Negative regulation of type I interferon signaling by integrin-linked kinase permits dengue virus replication

Dengue virus (DENV) infection can induce life-threatening dengue hemorrhagic fever/dengue shock syndrome in infected patients. DENV is a threat to global health due to its growing numbers and incidence of infection in the last 50 years. During infection, DENV expresses ten structural and nonstructural proteins modulating cell responses to benefit viral replication. However, the lack of knowledge regarding the cellular proteins and their functions in enhancing DENV pathogenesis impedes the development of antiviral drugs and therapies against fatal DENV infection. Here, we identified that integrin-linked kinase (ILK) is a novel enhancing factor for DENV infection by suppressing type I interferon (IFN) responses. Mechanistically, ILK binds DENV NS1 and NS3, activates Akt and Erk, and induces NF-κB-driven suppressor of cytokine signaling 3 (SOCS3) expression. Elevated SOCS3 in DENV-infected cells inhibits phosphorylation of STAT1/2 and expression of interferon-stimulated genes (ISGs). Inhibiting ILK, Akt, or Erk activation abrogates SOCS3 expression. In DENV-infected mice, the treatment of an ILK inhibitor significantly reduces viral loads in the brains, disease severity, and mortality rate. Collectively, our results show that ILK is a potential therapeutic target against DENV infection.


Introduction
Dengue virus (DENV) in the Flaviviridae family is the most common cause of mosquitoborne diseases transmitted by Aedes aegypti and Aedes albopictus mosquitoes. DENV infection is often asymptomatic or induces a wide range of self-limited symptoms, including fever, skin rashes, and pains behind the eyes. Occasionally, patients can develop fatal dengue hemorrhagic fever/dengue shock syndrome (DHF/DSS). It is believed that the overt production in patients of cytokines and chemokines, such as interleukin (IL)-1β, IL-4, IL-6, IL-8, tumor necrosis factor (TNF)-α, CXCL10, and macrophage migration inhibitory factor (MIF), leads to DHF/DSS [1][2][3][4]. According to a report of the World Health Organization, the incidence of dengue has increased by 30-fold over the last 50 years. Currently, it is estimated that there are 390 million dengue cases per year [5]. More than 100 countries worldwide include endemic regions of dengue, including Taiwan. The recent outbreaks in Taiwan from 2014 to 2015 caused 60,000 cases of dengue fever and 250 deaths [6]. Despite the fast-emerging and wide-spread nature of dengue infection, there are as yet no antiviral drugs or therapies available to prevent or ameliorate DENV-induced diseases.
There are four serotypes of DENV in which DENV2 infection is more likely to induce severe diseases [7]. Mononuclear phagocytes are the primary target cells for DENV infection in humans [8], and several cell lines support DENV infection in vitro, such as A549, Huh-7, and U937 cells. During infection, DENV binds to cellular receptors, including β1-and β3-integrin [9,10] and DC-SIGN [11], to facilitate virus binding and entry. After entry, its positive, single-stranded RNA is translated into a polypeptide and processed by cellular and viral proteases to yield three structural (E, envelope; prM, precursor membrane; C, capsid) and seven nonstructural (NS) proteins (NS1, NS2A, NS2B, NS3, NS4A, NS4B, and NS5). DENV infection activates Akt by phosphorylating both Thr308 and Ser473 to enhance virus replication [12][13][14]. Furthermore, the NS proteins interact with cellular proteins to help DENV infection by altering ER and autophagic membrane to form replication complexes or by suppressing antiviral responses [15][16][17]. However, most of the cellular proteins bound to NS proteins and their roles in DENV pathogenesis remain elusive.
DENV infection elicits innate immune recognition by Toll-like receptor (TLR) 3, 7, and 8 in endosomes and retinoic acid-inducible gene I (RIG-I) and melanoma differentiation-associated gene 5  in the cytoplasm [18][19][20]. Recognition of DENV by these receptors activates nuclear factor κ B (NF-κB) and interferon (IFN) regulator factors 3 or 7 to produce proinflammatory cytokines and type I (α and β) and III (λ) IFN, respectively. Secreted IFN binds to receptors and induces phosphorylation of Janus kinase (JAK) and signal transducer and activator of transcription (STAT) 1 and 2. The phosphorylated STAT1/2 translocates into the nucleus and induces IFN-stimulated gene (ISG) expression to restrict virus infection. The suppressor of cytokine signaling (SOCS) proteins, especially SOCS1 and SOCS3, are endogenous negative regulators that suppress JAK-mediated STAT1/2 phosphorylation [21]. DENV is shown to increase SOCS1 and SOCS3 expression in infected epithelial cells [22] and mononuclear phagocytes [23] to suppress antiviral responses. In patients with DHF, the level of SOCS3 in the blood is significantly elevated and is suggested to be a predictive biomarker [24]. However, few studies focus on how DENV increases SOCS expression.
Integrin-linked kinase (ILK) is a 59-kDa serine/threonine protein kinase binding on the cytoplasmic domain of β1 integrin [25]. ILK comprises three conserved domains, including Nterminal ankyrin (ANK) repeats, a central pleckstrin homology (PH)-like domain, and a C-terminal kinase domain [26]. ILK functions as an adaptor protein to link integrins with cytoskeletons. As a kinase, ILK is activated by phosphoinositide 3-kinase (PI3K) to phosphorylate Akt on Ser473, GSK3β on Ser9, or NF-κB subunit p65 on Ser539 [27][28][29]. The phosphorylated p65 is re-localized into the nucleus to initiate NF-κB-regulated gene expression. Accordingly, ILK plays a vital role in regulating various cellular processes, even virus infection. Currently, although it has been revealed that ILK participates in RNA and DNA virus infection [30][31][32], its role in DENV infection has not yet been addressed.
In this study we have identified that ILK enhances DENV infection. ILK binds DENV NS1 and NS3 to enhance SOCS3 expression via the Akt-Erk-NF-κB pathway. The enhanced SOCS3 expression results in decreased STAT1/2-driven ISG expression and increased DENV replication. Blocking ILK by the inhibitor, OSU-T315, significantly improves the survival rate of DENV-infected mice, with reduced viral loads in the mouse brains and reduced disease severity.

