Characterization of time-dependent IFNα and IFNγ responses
To obtain further insight into the overlapping IFN-I and IFN-II activated transcriptional responses in Huh7.5 WT cells, we first characterized the IFNα and IFNγ induced expression and phosphorylation of ISGF3 and GAF components i.e. STAT1, STAT2 and IRF9 and IRF1 (Fig 1A). IFNα treatment for different time-points (0, 0.5, 1, 2, 4, 8, 24, 36, 48, 72 h) resulted in a transient increase in phosphorylation of STAT1 and STAT2 between 0.5 and 4 h, after which it rapidly diminished to lower but still detectable levels even after 72h. IRF1 expression clearly correlated with the transient phosphorylation pattern of STAT1 and STAT2, with a maximum expression at 2h followed by a decrease in time. In contrast, expression of STAT1 and STAT2, as well as IRF9, exhibited a prolonged character and increased in expression until 72h after treatment. Compared to IFNα, time-dependent IFNγ treatment resulted in a similar, but more prolonged behavior of STAT1 phosphorylation, but not STAT2, and accumulation of IRF1, STAT1, STAT2 and IRF9 (Fig 1A).
This marks the positive feedback regulation of the ISGF3 and GAF components observed in response to IFNα as well as IFNγ.
The known GAF-target and GAS-containing genes ICAM1 and IRF1 showed an early and transient response to both types of interferon (Fig 1B), with a stronger and more delayed effect upon IFNγ stimulation. This correlated with the phosphorylation pattern of STAT1 and/or STAT2 depending on the type of IFN (Fig 1A). The expression of the known ISGF3 and IRF1-target and ISRE-containing genes MX1 and OAS2, on the other hand, followed a more delayed and prolonged pattern, with a much stronger effect upon IFNα treatment (Fig 1B). This followed the presence of phosphorylated STAT1 and/or STAT2 in combination with the increased expression of IRF9 and IRF1 induced by the different IFNs (Fig 1A). The Composite (GAS+ISRE-containing) genes APOL6 and DTX3L, on the other hand, exhibited a more intermediate response to both types of interferon, with a comparable efficacy (Fig 1B). This could point to the combined involvement of both GAS and ISRE sites with GAF, ISGF3 and IRF1 complexes, depending on time and type of IFN.
GAS, ISRE and composite genes are commonly involved in IFNα and IFNγ mediated transcriptional responses
To obtain detailed insight into the time-dependent IFN-I and IFN-II activated transcriptional responses and chromatin interactions, we first performed RNAseq on RNA isolated from Huh7.5 WT cells treated with IFNα or IFNγ for 0-72h. Using differential expression analysis (DESeq) we identified 901 IFNα and 451 IFNγ induced genes (Fig 1C), of which 319 genes that were commonly up-regulated (Fig 1C). In general, the potency of transcriptional responses to IFNα was higher in comparison to IFNγ, whereas a more transient expression profile could be observed for IFNα up-regulated genes over time (Fig 1D).
Interestingly, heatmaps presenting the expression pattern of IFNα and IFNγ upregulated genes in time, identified different clusters and further illustrated important overlap between transcriptional responses to both types of IFN. Closer inspection of a pre-selection of these different types of ISGs identified analogical clusters in IFNα- and IFNγ-dependent heatmaps (Fig 1E). Moreover, it confirmed the early and transient nature of the expression of GAS-containing genes (IRF1, ICAM1, CXCL2) as compared to the more delayed and prolonged pattern of ISRE-containing genes (ISG15, MX1, OAS2) in response to both types of IFN. The composite (GAS+ISRE-containing) genes APOL6 and DTX3L, PARP14, on the other hand, exhibited a more intermediate as well as transient response to both types of IFN (Fig 1E). Moreover, GO analysis of IFNα and IFNγ up-regulated genes revealed significant enrichment in biological terms connected to, innate, adaptive and humoral immune response, response to stress, cytokine production or immune response-activating signal transduction. They all reflect anti-viral and pro-inflammatory biological functions and also pointed to the functional overlap between IFNα and IFNγ responses (Fig 1F).
