Cell type-specific consequences of IRF3 deficiency were reported by Daffis et al. in a study on WNV infection in mouse BMDM and PCNs [17]. Upon infection, Irf3-deficient PCN showed reduced IFN-I gene expression and concentrations of IFNα and IFNβ at 24 and 48 hpi in the supernatants compared with WT cells, associated with a moderate increase in viral load at 48 hpi. Irf7 induction was also delayed and decreased in Irf3-deficient PCN, which may be the cause of the low levels of IFN-I. Conversely, in BMDM, while the viral titer was higher in Irf3-deficient cells than in WT cells at 24 to72 hpi, Irf3-deficient BMDM had earlier and higher expression of IFN-I genes than WT BMDM. This induction was possibly driven by the increased expression of Irf7 in Irf3-deficient BMDM, in response to increased viral infection. While it is possible that the time points studied missed the early IRF3-dependent IFN-I induction which was observed before 24 h in other studies on Irf3 KO BMDM [25, 49, 57], these results demonstrate that the consequences of IRF3 deficiency on the control of viral infection in vitro may be cell-type specific, in particular regarding the compensatory role of Irf7.

As with in vitro experiments, mouse in vivo studies have been performed under diverse conditions, in particular using different viruses, doses and routes of infection (most often intraperitoneal, subcutaneous, intravenous and intranasal, occasionally intracranial or corneal). The mechanisms driving the progression of in vivo infections are more complex than in in vitro models as they involve interactions between multiple cell types and tissues. Importantly, as for in vitro experiments, studies comparing Irf3 WT and Irf3 KO mice were performed on the C57BL/6 J background and were therefore not affected by the influence of other genetic susceptibility factors.

A compilation of the in vivo results published using Irf3 KO mice is presented in Table 2, which shows less consistency in the consequences of IRF3 deficiency across mouse models than in in vitro studies. In some experiments, Irf3 KO mice showed higher viral load compared with WT mice in the tissues tested, while others did not detect significant differences. When measured, IFN-I levels were often found to be decreased but comparisons between studies are rarely possible due to different experimental conditions or different time points. As observed in vitro, Irf3-deficient and Irf3 Bcl2l12 double-deficient mice showed the same mortality following EMCV infection, indicating that Bcl2l12 loss of function did not modify the effect of Irf3 KO in these models [14, 16, 49].

Interestingly, a few studies have examined both in vitro and in vivo phenotypes and found consistent effects of IRF3 deficiency (see Tables 1 and 2). For instance, Yanai et al. have reported that HSV-1, VSV, and EMCV infection of Irf3 KO MEF, BMDC and BMDM resulted in lower Ifnb1 expression than in WT cells. After intravenous infection with EMCV, Irf3-deficient mice exhibited a higher mortality rate than WT mice and lower levels of IFNβ in the serum [49].

Similarly, HSV-1 infection of Irf3-deficient BMDC led to higher viral titer and lower levels of IFNβ in the supernatant than in WT BMDC [15]. Irf3 KO mice infected with HSV-1 either by the corneal and intracranial route showed a higher mortality rate than WT mice and higher viral titers in the brain. After intracranial infection, Irf3 KO mice also presented increased levels of cytokines in the brain, such as TNFα, IL10 and CCL5 [61].

In other reports, the susceptibility of Irf3-deficient cells was not associated with an increased susceptibility in vivo. Upon HSV-1 infection, Irf3 KO MEF showed a decreased induction of IFN-I, especially IFNβ, compared with WT MEF. However, both Irf3 KO mice and WT mice survived infection and exhibited the same level of serum IFNα [14]. Similarly, infection of Irf3 KO MEF with Chikungunya virus (CHIKV) induced lower expression of Ifnb1 compared with WT MEF, while Irf3 KO and WT mice had the same viral load in the blood and equally survived the infection [18].

These studies, along with others [14, 16, 17, 49, 58], have reported increased susceptibility in Irf3 KO mice mostly in cases where WT mice displayed clinical signs of illness and mortality. On the other hand, studies that found no effect of IRF3 deficiency on the outcome of viral infection were generally performed under conditions where infection had no effect on WT mice [14, 18, 62,63,64,65,66] as a consequence of the low intrinsic virus pathogenicity or of the route and dose of infection.

Differences between studies can be caused by different routes of infection, as exemplified with HSV-1. Honda et al. infected mice intravenously and observed 100% survival and the same levels of serum IFNα in WT and Irf3 KO mice [14]. On the other hand, Canivet et al. infected mice intranasally and reported a 10% survival rate in Irf3 KO mice while 70% of WT mice survived infection. Irf3 KO mice also showed increased body weight loss and viral titers as well as higher concentrations of inflammatory cytokines in the brain compared with WT mice [67]. Corneal and intracranial HSV-1 infection also led to higher mortality rates in Irf3-deficient mice than in WT mice [15].