ILK knockdown reduces DENV infection
Previous studies showed that DENV infection activates Akt, and Akt inhibitors reduce DENV infection [12][13][14]. We determined Akt phosphorylation (p-) on Ser473 (p-Akt) in DENV serotype 2 (DENV2)-infected A549 cells and found an increase in phosphorylated Akt (S1A Fig). While it is known that ILK phosphorylates Akt on Ser473 [27,28], the roles of ILK in DENV infection remain elusive. To address this issue, we established A549 cells with control shRNA (shLuc) or shRNA targeting ILK 3'UTR (shILK). The ILK-specific shRNA abolished corresponding protein expression (Fig 1A) without affecting cell viability (S1B Fig). In infected control cells, the p-Akt levels were increased from 3 to 24 hours post-infection (hpi), and viral proteins NS1 and NS4B were detected at 24 hpi (Fig 1A). ILK knockdown attenuated p-Akt levels about 1.5-fold in mock-infected cells and infected cells from 3 to 24 hpi as well as NS1 and NS4B in infected cells at 24 hpi. OSU-T315 is an ILK inhibitor suppressing ILK-mediated Akt phosphorylation on Ser473 [33]. Treatment with 1.5 to 3 μM OSU-T315 in A549 cells diminished p-Akt level without affecting cell viability (S1C and S1D Fig, respectively). In infected A549 cells, OSU-T315 decreased the expression of viral proteins, including NS1, NS3, NS4B, and prM, in a dose-dependent manner (Fig 1B). These results show that knockdown or inhibition of ILK reduced Akt phosphorylation and DENV2 infection.
To determine if DENV2 yields were reduced in ILK knockdown cells, we detected doublestranded RNA (dsRNA), an intermediate generated during DENV genome duplication, in infected cells using immunofluorescence staining. ILK knockdown not only diminished the dsRNA intensity in cells but also reduced the percentage of dsRNA-positive cells by 50% ( Fig  1C). The mean viral yield was lower in ILK knockdown cells than in control cells by 55-fold. Similar results were observed in cells infected with other serotypes of DENV (Fig 1D), showing that ILK knockdown reduces DENV infection independent of virus serotypes.

