key: cord-0426180-vl95csp4
authors: Weixler, Lisa; Voorneveld, Jim; Aydin, Gülcan; Bolte, Timo M. H. R.; Momoh, Jeffrey; Bütepage, Mareike; Golzmann, Alexandra; Lüscher, Bernhard; Filippov, Dmitri V.; Žaja, Roko; Feijs, Karla L. H.
title: Systematic analysis of ADP-ribose detection reagents and optimisation of sample preparation to detect ADP-ribosylation in vitro and in cells
date: 2022-02-22
journal: bioRxiv
DOI: 10.1101/2022.02.22.481411
sha: b7acd9d92e5240731a73ba953483dee4ac0c86e3
doc_id: 426180
cord_uid: vl95csp4

Recent evidence suggests that modification of substrates with a single ADP-ribose (ADPr) is important in for example antiviral immunity and cancer. However, the endogenous substrates and the extent of mono-ADP-ribosylation are still largely unclear. Several reagents were developed to detect ADP-ribosylation but it is unknown whether they recognise only ADPr, amino acid-ADPr linkages or a combination of ADPr with a protein backbone. We screened the affinity of selected reagents for enzymatically, chemically and in cell generated ADP-ribosylation on glutamate, cysteine, serine, arginine, threonine and RNA by blotting, as well as analysed the subcellular sites of ADP-ribosylation using immunofluorescence confocal microscopy. We furthermore observed that the modification is heat-labile and optimised sample preparation procedures. Our comparison of the available reagents, as well as optimisation of sample preparation, will allow future work further dissecting the function of ADP-ribosylation in cells, both on protein and on RNA substrates.

The posttranslational modification of proteins is a well-known way to regulate proteins in response to 32 changes in nutrient availability, viral infection, DNA damage and many other signals. Proteins with 33 catalytic activity can be switched on or off, others can change their localisation within the cell or interact 34 with different molecules. ADP-ribosylation is a posttranslational modification which is mediated in cells 35 by ADP-ribosyltransferases (ARTs) of the ARTD family, which add ADP-ribose (ADPr) to their targets 36 while releasing nicotinamide from co-factor NAD + (1, 2). Best studied is the modification with chains 37 of ADPr, termed poly(ADP-ribosyl)ation (PARylation), which amongst other processes has a 38 demonstrated role in the DNA damage response, regulation of protein stability and Wnt signalling. PARP13 (6, 7) . Despite the potentially misleading name PARP for MARylating enzymes, this term was 46 kept for historical reasons (8). MARylation may play a role in immunity and transcription amongst other 47 functions (2, (9) (10) (11) (12) , which are very varied: PARP10 may regulate kinase activity (13-15) and replication 48 (16) , PARP12 localises to the Golgi and stress granules and may function there (17, 18) , PARP14 was 49 reported to act as transcriptional co-activator (19) , but also to be present at focal adhesions (20) and 50 involved in DNA replication and repair (21, 22) . Recent work on PARP7 points to a role in both 51 immunity, regulation of the cytoskeleton and suggests that it may serve as anti-cancer drug target (23-52 25). These highly diverse functions of MARylation and associated transferases have been reviewed in 53 detail (2, 10-12).

Different amino acids were identified as ADPr acceptors in recent years using different approaches 55 including mass spectrometry: glutamates, serines, tyrosines, histidines and most recently cysteines (24, 56 [26] [27] [28] [29] . In addition to the modification of proteins with ADPr, recent in vitro data indicate that some of 57 the mammalian PARPs can also modify nucleic acids as summarised elsewhere (30, 31), although it is 58

not clear yet what role RNA and DNA MARylation has in cells.

