key: cord-0893719-k6ub78vt
authors: Roy, Anu; Alhammad, Yousef M.; McDonald, Peter; Johnson, David K.; Zhuo, Junlin; Wazir, Sarah; Ferraris, Dana; Lehtiö, Lari; Leung, Anthony K.L.; Fehr, Anthony R.
title: Discovery of compounds that inhibit SARS-CoV-2 Mac1-ADP-ribose binding by high-throughput screening
date: 2022-03-02
journal: bioRxiv
DOI: 10.1101/2022.03.01.482536
sha: 6348129f8e9c98da6a5287e4392584b48f5c9e39
doc_id: 893719
cord_uid: k6ub78vt

The emergence of several zoonotic viruses in the last twenty years, especially the pandemic outbreak of SARS-CoV-2, has exposed a dearth of antiviral drug therapies for viruses with pandemic potential. Developing a diverse drug portfolio will be critical for our ability to rapidly respond to novel coronaviruses (CoVs) and other viruses with pandemic potential. Here we focus on the SARS-CoV-2 conserved macrodomain (Mac1), a small domain of non-structural protein 3 (nsp3). Mac1 is an ADP-ribosylhydrolase that cleaves mono-ADP-ribose (MAR) from target proteins, protects the virus from the anti-viral effects of host ADP-ribosyltransferases, and is critical for the replication and pathogenesis of CoVs. In this study, a luminescent-based high-throughput assay was used to screen ∼38,000 small molecules for those that could inhibit Mac1-ADP-ribose binding. We identified 5 compounds amongst 3 chemotypes that inhibit SARS-CoV-2 Mac1-ADP-ribose binding in multiple assays with IC50 values less than 100µM, inhibit ADP-ribosylhydrolase activity, and have evidence of direct Mac1 binding. These chemotypes are strong candidates for further derivatization into highly effective Mac1 inhibitors.

COVID-19, caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), 41 is one of the most disruptive and deadly pandemics in modern times, with greater than 385 42 million cases and having led to greater than 5.7 million deaths worldwide. SARS-CoV-2 is the 43 third CoV to emerge into the human population in the last 3 decades, following outbreaks of 44 SARS-CoV in 2002 -2003 and Middle East respiratory syndrome MERS-CoV in 2012. These 45 outbreaks highlight the potential for CoVs to cross-species barriers and cause severe disease in a 46 new host. There is a tremendous need to develop broad-spectrum antiviral therapies capable of 47 targeting a wide range of CoVs to prevent severe disease following zoonotic outbreaks. 48

Coronaviruses encode for 16 highly conserved, non-structural proteins that are processed 49 from two polyproteins, 1a and 1ab (pp1a and pp1ab) (1). The largest non-structural protein is 50 non-structural protein 3 (nsp3) that encodes for multiple modular protein domains. Both the 51 SARS-CoV and the SARS-CoV-2 nsp3 proteins include three tandem macrodomains, Mac1, 52

Mac2, and Mac3 (2). Mac1 is present in all CoVs, unlike Mac2 and Mac3, and contains a 53 conserved three-layered α/β/α fold, a common feature amongst all macrodomains. All CoV Mac1 54 proteins tested have mono-ADP-ribosylhydrolase (ARH) activity, though it remains unclear if 55 they have significant poly-ARH activity (3) (4) (5) (6) (7) (8) . In contrast, Mac2 and Mac3 fail to bind ADP-56 ribose and instead bind to nucleic acids (9,10). Mac1 homologs are also found in alphaviruses, 57

Hepatitis E virus, and Rubella virus, indicating that ADP-ribosylation may be a potent anti-viral 58 post-translational modification (PTM) (11, 12) . All are members of the larger MacroD-type 59 macrodomain family, which includes human macrodomains Mdo1 and Mdo2 (13) . 60 ADP-ribosylation is a post-translational modification catalyzed by ADP-61 ribosyltransferases (ARTs, also known as PARPs) through transferring an ADP-ribose moiety 62 with minimal success (27) (28) (29) (30) . The only compounds identified thus far that inhibit SARS-CoV-2 86 Mac1 with IC50 less than 100 μM are Suramin, which inhibited Mac1-ADP-ribose binding in a 87

