ABSTRACT
In recent years, Asian lineage Zika virus (ZIKV) strains emerged to cause pandemic outbreaks associated with a high rate of congenital ZIKV syndrome (CZVS). The reasons for the enhanced spread and severe disease caused by newly emerging strains are not fully understood. Here we compared viral sequences, viral replication, and innate immune signaling induction of three different ZIKV strains derived from African and Asian lineages and West Nile virus, another flavivirus. We found pronounced differences in activation of innate immune signaling and inhibition of viral replication across ZIKV strains. The newly emerged Asian ZIKV strain Brazil Fortaleza 2015, which is associated with a higher rate of neurodevelopmental disorders like microcephaly, induced much weaker and delayed innate immune signaling in infected cells. However, superinfection studies to assess control of innate immune signaling induced by Sendai virus argue against an active block of IRF3 activation by the Brazilian strain of ZIKV and rather suggest an evasion of detection by host cell pattern recognition receptors. Compared to the Asian strain FSS13025 isolated in Cambodia, both ZIKV Uganda MR766 and ZIKV Brazil Fortaleza appear less sensitive to the interferon-induced antiviral response. ZIKV infection studies of cells lacking the different RIG-I-like receptors identified RIG-I as the major cytosolic pattern recognition receptor for detection of ZIKV.
IMPORTANCE Zika Virus (ZIKV), discovered in 1947, is divided into African and Asian lineages. Pandemic outbreaks caused by currently emerging Asian lineage strains are accompanied by high rates of neurological disorders and exemplify the global health burden associated with this virus. Here we compared virological and innate immunological aspects of two ZIKV strains from the Asian lineage, an emerging Brazilian strain and a less-pathogenic Cambodian strain, and the prototypic African lineage ZIKV strain from Uganda. Compared to the replication of other ZIKV strains, the replication of ZIKV Brazil was less sensitive to the antiviral actions of interferon (IFN), while infection with this strain induced weaker and delayed innate immune responses in vitro. Our data suggest that ZIKV Brazil directs a passive strategy of innate immune evasion that is reminiscent of a stealth virus. Such strain-specific properties likely contribute to differential pathogenesis and should be taken into consideration when choosing virus strains for future molecular studies.
INTRODUCTION
Zika virus (ZIKV) is a positive-sense RNA virus belonging to the Flavivirus genus within the Flaviviridae family. It was first identified in Africa in 1947 (1). Two different lineages exist: an African lineage with the prototype strain MR766 isolated in Uganda and an Asian lineage which has caused increasing public health concern due to epidemic outbreaks in Micronesia (2007) and French Polynesia (2013) and which is now emerging within South and Central America (from 2014 on) (2, 3). ZIKV is transmitted mainly by Aedes sp. mosquitoes, but during recent outbreaks, sexual and maternal-to-fetal transmission have also been reported (4, 5). In adult humans, infection is usually asymptomatic or causes mild febrile illness (3). However, during recent outbreaks, an increase in neurological diseases has been observed. In particular, Asian lineage ZIKV has been associated with Guillain-Barré syndrome during the French Polynesian outbreak, and high case numbers of microcephaly have been reported for newborns during its spread throughout Brazil in 2015/2016 (3, 6, 7). Thus, African and Asian lineage ZIKV strains appear to differ in various aspects, with the newly evolved Asian/American lineage posing an increasing global health concern.
Like other members of the Flaviviridae family, ZIKV induces rearrangements of the endoplasmic reticulum to establish viral replication sites within the host cell (8). During flavivirus infection, viral RNA replication occurs via a negative-strand intermediate produced by the viral RNA-dependent RNA polymerase (RdRp). This replication intermediate forms a double-stranded RNA (dsRNA) complex with the viral genomic RNA template. Viral RNAs accumulate in the infected cells, including dsRNA and single-stranded RNA (ssRNA) products (9), and can be detected by host cell pattern recognition receptors (PRRs) as pathogen-associated molecular patterns (PAMPs). PRRs relevant to flavivirus infection include Toll-like receptor 3 (TLR3), TLR7 (10–13), and the RIG-I-like receptors (RLRs), including retinoic acid-inducible gene I (RIG-I) and melanoma differentiation-associated gene 5 (MDA5) (14, 15). PRR signaling activates downstream transcription factors, including interferon regulatory factor 3 (IRF3), IRF7, and nuclear factor kappa B (NF-κB), to drive innate immune activation and the expression of antiviral genes, including type I and III interferon (IFN) (9). In particular, during infection by West Nile virus (WNV), an emerging flavivirus related to ZIKV, TLR signaling has a minor role in PAMP recognition and innate immune signaling in vitro and in vivo (10, 13). In contrast, RLR signaling by RIG-I and MDA5 is essential for recognition of viral RNA to trigger innate immune activation from within the infected cell, ultimately driving systemic protection against infection and disease (14). The RLRs have also been shown to play a direct and essential role in the recognition of other flaviviruses, including dengue virus and Japanese encephalitis virus (16, 17). A recent study revealed that RNA isolated from cells infected with a ZIKV isolate from Brazil could trigger innate immune activation through RIG-I and the RLR signaling adaptor protein, mitochondrial antiviral-signaling protein (MAVS) (18). Flaviviruses, including ZIKV, overcome the actions of IFN induced by PRR signaling by targeting and regulating the Jak-STAT signaling pathway (19–21). In this context of Jak-STAT regulation, the processes of PRR signaling and IRF activation impart the major antiviral defense program against flaviviruses and play essential roles in controlling flavivirus infection and immunity (9).
In the present study, we conducted comparative viral genomic, virologic, and innate immune activation phenotype analyses of human cell culture infection by African and Asian lineage ZIKV strains, including the African lineage prototype MR766 (MR766; ZIKV/Uganda), the Asian lineage strain FSS13025 isolated in Cambodia in 2010 (FSS13025; ZIKV/Cambodia), and the newly emerging Asian lineage strain Brazil Fortaleza 2015 isolated in northeastern Brazil in 2015 (Brazil Fortaleza 2015; ZIKV/Brazil), an area where the majority of ZIKV-associated microcephaly cases were reported (22). We reveal differences in viral sequences, viral replication, and innate immune activation dynamics between ZIKV strains. By viral sequence analysis, ZIKV/Brazil was assigned to clade Z and identified as a carrier of the NS1 S139N mutation that is linked to microcephaly (23, 24). We further demonstrate that innate immune signaling induced by ZIKV infection is dependent upon RIG-I.