In vivo studies in mice have also allowed investigating the impact of IRF3 deficiency on the kinetics of IFN-I response over several days. For example, in their above-mentioned study, Canivet et al. reported that, after intranasal HSV-1 infection, levels of IFNβ in brain homogenates were slightly lower at 3 days post-infection (dpi) but higher at 5 dpi in Irf3 KO mice compared with WT mice. The authors suggested the implication of alternative IFN induction pathways to explain the increased IFN-I production at 5 dpi [67].

Notably, the role of non-transcriptional activities of IRF3 in viral infections has been investigated and characterized in vivo. Mice carrying a mutated allele preserving functional RIPA and RIKA but not IRF3 transcriptional activity were more susceptible to SeV than Irf3 WT mice, but less than Irf3 KO mice. Moreover, Irf3-deficient mice were more susceptible to SeV infection than Ifnar1-deficient mice, which are unable to respond to IFN-I signaling. These results show that the non-transcriptional activities of IRF3 provide complementary antiviral pathways to the IFN-I induction and are required for a full IRF3-mediated protection against viral infection [21].

IRF3 deficiency did not always result in increased mortality rate, viral loads, or decreased IFN-I titers, depending on the virus studied or on the experimental conditions. However, other parameters may be modified, in particular the inflammatory response to infection which was found to be increased in several studies (Table 2). For instance, after infection with Sindbis virus (SINV), Irf3-deficient mice showed higher expression of Ccl2, Il1b, Tnfa and Il10 at 5 dpi in the brain and/or the spinal cord than WT mice. Moreover, the histological inflammation score was higher in Irf3 KO than in WT mice [68]. Likewise, Irf3-deficient mice showed higher levels of IL1α, IL1β, IL6, IL12p40, IL12p70 and IFNγ at 5 dpi after HSV-1 infection than WT mice [67]. Whether these effects are due to the absence of RIKA activity or indirectly caused by the lack of IRF3-dependent IFN-I expression was not established but could be investigated using an Irf3 allele in which the RIKA domain has been altered.

Finally, a few in vivo studies have revealed the role of IRF3 deficiency on the adaptative immune responses. Indeed, IRF3-mediated IFN-I signaling has been shown to mediate T cell differentiation in the context of infections [69]. Under basal conditions, the composition of lymphocyte populations is similar between Irf3 KO and WT mice [16]. However, after TMEV or IAV infection, Irf3-deficient mice showed reduced proportion of circulating CD8 T cells expressing granzyme B, an enzyme involved in the cytotoxicity of CD8 T cells, and impaired development of memory T cell population [70]. After SINV infection, Irf3-deficient mice showed increased infiltration of CD4 T cells and CD19 B cells in the brain [68]. These two examples underscore the importance of investigating broadly the immune response to infection to get a complete overview of the impact of IRF3 deficiency.

Human in vivo investigations were based on patients with severe viral disease and found to carry mutations in the IRF3 gene. For example, two dominant mutations were found in patients with herpes simplex encephalitis (HSE). The c.854G > A (R285Q) mutation causes IRF3 to fail to be phosphorylated, to undergo dimerization, and therefore to activate transcription, resulting in impaired IFN-I responses in peripheral blood mononuclear cells (PBMC) infected with HSV-1 [50]. The c.829G > A (A277T) was also found to inhibit induction of IFNB1 and CXCL10 expression in HSV1-infected PBMC [52]. A non-coding IRF3 variant (rs2304204; -925A > G) was associated with liver cirrhosis induced by hepatitis C virus in Egyptian patients, with the heterozygous A/G genotype being protective [55]. This polymorphism was not found to be associated with chronic hepatitis B virus infection in a cohort of Chinese patients [56].

A dominant c.1576C > T variant in the 3'UTR of IRF3 was found by whole exome sequencing in a patient with severe IAV disease. PBMC showed normal mRNA expression but reduced amounts of IRF3 protein, suggesting that the variant could affect IRF3 mRNA translation into protein. Upon infection with IAV, PBMC exhibited impaired and almost abolished induction of IFN-I and IFN-λ compared with controls [51]. Similarly, two dominant mutations (p.Glu49del/WT and p.Asn146Lys/WT) were found in patients with severe COVID19, which importance was confirmed by decreased IFNB1 induction in transfected HEK293T cells [53]. Lastly, in the case of WNV infection, the alternative C allele for the rs2304207 SNP in intron 2 of IRF3 was overrepresented in symptomatic cases compared with asymptomatic cases [54]. In conclusion, unlike mice in which Irf3 variants were shown to have a variable impact depending on the virus tested, human studies on IRF3 have been restricted to cases where deficiencies result in severe disease following viral infection.