FRET assay with an IC50 of 8.7 μM, and Dasatanib, which inhibited Mac1 mono-ARH activity 88

with an IC50 of ~50 μM. Suramin targeted several divergent macrodomains and is known to have 89 additional targets, and thus is not suitable for further evaluation (30). Dasatinib is not a candidate 90 for a Mac1 inhibitor as it is toxic to mammalian cells, though it may provide a scaffold for 91 further inhibitor development. None of the identified compounds have been tested for their 92 ability to inhibit Mac1 in cell culture or in animal models of disease. 93

Here, we optimized two high-throughput macrodomain-ADP-ribose binding assays, a 94 previously described luminescent-based AlphaScreen TM assay, and a novel fluorescence 95 polarization assay (31, 32) , and used the AlphaScreen TM assay to screen ~38,000 compounds for 96 their ability to inhibit SARS-CoV-2 Mac1-ADP-ribose binding. We identified 5 compounds 97 from 3 chemotypes that inhibited ADP-ribose binding by the SARS-CoV-2 Mac1 protein in both 98 assays, some with IC50 values as low as 5-10 μM. These compounds also demonstrated some 99 inhibition of ARH activity and have evidence of direct binding to Mac1. The profiling of the 100 most potent inhibitor against a panel of virus and human MAR binding and hydrolyzing proteins 101 revealed the remarkable selectivity of the inhibition of SARS-CoV-2 Mac1. These compounds 102 represent several series that can be further developed into potent Mac1 inhibitors and potential 103 therapeutics for SARS-CoV-2 and other CoVs of interest. 104 macrodomain proteins ( Fig. 1A-C) . First, we adopted a previously published AlphaScreen TM 109 (AS) assay, where a short peptide was modified at a leucine residue with ADP-ribose through an 110 amino-oxyacetic acid linkage, and at a second leucine residue with biotin ( Fig. 1A) (32) . 111

Streptavidin donor beads and Ni 2+ acceptor beads induce a light signal if the His-tagged Mac1 112 protein interacts with the biotinylated peptide (Fig. 1B) . We also developed a fluorescent 113 polarization (FP) assay as an orthogonal assay to evaluate interactions of macrodomains with 114 ADP-ribosylated peptide. This assay used the same peptide but with fluorescein attached instead 115 of biotin and measures polarization of the fluorescent signal (Fig. 1C) . We then tested 4 separate 116 macrodomains for their ability to bind to these peptides, the human macrodomain Mdo2, and 117

Mac1 from SARS-CoV, MERS-CoV, and SARS-CoV-2. All 4 macrodomains bound to the 118 ADP-ribosylated control peptides better than to non-ADP-ribosylated peptides (Fig. 1D ,G). The 119

AS assay had an especially strong signal-to-background ratio, ranging from ~0.75-2×10 3 . To 120 further study the binding of Mac1 proteins to AS and FP peptides, we evaluated binding in a 121 dose-dependent assay. Of these four proteins, the human MDO2 demonstrated the highest 122 affinity in both assays, with a KD of 1.1 ± 0.3 µM in the FP assay and reached a maximum signal 123 in the AS assay at 40 nM ( Fig. 1E,H) . The SARS-CoV-2 Mac1 had a KD of 3.4 ± 0.4 µM in the 124 FP assay and reached a maximum signal in the AS assay at 0.625 µM, while the SARS-CoV and 125 MERS-CoV Mac1 both reached their maximum signal in the AS assay at ~1.25-2.5 µM (AS) 126 and had KD's of 7.7 ± 1.3 µM and 19.9 ± 3.3 µM in the FP assay, respectively (Fig. 1F,I) . 127

Next, we tested the ability of free ADP-ribose to inhibit the binding of Mac1 to the ADP-128 ribosylated peptide. For these displacement assays, the amount of beads, peptide, and Mac1 129 protein amounts to be used were optimized to obtain a robust signal while limiting the amount of 130 reagents used for screening purposes (see Methods). The addition of free ADP-ribose, but not 131 ATP, into the AS and FP assays inhibited human macrodomain and CoV Mac1 binding to the 132 ADP-ribosylated peptides, confirming that these assays can be used to identify macrodomain 133 binding inhibitors (Fig. 2) . IC50 values for free ADP-ribose ranged between 0.24 µM with SARS-134