RESULTS
Sequence comparison of distinct ZIKV strains suggests different association with CZVS.We evaluated the properties of three ZIKV strains, including ZIKV/Uganda (GenBank no. NC_012532), ZIKV/Cambodia (GenBank no. MH368551), and ZIKV/Brazil (GenBank no. KX811222). To phylogenetically assign these strains, we first subjected RNA isolated from each virus stock to next-generation deep sequencing analysis. After contig assembly and sequence alignment, we confirmed the presence of a previously described deletion within the E glycoprotein coding region of the ZIKV/Uganda E protein (E153-156, corresponding to positions 443 to 446 in the polyprotein) (Fig. 1A). This region contains a glycosylation site (E154) that has been reported to correlate with enhanced in vivo infectivity (25, 26). Furthermore, alignment of the viral polyprotein open reading frame (ORF) of each strain demonstrated a nucleotide divergence of 12.7% between the sequence of ZIKV/Uganda and those of ZIKV/Cambodia and ZIKV/Brazil, which translated into a 3.3% amino acid divergence (110 amino acid exchanges plus 4 deleted residues, ZIKV/Uganda versus ZIKV/Brazil) (Fig. 1B and C) (Table 1), while amino acid sequences of ZIKV/Cambodia and ZIKV/Brazil demonstrated high identity (99.5%) and differed in only 18 out of 3,423 residues (Table 2). Eight of these changes were shared between ZIKV/Cambodia and ZIKV/Uganda (Tables 1 and 2, bold residues). Alignment of single protein sequences of ZIKV/Cambodia and ZIKV/Uganda with ZIKV/Brazil revealed that the highest divergence was contained within the prM protein (Fig. 1D). Overall, nonstructural proteins were better conserved across the ZIKV strains than the structural proteins, which might primarily impact differences regarding entry, maturation, and release of virus particles. According to a recently published phylogenetic study, we assigned ZIKV/Cambodia to node C based on the emergence of the T777M substitution in E glycoprotein and ZIKV/Brazil to node G (appearance of 2634V in NS5 RNA polymerase) (27). Furthermore, mutation 1143V in NS1 identified ZIKV/Brazil as a member of clade Z within node G, a group which is associated with a high rate (33%) of congenital ZIKV syndrome (CZVS) (Fig. 1A) (23). ZIKV/Brazil mutation A982V in NS1 is reported to be associated with enhanced infectivity in Aedes sp. mosquitoes, possibly contributing to the recent spread of ZIKV (Fig. 1A) (28). In addition, ZIKV/Brazil carries a S139N substitution in prM (Fig. 1A), a mutation which has been reported to result in higher infectivity of human and mouse neural progenitor cells and more severe microcephaly in the mouse fetus (24).
Sequence comparison of African and Asian lineage ZIKV strains. (A) Schematic of the ZIKV polyprotein. An asterisk indicates the position of a deletion in ZIKV/Uganda E protein. Arrowheads indicate positions of mutations referred to in Results. (B and C) After deep sequencing of viral stocks and contig generation, nucleotides from the open reading frames (ORFs) (B) or amino acid sequences (C) of ZIKV strains and WNV TX were aligned by the Jotun Hein method using MegAlign software (DNASTAR). The matrix presents percent identity and percent divergence, as well as nucleotide (B) or amino acid (aa) sequence (C) length. (D) Amino acid sequence comparison of single proteins of ZIKV/Uganda and ZIKV/Cambodia to ZIKV/Brazil depicted as percent divergence.
Amino acid residues differing between ZIKV/Uganda and ZIKV/Brazila
Amino acid changes between ZIKV/Cambodia and ZIKV/Brazila
Dynamics of viral growth and innate immune activation.To determine how viral sequence distinctions relate to ZIKV replication and innate immune activation of the host cell, we examined viral replication and host cell innate immune gene induction in immunocompetent A549 human epithelial cells. As a first step, we infected A549 cells with viral titers derived from plaque assays performed with Vero cells, which lack type I IFN production and are a standard cell line to measure viral titers for Flaviviridae. However, immunofluorescent staining for dsRNA, used as a marker for infected cells, demonstrated significant variation in infection efficiency of A549 cells, with ZIKV/Brazil infecting at the lowest efficiency (Fig. 2A and B). To allow more consistent infection of A549 cells across the different viral strains, we measured the titers of viral stocks by focus-forming unit (FFU) assays of infection in A549 cells and in Vero cells. This comparison established a titration analysis of each virus in the two cell lines and allowed us to define infection efficiency in each cell line across viral strains. We found that ZIKV/Brazil indeed infected A549 cells with lower efficiency than Vero cells (Fig. 2C). Twenty-four hours after ZIKV challenge (multiplicity of infection [MOI] = 1, based on viral titers derived from an FFU assay performed with A549 cells), we examined infection rates of A549 cells (Fig. 2D and E). We found that ZIKV/Uganda and ZIKV/Brazil display significant differences in infection rates (mean infection rates as follows: ZIKV/Uganda, 85.6%; ZIKV/Brazil, 59.0%; ZIKV/Cambodia, 71.9%) but with a minor difference in comparison to infection of cells using an MOI based on Vero cell-derived viral titers (mean infection rates as follows: ZIKV/Uganda, 80.0%; ZIKV/Brazil, 29.9%; ZIKV/Cambodia, 49.4%) (compare Fig. 2B and E). Next, we conducted one-step growth analyses of each viral strain in A549 cells. As a comparison, we also included analyses of West Nile virus (WNV) TX-02 (termed TX), a North American emerging flavivirus, and Sendai virus (SeV), a member of the Paramyxoviridae family (29). While production of viral RNA (intracellular) and infectious virus (extracellular) by ZIKV/Cambodia peaked by 24 h, virus production by ZIKV/Uganda peaked at 48 h and ZIKV/Brazil increased virus production up to the 72-h time point (Fig. 2F and G). One striking difference between the strains was the significant drop in viral RNA and infectious virus particle production for ZIKV/Cambodia after 24 h of infection (Fig. 2F and G). This observation was confirmed by examining ZIKV NS5 protein abundance over the infection time course (Fig. 2H). Thus, the ZIKV strains phenotypically differ in their viral replication properties in our epithelial cell infection model, with ZIKV/Uganda and ZIKV/Brazil exhibiting a higher replication fitness than that of ZIKV/Cambodia.
