A pivotal role of serine 225 phosphorylation in the function of hepatitis C virus NS5A revealed with the application of a phosphopeptide antiserum and super-resolution microscopy

NS5A is a multi-functional phosphoprotein that plays a key role in both viral replication and assembly. The identity of the kinases that phosphorylate NS5A, and the consequences for HCV biology, remain largely undefined. We previously identified serine 225 (S225) within low complexity sequence (LCS) I as a major phosphorylation site and used a phosphoablatant mutant (S225A) to define a role for S225 phosphorylation in the regulation of genome replication, interactions of NS5A with several host proteins and the sub-cellular localisation of NS5A. To investigate this further, we raised an antiserum to S225 phosphorylated NS5A (pS225). Western blot analysis revealed that pS225 was exclusively found in the hyper-phosphorylated NS5A species. Furthermore, using kinase inhibitors we demonstrated that S225 was phosphorylated by casein kinase 1α (CK1α) and polo-like kinase 1 (PLK1). Using a panel of phosphoablatant mutants of other phosphorylation sites in LCSI we obtained the first direct evidence of bidirectional hierarchical phosphorylation initiated by phosphorylation at S225. Using super-resolution microscopy (Airyscan and Expansion), we revealed a unique architecture of NS5A-positive clusters in HCV-infected cells - pS225 was concentrated on the surface of these clusters, close to lipid droplets. Pharmacological inhibition of S225 phosphorylation resulted in the condensation of NS5A-positive clusters into larger structures, recapitulating the S225A phenotype. Although S225 phosphorylation was not specifically affected by daclatasvir treatment, the latter also resulted in a similar condensation. These data are consistent with a key role for S225 phosphorylation in the regulation of NS5A function. Importance NS5A has obligatory roles in the hepatitis C virus lifecycle, and is proposed to be regulated by phosphorylation. As NS5A is a target for highly effective direct-acting antivirals (DAAs) such as daclatasvir (DCV) it is vital to understand how phosphorylation occurs and regulates NS5A function. We previously identified serine 225 (S225) as a major phosphorylation site. Here we used an antiserum specific for NS5A phosphorylated at S225 (pS225-NS5A) to identify which kinases phosphorylate this residue. Using super-resolution microscopy we showed that pS225 was present in foci on the surface of larger NS5A-positive clusters likely representing genome replication complexes. This location would enable pS225-NS5A to interact with cellular proteins and regulate the function and distribution of these complexes. Both loss of pS225 and DCV treatment resulted in similar changes to the structure of these complexes, suggesting that DAA treatment might target a function of NS5A that is also regulated by phosphorylation.

image of pS225 (green) and total NS5A (red) signals. As expected, the pS225 156 antiserum exhibited no reactivity against the S225A mutant. and there was a subset of total NS5A reactivity that did not stain for pS225. This 166 observation is developed further later in this manuscript (Figs 7,9). The specificity of 167 the pS225 antiserum was further confirmed by a lack of cross-reactivity with either the 168 S225A phosphoablatant NS5A or control cells (Fig 1D, E). Taken together these data 169 validated the subsequent use of this unique reagent as a tool to investigate the 170 properties of S225-phosphorylated NS5A in more detail.

171
Identification of kinases that phosphorylate S225 172 We proceeded to use the antiserum to investigate which cellular kinases might be by western blot for total NS5A and pS225-NS5A (Fig 2A, B). The effects of the 182 inhibitors on HCV genome replication were also assessed in parallel using Huh7 cells 183 stably harbouring an HCV SGR expressing firefly luciferase-neomycin 184 phosphotransferase (Feo) (Fig 2C). Cell viability was monitored by MTT assay (Fig   185   2D). 186 Both D4476 and SBE13-HCl significantly reduced pS225 and p58 abundance in a 187 dose-dependent manner (Fig 2A). Quantification of multiple western blots revealed 188 that at the maximal non-cytotoxic concentrations (D4476: 50 µM, SBE13-HCl: 40 µM) 189 phosphorylation of S225 was reduced by approximately 70% (Fig 2B). Consistent with the reduction in both pS225 and p58 abundance, D4476 or SBE13-HCl at the maximal 191 non-cytotoxic concentrations also resulted in a 75-fold or 11-fold reduction in RNA 192 replication respectively (Fig 2C). In contrast to the inhibitors of CKIα and PLK1, 193 treatment with GSK-3β Inhibitor VIII had no effect on pS225 phosphorylation or HCV 194 RNA replication (Fig S1). We conclude that S225 is a substrate for phosphorylation 195 by either of the two cellular kinases, CKIα and PLK1, but is not phosphorylated by 196 GSK-3β.

