The conserved macrodomains of the non-structural proteins of Chikungunya virus and other pathogenic positive strand RNA viruses function as mono-ADP-ribosylhydrolases

Human pathogenic positive single strand RNA ((+)ssRNA) viruses, including Chikungunya virus, pose severe health problems as for many neither efficient vaccines nor therapeutic strategies exist. To interfere with propagation, viral enzymatic activities are considered potential targets. Here we addressed the function of the viral macrodomains, conserved folds of non-structural proteins of many (+)ssRNA viruses. Macrodomains are closely associated with ADP-ribose function and metabolism. ADP-ribosylation is a post-translational modification controlling various cellular processes, including DNA repair, transcription and stress response. We found that the viral macrodomains possess broad hydrolase activity towards mono-ADP-ribosylated substrates of the mono-ADP-ribosyltransferases ARTD7, ARTD8 and ARTD10 (aka PARP15, PARP14 and PARP10, respectively), reverting this post-translational modification both in vitro and in cells. In contrast, the viral macrodomains possess only weak activity towards poly-ADP-ribose chains synthesized by ARTD1 (aka PARP1). Unlike poly-ADP-ribosylglycohydrolase, which hydrolyzes poly-ADP-ribose chains to individual ADP-ribose units but cannot cleave the amino acid side chain - ADP-ribose bond, the different viral macrodomains release poly-ADP-ribose chains with distinct efficiency. Mutational and structural analyses identified key amino acids for hydrolase activity of the Chikungunya viral macrodomain. Moreover, ARTD8 and ARTD10 are induced by innate immune mechanisms, suggesting that the control of mono-ADP-ribosylation is part of a host-pathogen conflict.


Analysis of released products from hydrolase assays by poly-acrylamide gel electrophoresis (PAGE)
Hydrolase assays were carried out with immunoprecipitated HA-ARTD1 that was labeled in the presence of 100 µM β-NAD + , 5 pmol annealed double stranded oligomers and 1 µCi 32 P-NAD + at 30°C for 30 min. Following the incubation step the supernatants were removed from the beads. The material bound to the beads and a fraction of the supernatant were used for SDS-PAGE analysis and Coomassie blue (CB) staining to visualize the amount of immunoprecipitated HA-ARTD1 and the vMDs that were included in the reactions. Furthermore the dried gels were subjected to autoradiography ( 32 P) to depict the remaining automodification of HA-ARTD1. The supernatant and thus the released products of the hydrolase assays were further subjected to PAGE to determine polymer length. The PAGE was carried out as previously described 1 . Chemical detachment and precipitation of the modifications were not necessary and therefore the reaction's supernatants were simply dried in a Speed Vac and resuspended in 5 µl of loading buffer (50% urea w/v, 25 mM NaCl, 4 mM EDTA, 0.02% xylene cyanol, 0.02% bromophenol blue).

In silico alanine scanning
The ABS-Scan web-server 2 systematically evaluates amino acids for their importance in protein-ligand interactions by in silico alanine scanning. The crystal structure of the Chikungunya macrodomain in complex with ADP-ribose (PDBID 3GPO) was used as starting structure 3 . A distance cut-off of 5 Å was chosen to define the binding site around the ADP-ribose. For each residue within the cut-off, all side chain atoms beyond C β were removed and the missing hydrogen was added, obtaining an alanine side chain. Modeler library was used on all selected residues, coupled with steps of energy minimization to ensure that no steric clashes occur between protein and ligand atoms 4 . The analysis and results derived from alanine scanning mutagenesis relies on two assumptions: (i) The introduced point mutation does not drastically change the structure of the protein and (ii) the mode of ligand interaction is unchanged 2 . The structural quality of the generated protein structures was estimated through Discrete Optimized Protein Energy (DOPE) score 5 , while the energetics of a protein-ligand complex was scored by using Autodock 4.1 forcefield 6 .
The contribution of a specific amino acid is determined by the difference in interaction score of mutant and wild-type protein (ΔΔG value). From this procedure three relevant residues for ADP-ribose binding were identified, namely N24, V33 and Y114 (hot-spot residues, hereafter).

Computational mutagenesis
The three residues identified in the in silico alanine scanning were systematically exchanged against all other 19 amino acids by using the Swiss-PdbViewer package 7 . The energetics of the protein-ligand complex of all protein variants of the three hot-spot residues (N24, V33 and Y114) were evaluated by the Amber score.
Amber score implements molecular mechanics Generalized Born/surface area simulations with traditional general Amber force field for ligand molecules 8 . The interaction between the ligand and the protein is represented by electrostatic and van der Waals energy terms, and the solvation energy is calculated using Generalized Born solvation model. In this protocol it is implicitly assumed that point mutations in the protein do not significantly affect the conformation of the mutated protein. During Amber score calculation, the input coordinates of the different amino acids for each hot-spot residue are minimized using the conjugate gradient method to remove poor contacts. This is followed by molecular dynamics simulation (Langevin dynamics at constant temperature), and a short minimization to obtain the final energy of the system. The Amber score is calculated as E(Complex) -[E(Protein) + E(Ligand)]. The entropic contribution is supposed to be constant in the mutated and wild type structure considering their similarity and was therefore not calculated, as discussed previously 9 . From these procedure two mutants for each hot-spot residue were identified to strongly destabilize the protein-ligand complex without introducing dramatic structural changes to the protein. These are N24Y, N24R, V33E, V33F, Y114V, and Y114W.

