Zinc-finger PARP proteins ADP-ribosylate alphaviral proteins and are required for interferon-γ–mediated antiviral immunity

Figures

• 

  Fig. 1.. A subset of human PARP proteins ADP-ribosylates alphavirus nsPs. (A) Domain architecture schematics and amino acid lengths of human PARP proteins used in this work. Only the domains shown in the legend are marked on each PARP protein. Domain annotations were obtained from www.uniprot.org. MW, molecular weight. (B) Schematic of the alphaviral nsPs that make up the nonstructural polyprotein. Highlighted is the position of the viral macrodomain in nsP3. (C to F) Western blots probed for ADP-ribosylated proteins with the indicated PARPs and nsPs cotransfected in HEK293T cells. Expected positions of each individual PARP protein and nsP are shown, with those showing no specific ADP-ribosylation indicated in gray and those that are ADP-ribosylated in at least one lane indicated in black. Automodification of PARP7 and PARP12 indicated by blue stars and specific labeling of nsP3 and nsP4 are indicated in the appropriate lanes. Blots with all other PARP proteins are found in fig. S2. The decrease in automodification and FLAG-tagged protein expression in all lanes expressing nsP2 is likely caused by the known role of nsP2 in inhibiting host gene expression (34). All blots are representative…

• 

  Fig. 2.. Evolutionary and functional prediction for altered catalytic activity of the SFV4 macrodomain. (A) Consensus motif from an alignment of 20 SFV nsP3 sequences surrounding position 48 (SFV4 numbering), indicating the high conservation of the macrodomain sequence. SFV4 contains the sole difference in this region of the alignment, replacing a well-conserved alanine at position 48 (polyprotein position 1384) with a glutamic acid. A list of all sequences used is found in data S1 (tab 1). The position of the conserved N21 residue used in this manuscript for inactivating macrodomain activity is indicated with an asterisk. (B) Alignment of the region surrounding residue 48 from diverse members of the “SFV complex” of alphavirus phylogeny. All members, except SFV4, have an A or S in position 48. Also included is the macrodomain from Eilat virus, an insect-specific virus that lacks the ability to replicate in vertebrate cells (39) and that lacks ADP-ribosylhydrolase activity (36). Among all alphaviruses, only the insect-specific viruses such as Eilat virus contain an E at position 48. A list of all sequences used is found in data S1 (tab 1). (C) Structure of the macrodomain from Getah virus…

• 

  Fig. 3.. The macrodomain activity of SFV nsP3 affects PARP-mediated labeling. (A) Assay for the ADP-ribosylhydrolase activity of the indicated macrodomain-containing protein. Coexpression of PARP10 with nsP3 from the original SFV4 sequence (“SFV4”), a version in which E48A substitution is introduced (“SFV4 E48A”), or a macrodomain of archaeal origin with high enzymatic activity (“Af1521”). The E48A substitution that restores SFV4 nsP3 to the consensus sequence increases ADP-ribosylhydrolase activity when tested against PARP10. Numbers at the bottom of the anti-HA blot indicate quantification of bands for macrodomain-containing proteins (nsP3 in lanes 2 and 3 and Af1521 in lane 4). (B) Coexpression of PARP15 with the indicated macrodomain-containing protein. Unlike the original SFV4 nsP3 protein, coexpression of PARP15 with SFV4 nsP3 E48A shows no accumulation of ADP-ribosylation, consistent with macrodomain-catalyzed removal of the PARP15-mediated ADP-ribosylation. (C) Coexpression of PARP12 with SFV nsP3 and nsP4. Coexpression of the active SFV4 nsP3 (E48A) results in a decreased signal of ADP-ribosylation on nsP4, indicative of macrodomain-catalyzed removal of PARP12-mediated…

• 

  Fig. 4.. The macrodomain activity of SFV nsP3 affects PARP-mediated antiviral activity. (A) Schematic of the SFV4 β-galactosidase reporter constructs (45) used in these studies. (B) Macrodomain sequences from the SFV consensus (see Fig. 2A), SFV4, and the “macro active” and “macro inactive” versions used in these studies. (C) Assay for ADP-ribosylhydrolase activity of the indicated macrodomain-containing SFV nsP3s on PARP10 automodification in HEK293T cells. Blots are representative of three independent biological replicate experiments. (D) β-Galactosidase output of SFV replication reporters with a catalytically active macrodomain (“macro active”) compared to catalytically inactive macrodomain (“macro inactive”) in HEK293T cells. “−” indicates a condition in which an empty cloning vector (pQCXIP) was transfected in place of the SFV4 reporter. (E) SFV4 replication reporters with active or inactive macrodomains tested against a variety of PARP proteins in HEK293T cells. PARP13S (ZAP-S) and PARP13L (ZAP-L) lack a catalytically active PARP domain and thus have anti-alphavirus activity that is independent of ADP-ribosyltransferase activity. In contrast, the antiviral activities of PARP7,…

• 

  Fig. 5.. SinV nsP4 is ADP-ribosylated by PARP7 and PARP12. (A) SinV nsP3 contains a macrodomain with ADP-ribosylation activity. The indicated nsP3 (WT or N21A mutant) from SinV, or Af1521, was coexpressed with PARP10. WT nsP3 and Af1521 can reverse PARP10 automodification, but the N21A mutant nsP3 cannot. (B and C) Overexpression of SinV nsP3 (WT or N21A mutant) or nsP4 with PARP7 (B) or PARP12 (C). No labeling of either version of nsP3 is observed, but nsP4 is modified during coexpression with PARP7 or PARP12. All experiments were performed in HEK293T cells, and blots are representative of at least three independent biological replicate experiments.

• 

  Fig. 6.. Zinc-finger PARPs, PARP7, PARP12, and PARP13, are important for the antiviral activity of IFN-γ. (A to C) Human Huh7 cells were pretreated with the indicated siRNA and concentration of IFN-γ and then infected with SinV (25,000 PFU per well; multiplicity of infection of ~0.1). Negative control conditions (“−”) were treated with a nontargeting negative control siRNA. siRNA silencing of either PARP7 (A), PARP12 (B), or PARP13 (C) attenuates the antiviral effects of IFN-γ against SinV replication. (D and E) Human Huh7 cells were treated with the PARP7-selective inhibitor Phthal01, the PARP1/2 inhibitor olaparib (59), or dimethyl sulfoxide control and the indicated concentration of IFN-γ or IFN-α and then infected with SinV. (A) to (E) N = 3 biological replicates. Data were analyzed using the two-way ANOVA with Šidák’s post test. ****P < 0.0001; n.s., not significant.

• 

  Fig. 7.. Ancient domain architecture involved in antiviral immunity. (A) Conservation of a zinc-finger and tandem WWE domain architecture across human PARP7, PARP12, and PARP13. Sliding window analysis of amino acid conservation across the three indicated human PARPs. A schematic of PARP12 is shown below to indicate the positions of the annotated zinc-finger and WWE domains (B). Amino acid alignments of the indicated human PARPs in the indicated protein regions. Residues in bold are conserved across all three proteins and define the CCCH-type zinc-finger and WWE domains. (C) Updated domain architecture of PARP7, PARP12, and PARP13 based on amino acid conservation in the indicated domains. Annotation of the PAR-binding domain, comprising a zinc-finger domain followed by two WWE domains, is based on structural and functional analyses performed in PARP13 (60). Colors of domains are as in Fig. 1A. (D) Presence of the PAR-binding-PARP domain architecture within eukaryotic genomes. Species clades in which one or more proteins with a PAR-binding-PARP domain architecture are present as shown with bold black text. Clades in which no proteins exist with that architecture are shown in black with red…