Poly(ADP-Ribosyl)ation Affects Histone Acetylation and Transcription

Poly(ADP-ribosyl)ation (PARylation) is a posttranslational protein modification catalyzed by members of the poly(ADP-ribose) polymerase (PARP) enzyme family. PARylation regulates a wide variety of biological processes in most eukaryotic cells including energy metabolism and cell death, maintenance of genomic stability, chromatin structure and transcription. Inside the nucleus, cross-talk between PARylation and other epigenetic modifications, such as DNA and histone methylation, was already described. In the present work, using PJ34 or ABT888 to inhibit PARP activity or over-expressing poly(ADP-ribose) glycohydrolase (PARG), we show decrease of global histone H3 and H4 acetylation. This effect is accompanied by a reduction of the steady state mRNA level of p300, Pcaf, and Tnfα, but not of Dnmt1. Chromatin immunoprecipitation (ChIP) analyses, performed at the level of the Transcription Start Site (TSS) of these four genes, reveal that changes in histone acetylation are specific for each promoter. Finally, we demonstrate an increase of global deacetylase activity in nuclear extracts from cells treated with PJ34, whereas global acetyltransferase activity is not affected, suggesting a role for PARP in the inhibition of histone deacetylases. Taken together, these results show an important link between PARylation and histone acetylation regulated transcription.


Introduction
PARylation is a posttranslational protein modification catalyzed by enzymes belonging to the PARP family. PARPs use NAD + as substrate and, upon cleaving off nicotinamide, they covalently transfer the ADP-ribosyl moiety to suitable acceptor proteins and, subsequently, elongate the chain by adding further ADP-ribose units. In this way, they are capable to modify the protein activity by creating a branched polymer, termed poly(ADP-ribose) (PAR), which can be rapidly degraded by PARG and by ADP-ribosylhydrolase 3 (ARH3) [1,2]. Free or proteinbound ADP-ribose polymers work as signal transducers by binding other proteins through their conserved PAR recognition modules, including PAR-binding motifs (PBMs), PAR-binding zinc finger (PBZF) domains, and macrodomains [3].
Nuclear extracts. Cells were washed twice with Phosphate-buffered saline (PBS), collected by centrifugation and treated for 30 min on ice with a nuclei specific buffer (10 mM Tris-HCl pH 7.9, 4 mM MgCl 2 , 1 mM EDTA, 0.5 mM DTT, 0.25 M sucrose, 1% Triton X-100),supplemented with protease inhibitors (complete EDTA-free, Roche Applied Science). Nuclei were collected by centrifugation at 10.000 xg at 4°C for 10 min. Pellets were lysed in RIPA bufferand processed as above.
HAT and HDAC activity were assayed using EpiQuick TM kits (Epigentek), according to manufacturer instructions.
Detection was carried out using the ECL Western blotting detection reagents (GE Healthcare). Densitometric analysis was performed using Image Lab software (Bio-Rad). For each lane, the value corresponding to the area of histone H3 or H4 band in the membrane probed with anti-acetyl-histone H3 or H4 was normalized by dividing for the value corresponding to the area of histone H3 or H4 band in the same membrane re-probed, after mild stripping, with anti-C-terminal region of histone H3 or H4.

qRT-PCR
Total RNA was purified by RNeasy mini kit (Qiagen). Concentration, purity and integrity of preparations were evaluatedspectrophotometrically, followed by agarose gel-ethidium bromide electrophoresis. Total RNA (1 μg) was subjected to retrotranscription using Bioscript Reverse Transcriptase (Bioline) and random hexamer mix. Amplification of cDNA was performed in triplicate for each sample, using SsoAdvanced SYBR Green supermix on a MiniOpticon Realtime PCR System (Bio-Rad). The values, obtained by three independent experiments, were normalized with Actβ and Hprt. PCR efficiency was 90-100% for each set of primers. Primers used were as follows:

