S C I E N C E A D VA N C E S | R E S E A R C H A R T I C L E

BIOCHEMISTRY Copyright © 2023

The Authors, some
PARP14 is a PARP with both ADP-ribosyl transferase and rights reserved;
exclusive licensee
hydrolase activities American Association
for the Advancement
of Science. No claim to
Nina Đukić1†, Øyvind Strømland1,2†, Jonas Damgaard Elsborg3, Deeksha Munnur1, Kang Zhu1, original U.S. Government
Marion Schuller1, Chatrin Chatrin1, Pulak Kar1, Lena Duma1, Osamu Suyari1, Works. Distributed
Johannes Gregor Matthias Rack1,4, Domagoj Baretić1, Dorian Richard Kenneth Crudgington1, under a Creative
Joséphine Groslambert1, Gerissa Fowler5, Sven Wijngaarden6, Evgeniia Prokhorova1, Commons Attribution
Jan Rehwinkel5, Herwig Schüler7, Dmitri V. Filippov6, Sumana Sanyal1, Dragana Ahel1, License 4.0 (CC BY).

Michael L Nielsen3, Rebecca Smith1*, Ivan Ahel1*

PARP14 is a mono–ADP-ribosyl transferase involved in the control of immunity, transcription, and DNA replica-
tion stress management. However, little is known about the ADP-ribosylation activity of PARP14, including its
substrate specificity or how PARP14-dependent ADP-ribosylation is reversed. We show that PARP14 is a dual-
function enzyme with both ADP-ribosyl transferase and hydrolase activity acting on both protein and nucleic
acid substrates. In particular, we show that the PARP14 macrodomain 1 is an active ADP-ribosyl hydrolase. We
also demonstrate hydrolytic activity for the first macrodomain of PARP9. We reveal that expression of a PARP14
mutant with the inactivated macrodomain 1 results in a marked increase in mono(ADP-ribosyl)ation of proteins
in human cells, including PARP14 itself and antiviral PARP13, and displays specific cellular phenotypes. More-
over, we demonstrate that the closely related hydrolytically active macrodomain of SARS2 Nsp3, Mac1,
efficiently reverses PARP14 ADP-ribosylation in vitro and in cells, supporting the evolution of viral macrodo-
mains to counteract PARP14-mediated antiviral response.

INTRODUCTION from NAD+ [nicotinamide adenine dinucleotide (oxidized form)]

Cells must quickly adapt to both internal and external changes. This onto a target, releasing nicotinamide. These ADP-ribose moieties
could be due to internal pressures such as changes in metabolic can either be added as a single moiety, known as mono-ADPr, or
demands that require alteration in transcription or protein transla- they can be attached as long branched polymers, so called
tion, DNA replication, or DNA damage repair, as well as from ex- poly(ADP-ribosyl)ation (6). A number of different reader
ternal pressures such as invasion of pathogenic bacteria and viruses. domains have been shown to recognize and bind ADP-ribose in-
For such changes to occur efficiently, cells have developed a number cluding macrodomains that primarily recognize either
of signaling pathways to transduce signals rapidly. This often in- mono–ADP-ribose or the terminal ADP-ribose moiety of a
volves posttranscriptional or posttranslational modification polymer or WWE domains and PAR-binding zinc fingers that spe-
(PTM) of nucleic acids and proteins, respectively. One such signal- cifically recognize poly(ADP-ribose) (7–10). Last, there are eraser
ing type is the adenosine 50 -diphosphate (ADP)–ribosylation proteins involved in the removal of ADP-ribose, primarily ADP-
(ADPr). ADPr has been shown to target both proteins (1) and ribosyl hydrolases (ARHs), or hydrolytic macrodomains, of which
nucleic acids, including RNA and DNA (2), with modification on four—poly(ADP-ribosyl) glycohydrolase (PARG), MacroD1,
proteins able to occur on different amino acid acceptors including MacroD2, and terminal ADP-ribose protein glycohydrolase
serine, glutamate, and arginine (3–5). (TARG1)—have been described in humans (11).
Like in most signaling pathways, there are proteins required for While there have been strides taken in understanding how
the writing, reading, and reversal of ADPr. The largest known several members of the PARP family function, there are still out-
family of proteins that are responsible for the addition of ADP- standing questions regarding the specificity of their ADPr activity,
ribose to proteins or nucleic acids are the poly(ADP-ribose) poly- as well as the hydrolases that remove their modification. The best
merases (PARPs). PARPs function by transferring ADP-ribose understood aspect of ADPr signaling in humans is its role in the
DNA damage response. Here, PARP1 or PARP2 will recognize
1
Sir William Dunn School of Pathology, University of Oxford, Oxford OX1 3RE, UK.
and bind to DNA breaks (12) where they interact with their auxil-
2
Department of Biomedicine, University of Bergen, 5020 Bergen, Norway. iary factor histone PARylation factor 1 (HPF1) (13). Together, they
3
Proteomics Program, Novo Nordisk Foundation Center for Protein Research, ADP-ribosylate proteins around the break site, including PARP1
Faculty of Health and Medical Sciences, University of Copenhagen, Blegdamsvej itself, on serine residues (14, 15). The modified proteins can be
3B, 2200 Copenhagen, Denmark. 4MRC Centre for Medical Mycology, University
of Exeter, Geoffrey Pope Building, Stocker Road, Exeter EX4 4QD, UK. 5Medical Re- also then recognized by ADP-ribose–binding repair proteins such
search Council Human Immunology Unit, Medical Research Council Weatherall In- as ALC1, XRCC1, and APLF, which are required for efficient
stitute of Molecular Medicine, Radcliffe Department of Medicine, University of DNA repair (16). Following the initial ADP-ribose signaling at
Oxford, Oxford OX3 9DS, UK. 6Leiden Institute of Chemistry, Leiden University, Ein-
steinweg 55, 2333 CC Leiden, Netherlands. 7Center for Molecular Protein Science,
DNA breaks, the ADP-ribose moieties are efficiently removed by
Department of Chemistry, Lund University, 22100 Lund, Sweden. PARG, which degrades long polymers of ADP-ribose, while the
*Corresponding author. Email: rebecca.smith@path.ox.ac.uk (R.S.); ivan.ahel@path. final ADP-ribose on serine residues is removed by ARH3 (17). Re-
ox.ac.uk (I.A.) cently, we have also begun to understand the complexity of ADPr
†These authors contributed equally to this work.

Đukić et al., Sci. Adv. 9, eadi2687 (2023) 13 September 2023 1 of 18

S C I E N C E A D VA N C E S | R E S E A R C H A R T I C L E

signaling on nucleic acids in bacterial systems, where a PARP-like has been shown to efficiently modify itself both in its catalytic
protein, DarT2, catalyses the ADPr of thymidine bases, which can region and isolated MD2 and MD3, but not MD1 (3, 38), and
be efficiently reversed by the hydrolytic macrodomain DarG (18, ADPr of endogenous PARP14 on acidic residues has also been de-
19). Several human PARPs have been suggested to ADP-ribosylate tected in IFN-γ–stimulated primary human macrophages (39).
50 and 30 phosphorylated single-stranded (ss) DNA and RNA (2, 20– PARP14’s close relative, PARP9, is also induced by interferon
22). This is reversed by endogenous ADP-ribosyl hydrolases, in- stimulation and is expressed from the same genetic locus as
cluding PARG, TARG1, MacroD1, MacroD2, and ARH3 (20, 21). PARP14 (chromosome 3q21.1). Structurally, PARP9 is similar to
Despite the wealth of knowledge on ADPr and PARP1, much PARP14 with two KH domains and two macrodomains at the N
less is known about most of the other human PARPs. One poorly terminus and a C-terminal ART domain (Fig. 1A). No ART activity
understood subgroup is the interferon-induced “antiviral PARPs,” of PARP9 has been detected to date, presumably because of the lack
which include PARP7, 9, 10, 11, 12, 13, 14, and 15. Several of several essential catalytic amino acids within the ART domain
members of this class have been reported to modify both protein (3). However, PARP9 forms a heterodimer with Deltex E3 ubiquitin
and nucleic acid substrates (20–22), while PARP9 and PARP13 ligase 3L (DTX3L), which ubiquitinates different cellular, as well as
are reported to be catalytically inactive (3). These antiviral PARPs viral substrates. For instance, ubiquitination of histones promotes
are interferon-inducible and were shown to confer resistance to a increased histone methylation, leading to chromatin remodeling
range of viruses, including coronaviruses, influenza, HIV, and and eventually enhanced expression of interferon-stimulated
Ebola (23–27). These PARPs are also under positive evolutionary genes (40). PARP9 has also been reported to serve as an RNA
selection, strongly suggesting coevolution with viruses as a conse- virus sensor, leading to activation of the phosphatidylinositol 3-
quence of host-virus conflicts (28). kinase/AKT3 signaling pathway and subsequent production type I
The largest of all the human PARPs is PARP14. The N terminus IFN (41). In addition to roles in viral defense, PARP9/DTX3L has
of PARP14 harbors three RRM (RNA recognition motif ) domains also been implicated in the maintenance of genome integrity. Spe-
and eight putative K homology (KH) domains, which may bind cifically, PARP9/DTX3L is recruited to sites of damage where
RNA or DNA, in addition to three ADPr-binding macrodomains DTX3L ubiquitinates different targets, including p53 (42, 43). Re-
(MD1, MD2, MD3) (Fig. 1A) (29). The C terminus of PARP14 con- cently, it was reported that DTX3L can also ubiquitinate ADP-
tains a WWE domain, followed by the catalytic ADP-ribosyl trans- ribose on proteins in vitro, presenting a PTM with unknown phys-
ferase (ART) domain. PARP14 has been reported to regulate several iological functions (44).
different pathways involved in immunity, inflammation, and Previous studies have suggested that Mac1 of SARS-CoV-2
genome stability. PARP14 was initially shown to be involved in tran- (SARS2) Nsp3 is closely related to macrodomain 1 of PARP9 and
scriptional regulation, acting as a molecular switch of interleukin-4 PARP14 (38). Given these similarities, we sought to examine
(IL-4)–regulated genes (30). In basal conditions, PARP14 represses whether PARP9 and PARP14 MD1 share the same hydrolytic func-
gene transcription by binding to IL-4–responsive promoters and re- tion. Here, we show that these macrodomains are active hydrolases
cruiting histone deacetylase 2/3 (HDAC2/3). In contrast, under IL- and can remove ADPr from both protein and nucleic acid sub-
4–stimulated conditions, PARP14 is activated, leading to the disso- strates. We also show that PARP14 modifies proteins in cells by
ciation of HDAC2/3 from the promoter regions. Consequently, this mono-ADPr and that this ADPr is reversed by its own macrodo-
allows the binding of the transcription factor signal transducer and main. In addition, we identify targets of PARP14 MD1 by mass
activator of transcription 6 (STAT6), as well as other transcription spectrometry (MS). Thus, we show that PARP14 is a PARP that
cofactors, to their target genes and allows efficient gene transcrip- acts both as a transferase and as a hydrolase. We further show
tion (31). PARP14 has also been shown to regulate transcription in that SARS2 Mac1 can reverse PARP14-dependent ADPr.
response to interferon-γ (IFN-γ) stimulation. Specifically, PARP14
has been suggested to ADP-ribosylate STAT1, inhibiting phosphor-
ylation of STAT1 and subsequent activation of proinflammatory RESULTS
gene expression. With its role in regulating genomic stability, PARP14 and PARP9 macrodomain 1 exhibit ADP-ribosyl
PARP14 was shown to regulate DNA repair by homologous recom- hydrolase activity on protein substrates
bination (HR) and in the replication stress response (32–34). PARP14, the largest of the human PARPs belonging to the macro-
PARP14 has also been reported to play an important role in the domain-containing PARPs, together with PARP9 and PARP15, is a
antiviral response (25, 35). In the context of coronavirus infection, potent mono (ADP-ribosyl) transferase (Fig. 1A) (3, 29, 38, 45).
PARP14 is required to enhance type I interferon production and PARP14 ADPr activity is efficiently reversed in vitro by SARS2
restrict replication of murine hepatitis virus, a model coronavirus Nsp3 Mac1 (38); however, human endogenous hydrolases that
(25). To combat the antiviral activity of these PARPs, severe acute can reverse PARP14 modification remain elusive. To identify po-
respiratory syndrome coronavirus (SARS-CoV) contains a hydro- tential human hydrolases with activity toward PARP14-mediated
lytic macrodomain (36, 37) within the nonstructural protein 3 ADPr, we compared human macrodomains to Mac1. Phylogenetic
(Nsp3), which has been suggested to remove PARP14-mediated analysis suggests that while there is an obvious homology to the
ADPr (38). Nsp3 macrodomain 1 (Mac1) is critical for virus repli- known hydrolases such as MacroD1, the closest orthologs are the
cation in vivo and viruses with mutated or absent macrodomains first macrodomain of PARP14 (PARP14 MD1) and the first macro-
are unable to hydrolyse host ADPr and therefore associated with domain of PARP9 (PARP9 MD1) (Fig. 1, B to D), while macrodo-
reduced viral loads and increased sensitivity to IFN-I treatment in mains 2 (MD2) and 3 (MD3) of PARP14 and macrodomain 2
PARP14-proficient cells (25, 26). However, the exact mechanism of (MD2) of PARP9 are more diverged. Comparison of ADP-ribose
activation and the molecular substrates of PARP14 and the viral complex structures of PARP14 MD1 and Mac1 reveals that the res-
macrodomains are still poorly characterized. Notably, PARP14 idues important for distal ribose coordination are also structurally

