The Naturally Attenuated Kunjin Strain of West Nile Virus Shows Enhanced Sensitivity to the Host Type I Interferon Response

Abstract

The host determinants that contribute to attenuation of the naturally occurring nonpathogenic strain of West Nile virus (WNV), the Kunjin strain (WNV_KUN), remain unknown. Here, we show that compared to a highly pathogenic North American strain, WNV_KUN exhibited an enhanced sensitivity to the antiviral effects of type I interferon. Our studies establish that the virulence of WNV_KUN can be restored in cells and mice deficient in specific interferon regulatory factors (IRFs) or the common type I interferon receptor. Thus, WNV_KUN is attenuated primarily through its enhanced restriction by type I interferon- and IRF-3-dependent mechanisms.

TEXT

West Nile Virus (WNV) is a positive-stranded RNA mosquito-borne virus in the Flaviviridae family and is responsible for sporadic outbreaks of encephalitis in humans worldwide (20). A highly pathogenic North American strain of WNV (WNV_NY99) has emerged and has caused over 30,000 diagnosed human cases, including more than 1,000 deaths over the past decade (http://www.cdc.gov/ncidod/dvbid/westnile/index.htm). In contrast to WNV_NY99, the indigenous Australian strain of WNV, the Kunjin strain (WNV_KUN), has limited pathogenic capacity, as only 18 nonfatal cases in humans have been reported since its discovery almost 50 years ago (10). Genomic sequence comparisons between WNV_NY99 and WNV_KUN reveal a close relationship, with 97.7% identity at the amino acid level and only 33 of 77 substitutions being nonconservative (references 14 and 23; M. Audsley and A. Khromykh, unpublished results). Although differences in structural protein N-linked glycosylation have been postulated to explain, in part, the lack of pathogenicity of WNV_KUN (3, 22), no study has definitively described the mechanism of natural attenuation.

The pathogenesis of WNV is greatly influenced by the type I interferon (IFN) host response (7, 12, 21). While both WNV_NY99 and WNV_KUN can antagonize this response in vitro (reviewed in reference 8), WNV_KUN is strongly attenuated in adult mice (2) and virulent only in very young mice, for which the host immune response to viruses has not fully matured. Although the less virulent phenotype of WNV_KUN could be attributed to several factors, including its slightly delayed replication kinetics in C6/36 insect cells or Vero cells (data not shown), we hypothesized that the major part of its attenuation could be explained by its distinct interaction with the type I IFN response.

Prior experiments showed that both WNV_NY99 and WNV_KUN could limit the type I IFN response by blocking the phosphorylation of STAT1 and STAT2 in cells (18). To define whether attenuation of WNV_KUN was linked to a reduced capacity to replicate after the induction of an IFN response, human A549 cells were treated with increasing amounts of human alpha IFN (IFN-α; 10⁰ to 10⁴ IU/ml) 6 h prior to or 3 h after infection with WNV_KUN or WNV_NY99 generated from infectious cDNA clones (13) at a multiplicity of infection (MOI) of 1. Viral production was measured 24 h after infection by a plaque assay using BHK21-15 cells. Pretreatment with 10⁰ to 10⁴ IU/ml of IFN-α decreased the yield of WNV_KUN 5- to 1,000-fold; in comparison, the same concentration of IFN-α reduced the infectivity of WNV_NY99 only 1- to 50-fold (P < 0.05) in these cells (Fig. 1A, left panel). Similarly, treatment with 10⁰ to 10⁴ IU/ml of IFN-α 3 h after infection reduced WNV_KUN titers 10- to 100-fold compared to 1- to 10-fold for WNV_NY99 (P < 0.05) (Fig. 1A, right panel). Thus, WNV_KUN exhibited enhanced sensitivity to IFN-α compared to WNV_NY99 regardless of whether it was added before or after infection.

Fig. 1.

