Chikungunya virus (CHIKV) is a recently re‐emerged arbovirus that triggers autophagy. Here, we show that CHIKV interacts with components of the autophagy machinery during its replication cycle, inducing a cytoprotective effect. The autophagy receptor p62 protects cells from death by binding ubiquitinated capsid and targeting it to autophagolysosomes. By contrast, the human autophagy receptor NDP52—but not its mouse orthologue—interacts with the non‐structural protein nsP2, thereby promoting viral replication. These results highlight the distinct roles of p62 and NDP52 in viral infection, and identify NDP52 as a cellular factor that accounts for CHIKV species specificity.
Chikungunya virus (CHIKV) is a re‐emerged arbovirus belonging to the Alphavirus genus that causes febrile arthralgia in humans . Alphaviruses are enveloped viruses with a positive‐strand RNA genome, encoding non‐structural (nsP1 to nsP4) and structural proteins, the capsid, 3 envelope glycoproteins (E1, E2 and E3) and 6k peptide. Alphaviruses replication in the cytoplasm of vertebrate cells is cytolytic . During infection, nsPs associate with viral RNA to form replicative complexes (RC), and allow replication through a double‐stranded (ds)RNA replicative intermediate. RCs are associated with membranous structures called cytopathic vacuoles located in the perinuclear area . RCs of Sindbis (SINV) and Semliki forest (SFV) viruses, two alphaviruses, harbor endo‐ or lysosomal compartment markers [4, 5]. They also use the plasma membrane as a site of replication [6, 7]. Alphavirus infection is associated with cell shutoff and apoptosis. nsP2, an essential and multifunctional component of RCs, serves as a trigger for cell shutoff and induction of apoptosis in SINV‐ and CHIKV‐infected cells [8, 9]. These functions are related to nsP2 nuclear location and assigned to its carboxy‐terminal domain [9–11].
Autophagy is a cellular catabolic process, which sequesters cytosolic components within double‐membrane vesicles and targets them for degradation in lysosomes . While autophagy was initially thought to be non‐selective, evidence suggests a selective autophagic degradation of cytosolic material, including intracellular pathogens . By simultaneously binding to ubiquitin and LC3/GABARAP proteins, autophagy receptors such as p62 (SQSTM1) and NDP52 (nuclear dot protein 52 kDa) can mediate docking of ubiquitinated targets to autophagosomes [14–16]. While recent studies suggest that p62 and NDP52 might mediate antibacterial autophagy through different pathways [16–19], their respective contributions to selective autophagy remains unclear.
Selective autophagy is a well‐recognized innate immune response to infection [20–22]. Autophagy might exert anti‐ or pro‐viral roles and its impact on alphaviruses has been investigated [23–26]. Autophagy has been reported to limit the pathogenesis of CHIKV‐infected mouse cells . Yet, in human cultured cells, CHIKV triggers an autophagic response that promotes viral replication . The molecular mechanisms underlying these species‐specific differences are not known, and the role of autophagy on CHIKV remains unclear. Here we have uncovered distinct but complementary roles for the autophagy receptors p62 and NDP52 in the context of viral infection, and provide molecular evidence for the species specificity of CHIKV.
RESULTS AND DISCUSSION
Autophagy promotes CHIKV infection in human cells
We first investigated whether CHIKV induces autophagy in HeLa cells. A decrease of p62 protein and an increased conversion of LC3‐I to LC3‐II, both indicative of autophagy induction, were observed in infected cells (Figs 1A,B). Puncta of p62 clustered around the cytosolic capsid (Fig 1C) and capsid colocalized partially with GFP‐LC3‐B (Fig 1D). Stochastic optical reconstruction microscopy provided a high‐resolution image of p62 association with capsid (supplementary Fig S1A online). Ultrastructural analysis revealed double‐membrane vesicles, containing and surrounded by nucleocapsids, immunolabelled for capsid and p62 (supplementary Figs S1B,C online). Moreover, CHIKV induced autophagy in a mouse model for CHIKV (Figs 1E,F) .