Đukić et al., Sci. Adv. 9, eadi2687 (2023) 13 September 2023 2 of 18

S C I E N C E A D VA N C E S | R E S E A R C H A R T I C L E

Fig. 1. PARP14 and PARP9 macrodomain 1 are similar to SARS2 Mac1. (A) Domain architecture of human PARP14 and PARP9. (B) Unrooted phylogenetic tree of
human and mouse macrodomains including of PARP9, 14, and 15 and viral macrodomains (highlighted in cyan). (C) Multiple sequence alignment showing conservation
of catalytic residues (magenta-framed) and residues involved in ADP-ribose coordination (cyan-framed) of human PARP14 and PARP9 macrodomains in comparison to
SARS2 Nsp3 Mac1. Numbers on top of the residues refer to human PARP14 MD1. (D) Pairwise sequence identity comparison of SARS2 Nsp3 Mac1 and human macro-
domains. (E) Crystal structure overlay of PARP14 MD1 [Protein Data Bank (PDB) ID: 3Q6Z] and SARS2 Nsp3 Mac1 (PDB ID: 7KQP) both in complex with ADP-ribose (root
mean square deviation of 1.02 Å over 214 Cα). The catalytic residues are highlighted in magenta.

conserved, suggesting that PARP14 MD1 and Mac1 potentially examined whether PARP14 MD1 could hydrolyse trans-modified
share the same catalytic ability and functional context (Fig. 1, D PARP14 MD3. We were able to observe that PARP14 MD1 and
and E). SARS2 Nsp3 Mac1 efficiently reversed PARP14 MD3 ADPr
These observations prompted us to investigate the potential (Fig. 2B), while PARP14 MD2 and PARP14 MD3 had no discern-
ADP-ribosyl hydrolase activities of the macrodomains in PARP9 ible effect on the ADP-ribosylation level of trans-modified
and PARP14. First, we used the automodified PARP14 catalytic PARP14 MD3.
fragment (WWE-CAT) as a model substrate and the isolated macro- PARP14 is involved in macrophage activation whereby gene ex-
domains derived from PARP14 (MD1 to MD3) as described previ- pression enabling the defense against pathogens is induced. In par-
ously (38). PARP14 MD1 and SARS2 Mac1 have a notable ticular, PARP14 has been suggested to ADP-ribosylate STAT1α,
hydrolysis activity on automodified PARP14 WWE-CAT, while preventing STAT1α phosphorylation, which is essential for
MD2 and MD3 did not exhibit activity (Fig. 2A and fig. S1A). To STAT1α to drive transcription of proinflammatory genes (47). Con-
strengthen our finding, we introduced a point mutation G832E in versely, PARP9 has been reported to antagonize the activation by
PARP14 MD1 that sterically blocks the active site in macrodomain inhibiting the ADPr of STAT1α (47). Given the similarity
hydrolases for ADP-ribose binding (46). As predicted the mutation between PARP9 MD1, PARP14 MD1, and SARS2 Mac1 and the re-
diminished the catalytic activity of PARP14 MD1 (Fig. 2A). Mutat- ported role of PARP9 in antagonizing PARP14 ADPr, we investigat-
ing the corresponding residue in PARP14 MD2 had no discernible ed whether PARP9 MD1 can reverse PARP14 automodification.
effect (Fig. 2A). PARP14 MD3 has also been reported to be robustly PARP9 MD1 could efficiently reverse PARP14 ADPr, while
modified by PARP14 (38). We used this to our advantage and PARP9 MD2 had no observable effect (Fig. 2C). We also tested

Đukić et al., Sci. Adv. 9, eadi2687 (2023) 13 September 2023 3 of 18

S C I E N C E A D VA N C E S | R E S E A R C H A R T I C L E

Fig. 2. PARP14 and PARP9 MD1 reverse glutamate-linked PARP14 auto- and trans-ADPr. (A) PARP14 WWE-CAT was auto–ADP-ribosylated using NAD+ spiked with
32
P NAD+. The ADP-ribosyl hydrolysis activity of PARP14 MD1, MD2, MD3, MD1mut (G823E), MD2mut (G1044E), and SARS2 Nsp3 Mac1 (SARS2 Mac1) was assessed upon
incubation with automodified PARP14 WWE-CAT. (B) PARP14 WWE-CAT and PARP14 MD3 were auto– and trans–ADP-ribosylated, respectively, using NAD+ spiked with
32
P NAD+. The trans-ADPr hydrolysis activity of PARP14 MD1, MD2, MD3, MD1mut, MD2mut, and SARS2 Mac1 was assessed upon incubation with the trans-modified
PARP14 MD3 and auto-modified PARP14 WWE-CAT. (C) PARP14 WWE-CAT was auto–ADP-ribosylated using NAD+ spiked with 32P NAD+. The ADP-ribosyl hydrolysis
activity of SARS2 Mac1, PARP14 MD1, PARP9 MD1, PARP9 MD2, MacroD1, MacroD2, and TARG1 was determined upon incubation with automodified PARP14 WWE-
CAT. Samples in (A to C) were analyzed by Coomassie brilliant blue staining and autoradiography. The arrows show the position of the indicated macrodomains. (D)
Hydrolysis of arginine-, serine-, and glutamate-linked mono(ADP-ribosyl)ation on synthetic peptides by PARP14 MD1, PARP14 MD1mut, PARP14 MD2, PARP14 MD3,
PARP9 MD1, PARP9 MD2, and SARS2 Mac1. Briefly, the released ADP-ribose was converted by NUDT5 to adenosine 50 -monophosphate (AMP), which subsequently
was detected by luminescence using the AMP-Glo assay (Promega). Samples are background-corrected and normalized to the positive control, ARH1 for arginine,
ARH3 for serine, and MacroD1 for glutamate. The data represent means ± SD measured in triplicates.

several known human hydrolases and saw that MacroD1, MacroD2, MD1 showed no activity on serine- and arginine-linked peptides
and TARG1, known to remove glutamate-linked ADP-ribose from in contrast to the cognate hydrolases ARH3 (17) and ARH1 (50).
target proteins (48, 49), exhibited pronounced hydrolytic activity on
automodified PARP14 (Fig. 2C), suggesting that the PARP14- PARP14 and PARP9 macrodomain 1 exhibit ADP-
derived ADPr could be glutamate linked. To investigate the ribosylhydrolase activity on nucleic acid substrates
amino acid specificity of the PARP9 MD1 and PARP14 MD1, we The domain architecture of PARP14 suggests that it is tightly linked
assessed their activity against chemically synthesized defined to nucleic acids because it harbors three RRM domains and eight
ADP-ribosylated peptide substrates modified on serine, arginine, KH domains that all are putative ssRNA or ssDNA binders
and glutamate, respectively (Fig. 2D).We observed that PARP14 (Fig. 1A). Furthermore, PARP14 is interferon induced and plays a
MD1 and PARP9 MD1 are specifically active on glutamate-ADPr, role in the innate immune response against viruses (51). Interferon-
similar to SARS2 Mac1 and MacroD1. PARP14 MD1 and PARP9 induced antiviral PARPs, such as PARP10 and PARP11, ADP-ribo-
sylate 50 and 30 phosphorylated ssRNA and ssDNA (21). On the

Đukić et al., Sci. Adv. 9, eadi2687 (2023) 13 September 2023 4 of 18

S C I E N C E A D VA N C E S | R E S E A R C H A R T I C L E

basis of these observations, we postulated that PARP14 MD1 could Last, in a competition assay where DNA-ADPr was in eightfold
reverse ssDNA and ssRNA ADPr. ssRNAs phosphorylated at either molar excess compared to an ADP-ribosylated protein substrate,
the 50 or at 30 were ADP-ribosylated with PARP14, reaction-inhib- we did not see any notable difference in protein-ADPr hydrolysis,
ited with PARP14i (RBN012759), and then used as potential sub- suggesting that in our experimental setup, DNA is not an efficient
strates for PARP14 macrodomains [Fig. 3, A to C (lane 2)]. competitor for protein-ADP-ribose hydrolysis (fig. S1A). Together,
PARP14 MD1 efficiently reversed the ADPr from both RNA sub- our data identify two additional ADP-ribosyl hydrolases in humans
strates, and as expected, PARP14 MD2 and MD3 did not (Fig. 3, A (MD1 of PARP9 and PARP14) and demonstrate that PARP14 rep-
and B). SARS2 Mac1, as reported previously, could also remove the resents a PARP enzyme that can reverse its own modification on
modification (Fig. 1, A and B) (21). Next, we tested whether both protein and nucleic acid substrates.
PARP14 MD1 could also reverse the ADPr of 50 phosphorylated
ssDNA. As for ssRNA, PARP14 MD1 reversed the ADPr on PARP14 shows ADP-ribosyl transferase and hydrolase
ssDNA, while PARP14 MD2 and MD3 did not (Fig. 3C and fig activity on different cellular substrates
S1B). When comparing hydrolase efficiencies against ADP-ribosy- We next studied PARP14 activity in human cells. To do so, we tran-
lated DNA or RNA, we did not observe major differences in siently transfected U2OS cells, which express PARP14 endogenous-
PARP14 activity for either substrate (fig S1C). We also determined ly, or 293T cells, which are naturally deficient in PARP14 (fig S3, A
whether PARP9 MD1 can reverse ADPr of 50 phosphorylated and B) (52), with yellow fluorescent protein (YFP)–tagged full-
ssDNA, finding that PARP9 MD1 but not MD2 reversed the mod- length PARP14 wild type (WT), PARP14 R1699A catalytic
ification (Fig. 3D). As reported previously, MacroD1 could also mutant, PARP14 G832E MD1, and PARP14 G1044E MD2
remove ADPr of 50 phosphorylated ssDNA, while the unrelated mutant, and examined changes in protein mono(ADP-ribosyl)ation
human hydrolase ARH1 could not. Last, we tested whether in cell extracts using a mono-ADPr–specific antibody (Fig. 4A and
PARP14 was capable of adding and removing ADPr on 50 or 30 fig. S3C) (53). In both cell lines, overexpression of WT PARP14, but
phosphorylated double-stranded DNA (dsDNA). Our data show not the R1699A ADPr-deficient mutant, resulted in a modest in-
that PARP14 modifies dsDNA substrates including blunt-ended, crease of mono-ADPr, indicating that this mutant is devoid of cat-
gapped, nicked, forked, and single-stranded overhangs equally effi- alytic activity. On the other hand, we observed a marked increase in
ciently at 50 or 30 phosphates, while PARP14 MD1 more efficiently mono(ADP-ribosyl)ation of a variety of protein sizes when we over-
removes ADPr from 50 phosphorylated dsDNA substrates (fig. S2). expressed the PARP14 MD1 mutant, suggesting that PARP14

Fig. 3. PARP14 and PARP9 MD1 reverse ADPr of ssRNA and ssDNA. (A) ssRNA with 50 phosphate and 30 cyanine 3 (Cy3), (B) ssRNA with 30 phosphate and 50 Cy3, and (C)
ssDNA with 50 phosphate and 30 Cy3 were ADP-ribosylated using PARP14 WWE-CAT. Subsequently, the ADPr was hydrolyzed by treating the modified oligos with PARP14
MD1, MD1mut, MD2, MD3, or SARS2 Mac1. (D) ssDNA with 50 phosphate and 30 Cy3 was ADP-ribosylated using PARP14 WWE-CAT. Following, the ADP ribose modification
was hydrolyzed by subjecting the ADP-ribosylated oligo to PARP14 MD1, SARS2 Mac1, PARP9 MD1 and MD2, MacroD1, or ARH1.

Đukić et al., Sci. Adv. 9, eadi2687 (2023) 13 September 2023 5 of 18

S C I E N C E A D VA N C E S | R E S E A R C H A R T I C L E

Fig. 4. PARP14 ADPr is reversed by its own macrodomain 1. (A) U2OS cells were transfected with the indicated plasmids in the presence or absence of PARP14
inhibitor (PARP14i). Cell lysates and green fluorescent protein (GFP)–immunoprecipitations (GFP-IPs) were examined by Western blotting using the indicated antibodies.
(B) U2OS cells were transfected with the indicated plasmids in the presence of PARP14 or PARG inhibitor. Cell lysates were examined by Western blotting with the
indicated antibodies. For all blots, tubulin was used as a loading control.

modifies a number of different proteins in the cells. In all cases, signal, suggesting that there is no major extension of PARP14-
treatment with a specific PARP14 inhibitor suppressed the increase derived mono-ADPr into polymers (Fig. 4B). PARP14 is an IFN-
in mono-ADPr, indicating that ADPr signal is a result of PARP14 induced protein (fig. S3B) (35). To test whether we can detect
catalytic activity. In contrast to protein ADPr, we were unable to PARP14-dependent ADPr endogenously, we stimulated an immu-
detect any nucleic acid ADPr under the same conditions (fig. nogenic cell line, A549, with IFN-γ and compared the pattern of
S3D). The ADPr pattern seen upon overexpression of the ADP-ribosylated proteins. After stimulation, we could again see
PARP14 MD1 mutant suggests that PARP14 modifies several differ- several strongly ADP-ribosylated proteins and that this was abol-
ent proteins in the cells, while the hydrolytic activity of MD1 ished upon treatment of cells with the PARP14 inhibitor (fig.
removes these modifications. A pull-down experiment further sug- S3E). Our data show that PARP14 is one of the most robustly
gested that the signal around 200 kDa belongs to automodified active ART of proteins in human cell extracts upon IFN stimulation.
PARP14. Addition of PARG inhibitor did not affect the PARP14 Together, these results suggest that PARP14 is a highly active mono-