MARylation of both proteins and nucleic acids is a reversible modification, with different hydrolases 60 removing the modification from different acceptor sites. ARH1 removes ADPr from arginine (32), COVID19 pandemic, the spotlight has recently been on the SARS-CoV-2 macrodomain protein Mac1, 68 which was shown to be an ADP-ribosylhydrolase with thus far unknown substrates (9, 37, 38) . Not only 69 certain transferases, but also a number of the mammalian hydrolases has been suggested to drive certain 70 aspects of tumorigenesis such as transformation, growth and invasiveness (39, 40) . This has been 71 difficult to verify, as antibodies for the hydrolases were poorly characterised and antibodies specific for 72 MARylation were not available at all.

Research in the area of MARylation has been held back by lack of tools for the analysis and detection 75 of modified substrates. In recent years two major breakthroughs have occurred: optimised mass 76 spectrometry methods, which are now reliably able to detect the modified proteins present in cells (29, 77 41, 42) , and the development of multiple antibodies and other reagents which detect this modification.

This provides a great opportunity which has to be enjoyed with caution: most of the reported detection 79 tools were tested on specific substrates, but their actual epitopes have been poorly mapped. It is thus not 80 clear whether it is possible to directly compare one study performed with one antibody to another study 81 using another detection tool. The first specialised readers of MARylation are the macrodomains of 82 PARP14, which have been used to detect intracellular MARylation using live-cell imaging (43, 44) .

This was turned into a commercially available detection reagent by fusion of specific macrodomains to 84 the Fc region of rabbit immunoglobulin (45). Next an antibody was generated against a peptide with a 85 MARylated lysine which appears to detect MARylation in cells efficiently as well as PARylation (46).

This was followed by the creation of a modified version of the MAR-hydrolase Af1521, which has 87 increased binding affinity compared to the wildtype protein but is still catalytically active (29). Next, 88 antibodies were generated against MARylated serines as well as a general ADP-ribose antibody (47).

Most recently, a polyclonal antibody raised against MARylated peptides was raised in rabbits for 90 detection of MARylation (48). Several reagents are thus available that can be utilised to study the in 91 vivo function of MARylation, however, none of these reagents have been compared with each other. It 92 is not clear whether there are differences in substrate recognition, whether some may preferentially bind 93 to specific MARylated amino acids or even recognise part of the protein backbone. It is highly probable 94 that the MAR-specific reagents do recognise either the specific ADPr-protein bond or part of the protein 95 surrounding, as otherwise they would be expected to be efficient tools to detect PARylation as well.