Growth kinetics and innate immune activation of African and Asian lineage ZIKV strains in A549 cells. (A and B) Infection efficiency of ZIKV strains in A549 cells. MOIs were calculated with viral titers derived from plaque assays performed with Vero cells. After infection of A549 cells with an MOI of 5 for 24 h, samples were analyzed by immunofluorescent staining for dsRNA. Per condition, three randomly selected fields of view representing at least 300 cells were analyzed. Images were acquired on a Nikon Eclipse Ti confocal microscope and manually analyzed in ImageJ using the multipoint tool. The number of dsRNA-positive cells was normalized to the total number of cells as determined by DAPI staining of nuclei and is presented as the percentage of infection efficiency. Depicted are means and standard deviations (SD) of the results of two independent experiments (n = 2). Statistical analysis was performed with a one-way analysis of variance (ANOVA), followed by Tukey’s range test. ns, not significant; *, P < 0.05; **, P < 0.01. (C) Focus-forming unit (FFU) assay performed with Vero and A549 cells to obtain viral titers for ZIKV and WNV stocks for both cell lines. Data are derived from two FFU assays performed independently and are presented as the means with SD on a log scale. (D and E) Infection efficiency of ZIKV strains and WNV TX in A549 cells after calculating the MOI with viral titers derived from an FFU assay performed with A549 cells (shown in panel C). At 24 h after virus challenge, infection rates were analyzed by immunofluorescent staining for viral protein NS4B and dsRNA. Per condition, at least five randomly selected fields of view representing at least 230 cells in total were analyzed. Images were acquired on a Nikon Eclipse Ti confocal microscope and manually analyzed in ImageJ using the multipoint tool. The number of NS4B-positive cells was normalized to the total number of cells as determined by DAPI staining of nuclei and is presented as the percentage of infection efficiency in panel E. Data are derived from two independent experiments and were tested for statistical significance by one-way ANOVA, followed by Tukey’s range test. ns, not significant; **, P < 0.01. (F to J) A549 cells were infected with SeV (40 hemagglutination units [HAU]/ml) or ZIKV or WNV (both at an MOI of 1) based on viral titers derived from an FFU assay of A549 cells. Cell lysates and culture supernatants were collected after 6, 24, 48, and 72 h to determine intracellular viral RNA, viral particles in the supernatant, and viral protein and for analysis of the innate immune response. Three independent experiments were performed (n = 3). (F) Analysis of ZIKV, WNV, and SeV RNA by SYBR green qPCR using virus-specific primer pairs and normalized to RPL13a housekeeping gene expression. Shown are the mean fold changes over the value at 6 h and SD calculated from three independent experiments performed in triplicate and presented on a log scale (log10). A two-way ANOVA followed by Tukey’s range test was used to test for statistical significance. ns, not significant; *, P < 0.05; **, P < 0.01; ***, P < 0.001. p.i., postinfection. (G) Viral titers in supernatants were measured by an FFU assay performed on Vero cells. Presented are means and SD calculated from three independent experiments on a log scale. A two-way ANOVA followed by Tukey’s range test was used to test for significance. ns, not significant; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001. (H and I) Western blot analysis of cell lysates collected at the indicated time points after infection. Viral protein levels were measured using antibodies specific for ZIKV NS5, WNV NS3, or parainfluenza virus (SeV). The innate immune response was analyzed with antibodies targeting IRF3 phosphorylated at serine 386 (IRF3 S386), total IRF3 (IRF3), and the ISGs IFIT1 and MxA. The actin level served as a loading control. Depicted is one representative Western blot of three independent experiments. (J) IFN-β and IFIT1 mRNA expression analyzed by SYBR green qPCR and normalized to RPL13a housekeeping gene expression using the total RNA harvested from infection experiments described above. Presented are means and SDs of the results of three independent experiments performed in triplicate and depicted on a log scale. A two-way ANOVA followed by Tukey’s range test was used to test for statistical significance. ns, not significant; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001.
As a readout for innate immune activation in A549 cells, immunoblot analyses were conducted to evaluate IRF3 activation as indicated by serine 386 and serine 396 phosphorylation (IRF3 S386, IRF3 S396). ZIKV/Uganda and ZIKV/Cambodia induced IRF3 phosphorylation/activation by 24 h, while IRF3 activation appeared delayed and remained weak over the whole time course in cells infected with ZIKV/Brazil (Fig. 2H). This observation was surprising to us considering the consistently high viral RNA, virus particle, and viral protein levels produced in cells infected with ZIKV/Brazil. In contrast, infection with ZIKV/Uganda demonstrated viral RNA, virus particle, and viral protein levels comparable to those of ZIKV/Brazil but induced strong IRF3 activation at 24, 48, and 72 h of infection (Fig. 2F to H). This delay and overall weak induction of IRF3 phosphorylation by ZIKV/Brazil, despite ongoing viral replication and presence of viral PAMPs, could point to an evasion of IRF3 activation by this ZIKV strain. With ZIKV/Cambodia, a strong decrease of the IRF3 S386 signal coincided with a drop in viral RNA, virus particle, and NS5 protein levels, suggesting an inhibition of ZIKV/Cambodia replication by the induced immune response and a rapid resolution of IRF3 signaling (Fig. 2F to H). Similarly, WNV infection induced strong IRF3 phosphorylation at 24 h that resolved over 48 and 72 h with a drop in virus replication (Fig. 2F, G, and I). In comparison, SeV, a model virus for innate immune activation, induced early and brief IRF3 activation in A549 cells, which mostly resolved at 48 h after infection (Fig. 2I). In line with this observation, the strong IRF3 activation induced by ZIKV/Uganda, ZIKV/Cambodia, WNV, and SeV but not ZIKV/Brazil infection was followed by a reduction in total IRF3 over time (Fig. 2H and I).
To determine how each ZIKV strain engages the host cell to trigger innate immune activation and antiviral defenses, we assessed the expression of IFIT1 and IFN-β mRNA as well as IFIT1 and MxA protein (Fig. 2H to J). Expression of both IFN-β and IFIT1 is induced in large part by activated IRF3, while IFIT1 and MxA are induced by IFN-β or IFN-α as interferon-stimulated genes (ISGs) via activation of the transcription factor interferon-stimulated gene factor 3 (ISGF3) (30). Protein expression of IFIT1 (IRF3 and ISGF3 induced) and MxA (ISGF3 induced) and transcript level of IFN-β (IRF3 induced) and IFIT1 were strongly increased 24 h after ZIKV/Cambodia infection, concomitant with reduced viral protein abundance (Fig. 2H and J). In line with the weak and delayed IRF3 activation after ZIKV/Brazil infection, induction of the IFN-β transcript level remained weak over the whole time course, and despite a stronger induction of the IFIT1 transcript level at 24 h, the ISG protein level increased only at 48 h of ZIKV/Brazil infection and did not result in reduced viral replication (Fig. 2H and J). ZIKV/Uganda infection induced rapid IFN-β and IFIT1 mRNA expression; however, this did not translate into strong IFIT1 or MxA protein expression, which is likely due to the overall higher infection efficiency and inhibition of Jak-STAT signaling in infected cells (Fig. 2D, E, H, and J). In comparison, analysis of WNV infection demonstrated viral replication and innate immune activation kinetics similar to that of ZIKV/Uganda, resulting in strong induction of ISG expression within 24 h after virus challenge. As a control, SeV infection induced a much faster innate immune response, with the highest IFN-β and IFIT1 mRNA expression at 6 h postinfection, the earliest time point measured, showing that our A549 cell model exhibits robust innate immune signaling programs. In summary, our analyses of virus replication and activation of innate immune signaling reveal striking differences between ZIKV strains, with a much weaker and delayed induction of innate immune activation by ZIKV/Brazil.