CoV Mac1 to 1.5 µM with SARS-CoV-2 using the free ADP-ribose in the AS assay ( Fig. 2A) . 135 Similar results, albeit higher IC50 values were observed in the FP assay, likely because of higher 136 amount of Mac1 used in this assay (4 µM vs 250 nM), with IC50 values ranging from 2.3 µM to 137 9.74 µM (Fig. 2B) . 138 139 High-throughput screening (HTS) for SARS-CoV-2 Mac1 inhibitors. We next performed a 140 small pilot screen of ~ 2,000 compounds from the Maybridge Mini Library of drug-like scaffolds 141 at 10 µM using both AS and FP assays ( Fig. 3A-B) . We identified 39 compounds that 142 significantly inhibited Mac1-ADP-ribose binding at >3 standard deviations (3SD) plus the plate 143 median ( Fig. 3A-B) . After performing dose-response curves we found that two compounds 144 inhibited binding in both assays (Fig. 4A ). We then tested these compounds in a counter screen, 145 which is also an AS assay that utilizes a biotinylated-His peptide that gives off a strong signal 146 with the addition of streptavidin donor and nickel acceptor beads. These two compounds did not 147 affect the signal from our counter screen indicating that they do not intrinsically inhibit the assay. 148

After this initial validation of our screen, three additional libraries were chosen to include a total 149 number of 35,863 compounds from the Analyticon, 3D BioDiversity, and Peptidomimetics 150 libraries (Fig. 3A) . We chose the AS assay as our primary HTS assay, as the average Z' score for 151 the AS was higher than the Z' score from the FP assay in our original screen (0.82 vs 0.67). In 152 this larger screen, the average Z' was 0.89±0.05, indicating a strong separation between positive 153 and negative controls (Fig. 3C ). Using the same hit criteria described above for each individual 154 library, we identified 406 hits resulting in a 1% hit rate (Fig. 3D ). Of note, the Analyticon library 155 produced a lot of non-specific inhibitors, indicating a lot of these compounds likely inhibit the 156 assays themselves (Fig. 3B) . We next performed dose-response (10-40 µM) curves of these 406 157 compounds in our primary (AS), orthogonal (FP), and counter screen (Bn-His6) assays (Fig. 3) . 158

From the 406 original hits, 26 compounds were identified that inhibited SARS-CoV-2 Mac1-159 ADP-ribose binding in the AS assay in a dose-dependent fashion, and 6 compounds were 160 identified that inhibited Mac1 binding in both AS and FP assays (Fig. 3D ). Of these 32 hit 161 compounds, we re-purchased 17 of them, excluding 15 based on several selection criteria, 162

including substantial inhibition of the counter screen, high IC50 values in the AlphaScreen, pan-163 assay interference compounds, and compound availability (Fig. 3D ). The remaining 17 164 compounds along with 4 analogs were repurchased or resynthesized (see Methods). 165

Re-purchased compounds were evaluated in dose-response assays against both SARS-166

CoV-2 Mac1 and human MDO2 protein. Our cutoff criteria included: i) compound must inhibit 167 both primary and orthogonal assays with at least 75% inhibition in AS assay and at or near 50% 168 inhibition in the FP assay, and ii) less than 30% inhibition of the Bn-His6 counter screen. Among 169 the 17 selected and the 4 analogs compounds, six compounds inhibited ADP-ribose binding of 170 SARS-CoV-2 Mac1 in both AS and FP assays with no substantial inhibition of the Bn-His6 171 counter screen. These were compounds 1,2,6,7,10, and 11 (Table 1) . IC50 values ranged from 6.2 172 µM to 112.2 µM in AS assay and 7.3 µM to 159.4 µM in FP assay (Table 1, Fig. 4 ). Compounds 173 1, 10, and 11 also had some inhibitory activity against the MERS-CoV Mac1 protein, though the 174 inhibition of MERS-CoV Mac1 was lower than the inhibition demonstrated against SARS-CoV-175 2 (Table 1 ). In addition, only compound 2 inhibited MDO2, indicating that these compounds 176 were broadly specific for viral macrodomains. 177 Selected compounds demonstrate evidence of SARS-CoV-2 Mac 1 binding. Next, we set out 179 to test the hypothesis that these compounds inhibit Mac1-ADP-ribose by binding to Mac1, and 180 not other components of the assay, such as the peptide. To test for Mac1 binding, we used a 181 differential scanning fluorimetry (DSF) assay as previously described (8) and tested our top 6 hit 182 compounds (Fig. 5 , S1) and compounds 8 and 9, as they are analogs of 6 and 7 (Fig. S2 ). In this 183 assay, compound binding to Mac1 should increase the melting temperature of Mac1. The 184 addition of free ADP-ribose, which binds to Mac1, showed a dose-dependent increase of 185 approximately 4℃ in the melting temperature of Mac1, while the negative control, ATP, had no 186 effect, as previously demonstrated (8). 1, 6, 7, 10, and 11 showed dose-dependent shifts in the 187 melting temperature of Mac1 ranging from 0.2 -1.5℃, providing strong evidence that these 188 compounds bind to Mac1, albeit not with the same affinity as ADP-ribose. On the other hand, 189 compound 2 resulted in highly irregular thermal shift curves, indicating that this compound may 190 not be a true Mac1 binder (Fig. 5, S1 ). These results provide evidence that 5 of our 6 hit 191 compounds (1, 6, 7, 10, and 11) directly bind to Hit compounds inhibit ADP-ribosylhydrolase activity in vitro. SARS-CoV-2 Mac1 is a 194 mono-ADP-ribosylhydrolase that removes mono-ADP-ribose from target proteins (8). Next, we 195 examined the ability of some of our top 5 hit compounds to inhibit the enzymatic activity of 196 SARS-CoV-2 Mac1 using two distinct assays. The first approach was a gel-based Mac1 ADP-197 ribosylhydrolase assay where we tested each compound against the SARS-CoV-2 Mac1 protein 198 (8). Compound 1 tended to precipitate in these assays at higher concentrations, and so we used 199 lower concentrations for this compound than others. Compounds 1, 6, and 7 exhibited a dose-200 dependent inhibition of Mac1 ADP-ribosylhydrolase activity (Fig. 6A ). We were unable to detect 201 any significant inhibition with 10 and 11 in this assay. 202

