**4. Discussion**

Previous studies have described that IFNβ and TNF synergize to elicit a specific delayed transcriptional program that differs from the one induced by either cytokine alone [13,36]. The mechanisms underlying the transcriptional induction of genes specifically regulated by IFNβ and TNF remain poorly defined. The present study was specifically designed to document the functional relevance of a previously observed delayed gene expression induced by IFNβ in the presence of TNF in the absence of STAT1 [14] and to document the role of STAT2 and IRF9 in this response.

The observation that STAT2 and IRF9 activation in response to IFNβ + TNF is reduced in STAT1-deficient U3A cells compared to the wild-type 2ftGH parental cells and that IFNβ + TNF-mediated induction of the STAT2- and IRF9-dependent *CXCL10* promoter exhibits partial dependence on STAT1 support a model in which a canonical ISGF3 pathway is engaged downstream of the costimulation. Importantly, it also implied the existence of a STAT2- and/or IRF9-dependent transcriptional response occurring in the absence of STAT1. The human STAT1-deficient U3A cell model offered a unique opportunity to specifically pinpoint this STAT1-independent response. In this model, genome wide RNA sequencing highlighted that the transcriptional program induced by IFNβ + TNF in the absence of STAT1 encompasses a wide range of immunoregulatory and antiviral functions. The functional relevance of this response was confirmed by the observation that the treatment with IFNβ + TNF induced an antiviral state capable of restricting VSV replication in the absence of STAT1. This points to a significant role of the STAT1-independent pathway in the establishment of the antiviral state induced by the synergistic action of IFNβ and TNF that enhances the restriction of VSV (Figure 4D and [12]), Myxoma virus [13], and paramyxoviruses [14]. Although previous reports have shown that type I IFNs, mostly IFN<sup>α</sup>, alone can trigger STAT1-independent responses [8], we neither observed

establishment of an IFNβ-induced antiviral state against VSV, nor activation of the *CXCL10* promoter in the absence of STAT1 in our model (Figures 4D and 6).

We previously reported that IFNβ + TNF induces the *DUOX2* gene via a STAT2- and IRF9-dependent pathway in the absence of STAT1 [14]. To what extent this pathway contributes to the STAT1-independent transcriptional response elicited by IFNβ + TNF remained to be addressed. Here, we demonstrate that IFNβ + TNFα-induced DEGs segregate into seven categories that reflect distinct contributions of STAT2 and/or IRF9, thereby highlighting an unexpected heterogeneity of the STAT1-independent pathways engaged downstream of IFNβ + TNF. Importantly, only one anecdotic gene was found in categories implying inverse regulation by STAT2 and IRF9 (categories H and I) pointing to convergen<sup>t</sup> functions of STAT2 and IRF9 when both are engaged in gene regulation. We can rule out that these distinct regulation mechanisms reflect specific induction profiles by IFNβ + TNF as *CXCL10*, *IL33*, *CCL20,* and *ISG20* all exhibit synergistic induction by IFNβ + TNF, but are differentially regulated by STAT2 and/or IRF9; while *CXCL10* is dependent on STAT2 and IRF9, *IL33* is independent on STAT2 and IRF9, and *CCL20* and *ISG20* are STAT2-independent but IRF9-dependent (Figure 5; Supplemental Table S1). Consistent with our previous observation [14], we found several STAT1-independent genes positively regulated by STAT2 and IRF9 (Category A). DEGs in this category encompass most of the functions induced in response to IFNβ + TNF, with the notable exception of cell cycle and growth arrest and inflammasome and receptor signaling functions. Genes negatively regulated by STAT2 and IRF9 were also identified (Category B). Formation of an alternative STAT2/IRF9-containing complex mediating gene expression in the absence of STAT1 [37–41] has been reported, but with limited DNA-binding affinity for the typical ISRE sequence [37]. The existence of a STAT2/IRF9 complex is also supported by our recent observation of a high affinity of IRF9 for STAT2 with an equilibrium dissociation constant (Kd) of 10 nM [42]. A recent report of experiments, performed in murine bone marrow-derived macrophages proposes a model in which murine STAT2/IRF9 complex drives basal expression of ISGs, while IFNβ-inducible expression of ISGs depends on a switch to the ISGF3 complex [43]. This differs from our results as silencing of either STAT2 or IRF9 did not alter basal gene expression (Supplemental Figure S1). Further analysis of the *CXCL10* promoter demonstrates a restricted usage of ISRE sites by the STAT2/IRF9 pathway compared to the ISGF3 pathway. Further large-scale studies will be needed to identify the parameters allowing binding of ISGF3, but not STAT2/IRF9, to specific ISRE sequences upon IFNβ + TNF.

The observation that some IFNβ + TNF-induced genes were solely dependent on STAT2 (either positively or negatively) but not on IRF9, (Categories D and E) is a rare genome wide demonstration of gene regulation by STAT2 independently of STAT1 and IRF9. Previous reports have identified ISGF3-independent, STAT2-dependent genes but the association with IRF9 was not formally excluded [44–47]. STAT2 was shown to associate with STAT3 and STAT6, but it is not clear whether IRF9 is also part of these alternative complexes [44,46]. Transcriptional module analyses demonstrated that the functional distribution of genes negatively regulated by STAT2 is very limited compared to other categories; only a virus-sensing module was enriched in this category. In contrary, IRF9-independent genes positively regulated by STAT2 mediate broader antiviral and immunoregulatory functions.