Autophagy induction with rapamycin in infected human cells significantly increased CHIKV replication (15 h: 1.6±0.2‐fold, P⩽0.005; 24 h: 1.6±0.1‐fold, P⩽0.01) and production (15 h: 2.4±0.3‐fold, P⩽0.05; 24 h: 2.9±0.3‐fold, P⩽0.05), as well as the level of capsid (15 h: 1.6±0.06‐fold, P⩽0.001; 24 h: 1.5±0.11‐fold, P⩽0.01) (Figs 1G–I). No significant difference in cell viability between treated and untreated cells was observed (supplementary Fig S1D online). In contrast, treatment with wortmannin, a well‐characterized inhibitor of PI‐3 kinase activity and autophagy, restricted infection (supplementary Figs S1E,F online). Depletion of canonical mediators of autophagy, Beclin1 (Atg6) and Atg7, significantly increased virus‐induced cell death (Beclin1: 1.7±0.3‐fold, P⩽0.05; Atg7: 1.8±0.1‐fold, P⩽0.001) (Fig 1J) and decreased CHIKV replication (Beclin1: 1.3±0.2‐fold, P⩽0.05; Atg7: 1.8±0.04‐fold, P⩽0.001), production (Beclin1: 1.7±0.2‐fold, P⩽0.05; Atg7: 2.4±0.4‐fold, P⩽0.05) (Figs 1K,L), and level of capsid (Beclin1: 1.5±0.1‐fold, P⩽0.001; Atg7: 3.1±1.1‐fold, P⩽0.01) (Fig 1M). Altogether, these data suggest a cytoprotective effect and an overall proviral role of autophagy in CHIKV‐infected human cells.
Distinct effects of p62 and NDP52 on CHIKV infection
As NDP52 is thought to function similarly to p62, we investigated its subcellular location in infected HeLa cells. NDP52 was observed in close proximity to a fraction of capsid (Fig 2A; supplementary Figs S2A,B online) in association with the TGN46‐positive perinuclear region identified as the trans‐Golgi network (TGN) (supplementary Fig S2C online). Strikingly, NDP52 formed clusters with capsid in cytoplasmic areas where p62 was absent (Fig 2B; supplementary Fig S2D online). Moreover, NDP52‐capsid clusters did not colocalize with GFP‐LC3‐B (supplementary Fig S2E online), but clustered with LC3‐C (Fig 2C). E2 also localized near NDP52 in the perinuclear region (supplementary Fig S2F online) but neither with GFP‐LC3‐B nor with p62 (supplementary Figs S3A,B online). As p62 and NDP52 are both ubiquitin‐binding proteins, we investigated their location relative to capsid and ubiquitin. Co‐localization of p62 with capsid was detected in 71%±2.0% of infected cells, in which p62 puncta were observed, and in 98%±0.8% of these cells, ubiquitin partially colocalized with capsid co‐clustered with p62 (Fig 2D; supplementary Fig S3C online). Similar results were obtained in primary human labial fibroblasts (HLFs) (supplementary Fig S3D online). In contrast to p62 (Fig 2D), NDP52 did not localize with ubiquitinated capsid (Fig 2E).
We next investigated the respective roles of p62 and NDP52 during CHIKV infection. As with depletion of canonical autophagy proteins Beclin1 and Atg7, p62 depletion significantly increased virus‐induced cell death (1.7±0.1‐fold, P⩽0.01) (Fig 2F). In contrast to canonical autophagy proteins, p62 depletion significantly increased CHIKV replication (GFP: 1.3±0.05‐fold, P⩽0.01; quantitative reverse transcription polymerase chain reaction (qRT‐PCR): 1.9±0.4‐fold, P⩽0.01), production (1.5±0.1‐fold, P⩽0.01) (Figs 2G,H) and capsid level (1.5±0.08‐fold, P⩽0.001) (Fig 2I). Similar results were obtained in HLFs (supplementary Figs S3E,F online). Depletion of NDP52 significantly increased virus‐induced cell death (1.7±0.2‐fold, P⩽0.01) (Fig 2J). However, in contrast to p62 depletion, NDP52 depletion significantly decreased CHIKV replication (GFP: 1.4±0.2‐fold, P⩽0.05; qRT‐PCR: 2.3±0.1‐fold, P⩽0.001), production (1.9±0.1‐fold, P⩽0.01) (Figs 2K,L) and capsid level (2.1±0.6‐fold, P⩽0.01) (Fig 2M). Similar results were obtained in HLFs (supplementary Figs S3G,H online). Together, these results show that p62 and NDP52 protect cells from virus‐induced death, but that they differentially affect CHIKV infection in human cells.