Đukić et al., Sci. Adv. 9, eadi2687 (2023) 13 September 2023 6 of 18

S C I E N C E A D VA N C E S | R E S E A R C H A R T I C L E

ART for proteins, but the level of ADPr is tightly regulated by its MS-based affinity purification reveals ADPr targets
own hydrolytic macrodomain and by induction of immune regulated by PARP14
responses. To determine the ADPr targets of the PARP14 and, in particular, of
the hydrolytic MD1, we expressed PARP14 WT or PARP14 MD1
PARP14-derived ADPr can be reversed by hydrolytic mutant in 293T cells and enriched ADP-ribosylated proteins with
macrodomains the Af1521 macrodomain (56), a specific ADP-ribose binding
While PARP14 MD1 appears to be the dominant hydrolase control- domain (Fig. 7A). The enriched proteins were then digested into
ling PARP14-catalyzed ADPr in cells, we also tested whether peptides and analyzed by MS to identify differential binders.
PARP14 ADPr could be reversed by several other human hydrolases Overall, the coefficient of variation was low (<15%) within groups
in a cellular context. For this, we coexpressed the PARP14 MD1 (fig. S4A), and they show a high correlation (fig. S4B), suggesting
mutant, which shows the strongest increase in mono-ADPr signal excellent reproducibility, while distinct clusters can be observed
upon overexpression, together with different FLAG-tagged human between groups by principal components analysis (fig. S4C).
hydrolases (Fig. 5). First, we confirmed a strong increase in mono- ADP-ribosylated target proteins were efficiently purified
ADPr upon expression of PARP14 MD1 mutant and that this could (Fig. 7B), with hundreds of proteins significantly enriched over
be inhibited by treatment with PARP14i. Next, we compared the background control. As a whole, the majority of the proteins
PARP14-dependent ADPr in the presence of the human hydrolases are similarly enriched by the Af1521 macrodomain in PARP14
MacroD1, MacroD2, TARG1, and PARG. While MacroD1 could WT– and PARP14 MD1 mutant–expressing cells (Fig. 7B and fig.
quite efficiently remove PARP14 auto- and trans-ADPr, the activity S4B). We were able to identify a subset of ADP-ribosylated proteins
of MacroD2 and TARG1 could only modestly remove PARP14-de- that were enriched in PARP14 MD1 expressing cells compared to
pendent ADPr. Last, PARG overexpression did not reduce ADP- WT, suggesting that their ADPr pattern is regulated by PARP14
ribose, suggesting that PARP14 catalyses largely mono-ADPr as MD1 (Fig. 7C). We confirmed the ADPr of the only characterized
previously suggested (3, 38). These results are consistent with our PARP14 target, PARP13 (Fig. 6) (55).
in vitro data (Fig. 2D) showing that PARP14 auto- and trans-mod- Generally, the ADPr target proteins enriched in cells expressing
ification targets primarily acidic residues for mono-ADPr and that PARP14 MD1 mutant were functionally highly interconnected
its modifications are reversed by ADP-ribosyl hydrolases with activ- (Fig. 7D) and were enriched for terms related to not only ADPr
ity toward acidic residues such as PARP14 MD1 and MacroD1 (54). but also ubiquitin signaling, linking processes such as immunity
Given that SARS2 Mac1 hydrolyses PARP14 ADPr in vitro (PARP13, RNF114, and RNF166), DNA repair (RPA1, XRCC1,
(Fig. 2) (38), we wanted to examine whether this also held true in and PARP1), and tankyrase (TNKS) biology [TNKS1, TNKS2,
a cellular context. When we coexpressed SARS2 Mac1 together with AMOTL1 (angiomotin-like 1), and ZNRD2/SSSCA1 (Sjögren syn-
PARP14 MD1 mutant, we saw a marked reduction in PARP14- drome/scleroderma autoantigen 1); Fig. 7E], biological processes
derived ADPr (Fig. 5, lane 9), again confirming our results that that are known to be functionally coupled to ADPr signaling and
these two hydrolases, PARP14 MD1 and SARS2 Mac1, can both the previously reported functions of PARP14 (25, 34). Together,
remove ADPr catalyzed by PARP14. This supports the available our MS-based methodology and results will enable us to dissect
genetic data showing that PARP14 and SARS2 Mac1 act as a pair the various functional roles in which PARP14 and its MD1
where PARP14-driven ADPr acts to suppress virus proliferation ADPr-hydrolase domain are involved.
while SARS2 Mac1 counteracts PARP14 antiviral activity through
ADP-ribosyl hydrolysis (25). Macrodomain 1 of PARP14 and PARP9 regulate different
cellular activities
PARP13 is a target of both ADP-ribosyl transferase and Next, we aimed to investigate the role of PARP14 MD1 in a cellular
hydrolase activity of PARP14 context. First, we compared the localization of YFP-tagged PARP14
We then sought to examine the hydrolytic activity of PARP14 MD1 and its mutants expressed in U2OS cells. Here, we saw that YFP-
on a known target of PARP14 ADPr. It has previously been reported PARP14 would localize primarily to the cytoplasm, that PARP14
that PARP14 can modify another antiviral PARP, i.e., PARP13 (55). WT would tend to occasionally form large foci around the periph-
To test this, we coexpressed green fluorescent protein (GFP)–tagged ery of the nucleus, and that these showed weak positive staining for
PARP13 with PARP14 WT or PARP14 MD1 mutant (Fig. 6). We ADPr (Fig. 8A). This localization was dependent on the catalytic
observed a strong induction of ADPr on a protein the size of activity of PARP14, as expression of catalytically inactive PARP14
GFP-PARP13 in lysates of cells expressing PARP14 MD1 mutant. resulted in no foci formation. Notably, the foci seen upon expres-
We then performed immunoprecipitation to pull-down GFP- sion the YFP-PARP14 MD1 mutant were smaller, numerous, and
PARP13 and examined its ADP-ribosylation state. We found that stained strongly for ADPr (Fig. 8A). This result suggests the inter-
expression of GFP-PARP13 together with WT PARP14 could mod- play between PARP14 ART activity and PARP14 MD1 hydrolysis
estly increase PARP13 mono-ADPr levels compared to expression activity modulate the cellular localization of PARP14.
with YFP alone. However, PARP13 ADPr was greatly increased We also explored whether PARP14 MD1 would also regulate
upon expression of the PARP14 MD1 mutant. Together, this con- protein interactions. We performed GFP-immunoprecipitation ex-
firms that PARP14 ADP-ribosylates PARP13 in cells and that the periments using 293T cells overexpressing PARP14 WT, MD1
hydrolytic macrodomain MD1 of PARP14 reverses this ADPr. mutant, and the catalytically inactive mutant. We tested for the in-
teraction with DEAD-Box Helcase 6 (DDX6), an RNA helicase that
plays important roles in processing bodies (P-bodies) and in the
immune response to viruses, for example, SARS2 (57), and has
been shown to colocalize with PARP14 (55). Here, we saw that

Đukić et al., Sci. Adv. 9, eadi2687 (2023) 13 September 2023 7 of 18

S C I E N C E A D VA N C E S | R E S E A R C H A R T I C L E

Fig. 5. MacroD1 and SARS2 Mac1 can reverse PARP14 ADPr. U2OS cells were
transfected with the indicated PARP14 plasmid together in the presence or
absence of PARP14i or with FLAG-tagged MacroD1, MacroD2, TARG1, PARG, or
SARS2 Mac1. Cell lysates and GFP-IPs were examined by Western blotting using
the indicated antibodies. Tubulin was used as a loading control.

mutation of PARP14 MD1 resulted in a stronger interaction with

Fig. 6. PARP14 acts to add and remove ADPr on PARP13. 293T cells were co-
DDX6 (Fig. 8B). Moreover, we observed a strong colocalization transfected with GFP-PARP13 and the indicated YFP plasmid. Cell lysates and GFP-
between DDX6 and YFP-PARP14 MD1 mutant by immunofluores- IPs were examined by Western blotting using the indicated antibodies. Ponceau S
cence and that this also colocalized with strong ADPr signal (fig. staining was used to indicate equal loading.
S5A). A stronger interaction with the PARP14 MD1 mutant com-
pared to WT was also observed for DDX3, another P-body element
with PARP14 (31), did not show stable interaction with PARP14,
(Fig. 8B). PARP14 has also been identified as an interactor of the
regardless of MD1 activity (Fig. 8B).
DNA binding DNA repair/replication factor RPA using an MS
PARP14 has been suggested to modulate STAT1 signaling and,
pull-down approach (58). We observe an increased interaction
therefore, modulate IFN-β production (35). To test the role of the
between PARP14 and RPA when MD1 is mutated (Fig. 8B). On
catalytic activity or the hydrolytic activity of PARP14 MD1 on IFNβ
the other hand, HDAC2, which has been suggested to interact

Đukić et al., Sci. Adv. 9, eadi2687 (2023) 13 September 2023 8 of 18

S C I E N C E A D VA N C E S | R E S E A R C H A R T I C L E

Fig. 7. MS identification of ADP-ri-

bosylated proteins regulated by
PARP14 macrodomain 1. (A) Over-
view of the experimental design. (B)
Scatter plot analysis of proteins en-
riched specifically by the ADPr-binding
Af1521 macrodomain. The mean diff-
erence in abundance between proteins
enriched by the Af1521 compared to
control beads for cells expressing WT
PARP14 is plotted against cells ex-
pressing PARP14 MD1 mutant. The
color scale represents the normalized
kernel density estimation of the data.
(C) Analysis of proteins specifically en-
riched for ADPr in cells expressing
PARP14 MD1 mutant compared to WT.
The volcano plot shows the sample
conditions (x axis), plotted against the
corresponding P value resulting from
two-tailed Student’s t testing (y axis).
Proteins significantly down- or up-reg-
ulated [false discovery rate (FDR) < 0.05,
s0 = 0.1] are represented as blue or red
dots, respectively. N = 5. (D) STRING
network visualizing functional interac-
tions (edges) between proteins (nodes)
significantly enriched in cells express-
ing PARP14 MD1 mutant over WT. The
thickness of the edges corresponds to
their score, and the default STRING
clustering confidence score cutoff of
0.4 was used to determine whether two
nodes were functionally related. Pro-
teins were colored according to the
UniProt keyword displayed on the
figure legend if it was significantly en-
riched. (E) Gene set enrichment analy-
sis showing UniProt keywords
functionally enriched in the MD1-
specific network [highlighted in (D)].
Significantly enriched terms were de-
termined by Fisher exact testing,
testing whether categorical terms
found in the MD1-specific network
were functionally enriched over terms
found in the background, which was
defined as all proteins significantly en-
riched by Af1521 over bead controls.
Terms were determined to be signifi-
cant with a Benjamini-Hochberg mul-
tiple-hypotheses corrected P value
<0.05. The terms are ranked by their
functional enrichment over the background in descending order and colored by their corresponding Benjamini-Hochberg–adjusted P values.

expression, we expressed YFP-PARP14 WT, MD1 mutant, or cata- Last, we looked at the model of PARP9 and examined the role of
lytic mutant and used quantitative polymerase chain reaction its macrodomain 1 (MD1). It has been previously shown that
(qPCR) to examine the changes to IFNβ after poly(I:C) treatment, PARP9, as well as its binding partner DTX3L, recruits to sites of
which has been previously shown to induce IFNβ expression (59). DNA damage and will modify proteins at DNA breaks (42, 43).
We found that expression of PARP14 catalytic mutant did not alter Here, we examined the recruitment of YFP-PARP9 WT and the
IFNβ expression compared to YFP alone, while both PARP14 WT MD1 mutant to sites of laser microirradiation. While we observed
and the MD1 mutant dampened IFNβ expression (fig. S5B). a robust recruitment of PARP9 WT to sites of damage, mutation of
PARP9 MD1 resulted in a strong impairment of recruitment to sites

Đukić et al., Sci. Adv. 9, eadi2687 (2023) 13 September 2023 9 of 18

S C I E N C E A D VA N C E S | R E S E A R C H A R T I C L E

Fig. 8. PARP14 and PARP9 macrodomain 1 regulates various cellular functions. (A) Confocal images of U2OS cells expressing YFP or YFP-PARP14 WT, MD1 mutant
(MD1 mut), or catalytically inactive mutant (cat mut) stained with Hoechst (blue) and YFP (green) and for ADPr (poly/mono) (red). Scale bars, 5 μm. (B) 293T cells were
cotransfected with the indicated YFP plasmid. Cell lysates and GFP-IPs were examined by Western blotting using the indicated antibodies. Ponceau S staining was used to
indicate equal loading. (C) Confocal images and (D) recruitment kinetics of YFP-PARP9 and YFP-PARP9 macrodomain 1 mutant (MD1 mut) to sites of laser irradiation in the
absence or presence of 1 μM olaparib. Scale bars, 5 μm.

of DNA damage (Fig. 8C). Furthermore, the recruitment of both DISCUSSION

WT and MD1 mutant was completely abolished upon treatment PARP14 is involved in the regulation of several cellular processes
with the PARP1/2 inhibitor, olaparib, suggesting that the recruit- including DNA repair, immune and antiviral response, and RNA
ment of PARP9 to sites of damage is through direct binding to stability (25, 26, 30, 31, 34, 35, 47, 51, 60). However, it is still
ADP-ribose. Together, our results highlight the complexity of largely unclear how PARP14 activity is achieved and regulated.
both ART and ADP-ribosyl hydrolase activity of PARP14 and While PARP14 macrodomain 2 and 3 have been identified as
PARP9 in different cellular functions. binders of mono-ADPr substrates (61), the function of the first

