Constitutive IFNα/β signaling maintains expression of signaling intermediaries for efficient cytokine responses

ABSTRACT

Interferons (IFNs) are a family of immunoregulatory cytokines with important roles in anti-viral and anti-tumor responses. Type I and II IFNs bind distinct receptors and are associated with different stages of the immune response. There is however, considerable crosstalk between these two cytokines with enhancement of IFNγ responses following IFNα/β priming and loss of IFNα/β receptor (IFNAR) resulting in diminished IFNγ responses. In this study, we sought to define the mechanism of crosstalk between the type I and II IFNs. Our previous reports demonstrated reduced expression of the canonically activated transcription factor signal transducer and activator of transcription (STAT)1, in cells lacking the IFNAR α chain (IFNAR1). Therefore, we used microarray analysis to determine whether reconstitution of STAT1 in IFNAR1-deficient cells was sufficient to restore IFNγ responses. We identified several biological pathways, including the MHC class I antigen presentation pathway, in which STAT1 reconstitution was able to significantly rescue IFNγ-mediated gene regulation in Ifnar1^−/− cells. Notably, we also found that in addition to low basal expression of STAT1, cells lacking the IFNAR1 also had aberrant expression of multiple other transcription factors and signaling intermediaries. The studies described herein demonstrate that basal and regulated expression of signaling intermediaries is a mechanism for crosstalk between cytokines including type I and II IFNs.

KEYWORDS:

• cross talk
• gene expression
• interferon
• priming
• signal transduction
• STAT1
• transcription factors

INTRODUCTION

The exposure of cells to cytokines, chemokines and other extracellular signals during pathogenic infection, injury and neoplasia allows for potent, and often synergistic, host immune responses.^1,2 IFNs, a family of immunoregulatory cytokines with important roles in anti-pathogen and anti-tumor responses, can synergise with other stimuli, such as tumor necrosis factor (TNF) α and interleukin (IL) 1α for more potent immune responses.^3-10 Upon IFN binding, receptor-associated janus activated kinases (JAKs) are phosphorylated leading to activation of latent signal transducer and activator of transcription (STAT) molecules. Pivotal to IFN signaling is the transcription factor STAT1, the loss of which severely impairs IFN-dependent cellular responses.^11,12 The canonical, JAK/STAT1-dependent IFNγ signaling pathway has been extensively studied, however more recently alternate signaling pathways triggered by IFNγ such as mitogen activated protein kinase (MAPK) and nuclear factor κB (NFκB) pathways have been identified.^13-16

Although specific IFN family members bind to selective receptors and activate distinct signaling pathways, significant crosstalk between these pathways is evident.^1-3,6,17,18 Pre-treatment of cells with low levels of type I IFNs (IFNα/β) leads to more potent response to type II IFN (IFNγ) and interestingly loss of the type I IFN receptor (IFNAR) results in diminished type II IFN responses.^19,20 Although type I IFNs are secreted in abundance in response to viral infection, uninfected cells also constitutively secrete low concentrations of IFNα/β.^21-24 Studies by our laboratory and others, show that cells unable to produce or respond to constitutive IFNα/β have attenuated STAT1 expression. This led to a model of cytokine crosstalk which postulates that priming levels of IFNβ regulate STAT1 expression, which is important for mediating IFNγ responses.^23,25,26 Moreover, we have demonstrated that reconstitution of STAT1 into Ifnar1^−/− cells could restore IFNγ-induced STAT1 DNA binding and partially rescued IFNγ-dependent anti-viral responses.²³ In this study we used global gene expression analysis to examine the contribution of STAT1 to the crosstalk between tonic IFNα/β and IFNγ. Our data revealed that STAT1 alone regulates only a small proportion of IFN crosstalk. Moreover, loss of tonic IFNα/β signaling results in aberrant expression of numerous transcription factors and intracellular signaling molecules, thus tonic IFNα/β signaling regulates the expression of multiple signaling intermediaries involved in cytokine signaling.

RESULTS

STAT1 rescue of IFNγ responsiveness in Ifnar1^−/− cells is enhanced at earlier time points