Next, we characterized the genome-wide binding of the GAF and ISGF3 components and IRF1 to the regulatory regions of IFNα and IFNγ up-regulated genes. Thus, we performed ChIP-seq on chromatin from Huh7.5 cells exposed to IFNα (0, 0.5, 2, 8, 24 and 72h; (pSTAT1, pSTAT2, IRF9 and IRF1) and IFNγ (0, 0.5, 4, 24 and 72h; pSTAT1, IRF9 and IRF1). Because of significant lower ChIP quality results with the IRF9 antibody, the number of identified IRF9 peaks was noticeably lower than compared to other antibodies and also associated with lower scores. Under these conditions, the peak score distribution followed a transient pattern for pSTAT1, pSTAT2, IRF1 and IRF9 in response to IFNα, being the highest for pSTAT1, pSTAT2 and IRF1 at 2h and for IRF9 at 8h (Fig 1G). Likewise, exposure to IFNγ increased the genome-wide number of pSTAT1, IRF9 and IRF1 binding peaks/sites, however with a more prolonged character. pSTAT1 displayed the strongest signal at 4h, whereas peak scores for IRF9 and IRF1 were maximum at 24h. Opposite to IRF9, which peak score values were clearly lower than after IFNα stimulation, a higher number of annotated peaks for IRF1 could be observed in comparison to IFNα (Fig 1G). Although a more transient binding pattern was detected for pSTAT1 and/or pSTAT2, IRF1 and IRF9 in response to IFNα as compared to IFNγ, recruitment was still clearly detectable after 72h (Fig 1G). Strikingly, only for IRF1 high peak scores could already be identified in the absence of IFN (Fig 1G).
Further analysis indicated that the pSTAT1, pSTAT2, IRF9 and IRF1 binding peaks were located predominantly in introns and promoters of IFNα- and IFNγ-responsive genes (Fig S 1A, B). These regions corresponded to the presence of individual GAS sites, commonly bound by pSTAT1 and pSTAT2 in response to IFNα, and pSTAT1 after IFNγ treatment, or ISRE sites, recruiting pSTAT1, pSTAT2, IRF9 and IRF1 after IFNα and pSTAT1, IRF9 and IRF1 after IFNγ treatment (Fig S 1C, D; Fig S 2 and 3). In addition, binding regions also contained GAS+ISRE composite sites, which showed similar binding characteristics as ISRE genes (Fig S 1C, D). This offers clear evidence for the existence of a GAS+ISRE composite gene group, which together with GAS and ISRE genes are commonly involved in IFNα and IFNγ mediated transcriptional responses.
Time-dependent IFNα and IFNγ transcriptional responses depend on differential binding of GAF, ISGF3 and IRF1 complexes to GAS, ISRE and composite genes
Subsequently, we performed an integrative analysis of our RNAseq-ChIPseq data. By concentrating on the promoter/5’UTR regions with annotated GAS and/or ISRE motifs, our multi-omics data integration identified 319 IFNα-responsive genes that bound pSTAT1, pSTAT2, IRF9 and/or IRF1 at least at one time-point. Likewise, 286 IFNγ-inducible genes were selected that bound pSTAT1, IRF9 and/or IRF1 (Fig S 4).
Comparing their expression profiles, 3 important clusters could be recognized in response to IFNα and IFNγ that distinguished early (maximum 2-4h), intermediate (max. 4-8h) and late (max. >8h) profiles (Fig 2A). Interestingly, for both IFNα and IFNγ responses GAS-containing genes were predominantly present in the early cluster, either as a solitary element or as GAS+ISRE composite site. In contrast, ISRE-containing genes formed the majority in the intermediate and late clusters, mostly as single motif sites or as composite sites (Fig 2A). This pointed to a mechanistic overlap and correlated with an early role of GAS-targeting GAF complexes and the later importance of ISGF3 and/or IRF1.