We have compared the above-mentioned different reagents to map their respective specificities. For this 98 purpose, we enzymatically generated protein substrates MARylated on specific amino acids, such as 99 serine, arginine and glutamate, or peptides chemically modified on serine, cysteine and threonine, as 100 well as MARylated nucleic acid substrates, and tested all the detection reagents for ADP-ribosylation 101 that were available to us. We verified the specific modification of our substrates by using a panel of 102 hydrolases, which reverse the modification only from specific targets as expected. We have included a 103 detection reagent based on murine PARP14 macrodomains, which was previously described to have 104 higher affinity for ADPr than the human macrodomains (43) and appears to be an efficient tool for 105 immunoprecipitation of ADP-ribosylated nucleic acids. After determining the specificities of the 106 antibodies on in vitro generated substrates, we next asked whether differences exist in their recognition 107 of the modification introduced by different PARPs overexpressed in cells and tested their suitability for 108 immunofluorescence using different fixation methods. To be able to do this, we optimised the 109 preparation of cell lysates to ensure maximum retention of the ADPr signal. Collectively, our work 110 deciphers which reagents are suitable for which purpose and has optimised sample preparation 111 procedures which allow detection of low ADP-ribosylation levels. To be able to compare the currently available ADP-ribosylation detection reagents, we required suitable, 118 well-defined substrates. Therefore, we assembled a collection of ADP-ribosyltransferases with known 119 substrate specificity. Using these enzymes, we can generate proteins specifically modified on cysteine 120 by pertussis toxin (PT), acidic amino acids by PARP10, arginine by mART2.2 and SpvB, serine by 121 PARP1/HPF1 or PAR by PARP1. Recent studies reported modification of other amino acids, such as 122 histidine or tyrosine, however, the respective enzymes are not yet known and could therefore not be 123 included. The majority of proteins employed here were purified from bacterial expression systems, with 124 the exception of PARP1, which was immunoprecipitated using a GFP-tag from HEK293T cells ( Figure   125 1a). All purified proteins are active and either automodify or ADP-ribosylate a specific target 126 (Supplementary Figure 1) , with the exception of the arginine-ART mART2.2, which modifies many 127 proteins present in a cytosolic extract. Nuclei and mitochondria were removed as they contain largest 128 amounts of PARP1, TARG1 and MACROD1 and could confound the ADP-ribosylation assay. We 129 noticed automodification also for pertussis toxin, which is in line with previous observations using this 130 truncated version of the protein (49). To confirm the specificity of the ADP-ribosylation reactions, we 131 incubated these substrates with a panel of hydrolases. For PARP10 and PARP1/HPF1, we made use of 132 specific inhibitors to stop the reactions before adding hydrolases (50). ARH1 is only capable of reversing 133 arginine modification, whereas ARH3 hydrolyses both PAR as well as removes ADP-ribose linked to 134 serine. PARG is only capable of hydrolysing the glycosidic bond between ADP-ribose moieties and 135 should not be able to remove the ADP-ribose linked to the proteins. MACROD1, MACROD2 and 136 TARG1 have been described to reverse the modification of acidic residues (13, 34). For the cysteine-137 linked ADP-ribosylation introduced by PT no erasers are known to date. We purified the known erasers 138 of ADP-ribosylation from bacteria: MACROD1, MACROD2, TARG1, PARG, ARH1 and ARH3 139 (Figure 1a ). The majority of described activities we could confirm, for example MACROD1, 140 MACROD2 and TARG1 reverse the modification introduced by PARP10 although not fully, as has 141 been seen before (Figure 1b and Supplementary Figure 2 ) (6, 51). We could furthermore confirm that 142 ARH3 reverses both PAR and serine-linked MARylation (Figure 1c) , but also appears to have some 143 activity towards PARP10 (Figure 1b) ,. This was hinted at in earlier work studying reversal of ADP-144 ribosylation by PARP10 (6) and would be expected if PARP10 also automodifies on serine as reported 145 (51) . ARH1 reverses the arginine modification introduced by mART2.2 very efficiently, but has no 146 activity towards other modified amino acids (Figure 1d) . As the hydrolases reverse the diverse 147 modifications as expected, we concluded that our in vitro substrates are modified on the expected amino 148 acids. None of the hydrolases were able to reverse the modification from cysteine as generated by 149 pertussis toxin (Figure 1e) . This raises the possibility that additional mammalian intracellular 150 hydrolases remain to be discovered. Alternatively, the automodification of the truncated pertussis toxin 151 we used may not represent an accessible substrate for the hydrolases tested and we can thus not exclude 152 that they are capable of reversing cysteine MARylation on different substrates.

In addition, we also noted that none of the hydrolases is able to completely reverse PARP10 modification 154 and speculated that PARP10 may be promiscuous and able to modify more than one type of amino acid.

Partial reversal of PARP10 automodification by PARG and ARH3 was observed before, which may be 156 reversal of oligomers that PARP10 was reported to generate (6). A more recent study using recombinant 157 proteins also hinted at PARP10 promiscuity, as the modification of serine, arginine and glutamate by 158 PARP10 was detected using mass spectrometry (51). We incubated the protein with different hydrolases 159 and combinations thereof. We were not able to further decrease the modification after removal of the 

Having thus confirmed the specific modification of our enzymatically generated substrates, we next 178 performed western blots to test the specificities of the antibodies and detection reagents available to us.