Đukić et al., Sci. Adv. 9, eadi2687 (2023) 13 September 2023 10 of 18

S C I E N C E A D VA N C E S | R E S E A R C H A R T I C L E

macrodomain has not yet been characterized. Our biochemical and MD1–regulated proteins, have been reported to control a large
cellular data along with recent data from others demonstrate that variety of pathways (45) including inhibiting the innate antiviral re-
the PARP14 MD1 has hydrolytic activity and is a major cellular sponse by ADP-ribosylating virus-induced signal adaptor/ mito-
enzyme that controls the levels of PARP14 ADPr (62, 63). This rep- chondrial antiviral signaling protein, which facilitates the
resents a rare example of an ART that also has ADP-ribosyl hydro- recruitment of the E3 ligase RNF146 for subsequent ubiquitination
lase activity. Analogously, PARP9 MD1 also has ADP-ribosyl and degradation (66).
hydrolase activity, but PARP9 lacks the transferase activity (3). One of the major effects of macrodomain 1 from PARP14 that
PARP9 MD1 may contribute to the control of PARP14 ADPr we have been able to identify is the change in cellular localization.
levels, but it could equally control the ADPr levels of other transfer- Ectopic expression of PARP14 with active ART and MD1 domains
ases or under specific cellular conditions. However, the fact that results in large cellular bodies with unknown function in the cyto-
PARP9 and PARP14 appear to be in the same complex and that plasm, which show low levels of colocalization with ADPr (Fig. 8A).
they are expressed from the same genomic region (chromosome Loss of PARP14 ART activity resulted in loss of foci formation and
3q21.1) suggests that the activities of these two PARPs regulate loss of ADPr signal. Mutation of the MD1 ADP-hydrolase domain
the same pathways (35, 47). Together, our findings reveal two addi- resulted in the formation of numerous smaller foci with strong
tional ADP-ribosyl hydrolase enzymes that are present in humans staining with ADPr. Given that viral replication can depend on
and many higher organisms in addition to the already known six the hijacking of cellular vesicles, it would be of interest to identify
ADP-hydrolases: PARG, ARH1/3, MacroD1/2, and TARG1 (11). what pathways or vesicles are altered with the changes to PARP14
Our discovery that PARP14 MD1 and PARP9 MD1 are specific activities. Our data does suggest a possible link to P-bodies with an
for glutamate ADPr may suggest some redundancy with the other increase interaction and colocalization with DDX6, an RNA heli-
glutamate-ADPr targeting enzymes TARG1, MacroD1, and case found in P-bodies (Fig. 8B and fig. S5A) (57); however,
MacroD2 (48, 49). future studies will be needed to confirm this link. The localization
PARP14 MD1 and PARP9 MD1 are the most closely related of PARP9 protein is also dependent on its macrodomain 1. Specif-
human enzyme to SARS2 Nsp3 Mac1, more closely than to the ically, we demonstrate that mutation of macrodomain 1 from
human paralogs MacroD1 and TARG1 (Fig. 1). It is conceivable PARP9 affects its ability to stably associate with sites of DNA
that coronaviruses and some other viruses bearing macrodomains damage (Fig. 8C).
such as alphaviruses and hepatitis E (37, 64) hijacked MD1 at some Our data demonstrate that PARP14 is one of the major mono-
point in evolution and now use it to oppose PARP14 ADPr antiviral ARTs in IFN-stimulated cells (fig. S3E). PARP14 has been shown to
activities. However, it cannot be ruled out that the specificities of modulate both STAT1 phosphorylation and IFNβ transcription (25,
these two macrodomains, at least on some ADPr sites in macromol- 47). In poly(I:C)-stimulated 293T cells, we observe that PARP14
ecules, are different and may have diverged through evolution in the overexpression leads to lower levels of IFNβ transcription, but we
host-virus arms race. Linking to this, both PARP14 and other anti- further demonstrate that this effect is dependent on PARP14 cata-
viral PARPs and viral macrodomains are under positive natural se- lytic activity but not MD1 hydrolytic activity. The decrease in IFNβ
lection (28, 38). Recent genetic data convincingly shows that transcription may be due to lower levels of STAT1 phosphorylation,
PARP14 functions as an antiviral enzyme in a murine coronavirus as PARP14 has been reported to block STAT1 phosphorylation (47).
model and that Mac1 counteracts this activity, antagonizing the in- However, in another model (unstimulated delayed brain tumor
terferon response and enabling viral replication (25, 36). cells), PARP14 overexpression led to an increase in IFNβ expression,
PARP14 has been shown to play important roles in immunity suggesting that the effect of PARP14 can vary between different cell
and replication stress (34, 51). We have used our published MS ap- models and types of stimulation (25).
proach to unbiasedly identify the PARP14 modification protein PARP14 has been associated with the development of inflamma-
targets (56). In particular, we identified a subset of ADP-ribosylated tory diseases such as allergic asthma (31) and inflammatory arterial
targets that are reversed via PARP14 MD1 hydrolase activity diseases (47) and various types of cancer including B cell lymphoma
(Fig. 7). Fittingly, our results show that a majority of the identified (67), prostate cancer (68), and hepatocellular carcinoma (69).
proteins are involved in immunity, for example, PARP13, RNF114, Therefore, PARP14 has emerged as a potential drug target prompt-
and RNF166, which fits the finding that this domain is highly ho- ing the development of several PARP14 inhibitors (39, 70), although
mologous to the viral SARS2 Mac1 (Fig 1). Many of the targets are none targeting PARP14 macrodomain 1. The discovery of the hy-
also associated with the DNA damage response such as PARP1, drolytic activity of PARP14, as well as PARP9, macrodomain 1 po-
XRCC1, and RPA1. Our data also confirms that PARP14 can revers- tentially presents a druggable target, together with SARS2 Nsp3
ibly ADP-ribosylate PARP13 (Figs. 6 and 7C). Although not cata- Mac1, that could be used to manipulate PARP9- and PARP14-de-
lytically active, PARP13 has been implicated in inhibiting the pendent pathways or function as potent antivirals.
replication of multiple classes of viruses including retroviruses Together, our data identify PARP14 as a complex protein with
(65), alphaviruses (23), flaviviruses, and filoviruses (27). It is tempt- the specific domains that enables it to function as a writer (ART),
ing to speculate that PARP14-induced ADPr of PARP13 might be reader (MD2 and MD3) (61), and eraser (MD1) of ADPr in addi-
involved in the regulation of PARP13 activity; namely, increased tion to nucleic acid binding domains (Fig. 1A). This, together with
ADPr of PARP13 might affect its stability and/or binding affinity. the interplay of the PARP9/DTX3L complex and ubiquitylation sig-
However, further studies are required to understand the cross-talk naling (44, 47), is expected to have far-reaching consequences on
between these two antiviral PARPs. the physiology of the cell and human disease.
Furthermore, several PARP14 ADPr hits from our MS analysis
were TNKS PARPs or their targets such as AMOTL1 and ZNRD2/
SSSCA1 (Fig. 7). TNKS1 and TNKS2, both on the list of PARP14

Đukić et al., Sci. Adv. 9, eadi2687 (2023) 13 September 2023 11 of 18

S C I E N C E A D VA N C E S | R E S E A R C H A R T I C L E

MATERIALS AND METHODS subsequently added, and the samples were incubated at 95°C for 3
Hydrolase activity analysis using luminescence detection min. The samples were run on a prerun denaturing urea PAGE gel
of ADPr [20% (w/v) polyacrylamide, 8 M urea, and 1× TBE] at 7 W per gel in
The ADP-ribosylated peptides were chemically synthesized (table 0.5× TBE. The gels were visualized with laser excitation for Cy3 at
S1). Serine-ADPr and arginine-ADPr peptides were synthesized 532 nM using a PharosFX Molecular Imager (Bio-Rad).
as previously described (71, 72). Chemical synthesis of the gluta- In fig. S1B, purified ADP-ribosylated ssDNA was generated and
mate-ADPr peptide was previously described (73). The hydrolytic used as a substrate for the hydrolysis assay. Here, 50 μM ssDNA
assay against the ADP-ribosylated peptides was performed as previ- were incubated with 20 μM PARP14 WWE-CAT and 10 mM
ously described (74). Briefly, 10 μM substrate peptides (arginine- NAD+ in reaction buffer [20 mM Hepes-KOH (pH 7.6), 5 mM
ADPr, serine-ADPr, or glutamate-ADPr) was hydrolyzed using 1 MgCl2, and 1 mM DTT]. The reactions were incubated at 37°C
μM PARP14 MD1, PARP14 MD1mut, PARP14 MD2, PARP14 for 3 hours and stopped by adding proteinase K (50 ng/μl) and
MD3, PARP9 MD1, PARP9 MD2, or S2 Mac1. ARH1, ARH3, 0.15% SDS, followed by incubation at 50°C for 30 min. Then, the
and MacroD1 served as positive controls for arginine-ADPr, reaction mixture was incubated at 95°C for 5 min to inactivate pro-
serine-ADPr, and glutamate-ADPr, respectively. Hydrolysis was teinase K. The reaction was further passed onto a preequilibrated G-
carried out for 1 hour at 30°C in assay buffer [50 mM tris-HCl 25 column to remove the excess NAD+.
(pH 7.5), 200 mM NaCl, 10 mM MgCl2, 1 mM dithiothreitol For dsDNA annealing, 5 μM Cy3-labeled 50 phosphorylated or 30
(DTT), and 0.2 μM Nudix hydrolase 5 (NUDT5) for arginine- phosphorylated E21 ssDNA was annealed with 10 μM various non-
ADPr and serine-ADPr and 50 mM Pipes (pH 6.9), 200 mM phosphorylated oligonucleotides (1:2 molar ratio) to create blunt,
NaCl, 10 mM MgCl2, 1 mM DTT, and 0.2 μM NUDT5 for gluta- gapped, nicked, overhang, and forked dsDNA (fig. S2). The oligo-
mate-ADPr]. The reactions were analyzed using the AMP-Glo assay nucleotides were mixed in annealing buffer [10 mM tris (pH 8.0), 50
kit (Promega) following the manufacturer’s recommendations. Lu- mM NaCl, and 0.5 mM EDTA] and incubated at 95°C for 5 min,
minescence was read using a SpectraMax M5 plate reader with the followed by gradual cooling to 25°C over 1 hour. The samples
SoftMax Pro software (Molecular Devices). Data were analyzed were run on native 20% acrylamide gel to check for annealing
using GraphPad Prism. completion.
For dsDNA modification reaction, 0.25 μM annealed dsDNA
In vitro protein (ADP-ribosyl) hydrolase assay was mixed with 5 μM PARP14-WWE-Cat and 1 mM NAD+ in re-
PARP14 WWE-CAT (1 μM) with or without PARP14 MD3 (2 μM) action buffer [20 mM Hepes (pH 7.5), 50 mM KCl, 5 mM MgCl2,
was incubated with 50 μM NAD+ (spiked with 32P NAD+) in reac- and 1 mM DTT] at 37°C for 90 min. PARP14 inhibitor (0.5 μM) was
tion buffer [50 mM tris-HCl (pH 8.0), 100 mM NaCl, and 2 mM added to stop the reactions. For macrodomain treatment, 4 μM
MgCl2]. Reactions were incubated at 37°C for 3 hours, then PARP14-MD1 was added to the reaction mixture and incubated
stopped by the addition of 0.5 μM PARP14i, and further passed at 37°C for 60 min. Samples were incubated with proteinase K (50
through a preequilibrated G-25 column to remove excess of ng/μl) and 0.15% SDS at 37°C for 30 min, mixed with 2× TBE-urea
NAD+. The flow-through was ADP-ribosylated PARP14 WWE- loading buffer, and incubated at 95°C for 3 min. Subsequently,
CAT and PARP14 MD3, which were used for hydrolase assays. samples were run on a 20% acrylamide TBE-urea gel. The gels
Next, ADP-ribosylated substrates were incubated with PARP14 were visualized with laser excitation for Cy3 at 532 nM using a Phar-
and PARP9 macrodomains (2 μM) for 1 hour. The reactions were osFX Molecular Imager (Bio-Rad).
subsequently stopped by the addition of 4× LDS sample buffer (Life
Technologies) and incubation at 95°C for 5 min. Samples were then Competitive hydrolase assay
analyzed by SDS–polyacrylamide gel electrophoresis (SDS-PAGE) The purified ADP-ribosylated ssDNA was generated as described
and autoradiography. above. The reaction was further passed onto preequilibrated G-25
column to get rid of the excess of NAD+. The flow-through was
In vitro DNA and RNA (ADP-ribosyl) hydrolase assay the mixture of modified and unmodified ssDNA and used in the
DNA and RNA (ADP-ribosyl) hydrolase assays were carried out as competitive hydrolase assay.
described previously (21). All buffers were prepared using deoxyri- ADP-ribosylated PARP14 WWE-CAT (0.5 μM) were incubated
bonuclease/ribonuclease (RNase)–free water and filter-sterilized with or without PARP14 MD1 (2 μM) in the presence or absence of
before use. Briefly, 0.25 μM Cy3–labeled RNA or DNA (table S2) ADP-ribosylated ssDNA (4 μM) in reaction buffer [20 mM Hepes-
was mixed with 500 μM NAD+ and 2 μM PARP14 WWE-CAT. Re- KOH (pH 7.6), 5 mM MgCl2, and 1 mM DTT] at 37°C for indicated
actions were incubated for 1 hour at 30°C. The ADPr reaction was times. The reactions were subsequently stopped by the addition of
terminated by the addition of 0.1 μM PARP14i; the reaction prod- 4× LDS sample buffer (Life Technologies) and incubation at 95°C
ucts were not purified before the addition of the macrodomains for 5 min. Samples were then analyzed by SDS-PAGE and
unless stated specifically (see the purification procedure below). Hy- autoradiography.
drolysis of the ADP-ribosylated DNA or RNA was initiated by the
addition of 4 μM macrodomains from PARP14, PARP9, or SARS2 Plasmids and mutagenesis
Mac1, followed by incubation of the reactions for 30 min at 30°C. Full-length PARP14 cloning was performed by Gateway cloning
Hydrolysis was stopped by adding proteinase K (50 ng/μl) and (Invitrogen) according to the manufacturer’s instructions.
0.15% SDS, followed by incubation for 30 min at 50°C. Tris- PARP14-encoding pEZ-M11 mammalian expression vector (300
borate EDTA urea sample buffer [2×; 8 M urea, 20 μM EDTA ng) obtained from GeneCopoeia was directly set up for BP recom-
(pH 8.0), 2 μM tris-HCl (pH 7.5), and bromophenol blue] was bination reaction with pDONR221 vector without PARP14 insert
amplification. The reaction was stopped by adding 1 μl proteinase

