10 Phosphorylation\nWe conducted reciprocal experiments based on our above findings, this time modifying histone H3 peptide first with PARP1/HPF1 complex (Figure S5), then subsequently incubating the reaction products in acetylation, phosphorylation, and methylation reaction mixtures using the purified catalytic domain of p300, the activated fragment of Aurora B kinase (Baronase; to phosphorylate H3S10) and Dim5 methyltransferase (Nunes Bastos et al., 2013, Moonat et al., 2013, Zhang et al., 2003). We detected the acetylated products of Ser-ADPr H3 peptides using a specific H3K9ac antibody and observed that K9ac is effectively prevented if the peptide is previously ADPr (Figure 4A, lane 5). To control for any possible interference of ADPr with western blot detection, we incubated the Ser-ADPr H3 peptide with p300, stopped the reaction, and removed Ser-ADPr from the peptide using ARH3. This assay showed only a negligible amount of H3K9Ac (Figure 4A, lane 6). In a similar experiment, we saw that Ser-ADPr of H3 peptide prevented subsequent H3S10 phosphorylation (Figure 4B, lane 5). Finally, we incubated an Ser-ADPr H3 peptide in an Lys methylation reaction and found that H3S10ADPr did not preclude the incorporation of H3K9me3, although it did substantially reduce the efficiency of the reaction compared to the unmodified H3 peptide (Figure 4C, lane 3 versus lane 5).\nFigure 4 H3S10ADPr Reduces the Efficiency of Subsequent H3K9 Acetylation and H3S10 Phosphorylation\n(A) Western blot showing PARP1/HPF1 ADPr of H3 (1–20aa) peptide and subsequent p300-mediated acetylation. One reaction was stopped after p300 incubation, then supplemented with ARH3 to remove ADPr before signal detection. Membrane probed with H3K9ac antibody, with H3K9ac peptide included as a positive marker.\n(B) Western blot showing PARP1/HPF1 ADPr of H3 (1–20aa) peptide and subsequent Baronase-mediated phosphorylation. Control sample excludes NAD from the PARP1/HPF1 reaction. Membrane probed with H3S10ph antibody, with H3S10ph peptide included as a positive marker.\n(C) Western blot showing PARP1/HPF1 ADPr of H3 (1–20aa) peptide and subsequent Dim5-mediated methylation. Control sample excludes NAD from the PARP1/HPF1 reaction. Membrane probed with H3K9me3 antibody, with H3K9me3 peptide included as a positive marker.\n\nAn Approach for Rapid and Easy Analysis of ADPr Peptides\nOur approaches above use [32P]NAD as a detection method, but this radioactive technique is expensive and requires strict safety procedures. Furthermore, using [32P]NAD and standard gel electrophoresis only allows the detection of modified product, rather than an analysis of unmodified and modified peptides together (i.e., substrates and products). These limitations, together with the clear importance of studying the interplay of Ser-ADPr and other known histone marks, motivated us to look for a simpler approach that could be implemented in virtually any biological laboratory. Given that ADP-ribose is a nucleotide, we reasoned that an electrophoresis system capable of resolving a one-nucleotide difference in the length of oligonucleotides would allow a clear separation of ADPr and unmodified substrate peptides. However, the negatively charged nucleic acids are separated by migrating toward the positively charged anode. In contrast, the histone tail peptides have a net positive charge, even when modified by ADP-ribose and would therefore migrate in the wrong direction. By changing the polarity of the electrodes, the positively charged substrate peptides can be driven into gels intended for electrophoresis of short nucleic acids and be separated according to their charge. Following ADPr, peptides become less positively charged and therefore migrate more slowly, which allows a clear spatial separation between the bands of the modified and unmodified peptides (Figure 5A). After the run, both species (unmodified and modified) can be clearly visualized and quantified by Coomassie-based staining, which reveals by band shift how much of the starting peptide has been ADPr (Figure 5A). Figure 5B shows an exemplar of this technique, comparing unmodified H3 peptide during a time course with PARP1, HPF1, and H3 peptide, with the modified peptide shifted upward at later time points. Incubating modified H3 peptide with ARH3 reverses the band shift to the unmodified state (Figure 5C). We then expanded this method to investigate ADPr efficiency