Chromatin immunoprecipitation (ChIP)
ChIP analyses were performed on chromatin extracts using MAGnify Chromatin Immunoprecipitation System kit (Invitrogen), according to manufacturer's specifications. Cell cultures (about 1×10 6 cells/ml) were cross-linked, in standard culture dishes, at room temperature for 10 min by formaldehyde 37% (final concentration 1%). Reaction was stopped by 5 min incubation in 0.125 M Glycine. Cell monolayer was harvested by scraping in ice-cold PBS containing protease inhibitors. After cell lysis (final concentration of cell: 10 6 cells/50 μl) chromatin was sonicated using Bioruptor NextGen (Diagenode) to High Power, 18 cycles for 30 seconds ON, 30 seconds OFF. Average size of sonicated DNA was around 400 bp, as measured by agarose gel electrophoresis. Aliquots containing 200.000 cells were snap-freezed and stored at -80°C. Sheared chromatin was immunoprecipitated with anti-acetyl-Histone H3 or anti-acetyl-Histone H4, or rabbit IgG as negative control. DNA amplification was performed using SsoAdvanced SYBR Green supermix on a MiniOpticon Real-time PCR System (Bio-Rad).
The Ct values for each gene promoter, obtained from three biological replicates of samples analysed in triplicate, were normalized with an internal region of Actβ and INPUT DNA, as follows: first, the Ct value of the immunoprecipitated (IP) target gene was corrected subtracting the Ct value of the Actβ IP; then, the Ct value of the target gene INPUT was corrected subtracting the Ct value of the Actβ INPUT; finally, the normalized target gene IP value was corrected subtracting the normalized INPUT value.
A map describing the position of the promoter fragments analysed for each target gene is presented in S2 Fig.

Statistical analysis
Statistical analysis was carried out by the Student's t-test. Differences were considered significant when p0.05.

Global decrease of histone H3 and H4 acetylation level following PARP inhibition
To determine the influence of PARP activity on the global histone acetylation level, we initially analysed by western blot the extent of both histones H3 and H4 lysine acetylation in conditions in which PARP enzymes are inhibited by PJ34. Mouse NIH3T3 cells were treated with PJ34 for 30 min, 1 hand 3 h. Whole cell extracts were prepared, and aliquots were run on SDS-PAGE gels, transferred to a membrane and probed with antibodies recognizing acetylated lysines of histone H3 and H4 tails. The results are shown in Fig 1, panels A and B for H3 and H4, respectively. Following inhibition of PARPs activity with PJ34, we observed a decrease of lysine acetylation in both histone H3 and H4 tails, within 3 h treatment. As a control for reduced level of ADP-ribose polymers, we utilised on the same cell extracts an antibody against PAR (Fig 1,  panel C). In order to exclude a non specific effect of the drug utilized to inhibit PARPs, we repeated the same experiment with the inhibitor ABT888, and the results are shown in Fig 1, panels D and E for H3 and H4, respectively. PARPs inhibition by ABT888, as shown by decreased PAR levels (Fig 1, panel F), leads to decreased H3 and H4 acetylation. Taken together, the results obtained with the two different inhibitors underline the true involvement of PARP activity rather than reflecting the effect of the drug utilized.
Moreover, the decrease of histone H3 and H4 acetylation level, following PARP inhibition, was not a peculiarity of NIH3T3 cells, because it was observed also in another mouse fibroblast cell line, L929, as well as in the neuroblastoma cell type N2a (see S4 Fig).