Differential IRF3 activation kinetics and IFN sensitivity of ZIKV strains.To determine if the observed delayed and overall weak IRF3 activation induced by ZIKV/Brazil was derived from its slightly lower infection efficiency (ZIKV/Uganda, 85.6%; ZIKV/Brazil, 59.0%; ZIKV/Cambodia, 71.9%) (Fig. 2D and E) or an inherent feature of the infection model, we used immunofluorescence analysis to simultaneously detect IRF3 nuclear translocation and dsRNA products of viral replication in A549 cells (Fig. 3A). Nuclear translocation of IRF3 was observed in more than 60% of cells containing dsRNA, a marker of viral replication, after 24 h of infection with ZIKV/Uganda or ZIKV/Cambodia, and in more than 95% of cells for SeV infection (all cells were considered SeV infected as demonstrated by anti-parainfluenza virus staining) (Fig. 3A and B). However, the majority of cells in cultures infected with ZIKV/Brazil did not contain activated/nuclear IRF3 at 24 h despite the presence of the viral dsRNA PAMP (Fig. 3A, top row; Fig. 3B). By 48 h of infection, 51% of ZIKV/Brazil dsRNA-positive cells had detectable nuclear IRF3 in comparison to 83% for ZIKV/Uganda, 80% for ZIKV/Cambodia, and 81% for WNV (Fig. 3A, row 2 and boxed areas; Fig. 3B). Thus, in comparison to two other ZIKV strains, ZIKV/Brazil displays significantly reduced IRF3 activation, despite the presence of dsRNA PAMP within infected cells.
ZIKV-induced IRF3 activation and type I IFN sensitivity. (A) Immunofluorescence analysis of IRF3 activation in A549 cells after infection with ZIKV (MOI = 5), WNV TX (MOI = 5), or SeV (20 HAU/ml). At 24 or 48 h postinfection (p.i.), cells were fixed and analyzed with antibodies targeting IRF3 and dsRNA or parainfluenza virus to control for SeV infection. The bottom row shows enlarged images of the boxed areas (insets: 48 h for ZIKV and WNV, 24 h for SeV). (B) Quantification of immunofluorescence shown in panel A. dsRNA/nuclear IRF3 double-positive cells were counted for 150 to 300 dsRNA-positive cells per condition and expressed as a percentage of double-positive cells. The graph depicts the means and SD of the results of three independent experiments (n = 3), with one data point representing one analyzed field of view. For SeV, cells with nuclear IRF3 were counted and compared to the total cell number, since at 20 HAU/ml, the infection efficiency was ∼100% as determined by anti-parainfluenza virus staining. Statistical analysis was performed with one-way ANOVA, followed by Tukey’s range test. ns, not significant; ***, P < 0.001; ****, P < 0.0001. (C) A549 cells were infected with ZIKV/Brazil (MOI = 1), and at 7 h p.i., cells were superinfected with SeV (20 HAU/ml) for 17 h. After 24 h of infection with ZIKV/Brazil, cells were fixed for immunofluorescent staining with antibodies targeting dsRNA to detect ZIKV-infected cells, IRF3, or IFIT1. The right columns show enlarged images of the boxed areas (insets). (D) Quantification of immunofluorescence shown in panel C. dsRNA/nuclear IRF3 double-positive cells were counted for 200 to 270 cells per condition and expressed as a percentage of double-positive cells. The graph depicts the means and SD of the results of two independent experiments (n = 2), with one data point representing one analyzed field of view. A one-way ANOVA analysis followed by Tukey’s range test was used to test for statistical significance. ns, not significant; ****, P < 0.0001. (E) Type I IFN sensitivity of ZIKV strains in Vero cells. Vero cells were treated with IFN-β (500 IU/ml) for 4 h, followed by infection with the indicated ZIKV strains (MOI = 5). Cells were harvested at the indicated time points, and lysates were analyzed by Western blotting. One representative Western blot of three independent experiments is shown. (F and G) ZIKV RNA and IFITM1 mRNA expression measured by qPCR analysis of experiments performed analogous to those described for panel E, with 1 h of IFN-β pretreatment before ZIKV infection. Values for ZIKV RNA were normalized to those of the RLP13a housekeeping gene and the respective untreated control. IFITM1 transcript levels were normalized to those of the RLP13a housekeeping gene and are presented on a log scale. Depicted are the means and SD of the results from three independent experiments (n = 3). Statistical analysis was performed with a two-way ANOVA followed by Tukey’s range test. ns, not significant.
To determine if ZIKV/Brazil imparts a blockade to antagonize IRF3 activation during infection, we assessed its ability to block SeV-induced IRF3 activation in a superinfection model. In this approach, the cells are first infected with ZIKV/Brazil, followed by superinfection with SeV 7 h after ZIKV/Brazil challenge. Immunofluorescent analysis clearly showed that ZIKV/Brazil virus replication did not block nuclear translocation of IRF3 induced by SeV (Fig. 3C and D). As observed before, ZIKV/Brazil infection alone triggered IRF3 activation in only a small number of infected, dsRNA-positive cells (9%), and IFIT1 protein expression occurred primarily in uninfected bystander cells, likely through paracrine signaling by IFN produced from infected cells (Fig. 3C and D). However, infection with SeV alone and superinfection with SeV resulted in IRF3 nuclear translocation in a majority of cells (91% and 87%, respectively) (Fig. 3C and D). Therefore, it was unlikely that the weak and delayed induction of IRF3 activation upon ZIKV/Brazil infection was caused by an early active blockade of IRF3 activation by this ZIKV strain.