on H3 peptides with a variety of histone marks. We observed that H3K4me mildly reduced ADPr levels compared to WT, whereas H3R8me peptides were modified efficiently (Figure 5D). H3K9ac, H3K9me, and K14ac modification profiles were comparable to the [32P]NAD experiments, reinforcing the value of this Coomassie-based approach for estimating the efficiency of a reaction. Additionally, we examined an H3T11ph peptide, which showed only a very slight ADPr band, suggesting a strong inhibition of HPF1/PARP1-catalyzed Ser-ADPr by the adjacent phosphorylation (Figure 5D). These combined experiments produced a map of the histone marks within a local region around H3S10 that affect the efficiency of H3S10ADPr (Figure 5E). Notably, histone marks other than ADPr also generated a band shift compared to the unmodified counterpart peptide (Figure 5D, left). This implies that the utility of our approach is not limited to ADPr and that this technique can be used to study the dynamics of other histone marks at the peptide level, such as the interplay between phosphorylation and acetylation (Latham and Dent, 2007).\nFigure 5 A Technique to Rapidly Analyze ADP-Ribosylated Peptides\n(A) Schematic representation of the approach to rapidly and easily analyze the modification status of positively charged histone tail peptides.\n(B) Imperial stained gel showing ADPr of H3 (1–21aa) peptides after addition of PARP1/HPF1 during a 6-hr time course. The upward band shift denotes ADPr of the H3 peptide.\n(C) Imperial stained gel showing ADPr of H3 (1–21aa) peptides after addition of PARP1/HPF1 and subsequent addition of ARH3. The upward band shift denotes ADPr of the H3 peptide.\n(D) Imperial stained gel showing H3 (1–21aa) WT, K4me1, K4me2, K4me3, R8me1, R8me2a, K9ac, K9me1, K9me2, K9me3, S10ph, T11ph, and K14ac peptides and subsequent ADPr following addition of PARP1 and HPF1. The upward band shift denotes ADPr of the H3 peptide.\n(E) A schematic showing a map of histone H3 1–20aa with histone marks that interfere with Ser-ADPr on H3 peptide in vitro.\n\nH3K9ac and S10ADPr Are Mutually Exclusive Histone Marks in Human Cells\nWe sought to assess whether the results generated using histone H3 peptides could be replicated with intact human nucleosomes in vitro. WT and H3K9ac recombinant human mononucleosomes were incubated with PARP1 in the presence and absence of HPF1. We observed a clear contrast between the WT and H3K9ac nucleosomes when incubated with PARP1/HPF1, with WT displaying a higher level of Ser-ADPr (Figure 6A). The H3K9ac nucleosomes were still significantly modified, albeit to a lower degree, presumably due to modifications of other previously observed histone tail sites, such as H3S28 and H2BS6 (Leidecker et al., 2016). Similarly, we performed an assay using a nucleosome substrate to test the reciprocal reactions, namely Ser-ADPr, and then acetylation of the nucleosome (Figure 6B). We saw that prior Ser-ADPr reduced subsequent H3K9 acetylation, as detected by the specific anti-H3K9ac antibody (Figure 6B). These results suggest that the interplay that we observe between Ser-ADPr and acetylation of neighboring Lys residues on the peptide level also occurs in the context of whole nucleosomes and in vivo. We also observed that prior H3S10 phosphorylation of the nucleosome also significantly reduced subsequent p300-mediated acetylation of H3K9 (Figure 6B).\nFigure 6 H3K9ac and S10ADPr Are Mutually Exclusive Histone Marks in Human Cells\n(A) Autoradiogram showing PARP1 mediated ADPr in the presence of absence of HPF1, with either WT or K9ac human recombinant nucleosomes. Coomassie staining of the SDS-PAGE is included.\n(B) Western blot showing PARP1/HPF1 ADPr of recombinant human nucleosome and subsequent p300-mediated acetylation. One reaction includes Baronase incubation instead of ADPr reaction, before p300 acetylation reaction. Membrane probed with H3K9ac antibody, with commercially obtained recombinant human H3K9ac nucleosome included as a positive marker.\n(C) High-resolution ETD fragmentation spectrum of a H3 peptide modified by methyl on lysine 9 and ADP-ribose on serine 10 obtained from Leidecker et al. (2016). The chemical structure of methyl and ADP-ribose are depicted.