Comparative analysis accordingly identified 211 IFNα and IFNγ commonly integrated genes (Fig S 4), including 86 GAS, 82 ISRE and 43 composite genes. Comparing recruitment of pSTAT1, pSTAT2, IRF9 and/or IRF1 to these 211 common genes revealed different binding characteristics (Fig 2B and C). For example, GAS genes bound pSTAT1 alone or pSTAT1 together with pSTAT2 in response to IFNα, which coincided with individual or combined roles of GAF and GAF-like complexes (Fig 2B). Upon IFNγ treatment only pSTAT1 was recruited, which pointed to the sole role of GAF (Fig 2C). In contrast, binding characteristics to ISRE genes in response to IFNα, involved collective recruitment of all ISGF3 components (pSTAT1, pSTAT2, IRF9) with or without IRF1, or only IRF1 (Fig 2B). IFNγ predominantly directed IRF1 binding to these genes, but less frequently also IRF9 and pSTAT1 (Fig 2C). Interestingly, Composite genes exhibited combined features of GAS and ISRE-containing genes, with the collective recruitment of pSTAT1, pSTAT2, IRF9 and IRF1 (and less frequently only ISGF3 components or only IRF1) after IFNα stimulation (Fig 2B) and pSTAT1+IRF1 and less frequently IRF9 after IFNγ (Fig 2C). This could point to the involvement of GAF and/or GAF-like complexes in collaboration with ISGF3 and/or IRF1.
We also performed a cluster analysis of commonly IFNα and IFNγ integrated genes based on their time-dependent recruitment of pSTAT1 and pSTAT2 upon IFNα treatment and pSTAT1 and IRF1 after IFNγ treatment (Fig 3A). Interestingly, the binding profile of GAS genes in response to both IFNα and IFNγ displayed a clear early character, corresponding with early expression. In comparison, the binding profile of Composite and ISRE genes was delayed and prolonged, which correlated with a more intermediate and later expression pattern to both types of IFN (Fig 3A)
Finally, enrichment analysis of these groups revealed significant enrichment in biological terms connected to defense response to virus and IFN and cytokine signaling (Fig 3B). Particularly, GAS genes reflected more specific biological terms connected to complement activation and blood coagulation or establishment of endothelial barrier, whereas composite genes were explicitly linked to tumour necrosis factor-mediated signaling or further positive amplifying the cytokine-dependent signal. Finally, ISRE genes were specifically associated with antigen processing and presentation, T cell receptor signaling and IFN-I signaling.
STAT1, STAT2 and IRF9 control onset and progression of IFN responses
To obtain further insight into the role of STAT1, STAT2, IRF9 and IRF1 in IFN-I and IFN-II activated transcriptional responses, we generated STAT1-, STAT2-, IRF9-, IRF1- and IRF9/IRF1-mutant Huh7.5 cells (Fig S 5). As compared to wt cells (Fig S 5A), knocking out STAT1 dramatically delayed and prolonged the phosphorylation pattern of STAT2 in response to IFNα, which correlated with the prolonged expression pattern of STAT2, IRF1 and IRF9 (Fig S 5B). The response to IFNγ in these cells was completely abrogated, as marked by the absence of STAT2, IRF1 and IRF9 expression (Fig S 5B). Knocking out STAT2 resulted in a significant decrease in STAT1 phosphorylation levels, displaying a bi-phasic character in response to IFNα and a clear transient pattern after IFNγ treatment. IRF1 expression clearly correlated with this phosphorylation pattern of STAT1, with a more prolonged expression after IFNα treatment as compared to early and transient in response to IFNγ. In contrast, the expression of STAT1 and IRF9 under these conditions exhibited a prolonged character and increased even after long-term treatment, similar to wt cells (Fig S 5C). Knocking out IRF9 predominantly effected the IFNα response, with a weaker and more transient phosphorylation pattern of STAT1 and STAT2 slowly decreasing over time until 72h. IRF1 expression clearly correlated with this phosphorylation pattern of STAT1 and STAT2, being transient but more prolonged as compared to wt cells. In contrast, the expression of STAT1 and STAT2 under these conditions exhibited a prolonged character and increased even after long-term treatment, similar to wt cells. The response to IFNγ in these cells was similar to wt cells, displaying a transient phosphorylation of STAT1 reflecting the increase in IRF1 expression. Expression of STAT1 and STAT2 exhibited a prolonged character and increased even after long-term treatment (Fig S 5D). IRF1KO cells responded opposite to IRF9KO cells, with a transient phosphorylation pattern for STAT1 and STAT2 after IFNα treatment (that mirrored wt cells), and prolonged phosphorylation of STAT1 after IFNγ stimulation. In these cells, both types of IFN induced a similar expression pattern of STAT1, STAT2 and IRF9, however being more prolonged as observed in IRF9KO cells (Fig S 5D). Finally, in IRF1/IRF9 double KO cells, phosphorylation of STAT1 and STAT2 in response to IFNα and STAT1 in response to IFNγ exhibited an extremely prolonged pattern. More important, in IRF1/IRF9 double KO cells the IFNα- and IFNγ-mediated increase in STAT1 and STAT2 proteins was still observed (Fig S 5E).
