Non-lethal stressors including thermal stresses activated the synthesis of HSPs and protected fish from bacterial infections [65, 66]. For example, treatments of the chemical compound TEX-OE and its variant induced HSPs non-lethally in salmon and gilthead sea bream (Sparus aurata), which protected these fish species from Vibrio anguillarum infection . Bacterial infections up-regulated HSP70 in rainbow trout (Oncorhynchus mykiss) and sea bream (Sparus sarba), which were suggested as defense responses of the hosts against the bacterial pathogens [67–71]. However, these inhibitory functions of HSPs in bacterial infections are not necessarily true for viral infections. Relatively more fish HSPs are thus far described as positive HFs for the relevant viruses [e. g., 16–20] compared to those designated as negative factors [21, 22]. As for HSPs involved in betanodavirus infection, HSP70-5 supports RGNNV multiplication in the grouper (Epinephelus coioides) cells GF-1  and HSP90-β also serves as a positive HF for the virus in GF-1 cells  and in the marine medaka (Oryzia melastigma) cells hMMES1 , which are consistent with our overexpression and knockdown experiments using OLHNI-2 cells (Figs. 5 and 6). In contrast, HSP27 in sea perch (Lateolabrax japonicus) exerts anti-RGNNV activities by regulating the apoptosis signaling pathway . In the present study, HSP70-1, HSP70-2, HSP90-α1, and HSP90-α2 as well as HSP70-5 and HSP90-β facilitated RGNNV multiplication in medaka cells (Figs. 5 and 6). In contrast, HSP70-8, HSP70-9, GRP94, and TRAP1 suppressed virus multiplication though further studies including knockdown each of the four genes are required to confirm this activity. One curious observation is that overexpression of HSP70-8 suppressed RGNNV multiplication in medaka cells, which is apparently contradict to the precedent report by Chang and Chi . A possible explanation for the discrepancy is that our data were obtained using OLHNI-2 cells derived from medaka, the host specificity of which is different from that of grouper used by Chang and Chi . Medaka is susceptible to SJNNV as well as RGNNV though the levels of SJNNV infections are less than those of RGNNV infections . Conversely, groupers are susceptible to RGNNV but entirely resistant to SJNNV [42, 43]. Chang and Chi  indicated that HSP70-8 functions as an RGNNV receptor or coreceptor which probably distinguishes the difference between SJNNV and RGNNV. Therefore, some protein other than HSP70-8 might function as a RGNNV receptor in medaka. In any event, more studies are required to know the whole picture of HSP isoforms involved in fish virus multiplication.
Among the three inhibitors used in this study, radicicol and 17-AAG inhibit specifically HSP90 activity through the inhibition of ATP binding to HSP90 . Quercetin inhibits the activity of the transcription factor, heat shock factor 1, resulting in down-regulation of broader members of HSPs including HSP70 and HSP90 [74, 75]. In this study, radicicol, 17-AAG, and quercetin had inhibitory effects on the RNA replication and progeny virus production of RGNNV (Fig. 1). These results are consistent with the previous studies that the treatment of the marine medaka cells hMMES1 with ganetespib or NVP-AUY922, the inhibitors of HSP90ab1, significantly decreased RGNNV entry into the cells . Furthermore, geldanamycin (GA, a prototype of 17-AAG) and radicicol negatively affected the replicative and multiplicative competence of frock house virus (FHV) belonging to the same family Nodaviridae as RGNNV . Similar negative effects of the HSP inhibitors were found for other virus species such as hepatitis viruses [76–79], influenza virus , Ebola virus , and herpes simplex virus . However, in contrast, the treatment of GF-1 cells with GA facilitated the multiplication of RGNNV . One possible reason for these contradictory findings is the difference in the dose of inhibitors used. While we treated the 5 µM concentration of 17-AAG, Chen et al  used the relatively low concentration of GA (1.5 uM). Since inhibition of HSP90 by GA is known to trigger simultaneously up-regulation of HSP90 , the concentration of GA used by Chen et al  might be sufficient to up-regulate HSP90 but not to inhibit HSP90 activity. Alternatively, a treatment of GF-1 cells with GA could give some unexpected effects on virus multiplication. Although the cytotoxic effect of GA on GF-1 cells was not tested by Chen et al , we confirmed that GA gave a harmful effect on medaka cells at the concentration of as little as 1 µM within 12 h, which prompted us to use 17-AAG instead of GA in this study.
In this study, overexpression and knockdown experiments on the HSP70 and HSP90 isoforms revealed that HSP70-1, HSP70-2, HSP70-5, Hso90-α1, Hso90-α2, and HSP90-β facilitated virus multiplication. Overexpression of these six isoforms increased the levels of viral multiplication from 1.69 to 3.63 times more than that of the control (Fig. 5B). However, the increasing rates were less than those observed in the previous study . One possible explanation for this discrepancy is that the expression levels of the exogenous HSP genes were not high enough to give good increasing rates in viral titer in this study because of the low transfection efficiencies (Fig. 5A). Similarly, knockdown of the endogenous genes encoding the six HSP isoforms decreased the levels of viral multiplication at the rates from 0.26 to 0.55 of that of the control (Fig. 6B), which is less efficient than the previous results reported by Su et al . Since the knockdown levels of the isoform mRNAs obtained in this study were around 40% of the control (Fig. 6A), more efficient knockdown of the HSPs should impair more efficiently virus multiplication.
We showed that HSP70-1, HSP70-2, HSP70-5, HSP90-α1, HSP90-α2, and HSP90-β played positive roles for betanodavirus multiplication. Surprisingly, all of these but HSP70-5 were up-regulated by virus infection exclusively among the ten HSP isoforms tested, though the ten HSP isoforms are known to be up-regulated by various physiological or pathological stimuli in human [26, 61, 82]. Su et al  reported that up-regulation of HSP70-5 occurred in GF-1 cells at 48 and 72 h post inoculation with RGNNV (MOI = 5). We could not examine the expression of HSP70-5 in medaka cells at 48 h post inoculation because almost all of the cells were disrupted by a cytopathic effect at this period (data not shown). HSP90-β in GF-1 cells was up-regulated by inoculation with RGNNV (MOI = 0.1) at as early as 6 h post inoculation , which is well consistent with our data that HSP90-β was up-regulated in medaka cells at 6 h post RGNNV inoculation (Fig. 4). Collectively, these results imply that RGNNV actively induces those six HSP isoforms to facilitate its multiplication. The viral manipulation of host cells via alteration of the gene expression has been reported in various virus species [83–86]. Nevertheless, we still can not rule out the possibility that induction of some or all of the virologically important HSP isoforms were just results of host stress responses against virus infection and were incidentally utilized for virus multiplication.