ILK knockdown reduces DENV infection after virus binding and entry
To identify DENV infection steps controlled by ILK, we first assessed the levels of DENV binding to control and ILK knockdown cells. Briefly, A549 cells were infected with DENV at a multiplicity of infection (MOI) of 25 for 2 hours at 4˚C, a temperature that allows virus binding but not entry, washed with PBS to remove the unbound virus, stained with an antibody against viral envelope (E) protein, and analyzed by flow cytometry as previously described [12].

PLOS PATHOGENS
Integrin-linked kinase antagonizes type I interferon signaling permits dengue virus replication Next, we assessed the effect of ILK in DENV infection using a previously established Huh7 cell line carrying DENV replicon expressing firefly luciferase [34], in which virus entry is bypassed. The Huh7 replicon cells were transduced with control shRNA (shLacZ) or shRNA targeting ILK 3'UTR (ILK-1) or CDS (ILK-2). In Huh7 replicon cells with knocked-down ILK, the luciferase activity (Fig 2A) and NS1 and NS4B expressed by DENV replicon (Fig 2B) were

PLOS PATHOGENS
Integrin-linked kinase antagonizes type I interferon signaling permits dengue virus replication dramatically decreased when compared with those in control cells. Treatment with OSU-T315 inhibited luciferase activity with significant differences found at 1.25 and 2.5 μM (Fig 2C) without affecting cell viability (Fig 2D). Consistently, OSU-T315 with a concentration of 1.25 and 2.5 μM attenuated the expression of viral proteins in DENV-infected Huh7 cells (Fig 2E). Accordingly, these findings show that ILK knockdown decreases DENV replication after virus entry.
DENV has been known to induce autophagy to benefit viral replication [35][36][37]. The PI3K inhibitors decrease autophagosome formation and DENV replication [35,38]. Since ILK is activated by PI3K, we determined whether ILK facilitates autophagosome formation in DENV-infected cells by detecting LC3 I/II conversion. After 3 hpi, DENV infection increased the conversion of LC3 I to II in control cells. However, there was no difference in the LC3 I/II conversion between infected control and ILK knockdown cells (S3A Fig), indicating that ILK facilitates DENV replication in a manner independent of autophagosome formation. Additionally, activation of Akt has been reported to facilitate DENV replication by blocking apoptosis [13]. Thus, we determined the levels of cleaved caspase-3, a marker of apoptotic cell death, in DENV-infected control or ILK knockdown cells. In control cells, the cleaved caspase-3 was detected concomitantly with NS4B expression at 24 hpi. Nonetheless, there was no increase in the cleaved caspase-3 levels in ILK knockdown cells after infection (S3B