Đukić et al., Sci. Adv. 9, eadi2687 (2023) 13 September 2023 12 of 18

S C I E N C E A D VA N C E S | R E S E A R C H A R T I C L E

K (Invitrogen) and after incubation for 10 min at 37°C. Competent (RBN012759, MedChemExpress), 5 μM PARGi (PDD00017273,
Stable E. coli (New England Biolabs) was transformed with 2 μl of Sigma-Aldrich), or IFN-γ (100 ng/ml; Merck) for 24 hours.
the BP reaction mix. For transfer into the destination vector, 100 ng
of positive pENTR clone DNA was incubated with 100 ng of Immunoprecipitation
pDEST-N-YFP/FRT/TO pcDNA5 and LR Clonase enzyme mix 293T and U2OS cells were collected 24 or 48 hours after transfection
for 2 hours at room temperature. Plasmid DNA was isolated for pos- and washed two times with phosphate-buffered saline (PBS). Cells
itive clones and verified by Sanger sequencing. PARP9 was cloned were lysed with Triton X-100 lysis buffer [50 mM tris-HCl (pH 8.0),
into of pDEST-N-YFP/FRT/TO pcDNA5 using gateway cloning. 100 mM NaCl, and 1% Triton X-100] supplemented with 5 mM
PARP14 and PARP9 macrodomains were cloned into a pNIC28- MgCl2, 0.1% Benzonase (Sigma-Aldrich), protease and phosphatase
Bsa4 vector, which adds an N-terminal His6-TEV cleavage site to inhibitors (Roche), olaparib (Cayman Chemical; 1 μM for U2OS
the proteins to aid protein purification. PARP14 and PARP9 and 2 μM for 293T cells), and 1 μM PARGi PDD00017273
point mutations were introduced through site-directed mutagenesis (Sigma-Aldrich) for 30 min at 4°C. Protein concentrations were de-
PCR using the QuikChange Lightning kit (Agilent) or the Q5 Site- termined by Bradford Protein Assay (Bio-Rad) and normalized for
Directed Mutagenesis Kit (New England Biolabs) with primers de- equal protein amounts. Cell lysates were incubated with GFP-Trap
scribed in table S3 and confirmed by Sanger sequencing. magnetic agarose beads (ChromoTek) on an orbital rotator for 2
Mammalian expression vectors encoding FLAG-MacroD1, hours at 4°C. Beads were pelleted using a magnetic separation
FLAG-MacroD2, FLAG-PARG, FLAG-SARS2 Mac1, and GFP- rack and washed five times with Triton X-100 lysis buffer [50 mM
PARP13 were generated by gateway cloning as described previously tris-HCl (pH 8.0), 800 mM NaCl, and 1% Triton X-100]. Proteins
(54). PaTagRFP-H2B (histone 2B) (75) and GFP-DarT2 (Thermus were eluted with 2× NuPAGE LDS sample buffer (Invitrogen) sup-
aquaticus) (76) were previously described. plemented with DTT (Sigma-Aldrich), boiled for 5 min at 95°C, and
analyzed by Western blotting.
Protein expression and purification
BL21(DE3)-R3-pRARE cells were transformed with PARP14 and Dot blot
PARP9 macrodomain encoding constructs and grown at 37°C in For genomic DNA and total RNA extraction, 1.2 × 106 of U2OS WT
LB medium supplemented with appropriate antibiotics until cells were seeded on a 10-cm dish. Cells were transfected using
optical density at 600 nm (OD600) 0.5 to 0.6, then cooled to 18°C, TransIT-LT1 (Mirus, MIR2300) according to the manufacturer’s
and supplemented with 0.5 mM isopropyl-β-D-thiogalactopyrano- instructions. TARG1 KO cells were transfected with GFP-DarT2
side at an OD600 of 0.8 to induce protein expression overnight. Cells was used as a positive control for DNA-ADPr (76). Cells were har-
were harvested by centrifugation, resuspended in lysis buffer [50 vested in ice-cold PBS 24 hours after transfection and pelleted at
mM Hepes (pH 7.5), 500 mM NaCl, 20 mM imidazole, 5% glycerol, 300g for 5 min in a benchtop centrifuge.
0.5 mM tris(2-carboxyethyl)phosphine (TCEP), and 1:2000 Calbio- Genomic DNA was extracted using the DNeasy Blood & Tissue
chem protease inhibitor cocktail set III), and lysed by sonication. Kit (QIAGEN) according to the manufacturer’s instructions with
Proteins were purified by Ni2+–nitrilotriacetic acid chromatography the addition of RNA digestion with RNase A (10 μg/ml; Invitrogen,
(Jena Bioscience) and eluted stepwise in binding buffer with 40 to 8003089) for 5 min at room temperature before the addition of pro-
250 mM imidazole. Proteins were further purified by size exclusion teinase K. DNA concentration was measured using a spectropho-
chromatography (Superdex 75, GE HealthCare) in a buffer consist- tometer (DeNovix, DS-11 FX), and the concentration was
ing of 50 mM Hepes (pH 7.5), 300 mM NaCl, 5% glycerol, and 0.5 normalized across the samples. Total RNA was extracted from
mM TCEP. PARP14 MD1 was additionally purified by ion ex- cells using the RNeasy Mini Kit (QIAGEN) according to the man-
change chromatography using a HiTrap SP HP (5 ml; GE Health- ufacturer’s instructions. RNA concentration was measured using a
Care) equilibrated in 25 mM Hepes (pH 7.5), 75 mM NaCl, and 0.5 spectrophotometer, and the concentration was normalized across
mM TCEP. The purity of protein preparations was assessed using the samples using RNase-free water (Invitrogen).
SDS-PAGE and Coomassie Brilliant Blue staining, and aliquots Equal amounts, 100 ng, of nucleic acids were dotted onto a ni-
were stored at −80°C until use. trocellulose membrane (0.45 μm; Amersham Protran) and cross-
linked using 1200 J at 254 nm with an ultraviolet cross-linker (Star-
Cell culture linker) (19, 76). The membranes were blocked with blocking buffer
Human U2OS osteosarcoma [American Type Culture Collection [5% nonfat dried milk (w/v) in PBST] for 1 hour before the addition
(ATCC), HTB-96], embryonic kidney 293T (ATCC, CRL-3216), of primary antibodies diluted in blocking buffer for 1 hour at room
and A549 (ATCC, CCl-185) cell lines were purchased from temperature (table S4). The membranes were washed three times
ATCC. Cells were grown in Dulbecco’s modified Eagle’s medium with PBST and then incubated with HRP-conjugated secondary an-
(Sigma-Aldrich) supplemented with 10% fetal bovine serum (FBS; tibodies diluted in blocking buffer for 1 hour at room temperature.
Gibco) and penicillin-streptomycin (100 U/ml; Gibco). All cell lines Blots were washed three times in PBST. Chemiluminescence was de-
were cultured in a humidified atmosphere at 37°C with 5% CO2. tected using the SuperSignal West Dura Extended Duration Sub-
Human embryonic kidney 293T (293T) and U2OS cells were strate (Thermo Fisher Scientific) on Hyperfilm ECL film (Cytiva).
plated in 10-cm dishes 24 hours before cells were transfected with The RhsP2-positive control was created using RhsP2 protein
the indicated plasmids. 293T cells were transfected using PolyFect provided by J. Whitney (77). The RNA-ADPr was synthesized by
(QIAGEN), while U2OS cells were transfected using TransIT-LT1 incubating double-stranded RNA oligonucleotide with 1 μM
Transfection Reagent (Mirus Bio), according to the manufacturer’s RhsP2 and 1 mM NAD+ in 20 mM Hepes-KOH (pH 7.6), 50 mM
protocol. Cells were treated with DMSO, 0.5 μM PARP14i KCl, 5 mM MgCl2, and 1 mM DTT buffer for 1 hour at 37°C.

Đukić et al., Sci. Adv. 9, eadi2687 (2023) 13 September 2023 13 of 18

S C I E N C E A D VA N C E S | R E S E A R C H A R T I C L E

Western blotting purification of the GST-Af1521 macrodomain was performed es-

Cells were lysed, and protein concentration was measured as de- sentially as described previously (56, 84). The beads were incubated
scribed above. Proteins were boiled in 1× NuPAGE LDS sample with the lysates for 8 hours at 4°C tumbling end over end to facilitate
buffer (Invitrogen) with 60 mM DTT (Sigma-Aldrich) and resolved binding between the ADP-ribosylated proteins and the Af1521-
on NuPAGE Novex 4 to 12% bis-tris gels (Invitrogen) in 1× containing beads. After binding, the beads were transferred to
NuPAGE MOPS SDS Running Buffer (Invitrogen) at 150 new tubes three times and were washed two times after each tube
V. Proteins were transferred onto nitrocellulose membranes (Bio- change in RIPA buffer without supplements. Proteins were digested
Rad) using Trans-Blot Turbo Transfer System (Bio-Rad). The mem- on beads with RIPA buffer supplemented with 500 ng of trypsin
branes were stained with Ponceau S Staining Solution (Thermo (sequencing grade; Promega) per sample and incubated overnight
Fisher Scientific) to check the transfer quality, rinsed with water, at 30°C with mild shaking. To separate the peptides from the beads,
and blocked in 5% (w/v) nonfat dried milk in PBS buffer with samples were spun through a 0.45 μM filter (Ultrafree-MC, Milli-
0.1% (v/v) Tween 20 (PBST) for 1 hour at room temperature. pore), and then reduced, and alkylated with 5 mM TCEP and 5
This was followed by overnight incubation with primary antibody mM chloroacetamide for 30 min at room temperature.
as indicated in table S4 at 4°C. The next day, membranes were Following procedures previously described in (85), peptides
washed in PBST and incubated with HRP-conjugated antibodies were subjected to StageTip clean-up by desalting and purifying
at room temperature for 1 hour. Membranes were visualized on Hy- them on C18 discs at high pH. Briefly, quad-layer StageTips were
perfilm ECL films (Cytiva) after adding Pierce ECL Western Blot- prepared using four punch-outs of C18 material (C18, 47 mm;
ting Substrate (Thermo Fisher Scientific). Sigma-Aldrich, Empore SPE Disks). StageTips were equilibrated
using 100 μl of methanol, 100 μl of 80% acetonitrile (ACN) in
Sequence and phylogenetic analysis 200 mM ammonium hydroxide, and 75 μl of 50 mM ammonium
For multiple-sequence alignments, Jalview v2 (78) and MAFFT7 hydroxide two times. Samples were supplemented with one-tenth
(79) were used. The phylogenetic tree of macrodomains was gener- volume of 200 mM ammonium hydroxide (pH, >10), just before
ated with SplitsTree4 (v4.15.1) (80) using the neighbor-joining loading them on the StageTip. The StageTips were subsequently
method and confidence levels estimated using 1000 cycles of the washed twice with 150 μl of 50 mM ammonium hydroxide and af-
bootstrap method. Pairwise identities were determined using the terward eluted using 80 μl of 30% ACN in 50 mM ammonium hy-
Needleman-Wunsch algorithm implemented as part of the Europe- droxide. All samples were dried to completion in Protein LoBind
an Molecular Biology Laboratory–European Bioinformatics Insti- tubes (Eppendorf ) using a SpeedVac for 2 hours at 60°C, after
tute search and sequence analysis server (81). Structural which the dried peptides were dissolved using 11 μl of 0.1%
alignments and analyses, as well as figure preparation, were formic acid and stored at 20°C until MS analysis.
carried out using PyMOL (Molecular Graphics System, version
2.3.3; Schrӧdinger LLC). PARP domain architecture was visualized MS analysis
using IBS illustrator (82). Following procedures previously described in (86), MS samples
were analyzed on an EASY-nLC 1200 system (Thermo Fisher Sci-
Reverse transcription qPCR entific) coupled to an Orbitrap Exploris 480 mass spectrometer
Following procedures previously described in (83), total RNAs from (Thermo Fisher Scientific). Separation of peptides was performed
293T cells were purified with the RNeasy Plus Micro Kit using 20-cm columns (internal diameter, 75 μm) packed in-house
(QIAGEN), and then, 0.5 μg of total RNA was used for cDNA syn- with ReproSil-Pur 120 C18-AQ 1.9-μm beads (Dr. Maisch). Elution
thesis with QuantiTect Reverse Transcription Kit according to the of peptides from the column was achieved using a gradient ranging
manufacturer’s instructions. The cDNAs were detected by quanti- from buffer A (0.1% formic acid) to buffer B (80% ACN in 0.1%
tative real-time PCR using the Rotor-Gene SYBR Green PCR Kit formic acid) at a flow of 250 nl/min. The gradient length was 80
and the Rotor-Gene Q (QIAGEN). Primer pairs for reverse tran- min per sample, including ramp-up and wash-out, with an analyt-
scription qPCR are given in table S5. The relative gene expression ical gradient of 60 min ranging from 5% B to 38% B. Analytical
analysis of INFβ was performed using the ddCt method normalized columns were heated to 40°C using a column oven, and ionization
to hypoxanthine phosphoribosyltransferase 1 (HPRT1). was achieved using a Nanospray Flex NG ion source. Spray voltage
was set to 2 kV; ion transfer tube temperature was set to 275°C, and
Enrichment of ADPr proteins for MS analysis radio frequency funnel level was set to 40%. Full scan range was set
293T cells were cultured as five technical replicates in a humidified to 300 to 1300 mass/charge ratio (m/z). MS1 resolution was set to
incubator at 37°C with 5% CO2 in 15-cm dishes. The cells were 120,000. MS1 AGC target was set to “200” (2,000,000 charges), and
transfected with plasmids encoding either YFP-PARP14 WT or MS1 maximum injection time was set to “Auto.” Precursors with
YFP-PARP14 MD1 mutant sequences as described above. Cells charges 2 to 6 were selected for fragmentation using an isolation
were gently washed with PBS 24 hours after transfection and then width of 1.3 m/z and fragmented using higher-energy collision dis-
immediately lysed in radioimmunoprecipitation assay (RIPA) association with normalized collision energy of 25. Precursors were
buffer [50 mM tris-HCl (pH 8.0), 100 mM NaCl, and 1% Triton excluded from resequencing by setting a dynamic exclusion of 80 s.
X-100] supplemented with 5 mM MgCl2, 0.1% Benzonase MS2 resolution was set to 30,000. MS2 AGC target was set to 200
(Sigma-Aldrich), protease and phosphatase inhibitors (Roche), 2 (200,000 charges). Intensity threshold was set to 360,000 charges
μM olaparib (Cayman Chemical), and 1 μM PARGi per second. MS2 maximum injection time was set to Auto, and
PDD00017273 (Sigma-Aldrich) for 30 min at 4°C. TopN was set to 13.
Lysates were split evenly across either Af1521-linked glutathione
Sepharose 4B beads or empty control beads (Cytiva). The

