The Chikungunya virus nsP3 macro domain inhibits activation of the NF-κB pathway

The role of the chikungunya virus (CHIKV) non-structural protein 3 (nsP3) in the virus lifecycle is poorly understood. The protein comprises 3 domains. The N-terminus is a macro domain, biochemically characterised to bind both RNA and ADP-ribose, and to possess ADP-ribosyl hydrolase activity – an enzymatic activity that removes ADP-ribose from mono-ADP-ribosylated proteins. As ADP-ribosylation is important in the signalling pathway leading to activation of the transcription factor NF-κB, we sought to determine if the macro domain might perturb NF-κB signalling. We first show that CHIKV infection did not induce NF-κB activation, and could not block exogenous activation of the pathway via TNFα, although TNFα treatment did reduce virus titres. Ectopic expression of nsP3 was able to block TNFα-mediated NF-κB activation and this was dependent on the macro domain, as mutations previously shown to disrupt either ADP-ribose binding or hydrolase activity lost the ability to inhibit NF-κB activation. Lastly, we determined the phenotype of the macro domain mutants in the context of virus infection in a range of cell types. Our data are consistent with cell- and species-dependent roles of the macro domain, however, these phenotypes do not correlate with the ability to inhibit NF-κB activation suggesting that the macro domain plays multiple independent roles in the virus lifecycle.


Introduction
Technologies). Luciferase samples were collected and quantified using the Dual Luciferase 110 reporter assay system (Promega). 111 Immunofluorescence. Cells were grown on coverslips prior to experimentation. Cells were 112 washed with PBS, fixed with 4% paraformaldehyde for 10 min at room temperature, washed 113 with PBS and permeabilised in 0.5% Triton-X-100 for 10 min, washed with PBS, blocked in 114 2% BSA for 1 h, washed with PBS and incubated with primary antibodies (rabbit anti-nsP3 115 kindly provided by Andres Merits, University of Tartu, or mouse anti-p65 Santa Cruz SC-8008) 116 for 1 h. After washing in PBS, secondary antibodies (Life Technologies) were applied for 1 h. 117 Coverslips were finally washed with PBS, rinsed in dH2O then mounted in Prolong-gold with 118 DAPI. Cells were imaged using a Zeiss LSM 700 confocal microscope, images were 119 processed using Zen black software. 120 Western blotting. Protein concentration of lysed cell samples was calculated using a Bradford 121 assay and 30 µg of protein in Laemmli buffer loaded per well on a 12%, 10%, or 7.5% SDS-122 PAGE gels for B14, nsP3 and phospho-p105 respectively. Protein were transferred from gels 123 onto membrane (Immobilon FL, Merck) via semi-dry blotter for 1 h at 15 V. Membranes were 124 blocked using blocking buffer (LICOR) for 20 min at room temperature prior to incubation with 125 primary antibody prepared in TBS (rabbit anti-nsP3, anti-phospho-p105 NEB 4806, anti-Flag 126 Sigma F3165, or anti-actin Sigma A1978) rocking overnight at 4 C. Membranes were washed 127 in TBS then secondary antibodies (LICOR) added for 1h at room temperature. Membranes 128 were then washed in TBS + 0.1% Tween, then dH 2O prior to drying and imaging using the 129 LICOR Odyssey scanner.