ILK regulates SOCS3 expression to attenuate IFN signaling
Type I and III IFNs stimulate the expression of ISGs and suppress virus replication. Therefore, we assayed if ILK suppresses the induction of type I (α and β) and III (λ) IFNs and ISGs (MxA, ISG54, and ISG56) in DENV-infected cells. At 24 hpi, the levels of IFN-α, -β, and -λ in control cells were significantly increased and were not different from those in ILK knockdown cells (Figs 3A and S4A and S4B). However, the mRNA level of MxA was significantly increased in ILK knockdown cells before and, especially, after 24 hours of DENV infection by 2-and 60-fold, respectively (Fig 3B). Similar results were observed in the mRNA levels of ISG54 and ISG56 (S4C and S4D Fig), showing that ILK knockdown enhances ISG rather than type I/III IFN expression. We further assayed the levels of phosphorylated (p-) STAT1 and STAT2 in control and ILK knockdown cells. Upon IFN-β stimulation, p-STAT1 and p-STAT2 were increased in control and ILK knockdown cells, with higher levels found in ILK knockdown cells (S5 Fig). In DENV-infected cells, p-STAT1 and p-STAT2 were detected at 24 hpi, with levels further enhanced by ILK knockdown (Fig 3C). Treatment of OSU-T315 enhanced STAT1 phosphorylation and MxA expression and reduced NS1 expression in infected cells (Fig 3D). These results show that ILK attenuates STAT1/2-activated ISG expression.
Previous studies showed that SOCS1 and SOCS3 suppress IFN signaling in DENV-infected cells [22,23]. We found that DENV infection increased SOCS3 mRNA and protein levels, while SOCS1 expression was unaffected (Figs S4E and 4A, respectively). SOCS3 knockdown via shRNA abolished SOCS3 mRNA expression (Fig 4B upper panel), dramatically increased ISG54 and ISG56 mRNA levels and STAT1 phosphorylation in cells without DENV infection (Fig 4B lower panel and Fig 4C, respectively), and reduced NS1 and NS4B expression and viral yields in infected cells (Fig 4C and 4D). These results show that DENV enhances SOCS3 expression to attenuate IFN signaling. Infection of the UV-inactivated DENV, a virus able to bind and enter cells but unable to replicate, failed to increase SOCS3 expression in cells (S6 Fig), indicating the essential role of DENV replication in SOCS3 induction.
We further assessed if ILK regulates SOCS1 and SOCS3 expression. In mock-infected cells, ILK knockdown abrogated SOCS3 but not SOCS1 expression (Fig 4A), indicating that ILK is required for SOCS3 expression. Notably, ILK knockdown and OSU-T315 treatment diminished DENV-induced SOCS3 expression (Figs 4A and S4E and 4E). Similar results were observed in U937 cells, the primary target for DENV infection. Knocking down ILK in U937 cells reduced SOCS3 and viral protein expression and viral yields and enhanced STAT1 phosphorylation at 24 hpi (S7 Fig). The results suggest that DENV hijacks ILK-regulated SOCS3 expression to attenuate type I IFN signaling.

ILK regulates SOCS3 expression via Akt-Erk-NF-κB pathway
Influenza A virus infection-induced SOCS3 expression is dependent on NF-κB activation [39]. Interestingly, blocking ILK reduces NF-κB activity in cells by attenuating phosphorylation of subunit p65 on Ser536 [29]. Accordingly, we aimed to determine if ILK regulates SOCS3 expression via NF-κB. Our data showed that DENV infection increased the phosphorylation of p65 on Ser536 (p-p65) in control cells at 3 to 24 hpi, while ILK knockdown abrogated p65 phosphorylation in mock-and DENV-infected cells (Fig 5A). Additionally, treatment of NF-

PLOS PATHOGENS
Integrin-linked kinase antagonizes type I interferon signaling permits dengue virus replication Treatment of the Akt1/2 kinase inhibitor reduced DENV-induced p-Erk as well as Erk-NF-κB-driven SOCS3 expression and NS4B expression (Fig 5E). Furthermore, DENV yields were reduced in cells treated with inhibitors of NF-κB, Erk, and Akt (Fig 5F). This shows that ILK activates the Akt-Erk-NF-κB pathway to increase SOCS3 expression and facilitate DENV infection.

PLOS PATHOGENS
Integrin-linked kinase antagonizes type I interferon signaling permits dengue virus replication β-triggered p-STAT1, ISG56, and MxA (Fig 6B and 6C). These results indicate that DENV NS1 and NS3 are involved in ILK-mediated IFN signaling inhibition. To understand how DENV NS1 and NS3 affect ILK, we monitored their cellular distribution. The co-localization of ectopic expressed DENV NS1 or NS3, but not NS5, with ILK was clearly observed (S9A Fig). The interaction between ectopic expressed DENV NS1 or NS3 with ILK was confirmed by co-immunoprecipitation analysis (Fig 6D). In line with these observations, we also found a similar co-localization of DENV NS1 or NS3, but not NS5, with endogenous ILK in DENVinfected A549 cells (S9B Fig). The interaction between DENV NS1 or NS3 with endogenous ILK was also further confirmed in DENV-infected A549 cells by co-immunoprecipitation analysis (Fig 6E) as well as proximity ligation assay, a powerful tool with high specificity and sensitivity to detect endogenous protein interaction (S9C Fig). ILK contains three domains, including N-terminal ankyrin (ANK) repeats, plekstrin homology (PH) domain and a C-terminal kinase domain, in which ANK and kinase domains are able to bind several cellular proteins such as ILK-associated phosphatase and Akt, respectively, to regulate ILK activity [44]. To understand whether ANK or kinase domain is responsible for interacting with DENV NS1 and NS3, we introduced two ILK constructs (ILK  or ILK 171-452 ) into NS1 or NS3 expressing A549 cells. By co-immunoprecipitation analysis, we found that only ILK 171-452 was able to interact with DENV NS1 and NS3, indicating that ILK kinase domain but not ANK domain contributes to this interaction (Fig 6F). Furthermore, we also observed the increased Akt recruitment to ILK complex in the presence of DENV NS1 and NS3 (Fig 6G). Collectively, these results reveal that DENV NS1 and NS3 interact with ILK to attenuate type 1 IFN signaling.