Đukić et al., Sci. Adv. 9, eadi2687 (2023) 13 September 2023 14 of 18

S C I E N C E A D VA N C E S | R E S E A R C H A R T I C L E

Data processing numerical aperture oil-immersion objective lens and a Prime BSI
All RAW files were analyzed using MaxQuant software (version scientific complementary metal-oxide semiconductor camera. The
1.5.3.30) using default settings, except with label-free quantification fluorescence of YFP and PaTagRFP-H2B was excited with 488-nm
enabled (87). The human FASTA database used in this study was and 561-nm solid state lasers, respectively, and fluorescence detec-
downloaded from UniProt on 25 June 2023 (UP000005640). All tion was achieved with band-pass filters adapted to the fluorophore
data were filtered by posterior error probability to achieve a false emission spectra. Laser microirradiation at 405 nm was performed
discovery rate (FDR) of <1% (default) at both the peptide-spectrum along a 15-μm line through the nucleus for 250 ms using a single-
match and the protein assignment levels. point scanning head (Olympus cellFRAP) coupled to the epifluor-
escence backboard of the microscope. To ensure reproducibility,
Data filtering and statistical analysis laser power at 405 nm was measured at the beginning of each exper-
Beyond automatic filtering and FDR correction applied by Max- iment and set to 110 μW at the sample level. For time course exper-
Quant during data processing, data S1 and S2 were manually strin- iments, images were collected every 5 s. For the live-cell imaging
gently filtered to ensure robust quantification of differential experiments, cells were maintained at 37°C with a heating
experimental groups using the freely available Perseus software chamber. Protein accumulation at sites of damage (Ad) was then cal-
(88). This filtering includes log2 transformations, n = 5 filtering culated as
within at least one group, filtering proteins identified with less
Id I bg
than two peptides, column-wise imputation (down shift, 1.8; Ad ¼
width, 0.3), two-sample t tests for differential expression, and en- In I bg
richment analysis through FDR-controlled Fisher exact testing. The intensity within the microirradiated area was then normal-
Principal components analysis was performed using the R ized to the intensity before damage induction. Photoactivated H2B
(version 4.3.1 “Beagle Scouts”) function prcomp, after transposing was used as a reference to indicate where irradiation had occurred.
and z score normalizing the rows following the filters described For immunofluorescence, U2OS cells were plated on an eight-
above. Plots were generated using R and the “ggplots” package well μ-slide glass bottom chamber slide, while 293T cells were
(version 3.4.2). plated on poly-L-lysine–coated coverslips. Cells were transfected
UniProt entries complete with Gene Ontology and keywords as described above. Cells were fixed 24 hours after transfection for
were downloaded concomitantly with the fasta file used to build 20 min at −20°C with ice-cold methanol:acetone (1:1). Cells were
the search space in MaxQuant, and these were mapped to the fil- washed twice with PBS before being blocked for 60 min in blocking
tered dataset to perform gene set enrichment analysis. To this buffer (3% bovine serum albumin in PBS + 0.2% Tween). Cells were
end, proteins that were significantly enriched in the Af1521 group incubated in primary antibody overnight at 4°C before being
over the control were assigned as the background dataset, while the washed three times with PBS + 0.1% Triton. Cells were incubated
proteins that were significantly up-regulated in the MD1 samples with secondary antibody in blocking buffer with Hoechst 33342
over the WT was considered foreground. The significantly enriched (1 μg/ml). Cells were washed three times with PBS + 0.1% Triton
terms, found in data S3, was found by Fisher’s exact test with Ben- before being mounted on slides using MOWOIL (Merck) or
jamini-Hochberg FDR corrected P values >5%. being imaged directly. Immunofluorescence was carried on
The online STRING database (version 11.5) was used for gener- Olympus IX-83 inverted microscope as described above using
ation of protein interaction (89), and Cytoscape (version 3.10.0) was 405-, 488-, and 633-nm solid-state lasers and with band-pass
used for manual annotation and visualization of the STRING (90) filters adapted to the fluorophore emission spectra.
together with the Omics Visualizer App (91).

Data availability Supplementary Materials

The MS proteomics data have been deposited to the ProteomeX- This PDF file includes:
change Consortium via the PRIDE (92) partner repository with Figs. S1 to S5
the dataset identifier PXD043452. Tables S1 to S5
Legends for data S1 to S3

Microscopy Other Supplementary Material for this

Following procedures previously described in (83), U2OS cells were manuscript includes the following:
plated on an eight-well μ-slide glass bottom chamber slide (ibidi) Data S1 to S3
and transfected with an expression plasmid for YFP-PARP9 WT
or MD1 mutant together with a plasmid for expression of photoac-
tivable tagged H2B (PaTR-H2B) 48 hours before imaging. For cell REFERENCES AND NOTES
sensitization before laser irradiation at 405 nm, growth medium was 1. R. Gupte, Z. Liu, W. L. Kraus, PARPs and ADP-ribosylation: Recent advances linking mo-
lecular functions to biological outcomes. Genes Dev. 31, 101–126 (2017).
aspirated from the Lab-Tek and replaced with fresh medium con-
2. J. Groslambert, E. Prokhorova, I. Ahel, ADP-ribosylation of DNA and RNA. DNA Repair (Amst)
taining Hoechst 33342 (0.3 μg/ml). Immediately before imaging, 105, 103144 (2021).
the Hoechst containing medium was replaced with imaging 3. S. Vyas, I. Matic, L. Uchima, J. Rood, R. Zaja, R. T. Hay, I. Ahel, P. Chang, Family-wide analysis
media [phenol red–free Leibovitz’s L-15 medium (Life Technolo- of poly(ADP-ribose) polymerase activity. Nat. Commun. 5, 4426 (2014).
gies) supplemented with 20% FBS, penicillin (100 μg/ml), and strep- 4. B. Lüscher, I. Ahel, M. Altmeyer, A. Ashworth, P. Bai, P. Chang, M. Cohen, D. Corda, F. Dantzer,
tomycin (100 U/ml). Live-cell microscopy was carried out on an M. D. Daugherty, T. M. Dawson, V. L. Dawson, S. Deindl, A. R. Fehr, K. L. H. Feijs, D. V. Filippov,
J.-P. Gagné, G. Grimaldi, S. Guettler, N. C. Hoch, M. O. Hottiger, P. Korn, W. L. Kraus,
Olympus IX-83 inverted microscope equipped with a Yokogawa
A. Ladurner, L. Lehtiö, A. K. L. Leung, C. J. Lord, A. Mangerich, I. Matic, J. Matthews, G.-
SoRa superresolution spinning-disk head, a UPlanAop 60x/1.5