Next, we utilized a recently published high-throughput luminescence-based ADP-203 ribosylhydrolyase assay (33). Here we found that 1, 6, 7, 10 and 11 all showed dose-dependent 204 inhibition of ADP-ribosylhydrolase activity (Fig. 6B) . 6 was clearly the most efficient inhibitor, 205 as it had a peak of ~60% inhibition, similar to Dasatinib which we previously identified in a 206 separate HTS (33). In contrast to the gel-based assay, 10 and 11 did inhibit ADP-207 ribosylhydrolase activity in this assay, likely reflecting the increased sensitivity of this assay 208 (Table 1 ). These results indicate that this compound is highly selective for the 220 SARS-CoV-2 Mac1 protein. structures. In addition to our 5 hit compounds, we also docked compounds 8 and 9 as they are 227 analogs of 6 and 7 and could give further insight into SAR, even though we either detected 228 minimal or no direct Mac1 binding by these compounds. These seven compounds were docked 229 against the ADP-ribose bound structure of SARS-CoV-2 Mac1 (PDB 6WOJ) as well as three apo 230 structures of Mac1 were used (PDB 7KR0, 7KR1, 6WEY). Docking and glide emodel scores 231

were calculated for each compound against all four structures and the best structure was chosen 232 based on these scores (Table S1 ). Analog compounds 6, 7, 8, and 9 were assessed both based on 233 score and visual inspection, and were re-docked using a core constraint to a high scoring, 234 intuitive pose of compound 7. All top scoring poses were subsequently minimized using Prime, 235 allowing flexibility within 5 Å of the ligand. Compound 1 was its own chemotype but has a 236 sulfonohydrazide that is also found in a compound identified in a previous screen for Mac1 237 compounds (34). It also has a thienopyrimidine that is similar to the pyrrolopyrimidine found in 238 of the compounds identified in the fragment screen by Schuller et al (31) . It makes a hydrogen 239 bond with a backbone amine of D22, pie-stacking interactions with F156, and extends with a 240 benzene ring into the distal ribose pocket inserting in between the GGG and GIF loops (Fig. 8A) . 241

Compounds 10 and 11 are close analogs with a single difference of positioning in the 242 bromobenzoyl moiety on the piperidine ring (Fig. 4C) . These compounds had similar activity 243 across the board in our assays, making it difficult to analyze their SAR. While they docked into 244 the binding pocket, these docking poses only indicate a single hydrogen bond with the backbone 245 amino of D22. In contrast, compounds 6, 7, 8, and 9 are close analogs of each other and have a 246 wide-range of inhibitory and binding activity. IC50 values for these compounds range from 10 to 247 several hundred µM (Table 1) . Direct binding also varied substantially, with Tm's ranging from 248 ~1.7 ℃ (6) to undetectable binding (8). These compounds all have the same base structure, 249 including a beta-alanine core substituted with a N-benzyl or N-chlorobenzyl group, a methoxy 250 benzoyl group and a piperazine amide. The main difference between 6 and its analogs are the 251 addition of a methoxy group on the benzoyl group (7), the loss of a chlorine (8), and a missing 252 methoxy group (9). Each of these changes reduces the activity of this series indicating that i) the 253 orientation of the methoxy groups on 6 is likely important for its increased activity, ii) 254 reorienting 7 to accommodate the 4-methoxy group likely decreases activity due to the disruption 255 of multiple interactions, and iii) the chlorine likely makes a critical halogen bond with a 256 backbone amino group of L126 in the binding pocket. 257

