Autophagy Modulation in Mammarenavirus Infection

Mammarenavirus genus groups viruses causing human haemorrhagic diseases, including the New World (NW) Junín virus (JUNV), and the Old World (OW) viruses Lassa (LASV), among others. The high mortality and morbidity rates associated to pathogenic mammarenaviruses, the absence of vaccines and the constant threat of new emerging species, make these viruses a public health concern in endemic areas. Autophagy is a widely-known intracellular metabolic pathway involved in maintaining the cellular homeostasis in response to several stress conditions. Autophagy Modulation in Mammarenavirus Infection


Introduction
Junín virus (JUNV), a member of Mammarenavirus genus within the Arenaviridae family, is the etiological agent of Argentine Haemorrhagic Fever (AHF), a potentially lethal, endemic-epidemic disease affecting the population of the most fertile farming land of Argentina 1 . Mammarenavirus genus groups viruses causing human HF diseases, including the New World (NW) viruses JUNV, Machupo, Guanarito, Sabia, and Chapare, and the Old World (OW) viruses Lassa (LASV) and the recently emerging Lujo virus. These two groups of mammarenavirus are divided based on phylogenetic, serological, and geographical distribution. Mammarenaviruses are negative-sense single strand RNA virus, with a bi-segmented genome consisting of a large (L) and small (S) molecule 2 . The L segment encodes the RNA-dependent RNA polymerase and the matrix protein (Z), a RING finger protein essential for viral morphogenesis and replication 3,4 . The S segment encodes the nucleoprotein (NP), responsible for nucleocapsid formation, and the Glycoprotein Precursor Complex (GPC). This polypeptide is exposed to proteolytic cleavage by cellular proteases into a stable signal peptide (SSP), GP1, and GP2 which remains associated and mediates recognition and entry into the target cell 5 . Mammarenaviruses represents a global threat with the emergence of new species as an increasingly possible event as the human population reaches new rural or previously inhabited territories. This aspect, together with the high mortality and morbidity rates associated to pathogenic mammarenaviruses and the absence of vaccines, makes these viruses a public health concern in endemic areas and led to their classification as category A pathogens by the U. S. National Institutes of Health (Emerging Infectious Diseases). To date, only Candid#1, a live-attenuated vaccine, is available in JUNV endemic areas from Argentina 6 .
Autophagy is a widely-known intracellular metabolic pathway involved in maintaining the cellular homeostasis in response to several stress conditions 7,8 . However, autophagy also comprises an important activator of the innate and adaptive immunity 9,10 . Indeed, autophagy triggering is a commonly-used strategy by the cells to restrict viral infections, which can directly degrade the virus, and concomitantly regulate the innate and adaptive immunity to promote virus clearance 11,12 . This virus degradation leads to the activation of the Pattern   Recognition Receptor signalling-cascade to finally induce  type I interferon (IFN-I)-mediated viral elimination 11,12 .  Thus, although the main function of autophagy during a  viral infection is to counteract the infection, many viruses,  most of them single-stranded RNA viruses, are able to  hijack the autophagic pathway to facilitate viral replication,  immune evasion and release 13,14 .