Inhibition of ILK reduces viral loads and the mortality rate of DENV2-infected mice
To test whether inhibition of ILK protects hosts against DENV infection, the therapeutic effect of OSU-T315 in DENV-infected mice was examined. 7-day-old C57BL/6 mice were infected with 6 × 10 5 PFU of DENV2 via intraperitoneal and intracranial injections. Infected mice were treated with or without 12.5 mg/kg OSU-T315 on the day of infection via intraperitoneal injection. In the control-treated mice, all infected mice displayed neurological symptoms, including hunched back, seizure, and paralysis in both fore-and hind-limbs after 6 days of infection ( Fig  7A) and succumbed to death (0/8) within 10 days post-infection (dpi) (Fig 7B). With OSU-T315 treatment, infected mice exhibited a delayed onset of neurological symptoms, and half of them exhibited little or no neurological symptoms (Fig 7A). The survival rate of OSU-T315-treated mice was 50% (4/8) by 21 dpi, which was significantly higher than that in control-treated mice (0/8) (Fig 7B), showing that inhibition of ILK protects mice from lethal DENV infection. We further determined the viral loads and NS3 expression in the brains of infected mice on 7 dpi, when virus load reaches a peak. In the brains of OSU-T315-treated mice, the mean viral load (Fig 7C) and the NS3 expression (Fig 7D) were both reduced compared with those in control-treated mice. Additionally, the levels of TNF-α, as well as IL-6, were significantly lower in the brains of OSU-T315-treated mice than those of control-treated mice on 7 dpi, indicating OSU-T315 attenuates DENV-induced inflammation in mouse brains (Fig  7E). Taken together, our findings revealed that inhibition of ILK reduces not only viral loads but also inflammation in infected brains and improves the survival of DENV-infected mice.

Discussion
In the present study, we revealed that ILK is an enhancing factor for DENV replication (Fig  7F). ILK increases the expression of SOCS3 via the Akt-Erk-NF-κB pathway by interacting with DENV NS1 and NS3. Elevated SOCS3 expression antagonizes STAT1/2 activation and

PLOS PATHOGENS
Integrin-linked kinase antagonizes type I interferon signaling permits dengue virus replication

