PARP-1 and PARG coordinately regulate global patterns of gene expression in MCF-7 cells. A, Venn diagram of PARP-1- and PARG-regulated genes in MCF-7 cells as defined by shRNA-mediated knockdown and expression microarrays. Genes passing both present call and p value <0.05 criteria for at least one factor represent the union (see also supplemental Fig. 1), whereas genes regulated by both factors represent the commonly regulated genes or intersection. B, heat map showing the expression profiles of the commonly regulated genes (i.e. intersection; 485 genes) from A. The genes are ranked in the heat map by log₂ -fold change in the PARP-1 knockdown cell line (see color scale). C, correlation analysis of the commonly regulated genes (i.e. intersection; 485 genes) from A. The Spearman correlation coefficient (c.c.) and p value are indicated.

PARP-1 and PARG localize to the promoters of regulated target genes. PARP-1 and PARG show similar patterns of localization at the promoters of commonly regulated target genes. The occupancy of PARP-1 (black bars) and PARG (gray bars) at the promoters of coregulated genes was examined by ChIP analyses. Each bar represents the mean + S.E. (error bars) from three or more independent determinations. Bars that are not marked with an asterisk are statistically different from the no antibody (NA) control, as determined by a Student's t test with a p value threshold of <0.05. SEMA4G is a gene not regulated by knockdown of PARP-1 or PARG (Fig. 3D). The effects of PARP-1 and PARG knockdown (from Fig. 3D) are indicated for comparison: 1) −, no effect; 2) ↑, up-regulated by both PARP-1 and PARG knockdown; 3) ↓, down-regulated by both PARP-1 and PARG knockdown; and 4) ↑ ↓, differentially regulated by PARP-1 and PARG knockdown. Rel. Enrichment, relative enrichment.

PARP-1 and PARG can affect each other's binding at target gene promoters. To determine whether PARP-1 and PARG can affect each other's binding, the occupancy of PARP-1 and PARG was examined by ChIP analyses in the control and knockdown (KD) cell line of the opposing factor. A, the occupancy of PARP-1 was examined by ChIP analyses in the Luc (black bars) and PARG (gray bars) knockdown cell lines. Each bar represents the mean + S.E. (error bars) from three or more independent determinations. B, the occupancy of PARG was examined by ChIP analyses in the Luc (black bars) and PARP-1 (white bars) knockdown cell lines. Each bar represents the mean + S.E. (error bars) from three or more independent determinations. The data revealed three groups of PARP-1 and PARG binding patterns, as indicated. All changes in occupancy for PARP-1 (Group I and Group II) and PARG (Group I) are statistically different between the control and knockdown cell line, as determined by a Student's t test with a p value threshold of <0.05.

Catalytically inactive mutants of PARP-1 and PARG support the wild-type expression patterns of some, but not all, target genes. RNA interference-resistant wild-type (Wt) or catalytically inactive (CatMut) PARP-1 and PARG were stably expressed in their respective MCF-7 knockdown cells using retrovirus-mediated gene transfer, as shown in Fig. 7. Total RNA was isolated from the cells, reverse-transcribed, and subjected to qPCR using gene-specific primers. Each bar represents the mean + S.E. (error bars) from three or more independent determinations. Bars that do not share at least one lowercase letter marking (a, b, or c) within each graph are statistically different, as determined by analysis of variance with a p value threshold of <0.05. A, re-expression of PARP-1 after PARP-1 knockdown restores the wild-type expression pattern of some PARP-1 target genes. B, re-expression of PARG after PARG knockdown restores the wild-type expression pattern of some PARG target genes.

shRNA-mediated knockdown of PARP-1 and PARG in MCF-7 cells. A, whole cell lysates collected from Luc, PARP-1, and PARG stable shRNA-mediated knockdown cell lines were subjected to Western blotting analyses for PARP-1 and PARG. SIRT1 was also analyzed as a loading control. PAR levels were analyzed by Western blotting using nuclear extracts from the same cell lines. B, RT-qPCR analysis confirms the knockdown PARP-1 and PARG mRNA in MCF-7 cells. Total RNA was isolated from Luc, PARP-1, and PARG knockdown cells, reverse-transcribed, and subjected to qPCR using gene-specific primers to PARP-1 and PARG. Each bar is the mean + S.E. (error bars) for three independent RNA isolations. Bars that are marked with an asterisk are statistically different from the Luc control, as determined by a Student's t test with a p value threshold of <0.05. C, total cellular NAD⁺ levels (black bars) in Luc, PARP-1, and PARG knockdown cells were measured using a quantitative HPLC/mass spectrometry method with ¹⁸O standards. PAR levels (gray bars) were quantified from Western blots (from A, above) by densitometry. The data are shown as the mean + range or S.E. (error bars) from two or more independ ent biological replicates. Bars that are marked with an asterisk are statistically different from the Luc control, as determined by a Student's t test with a p value threshold of <0.05.