\n(D) High-resolution ETD fragmentation spectrum of a H3 peptide modified by acetylation on lysine 9 and lysine 14 obtained from Leidecker et al. (2016). The chemical structure of acetylation is depicted. ∗1, Peak corresponding to an unfragmented co-eluting, co-isolated +2 precursor deconvoluted into the +1 state. ∗2, Peak corresponding to an unfragmented co-eluting, co-isolated +3 precursor deconvoluted into the +1 state.\n(E) Schematic summary of canonical histone H3 marks and their interactions with Ser-ADPr based on the mass spectrometry (MS) analysis of U2OS cell extracts from Leidecker et al. (2016). The marks depicted on the top are H3 marks that can coexist with Ser10 or Ser28 ADPr in vivo, while the H3 marks depicted on the bottom are mutually exclusive with ADPr on Ser10 or Ser28.\nTo compare the interplay observed in our in vitro system with that in cells, we analyzed U2OS cell extracts by high-resolution electron-transfer dissociation (ETD) mass spectrometry (Leidecker et al., 2016). We identified H3S10ADPr in the presence of mono-, di-, and trimethylation of H3K9 and with H3K14ac, but never with H3K9ac (Figures 6C and S6A–S6C). Any detection of H3K9ac was in the absence of H3S10ADPr, although we were able to detect H3K9ac in co-existence with marks other than H3S10ADPr, such as H3K14ac (Figure 6D). To test whether our failure to detect H3K9ac and H3S10ADPr together was due to technical limitations, we purified small amounts of H3K9acS10ADPr peptide generated from a highly inefficient reaction and analyzed it by mass spectrometry. We found that we were able to detect both histone marks on the same peptide (Figure S6D), further indicating that the apparent non-coexistence of these marks is due to the mutual exclusivity in vivo rather than our technical inability of detecting doubly modified H3K9ac/H3S10ADPr peptides. Our findings define two groups of histone H3 PTMs that can either coexist with or are mutually exclusive to Ser-ADPr (Figure 6E).\nTo further characterize the interplay between histone Ser-ADPr and other PTMs in vivo, we assessed the levels of H3S10ph, H3K9ac, H3K9me3, and several other PTMs around the H3S10ADPr site in 293T cells following DNA damage (Figure 7A). Our results confirmed previously published data showing reduction of H3K9ac in response to DNA damage (Tjeertes et al., 2009), because we also observed striking specific deacetylation of the H3K9 site after 120 min of treatment (Figure 7A). We also observed significant deacetylation of H3K14 under the same conditions (Figure 7A). DNA damage-induced deacetylation of both H3K9 and H3K14 was completely blocked by pre-treatment with a PARP inhibitor, olaparib. We did not observe DNA damage-induced deacetylation of H3K27ac or K36ac, or demethylation of H3K9me3 or H3K27me3, among others (Figure 7A). As evident from the patterns for the cell-cycle proteins cyclin A, B1, E1, Cdc2 T15P, PRC1 T481P, and p21 (Figure 7A), the cell cycle was unaffected by olaparib treatment in our experimental settings.\nFigure 7 Histone Mark Response to DNA Damage with PARP Inhibition and Persistent Ser-ADPr\n(A) 293T cells were pretreated with DMSO or olaparib and treated with H2O2. Western blotting analysis of the changes in histone H3 K9ac, K9me3, S10P, K14ac, K27ac, K27me3, and K36ac, as well as total pan-Kac histone acetylation and cell-cycle protein levels was performed at the indicated times after the induction of DNA damage.\n(B) U2OS WT and ARH3 KO cells were treated with H2O2. The levels of H3K9ac, H3K9me3, and pan-Kac were examined by western blotting at the indicated time points.\nTo test our hypothesis that Ser-ADPr specifically affects these canonical histone marks, we performed similar experiments in ARH3 KO cells. We have previously demonstrated that these cells have chronically increased histone ADPr, including the Ser10 site (Fontana et al., 2017, Palazzo et al., 2018). DNA damage-induced deacetylation was more robust in these cells, which was especially obvious at 10 and 120 min post-DNA damage (Figure 7B). It is worth mentioning that some other acetylated proteins, detected by pan-acetylation antibody, displayed a different profile of increasing acetylation after DNA damage treatment (Figure 7B). These results combined suggest that interplay between histone ADPr and K9 acetylation and some other forms of histone modifications takes place in living cells. This knowledge can offer a framework for the further investigation of crosstalk between Ser-ADPr and other histone marks, and onward toward a wider understanding of the physiological function of Ser-ADPr as a histone mark and as a PTM.\n"}