The observed levels of viral RNA, protein, and infectious particles generated across ZIKV strains in our infection experiments of A549 cells (Fig. 2F to H) suggest that ZIKV/Cambodia replication is suppressed by innate immune actions of the host cell. We therefore conducted analyses of IFN sensitivity across the different ZIKV strains. As a cell model for analysis of viral sensitivity to IFN, we chose Vero cells, because they are deficient in IFN production but retain an intact response to IFN (31). These properties of Vero cells prevent potential experimental bias that might otherwise occur when cells produce IFN in response to virus infection. Therefore, Vero cells were pretreated with IFN-β for 4 h (for assessment of viral protein) (Fig. 3E) or 1 h (for assessment of viral RNA levels) (Fig. 3F). Analysis of viral protein and RNA levels revealed that ZIKV/Cambodia is more sensitive to inhibition of viral replication by IFN, with reduced viral protein and RNA levels in infected cells pretreated with IFN (Fig. 3E and F). In contrast, ZIKV/Uganda and ZIKV/Brazil protein and viral RNA levels remained relatively stable over 72 h of treatment during Vero cell infection and were comparable to levels in untreated cells (Fig. 3E and F). ISG induction levels by IFN-β pretreatment were comparable across infection groups (Fig. 3G). All together, these results point to a higher sensitivity of ZIKV/Cambodia to the antiviral actions of IFN-β.
RIG-I plays a major role for induction of antiviral innate immunity during ZIKV infection.To evaluate the contribution of RLRs in detecting ZIKV infection and triggering innate immune activation, stable knockout lines of A549 cells independently lacking RIG-I, MDA5, or LGP2 were produced and analyzed. For controls, we generated A549 cell lines lacking MAVS (positive control; MAVS knockdown [KD]) or expressing nontargeting guide RNA (negative control; gNT). Knockout (KO) or KD efficiencies of RLR and MAVS proteins were analyzed by Western blotting after stimulating the cells with IFN-β and tumor necrosis factor alpha (TNF-α) to induce RLR expression (Fig. 4A). RIG-I, MDA5, and LGP2 proteins could not be detected in the respective knockout cell lines even after stimulation. However, a longer exposure revealed residual amounts of MAVS protein in control cells so that only a reduction of innate immune signaling could be expected for this MAVS KD cell line. IFIT1 was induced in response to IFN-β/TNF-α treatment, demonstrating that each cell line responds to stimulation with these cytokines (Fig. 4A). Analysis of innate immune activation after infection with ZIKV/Cambodia or WNV revealed strong IRF3 activation and ISG induction in the absence of MDA5 or LGP2, with each response being comparable to that of control cells (gNT) (Fig. 4B, C, E, and F). In contrast, in RIG-I KO cells, IRF3 activation, as well as IFIT1 and MxA protein expression, was strongly reduced, similar to that in MAVS KD cells (Fig. 4B). By 48 h of infection, the ZIKV/Cambodia NS5 protein level was found to be reduced in all cell lines except RIG-I KO and MAVS KD cells (Fig. 4B). We also measured IFIT1 and IFN-β mRNA expression over the infection time course in the knockout cells and observed a trend toward reduced IFIT1 mRNA and significantly reduced IFN-β mRNA expression in RIG-I KO cells upon 24 h of ZIKV infection, which indicated a dominant role of RIG-I for innate immune signaling during ZIKV infection in A549 epithelial cells at the investigated time points (Fig. 4C). Interestingly, 24 and 48 h after infection with ZIKV, a slight induction of ISG proteins and IFIT1 and IFN-β mRNA could be detected in RIG-I KO cells (Fig. 4B and C), which might be due to induction of innate immune signaling by MDA5 or TLR3 in these cells. These findings are in line with previous reports on MDA5-dependent late induction of innate immune signaling in response to infection with members of the Flaviviridae family (14, 18, 32). In line with this strong reduction in antiviral immune responses in the absence of RIG-I, 48 h after infection, ZIKV RNA level and virus particle production were significantly enhanced in RIG-I KO cells (Fig. 4D). Similarly, upon challenge with WNV, IRF3 activation and ISG protein expression were strongly reduced in cells lacking RIG-I (Fig. 4E). In addition, RIG-I KO cells demonstrated a significant reduction in IFIT1 mRNA expression and trended toward reduced IFN-β transcript levels 24 and 48 h after infection (Fig. 4F). As observed for ZIKV/Cambodia challenge, IFIT1 and IFN-β mRNA were induced at 24 and 48 h after WNV infection in RIG-I KO cells, though this did not result in the detectable production of IFIT1 or MxA protein (Fig. 4E and F). As with ZIKV/Cambodia, the weaker immune response in RIG-I KO cells translated into a trend toward higher WNV RNA and virus particle level (Fig. 4G). In nontargeting control cells, as well as MDA5 and LGP2 KO cells, viral RNA and infectious virus particles plateaued or dropped after 24 h of infection (Fig. 4D and G). In contrast, RIG-I KO and MAVS KD cells supported ongoing viral replication even beyond 24 h of infection for both flaviviruses, underlining the importance of RIG-I and IRF3 activation for restriction of ZIKV and WNV replication (Fig. 4D and G).
RLR dependence of innate immune signaling during ZIKV and WNV infection of A549 cells. (A) A549 cells with single RLR knockout (KO) or MAVS knockdown (KD) were generated by CRISPR Cas9 and analyzed for RLR, MAVS, and ISG protein level upon stimulation with IFN-β (100 IU/ml) and TNF-α (50 ng/ml) for 24 h. An asterisk indicates prolonged exposure. (B) A549 RLR KO and MAVS KD cells were infected with ZIKV/Cambodia (MOI = 5). At the indicated time points, cell lysates were generated and analyzed by Western blotting for IRF3 activation (IRF3 S396), ISG induction, and ZIKV NS5 protein level. Depicted is one representative blot of three independent experiments (n = 3). An asterisk indicates prolonged exposure. (C) IFIT1 and IFN-β mRNA expression measured by qPCR analysis of experiments described for panel B. The graph depicts the means and SD of the results of three independent experiments (n = 3) performed in triplicate and presented on a log scale (log10). Data were normalized to those of the GAPDH housekeeping gene. (D) Viral growth kinetics of infection experiments described for panel B determined by qPCR (left) and plaque assay (Vero cells) (right). (Left) A graph presents the means and SD of the results of three independent experiments (n = 3) measured in triplicate and depicted on a log scale (log10). (Right) Supernatants were collected at the indicated time points, and PFU/ml were determined by plaque assay on Vero cells. The graph presents the means and SD of the results of three independent experiments (n = 3) on a log scale (log10). (E) Analogous to infection experiments described for panel B, A549 RLR KO or MAVS KD cells were infected with WNV TX (MOI = 5), and cell lysates were harvested at the indicated time points and analyzed by Western blotting for innate immune activation (IRF3 Ser396, IFIT1, MxA) and WNV NS3 protein level. Depicted is one representative blot of three independent experiments (n = 3). (F) qPCR analysis of IFIT1 and IFN-β mRNA expression upon infection of A549 KO or KD cells with WNV as described for panel E. The graph presents the mean and SD of three independent experiments (n = 3) performed in triplicate and presented on a log scale (log10). (G) WNV growth kinetics of A549 KO or KD cell infection experiments described for panel E. (Top) WNV RNA measured by qPCR using WNV-specific primer pairs. Data were normalized to GAPDH. The means and SD of the results of three independent experiments (n = 3) performed in triplicate are presented on a log scale (log10). (Bottom) Analysis of viral supernatants by plaque assay. Shown are the means and SD of the results of three independent experiments (n = 3) depicted on a log scale (log10). Statistical analysis of all qPCR and plaque assay data was performed with two-way ANOVA, followed by Bonferroni’s multiple-comparison test. ns, not significant; *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001.
