Serine ADP-Ribosylation Depends on HPF1

Abstract

ADP-ribosylation (ADPr) regulates important patho-physiological processes through its attachment to different amino acids in proteins. Recently, by precision mapping on all possible amino acid residues, we identified histone serine ADPr marks in the DNA damage response. However, the biochemical basis underlying this serine modification remained unknown. Here we report that serine ADPr is strictly dependent on histone PARylation factor 1 (HPF1), a recently identified regulator of PARP-1. Quantitative proteomics revealed that serine ADPr does not occur in cells lacking HPF1. Moreover, adding HPF1 to in vitro PARP-1/PARP-2 reactions is necessary and sufficient for serine-specific ADPr of histones and PARP-1 itself. Three endogenous serine ADPr sites are located on the PARP-1 automodification domain. Further identification of serine ADPr on HMG proteins and hundreds of other targets indicates that serine ADPr is a widespread modification. We propose that O-linked protein ADPr is the key signal in PARP-1/PARP-2-dependent processes that govern genome stability.

Keywords: ADP-ribosylation; DNA damage; HPF1; PARP-1; PARP-2; genome stability; histones; proteomics; serine ADP-ribosylation.

Graphical abstract

Figure 1

Histone Serine ADPr Is Dependent on HPF1 (A) SILAC strategy to quantify core histone ADPr marks after DNA damage in WT and ΔPARP-1 U2OS cells (left) and a representative western blot of total protein poly-ADP-ribosylation prior to mixing light and heavy lysates from each SILAC experiment (right). Anti-GAPDH was used as a loading control. (B) MS1 of a PARP-1-sensitive modified H2B peptide. The heavy peptide was derived from WT cells, and the light peptide was derived from ΔPARP-1 cells (both stimulated with H₂O₂). The inset (right) shows an ∼1:1 ratio (heavy/light) of a non-ADP-ribosylated peptide from the same experiment. (C) Autoradiogram shows ADP-ribosylation of recombinant H3 by PARP-1 in the presence of HPF1. (D) Autoradiogram shows ADP-ribosylation of two synthetic peptide variants corresponding to amino acids 1–21 of human H3. (E) Autoradiogram shows histone tetramer ADP-ribosylation by PARP-1, PARP-2, or PARP-3 in the presence of HPF1. (F) SILAC strategy to quantify core histone ADPr marks upon DNA damage in WT and ΔHPF1 U2OS cells (left) and a representative western blot of total protein poly-ADP-ribosylation levels prior to mixing light and heavy lysates from each SILAC experiment (right). Anti-GAPDH was used as a loading control. (G) MS1 of an HPF1-sensitive H2B-modified peptide. The heavy peptide was derived from the WT cells, and the light peptide (very low intensity) was derived from ΔHPF1 cells (both were stimulated with H₂O₂). The inset (right) shows an ∼1:1 ratio (heavy/light) of a non-ADP-ribosylated peptide from the same experiment. See also Figure S1.

Figure 2

HPF1 Changes PARP-1 Amino Acid Specificity toward Serine (A) High-resolution ETD fragmentation spectrum of a PARP-1 peptide modified by ADP-ribose on serine 499. The chemical structure of ADP-ribose is depicted. (B) Schematic representation of PARP-1 with the six novel serine ADPr sites. The three sites in the unstructured part of the automodificaton region were confirmed in vivo (underlined serines). Zn I/II/III, zinc-finger domains; BRCT, breast cancer suppressor protein-1 domain; WGR, WGR domain; HD, α-helical subdomain; ART, ADP-ribosyl transferase subdomain. (C) Analysis of the ADP-ribosylation of two different synthetic peptides corresponding to amino acids 494–524 of human PARP-1 is shown. (D) Autoradiogram of the ADP-ribosylation of three different variants of the PARP-1 automodification domain (374–525). In vitro ADP-ribosylation of recombinant H1 served as a positive control. See also Figure S2.

Figure 3

Additional Targets of Serine ADPr (A) High-resolution ETD fragmentation spectrum of an HMGB1 peptide with ADP-ribose on serine 181 is shown. (B) Schematic representations of HMGB1 (upper) and HMGA1 (lower). The three novel serine ADPr sites were identified in vivo. A box and B box, positively charged homologous DNA-binding structures; acidic tail, negatively charged region composed of 30 glutamic and aspartic acids, exclusively; AT, AT-hook with the Arg-Gly-Arg-Pro (RGRP) core motif. (C) Serine ADPr sites identified in histone-depleted fractions. ADPr sites were identified using ETD mass spectrometry. Modified serines are in red. (D) High-resolution ETD fragmentation spectrum of a PARP-1 peptide with ADP-ribose on serine 499, obtained by reprocessing a published dataset (Phanstiel et al., 2011). The chemical structure of ADP-ribose is depicted. See also Figure S3.

Figure 4

Serine ADPr Is a Widespread PTM (A) Gene ontology analysis of the serine ADPr proteins obtained by reprocessing a published dataset (Martello et al., 2016). Biological processes enriched in the serine ADPr proteins are shown. (B) Schematic representation of HMGN1 (upper panel) and sequence alignment of the highly conserved N-terminal region of the HMGN family (lower panel). Note that the serine ADPr site (red) is conserved. NLS, nuclear localization signal; NBD, nucleosomal binding domain; RD, regulatory domain. (C) HCD fragmentation spectrum of an HMGN1 peptide modified by ADP-ribose, obtained by reprocessing a published dataset (Martello et al., 2016). This HCD spectrum contains sufficient localization information to assign serine 25 as the modified residue. ^∗AMP neutral loss. (D) High-resolution ETD fragmentation spectrum of an HMGN1 peptide ADP-ribosylated on serine 25 in vitro in the presence of PARP-1 and HPF1 is shown. (E) Bar plot shows the occurrence of the different amino acids at N termini of ADPr peptides relative to non-ADPr peptides, obtained by reprocessing a published dataset (Martello et al., 2016). See also Figure S4.