CHIKV infection does not activate the NF-кB pathway 134
The NF-кB pathway is activated as part of the innate response to viral infection, leading to 135 expression of antiviral genes, and is frequently targeted for viral immune evasion 28,29 . Given 136 the importance of this pathway it is surprising that the interplay between CHIKV infection and 137 the NF-кB pathway is poorly understood. One study proposed that CHIKV infection led to 138 upregulation of miR-146a, a regulator of expression of NF-кB pathway components, thereby 139 blocking the NF-кB response 23 . As no other studies have been published to date on CHIKV 140 and the NF-кB pathway, we sought to address this gap in our understanding. 141 To test this, we examined the effect of CHIKV infection on the NF-кB pathway using a reporter 142 plasmid containing firefly luciferase (Fluc) under the control of an NF-кB-sensitive promoter 143 with a second plasmid as a transfection control, renilla luciferase (Rluc) under the control of 144 the constitutive thymidine kinase (TK) promoter. We chose to use Huh7 (human hepatoma) 145 cells for this analysis as they are efficiently infected with CHIKV 30 and respond well to 146 exogenous (e.g. TNFα) activation of the NF-кB pathway (Fig 1a). Following transfection cells 147 were incubated for 16 h, then either infected with CHIKV (MOI=5) or treated with TNF-α. NF-148 кB activation was determined by normalising the Fluc values to Rluc. As shown in Fig. 1a, 149 CHIKV infected cells showed no increase in NF-кB activation up to 24 h post infection, 150 indistinguishable from the control (neither infected nor TNF-α treated). In contrast, cells treated 151 with TNFα demonstrated a constant increase in NF-кB activation over 24 h. 152 NF-кB activation involves nuclear translocation of the p50-p65 heterodimer. Huh7 cells either 153 treated with TNFα or infected with CHIKV for 24 h were analysed by immunofluorescence 154 using antibodies to p65 to assess NF-кB activation, and nsP3 to detect CHIKV infection. Fig.  155 1b shows that nsP3-positive (CHIKV infected) cells exhibited a cytoplasmic localisation of p65, 156 consistent with the lack of NF-кB activation following CHIKV infection. As expected, TNFα treatment resulted in translocation of a proportion of p65 into the nucleus of cells, with some 158 protein retained in the cytoplasm. 159 We then sought to determine whether the pathway was active at an earlier stage. We therefore 160 assessed whether an active IKK complex was able to form in CHIKV infected cells, indicated 161 by the phosphorylation of the NF-кB precursor/inhibitor p105. Western blotting with an 162 antibody to phosphorylated p105 revealed that this was induced by TNFα treatment from 15 163 min to 1 h.p.t. (Fig. 1c). In contrast, CHIKV infection did not induce the phosphorylation of 164 p105 at any time. These data collectively demonstrate that CHIKV infection does not result in 165 activation of the NF-кB pathway. We proceeded to ask whether exogenous activation of the 166 pathway would have an effect on CHIKV virus production. Treatment of infected Huh7 cells 167 with TNFα resulted in a significant reduction (10-fold) in virus production, as determined by 168 plaque assay titration of virus released into the culture supernatant (Fig. 2a). The data shown 169 in Fig. 2b confirm that TNFα treatment of CHIKV infected cells resulted in p65 nuclear 170 translocation of NF-кB, consistent with NF-кB activation. These data demonstrate that 171 exogenous activation of the NF-кB pathway can inhibit the CHIKV lifecycle and reduce 172 production of progeny virus. 173

CHIKV infection does not inhibit exogenous activation of the NF-кB pathway 174
Given that NF-кB activation inhibited CHIKV virus production, we sought to investigate 175 whether CHIKV infection had any effect on exogenous activation of the NF-кB pathway. To 176 test this, Huh7 cells were transfected with the NF-кB reporter plasmid and infected with CHIKV 177 at 16 h post-transfection. Immediately after infection, cells were treated with TNFα, and cell 178 lysates were harvested over a 24 h time period. This analysis revealed that CHIKV infection 179 was not able to inhibit the exogenous activation of NF-кB (Fig. 3a). To confirm this result, cells 180 were analysed by immunofluorescence for nsP3 and p65. As shown in Fig. 3b, CHIKV 181 infection did not block the nuclear translocation of p65 following TNFα treatment. Taken 182 together, these data demonstrate that CHIKV infection is unable to suppress exogenous 183 activation of the NF-кB pathway by TNFα.

pathway. 186
We originally hypothesised that by virtue of its ADP-ribosyl-binding and hydrolase activities, 187 the CHIKV nsP3 macro domain might be able to modulate NF-кB activation.  Conversely, p65 was excluded from the nucleus in cells expressing nsP3-F, indicating that 208 nsP3 was inhibiting nuclear translocation of NF-кB, and therefore activation of the pathway. 209