We generated large amount of substrate (Supplementary figure 1) , stored this at -20ºC and proceeded 180 with western blots once radioactivity decayed with the detection reagents described in Table 4 . We first 181 used the macrodomain-based detection reagents on these substrates, Reagents I-III, which are based on 182 either human or murine PARP14 macrodomains or contain the aforementioned Af1421 fused to an Fc 183 (Figure 2a-d) . We could confirm their reported specificity for MARylation over PARylation, as also 184 for example the PARP1-HPF1 sample resulted in a more specific band instead of a smear. We developed 185 the murine PARP14 macrodomain-based detection reagent, Reagent III or mPARP14-m2m3-Fc, as 186 affinity for ADPr was reported to be higher for the mouse than human macrodomain proteins (43, 44).

We also generated a control, Reagent IIIm or mPARP14-m2m3-GE-Fc, which has impaired ADPr Autoradiogram an antibody which was described to detect PARylation, Reagent VII. It recognises PARylation as stated, 192 but also for MARylation bands are detectable (Figure 2h) , which raises the question whether a fraction 193 of the modification generated by these enzymes are oligomers, or whether the antibody can detect single 194 ADPr moieties. It for example detects the modification introduced by PARP10, which could reflect a 195 minor oligo-ADPr-transferase activity which PARP10 was reported to possess (6) After having determined their specificity on in vitro substrates, we next determined which signals the 230 antibodies detect in HEK293T lysates. We overexpressed almost all full-length GFP-tagged PARPs, as 231 well as GFP alone, and first confirmed expression of these proteins (Figure 3a) . PARP2 was not 232 included due to its expected redundancy with PARP1 and for PARP14 we were not able to generate a 233 full-length construct containing an intact N-terminus. Overexpression of these enzymes led to 234 degradation of some enzymes, apparent from the smaller products visible in the anti-GFP blot. We I II III  I II III  I II III   I II III  I II III  I II 

Resulting western blots were analysed with an ADPr antibody Reagent V. (f) ADP-ribosylated peptides and automodified 287 PARP10 catalytic domain were slot-blotted untreated, or heated at 60ºC and 95ºC for 2, 5 or 10 minutes. The blot was analysed show a robust modification signals for most enzymes (Figure 4a-b) . In general, the antibody Reagents 296 IV-VI detect the arginine modification introduced by mART2.2 well (Figure 4a-c) . In contrast to these, 297 a TNKS2 signal is present when using anti-PAR Reagent VII as expected (Figure 4d) . 

Only some reagents are suitable to detect RNA ADP-ribosylation

As recent publications have shown that also nucleic acids can be MARylated at least in vitro, we next 318 determined whether these different reagents can be used to detect the modified nucleic acids. We 319 generated MARylated RNA oligonucleotides and confirmed their modification using denaturing urea-320 PAGE and SYBR gold staining to visualise the nucleic acids in gel (Figure 5a) . We purified the 321 different RNA species and slot-blotted them alongside automodified PARP10 as control (Figure 5b) .