JUNV Promotes Autophagy
In our previous work, we analysed the autophagic response in JUNV infection by using the IV 4454 JUNV strain 15 . We showed that JUNV triggers the accumulation of autophagic vesicles in a Beclin-1 and Atg5-dependent manner in permissive human A549 cells from 2 h postinfection (p.i.), indicating the early activation of the autophagic pathway after viral infection. Even though the precise role of autophagy during JUNV infection remains unclear to date, our results indicate that autophagy promotion has a proviral role toward JUNV replication, providing important knowledge to a previously unknown association in the field of host cell-arenavirus interactions 15 . Almost simultaneously, but using the more-virulent strain P3441 of JUNV as the viral model, the group of Perez Vidakovics also reported the autophagy-triggering by JUNV and both groups observed, in turn, that autophagy promotion enhances JUNV fitness in permissive A549 cells 16 . However, there were some differences in the results reported by both groups.
On one side, analysing early time-points after cell exposure, i.e. 2 and 6 h p.i., we observed that UV-inactivated JUNV infected cells were unable to promote autophagy, showed, by western blot and fluorescence microscopy, an increased level of LC3-II lipidation (a well-established autophagosome indicator) after 24 h of cell exposure to UV-inactivated JUNV P3441 particles, suggesting that the autophagy activation was independent of viral replication. This discrepancy could be explained, in part, by the difference in the virulence between both strains used by us and Perez Vidakovics´s group. While we used the naturally-attenuated IV 4454 JUNV strain (isolated from a mild human case of AHF in 1970 17 ), Perez Vidakovics et al. used the P3441 strain, isolated from another patient with AHF 18 . Certainly, virus virulence is determined by the virus-cell host net of interactions, so it becomes perfectly suitable to hypothesize that P3441 strain is able to induce autophagy by an alternative mechanism independently of viral replication, that is, by direct interaction of a structural component of the viral particle with the autophagy machinery once inside the cell. However, according to our results, GP signal was no longer observed 6 h p.i. by confocal laser scanning microscopy (CLSM), suggesting that the non-replicative inactivated-JUNV particles would be already degraded by 6 h p.i. This observation was not surprising to us since the UV light exposure generates RNA-protein cross-link that blocks genome transcription, eliminating any standard virus activity, leaving the particles to the mercy of cellular degradation mechanisms 19,20 . Indeed, it would be interesting to confirm the presence of intact inactivated viral particles after 24 h inside the cell, and dissect the mechanism that mediates the autophagyinduction by these particles.
Baird et al. found JUNV replication-transcription complexes (RTCs) associated with cellular membranes containing NP and mRNA , thus it was feasible to postulate that JUNV triggers autophagy to stimulate an intracellular membrane reprogramming that provides membrane scaffolds necessary to the formation of RTCs 21 . With this in mind, we analysed the possible co-localization of JUNV NP with the autophagic marker LC3 protein along the first 24 h of infection. While we did not detect co-localization either at early or late times p.i., i.e. 2 and 24 h p.i., respectively, Perez Vidakovics et al. observed co-localization of both structures at 24 h p.i. Regarding this bifurcated results in the distribution of viral and autophagic structures, in addition to the already-mentioned difference in strains virulence that could explain these observations, we found in methodological aspects the most feasible explanation. While we infected the cells with a multiplicity of infection (MOI) of 1, Perez Vidakovics et al. used a MOI of 3 and reported a Pearson's coefficient of 0.71, so it could be possible that we missed the detection of NP-LC3 colocalization due to sensitivity issues within our approach. The authors suggested that the co-localization detected may implicate that the autophagosome membranes are involved in the RTCs assembly 16 . However, in the work by Baird and colleagues, they failed to detect co-localization of Candid#1 RTCs with LC3 by CLSM 21 . Indeed, further experiments are needed to elucidate the biological meaning of NP interaction with components of the autophagy machinery.

JUNV, Autophagy and ER Stress
Autophagy-triggering occurs by different stress signals to restore cellular homeostasis. Nutrient or growth factors deprivation, an excess of reactive oxygen species, aggregated or misfolded proteins, old or damaged organelles and several pathologic conditions are some of them 7,22 . A common effect of viral infections is the stress of the endoplasmic reticulum (ER), which is produced either by the exacerbated accumulation of viral proteins or the exploitation of ER membranes for viral replication 10,23 . To counteract the ERstress, cells have evolved with the unfolded protein response (UPR) 10,23,24 . For JUNV, Paessler's group recently showed that Candid#1 GPC (the protein that contains the most studied attenuation determinants of this strain) induces ER-stress promoting its degradation within lysosomal compartments 25,26 . They reported that a single amino acid substitution (T168A) in Candid#1 GPC resulted in the loss of an N-linked glycosylation motif and was the primarily responsible for the GPC retention in the ER, promoting the stress of the organelle. In our work, the JUNV-induced ER-stress was evaluated at early times p.i. by detecting the intracellular level of calnexin (CNX), a well-known ERstress-induced protein 27 . However, we observed similar CNX levels in infected and non-infected cells, suggesting that the IV 4454 JUNV strain triggers the autophagy pathway independently of the ER-stress. A possible explanation for these differences resides in the pathway of attenuation of both strains. While IV 4454 is a naturally-attenuated strain, Candid#1 was attenuated after several animal passages and tissue culture cloning from the more pathogenic strain XJ 44 28 . In fact, we observed that IV 4454 GPC does not possess the T168A substitution found in Candid#1.