Đukić et al., Sci. Adv. 9, eadi2687 (2023) 13 September 2023 15 of 18

S C I E N C E A D VA N C E S | R E S E A R C H A R T I C L E

L. Moldovan, J. Moss, G. Natoli, M. L. Nielsen, M. Niepel, F. Nolte, J. Pascal, B. M. Paschal, 29. M. J. Suskiewicz, D. Munnur, Ø. Strømland, J.-C. Yang, L. E. Easton, C. Chatrin, K. Zhu,
K. Pawłowski, G. G. Poirier, S. Smith, G. Timinszky, Z.-Q. Wang, J. Yélamos, X. Yu, R. Zaja, D. Baretić, S. Goffinont, M. Schuller, W.-F. Wu, J. M. Elkins, D. Ahel, S. Sanyal, D. Neuhaus,
M. Ziegler, ADP-ribosyltransferases, an update on function and nomenclature. FEBS J. 289, I. Ahel, Updated protein domain annotation of the PARP protein family sheds new light on
7399–7410 (2022). biological function. Nucleic Acids Res., gkad514 (2023).
5. O. Leidecker, J. J. Bonfiglio, T. Colby, Q. Zhang, I. Atanassov, R. Zaja, L. Palazzo, A. Stockum, 30. S. Goenka, M. Boothby, Selective potentiation of Stat-dependent gene expression by
I. Ahel, I. Matic, Serine is a new target residue for endogenous ADP-ribosylation on his- collaborator of Stat6 (CoaSt6), a transcriptional cofactor. Proc. Natl. Acad. Sci. U.S.A. 103,
tones. Nat. Chem. Biol. 12, 998–1000 (2016). 4210–4215 (2006).
6. D. D’Amours, S. Desnoyers, I. D’Silva, G. G. Poirier, Poly(ADP-ribosyl)ation reactions in the 31. P. Mehrotra, J. P. Riley, R. Patel, F. Li, L. Voss, S. Goenka, PARP-14 functions as a transcrip-
regulation of nuclear functions. Biochem. J. 342 (Pt. 2), 249–268 (1999). tional switch for Stat6-dependent gene activation. J. Biol. Chem. 286, 1767–1776 (2011).
7. G. I. Karras, G. Kustatscher, H. R. Buhecha, M. D. Allen, C. Pugieux, F. Sait, M. Bycroft, 32. C. M. Nicolae, E. R. Aho, K. N. Choe, D. Constantin, H.-J. Hu, D. Lee, K. Myung, G.-L. Moldovan,
A. G. Ladurner, The macro domain is an ADP-ribose binding module. EMBO J. 24, A novel role for the mono-ADP-ribosyltransferase PARP14/ARTD8 in promoting homolo-
1911–1920 (2005). gous recombination and protecting against replication stress. Nucleic Acids Res. 43,
8. P. A. DaRosa, Z. Wang, X. Jiang, J. N. Pruneda, F. Cong, R. E. Klevit, W. Xu, Allosteric activation 3143–3153 (2015).
of the RNF146 ubiquitin ligase by a poly(ADP-ribosyl)ation signal. Nature 517, 33. A. Dhoonmoon, E. M. Schleicher, K. E. Clements, C. M. Nicolae, G. L. Moldovan, Genome-
223–226 (2015). wide CRISPR synthetic lethality screen identifies a role for the ADP-ribosyltransferase
9. I. Ahel, D. Ahel, T. Matsusaka, A. J. Clark, J. Pines, S. J. Boulton, S. C. West, Poly(ADP-ribose)- PARP14 in DNA replication dynamics controlled by ATR. Nucleic Acids Res. 48,
binding zinc finger motifs in DNA repair/checkpoint proteins. Nature 451, 81–85 (2008). 7252–7264 (2020).
10. F. Teloni, M. Altmeyer, Readers of poly(ADP-ribose): Designed to be fit for purpose. Nucleic 34. A. Dhoonmoon, C. M. Nicolae, G.-L. Moldovan, The KU-PARP14 axis differentially regulates
Acids Res. 44, 993–1006 (2016). DNA resection at stalled replication forks by MRE11 and EXO1. Nat. Commun. 13,
11. J. G. M. Rack, L. Palazzo, I. Ahel, (ADP-ribosyl)hydrolases: Structure, function, and biology. 5063 (2022).
Genes Dev. 34, 263–284 (2020). 35. G. Caprara, E. Prosperini, V. Piccolo, G. Sigismondo, A. Melacarne, A. Cuomo, M. Boothby,
12. M.-F. Langelier, J. L. Planck, S. Roy, J. M. Pascal, Structural basis for DNA damage-dependent M. Rescigno, T. Bonaldi, G. Natoli, PARP14 controls the nuclear accumulation of a subset of
poly(ADP-ribosyl)ation by human PARP-1. Science 336, 728–732 (2012). type I IFN-inducible proteins. J. Immunol. 200, 2439–2454 (2018).
13. I. Gibbs-Seymour, P. Fontana, J. G. M. Rack, I. Ahel, HPF1/C4orf27 is a PARP-1-interacting 36. R. Fehr Anthony, R. Channappanavar, G. Jankevicius, C. Fett, J. Zhao, J. Athmer,
protein that regulates PARP-1 ADP-ribosylation activity. Mol. Cell 62, 432–442 (2016). K. Meyerholz David, I. Ahel, S. Perlman, The conserved coronavirus macrodomain pro-
14. M. J. Suskiewicz, F. Zobel, T. E. H. Ogden, P. Fontana, A. Ariza, J. C. Yang, K. Zhu, L. Bracken, motes virulence and suppresses the innate immune response during severe acute respi-
W. J. Hawthorne, D. Ahel, D. Neuhaus, I. Ahel, HPF1 completes the PARP active site for DNA ratory syndrome coronavirus infection. mBio 7, e01721-16 (2016).
damage-induced ADP-ribosylation. Nature 579, 598–602 (2020). 37. C. Li, Y. Debing, G. Jankevicius, J. Neyts, I. Ahel, B. Coutard, B. Canard, Viral macro domains
15. J. J. Bonfiglio, P. Fontana, Q. Zhang, T. Colby, I. Gibbs-Seymour, I. Atanassov, E. Bartlett, reverse protein ADP-ribosylation. J. Virol. 90, 8478–8486 (2016).
R. Zaja, I. Ahel, I. Matic, Serine ADP-ribosylation depends on HPF1. Mol. Cell 65, 38. J. G. M. Rack, V. Zorzini, Z. Zhu, M. Schuller, D. Ahel, I. Ahel, Viral macrodomains: A structural
932–940.e6 (2017). and evolutionary assessment of the pharmacological potential. Open Biol. 10,
16. L. Palazzo, M. J. Suskiewicz, I. Ahel, Serine ADP-ribosylation in DNA-damage response 200237 (2020).
regulation. Curr. Opin. Genet. Dev. 71, 106–113 (2021). 39. L. B. Schenkel, J. R. Molina, K. K. Swinger, R. Abo, D. J. Blackwell, A. Z. Lu, A. E. Cheung,
17. P. Fontana, J. J. Bonfiglio, L. Palazzo, E. Bartlett, I. Matic, I. Ahel, Serine ADP-ribosylation W. D. Church, K. Kunii, K. G. Kuplast-Barr, C. R. Majer, E. Minissale, J. R. Mo, M. Niepel, C. Reik,
reversal by the hydrolase ARH3. eLife 6, e28533 (2017). Y. Ren, M. M. Vasbinder, T. J. Wigle, V. M. Richon, H. Keilhack, K. W. Kuntz, A potent and
selective PARP14 inhibitor decreases protumor macrophage gene expression and elicits
18. M. Schuller, R. E. Butler, A. Ariza, C. Tromans-Coia, G. Jankevicius, T. D. W. Claridge,
inflammatory responses in tumor explants. Cell Chem. Biol. 28, 1158–1168.e13 (2021).
S. L. Kendall, S. Goh, G. R. Stewart, I. Ahel, Molecular basis for DarT ADP-ribosylation of a
DNA base. Nature 596, 597–602 (2021). 40. Y. Zhang, D. Mao, W. T. Roswit, X. Jin, A. C. Patel, D. A. Patel, E. Agapov, Z. Wang,
R. M. Tidwell, J. J. Atkinson, G. Huang, R. McCarthy, J. Yu, N. E. Yun, S. Paessler, T. G. Lawson,
19. G. Jankevicius, A. Ariza, M. Ahel, I. Ahel, The toxin-antitoxin system DarTG catalyzes re-
N. S. Omattage, T. J. Brett, M. J. Holtzman, PARP9-DTX3L ubiquitin ligase targets host
versible ADP-ribosylation of DNA. Mol. Cell 64, 1109–1116 (2016).
histone H2BJ and viral 3C protease to enhance interferon signaling and control viral in-
20. D. Munnur, I. Ahel, Reversible mono-ADP-ribosylation of DNA breaks. FEBS J. 284,
fection. Nat. Immunol. 16, 1215–1227 (2015).
4002–4016 (2017).
41. J. Xing, A. Zhang, Y. Du, M. Fang, L. J. Minze, Y. J. Liu, X. C. Li, Z. Zhang, Identification of
21. D. Munnur, E. Bartlett, P. Mikoľević, I. T. Kirby, J. G. M. Rack, A. Mikǒ, M. S. Cohen, I. Ahel,
poly(ADP-ribose) polymerase 9 (PARP9) as a noncanonical sensor for RNA virus in dendritic
Reversible ADP-ribosylation of RNA. Nucleic Acids Res. 47, 5658–5669 (2019).
cells. Nat. Commun. 12, 2681 (2021).
22. I. Talhaoui, N. A. Lebedeva, G. Zarkovic, C. Saint-Pierre, M. M. Kutuzov, M. V. Sukhanova,
42. Q. Yan, J. Ding, S. J. Khan, L. N. Lawton, M. A. Shipp, DTX3L E3 ligase targets p53 for
B. T. Matkarimov, D. Gasparutto, M. K. Saparbaev, O. I. Lavrik, A. A. Ishchenko, Poly(ADP-
degradation at poly ADP-ribose polymerase-associated DNA damage sites. iScience 26,
ribose) polymerases covalently modify strand break termini in DNA fragments in vitro.
106444 (2023).
Nucleic Acids Res. 44, 9279–9295 (2016).
43. Q. Yan, R. Xu, L. Zhu, X. Cheng, Z. Wang, J. Manis, M. A. Shipp, BAL1 and its partner E3 ligase,
23. J. A. Kerns, M. Emerman, H. S. Malik, Positive selection and increased antiviral activity as-
BBAP, link poly(ADP-ribose) activation, ubiquitylation, and double-strand DNA repair in-
sociated with the PARP-containing isoform of human zinc-finger antiviral protein. PLOS
dependent of ATM, MDC1, and RNF8. Mol. Cell. Biol. 33, 845–857 (2013).
Genet. 4, e21 (2008).
44. K. Zhu, M. J. Suskiewicz, A. Hloušek-Kasun, H. Meudal, A. Mikǒ, V. Aucagne, D. Ahel, I. Ahel,
24. Y. Zhu, G. Chen, F. Lv, X. Wang, X. Ji, Y. Xu, J. Sun, L. Wu, Y.-T. Zheng, G. Gao, Zinc-finger
DELTEX E3 ligases ubiquitylate ADP-ribosyl modification on protein substrates. Sci. Adv. 8,
antiviral protein inhibits HIV-1 infection by selectively targeting multiply spliced viral
eadd4253 (2022).
mRNAs for degradation. Proc. Natl. Acad. Sci. U.S.A. 108, 15834–15839 (2011).
45. A. G. Thorsell, T. Ekblad, T. Karlberg, M. Löw, A. F. Pinto, L. Trésaugues, M. Moche,
25. M. E. Grunewald, Y. Chen, C. Kuny, T. Maejima, R. Lease, D. Ferraris, M. Aikawa, C. S. Sullivan,
M. S. Cohen, H. Schüler, Structural basis for potency and promiscuity in poly(ADP-ribose)
S. Perlman, A. R. Fehr, The coronavirus macrodomain is required to prevent PARP-mediated
polymerase (PARP) and tankyrase inhibitors. J. Med. Chem. 60, 1262–1271 (2017).
inhibition of virus replication and enhancement of IFN expression. PLOS Pathog. 15,
46. D. Chen, M. Vollmar, M. N. Rossi, C. Phillips, R. Kraehenbuehl, D. Slade, P. V. Mehrotra, F. von
e1007756 (2019).
Delft, S. K. Crosthwaite, O. Gileadi, J. M. Denu, I. Ahel, Identification of macrodomain
26. L. S. Voth, J. J. O’Connor, C. M. Kerr, E. Doerger, N. Schwarting, P. Sperstad, D. K. Johnson,
proteins as novel O-acetyl-ADP-ribose deacetylases. J. Biol. Chem. 286,
A. R. Fehr, Unique mutations in the murine hepatitis virus macrodomain differentially
13261–13271 (2011).
attenuate virus replication, indicating multiple roles for the macrodomain in coronavirus
47. H. Iwata, C. Goettsch, A. Sharma, P. Ricchiuto, W. W. B. Goh, A. Halu, I. Yamada, H. Yoshida,
replication. J. Virol. 95, e0076621 (2021).
T. Hara, M. Wei, N. Inoue, D. Fukuda, A. Mojcher, P. C. Mattson, A.-L. Barabási, M. Boothby,
27. S. Müller, P. Möller, J. Bick Matthew, S. Wurr, S. Becker, S. Günther, M. Kümmerer Beate,
E. Aikawa, S. A. Singh, M. Aikawa, PARP9 and PARP14 cross-regulate macrophage activation
Inhibition of filovirus replication by the zinc finger antiviral protein. J. Virol. 81,
via STAT1 ADP-ribosylation. Nat. Commun. 7, 12849 (2016).
2391–2400 (2007).
48. F. Rosenthal, K. L. Feijs, E. Frugier, M. Bonalli, A. H. Forst, R. Imhof, H. C. Winkler, D. Fischer,
28. M. D. Daugherty, J. M. Young, J. A. Kerns, H. S. Malik, Rapid evolution of PARP genes
A. Caflisch, P. O. Hassa, B. Lüscher, M. O. Hottiger, Macrodomain-containing proteins are
suggests a broad role for ADP-ribosylation in host-virus conflicts. PLOS Genet. 10,
new mono-ADP-ribosylhydrolases. Nat. Struct. Mol. Biol. 20, 502–507 (2013).
e1004403 (2014).