Several reagents can be used to detected MARylated RNA: both the antibodies Reagent IV and V, as 323 well as the macrodomain-based Reagent II and III detects the modification. We performed a similar 324 experiment with NAD + -capped RNA, which we produced as described before (54) procedure, we generated RNA with a radiolabelled NAD + -cap and subjected this to DXO treatment, 326 which degrades specifically NAD + -capped RNAs (54) (Figure 5c) . None of the antibodies tested can 327 be used to detect the NAD + -cap (Figure 5d ), which agrees with dot-blots where NAD + could not be 328 detected by the anti-ADPr antibody (48). Adenylylation is another modification which closely resembles 329 ADP-ribosylation and may be recognised by these reagents. We generated 5'-adenylated ssRNA and 330 slot-blotted these alongside ADP-ribosylated RNA and noted that Reagents I, V and IV can detect the 331 modification, whereas the macrodomain-based reagents II and III are not able to bind this. Figure 6) . This further confirms that the reagents recognise significantly different 333 epitopes. Lastly, we enriched in vitro ADP-ribosylated ssRNA using GFP-mPARP14-m2m3-Fc 334 wildtype or GE mutant. mPARP14-m2m3 efficiently binds the modified ssRNA, but not the unmodified 335 RNA (Figure 5f ). The binding-mutant does not interact with either modified or non-modified RNA. In 336 this pull-down, an additional band is enriched with m2m3 wildtype, which could imply that PARP10cat 337 can generate an oligomer, which the module preferentially binds, or that it modifies multiple sites on 338 each RNA, also leading to enhanced precipitation. In addition to their usage to study MARylated 339 proteins, a subset of the available antibodies and detection reagents can thus be used to start studying reported before with Reagent V ( Figure. 6a) methanol fixed samples and leads to low staining intensity with most reagents, but gives rise to nucleolar 367 staining with the CST antibody (Figure 6a) . Upon H2O2 treatment, the ADP-ribosylation signal 368 becomes strongest in the nucleus (Figure 6b) . The discrepancy between cytoplasmic stainings in 369 untreated cells fixed with PFA can have several possible explanations: either the respective epitopes are 370 not exposed after fixing with methanol, or something is stained which is washed out by methanol but 371 crosslinked by PFA, such as small metabolites. We expected similar staining patterns between PFA and 372 glyoxal, as both are cross-linking agents. One key difference, however, is that glyoxal does not crosslink 373 RNA as well, in contrast to PFA. In theory, it is possible that the strong cytoplasmic signal observed 374 with a number of the reagents is mitochondrial RNA. Both antibodies which efficiently recognise slot-375 blotted MARylated ssRNA, also stain cytoplasmic structures after PFA fixation but not after glyoxal.

We incubated the slides with either RNAse or DNAse to remove signals derived from potentially 377 MARylated RNA or DNA, neutral hydroxylamine to reverse modification of acidic residues or mercury 378 chloride to reverse modification of cysteines. All treatments lead to some reduction of the signal, 379 however, the strongest change is visible in the hydroxylamine treated samples (Figure 6c) . In this work, we set out to compare the affinities of the different reagents available to detect 396 MARylation. They have different strengths and weaknesses: many also recognise PAR, making them 397 less suitable to study exclusively MARylation. One other aspect which may interfere with the blots 398 performed with these reagents, is the existence of AMPylation as protein modification (56) Many plasmids used to express proteins were gifts from other labs, as indicated below (Table 1) . For 465 expression of GFP-ARTs in mammalian cells, we amplified the different genes using appropriate 466 primers for the full-length gene products from HeLa cDNA where possible or used gBlocks from IDT.

These were either transferred into the Gateway system. All ARTs harbour the GFP-tag on the N-468 terminus in these constructs. Full sequences are available with the plasmids on Addgene. The generation 469 of pDONR-mPARP14-m2m3 constructs was described before (43). These we transferred using the 

The majority of ARTs used in this study were purified from bacterial expression systems with N-477 terminal His or GST-tags, with the exception of PARP1 which was produced as GFP-fusion protein 478 from HEK293T cells as described below. As we found that many of the proteins are toxic to the bacteria, 479 we optimised expression conditions and bacterial strains for each protein separately, with the optimal 480 conditions summarised below (Table 2) . benzonase and olaparib. Proteins were separated on 10-15% gels and blotted onto nitrocellulose 551 membranes using a BioRad TurboBlot apparatus using the high-molecular weight 10-minute program.

A step-by-step protocol for our western blotting procedure is available online (67). Membranes were 553 blocked with 5% non-fat milk in TBST for 30-60 minutes at RT, primary antibodies were diluted in 554 TBST as indicated below and incubated overnight at 4°C, secondary antibodies were diluted 1:5000 in 555 5% non-fat milk in TBST and incubated for 30-60 minutes at RT. Some of the antibodies could be 556 stored and reused multiple times as indicated in Table 2 . Multiple wash steps were performed in between 557 and after antibody incubations with TBST at RT for at least 5 minutes. Chemiluminescent signals were 558 detected by either exposure to film or using the Azure600. Additional antibodies used: anti-GFP 1:2500 559 (600-101-215 Rockland), anti-HSP60 1:2500 (12165 Cell Signaling Technology), anti-PARP1 1:1000

(1 835 238 Roche), anti-tubulin 1:5000 (B-5-1-2 Santa Cruz). For slot blotting, 2µM of each peptide 561 was slot blotted using a PR648 Slot Blot Blotting Manifold (Hoefer) onto nitrocellulose membrane.