Together, this clearly shows that in the absence of any one of the components, IFNα and IFNγ-activated responses still occur, except in IFNγ treated STAT1 mutant cells, and that the positive feedback regulation of the ISGF3 and GAF components is preserved.
STAT1 and STAT2 play a dual role in transcriptional regulation of GAS genes
A striking observation after comparing IFNα and IFNγ commonly up-regulated GAS genes revealed the differential recruitment of GAF and GAF-like complexes depending on the type of IFN (Fig 2B). Close examination of the known GAS-containing genes IRF1, ICAM1, TOP1, ANXA2R, AGT and GNB2 by ChIP-seq (Fig 4A) and of IRF1 and ICAM1 by ChIP-PCR (Fig 4B) confirmed a correlation between early and transient expression & recruitment of both pSTAT1 and pSTAT2 in response to IFNα, and only pSTAT1 upon IFNγ treatment (Figure 4A, B). Site-directed mutagenesis in combination with promoter-luciferase expression analysis was used to confirm the functionality of the proximal GAS sites in the promoters of the IRF1 and ICAM1 genes (Fig 4C). Constructs containing wild-type (WT) promoters of IRF1 and ICAM1 showed high luciferase activity in response to both IFNs. However, this effect was more prominent after treatment with IFNγ, which correlated with stronger pSTAT1 recruitment (Fig 4A) and higher expression (Fig 1B, 4B) after IFNγ stimulation. This effect was completely abolished by introducing mutations in the GAS sequences in both promoters.
To further characterize the dependence of IRF1 and ICAM1 on STAT1 and STAT2, we compared their expression in wt and STAT1-, STAT2-, IRF9-, IRF1- and IRF9/IRF1-mutant Huh7.5 cells (Fig 4C). As expected, lack of STAT1 completely abrogated the expression of IRF1 and ICAM1 upon IFNγ treatment (Fig S 6). Otherwise, the expression of IRF1 and ICAM1 clearly followed the phosphorylation pattern of STAT1 and/or STAT2 in the different cell-lines in response to the different types of IFN (Fig S 5). Remarkably, IFNα-induced expression was only partially lost in the STAT2-KO cells, which coincided with lower phosphorylated STAT1 levels (Fig S 5C). Moreover, their expression was not dependent on IRF9 or IRF1 (Fig 4B). These results were further verified by quantitative ChIP-PCR, and demonstrated to correlate with binding of both pSTAT1 and pSTAT2 in response to IFNα, and only pSTAT1 upon IFNγ treatment (Fig 4B). Surprisingly, in the STAT1KO cells a response to IFNα could still be detected, with a clear shift in expression of these genes to later time-points (Fig 4B). High STAT2 phosphorylation levels at later time points (Fig S 5B), correlated with weak, but significantly increased pSTAT2 recruitment to GAS motifs in IRF1 and ICAM1 (Fig 4B), in the absence of STAT1.
