Proteomic analyses identify ARH3 as a serine mono-ADP-ribosylhydrolase

Abstract

ADP-ribosylation is a posttranslational modification that exists in monomeric and polymeric forms. Whereas the writers (e.g. ARTD1/PARP1) and erasers (e.g. PARG, ARH3) of poly-ADP-ribosylation (PARylation) are relatively well described, the enzymes involved in mono-ADP-ribosylation (MARylation) have been less well investigated. While erasers for the MARylation of glutamate/aspartate and arginine have been identified, the respective enzymes with specificity for serine were missing. Here we report that, in vitro, ARH3 specifically binds and demodifies proteins and peptides that are MARylated. Molecular modeling and site-directed mutagenesis of ARH3 revealed that numerous residues are critical for both the mono- and the poly-ADP-ribosylhydrolase activity of ARH3. Notably, a mass spectrometric approach showed that ARH3-deficient mouse embryonic fibroblasts are characterized by a specific increase in serine-ADP-ribosylation in vivo under untreated conditions as well as following hydrogen peroxide stress. Together, our results establish ARH3 as a serine mono-ADP-ribosylhydrolase and as an important regulator of the basal and stress-induced ADP-ribosylome.

Fig. 1

ARH3 has mono-ARH activity. a Left panel: Recombinant H3 histone tail was in vitro ADP-ribosylated using recombinant ARTD1 in the presence of ³²P-labeled NAD⁺. Equal fractions were left untreated (Input) or were treated with PARG or ARH3. Above: radioactivity exposure, below: Coomassie Blue-stained poly-acrylamide gel. Right panel: Quantification of a expressed as demodification activity (=reduction of the radioactive signal, normalized to amount of protein). Data represent means ± SEM for n = 3 independent experiments, ***P < 0.0001 as determined by ANOVA. b Left panel: Automodification of recombinant ARTD8 in the presence of ³²P-labeled NAD⁺ results in MAR-labeled ARTD8 (Input) that was subsequently treated with recombinant PARG or ARH3. Above: radioactivity exposure, below: Coomassie Blue-stained poly-acrylamide gel. Right panel: Quantification of b expressed as demodification activity (=reduction of the radioactive signal, normalized to amount of protein). Data represent means ± SEM for n = 3 independent experiments, ***P < 0.0001 as determined by ANOVA. c Left panel: Structural overlap of human ARH3 (green) and the DraG/ADPr complex (cyan). Right panel: Zoom in the active site of ARH3 with side chains of key residues which were mutated (labels). The binding mode of ADPr (carbon atoms in cyan) was obtained by energy minimization starting from the pose obtained by the structural overlap. d Coomassie Blue-stained membrane of pull-downs of GST-ARH3 (left) or His-PARG using the biotinylated peptides with (H2B-ADPr) and without (H2B) modification. Af1521 served as positive control. e ARTD1 automodified in the presence of ³²P-labeled NAD⁺ was subjected to demodification using WT and different ARH3 mutants. Red labels: mutants deficient in binding to H2B-ADPr, green labels: mutants retaining binding to H2B-ADPr. f ARTD8 automodified in the presence of ³²P-labeled NAD⁺ was subjected to demodification using WT and different ARH3 mutants. Red labels: mutants deficient in binding to H2B-ADPr, green labels: mutants retaining binding to H2B-ADPr

Fig. 2

ARH3 mainly hydrolyzes ADP-ribosylated serines in vitro. a Number of ADP-ribosylated peptide spectra matches (PSMs) or demodified peptide spectra matches after PARG or ARH3 treatment. Data represent means ± SEM for n = 3 independent demodification experiments. b Venn diagrams of unique ADPr peptides and proteins. c Volcano plot of ARH3- and PARG-treated samples. “Unmodified peptides” are shown as open circles and “ADP-ribosylated peptides” as filled circles. ADP-ribosylation sites confirmed by EThcD spectra are annotated and color coded in red as S-ADPr and in blue as R-ADPr sites. ADP-ribosylated peptides with uncertain ADP-ribosylation site localization are shown in black. The black hyperbolic line represents a permutation-based false discovery rate (FDR) of 5% and a minimal fold change of 2. d Normalized abundance of individual Ser- and Arg-ADPr peptides after PARG or ARH3 treatment. Data represent means ± SEM for n = 3 independent demodification experiments

Fig. 3

ARH3 regulates basal and hydrogen peroxide-induced serine ADP-ribosylation in vivo. a Venn diagrams of unique ADP-ribosylated peptides of wild type (WT) and ARH3 KO MEF cells under basal and H₂O₂-treated conditions. b Unique ADP-ribosylation sites detected by EThcD fragmentation in the different samples. c Gene ontology analysis of the identified ADPr-modified proteins using the PANTHER database. Shown on the left are the P-values and on the right the number of identified and annotated ADP-ribosylated proteins. d Validation of mono-ARH activity of ARH3 on the nuclear protein HMGB1. Recombinant HMGB1 was in vitro ADP-ribosylated using recombinant ARTD1 in the presence of ³²P-labeled NAD⁺. Equal fractions were left untreated (Input) or were treated with PARG or ARH3. Above: radioactivity exposure, below: Coomassie Blue-stained poly-acrylamide gel. e Motif searches for ADP-ribosylated peptides with a mascot site localization score >80% in MEF cells using Weblogo. f Energy minimized binding mode of an acetyl-KSG peptide with ADPr-Ser modification. The surface of ARH3 (including the binding-site magnesium ions) is colored according to electrostatic potential (on a scale of −5 to 5 kT/e). The positively charged amino group of the K side chain and the backbone amide groups point toward the region of the surface with negative potential