WNV_KUN antagonizes IFN-α antiviral effects less efficiently than WNV_NY99. (A) Effects of pre- or posttreatment of IFN-α on WNV production. A549 cells were treated with the indicated doses of IFN-α 6 h before (left panel) or 3 h after (right panel) infection with WNV_KUN or WNV_NY99, and virus production was evaluated at 24 h by plaque assay. (B) WNV_KUN prevents IFN-α-dependent nuclear translocation of STAT1 less efficiently than WNV_NY99. A549 cells were infected with WNV_KUN or WNV_NY99 for 48 h, treated with IFN-α for 30 min, fixed, permeabilized, and stained with anti-WNV E and -STAT1 antibodies. DAPI, 4′,6′-diamidino-2-phenylindole. (C) WNV_KUN prevents IFN-α-dependent phosphorylation of STAT1 less efficiently than WNV_NY99. A549 cells were infected with WNV_KUN or WNV_NY99 for 48 h, treated with IFN-α for 30 min, and lysed. Levels of phosphorylated STAT1 (P-STAT1), STAT1, WNV E, and actin were examined by immunoblot analysis.

Since the inhibition of IFN signaling by WNV_NY99 and WNV_KUN is mediated, in part, through the blockade of STAT1 and STAT2 activation, we hypothesized that the enhanced sensitivity to IFN-α may be due to a decreased capacity of WNV_KUN to antagonize the IFN response by inhibiting these proteins. To evaluate this, A549 cells were infected with WNV_KUN or WNV_NY99 for 48 h and treated with 5,000 IU/ml of IFN-α for 30 min, and the levels of activated nuclear and phosphorylated STAT1 were assessed by an immunofluorescence assay (IFA) or by immunoblotting. In WNV_NY99-infected cells, nuclear localization of STAT1, in response to IFN-α treatment, was not detected by IFA, confirming that WNV_NY99 efficiently antagonized IFN-α signaling. In contrast, in WNV_KUN-infected cells, the percentage of infected cells with nuclear STAT1 was appreciably higher (Fig. 1B). Correspondingly, the level of phosphorylated STAT1 in WNV_KUN-infected cells after IFN-α treatment was greater than that observed for WNV_NY99-infected cells (Fig. 1C). Thus, in A549 cells, WNV_KUN has an increased sensitivity to the antiviral effects of IFN-α compared to that of WNV_NY99, due in part to a reduced relative capacity to block STAT1 phosphorylation and nuclear translocation.

Since WNV_KUN attenuation correlated with an enhanced sensitivity to IFN-α in vitro, we questioned whether the absence of host factors involved in IFN induction or signaling would restore WNV_KUN pathogenesis in vivo. To evaluate this, 8- to 10-week-old wild-type, interferon regulatory factor 3 knockout (IRF-3^−/−), IRF-7^−/−, IRF-3^−/− × IRF-7^−/− double-knockout, and IFN-α/β receptor knockout (IFN-αβR^−/−) C57BL/6 mice were infected subcutaneously with 10³ PFU of Vero cell-derived WNV_KUN or WNV_NY99 and monitored for survival. Consistent with previous studies, wild-type mice infected with WNV_NY99 exhibited an ∼60% survival rate (Fig. 2A) (5). In contrast, infection with WNV_KUN resulted in no appreciable morbidity or mortality (Fig. 2B). Viral burden analysis of spleen and serum samples from wild-type mice at several time points after infection confirmed the attenuated phenotype of WNV_KUN compared to that of WNV_NY99 (Fig. 3A and E). Analogous to previous studies (7, 21), IFN-αβR^−/− or IRF-3^−/− × IRF-7^−/− mice infected with WNV_NY99 showed a 100% mortality rate with accelerated kinetics (mean time to death [MTD] of 3.4 ± 0.5 days [n = 9 mice] and 5.8 ± 0.4 days ([n = 10], respectively). Remarkably, infection of IFN-αβR^−/− or IRF-3^−/− × IRF-7^−/− mice with WNV_KUN resulted in a 100% mortality rate with only a slight delay in kinetics (MTD of 5.0 ± 0.4 days [n = 12] and 6.7 ± 0.5 days [n = 11], respectively) (Fig. 2B). Consistent with this result, viral titers from IRF-3^−/− × IRF-7^−/− mice infected with WNV_KUN were markedly greater in all organs tested than those for wild-type mice, although in the brain, they did not reach those observed with WNV_NY99 (Fig. 3D, H, and L).