p62 targets capsid to autophagolysosomes
p62 localized with ubiquitinated capsid in infected cells (Fig 2D). In addition, capsid and ubiquitin were reciprocally co‐immunoprecipitated from lysates of infected cells (Figs 3A,B) and from capsid and ubiquitin co‐transfected cells (Fig 3C). p62 co‐immunoprecipitated with capsid (Fig 3D), which was recognized by an anti‐ubiquitin antibody (Fig 3E). p62 immunoprecipitation of capsid depends on its ubiquitination, as p62deltaUBA mutant (which lacks the p62 ubiquitin binding region) did not immunoprecipitate capsid, in contrast to p62WT (Fig 3F). Moreover, the proportion of ubiquitinated capsid increased upon depletion of p62 but not upon depletion of SMURF1, an ubiquitin ligase implicated in alphaviral infection (Fig 3G; supplementary Fig S4 online) . p62 and capsid were associated in LAMP‐1‐positive compartments (Fig 3H) and capsid level increased in bafilomycin‐treated infected cells, confirming that it is degraded by autophagy (Fig 3I).
To elucidate the mechanism of virus‐induced cell death upon p62 depletion, we investigated the cytotoxicity of capsid. Capsid was cytotoxic (Fig 3J), in a dose‐dependent manner (Fig 3K) and its cytotoxicity is increased in p62‐depleted cells (Fig 3L). Moreover, infected cells overexpressing p62WT displayed lower virus‐induced cell death (1.6±0.1‐fold, P⩽0.01), while overexpression of a p62 variant, either p62deltaUBA or p62deltaLIR (p62 lacking the LC3 binding region), displayed higher virus‐induced cell mortality (p62deltaUBA: 1.6±0.1‐fold, P⩽0.01; p62deltaLIR: 1.5±0.1‐fold, P⩽0.01) (Fig 3M).
Taken together, these results suggest that p62 targets ubiquitinated capsid for autophagic degradation and that the cytoprotective effect of p62 is linked to its ability to bind ubiquitinated capsid. It has been reported that p62 targets SINV capsid in an ubiquitin‐independent and SMURF1‐dependent manner [24, 30]. These observations highlight differences between CHIKV and SINV, and are consistent with the ability of p62 to target both ubiquitinated and non‐ubiquitinated substrates to autophagic degradation .
Human NDP52 binds nsP2
We next investigated the role of NDP52 in infected human cells. We have identified NDP52 as an interactor of nsP2 by yeast two‐hybrid screen . We show here that NDP52 co‐immunoprecipitates with nsP2 in infected cells (supplementary Fig S5A online). By using the yeast two‐hybrid system, we found that the coiled‐coil domain of NDP52 and the C‐terminal domain of nsP2 are both required for NDP52–nsP2 interaction (Fig 4A; supplementary Fig S5B online). Moreover, NDP52 colocalized partially with nsP2 at the perinuclear region (Fig 4B), in the vicinity of the areas enriched in capsid (supplementary Fig S5C online). Co‐localization of NDP52 with nsP2 was detected in 68.9%±1.1% of infected cells (supplementary Fig S5D online). Similar results were obtained in HLFs (supplementary Fig S5E online).
Previous studies have shown that the cytoplasmic fraction of nsP2 is an essential component of RCs . We assessed whether NDP52 and nsP2 localized with other components of RCs, such as nsP3 and dsRNA. nsP2 localized with nsP3 in a TGN46‐positive perinuclear region, as does NDP52 (Fig 4C). NDP52 clustered with dsRNA in a perinuclear area enriched in capsid (supplementary Fig S5F online), where protein translation was detected (supplementary Fig S5G online). TGN‐associated RCs (labeled by nsP2), in the vicinity of de novo protein synthesis (labeled by puromycin) where NDP52 concentrates, were detected in 68.9%±2.9% of cells. Strikingly, depletion of NDP52 resulted in the disappearance of nsP2 and puromycin co‐labeling, and TGN‐associated RCs in the vicinity of protein synthesis were only detected in 30.0%±1.9% of cells (P⩽0.001) (Fig 4D). These data indicate that NDP52 localizes in the vicinity of TGN‐associated RCs, which contain nsPs and dsRNA, near sites of protein translation, and suggest that NDP52 promotes CHIKV replication by its interaction with nsP2. RNA viruses target their proteins to specific cellular compartments to anchor their RCs and establish assembly sites. RCs are targeted to these subcellular membranes by nsPs . Although the precise mechanism by which NDP52 promotes CHIKV RNA replication remains to be fully understood, our results suggest that nsP2, via NDP52 that is itself linked to LC3‐C, allows the anchorage of RCs to the TGN membrane.