Subsequently, the membrane was processed identical to the described processing of western blots and 563 detected using exposure to film. were cross-linked using UV light (120 milijoules/cm2). The membrane was blocked with 5% non-fat 574 milk in PBST for 60 minutes at RT, primary antibodies were diluted in PBST as indicated and incubated 575 overnight at 4°C, secondary antibodies were diluted 1:5000 in 2% non-fat milk in PBST and incubated 576 for 30 minutes at RT. Multiple wash steps with PBST were performed after both antibody incubations 577 for at least 5 minutes. Chemiluminescent signals were detected using the Azure600. Peptides were 578 blotted onto nitrocellulose activated in water, followed by blocking in 5% non-fat milk in PBST and 579 antibody incubations. Antibody dilutions and wash steps were identical to the processing of western 

New insights into the molecular and cellular functions of 606 poly(ADP-ribose) and PARPs

Multifaceted Posttranslational Modification Involved in the Control of Cell Physiology in 609 Health and Disease

PARP and PARG inhibitors in cancer treatment

Toward a unified 613 nomenclature for mammalian ADP-ribosyltransferases

ADP-ribose): novel functions for an 616 old molecule

Substrate-assisted catalysis by PARP10 limits its activity to mono-ADP-ribosylation

Family-wide analysis of 621 poly(ADP-ribose) polymerase activity

ADP-623 ribosyltransferases, an update on function and nomenclature

The impact of PARPs and 625 ADP-ribosylation on inflammation and host-pathogen interactions

MARTs and MARylation in the Cytosol: Biological 630 Functions, Mechanisms of Action, and Therapeutic Potential

Uncovering the Invisible: Mono-ADP-ribosylation Moved into 632 the Spotlight

ARTD10 637 substrate identification on protein microarrays: regulation of GSK3beta by mono-ADP-638 ribosylation

PARP10 suppresses tumor 640 metastasis through regulation of Aurora A activity

PARP10 642 promotes cellular proliferation and tumorigenesis by alleviating replication stress

PARP1-645 produced poly-ADP-ribose causes the PARP12 translocation to stress granules and 646 impairment of Golgi complex functions

PARP12, an interferon-648 stimulated gene involved in the control of protein translation and inflammation

Collaborator of Stat6 (CoaSt6)-associated poly(ADP-651 ribose) polymerase activity modulates Stat6-dependent gene transcription

Genome-657 wide CRISPR synthetic lethality screen identifies a role for the ADP-ribosyltransferase 658 PARP14 in DNA replication dynamics controlled by ATR

Identification of PARP-7 substrates reveals a role for MARylation in microtubule control in 665 ovarian cancer cells

e10. 672 26. Suskiewicz MJ, Palazzo L, Hughes R, Ahel I. Progress and outlook in studying the 673 substrate specificities of PARPs and related enzymes

Gas-678 Phase Fragmentation of ADP-Ribosylated Peptides: Arginine-Specific Side-Chain Losses and 679 Their Implication in Database Searches

Engineering Af1521 improves ADP-ribose binding and identification of ADP-ribosylated 682 proteins

ADP-ribosylation of 684 RNA and DNA: from in vitro characterization to in vivo function

ADP-ribosylation of DNA and RNA

Molecular 689 and immunological characterization of ADP-ribosylarginine hydrolases

Proteomic analyses 692 identify ARH3 as a serine mono-ADP-ribosylhydrolase

Macrodomain-containing proteins: 697 regulating new intracellular functions of mono(ADP-ribosyl)ation