Defining the genes most robustly regulated by PARP-1 and PARG. A, Venn diagram of PARP-1- and PARG-regulated genes after applying a -fold change cutoff of log₂ <−0.5 or >0.5. Of the union of 371 genes most robustly regulated by either PARP-1 or PARG, 50 are commonly regulated. B, the distribution of the most robustly up-regulated or down-regulated genes upon knockdown of PARP-1 or PARG is indicated. C, correlation analysis comparing the magnitude and direction of regulation of the 371 most robustly regulated genes shown in A and B. The Spearman correlation coefficient (c.c.), p value, and -fold change cutoff are indicated. D, gene-specific confirmation of the expression microarray results by RT-qPCR. Luc, PARP-1, and PARG knockdown MCF-7 cells were seeded and grown under the same conditions used for the expression microarrays. Total RNA was isolated, reverse-transcribed, and subjected to qPCR using gene-specific primers. Genes were considered to be regulated if the log₂ -fold change was <−0.5 or >0.5 (values falling outside of the shaded box). Each bar represents the mean + S.E. (error bars) from three or more independent determinations. All bars with a mean value falling outside of the shaded box are statistically different from the Luc control, as determined by a Student's t test with a p value threshold of <0.05.

PARP-1 and PARG binding patterns are similar across individual target gene promoters. PARP-1 and PARG binding patterns are similar across individual promoters with both high and low occupancy. A, the occupancy of PARP-1 (black bars) and PARG (gray bars) at two regions (∼1 kb apart) of the GDF15, PVALB, LGALS3BP, and NELL2 promoters was determined by ChIP analyses. Each bar represents the mean + S.E. (error bars) from three or more independent determinations. Rel. Enrichment, relative enrichment. TSS, transcription start site. B, PARP-1 (solid line) and PARG (dotted line) occupancy was examined by ChIP-tiling analyses over a 2.5-kb region of the PVALB and LGALS3BP promoters. Each point represents the mean + S.E. (error bars) from three or more independent determinations; all values are statistically different from the no antibody control, as determined by a Student's t test with a p value threshold of <0.05.

Stable re-expression of shRNA-resistant PARP-1 and PARG in their cognate knockdown cell lines. A, schematic diagram of the human PARP-1 structural and functional domains. The DNA binding domain (DBD), zinc fingers (Zn Fingers), nuclear localization signal (NLS), third zinc binding domain (Zn3), automodification domain (AMD), and PARP “signature” motif are shown. Open triangles indicate the location of the 21-nucleotide shRNA recognition sequences used for knockdown. Filled circles indicate the location of the silent point mutations engineered into the cDNAs to make them resistant to the shRNAs. The open circle indicates the location of the inactivating point mutation (Glu 988 to Lys) that inhibits PARP-1 enzymatic activity in the catalytically inactive mutant (CatMut). Wt, wild type. B, FLAG-tagged RNA interference-resistant wild-type or catalytically inactive PARP-1 was stably expressed in MCF-7 PARP-1 knockdown cell lines. Re-expression was confirmed by Western blotting for PARP-1 and FLAG. An empty vector was used as a control (Empty) in both Luc and PARP-1 knockdown cell lines. C, schematic diagram of the rat PARG structural and functional domains. The regulatory domain, catalytic domain, active site, nuclear localization signal, and nuclear export signal (NES) are shown. Open triangles indicate the location of the 21-nucleotide shRNA recognition sequences used for knockdown. Filled circles indicate the location of the silent point mutation engineered into the cDNAs to make them resistant to the shRNA. The open circles indicate the location of the inactivating point mutations (Tyr-788 to Phe and Tyr-791 to Ala) that inhibit PARG enzymatic activity in the catalytically inactive mutant. D, FLAG-tagged RNA interference-resistant wild-type or catalytically inactive PARG was stably expressed in MCF-7 PARG knockdown cell lines. Re-expression was confirmed by Western blotting for PARG and FLAG. An empty vector was used as a control (Empty) in both Luc and PARP-1 knockdown cell lines.