The nuclear fraction of nsP2 mediates cell shutoff and death . When expressed in HeLa cells, nsP2 was cytotoxic (3.7±0.2‐fold, P⩽0.001) (Fig 4E). This cytotoxicity depended on its role in cell shutoff, as expression of a mutated nsP2 that fails to induce shutoff (nsP2R606A)  did not increase mortality (Fig 4E). Strikingly, depletion of NDP52 resulted in increased amount of nuclear nsP2 (Fig 4D) and an increased ratio of nuclear/cytosolic nsP2 (Fig 4F). NDP52 depletion led to a reduction in the total amount of puromycin‐labeled proteins either upon nsP2 transfection (supplementary Fig S5H online) or upon CHIKV infection (Fig 4G), suggesting that NDP52‐nsP2 interaction limits cell shutoff induced by nsP2. Moreover, nsP2‐mediated cell mortality in NDP52‐depleted cells was increased (Fig 4H) and expression of NDP52 in nsP2‐transfected cells showed a dose‐dependent decrease of the toxicity of ectopically expressed nsP2 (supplementary Fig S5I online), indicating that nsP2‐induced cell death is modulated by NDP52. Together, these results indicate that NDP52–nsP2 interaction in the cytoplasm restricts the amount of nuclear nsP2 and correlates with a reduced cell shutoff that limits nsP2‐induced cell death.
NDP52–nsP2 interaction is species specific
Autophagy plays an antiviral role in CHIKV‐infected mouse cells  and a proviral role in CHIKV‐infected human cells . To better understand these apparently conflicting results, we investigated the role of mouse NDP52 (mNDP52). In contrast to what was observed in human cells, we observed no differences in virus‐induced cell death and viral production in murine embryonic fibroblast (MEF) cells depleted for NDP52 versus control MEFs (Figs 5A,B). Sequence alignment revealed that mNDP52 exhibits 31% homology with human NDP52 (hNDP52) and lacks part of the corresponding hNDP52 C‐terminal region (supplementary Fig S6A online). No interaction between nsP2 and mNDP52 was observed by yeast two‐hybrid assay (Fig 5C). To further investigate the effect of mNDP52 and hNDP52 on CHIKV infection, hNDP52 was expressed in mouse cells. hNDP52 co‐immunoprecipitated with nsP2 in mouse cells (Fig 5D), yet its expression did not increase viral production (Fig 5E). It has recently been shown that hNDP52 binds LC3‐C (supplementary Fig S6B online)  that is expressed in human but not in mouse cells, in contrast to LC3‐A and LC3‐B. Co‐expression of hNDP52 and human LC3‐C (hLC3‐C) in mouse cells led to a strong increase in viral production (6.1±0.25‐fold, P⩽0.001), as compared to control cells or cells expressing either hNDP52 or hLC3‐C (Fig 5E; supplementary Fig S6C online), without affecting mortality (Fig 5F). Consistent with this latter observation, nsP2 did not modulate host protein synthesis in MEFs (supplementary Fig S6D), and therefore did not affect cell survival, as it did in human cells.
Consistent with an absence of a proviral role of the autophagy machinery in mouse cells, we observed an increase in virus‐induced cell death (2.5±0.1‐fold, P⩽0.01) (Fig 5G) and viral production (18.4±3.1‐fold, P⩽0.01) (Fig 5H) in atg5−/− MEFs, as compared to wt MEFs. As in human cells (Fig 2), mouse p62 immunoprecipitated capsid (Fig 5I) and depletion of p62 in mouse cells significantly increased virus‐induced cell death (1.6±0.03‐fold, P⩽0.01) and viral production (5.5±0.2‐fold, P⩽0.01) (Figs 5J,K). Together, these results show that, in contrast to human cells, autophagy is antiviral in mouse cells and indicates that CHIKV species specificity is linked, at least in part, to the ability of NDP52 to bind both nsP2 and LC3‐C, leading to a proviral effect. Recently, it has been shown that mNDP52 targets Mycobacterium tuberculosis to autophagic degradation, showing that mNDP52 might act as a canonical autophagy receptor in mouse cells . These data indicate that the ability of hNDP52 to bind both hLC3‐C and nsP2 is implicated in its species‐specific proviral effect on CHIKV. These results reconcile the two apparently conflicting reports, which showed that autophagy plays an antiviral role in mouse cells  but a proviral role in human cells . Moreover, these results suggest that the lower permissiveness to CHIKV of the mouse relative to humans  might result, at least in part, from the absence of the proviral role of NDP52 in the mouse.