197
In parallel inhibitor-treated HCV-infected cells were analysed using high resolution 198 confocal microscopy (Airyscan) to examine effects on the subcellular distribution of 199 total NS5A, pS225-NS5A and LDs. The latter were chosen as we, and others, had LDs. (Fig 3B).

208
In comparison, treatment with SBE13-HCl had a less dramatic effect on the distribution 209 of pS225-NS5A (Fig. 4A), although a degree of LD accumulation was observed.  In a previous study (43) we generated an antibody specific to pS222. Although this 237 reagent was in scarce supply, we had sufficient to interrogate the same samples for 238 the presence of pS222. As expected, S222A exhibited significantly reduced levels of pS222 within the p58 species (Fig 5B), however this mutant retained wildtype level of 240 reactivity for the p56 species. We believe this reflects reactivity against the non-241 phosphorylated sequence and is due to the fact that this serum was not affinity purified 242 using the phospho-peptide (as had the pS225 antiserum). Importantly, the only other 243 phosphoablatant mutant that blocked pS222 was S225A. These observations are 244 consistent with the suggestion that S225 phosphorylation is the 'priming' event that 245 leads to a bi-directional hierarchical cascade of phosphorylation events. 246 We also used confocal microscopy with Airyscan to analyse the distribution of both  Fig 8C). We therefore applied this technique to interrogate pS225-NS5A and total 287 NS5A distribution in Huh7 cells electroporated with mJFH-1 RNA at different times.
As observed using Airyscan, ExM analysis revealed a diffuse distribution of NS5A 289 throughout the cytoplasm ( Fig 9A) with significant co-localisation between NS5A and 290 pS225 (merge panel). Over time, the diffuse distribution resolved into a more 291 clustered appearance with an overall increase in the size of NS5A-positive structures.

294
For comparison, cells harbouring wildtype or S225A mutant SGR-Neo-JFH1 were also 295 analysed by ExM. As expected wildtype NS5A was distributed throughout the 296 cytoplasm ( Fig 10A), whereas S225A was restricted to the perinuclear region (Fig   297   10B). Interestingly, the proportion of NS5A punctae that were pS225 positive was 298 higher than for virus-infected cells (Fig 10B green bars). Wildtype NS5A positive 299 punctae were on average smaller (peak 0.09-0.12 μm 2 ) than those in infected cells,   To quantify this we measured intensity profiles across the clusters (Fig 12). In both in the serine-rich LCSI between DI and DII, for reasons outlined in the introduction.

347
Here we describe the use of an unique reagent -an antiserum specific for pS225 that 348 has allowed us for the first time to directly interrogate the role of S225-phosphorylated 349 NS5A.

350
Firstly, we showed that pS225 was exclusively present on the p58 NS5A species and 351 identified CKIα and PLK1 as candidate S225 kinases. Pharmacological inhibition of 352 either kinase resulted in a loss of pS225 and p58, but also resulted in a reduction in 353 the overall levels of NS5A. This is likely due to an effect on genome replication and is 354 consistent with the 10-fold reduction in replication seen previously with the S225A 355 phosphoablatant mutant (37). We also observed that inhibition of these kinases 356 resulted in a loss of co-localisation of NS5A with LDs, suggesting that this association 357 is required for efficient genome replication, in addition to its requirement for virus 358 assembly (19, 47).

359
Previously we had observed that the S225A phosphoablatant mutant had an extensive 360 phenotype (23, 37) and we considered this a disproportionate effect of a single serine-alanine substitution in a highly serine-rich region (Fig 1A). This prompted us to  We then combined the pS225 antiserum with two super-resolution microscopy 375 approaches -Airyscan and ExM -techniques which overcome the light diffraction limit 376 of conventional microscopy, to interrogate the subcellular distribution of pS225-NS5A 377 in comparison to the total pool of NS5A. This revealed that pS225-NS5A was only a 378 subset of the total, consistent with the presence of both p56 (lacking pS225) and p58 379 (pS225-positive) species by western blot (Fig 1B). However, surprisingly we observed 380 that pS225 was not uniformly distributed across the total NS5A (Fig 7). Three-381 dimensional reconstruction of NS5A clusters imaged by ExM revealed that pS225 foci 382 were surface exposed and in many cases were close to holes that extended through 383 the cluster (Fig 11). These holes are likely to represent LDs which cannot be directly    to the cluster that is subject to 3D-reconstruction in Fig 11. (B) The size of individual 573 NS5A and pS225-NS5A particles was determined and plotted as a frequency