In conclusion, we developed multiple high-throughput ADP-ribose binding assays and 258 performed HTS to identify high-quality Mac1 inhibitors. We followed these screens with several 259 additional assays to measure their ability to inhibit ADP-ribosylhydrolase activity and their direct 260 binding to Mac1. We have identified 5 compounds that inhibit both the primary and orthogonal 261 assays without inhibiting the counter screen and demonstrate dose-dependent inhibition of Mac1 262 enzymatic activity. Compounds 1 and 6 are particularly effective with IC50 values of ~10 µM in 263 the AS assay, along with thermal shifts and docking poses that indicate direct binding to Mac1. 264

Compound 6 shows excellent selectivity towards SARS-CoV-2 over the human macrodomains 265 guiding further development of the compound. We expect that these compounds could be 266 utilized for further derivatization and optimization into more potent Mac1 inhibitors. 267

Reagents 269

All plasmids and proteins used were expressed and purified as previously described 270

(30,35-37). All compounds were repurchased from MolPort except for compounds 6 and 10, 271 which were repurchased from ChemDiv. After reordering once, compounds 10 and 11 became 272 unavailable and thus were resynthesized according to the literature (38). ADP-ribosylated 273 peptides were purchased from Cambridge peptides. 

The AlphaScreen reactions were carried out in 384-well plates (Alphaplate, PerkinElmer, 287

Waltham, MA) in a total volume of 40 μL in buffer containing 25 mM HEPES (pH 7.4), 100 288 mM NaCl, 0.5 mM TCEP, 0.1% BSA, and 0.05% CHAPS. All reagents were prepared as 4X 289 stocks and 10 μL volume of each reagent was added to a final volume of 40 μL. All compounds 290 were transferred acoustically using ECHO 555 (Beckman Inc) and preincubated after mixing 291 with purified His-tagged macrodomain protein (250 nM) for 30 min at RT, followed by addition 292 of a 10 amino acid biotinylated and ADP-ribosylated peptide [ARTK(Bio)QTARK(Aoa-293 RADP)S] (Cambridge peptides) (625 nM). After 1h incubation at RT, streptavidin-coated donor 294 beads (7.5 μg/mL) and nickel chelate acceptor beads (7.5 μg/mL); (PerkinElmer AlphaScreen 295

Histidine Detection Kit) were added under low light conditions, and plates were shaken at 400 296 rpm for 60 min at RT protected from light. Plates were kept covered and protected from light at 297 all steps and read on BioTek plate reader using an AlphaScreen 680 excitation/570 emission 298 filter set. For counter screening of the compounds, 25 nM biotinylated and hexahistidine-tagged 299 linker peptide (Bn-His6) (PerkinElmer) was added to the compounds, followed by addition of 300 beads as described above. 301

The FP assay was performed in buffer containing 25 mM Tris pH7.5, NaCl 50 mM, 303 0.025% TritonX-100. All reagents were prepared as 2X stocks and 10 μL volume of each reagent After mixing for a minute, the plate was incubated at 25°C, protected from light and fluorescence 308 polarization was read after 30 minutes, 1h and 2h using a plate reader. anti-mouse antibodies (LI-COR Biosciences). All immunoblots were visualized using 327

Odyssey ® CLx Imaging System (LI-COR Biosciences). The images were quantitated using the 328 LI-COR Image Studio software. 329