Đukić et al., Sci. Adv. 9, eadi2687 (2023) 13 September 2023 16 of 18

S C I E N C E A D VA N C E S | R E S E A R C H A R T I C L E

49. R. Sharifi, R. Morra, C. D. Appel, M. Tallis, B. Chioza, G. Jankevicius, M. A. Simpson, I. Matic, 70. M. Schuller, K. Riedel, I. Gibbs-Seymour, K. Uth, C. Sieg, A. P. Gehring, I. Ahel, F. Bracher,
E. Ozkan, B. Golia, M. J. Schellenberg, R. Weston, J. G. Williams, M. N. Rossi, H. Galehdari, B. M. Kessler, J. M. Elkins, S. Knapp, Discovery of a selective allosteric inhibitor targeting
J. Krahn, A. Wan, R. C. Trembath, A. H. Crosby, D. Ahel, R. Hay, A. G. Ladurner, G. Timinszky, macrodomain 2 of polyadenosine-diphosphate-ribose polymerase 14. ACS Chem. Biol. 12,
R. S. Williams, I. Ahel, Deficiency of terminal ADP-ribose protein glycohydrolase TARG1/ 2866–2874 (2017).
C6orf130 in neurodegenerative disease. EMBO J. 32, 1225–1237 (2013). 71. J. Voorneveld, J. G. M. Rack, I. Ahel, H. S. Overkleeft, G. A. van der Marel, D. V. Filippov,
50. J. Moss, M. K. Jacobson, S. J. Stanley, Reversibility of arginine-specific mono(ADP-ribosy- Synthetic α- and β-Ser-ADP-ribosylated peptides reveal α-Ser-ADPr as the native epimer.
l)ation: Identification in erythrocytes of an ADP-ribose-L-arginine cleavage enzyme. Proc. Org. Lett. 20, 4140–4143 (2018).
Natl. Acad. Sci. U.S.A. 82, 5603–5607 (1985). 72. J. Voorneveld, M. S. Kloet, S. Wijngaarden, R. Q. Kim, A. Moutsiopoulou, M. Verdegaal,
51. S. Parthasarathy, A. R. Fehr, PARP14: A key ADP-ribosylating protein in host–virus inter- M. Misra, I. Đikić, G. A. van der Marel, H. S. Overkleeft, D. V. Filippov, G. J. van der Heden van
actions? PLOS Pathog. 18, e1010535 (2022). Noort, Arginine ADP-ribosylation: Chemical synthesis of post-translationally modified
52. M. Mentz, W. Keay, C. D. Strobl, M. Antoniolli, L. Adolph, M. Heide, A. Lechner, S. Haebe, ubiquitin proteins. J. Am. Chem. Soc. 144, 20582–20589 (2022).
E. Osterode, R. Kridel, C. Ziegenhain, L. E. Wange, J. A. Hildebrand, T. Shree, E. Silkenstedt, 73. K. Tashiro, S. Wijngaarden, J. Mohapatra, J. G. M. Rack, I. Ahel, D. V. Filippov, G. Liszczak,
A. M. Staiger, G. Ott, H. Horn, M. Szczepanowski, J. Richter, R. Levy, A. Rosenwald, W. Enard, Chemoenzymatic and synthetic approaches to investigate aspartate- and glutamate-ADP-
U. Zimber-Strobl, M. von Bergwelt-Baildon, W. Hiddemann, W. Klapper, M. Schmidt- ribosylation. J. Am. Chem. Soc. 145, 14000–14009 (2023).
Supprian, M. Rudelius, D. Bararia, V. Passerini, O. Weigert, PARP14 is a novel target in STAT6 74. J. G. M. Rack, I. Ahel, A simple method to study ADP-ribosylation reversal: From function to
mutant follicular lymphoma. Leukemia 36, 2281–2292 (2022). drug discovery. Methods Mol. Biol. 2609, 111–132 (2023).
53. E. J. Longarini, H. Dauben, C. Locatelli, A. R. Wondisford, R. Smith, C. Muench, 75. H. Sellou, T. Lebeaupin, C. Chapuis, R. Smith, A. Hegele, H. R. Singh, M. Kozlowski,
A. Kolvenbach, M. L. Lynskey, A. Pope, J. J. Bonfiglio, E. P. Jurado, R. Fajka-Boja, T. Colby, S. Bultmann, A. G. Ladurner, G. Timinszky, S. Huet, The poly(ADP-ribose)-Dependent
M. Schuller, I. Ahel, G. Timinszky, R. J. O’Sullivan, S. Huet, I. Matic, Modular antibodies reveal chromatin remodeler Alc1 induces local chromatin relaxation upon DNA damage. Mol. Biol.
DNA damage-induced mono-ADP-ribosylation as a second wave of PARP1 signaling. Mol. Cell 27, 3791–3799 (2016).
Cell 83, 1743–1760.e11 (2023). 76. C. Tromans-Coia, A. Sanchi, G. K. Moeller, G. Timinszky, M. Lopes, I. Ahel, TARG1 protects
54. T. Agnew, D. Munnur, K. Crawford, L. Palazzo, A. Mikǒ, I. Ahel, MacroD1 is a promiscuous against toxic DNA ADP-ribosylation. Nucleic Acids Res. 49, 10477–10492 (2021).
ADP-ribosyl hydrolase localized to mitochondria. Front. Microbiol. 9, 20 (2018). 77. N. P. Bullen, D. Sychantha, S. S. Thang, P. H. Culviner, M. Rudzite, S. Ahmad, V. S. Shah,
55. I. Carter-O’Connell, A. Vermehren-Schmaedick, H. Jin, R. K. Morgan, L. L. David, M. S. Cohen, A. Filloux, G. Prehna, J. C. Whitney, An ADP-ribosyltransferase toxin kills bacterial cells by
Combining chemical genetics with proximity-dependent labeling reveals cellular targets modifying structured non-coding RNAs. Mol. Cell 82, 3484–3498.e11 (2022).
of poly(ADP-ribose) polymerase 14 (PARP14). ACS Chem. Biol. 13, 2841–2848 (2018). 78. A. M. Waterhouse, J. B. Procter, D. M. Martin, M. Clamp, G. J. Barton, Jalview version 2—A
56. S. Jungmichel, F. Rosenthal, M. Altmeyer, J. Lukas, M. O. Hottiger, M. L. Nielsen, Proteome- multiple sequence alignment editor and analysis workbench. Bioinformatics 25,
wide identification of poly(ADP-Ribosyl)ation targets in different genotoxic stress re- 1189–1191 (2009).
sponses. Mol. Cell 52, 272–285 (2013). 79. K. Katoh, J. Rozewicki, K. D. Yamada, MAFFT online service: Multiple sequence alignment,
57. Y. Ariumi, Host cellular RNA helicases regulate SARS-CoV-2 infection. J. Virol. 96, interactive sequence choice and visualization. Brief. Bioinform. 20, 1160–1166 (2019).
e0000222 (2022). 80. N. Saitou, M. Nei, The neighbor-joining method: A new method for reconstructing phy-
58. A. Maréchal, J. M. Li, X. Y. Ji, C. S. Wu, S. A. Yazinski, H. D. Nguyen, S. Liu, A. E. Jiménez, J. Jin, logenetic trees. Mol. Biol. Evol. 4, 406–425 (1987).
L. Zou, PRP19 transforms into a sensor of RPA-ssDNA after DNA damage and drives ATR 81. F. Madeira, Y. M. Park, J. Lee, N. Buso, T. Gur, N. Madhusoodanan, P. Basutkar, A. R. N. Tivey,
activation via a ubiquitin-mediated circuitry. Mol. Cell 53, 235–246 (2014). S. C. Potter, R. D. Finn, R. Lopez, The EMBL-EBI search and sequence analysis tools APIs in
59. G. A. Farina, M. R. York, M. Di Marzio, C. A. Collins, S. Meller, B. Homey, I. R. Rifkin, A. Marshak- 2019. Nucleic Acids Res. 47, W636–W641 (2019).
Rothstein, T. R. Radstake, R. Lafyatis, Poly(I:C) drives type I IFN- and TGFβ-mediated infl- 82. W. Liu, Y. Xie, J. Ma, X. Luo, P. Nie, Z. Zuo, U. Lahrmann, Q. Zhao, Y. Zheng, Y. Zhao, Y. Xue,
ammation and dermal fibrosis simulating altered gene expression in systemic sclerosis. J. Ren, IBS: An illustrator for the presentation and visualization of biological sequences.
J. Invest. Dermatol. 130, 2583–2593 (2010). Bioinformatics 31, 3359–3361 (2015).
60. F. J. Bock, T. T. Todorova, P. Chang, RNA regulation by poly(ADP-ribose) polymerases. Mol. 83. P. Fontana, S. C. Buch-Larsen, O. Suyari, R. Smith, M. J. Suskiewicz, K. Schützenhofer, A. Ariza,
Cell 58, 959–969 (2015). J. G. M. Rack, M. L. Nielsen, I. Ahel, Serine ADP-ribosylation in Drosophila provides insights
61. A. H. Forst, T. Karlberg, N. Herzog, A. G. Thorsell, A. Gross, K. L. Feijs, P. Verheugd, P. Kursula, into the evolution of reversible ADP-ribosylation signalling. Nat. Commun. 14, 3200 (2023).
B. Nijmeijer, E. Kremmer, H. Kleine, A. G. Ladurner, H. Schüler, B. Lüscher, Recognition of 84. H. A. Anagho, J. D. Elsborg, I. A. Hendriks, S. C. Buch-Larsen, M. L. Nielsen, Characterizing
mono-ADP-ribosylated ARTD10 substrates by ARTD8 macrodomains. Structure 21, ADP-ribosylation sites using Af1521 enrichment coupled to ETD-based mass spectrometry.
462–475 (2013). Methods Mol. Biol. 2609, 251–270 (2023).
62. S. E. Delgado-Rodriguez, A. P. Ryan, M. D. Daugherty, Recurrent loss of macrodomain ac- 85. M. Carraro, I. A. Hendriks, C. M. Hammond, V. Solis-Mezarino, M. Völker-Albert, J. D. Elsborg,
tivity in host immunity and viral proteins. Pathogens 12, 674 (2023). M. B. Weisser, C. Spanos, G. Montoya, J. Rappsilber, A. Imhof, M. L. Nielsen, A. Groth, DAXX
63. A. Torretta, C. Chatzicharalampous, C. Ebenwaldner, H. Schüler, PARP14 is a writer, reader adds a de novo H3.3K9me3 deposition pathway to the histone chaperone network. Mol.
and eraser of mono-ADP-ribosylation. J. Biol. Chem., 105096 (2023). Cell 83, 1075–1092.e9 (2023).
64. R. L. McPherson, R. Abraham, E. Sreekumar, S.-E. Ong, S.-J. Cheng, V. K. Baxter, 86. L. Schubert, I. A. Hendriks, E. P. T. Hertz, W. Wu, S. Sellés-Baiget, S. Hoffmann,
H. A. V. Kistemaker, D. V. Filippov, D. E. Griffin, A. K. L. Leung, ADP-ribosylhydrolase activity K. S. Viswalingam, I. Gallina, S. Pentakota, B. Benedict, J. Johansen, K. Apelt,
of Chikungunya virus macrodomain is critical for virus replication and virulence. Proc. Natl. M. S. Luijsterburg, S. Rasmussen, M. Lisby, Y. Liu, M. L. Nielsen, N. Mailand, J. P. Duxin, SCAI
Acad. Sci. U.S.A. 114, 1666–1671 (2017). promotes error-free repair of DNA interstrand crosslinks via the Fanconi anemia pathway.
65. G. Gao, X. Guo, S. P. Goff, Inhibition of retroviral RNA production by ZAP, a CCCH-type zinc EMBO Rep. 23, e53639 (2022).
finger protein. Science 297, 1703–1706 (2002). 87. J. Cox, M. Mann, MaxQuant enables high peptide identification rates, individualized p.p.b.-
66. Y. R. Xu, M. L. Shi, Y. Zhang, N. Kong, C. Wang, Y. F. Xiao, S. S. Du, Q. Y. Zhu, C. Q. Lei, range mass accuracies and proteome-wide protein quantification. Nat. Biotechnol. 26,
Tankyrases inhibit innate antiviral response by PARylating VISA/MAVS and priming it for 1367–1372 (2008).
RNF146-mediated ubiquitination and degradation. Proc. Natl. Acad. Sci. U.S.A. 119, 88. S. Tyanova, T. Temu, P. Sinitcyn, A. Carlson, M. Y. Hein, T. Geiger, M. Mann, J. Cox, The
e2122805119 (2022). Perseus computational platform for comprehensive analysis of (prote)omics data. Nat.
67. S. H. Cho, S. Goenka, T. Henttinen, P. Gudapati, A. Reinikainen, C. M. Eischen, R. Lahesmaa, Methods 13, 731–740 (2016).
M. Boothby, PARP-14, a member of the B aggressive lymphoma family, transduces survival 89. D. Szklarczyk, R. Kirsch, M. Koutrouli, K. Nastou, F. Mehryary, R. Hachilif, A. L. Gable, T. Fang,
signals in primary B cells. Blood 113, 2416–2425 (2009). N. T. Doncheva, S. Pyysalo, P. Bork, L. J. Jensen, C. von Mering, The STRING database in
68. S. B. Bachmann, S. C. Frommel, R. Camicia, H. C. Winkler, R. Santoro, P. O. Hassa, DTX3L and 2023: Protein-protein association networks and functional enrichment analyses for any
ARTD9 inhibit IRF1 expression and mediate in cooperation with ARTD8 survival and pro- sequenced genome of interest. Nucleic Acids Res. 51, D638–D646 (2023).
liferation of metastatic prostate cancer cells. Mol. Cancer 13, 125 (2014). 90. P. Shannon, A. Markiel, O. Ozier, N. S. Baliga, J. T. Wang, D. Ramage, N. Amin, B. Schwikowski,
69. V. Iansante, P. M. Choy, S. W. Fung, Y. Liu, J.-G. Chai, J. Dyson, A. Del Rio, C. D’Santos, T. Ideker, Cytoscape: A software environment for integrated models of biomolecular in-
R. Williams, S. Chokshi, R. A. Anders, C. Bubici, S. Papa, PARP14 promotes the Warburg teraction networks. Genome Res. 13, 2498–2504 (2003).
effect in hepatocellular carcinoma by inhibiting JNK1-dependent PKM2 phosphorylation 91. M. Legeay, N. T. Doncheva, J. H. Morris, L. J. Jensen, Visualize omics data on networks with
and activation. Nat. Commun. 6, 7882 (2015). omics visualizer, a cytoscape app. F1000Research 9, 157 (2020).

Đukić et al., Sci. Adv. 9, eadi2687 (2023) 13 September 2023 17 of 18

S C I E N C E A D VA N C E S | R E S E A R C H A R T I C L E

92. Y. Perez-Riverol, J. Bai, C. Bandla, D. García-Seisdedos, S. Hewapathirana, S. Kamatchinathan, funding of the MRC Human Immunology Unit (to J.R.), Novo Nordisk Foundation Center for
D. J. Kundu, A. Prakash, A. Frericks-Zipper, M. Eisenacher, M. Walzer, S. Wang, A. Brazma, Protein Research, the Novo Nordisk Foundation (NNF14CC0001 to M.L.N.), The Danish Council
J. A. Vizcaíno, The PRIDE database resources in 2022: A hub for mass spectrometry-based of Independent Research (0135-00096B, 2034-00311B, and 2032-00311B to M.L.N.), and The
proteomics evidences. Nucleic Acids Res. 50, D543–D552 (2022). Danish Cancer Society (R325-A18824 to M.L.N.). Author contributions: Conceptualization: I.A.
Methodology: D.V.F., S.W., and H.S. Investigation: N.Đ., Ø.S., J.D.E., D.M., K.Z., M.S., C.C., P.K., L.D.,
Acknowledgments: We would like to thank I. Matic for his gift of HRP-conjugated anti–mono- O.S., J.G.M.R., D.B., D.R.K.C., J.G., G.F., S.W., E.P., and R.S. Visualization: N.Đ., Ø.S., J.D.E., C.C., K.Z.,
ADPr antibody. We would like to thank A. Mikoc for critical reading of our manuscript. We would O.S., M.S., and R.S. Supervision: S.S., D.V.F., J.R., D.A., M.L.N., and I.A. Writing (original draft): R.S.,
also like to thank L. Deimel for critical discussion. We also thank the A. Wainman and the Dunn N.Đ., Ø.S., and I.A. Writing (review and editing): N.Đ., Ø.S., D.M., K.Z., M.S., P.K., C.C., J.G.M.R., D.B.,
School Bioimaging Facility for expert advice and access to the confocal microscope. We H.S., D.V.F., S.S., R.S., D.A., and I.A. Competing interests: E.P. is an employee of Vertex
acknowledge funding from the MRC Centre for Medical Mycology at the University of Exeter Pharmaceuticals and may own stock or stock options in that company. All other authors declare
(MR/N006364/2 and MR/V033417/1) and the NIHR Exeter Biomedical Research Centre. that they have no competing interests. Data and materials availability: The MS proteomics
Additional work may have been undertaken by the University of Exeter Biological Services Unit. data have been deposited to the ProteomeXchange Consortium via the PRIDE (92) partner
The views expressed are those of the authors and not necessarily those of the NIHR or the repository with the dataset identifier PXD043452. All data needed to evaluate the conclusions
Department of Health and Social Care. Funding: This work was supported by the in the paper are present in the paper and/or the Supplementary Materials.
Biotechnology and Biological Sciences Research Council (BB/R007195/1 and BB/W016613/1 to
I.A.), Wellcome Trust (210634 to I.A.), Oxford University Challenge Seed Fund (USCF 456 to I.A.), Submitted 14 April 2023
Edward Penley Abraham Research Fund (to I.A.), Ovarian Cancer Research Alliance (813369 to Accepted 10 August 2023
I.A.), Research Council of Norway (315849 to Ø.S.), Wellcome Trust (223107 to I.A. and S.S.), Published 13 September 2023
Swedish Research Council (2019-04871 to H.S.), MRC CDA (MR/X007472/1 to J.G.M.R.), MRC core 10.1126/sciadv.adi2687

Đukić et al., Sci. Adv. 9, eadi2687 (2023) 13 September 2023 18 of 18