PLOS PATHOGENS
Integrin-linked kinase antagonizes type I interferon signaling permits dengue virus replication inhibits ISG, but not type I and III IFN production. Knocking down ILK or SOCS3 enhances ISG expression and reduces viral yields in cells. Moreover, treatment of the ILK inhibitor, OSU-T315 decreases brain viral loads, disease severity, and mortality of DENV-infected mice.
The integrin β1 and β3 are shown to promote DENV binding and entry into cells [10,45]. Although ILK is bound to and activated by integrin β1 and β3, the level of DENV binding to cells is not affected by ILK. Moreover, in Huh7 cells with a luciferase-expressing DENV replicon, ILK knockdown attenuated luciferase activity and viral protein expression. These results indicate an enhancing role of ILK in DENV replication after virus entry. During replication, DENV activates PI3K to facilitate autophagy and activates Akt to prevent apoptosis of infected cells [13]. Nonetheless, there was no difference in the levels of LC3 I/II conversion and cleaved caspase-3 between wild-type and ILK knockdown cells after DENV infection. These results show that ILK-mediated enhancement in DENV infection is independent of autophagy or apoptosis.
DENV has been shown to disrupt IFN and ISG induction via increasing SOCS1 and SOCS3 expression [22,23], but the detailed mechanism is still lacking. In this study, we revealed that ILK interacts with DENV NS1 and NS3 to increase SOCS3 expression and subvert antiviral IFN signaling. This SOCS3 induction is mediated by ILK-Akt-Erk-NF-κB signaling. Pharmaceutical or genetic inhibition of ILK decreases SOCS3 expression in both mock-and DENVinfected cells. Notably, knocking down ATP6V0C, an irrelevant DENV host-dependency factor [46], reduced DENV-induced p-p65 as well as NS1 and NS3 protein levels and SOCS3 protein levels, while ILK levels remained unchanged (S10 Fig). It might partly result from decreased NS1/NS3 and ILK interaction, as we suggested, or other DENV-activated signalings that have not been identified, such as pleiotropic effects resulting from lower replication of DENV. Further studies are needed to address the issue. We found that SOCS3 mRNA levels were only 1.4-to 1.3-fold higher in control cells than in ILK knockdown cells after DENV infection (S4E Fig). However, the protein levels of SOCS3 were dramatically reduced in ILK knockdown cells with or without DENV infection compared to control cells (Fig 4A). A recent study revealed that ILK is able to increase oncogenic protein synthesis by positively regulating translation via AKT/mTOR/eIF4E pathways [47]. According to these observations, we suggest that ILK might upregulate both SOCS3 RNA expression and translation as well to support DENV infection. As one of ILK kinase substrate, Akt is able to be phosphorylated at Ser 473 by ILK for its activation. Accordingly, Akt has been found to bind to ILK via Rictor to be activated [48]. Our results showed that DENV NS1 and NS3 bind to the kinase domain of ILK, and also showed an increase of Akt recruitment to this complex. The involvement of Rictor in forming this DENV NS-ILK complex is worthy of further investigation. Notably, several studies have pointed out that dengue viral non-structure proteins may antagonize IFN-signaling. For example, DENV NS2A, NS4A, and NS4B proteins inhibit STAT1 phosphorylation [15,49], and NS1 and NS2B3 activate NF-κB-dependent transcription [41,50]. DENV NS5 co-opts ubiquitin protein ligase E3 component n-recognin 4 (UBR4) to degrade STAT2 and antagonize type 1 IFN signaling [51]. Our findings here explored the new roles of DENV NS1 and NS3 in triggering SOCS3-mediated IFN signaling inhibition via ILK. Interestingly, previous studies revealed that overexpression of DENV NS1 and NS3 is able to induce IL-6 and IL-10 production, respectively [52,53]. Both IL-6 and IL-10 transduce a STAT3-dependent signal pathway,

PLOS PATHOGENS
Integrin-linked kinase antagonizes type I interferon signaling permits dengue virus replication