Fig. 2.

Survival of mice infected with WNV_KUN or WNV_NY99. Eight- to 10-week-old wild-type and congenic deficient C57BL/6 mice were inoculated with 10³ PFU of WNV_KUN or WNV_NY99 by footpad injection and monitored for mortality for 21 days. (A) Survival data with IRF-3^−/−, IRF-7^−/−, IRF-3^−/− × IRF-7^−/−, or IFN-αβR^−/− mice infected with WNV_NY99 were statistically different (P < 0.05) compared to those for the wild-type controls. (B) Survival data with IRF-7^−/−, IRF-3^−/− × IRF-7^−/−, and IFN-αβR^−/− mice infected with WNV_KUN were statistically different (P < 0.05) compared to those for the wild-type controls. Kaplan-Meier survival curves were analyzed by the log- rank test.

Fig. 3.

Viral burden in peripheral and central nervous system (CNS) tissues from mice infected with WNV_KUN or WNV_NY99. WNV RNA in serum (A to D) and infectious WNV virus in the spleen (E to H) and brain (I to L) were determined from samples harvested on the indicated days using quantitative reverse transcription-PCR (qRT-PCR; serum) or viral plaque assay (spleen and brain). Data are shown as viral RNA equivalents or PFU per gram of tissue for 5 to 10 mice per time point. The error bars indicate standard error of the mean, and the dotted line represents the limit of sensitivity of the assay. Asterisks (*, P < 0.05; **, P < 0.01, ***, P < 0.001) represent differences that are statistically significant by the Mann-Whitney test.

In comparison, infection of IRF-7^−/− mice, which have a blunted systemic IFN-α response (6, 11), showed partial restoration of WNV_KUN virulence, with an ∼47% mortality rate compared to 100% mortality after infection with WNV_NY99 and delayed kinetics of death (MTD of dying animals, 11.0 + 1.3 days [n = 15]) (Fig. 2B). Replication of WNV_KUN was enhanced in tissues from IRF-7^−/− mice but also did not attain the levels in the brain observed for WNV_NY99-infected IRF-7^−/− mice. For example, although levels of WNV_KUN were similar to those of WNV_NY99 in the serum and spleen samples (Fig. 3C and G), the amount in the brain was ∼25- to 60-fold lower on days 4 and 8 after infection (Fig. 3K). This difference is likely because cells from IRF-7^−/− mice still have an intact IFN-β response (6), which could limit WNV_KUN virulence in a tissue-specific manner. A deficiency of IRF-3, which by itself does not affect WNV-induced production of type I IFN in some cells or in vivo (4, 5), did not restore WNV_KUN virulence, as 93% (n = 15) of IRF-3^−/− mice still survived infection. This compares with a 0% survival rate (n = 5) after infection with WNV_NY99, as observed previously (5). Somewhat surprisingly, viral titers for the serum and spleen samples from IRF-3^−/− mice infected with WNV_KUN and WNV_NY99 were comparable (Fig. 3B and F), although levels in the brain at day 10 were 400-fold lower in IRF-3^−/− animals infected with WNV_KUN (Fig. 3J), which likely explains the mortality results. Thus, IRF-3 had a differential and tissue-specific effect on the control of WNV_NY99 and WNV_KUN in the brain.