STAT1 is the canonical transcription factor activated in response to IFNγ and we have previously found that its expression is attenuated in Ifnar1^−/− cells.²³ To determine the extent to which STAT1 reconstitution alone could rescue molecular responses to IFNγ in Ifnar1^−/− cells, microarray-based gene expression analysis was performed over a 6 hr time course. Prior to microarray analysis STAT1 expression in wild type, Ifnar1^−/−, empty vector-transduced (Ifnar1^−/−MSCV) and HA-tagged STAT1α-transduced (Ifnar1^−/−HA-STAT1) Ifnar1^−/− mouse embryonic fibroblasts (MEFs) was confirmed by western blotting (Fig. 1A). Probes in which IFNγ induced a significant (p < 0.05) and greater than 1.2-fold increase or decrease in expression were considered to be regulated by IFNγ. Of the 362 probes which had induced expression in wild type cells following 1 hr of IFNγ treatment, 109 probes (30.1%) were also induced in Ifnar1^−/− and/or Ifnar1^−/−MSCV cells and Ifnar1^−/−HA-STAT1 cells (Fig. 1B). In contrast to induced gene expression, only 5 of the 440 probes which showed significant repression in response 1 hr of IFNγ treatment in wild type cells, were also repressed in wild type, Ifnar1^−/− and/or Ifnar1^−/−MSCV cells and Ifnar1^−/−HA-STAT1 cells (Fig. 1B). Similar patterns of gene expression were seen at the later time points with 18.2% and 40.3% of probes induced and 3.2% and 2.5% of probes repressed in wild type, Ifnar1^−/− and/or Ifnar1^−/−MSCV cells and Ifnar1^−/−HA-STAT1 cells in response to 3 and 6 hrs of IFNγ treatment respectively (Fig. 1C-D).

Figure 1 Reconstitution of STAT1 in Ifnar1^−/− MEFs partially rescues IFNγ responses. (A) Protein lysates from wild type, Ifnar1^−/−, Ifnar1^−/−MSCV and Ifnar1^−/−HA-STAT1 129/C3H MEFs were separated by SDS-PAGE and probed with antibodies specific for STAT1 and actin as a loading control. (B-G) wild type, Ifnar1^−/−, Ifnar1^−/−MSCV and Ifnar1^−/−HA-STAT1 129/C3H MEFs were treated with 100 U/ml IFNγ for 1, 3 and 6 hrs and mRNA expression was assessed by Affymetrix microarray (n = 3). (B-D) Venn diagram of the number of genes that are up-regulated (↑ green) or down-regulated (↓ red) relative to untreated (p < 0.05, >1.2-fold) following (B) 1 hr, (C) 3 hrs or (D) 6 hrs of IFNγ treatment and whether they are specific or common to each cell type. (E-G) Left panel. Heat maps of probes with attenuated IFNγ-dependent gene induction in Ifnar1^−/− MEFs relative to wild type MEFs (p > 0.05, <1.2-fold change relative to untreated or >1.2-fold attenuation in change compared to wild type). Right panel. Heat maps of probes with attenuated (as defined above) IFNγ-dependent gene induction in Ifnar1^−/− MEFs and Ifnar1^−/−MSCV MEFs relative to wild type MEFs and rescued (p < 0.05 and >1.2-fold change relative to untreated and <1.2-fold attenuation in change compared to wild type) in Ifnar1^−/−HA-STAT1 MEFs. The 10 attenuated (left panel) and rescued (right panel) probes with the greatest fold change in wild type MEFs following (E) 1 hr, (F) 3 hrs and (G) 6 hrs of IFNγ treatment are displayed. Heatmaps were generated using Java Treeview. Total number of probes with induced (↑) and repressed (↓) expression in wild type MEFs that had attenuated (left panel) or rescued (right panel) expression are listed below heatmaps.

To determine the genes for which loss of IFNAR1 expression attenuated IFNγ responsiveness, we compared the wild type and Ifnar1^−/− microarray data sets and considered probes which lacked a significant, >1.2-fold change in expression or had >1.2-fold attenuation of gene regulation compared to wild type cells as having lost IFNγ responsiveness. In response to 1, 3 and 6 hrs of IFNγ treatment, 85.6%, 93.6% and 75.3% of probes had attenuated gene induction in Ifnar1^−/− cells compared to wild type cells respectively (Fig. 1B-D). For IFNγ-dependent gene repression 99.3%, 99.8% and 98.1% of probes had attenuated responses following 1, 3 and 6 hrs of IFNγ treatment (Fig. 1B-D). These data indicate that the vast majority of IFNγ regulated genes are affected by the absence of IFNAR1.

STAT1α reconstitution of Ifnar1^−/− cells was considered to have rescued IFNγ responses if the following criteria were met: (i) The gene had lost IFNγ responsiveness in Ifnar1^−/− and Ifnar1^−/−MSCV cells compared to wild type cells (ii) The gene had wild type-equivalent responsiveness to IFNγ in Ifnar1^−/−HA-STAT1 cells (i.e. a significant and >1.2-fold change in expression relative to untreated and <1.2-fold attenuation of gene regulation in Ifnar1^−/−HA-STAT1 cells compared to wild type cells). Of the 310 probes in which IFNγ-induced gene expression was attenuated in Ifnar1^−/− cells following 1 hr of IFNγ treatment, 43 (13.8%) probes had rescued IFNγ responsiveness in Ifnar1^−/−HA-STAT1 cells (Fig. 1E). In response to 3 hrs and 6 hrs of IFNγ treatment, induction of gene expression of 62 (8.4%) and 38 (7.4%) probes with attenuated responses in Ifnar1^−/− cells was rescued in Ifnar1^−/−HA-STAT1 cells respectively (Fig. 1F-G). In contrast to the effects on IFNγ-dependent gene induction, Stat1α reconstitution of Ifnar1^−/− cells only rescued gene repression of 0.7%, 1.0% and 0.4% of probes following 1, 3 and 6 hrs of IFNγ treatment respectively. These data indicated that the effect of STAT1 reconstitution on IFNγ-dependent gene regulation was greatest during early phase gene induction.