PLOS PATHOGENS
Integrin-linked kinase antagonizes type I interferon signaling permits dengue virus replication which has been shown to negatively regulate type-I IFN responses [54]. Thus, DENV NS1 and NS3 might also antagonize IFN signaling by activation of STAT3-mediated actions. Overall, these observations underscore the conclusion that DENV develops multiple strategies to subvert the cellular antiviral response.
Besides DENV, ILK has also been shown to support the infection of coxsackievirus B3 and the hepatitis C virus [30,31]. Especially in hepatitis C virus-infected cells, ILK activation also attenuates ISG expression. Furthermore, infection of hepatitis C virus [55], influenza A virus [39,56], respiratory syncytial virus [57], human immunodeficiency virus [58], and herpes viruses [59,60], induce SOCS3 expression to antagonize STAT1/2 activation, implying SOCS3 can be a therapeutic target for the development of broad-spectrum antiviral drugs. Moreover, it is essential to determine if the ILK-Akt-Erk-NF-κB pathway we identified in DENV-infected cells is responsible for the induction of SOCS3 in cells infected with other viruses.
To study if ILK inhibitor, OSU-T315, protects the host from DENV infection, we chose the mouse model in which neonatal (7-day-old), immunocompetent mice were infected with DENV2 via intracranial and intraperitoneal injections [61,62]. DENV infection in immunocompetent mice fails to induce overt signs of diseases. Although the immunocompromised mice, such as mice with deficiency in type I and II IFN responses, support robust DENV replication, and display severe signs of disease [63][64][65], they are unsuitable for our study due to the lack of intact IFN responses. In contrast, in the immunocompetent neonatal mice, DENV replicates in the brains, induces neurological symptoms, such as seizure and paralysis, and results in death after infection, which makes it a feasible model for us to test the efficacy of blocking ILK as a therapeutic strategy. Furthermore, the central nervous system involvement is now considered as a criterion for severe dengue according to the World Health Organization classification [66], despite it having been thought an uncommon complication in the past two decades. Using this model, we demonstrated that inhibition of ILK by OSU-T315 improves the survival rate of DENV-infected mice with decreased viral loads in the brains, and reduced disease severity. These results suggest that ILK might serve as a promising therapeutic target for treating dengue.
Since the levels of TNF-α, IL-6, and SOCS3 in the blood are higher in dengue-infected patients with hemorrhagic fever or shock (severe dengue) than those in patients with fever only [24], it suggests that overt inflammatory responses are detrimental to DENV-infected hosts. The observation that treatment of anti-TNF serum protects mice from lethal DENV infection [67] supports this hypothesis. We found that the levels of TNF-α and IL-6 in infected mice were lower in OSU-T315-treated mice than in control mice, implying that inhibition of ILK reduces not only viral titers but also inflammatory cytokines. Others also demonstrate that inhibiting ILK abrogates NLRP3-mediated IL-1β production [68], that ILK knockdown reduces TNF-α production via NF-κB [69], and that ILK knockout in intestinal epithelial or myeloid cells reduces DSS-induced colitis in mice [70,71]. Further studies are needed to determine the cell types and mechanisms of ILK-induced inflammatory responses. Also, it would be essential to examine if ILK expression is elevated in DENV-infected patients with comorbidities who are susceptible to develop severe dengue. rate (B), viral loads in the mouse brains on day 7 post-infection (C), and levels of indicated proteins in the mouse sera day 7 post-infection (E) are shown. (D) The images of cortex regions stained with anti-NS3 (green) antibody from brain sections of mice mock-or DENV-infected and treated without (control) or with OSU-T315 on day 7 post-infection are shown. Nuclei were counterstained with DAPI. (F) The schematic diagram of mechanism mediated by ILK to promote DENV infection. This figure is created with Biorender.com. Data represent the mean ± SD (error bars) in panel A and the mean + SD (error bars) in panel E. Each symbol on the scattergram represents an individual sample. The black horizontal line represents the mean value for each group in panel C. � p < 0.01 and �� p < 0.001 of 5 samples per data points via two-way ANOVA with the appropriate post-hoc test for multiple comparisons in panel A and E, via log-rank test in panel B, and via Student's t test in panel C. https://doi.org/10.1371/journal.ppat.1011241.g007

PLOS PATHOGENS
Integrin-linked kinase antagonizes type I interferon signaling permits dengue virus replication In summary, we show that inhibition of ILK is a novel and effective therapeutic strategy against DENV replication in cells, and DENV lethality in mice, by reducing viral titers as well as inflammatory cytokines.

Ethics statement
All mouse experiment protocols were approved by the Institutional Animal Care and Use Committee of National Cheng Kung University (IACUC numbers 104062 and 105022). The C57BL/6 mice were maintained under a specific-pathogen-free condition in the Laboratory Animal Center of our college.

Infection and treatment of mice
The 7-day-old C57BL/6 mice were infected with a total of 6 × 10 5 plaque-forming unit (PFU) per mouse of DENV2 via intracerebral (1.5 × 10 5 PFU) and intraperitoneal (4.5 × 10 5 PFU) injections [61] and treated either with or without 12.5 mg/kg of OSU-T315 via intraperitoneal injection on the day of infection. Mice were monitored daily for survival rates and given disease scores as follows: 0, healthy; 1, hunched back; 2, seizure or wild rushing and jumping; 3, weakness or partial paralysis in limbs; 4, complete paralysis in hind and forelimbs; 5, moribund (score it as 5 rather than 4) or death. In separate experiments, the mouse brains were harvested on indicated days after infection to determine viral loads in the brain via plaque assay or for immunofluorescence staining.