PARG over-expression induces global decrease of histone H3 and H4 acetylation
To investigate the molecular mechanism underlying the global decrease of histone tail acetylation, observed in condition of inhibited PARP activity, we sought to analyse whether the same effect can be observed by reducing PAR through a different approach. In order to deplete the cells of PAR, we ectopically over-expressed the PARG enzyme. Whole cells extracts were prepared after transfection with the empty vector pCS2 or with Myc-PARG constructs, one over-expressing PARG and one containing a catalytically mutated PARG. We checked for exogenous protein expression by western blotting, as reported in Fig 2, panel A. Panel B shows that in cells containing plasmid pCS2-Myc-PARG no anti-PAR signal is visible, while the usual smear of PAR is present in cells containing the empty vector. In the extracts from cells containing plasmid pCS2-Myc-PARG, we observed reduced level of histone H3 and H4 tails acetylation (panel C and D, respectively), relative to the control. Interestingly, in the extracts from cells transfected with the plasmid containing catalytically mutated PARG, pCS2-Myc-PAR-G_E757N [20], decreased levels of histone H3 and H4 acetylation were observed (panel C and D, respectively), even though these cells are characterized by a greater amount of PAR (panel B).
In summary, the observed decrease of lysine acetylation, for both histone H3 and H4, is due to an unbalance in the level of PAR (see Discussion).
PARP inhibition down-regulates p300 and Pcaf expression by decreasing promoter histone H3 and H4 acetylation In order to investigate the link between histone acetylation and PARP activity, we next analysed the mRNA steady state level of the genes coding for two relevant enzymes responsible for the maintenance of histone acetylation, namely p300 and Pcaf by qRT-PCR analysis. As shown in Fig 3A, significant down-regulation of both genes was detected after 1 h of treatment with PJ34. However, the transcriptional effect observed at 1 h did not lead to decreased protein amount up to 3 h (S1 Fig), suggesting that the global H3 and H4 acetylation decrease reported was not due to reduced amount of p300 and PCAF. Rather, reduced acetyltransferase activity or increased deacetylase activity could have been involved. We hypothesized that hypoacetylation in the chromatin region spanning the TSS was responsible for decreased p300 and Pcaf mRNA accumulation. Therefore, we investigated by ChIP the promoter region of p300 and Pcaf genes, by using the same antibodies utilized in the western blot experiments, and the results are shown in Fig 3B. Acetylation of both histone H3 and H4 significantly decreased following 1 h of PARP inhibition.
We conclude that PARP activity promotes p300 and Pcaf transcription by maintaining the correct level of histone H3 and H4 acetylation at their TSS.
PARP inhibition leads to decreased histone H3 but not histone H4 acetylation at the Dnmt1 promoter In order to better investigate the relationship between promoter acetylation and PARP activity we tested by ChIP assay the Dnmt1 gene, coding for DNA methyltransferase 1, because the expression and the activity of this enzyme are known to be regulated by PARP-1 [17,21]. Since it was previously shown that the Dnmt1 promoter is occupied by poly(ADP-ribosyl)ated PARP-1, and that the decrease of PAR leads to aberrant methylation of the CpG island close to the TSS [22], we searched for the possible involvement of histone acetylation.
After PJ34 treatment for 1 h, a decrease of histone H3 acetylation at the Dnmt1 promoter region spanning the TSS was observed, while histone H4 acetylation was slightly increased. This pattern of histone tails acetylation changes did not affect Dnmt1 transcription (Fig 4).

PARP inhibition down-regulates basal Tnfα transcription
We then investigated if the mRNA accumulation of other genes known to be PARP-1-dependent was affected by PARP inhibition. In particular, we tested the Tnfα promoter, known to be involved in the inflammatory response to antigenic stimuli, and whose activation was shown to be impaired in PARP-1 knockout mice [23]. Cells were treated with PJ34 for 1 h, and the results are shown in Fig 5A: Tnfα basal transcription was affected by PARP inhibition, as indicated by significant decrease of mRNA accumulation. When the TSS region of the Tnfα promoter was investigated by ChIP after PARP inhibition for 1h, using antibodies against histone H3 or H4 acetylation, a slight but significant increase in histone acetylation levels was observed (see Fig  5B), suggesting that, in this case, PARylation is regulating transcription through a mechanism other than acetylation of H3 or H4 (see Discussion).

PARP inhibition up-regulates global HDAC activity
In order to gain a better insight on the molecular mechanism responsible for the observed effects, we tested nuclear extracts from NIH3T3 cells, treated with PJ34 for 1h, for global acetyltransferase or deacetylase activity, and the results are shown in Fig 6. Global deacetylase activity was found to increase upon PARPs inhibition (panel B), whereas acetyltransferase activity did not change when tested in the same conditions (panel A). Panel C shows the decrease of histone H3 and H4 acetylation in the same nuclear extracts, as expected.
These data suggest that in normal conditions PARP activity is required to prevent default histone deacetylation, thus allowing maintenance of constitutive transcription.