Arenavirus and Autophagy: State of the Art
The Z protein is a small protein with key roles during the replication cycle thanks to its capacity of interaction with viral and cellular factors 3,4 . Hallam et al. reported recently that a single mutation in the RING domain of the Candid#1 Z protein is able to confer attenuation to the pathogenic Romero JUNV strain. This demonstrates that, besides the extensively studied GP protein, the Z protein is also an important determinant in Candid#1 strain attenuation 25 Atg5 and Beclin-1 proteins, is required for efficient virus replication. JUNV NP and LC3-II co-localization is not completely clear 15,16 . MOPV induces the autophagy pathway, but not LASV. 7) OW LASV and MOPV Z proteins, a late gene product in the viral replication cycle, interact with the autophagic adaptors NDP52 and TAX1BP1, and the complete infectious cycle requires the presence of Atg5 30 . 8) Integrating the data, we propose that early p.i. the level of JUNV NP would be insufficient to reach its demonstrated type I interferon (IFN)-I antagonist role [32][33][34] . At that moment, the autophagy induction by the virus would offer augmented possibilities for Atg5 (from the Atg5-Atg12 complex) interaction with RIG-I to inhibit IFN-I response 31 . 9) GPC is translated in the endoplasmic reticulum (ER). JUNV Candid #1 GPC translation induces ER-stress, but not JUNV IV 4454 , thus JUNV IV 4454 -induced autophagy is due to an additional autophagy induction mechanism, unknown to date 15,25 . Solid and dotted-arrows indicate known and not completely known/proposed processes, respectively. and coimmunoprecipitation. They detected interaction with two well-known autophagic adaptors, NDP52 and TAX1BP1 30 . Also, they reported that, in Atg5 (an essential early autophagic protein) siRNA silenced HeLa cells, less viral RNA and fewer infection particles were produced upon MOPV and LASV infection. These results suggest that autophagy initiation structures, i. e. NDP52, TAX1BP1 and ATG5, play an important role in MOPV and LASV replication cycles 30 . Moreover, they observed that, 2 days after infection of permissive human HeLa cells, MOPV induced autophagy, which was required for efficient production of newly-formed infectious particles, suggesting a proviral role of autophagy. However, no autophagy-triggering was observed in LASV-infected cells. These results led the authors to suggest that autophagy may be required during different steps of MOPV and LASV replication cycles 30 .
A common observation between LASV, MOPV and JUNV with the autophagy machinery is their dependency on Atg5 expression for efficient virus replication 15,16,30 . Jounai et al. showed that autophagy may suppress the innate immune signalling via Atg5-Atg12 complex interaction with retinoic acid-inducible gene I pathway, leading to an impaired IFN-I response 31 . In their report, the authors observed an increase of IFN-I production in response to Vesicular Stomatitis Virus RNA in Atg5 deficient cells, inhibiting viral replication 31 . This IFN-I-inhibitory role of Atg5 suggests that autophagy may also have a regulatory role modulating the host innate antiviral response. In this context, one hypothesis is that the lower viral yield observed for JUNV, LASV and MOPV infection of Atg5 deficient cells, may be a consequence of enhanced IFN-I response rather than due to a deficiency in the autophagy flux. On the other side, it is well established that the NP protein of several mammarenaviruses acts as IFN-I antagonist in vitro 32 . So, putting all together and with the aim of depicting a unique scenario during JUNV infection in vitro, it becomes tempting to hypothesize that early in the infection (i.e. 2 h p.i.), the low levels of NP may not be sufficient to counteract the IFN-I response. Autophagy-triggering could comprise a viral strategy to promote an initial mechanism to downregulate the IFN-I response. Then, at later times during the infection, an increased level of NP due to viral genome replication and protein translation could "take on the task" to down-regulate IFN-I antiviral action 33,34 . We have included all the discussed data and proposed hypothesis in Figure 2.

Conclusion
Taken all together, autophagy has a proviral role during JUNV, LASV and MOPV infection where the early structural autophagic protein Atg5 becomes crucial. To date, there is increasing evidence suggesting a proviral role of autophagy for several RNA viruses. For instance, footand-mouth disease virus triggers autophagy early times p.i. which is dependent on Atg5 but, contrary to what we observed for JUNV, independently of viral replication 11,12,35 . Some flaviviruses, including Dengue and Zika viruses, induce LC3-II-containing vesicles to regulate the storage of cellular lipids (a process known as lipophagy) and, in the case of Zika, promote the virus release 11,12 . Others, like Hepatitis C virus, its infection triggers autophagy through the UPR, which restrict IFN-I induction and so, promotes the viral replication 12,35 . This ER-stress response could be similar to the one reported within Candid#1 infection but it seems not to be implicated during JUNV IV 4454 autophagy induction 15,26 .
Regardless of slight differences among the observations, the evidences discussed in this work robustly demonstrate that JUNV triggers autophagy early after infection to establish an efficient viral infection in a human permissive cell line. Deeper exploration is needed to dissect if JUNV promotes autophagy to assist its replication, to suppress the innate immune signalling or a combination of both. All these results highlight the extensive interaction that might exist within mammarenaviruses and autophagy, and the different strategies developed by OW and NW arenaviruses in order to improve their fitness. Finally, and most importantly, these findings shed light on a new field of arenavirus host-cell interaction widening the panel of targets to develop new strategies to counteract these pathogens.