Circular dichroism analysis
The buffer of the bacterially expressed and purified His 6 -tagged fusion proteins of the CHIKV-nsP3 macrodomain WT and mutants was exchanged for CD buffer (10 mM potassium phosphate pH 7.5, 100 mM (NH4) 2 SO 4 , 10% glycerol) using 10 kDa MWCO centrifugal filter devices (Amicon Ultra-0.5, Merck Millipore, Billerica, MA, USA) by centrifugation (14,000 g, 4°C, 15 min). Afterwards protein concentrations were determined using the "Bio-Rad DC™ Proteinassay Kits" (Bio-Rad) and adjusted to 0.5 mg/mL with CD buffer. Subsequently the protein purity was assessed by SDS-PAGE and the gels were stained with Coomassie blue (CB).
The sample volume in each circular dichroism (CD) measurement was 140 µL. Each sample was transferred onto a Hellma ® SUPRASIL cuvette (Hellma GmbH & Co. KG) with a pathlength of 0.5 mm. The sample analysis was carried out in a triplicate scan from 195 to 240 nm at room temperature with the Olis SDM 17 CD (Olis). CD buffer measurements provided the baseline and the scans were averaged. The CD spectra were smoothed employing the Savitzky-Golay filter (Olis Global Works software package) with a filter size of 11 10 . Gorry, P. A. General least-squares smoothing and differentiation by the convolution (Savitzky-Golay) method. Analytical chemistry 62, 570-573 (1990).
Supplementary Figure S1 (a) HeLa cells were transiently transfected with an expression plasmid for GFP-ARTD8-macro1-3. Cells were lysed and lysates separated by SDS-PAGE and immunoblotted for testing the ARTD8 antibody. The antibody, generated against a peptide derived from the sequence located between macrodomain 2 and 3, was able to detect GFP-ARTD8-macro1-3. As control the fusion protein was also detected by a GFP antibody.
(b) PMA differentiated THP-1 cells were stimulated with LPS for the indicated times and the mRNA expression of ARTD10 was determined using RT-qPCR (mean values ± SD of three experiments) and ARTD10 protein was evaluated by immunoblotting using mAb 5H11.
Supplementary Figure S2 (a) The GST-ARTD10cat domain was automodified in the presence of 32 P-NAD + . The proteins were then incubated with SINV-nsP3 macrodomain for the indicated times.
The proteins were stained using Coomassie blue (CB) and the radioactivity associated with the different substrates was assessed by autoradiography ( 32 P).
(b) As in panel a but with the automodified catalytic domain of ARTD8 as substrate.
(c) As in panel a but with the ONNV-nsP3 macrodomain.
(d) As in panel c but with the automodified catalytic domain of ARTD8 as substrate.
(e) As in panel a but with the FIPV-nsP3 macrodomain.
(f) As in panel e but with the automodified catalytic domain of ARTD8 as substrate.
Supplementary Figure S3 (a) Percent identity matrix generated with Clustal multiple sequence alignment based on sequence comparison of the indicated macrodomains.
(b) Phylogenetic tree created by using ClustalW2.
Supplementary Figure S4 (a) HA-ARTD1 was expressed in HEK293 cells and immunoprecipitated from lysates.
Automodification was carried out in presence of β-NAD + and double stranded oligomers. PARylated ARTD1 was then incubated with CHIKV-nsP3 macrodomain for the indicated times. PAR levels were determined using a PAR-specific antibody.
Proteins were visualized by Coomassie blue (CB) staining. For control whole cell lysates (WCL) were analyzed for the expression of HA-ARTD1 and α-Tubulin by immunoblotting.
(b) As in panel a but with the FIPV-nsP3 macrodomain.
Supplementary Figure S5 (a) HA-ARTD1 was expressed in HEK293 cells and immunoprecipitated from lysates.
Automodification was carried out in presence of 32 P-NAD + and double stranded oligomers. PARylated ARTD1 was then subjected to hydrolase assays with the indicated macrodomains. The supernatants were collected and fractions were either analyzed by thin layer chromatography (TLC) to visualize released ADPr and PAR chains or by SDS-PAGE and Coomassie blue (CB) staining to visualize the macrodomains that were used in the hydrolase assays. As control for the TLC analysis 32 P-NAD + was spotted onto PEI-F cellulose plates (indicated by *, very left lane).
(b) Fractions of the supernatants from panel a were analyzed on a sequencing PAGE to visualize released ADPr and PAR chains. As control 32 P-NAD + was used (indicated by *, very left lane).
(c) The remaining automodification of immunoprecipitated HA-ARTD1 from panel a (the beads) was analyzed by SDS-PAGE, stained with CB and exposed to X-ray film.
Supplementary Figure S6 (a) Transiently expressed HA-ARTD1 was immunoprecipitated from HEK293 cell lysates. Automodification was carried out in presence of 32 P-NAD + and double stranded oligomers and ARTD1 was incubated with the CHIKV macrodomain and the indicated mutants. The supernatants were analyzed by thin layer chromatography (TLC) to visualize released ADPr and PAR chains. As control 32 P-NAD + was spotted onto PEI-F cellulose plates (indicated by *, very left lane).
(b) Circular dichroism (CD) analysis of the CHIKV macrodomain WT and mutants.