DISCUSSION
We compared viral sequences, viral replication, and antiviral innate immunity for three ZIKV strains derived from African and Asian lineages. We provide evidence for a much weaker and delayed activation of innate immune signaling by ZIKV/Brazil, a strain with a high association with CZVS. In comparison to ZIKV/Uganda and ZIKV/Brazil, viral replication of ZIKV/Cambodia, an Asian strain with less severe pathological outcomes (33), demonstrates a higher sensitivity to the antiviral action of IFN-β. Finally, we identify RIG-I as a major PRR at early time points after ZIKV infection.
The observed reduction and delay in IRF3 activation by the more virulent ZIKV strain ZIKV/Brazil are reminiscent of observations made for pathogenic WNV strains, which were reported to delay IRF3 activation by evasion of detection by the host cell early after infection (34, 35). Similarly, evasion of detection by PRRs might also be a strategy employed by ZIKV/Brazil to reduce and delay IRF3 activation, since our data gained from SeV superinfection experiments argue against a direct and active block of IRF3 activation by this virus. A potential reason for the reduction and delay in IRF3 activation during ZIKV/Brazil infection could be the smaller amounts of viral PAMPs, although levels of released infectious viral particles or viral RNA during viral growth experiments were not reduced for ZIKV/Brazil, arguing against this notion. Another potential mechanism for the delay in IRF3 activation could be protection of ZIKV/Brazil PAMPs from detection by RIG-I or other PRRs within the viral replication compartment, as has been shown for hepatitis C virus (HCV) (36). Apart from delayed activation of IRF3, ZIKV/Brazil trended toward enhanced resistance to type I IFN antiviral actions in comparison to ZIKV/Cambodia, indicated by more consistent viral RNA and NS5 protein levels upon IFN treatment, suggesting that evasion of IFN actions might possibly contribute to the higher virulence of this virus strain. The difference in IFN sensitivity between ZIKV/Brazil and ZIKV/Cambodia is unlikely to be due to differences in viral replication fitness, something that has previously been described for IFN-resistant and IFN-sensitive WNV strains (29), since these virus strains demonstrate comparable replication kinetics up to 24 h postinfection. Also, a more efficient block of Jak-STAT signaling by ZIKV/Brazil is unlikely as an underlying mechanism, since equal amounts of the ISGs IFIT1 and MxA were induced by IFN-β treatment under the different infection conditions. An interesting possibility to define these differences might lie within the 18 amino acids that differ between ZIKV/Cambodia and ZIKV/Brazil and that could impact host innate immune effector functions (37), including rendering a delay in IRF3 activation. For ZIKV/Brazil, these features could contribute to the high association of ZIKV clade Z with CZVS. Two groups have reported binding of ZIKV RNA by RIG-I (18, 38). While Chazal et al. focused on RIG-I–RNA binding after ZIKV infection of HEK293 cells for 24 h, Hertzog et al. studied the contribution of RIG-I and MDA5 in HEK cells transfected with total RNA extracted from ZIKV-infected A549 cells, finding evidence for a role of RIG-I and MDA5 in directing IFN induction in response to ZIKV RNA, similar to our previous studies of WNV (14). Our observations from ZIKV infection in the different RLR knockout A549 cells support the findings by both groups. Our work does, however, reveal a dominant role of RIG-I in the detection of ZIKV PAMPs at early time points during infection of otherwise immunocompetent A549 cells. At later time points during infection, induction of innate immune signaling was detected even in the absence of RIG-I, suggesting a contribution of MDA5 to the antiviral innate immune response. A sequential contribution of RIG-I and MDA5 is in line with previous observations made for WNV and HCV (14, 32). We propose that RIG-I serves to recognize PAMPs that are presented early in the ZIKV infection cycle while MDA5-stimulatory viral PAMPs might accumulate at later time points during the viral replication cycle or become more accessible for detection by MDA5, possibly by morphological changes of viral replication compartments. In addition, low basal MDA5 protein levels might require an IFN-driven induction of MDA5 expression prior to a significant contribution of this PRR to innate immune signaling. We also note that we cannot exclude cell line- or cell type-specific differences in the contribution of each RLR to virus recognition and induction of an antiviral response which can occur differentially among tissues (9).
All together, our findings reveal striking differences in host cell innate immune responses and point to a difference in IFN sensitivity of distinct ZIKV strains. In summary, in contrast with other ZIKV strains, the emerging strain ZIKV/Brazil presented with characteristics of a stealth virus that does not robustly trigger innate immune activation, despite the accumulation of viral RNA/PAMPs in the infected cell. Moreover, ZIKV/Cambodia exhibits increased sensitivity to IFN compared to other ZIKV strains. These viral properties should be considered when choosing ZIKV strains for future studies of innate immune activation and response. Uncovering virus strain-specific features of innate immune regulation is critical for understanding ZIKV infection outcome and is paramount for developing therapeutic strategies that harness innate immunity for the control of ZIKV infection.