Since the attenuation of WNV_KUN in vivo correlated with the integrity of key regulatory components of the type I IFN response, we evaluated whether WNV_KUN and WNV_NY99 replicated differently in primary cells, which produce and respond to type I IFN after WNV infection (6), and compared this to growth curves for congenic IRF-3^−/−, IRF-7^−/−, IRF-3^−/− × IRF-7^−/− or IFN-αβR^−/− cells. In wild-type murine embryonic fibroblasts (MEF), multistep viral growth curves confirmed that WNV_KUN replication was attenuated compared to that for WNV_NY99 (∼10- to 80-fold decrease at 24 to 72 h postinfection; P < 0.05) (Fig. 4A). Similarly, WNV_KUN was attenuated (15- to 80-fold at 24 to 72 h postinfection; P < 0.05) in IRF-3^−/− or IRF-7^−/− MEF (Fig. 4B and C), which retain the capacity to produce IFN-α or -β after WNV infection (5, 6). WNV_KUN still replicated less efficiently than WNV_NY99 in IRF-3^−/− × IRF-7^−/− or IFN-αβR^−/− MEF, which are either virtually or completely deficient in type I IFN signaling (7) (Fig. 4D and E). Thus, in MEF, a loss of the ability to produce or respond to type I IFN augments but does not restore the infectivity of WNV_KUN to the levels observed with WNV_NY99. This phenotype could be explained by differential expression of antiviral IFN-stimulated genes (ISGs) in each of the deficient cells, some of which are induced differentially by IFN-dependent or IRF-3-dependent pathways (9). When we examined infection at 24 h in a second primary cell type, bone marrow-derived macrophages, we observed a similar phenotype: a complete loss of type I IFN signaling enhanced WNV_KUN infectivity but not to the levels (4-fold lower; P < 0.01) observed with WNV_NY99 (Fig. 4F). One apparent difference in macrophages was the relatively poor replication of WNV_KUN in IRF-3^−/− × IRF-7^−/− macrophages, which may be due to residual production of IFN-β in these cells after WNV infection through an IRF-3- and IRF-7-independent pathway (7).

Fig. 4.

Replication of WNV_KUN and WNV_NY99 in wild type, IRF-3^−/−, IRF-7^−/−, IRF-3^−/− × IRF-7^−/−, and IFN-αβR^−/− cells. MEF or macrophages were generated as previously described (7) and infected with WNV_KUN or WNV_NY99 (MOI of 0.01), and virus production was evaluated by plaque assay at the indicated times for MEF or 24 h after infection for macrophages. Values are averages from triplicate samples generated for three independent experiments. Error bars indicate the standard error of the mean, dashed lines denote the limit of sensitivity, and asterisks (*, P < 0.05; **, P < 0.01; ***, P < 0.001) represent differences that are statistically significant by an unpaired t test.

The experiments here establish that mice unable to produce or respond to type I IFN were vulnerable to infection with WNV_KUN and suggest that the relative lack of pathogenicity of this natural virus variant in wild-type mice or, possibly, humans can be attributed, in part, to its rapid control by the host type I IFN response. Data obtained ex vivo using genetically deficient primary fibroblasts showed that an absence of the type I IFN enhanced WNV_KUN viral replication but not to the level of WNV_NY99; this correlated with a viral burden analysis of mice, which showed increased WNV_KUN infectivity in genetically deficient animals that approached but was not equivalent to that for WNV_NY99. One explanation of this discrepancy could be that subsets of cells and tissues differentially express ISGs through IFN-dependent and -independent (yet IRF-3-dependent) pathways (5). Although the genetic basis of the relative attenuation of WNV_KUN remains a topic of further study, the phenotype is at least partially due to a reduced ability to antagonize IFN signaling and/or the antiviral activity of specific ISGs. Amino acid substitutions between both viruses are dispersed throughout the genome, with differences in the viral proteins NS1, NS2A, NS2B, NS4A, NS4B, and E of WNV_NY99 and WNV_KUN identified as possible antagonists in IFN production or signaling (1, 15–19, 24). Recently, WNV_NY99 but not WNV_KUN NS5 has been identified as a potent inhibitor of IFN-dependent JAK-STAT signaling; remarkably, a single-amino-acid substitution in WNV_KUN NS5 restored the inhibitory activity to that observed for WNV_NY99 NS5 (15). Nonetheless, the exact contribution of each viral protein in inhibiting the host response in vivo and the identification of the specific molecular determinants that explain differential pathogenesis warrant further studies.