Discussion
Eukaryotic gene regulation occurs at many steps during the process of gene expression: chromatin structure is one of the key control step. External stimuli are capable of modifying the epigenetic code associated with a specific promoter to allow for transcription and cell adaptation. This concept is easily demonstrated in simple organisms such as yeast where the availability of mutants allows the association of specific chromatin patterns to defined phenotypes under different environmental conditions. In more complex systems, depletion of proteins involved in epigenetic control, as well as the use of drugs capable of altering their activity are becoming increasingly useful: in addition to provide evidence of changes at the level of the specific mark, they could also shed light on the possible interplay among epigenetic marks [24][25][26][27].
Cross-talk between PARylation and other epigenetic marks was reported. Regulation of the activity of the DNA methyltransferase DNMT1 by PARP-1 affected genomic DNA methylation [17,18]. PARylation of KDM5B, a histone lysine demethylase acting on trimethyl H3 lysine 4 (H3K4me3), was shown to prevent the binding and demethylase activity of this enzyme [19]. The connection between PARP activity and histone acetylation, however, has  received less attention. Earlier in vitro reports suggested a correlation between the two posttranslational modification processes [28,29], while subsequent evidence points to a competition between them [30,31]. In cultured cortical neurons histone H4 acetylation was shown to be affected by PARP-1 activation [32]. More recently, an interplay between histone acetylation/deacetylation and poly(ADP-ribosyl)ation was shown [33][34]. In order to investigate the interdependence between PARylation and histone tail acetylation, we studied the effect of PARP inhibition both on global histone acetylation level and at specific promoters. We chose to utilize antibodies recognizing all acetylated lysines at the same time on either H3 or H4, instead of antibodies against single lysines, to get a general idea about the hypothesized correlation. We show that the global levels of histone H3 and H4 acetylation decrease as a consequence of alteration of PARP function by two different strategies: i) using PJ34 or ABT888 to inhibit PARP enzymatic activity (Fig 1); ii) over-expressing PARG to deplete the cell of ADPribose polymers (Fig 2). Moreover, we utilised a third experimental strategy, i.e., over-expression of a catalytically mutated PARG, to analyse the effect of increasing the amount of PARs relative to a normal condition, and we show that even in this case histone H3 and H4 acetylation decreases. Since we found that inhibition of PARylation caused an increase in global HDAC activity (Fig 6), we suggest that PARs reduction, due to PARP inhibitors or PARG over-expression lead to failed inhibition of HDACs, thus explaining histone acetylation decrease. To explain why the same effect is observed also in the presence of higher amount of PARs, present in cells transfected with the catalytically mutated PARG, we hypothesize that excess PARylation modifies PARP-1, causing it to be released from chromatin [35]. Presumably, chromatin bound PARP-1 is required to interact with remodelling factors on gene promoters and to recruit/regulate epigenetic marks. Thus, poly(ADP-ribose) levels that are either too high or too low lead to the same outcome in terms of global histone acetylation state: they both cause widespread histone hypoacetylation. These results underscore the critical importance of keeping balanced PAR levels, as already argued relatively to the cross-talk between poly(ADP-ribosyl)ation and DNA methylation [18]. When we used an alternative strategy to increase ADP-ribose polymers level, i.e., by over-expressing the architectural protein CTCF (CCCTC-binding factor), which strongly stimulates PARP-1 activity [36], we again observed global decrease of H3 acetylation (S5 Fig). In addition, we analysed extracts of cells treated with the PARG inhibitor gallotannin, and found a transient increase of histone H3 and H4 acetylation at 30 min, followed by a decrease within 3 h of treatment (S6 Fig). The transient increase observed was possibly due to inhibition of global HDAC activity by increased PARs. It has been shown that PARP-1 and PARG are localized at the promoters of positively and negatively regulated target genes, where they act in concert to control the amount of PARs [8]. Based on the results obtained by assaying HAT and HDAC activity upon PARP inhibition (Fig  6), we hypothesize that, in normal conditions, ADP-ribose polymers regulate histone lysine acetylation and transcription by inhibiting the function of deacetylases. PARP-1 was indeed described in complex with HDACs 1-3, but not HDACs 4-6, to inhibit their function [37]. Other nuclear PARPs could be implicated as well: PARP-14 was shown to interact with HDAC2 and HDAC3 [38]. We cannot exclude that, in defined conditions, ADP-ribose polymers could affect also the function of acetyltransferases. For example, PARP-1 was found in complex with p300 [39], suggesting a stimulatory effect on this enzyme.
Besides global effects, we described alteration of specific regulatory patterns at the level of defined promoters, as follows: i) PJ34-induced PARP inhibition leads to decreased acetylation at both histone H3 and H4 around the TSS region of p300 and Pcaf promoters (Fig 3B), and this effect correlates with decreased mRNA accumulation: histone tails acetylation is, in fact, responsible for keeping promoter nucleosomes accessible ( [40] and ref therein); ii) in the case of the Dnmt1 promoter, only H3 acetylation decreases, while histone H4 acetylation is slightly increased (Fig 4): this pattern of modifications does not parallel transcriptional down-regulation, suggesting a compensatory role for H4 over H3 acetylation, as proposed in the case of S. cerevisiae Adr1-regulated genes [41]; iii) Tnfα constitutive transcription is affected by PARP inhibition, as indicated by significant decrease of basal mRNA accumulation after PJ34 treatment (Fig 5A). When the TSS region of the Tnfα promoter is investigated by ChIP in noninducing conditions, a slight but significant increase of histone H3 and H4 acetylation is observed (Fig 5B), suggesting that, in this specific case, PARylation is likely regulating transcription through a mechanism other than acetylation of H3 or H4. For example, PARylation of histones could be implicated [31,[42][43][44]: since the acceptor sites for this modification are lysine residues, the possibility exists that PJ34 induced PAR reduction allows for the observed acetylation increase. An alternative explanation can be provided when considering that histone acetylation was found to be in continuous dynamic turnover [45,46]. When this turnover was inhibited, the expression of certain genes was no longer possible, despite the increase of histone acetylation levels at their promoters [45,46]. This indicates that cycles of histone acetylation and deacetylation are important for induction of certain genes in mammalian cells.
We argue that for each promoter different patterns of interactions between PARPs and histone acetyltransferases/deacetylases and chromatin remodelers occur.
Taken together, our results indicate that PARP activity affects transcription by modulating histone H3 and H4 acetylation at the level of specific promoters in a context-dependent manner, thus underlying gene-and cell type-specific effects in determining PARP-dependent outcomes, as already suggested [8,10].
The data presented in this work could help to address future investigations aimed at identifying the specific targets of PARP activity among the enzymes regulating histone acetylation, in particular deacetylases. In addition, it will be important to identify the specificity of acetylation/deacetylation sites and how are they regulated.  , value~1), run on a 15% SDS-PAGE, probed with anti-acetyl-histone H3 or anti-C-terminal of H3 antibodies to measure the relative level of H3 acetylation. Error bars indicate the standard deviation of data obtained from the three independent experiments. (B) Same as in (A) but hybridisation was performed with anti-acetyl-histone H4 or anti-C-terminal of H4 antibodies to measure the relative level of H4 acetylation. (C) The same extracts used in panels A and B were run on an 8% SDS-PAGE and probed with anti-PAR antibody to visualize ADP-ribose polymers level. Panels D, E and F: same as in A, B and C, respectively but the analysis was done with whole cell extracts from neuroblastoma N2a cells. The same extracts as in (A) were run on a 15% SDS-PAGE and probed with anti-acetyl-histone H3 or anti-C-terminal of H3 antibodies to measure the relative level of H3 acetylation. (C) Same as in (B) but hybridisation was performed with anti-acetyl-histone H4 or anti-C-terminal of H4 antibodies to measure the relative level of H4 acetylation. (TIF)