MATERIALS AND METHODS
Cell culture and virus stocks.A549 lung epithelial cells and Vero cells were maintained in Dulbecco’s modified Eagle medium (DMEM) supplemented with 10% fetal calf serum, 2 mM l-glutamine, 1 mM sodium pyruvate, penicillin-streptomycin solution, 1× nonessential amino acids (all from Fisher Scientific, NH, USA), and 10 mM HEPES (Corning, NY, USA) (complete DMEM). The following flaviviruses were used during this study: West Nile Virus, 2002, Texas, Hall County (termed WNV TX); Zika virus African strain, 1947, Uganda, MR766 (termed ZIKV/Uganda); Zika virus Asian strain, 2010, Cambodia, FSS13025 (termed ZIKV/Cambodia); Zika virus Asian strain, 2015, Brazil, Fortaleza, Fortaleza 2015 (termed ZIKV/Brazil). WNV TX (GenBank no. DQ176637) was isolated and working stocks were prepared as previously described (29). ZIKV/Cambodia (GenBank no. KU955593) was provided by the World Reference Center of Emerging Viruses and Arboviruses (WRCEVA, Galveston, TX, USA). ZIKV/Brazil (GenBank no. KX811222) was provided by M. Diamond (Washington University School of Medicine in St. Louis). ZIKV MR766 (Uganda) (GenBank no. NC_012532) was purchased from the ATCC (ATCC VR-84). All ZIKV working stocks were generated from plaque-purified isolates amplified in Vero cells, and viral titers are derived either from plaque assays performed with Vero cells or from FFU assays performed with A549 cells as stated in the text and figure legends. Sendai virus Cantell strain (SeV) was purchased from Charles River (Seattle, WA, USA). All viral stocks and cell lines tested negative for mycoplasma.
Deep sequencing of viral stocks and sequence alignments.RNA for whole-transcriptome sequencing (RNA-Seq) was isolated from virus stocks derived from plaque-purified isolates, passaged one time in Vero cells with the QIAamp viral RNA minikit (Qiagen, Germantown, MD, USA) according to the instructions of the manufacturer. The RNA was digested with DNase I, followed by purification with the Qiagen RNeasy minikit (Qiagen, USA), and stored at −80°C. RNA concentrations were measured on a Qubit fluorometer (Invitrogen, Carlsbad, CA, USA), and RNA quality was analyzed on the Agilent 2100 bioanalyzer (Agilent, Santa Clara, CA, USA). rRNA was depleted from each RNA sample by using the Ribo-Zero rRNA removal kit (Epicentre). Libraries were prepared from 150 ng of rRNA-depleted RNA by following the KAPA stranded RNA-Seq with RiboErase workflow for Total RNA-Seq libraries (KAPA Biosystems). Library quality was evaluated using the Qubit 3.0 fluorometer and the Agilent 2100 bioanalyzer instrument. Constructed libraries were sequenced on a NextSeq 500 Illumina platform, producing 2 × 75nt stranded paired-end reads. Raw RNA-Seq data were demultiplexed and checked for quality (FastQC version 0.11.3), and adapters were removed. After quality control, the sequencing files (Fastqs) were loaded onto the de novo assembler Trinity (version 2.4.0) and assembled using methods previously outlined (39, 40). Each sample had roughly 30 million raw reads. The Trinity assembler has several components (Jellyfish, Inchworm, Chysalis, Butterfly) that assemble sequences into contigs with k-mer clusters and build de Bruijn graphs. Trinity generated viral contigs from the strain-specific samples, and these sequences were deposited into GenBank under the following accession numbers: ZIKV/Brazil, KX811222.1; ZIKV/Uganda, MK105975; ZIKV/Cambodia, MH368551. Viral sequences were aligned using the Jotun Hein method of MegAlign software, version 10.0.1 (DNASTAR, Madison, WI, USA), to identify nucleotide and amino acid differences.
Plaque assay.A total of 3.7 × 105 Vero cells were seeded in 6-well dishes, and virus-containing supernatants were serially diluted in complete DMEM. Vero cell monolayers were incubated in duplicate with dilutions shaking at 37°C for 2 h. Monolayers were overlaid with 1% agarose, and 3 to 4 days later, plaques were visualized with a 1% agarose overlay containing 3% neutral red (Sigma, St. Louis, MO, USA).
FFU assay.A total of 1 × 104 Vero cells or 2 × 104 A549 cells were seeded in a 96-well plate in a final volume of 100 μl complete DMEM containing 10% fetal bovine serum (FBS). The next day, the medium was removed from the cells and replaced with 100 μl of serial dilutions of virus containing supernatant made in complete DMEM containing 2% FBS. After incubation of the cells for 2 h at 37°C and 5% CO2, the cells were covered with ∼120 μl minimum essential medium (MEM) supplemented with 2% FBS, 1% methylcellulose, a penicillin-streptomycin solution (all from Fisher Scientific, NH, USA), and 10 mM HEPES (Corning, NY, USA). At 20 h later, the overlay and medium were removed, and the cells were fixed for 20 min at room temperature (RT) by the addition of 200 μl 4% paraformaldehyde (PFA) per well. After washing the cells three times with phosphate-buffered saline (PBS), cells were permeabilized with Perm/Wash (BD biosciences) for 5 to 10 min at RT. After the removal of Perm/Wash, 50 μl of a fluorescein isothiocyanate (FITC)-conjugated anti-ZIKV-E 4G2 antibody (1:250) or an FITC-conjugated anti-ZIKV-E ZV13 antibody (cross-reactive against WNV) (1:1,000) diluted in Perm/Wash was added. After gently rocking the plates for 2 h at RT in the dark and three washes with PBS, foci of infected cells were detected in an IncuCyte Zoom analysis system and analyzed with the IncuCyte Zoom 2018A software (Sartorius, Essen BioScience, MI, USA). Anti-ZIKV-E 4G2 (derived from an anti-pan-flavivirus hybridoma cell line, ATCC HB-112) and anti-ZIKV-E ZV13 (41) (gift of Michael S. Diamond, Washington University School of Medicine in St. Louis) antibodies were conjugated to FITC using the Fluorescein-EX labeling kit (Thermo Fisher) according to the instructions of the manufacturer.
SYBR green quantitative PCR.Total RNA was isolated from cell lysates with the RNeasy kit (Qiagen, Germantown, MD, USA) by following the instructions of the manufacturer. All samples were DNase I digested. Five hundred nanograms of total RNA was subjected to cDNA synthesis with the iScript cDNA synthesis kit (Bio-Rad, Hercules, CA, USA) according to the manufacturer’s instructions. cDNA was diluted 1:8 in H2O, and SYBR green quantitative PCR (qPCR) was performed on an Applied Biosystems ViiA 7 real-time PCR machine using the SYBR select master mix (Thermo Fisher, Waltham, MA, USA) and gene-specific primers. Primers for specific detection of cellular target genes and viral genomes can be found in Table 3. qPCR analyses were run in triplicate, and data were analyzed using the 2–ΔΔCT method and compared to GAPDH or Rpl13a housekeeping genes.