IFNα- and IFNγ-mediated induction of ISRE genes differentially depends on ISGF3 and IRF1
In contrast to GAS genes, binding characteristics of ISRE genes clearly pointed to the regulatory role of ISGF3 together with IRF1. Close examination of the known ISRE-containing genes ISG15, MX1, IFIT12 and OAS2, by ChIP-seq (Fig 4D) and of ISG15 and IFIT12 by ChIP-PCR (Fig 4E) confirmed a correlation between more prolonged expression & recruitment of all ISGF3 components (pSTAT1, pSTAT2, IRF9) and IRF1 in a time-dependent manner (even after 72h) in response to IFNα, and mainly IRF1 upon IFNγ treatment. Interestingly, IFN-dependent binding of ISGF3 components and/or IRF1 to these ISRE genes appeared later as compared to GAF and GAF-like complexes to GAS genes. In contrast to the exclusive IFN-dependent binding of ISGF3 (pSTAT1, pSTAT2 and IRF9), IRF1 exhibited binding also under basal conditions (Fig 4D), although it varied between different genes.
Site-directed mutagenesis confirmed the functionality of the proximal ISRE sites in the promoters of the ISG15 and IFIT2 genes (Fig 4F), displaying higher luciferase activity in response to IFNα, as compared to IFNγ. This correlated with the differential involvement of ISGF3 vs IRF1 (Fig 4D) and the higher gene expression (Fig 4E) after IFNα stimulation. This effect was completely abolished by introducing mutations in the ISRE sequences in both promoters (Fig 4F).
Further characterization of the expression of ISG15 and IFIT2 in wt and STAT1-, STAT2-, IRF9-, IRF1- and IRF9/IRF1-mutant Huh7.5 cells (Fig 4E), revealed a dramatic loss of IFNα-induced expression in the STAT2- and IRF9KO cells. In contrast, IFNγ-induced expression of these genes was not affected in these cell lines. As expected, lack of STAT1 completely abrogated the expression of ISG15 and IFIT2 upon IFNγ treatment (Fig S 6), whereas the IFNα response was marked by a delayed response and shift in maximum gene expression towards 72h. Although, in IRF1KO cells IFN-induced expression of ISG15 and IFIT2 was normal, in the IRF9/IRF1KO cells, both genes responded neither to IFNα nor to IFNγ (Fig 4E).
These results were further verified by quantitative ChIP-PCR, and demonstrated to correlate with binding of all ISGF3 components and IRF1 to ISRE motifs in ISG15 and IFIT2 in response to IFNα (Fig 4E), and mainly IRF1 upon IFNγ treatment in wt cells (Fig 4E). In IFNα treated STAT2- and IRF9KO cells, the remaining IRF1 binding (Fig 4E) was not able to compensate for the loss of ISGF3-dependent transcription. In contrast, IFNγ-induced binding of IRF1 in these cell lines was similar to WT cells (Fig 4E). In STAT1KO cells, high STAT2 phosphorylation and IRF9 expression levels at later times (Fig S 5B), together with increased recruitment of pSTAT2 and IRF9 to ISRE motifs in ISG15 and IFIT2 in response to IFNα (Fig 4E), is in agreement with the functional role of the STAT2/IRF9 complex under these conditions, as a replacement of ISGF3 activity in wt cells [42]. On the other hand, prolonged STAT1 phosphorylation and IRF9 expression levels in IRF1KO cells (Fig S 5E), together with the increased pSTAT1 and IRF9 recruitment (Fig 4E), could point to the possible involvement of the ISGF3-like STAT1/IRF9 complex in the potent transcriptional regulation of ISRE genes in response to IFNγ [43].
The ISRE+GAS composite site shares features of GAS and ISRE genes and acts as a molecular/regulatory switch in response to IFNα and IFNγ
As mentioned above, composite genes exhibited combined features of GAS and ISRE-containing genes, with the collective recruitment of pSTAT1, pSTAT2, IRF9 and IRF1 after IFNα stimulation and pSTAT1+IRF1 and less frequently IRF9 after IFNγ (Fig 2B).