The 702 SARS-CoV-2 Conserved Macrodomain Is a Mono-ADP-Ribosylhydrolase

The Viral Macrodomain Counters Host Antiviral ADP-704 Ribosylation

ARH1 in Health and Disease. Cancers 706 (Basel)

The Controversial Roles of ADP-Ribosyl Hydrolases 708 MACROD1, MACROD2 and TARG1 in Carcinogenesis. Cancers (Basel)

An Advanced Strategy for Comprehensive 710 Profiling of ADP-ribosylation Sites Using Mass Spectrometry-based Proteomics

Proteome-713 wide identification of the endogenous ADP-ribosylome of mammalian cells and tissue

Recognition of 716 mono-ADP-ribosylated ARTD10 substrates by ARTD8 macrodomains

Assessment of 719 Intracellular Auto-Modification Levels of ARTD10 Using Mono

Macrodomains 2 and 3 of Murine Artd8

Generation and Characterization of 722

Recombinant Antibody-like ADP-Ribose Binding Proteins

Enabling drug discovery for the 725 PARP protein family through the detection of mono-ADP-ribosylation

An HPF1/PARP1-Based Chemical Biology Strategy for Exploring ADP-Ribosylation

Mitochondrial 731 NAD(+) Controls Nuclear ARTD1-Induced ADP-Ribosylation

Discovery 733 of Compounds Inhibiting the ADP-Ribosyltransferase Activity of Pertussis Toxin

Small-736 Molecule Chemical Probe Rescues Cells from Mono-ADP-Ribosyltransferase

ARTD10/PARP10-Induced Apoptosis and Sensitizes Cancer Cells to DNA Damage

PARP10 Multi-Site Auto-and Histone MARylation 740 Visualized by Acid-Urea Gel Electrophoresis

Philibert K, Zwiers H. Evidence for multisite ADP-ribosylation of neuronal 744 phosphoprotein B-50/GAP-43

Hopp AK, Hottiger MO. Investigation of Mitochondrial ADP-Ribosylation Via 749 Immunofluorescence

Enzymes Involved in AMPylation and deAMPylation

Antibodies Are Sensitive and Valuable Tools for Detecting Patterns of AMPylation. iScience

Identification 756 of macrodomain proteins as novel O-acetyl-ADP-ribose deacetylases

Nucleolar-nucleoplasmic shuttling of TARG1 and its control by DNA damage-induced poly-760 ADP-ribosylation and by nucleolar transcription

Activity-Based 762 Screening Assay for Mono-ADP-Ribosylhydrolases

A new 770 monoclonal antibody detects a developmentally regulated mouse ecto-ADP-771 ribosyltransferase on T cells: subset distribution, inbred strain variation, and modulation 772 upon T cell activation

Functional localization of two 779 poly(ADP-ribose)-degrading enzymes to the mitochondrial matrix

Unprocessed blots are presented either in the manuscript or in the Supplementary Data Files

RNA slot blots

generated the 795 murine Parp14-m2m3-Fc fusion constructs

conceived this study, designed the 796 experiments and analysed the data

wrote the manuscript 797 with input from R.Ž. and L.W. All authors have proof-read and agreed with the submission of the

We thank the researchers in the ADP-ribosylation community who have put their expression constructs 802 and detection reagents at our disposal to allow thorough characterisation of these materials

Hottiger provided the eAf1521 and ADPr antibody

Ivan Matic provided a pan-ADPr antibody

2 bacterial and human expression plasmids were a gift from Fritz Koch-Nolte

Purified recombinant HPF1 was provided by Patricia Korn. We are 807 grateful for helpful discussions with and suggestions from Michael O

This work was supported by the Confocal Microscopy Facility, a Core Facility of the Interdisciplinary 809

Center for Clinical Research (IZKF) Aachen within the Faculty of Medicine at RWTH Aachen 810

Funding was provided by the START program of the Medical Faculty of RWTH Aachen 811 University to K.F. (10/18) and R.Ž. (13/20) and by the German Research Foundation DFG to B.L. 812 (LU466/16-2)