Primer sequences and gRNA sequencesa
SDS-PAGE and Western blot analysis.Cells were lysed in a modified radioimmunoprecipitation (RIPA) buffer (150 mM NaCl, 50 mM Tris-HCl [pH 7.6], 1% Triton X-100, 0.5% sodium deoxycholate) with freshly added protease inhibitor cocktail (Sigma, St. Louis, MO, USA), phosphatase inhibitor cocktail (Millipore, Burlington, MA, USA), and okadaic acid (Thermo Fisher, Waltham, MA, USA). Protein concentrations were measured using the Pierce BCA protein assay kit (Thermo Fisher), and 12 μg protein was subjected to SDS-PAGE and blotted onto polyvinylidene difluoride (PVDF) membranes (Thermo Fisher). After a blocking step with a 4% bovine serum albumin (BSA) solution, the membranes were incubated with the following primary antibodies at 4°C overnight: rabbit anti-IRF3 S386 (1:500) (ab76493; Abcam, MA, USA), rabbit anti-IRF3 S396 (1:1,000) (no. 29047; Cell Signaling Technology [CST], Danvers, MA, USA), rabbit anti-IRF3 (1:1,000) (no. 4302; CST), rabbit anti-IFIT1 no. 971 (1:1,000) and rabbit anti-MxA no. 340B (1:500) (both made at UT Southwestern Antibody Core facility by repeated injections of synthetic peptides into rabbits), mouse anti-pan-actin C4 (1:3,000) (MAB1501; Millipore), rabbit anti-ZIKV-NS5 (1:2,500) (GTX133312; GeneTex, Irvine, CA, USA), goat anti-WNV-NS3 (1:1,000) (BAF2901; R&D Systems, Minneapolis, MN, USA), goat anti-parainfluenza virus 1 (B65121G-1; Amsbio, Cambridge, MA, USA), mouse anti-RIG-I AlmeI (1:1,000) (AG-20B-0009-C100; AdipoGen, San Diego, CA, USA), rabbit anti-MDA5 (1:500) (29020; IBL, Minneapolis, MN, USA), rabbit anti-LGP2 (1:500) (no. 29030; IBL), and rabbit anti-Cardif AT107 (1:1,000) (ALX-210-929-C100; Enzo Life Sciences, Farmingdale, NY, USA). Bound primary antibodies were detected with 1:1,000 diluted horseradish peroxidase-coupled secondary antibodies: donkey anti-mouse antibody (715-035-150), donkey anti-rabbit antibody (711-035-152), donkey anti-goat antibody (705-035-003) (all from Jackson Immunoresearch, West Grove, PA, USA), and an ECL prime reagent (GE Healthcare, Pittsburgh, PA, USA) on a ChemiDoc Imager (Bio-Rad). If necessary, membranes were stripped for reprobing with a harsh stripping buffer containing 2% SDS and 0.8% β-mercaptoethanol.
Immunofluorescence.A total of 1 × 105 A549 cells were seeded onto glass coverslips in a 24-well plate. The next day, cells were infected with ZIKV, WNV, or SeV and fixed with 4% PFA for 15 min at 24 or 48 h postinfection. Cells were permeabilized with 0.5% Triton X-100 for 15 min at RT. After blocking the cells with 3% BSA in PBS, immunofluorescent staining was performed for 1 h at RT with primary antibodies directed against dsRNA (J2, mouse, 1:800; no. 10010500; Scicons, Budapest, Hungary), parainfluenza virus (1:200, goat, no. B65121G; Amsbio, MA, USA), rabbit anti-ZIKV-NS4B cross-reacting with WNV (1:500, no. 133311; GeneTex, CA, USA), IFIT1 (no. 971, 1:100, rabbit), or IRF3 (AR1, 1:100, mouse), which has been described elsewhere (42). Nuclei were counterstained with 4',6-diamidine-2'-phenylindole dihydrochloride (DAPI; Thermo Fisher). Isotype-specific and fluorophore-coupled secondary antibodies (1:1,000; Thermo Fisher) were applied for 1 h at RT. After washing, samples were mounted onto glass slides using ProLong gold (Thermo Fisher). Images were acquired with a Nikon Eclipse Ti confocal microscope equipped with a 60× oil immersion objective using the Nikon confocal software. Images were merged and processed using the Nikon confocal analysis software (NIS Elements AR 5.11.00; Nikon, Melville, NY, USA). Further analyses of the images were performed in ImageJ (version 1.52; NIH, USA) (43).
Generation of gene knockouts by CRISPR Cas9.For CRISPR Cas9-mediated gene knockout, guide RNA (gRNA) sequences were designed with the CRISPR tool of benchling (Biology Software, 2017; https://benchling.com). Upon annealing of the gRNA, target oligonucleotides were cloned into a puromycin-selectable Cas9-t2a-pRRL lentiviral vector (44) by digesting the plasmid with SbfI and AfeI and using the In-Fusion cloning kit (TaKaRa, Mountain View, CA, USA). Production of lentiviral particles and generation of stable cell lines were performed as described previously (45). Sequences for gRNA-directed gene knockout are summarized in Table 3. Upon transduction, cells were kept under continuous selection with 1.5 μg/ml puromycin and knockouts were confirmed by Western blotting.
Statistical analysis.Statistical analysis of all data was performed using GraphPad Prism 7.03 (GraphPad, La Jolla, CA, USA) and is described in the figure legends.
Data availability.Deep sequencing data were deposited in the Sequencing Read Archive (SRA) under the following accession numbers: BioProject no. PRJNA498737; GenBank no. KU955593, ZIKV/Cambodia, as derived from the World Reference Center of Emerging Viruses and Arboviruses; GenBank no. NC_012532, ZIKV/Uganda, as purchased from the ATCC; GenBank no. MH368551, ZIKV/Cambodia, plaque-picked isolate (sequence identified in the present study); GenBank no. MK105975, ZIKV/Uganda, plaque-picked isolate (sequence identified in the present study); GenBank no. KX811222.1, ZIKV/Brazil, plaque-picked isolate (sequence identified in the present study); and GenBank no. DQ176637, WNV TX (29).
ACKNOWLEDGMENTS
This work was supported by NIH grants AI143265, AI145296, AI104002, AI083019, and AI100625.
We thank the Seattle Genomics sequencing group for RNA sequencing of ZIKV stocks.
K.E.-N. designed and performed experiments, interpreted data, and wrote the manuscript, L.D.A. designed and cloned all CRISPR Cas9 gRNA constructs and generated A549 MAVS KO cells, J.A.R. helped with IFN sensitivity experiments, M.R.F. helped with total RNA isolation and Western blotting, R.G. processed the deep sequencing data and assembled the contigs, and M.G. designed the concept of this study, interpreted the results, and wrote the manuscript. All authors contributed to writing the manuscript.
We declare no conflicts of interest.
FOOTNOTES
- Received 16 April 2019.
- Accepted 18 April 2019.
- Accepted manuscript posted online 24 April 2019.
- Copyright © 2019 American Society for Microbiology.