By more detailed analysis of the IFNα and IFNγ commonly up-regulated genes, promoters of a number of pre-selected composite genes (Fig 5A; exemplified by APOL6, PARP14, DTX3L, TRIM69, and UBE2L6) displayed a similar time-dependent binding pattern of pSTAT1, pSTAT2, IRF9 and IRF1 (except for TRIM69) in IFNα treated cells. Likewise, binding of pSTAT1, IRF9 and IRF1 was observed, except TRIM69 (no IRF1 binding was visible), in response to IFNγ (Fig 5A). This corresponded with the presence of an ISRE and a GAS element in close proximity in their promoters. Interestingly, IFNα-dependent binding of ISGF3 components to these composite genes appeared earlier (and stronger) as compared to ISRE genes, but was still clearly detectable after 72h. In contrast, IFNγ-dependent binding of GAF followed a similar pattern as compared to GAS genes. IRF1 binding patterns between ISRE and composite genes were comparable. As shown for ISRE genes, IRF1 exhibited binding to composite genes already under basal conditions (Fig 5A).
Site-directed mutagenesis in combination with promoter-luciferase expression analysis of the APOL6, UBE2L6 and TRIM69 gene promoters further highlighted the unique characteristics of composite genes (Fig 5B). For example, promoter constructs for all three genes showed high luciferase activity in response to both IFNs, which correlated with the comparative ISGF3, GAF and IRF1 binding patterns and equal expression after IFNα and IFNγ stimulation. (Fig 5C). More interestingly, ISRE mutated APOL6, UBE2L6 and TRIM69 promoters still responded to both types of IFN, where the remaining GAS site tended to respond more to IFNγ. The opposite pattern was visible in cells expressing mutant GAS constructs for APOL6 and UBE2L6, being more sensitive to IFNα. In contrast, the GAS mutated TRIM69 promoter did not significantly respond anymore to either type of IFN. Mutations introduced simultaneously in both regulatory elements resulted in the complete loss of APOL6, and TRIM69 promoter activity in response to IFNα and IFNγ (Fig 5B). These results provide further proof that both the ISRE and GAS sites present in composite genes are functionally important and can be used together or independently, and can respond to both types of IFN depending on the available components and transcription factor complexes.
In contrast to ISRE genes, composite genes sustained the ability to increase the expression in all mutant cell lines in response to both IFNs (Fig 5C). Importantly, combined recruitment of pSTAT1, pSTAT2, IRF9 and IRF1 in WT cells to all of these genes confirmed the importance of the ISRE site, and possibly the GAS site, after IFNα stimulation. However, collective pSTAT1+IRF1 and weak IRF9 binding after IFNγ, clearly pointed to the combined use of the GAS+ISRE composite site under these conditions (Fig 5C), except for TRIM69 where pSTAT1 binding (without IRF1) associated with a functional GAS site only.
As shown for the ISRE genes ISG15 and IFIT2, the IFNα response of APOL6, PARP14, DTX3L, UBE2L6 and TRIM69 in STAT1KO cells was marked by a delayed response and shift in maximum gene expression towards 72h (Fig 5C). Also, lower but intact IFNα-induced gene expression of APOL6, PARP14, UBE2L6 in the STAT2- and IRF9KO cells, in combination with pSTAT1 and IRF1 binding, pointed to a shift from an ISGF3/ISRE-dependent to a pSTAT1+IRF1/composite site-dependent mechanism (Fig 5C). For TRIM69 the absence of IRF1 binding under these conditions, pointed to a shift from an ISGF3/ISRE-dependent to a pSTAT1+pSTAT2/GAS (in case of IFNα) or pSTAT1/GAS (in case of IFNγ) site-dependent mechanism (Fig 5C). In the IRF1KO cells, intact expression of all genes in response to IFNα correlated with recruitment of all ISGF3 components to the ISRE site. The IFNγ response in these cells was marked by strong pSTAT1 and weak IRF9 recruitment (Fig 5C), which could point to the sole involvement of GAF binding to the GAS or a combined involvement of GAF and STAT1/IRF9 binding to the GAS+ISRE composite site. Likewise, active IFNα and IFNγ-mediated gene expression in IRF9/IRF1KO cells, together with pre-dominant pSTAT1 recruitment, implied a shift to the independent use of the GAS site, without the neighbouring ISRE site in response to both types of IFN. As expected, lack of STAT1 completely abrogated the expression of APOL6, PARP14, UBE2L6 and TRIM69 (Fig S 6) and binding upon IFNγ treatment (Fig 5C).
