**4. Discussion**

Our detailed pathway analyses revealed the fact that flavivirus-directed JAK/STAT signaling inhibition extends far beyond classical IFN responses, with antagonism of this pathway found in response to every cytokine we investigated. We proposed a mechanistic model of flavivirus JAK/STAT antagonism in which flavivirus NS5 (possibly in conjunction with other NS proteins) interacts with host HSP90 at the site of viral RNA replication. As a consequence of NS5 interaction, the HSP90-kinase client homeostasis is disrupted, leading to inappropriate chaperoning of JAK proteins, loss of their activity, protein kinase instability, and subsequent degradation via the proteasome. This process reduces abundance of JAKs, with the remaining pool of kinases lacking chaperone-supported enzymatic activity. Inactive JAKs cannot transmit pY to STAT proteins, and thus infected cells become refractory to cytokines that signal through the JAK/STAT pathway, including IFNs, proinflammatory cytokines, and immune regulatory cytokines.

HSP90 has been well described as a regulator of JAK activity and stability [34,35,38]. Inhibition of HSP90 leads to degradation of JAK1 and JAK2 to attenuate the pY-STAT1 and 2 response to IFNβ and IFNγ [34]. In Hodgkin lymphoma cells, inhibition of HSP90 led to loss of JAK1, JAK3, and Tyk2, and constitutive pY of STAT1, 3, 5, and 6 was ablated [35]. Thus, HSP90 inhibitors phenocopy JAK/STAT antagonism by flaviviruses, a similarity of function that we confirmed with side-by-side analyses. Our data revealed for the first time that flavivirus NS5 binds HSP90 to mediate broad STAT inhibition, which extends upon descriptive reports from other groups that identified JEV, DENV, and ZIKV NS5 interaction with HSP90, but did not define the relevance of this interaction [39–41]. Several DENV proteins have also been shown to bind HSP90 [37]. Our demonstration of JAK destabilization as a consequence of HSP90–NS5 interaction provides important functional insight into viral evasion of host innate immunity and inflammatory signaling. Moreover, we showed that Akt and ErbB2 levels decrease during flavivirus infection as a result of NS5-HSP90 interaction, implying that processes mediated by these and other HSP90-client kinases are likely altered during infection, possibly contributing to viral pathogenesis.

We showed that HSP90 chaperone activity is not required for flavivirus replication or protein stability but that HSP90 inhibition leads to slight enhancement of infection. This contrasts with a recent report showing activity of grp94 (an endoplasmic reticulum-resident paralog of HSP90) is required for DENV and ZIKV infection [42]. Vero cells were specifically chosen for these experiments as they lack IFN production [43,44], and thus the moderately enhanced viral growth that we observed was unlikely due to further dysregulation of antiviral signaling. Rather, we propose that induction of HSP70 likely contributes to the greater WNV and ZIKV replication seen here upon HSP90 inhibition, as HSP70 is known to enhance flavivirus replication [26–28]. Thus, flavivirus interaction with HSP90 may represent a common strategy among flaviviruses to dysregulate and suppress host IFN antiviral signaling rather than directly supporting viral replication and protein function. The broad inhibition of pY-STAT resulting from the NS5-HSP90 interaction then imparts collateral dysregulation of IFN-independent cytokine actions.

The remarkable breadth of cytokine signaling inhibition imposed by the NS5–HSP90 interaction may have consequences for flavivirus disease phenotypes outside of the context of innate antiviral defenses. In the context of an acute immune response, flavivirus-infected myeloid cells (important targets of infection) [13,45,46] would be less responsive to anti-inflammatory IL10 and IL4 signaling through STAT3 and 6, respectively, perhaps altering the immune phenotype polarization of myeloid cells and enhancing immune-mediated tissue damage. Additionally, reduced response of flavivirus-infected antigen-presenting cells to IFNγ through pY-STAT1 may a ffect adaptive immunity priming actions to alter T cell-mediated immunity.

Our findings of HSP90 interaction share similarities with recent reports on involvement of HSP70 in flavivirus infection and host inflammatory signaling [26–28], suggesting chaperones are key regulators of pathogenesis. Flaviviruses productively infect and transmit between diverse hosts from arthropods (ticks and mosquitoes) to vertebrates (birds and mammals) [47]. Hence, flaviviruses must exploit diverse systems to promote replication; a setting that may favor the targeting of factors conserved between reservoirs and vectors. HSP70 and HSP90 are highly conserved between these organisms, both in structure and function [48,49], and thus it is feasible that strategies developed to exploit these chaperones could be viable across diverse hosts. Indeed HSP70 inhibitors antagonize flavivirus replication in both insect and mammalian cells [26,27], and JAK/STAT inhibition by flaviviruses in mosquito cells is dependent on NS5 and involves proteasomal activity [50]. Thus, we propose flaviviruses target HSPs to facilitate infection in diverse hosts, with HSP70 and grp94 [42] activity usurped for viral replication and HSP90 activity disrupted to antagonize innate immune defense. This division of functions regarding flavivirus interactions with cytoplasmic HSPs (i.e., interaction with HSP70 to promote viral protein folding vs. targeting of HSP90 to disrupt host immunity) is striking. These chaperones normally operate in close cooperation with one another [49], and thus this distinct separation of roles warrants further investigation to discern the consequences of this division for host cell function. Whether a HSP antagonism strategy is utilized to promote pathogenesis of other virus groups also remains to be investigated. As chaperones support so many cellular functions, the unforeseen medical consequences of viral HSP antagonism are likely substantial.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2073-4409/9/4/899/s1: Figure S1: WNV infection inhibited STAT phosphorylation in response to multiple cytokines. Figure S2: Recombinant WNV protein expression did not induce pY-STAT3 or pY-STAT1. Figure S3: HSP90 interacted with NS5 and was localized to sites of viral dsRNA. Figure S4: HSP90 inhibition did not perturb WNV replication. Table S1: Reagents and resources used.

**Author Contributions:** J.A.R. conceived and performed experiments, wrote the manuscript, and secured funding. K.E.-N., E.C.D.-V., M.R.F., L.D.H., E.A.H., J.S., A.Y.L, F.W.S., B.C.K., and A.S. contributed to the design and performance of experiments. A.F., J.E.S.-B., and R.S. provided reagents, expertise, and feedback. M.G.J. conceived experiments, secured funding, and provided supervision. All authors edited and approved the final manuscript.

**Funding:** This research was funded by National Institutes of Health grants T32AI007509, AI145296, AI143265, AI104002, AI100625, AI127463, and AHA Award 17POST33660907 (EAH). A Perkins Coie Award for Discovery supported J.A.R.

**Acknowledgments:** In addition to generous gifts acknowledged in Table S1, the authors thank the members of University of Washington Immunology and the Center for Innate Immunity and Immune Disease for their helpful discussions and insight throughout this study.

**Conflicts of Interest:** The authors declare no conflict of interest.