The recently published assay, ADPr-Glo, was used to examine the impact of our top hit 331 compounds on SARS-CoV-2 enzymatic activity (33). Briefly, the compounds were preincubated 332 with SARS-CoV-2 Mac1 (2 nM) and NudF (125 nM) at ambient temperature for 30 min prior to F G H I Figure 1 . Coronavirus Mac1 binding to ADP-ribosylated peptides. A) Illustration of the amino-oxyacetic acid modified lysine-conjugated ADP-ribosylated peptide with an additional biotin conjugated to a different lysine residue and included are the amino acid sequences and modification sites of peptides used in this study. B-C) Cartoon diagrams depicting a bead-based AS (A) and FP (B) assays for measuring macrodomain interactions with an ADP-ribosylated peptide. D) Macrodomain proteins were incubated with peptide #1 or peptide #3 for 1 hour at RT and Alphacounts were determined as described in Methods. E-F) Peptide #1 was incubated with indicated macrodomains at increasing concentrations and Alphacounts were measured as previously described. G) Mac1 proteins were incubated at indicated concentrations with peptide #2 or peptide #4 and the plate was incubated at 25°C for 1 hr before polarization was determined. H-I) Peptide #2 was incubated with indicated macrodomain proteins at increasing concentrations and polarization was determined as previously described. All data represent the means ± SD of 2 independent experiments for each protein. . High-throughput screen for SARS-CoV-2 Mac1 inhibitors. A) List of libraries that were screened, the number of compounds from each library, and the type of compounds each library contains. B) Scatterplot showing the % inhibition of each compound in the screen. The cutoff for a hit was the plate median + 3 standard deviations. C) Z' scores were determined for each plate in the screen. The average Z' score was 0.89 ± 0.05. D) Dose response confirmation. From the original screen, we identified 406 potential hits, these hits were retested in a dose-response assay on both the AS and FP assays and were also counterscreened against a biotinylated 6His peptide. After these assays and other exclusion criteria, 17 hit compounds and 4 analogs were repurchased or resynthesized. (1) HTS01833 (2) F594-1001 (6) F594-1011 (7) Z269-0281 ( Figure 5 . Thermal stability of SARS-CoV-2 Mac1 after incubation with hit compounds. The top 6 hit compounds were tested for their ability to increase the thermal stability of SARS-CoV-2 Mac1 in a differential scanning fluorimetry assay (DSF). The data represent the means ± SD of the T m from two independent experiments. Inhibition % Figure 6 . Impact of hit compounds on SARS-CoV-2 ADP-ribosylhydrolase activity. A) Compounds were incubated at indicated concentrations for 30 minutes with the SARS-CoV-2 Mac1 protein prior to adding the PARP10 substrate and then were further incubated for 30 minutes. Proteins were analyzed by Immunoblotting with anti-GST (PARP10) and anti-MAR binding reagent (MABE1076). Gels were quantitated using Image Studio software. The bar graph above each immunoblots represent the mean inhibition ± SD from at least two independent experiments. (C) Compounds 10, 11. Yellow lines -hydrogen bonds; Cyan lines -pi-pi interactions; magenta lines -weak hydrogen bonds; and purple lineshalogen bond. Figure S1 . Thermal stability of SARS-CoV-2 Mac1 after incubation with hit compounds. The top 4 hit compounds were tested for their ability to increase the thermal stability of SARS-CoV-2 Mac1 in a differential scanning fluorimetry assay (DSF). Thermal profiles are shown for each compound at different concentrations. \ Figure S2 . Thermal stability of SARS-CoV-2 Mac1 after incubation with analog compounds. Two analogs of FS2MD-006 are shown here for their ability to increase the thermal stability of SARS-CoV-2 Mac1 in a differential scanning fluorimetry assay (DSF). Thermal profiles and change in T m are plotted for each compound at different concentrations.

Fragment binding to the 475 Nsp3 macrodomain of SARS-CoV-2 identified through crystallographic screening and 476 computational docking

Discovery of drug-like ligands for the Mac1 domain of 479 SARS-CoV-2 Nsp3

The SARS-CoV-2 Nsp3 483 macrodomain reverses PARP9/DTX3L-dependent ADP-ribosylation induced by 484 interferon signalling

A molecular toolbox for ADP-ribosyl binding proteins

Fragment 497 binding to the Nsp3 macrodomain of SARS-CoV-2 identified through crystallographic 498 screening and computational docking

Allosteric Inhibitor Targeting Macrodomain 2 of Polyadenosine-Diphosphate-Ribose 502 Polymerase 14

High-Throughput Activity Assay for Screening Inhibitors of the SARS-CoV-2

Identification of Poly(ADP-Ribose) Polymerase Macrodomain Inhibitors Using an 510 AlphaScreen Protocol

The SARS-CoV-2 Conserved Macrodomain Is a Mono-ADP-514 Ribosylhydrolase

Preparation of screening assays for ADP-ribosyl readers and erasers using the 517 GAP-tag as a binding probe