STAT1, STAT2 and IRF9 are functional composite genes that are part of a phosphorylation-dependent positive feedback-loop controlling long-term IFNα and IFNγ responses
As shown in Figure 1A, protein expression of STAT1, STAT2 and IRF9 exhibited a prolonged character and increased even after long-term treatment. This marks the positive feedback regulation of the ISGF3 and GAF components observed in response to IFNα as well as IFNγ.
Close examination of the STAT1, STAT2 and IRF9 genes by ChIP-seq (Fig 6A), revealed a similar binding pattern as seen for composite genes (Fig 5A). Indeed, binding of pSTAT1, pSTAT2, IRF9 and IRF1, was observed to the promoter of STAT1 and STAT2 in a time-dependent manner (even after 72h) in IFNα treated cells and pSTAT1 and IRF1 and weak IRF9 to these genes in response to IFNγ (Fig 6A). The same was true for IRF9, however under these conditions IRF1 binding could not be detected (Fig 6A). These binding characteristics for the STAT1, STAT2 and IRF9 genes corresponded with the presence of an ISRE and a GAS element in close proximity in the promoters of STAT2 and IRF9, and an ISRE in the proximal STAT1 promoter, combined with a distal ISRE and GAS composite site (Fig 6B).
Next, site-directed mutagenesis confirmed the functionality of the proximal ISRE and GAS sites in the promoters of the STAT2 and IRF9 genes (Fig 6C). To study the functionality of the proximal ISRE site and the distal composite site in the STAT1 gene (Fig 6B), we first compared constructs containing the proximal and distal fragments separately or in combination (Fig 6D). Clearly, both the proximal ISRE site and the distal composite site were required for an optimal response to both IFNα and IFNγ. This was confirmed by subsequent mutation analysis of individual or combined ISRE and GAS sites. Mutations introduced simultaneously in all three regulatory elements resulted in the complete loss of STAT1 promoter activity in response to IFNα and IFNγ (Fig 6D).
Remarkably, IFN-induced expression of STAT1, STAT2 and IRF9 was present in all mutant cell-lines (except in IFNγ treated STAT1KO cells: Fig S 5 and 6) (Fig 6E), which resembles the characteristics of composite-containing genes. More important, it marked the restoration of the positive feedback regulation loop of ISGF3 and GAF components under these conditions. Subsequent quantitative ChIP-PCR, further confirmed these observations (Fig 6E), with combined recruitment of pSTAT1, pSTAT2 and IRF9 with (STAT1prox, STAT1dist and STAT2) or without (IRF9) IRF1 in WT cells after IFNα stimulation. Likewise, collective pSTAT1+IRF1 and weak IRF9 binding could be observed after IFNγ (Fig 6E), except for IRF9 where only pSTAT1 binding (without IRF1) was detected. Also, the IFNa response of STAT1, STAT2 and IRF9 in STAT1KO cells was marked by a delayed response and shift in maximum gene expression towards 72h (Fig 6E). Moreover, lower but intact IFNa-induced gene expression of STAT1 and STAT2 in the STAT2- and IRF9KO cells, in combination with pSTAT1 (and sometimes pSTAT2) and IRF1 binding, pointed to a shift from an ISGF3/ISRE-dependent to a pSTAT1+IRF1/composite site-dependent mechanism. For IRF9 the absence of IRF1 binding under these conditions, pointed to a shift from an ISGF3/ISRE-dependent to a pSTAT1+pSTAT2/GAS (in case of IFNa) or pSTAT1/GAS (in case of IFNg) site-dependent mechanism. In the IRF1KO cells, intact expression of all three genes in response to IFNa correlated with recruitment of all ISGF3 components to the ISRE site. The IFNg response in these cells was marked by strong pSTAT1 and weak IRF9 recruitment (Fig 6E), which could point to the sole involvement of GAF binding to the GAS or a combined involvement of GAF and STAT1/IRF9 binding to the GAS+ISRE composite site. Finally, active IFNa and IFNg-mediated gene expression in IRF9/IRF1KO cells, together with pre-dominant pSTAT1 recruitment, implied a shift to the independent use of the GAS site, without the neighboring ISRE site in response to both types of IFN (Fig 6E).