The nsP3 macro domain contributes to inhibition of NF-кB activation 211
To assess whether the macro domain was involved with the nsP3-mediated inhibition of the 212 NF-кB pathway, we generated a panel of six alanine-substitutions of residues previously 213 shown to be critical for the various functions of this domain. The previously determined 214 phenotypes of these mutants are listed in Table 1  control were confirmed by western blot (Fig 6b). Interestingly the phenotype of the panel varied between the different cell types. In Huh7 cells 230 only two of the mutants (T111A and Y114A) were able to produce any detectable infectious 231 virus (Fig. 7a) and the titres of these were significantly reduced compared to WT. In BHK-21 232 cells (Fig. 7b), D10A and G112A failed to produce any infectious virus, V113A was attenuated 233 and the other 3 mutants exhibited a WT titre, and similar levels of nsP3 expression detected 234 by western blot (Supp. Fig. 1c). A similar picture was observed in C2C12 cells (Fig 7c), with 235 the exception that D10A was able to produce significant amounts of infectious virus. In U4.4 236 mosquito cells (Fig 7d) D10A, G112A and V113A failed to produce infectious virus but the functional RNAi response due to a frameshift mutation in the Dcr2 gene 32 , two of these mutants 239 (D10A and V113A) were able to produce infectious virus, suggestive of a role of the macro-240 domain in counteracting the insect antiviral system. Lastly, when we analysed the phenotype 241 of the mutant panel in the context of a dual-luciferase sub-genomic replicon (SGR) 30 in Huh7 242 cells (Fig 7f) we observed that all the mutants were able to replicate to some extent, although 243 only two (V113A and Y114A) retained WT levels of replication. We conclude from these data 244 In this study we sought to investigate the effect of CHIKV infection on NF-кB activity. We 258 observed that CHIKV infection could neither induce the pathway, nor inhibit activation via the 259 canonical pathway as a response to the external stimulus provided by TNFα treatment. 260 However TNFα treatment did result in a reduction in CHIKV virus production. We reasoned 261 that the lack of NF-кB activation was consistent with the hypothesis that CHIKV was able to 262 block this activation. A likely mediator of this effect was the macro domain of nsP3, and indeed ectopic expression of nsP3 and a panel of mutants in the active site of the macro domain 264 revealed that this was indeed the case. As shown in Table 1 those mutants previously shown  265 to abrogate ADP-ribosyl hydrolase activity (in particular D10A) also abolished the ability to 266 inhibit NF-кB activation, However, this was not an absolute correlation and in fact all mutants 267 had some effect, suggesting that the ability of the macro domain to block NF-кB activation is 268 not solely due to hydrolase activity.    Indeed both NEMO and ARTD10 were shown to be in vitro substrates for the ADP-279 ribosylhydrolase activity of the nsP3 macro domain 17 , and were efficiently de-mono-ADP-280 ribosylated (deMARylated). Although we have been unable to biochemically demonstrate an 281 interaction between nsP3 and either NEMO or ARTD10, we did observe partial colocalisation 282 of nsP3 with ARTD10 at late stages of CHIKV infection (Supp. Fig. S2). This is consistent propose therefore that the nsP3 macro domain directly interacts with one or more ADP-285 ribosylated proteins involved in the NF-кB pathway, likely candidates being NEMO and/or 286 ARTD10. However, given the previous observation that miR-146a is also involved in the 287 regulation of the NF-кB pathway by CHIKV, we believe it is highly likely that CHIKV, like many 288 other viruses, possesses multiple mechanisms to control this key signalling pathway. 289 Analysis of the nsP3 macro domain mutant panel in the context of infectious CHIKV revealed 290 that there was little correlation between the virus growth phenotype and the effect on the NF-291 кB pathway (Fig 7). Thus we conclude that, at least in the range of cells tested, inhibition of 292 NF-кB activation is not required for efficient production of infectious virus. However, it was 293 interesting to note that most mutants exhibited cell type specific phenotypes, for example 294 production of virus in Huh7 cells seemed particularly dependent on macro domain function as 295 4 mutants failed to grow in these cells, of these 3 (D10A, G32A and V113A) were able to grow 296 in other cell types. The growth of some mutants was not due to reversion to wildtype as, at 297 least for G32A, T111A and Y113A we did not observe any reversion to wildtype in C2C12 or 298 U4.4 cells (data not shown). However, due to low titres we were unable to RT-PCR and 299 sequence the other mutants. 300 In conclusion, we show here that one of the roles of the CHIKV nsP3 macro domain is to 301 contribute to the regulation of a critical antiviral factor, the transcription factor NF-кB. This   Huh7 cells were infected with CHIKV (MOI=5), mock infected or TNFα treated (50 ng/mL) fixed 331 at 24 h p.i. and stained for p65 (green) and nsP3 (red). Nuclear p65 indicates activation of the indicated timepoints. Lysates were western blotted for phospho-p105 with actin as control. 334