Finally, we investigated whether the ability of nsP2 to bind hNDP52 is common to Old World alphaviruses. We showed by yeast two‐hybrid assay that hNDP52 binds nsP2 of SINV and SFV (supplementary Fig S6E online), indicating that the proviral role of hNDP52 likely applies to Old World alphaviruses and suggesting a putative general involvement of hNDP52‐LC3‐C in alphaviral replication.
Here we provide insight into the mechanisms by which autophagy receptors target CHIKV. We have shown that the autophagy machinery globally prevents CHIKV‐induced cell death and that p62 and NDP52 play anti‐ and pro‐viral roles in human cells, respectively (supplementary Fig S7 online). We also reveal a novel mechanism by which NDP52 promotes viral infection in a species‐specific manner. The absence of NDP52–nsP2 interaction in mouse cells might account for the lower permissiveness of mice to CHIKV relative to humans . These results shed light on the cell biology of CHIKV, pave the way for humanized hNDP52/LC3‐C mouse model to study CHIKV infection and for the development of antiviral strategies against alphaviruses.
Infection. HeLa cells were infected at a multiplicity of infection (MOI) of 5, which results in a synchronized infection of 100% of cells during 15 or 24 h (supplementary Fig S8 online). All other cell lines were infected at the same MOI as HeLa cells.
CHIKV replication and production. CHIKV replication was determined by measuring the GFP mean intensity per live cell in mock‐infected and infected cells, and by real‐time qRT‐PCR. GFP mean intensity was analyzed by fluorescence‐activated cell sorting (FACS) with a FACSCalibur instrument (BD Biosciences) and the results were analysed with Cell Quest Pro software (Cell Quest). CHIKV production was determined by titration of the supernatant by tissue cytopathic infectious dose 50 (TCID50) on Vero cells. Viral titers are expressed as TCID50 per cell .
Cell viability assays. Viability of cells was determined using the Scepter handled automated Cell Counter (Ref no. PHCC00000 from Millipore) with 60 μm Scepter tips (Ref no. PHCC60050 from Millipore).
Ethics statement. This study was carried out in strict accordance with the recommendations from the Guide for the Care and Use of Laboratory Animals, as provided by the French Ministry of Agriculture and of the European Union.
Statistical analysis. Student's t‐tests were performed for all experiments. The level of significance is shown in each figure (*P⩽0.05, **P⩽0.01, ***P⩽0.001).
We are grateful to S. Higgs for providing CHIKV cDNA clones and to N. Mizushima for providing Atg5−/− and Atg5+/+ MEFs. We thank V. Sancho‐Shimizu, C. Prehault, M. Hamon, L. Radoshevich, P. Desprès, A. Merits, V. Meas‐Yedid Hardy, E. Meurs and A. Dufour for their generous gift of reagents or technical advices. We also thank members of the Biology of Infection Unit, F. Rey, P. Cossart and P. Codogno, for their helpful discussions and critical reading of the manuscript. This work was supported by Institut Pasteur, FRM, Ville de Paris, Fondation BNP‐Paribas, ICRES FP7, LabEx IBEID and the Programme Interdisciplinaire CNRS Maladies Infectieuses Emergentes. D.J. was supported by a doctoral fellowship from the Axa Research Fund and M.B. by a doctoral fellowship from Région Ile‐de‐France.
Author contributions: D.J., S.M., M.C.P., M.L., P.O.V., T.C. and M. Lecuit planned the experiments; D.J., S.M., M.B., N.G., M.L., M.L.H., N.C., M.C.P., P.O.V., T.C. performed the experiments, D.J., S.M., M.L., Y.J., M.C.P., P.P., F.T., C.Z., P.O.V., T.C. and M. Lecuit analysed the experiments; D.J., S.M., T.C. and M. Lecuit designed the overall research; D.J., S.M., T.C. and M. Lecuit wrote the manuscript.
Conflict of Interest
The authors declare that they have no conflict of interest.