ISGF3-independent functions of IRF9 have been proposed based on the study of IRF9 deficiencies [11,48]. However, IRF9 target genes in these contexts have been barely documented. Intriguingly, Li et al. [49] studied IFNα-induced genes and their dependency on the ISGF3 subunits. While they confirmed previous studies showing that IFNα can trigger a delayed and sustained ISG response via an ISGF3-independent pathway, it is very striking that they did not find STAT1- and STAT2-independent but IRF9-dependent genes. All identified IRF9-dependent genes were either STAT2- or STAT1-dependent. This result greatly differs with our study. Here, we found several IFNβ + TNF-induced DEGs independent of STAT1 and STAT2, but positively or negatively regulated by IRF9 (Categories F and G). Typically, IRF9 is considered a positive regulator of gene transcription. However, our findings are consistent with recent reports documenting the role of IRF9 in the negative regulation of the TRIF/NF-κB transcriptional response [50] or the expression of SIRT1 in acute myeloid

leukemia cells [51]. The molecular mechanisms underlying gene regulation by IRF9 without association with either STAT1 or STAT2 remain to be elucidated. To the best of our knowledge, no alternative IRF9-containing complex has ye<sup>t</sup> been described.

Our analysis showed that a large number of genes were induced by IFNβ + TNF independently of STAT2 and IRF9 (Category C). All transcriptional modules were enriched in this category pointing to a major role of this pathway in the establishment of a host defense and immunoregulatory response. The STAT2 and IRF9 independent genes does not solely reflects induction by TNF alone. For instance, *APOBEC3G* that is amongs<sup>t</sup> the STAT2- and IRF9- independent genes is not induced by TNF alone (Figure 2A). While NF-κB, a downstream effector of the TNF receptor, is an obvious candidate for the regulation of these DEGs, this might fall short in explaining the synergistic action of IFNβ + TNF as we did not observe enhanced NF-κB activation compared to TNF alone [14]. Alternatively, the potential role of AP-1 is supported by the finding that the AP-1 transcription network module is enriched amongs<sup>t</sup> IFNβ + TNF-induced DEGs. However, this module is not restricted to genes regulated independently of STAT2 and IRF9. It is also worth noting that two modules of IRF2-target genes were enriched, although again not specifically in the STAT2- and IRF9-independent category. A similar crosstalk was reported between IFNα and TNF in macrophages resulting in increased colocalized recruitment of IRF1 and p65 to the promoter of a subset of genes [52]. However, while IRF1 was found synergistically induced by IFNβ + TNF at early stages (Figure 1), we did not observe significant induction of IRF1 in the absence of STAT1 by RNASeq (Supplemental Table S1) or qRT-PCR (data not shown), suggesting that IRF1 is unlikely to be involved in our system. Further studies will be required to uncover these STAT2- and IRF9-independent pathways.

This study provides novel insight into the molecular pathways leading to delayed antiviral and immunoregulatory gene expression in conditions where elevated levels of both IFNβ and TNF are present. Altogether our results demonstrate that in addition to the engagemen<sup>t</sup> of an ISGF3-dependent canonical response, a broad transcriptional program is elicited independently of STAT1, and support a model in which STAT2 and IRF9 contribute to the regulation of this response through non-canonical parallel pathways involving their concerted or independent action (Figure 7).

**Figure 7.** Role of distinct STAT2 and/or IRF9-dependent pathways in the regulation of distinct subset of antiviral and immunoregulatory genes in response to IFNβ and TNF. Our data supports a model in which multiple pathways participate to the synergistic action of IFNβ + TNF. While the STAT1-dependent pathway, likely ISFG3, is engaged downstream of IFNβ and TNF, STAT1-independent pathways are also involved in the control of the delayed gene expression. STAT2 and IRF9 act not only in a concerted fashion, likely as a complex, but also independently. IRF9 is known to act as the DNA-binding subunit of the ISGF3 complex and therefore likely mediates binding of STAT2/IRF9 complexes and of alternative complexes devoid of STAT2. The mechanisms of STAT2-dependent regulation of gene expression remains to be characterized.

Consistent with accumulating evidence [8], these distinct STAT2 and IRF9 actions most likely result from the formation of specific complexes that coexist with ISGF3 upon IFNβ and TNF stimulation. Studies are underway to biochemically solve the complexity of the dynamic and specific mechanisms of activation of the alternative STAT2 and/or IRF9-containing complexes in a wild-type cell context to further characterize the transcriptional response induced by IFNβ and TNF.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2073-4409/8/8/919/s1, Table S1: List of I+T\_ DEG; Table S2: List of GO\_I+T \_DEG; Table S3: Module analysis\_DEG; Table S4: MatetMeth\_qPCR; Table S5: list\_all\_modules; Figure S1: qRT-PCR analysis of STAT2 and IRF9 regulated DEGs.

**Author Contributions:** Conceptualization, M.K.M. and N.G.; formal analysis, M.K.M., N.G., P.D., and A.W.R.S.; investigation, M.K.M., A.F., E.C., M.K., A.N.H., D.M.K., D.I.H., S.L.C.-O., and E.M.; writing—original draft preparation, N.G.; writing—review and editing, P.D., D.M.K. and D.I.H.

**Funding:** This research was funded by grants from NATURAL SCIENCES AND ENGINEERING RESEARCH COUNCIL OF CANADA (NSERC RGPIN/355306-2012 and RGPIN/2018-04279) and Research Chair in signaling in virus infection and oncogenesis from the Université de Montréal to N.G. A.W.R.S. acknowledges funds from Université de Montréal, NSERC and Merck Foundation. S.L.C.-O. was the recipient of a MITACS Globalinks studentship. N.G. was recipient of a Tier II Canada Research Chair.

**Acknowledgments:** We are very thankful to D. Proud (University of Calgary, Canada), G. Stark (Cleveland clinic, USA), and J. Bell (University of Ottawa, Canada) for sharing reagents used in this study. RNASeq analyses were performed at McGill University and Génome Québec Innovation Centre. Figure 6 was generated using biorender.com.

**Conflicts of Interest:** The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.