PLOS PATHOGENS
Integrin-linked kinase antagonizes type I interferon signaling permits dengue virus replication

Immunofluorescence staining
DENV-infected A549 cells were fixed, permeabilized, and stained with primary antibody against double-stranded RNA (SCICONS, clone J2). The mouse brains were harvested and embedded in OCT compound. The longitudinal mouse brain sections (10 μm) were prepared by the sliding freezing microtome, fixed in 3.7% formaldehyde, incubated with cold-acetone, and stained with primary antibody against NS3. Alexa Fluro 488-conjugated secondary antibodies were used to visualize the signal in mouse cortex regions under a fluorescence microscope. Nuclei were stained with DAPI.

Cell viability assay
The culture supernatants of cells treated with or without the indicated concentrations of OSU-315 for 24 hours were collected to determine the activity of released LDH using a cytotoxicity detection kit (Roche) according to the manufacturer's instructions. The cell viability of cells treated with 0.05% Triton-X 100 was set as 0%.

RT-PCR and RT-quantitative PCR
Total RNA was prepared via a rNeasy RNA Mini Kit (Qiagen) according to the manufacturer's instructions. The cDNA was reversely transcribed from 1 μg total RNA by random primers and subjected to quantitative real-time PCR reactions with a LightCycler FastStart DNA Master PLUS SYBR Green I kit (Roche), and primers used were as follows: MxA forward ACCAC AGAGG CTCTCAGCAT and reverse CTCAGCTGGTCCTGGATCTC; β-actin forward AAGGAGAAGCTGTGCTACGTCGC and reverse AGACAGCACTGTGTTGGCGTACA. The relative expression of a specific gene was calculated as 2 -ΔΔCT . For RT-PCR, equal amounts of cDNA were subjected to PCR reactions with primers as follows: ISG54 forward ATGTGCA ACCTACTGGCCTAT and reverse TGAGAGTCGG CCCATGTGATA; ISG56 forward GGGCAGACTGGCAGAAGC and reverse TATAGCGGAAGGGATTTGAAAGC; and βactin forward AAGGAGAAGCTGTGCTACGTCGC and reverse AGACAGCACTGTGT TGGCGTACA. PCR products were separated in 2% agarose DNA gel.

ELISA
The culture supernatants of cells, either mock-infected or infected with DENV at 24 hpi, and the mouse serum, were collected and subjected to tests to determine the levels of IFN-β in the culture supernatants, and TNF-α and IL-6 in the mouse serum, using ELISA kits (R&D Systems) according to the manufacturer's instructions.

Statistics
Data are expressed as mean + or ± standard deviation (SD). For statistical comparison, the luciferase activity or cell viability was analyzed via one-way ANOVA followed by Dunnett's post hoc test for multiple comparisons, while the MxA mRNA levels, cytokine levels in the mouse serum, and disease scores were analyzed via two-way ANOVA followed by Tukey post hoc test for multiple comparisons. The viral titers in cells or mouse brains were analyzed using Student's t test, and the Kaplan-Miller survival curves were analyzed via log-rank test. Analyses were performed using GraphPad Prism 4 software (GraphPad Software). All p values are for two-tailed significance tests. A p value of < 0.05 was considered statistically significant. The phosphorylated (p-) p65 and indicated proteins in the mock-or DENV-infected A549 cells transfected scramble control siRNA or siRNA targeting ILK or ATP6V0C at indicated hpi are shown. The ratios of phosphorylated to total p65 relative to mock-infected cells transfected with scramble siRNA at 0 hpi are shown. (C) The indicated A549 cells were infected with DENV (MOI of 2) for 0 and 24 hours. The NS1, NS3, SOCS3, and GAPDH protein expression in the indicated A549 cells are shown. The ratios of SOCS3 to GAPDH relative to mock-infected scramble control cells and the ratio of NS1 and NS3 to GAPDH relative to DENV-infected scramble control cells are shown. (TIF)

PLOS PATHOGENS
Integrin-linked kinase antagonizes type I interferon signaling permits dengue virus replication