STAT1 rescues IFNγ-mediated responses to a select range of biological pathways in IFNAR1 cells

Ontological analysis of the microarray data was performed to identify the biological pathways rescued following IFNγ treatment of Ifnar1^−/−HA-STAT1 cells. The 2106 genes with attenuated regulation by IFNγ in Ifnar1^−/− MEFs over the 6 hr time course of IFNγ treatment were analyzed using MetaCorZe's Genego. Genes with aberrant expression following IFNγ treatment in Ifnar1^−/− MEFs were involved in a diverse range of biological pathways, with the most significantly affected including developmental, immunological response, reproduction and apoptosis pathways (Fig 2A). Surprisingly, only one pathway directly associated with STAT signaling was found among the 10 most significantly affected biological pathways. Subsequently, genes with attenuated regulation by IFNγ in Ifnar1^−/− MEFs were compared to the genes in which IFNγ responses were significantly rescued by reconstitution of STAT1 using MetaCore's Genego comparative analysis for pathway maps. This analysis revealed, unsurprisingly, that among the 10 pathways most significantly affected by both the loss of IFNAR1 and reconstitution of STAT1 cells were the IFNα/β signaling pathway and multiple JAK/STAT signaling pathways (Fig. 2A). Of the 10 most significantly affected pathways, 7 were associated with immune responses with the most significantly affected pathway being the MHC class I antigen presentation pathway (Fig. 2A-B). These data suggested that despite genes involved in a diverse array of biological responses pathways being aberrantly regulated by IFNγ in the absence of IFNAR1, the pathways in which STAT1 reconstitution of IFNAR1 deficient cells rescued IFNγ-mediated gene regulation were enriched for immune response pathways.

Figure 2 Reconstitution of STAT1α in Ifnar1^−/− MEFs preferentially rescues immune and inflammatory IFNγ responses. (A-B) Probes from figure 1.E-G listed as attenuated (A-B) or rescued (B) were analyzed using MetaCore's Genego. The 10 most significantly affected pathways are displayed. (A) Pathway enrichment analysis of attenuated genes. EMT = epithelial-to-mesenchymal transition. (B) Comparative pathway enrichment analysis of attenuated and rescued genes. (C) Heatmaps displaying fold change in expression following 1, 3 or 6 hrs IFNγ treatment of genes involved in antigen presentation via MHC class I for wild type, Ifnar1^−/−, Ifnar1^−/−MSCV and Ifnar1^−/−HA-STAT1 MEFs. Heatmaps were generated using Java Treeview.

STAT1-reconstitution of Ifnar1^−/− cells rescues IFNγ-inducible protein expression of genes involved in the MHC class I antigen presentation pathway

Several components of the MHC class I-dependent antigen presentation pathway were significantly rescued by STAT1 reconstitution of Ifnar1^−/− cells as determined by microarray analysis (Fig. 2C). We had previously found that STAT1 reconstitution of Ifnar1^−/− cells rescued IFNγ-dependent induction of RNA expression for a key molecule in antigen presentation, β 2 microgloblin (B2M).²³ Therefore the MHC class I antigen presentation pathway represented an ideal candidate to validate the attenuation of IFNγ-dependent biological responses in IFNAR1 deficient cells and its rescue by STAT1 reconstitution.

We first examined the IFNγ-dependent induction of molecules involved in the transport of peptides to the endoplasmic reticulum, transporter associated with antigen processing (Tap)1 and Tap2, in Ifnar1^−/− cells. In corroboration with the microarray data, we found reduced IFNγ-mediated protein induction of TAP1 expression in Ifnar1^−/− MEFs compared to wild type MEFs following exposure to IFNγ for 6 hrs (Fig. 3A). There was also attenuation of TAP2 induction in Ifnar1^−/− MEFs in response to 6 hrs of IFNγ treatment and a minor effect of loss of IFNAR1 for cells treated for 24 hrs (Fig. 3B). Reconstitution of STAT1 in Ifnar1^−/−HA-STAT1 cells rescued IFNγ-mediated induction of TAP1 in response to 6 hrs and TAP2 following 6 and 24 hrs of treatment ((Fig. 3A–B). Similar to the TAP molecules, induction of proteasome subunit, β type (PSMB)9, a catalytic component of the immunoproteasome, was attenuated in Ifnar1^−/− cells following 6 hrs of treatment (Fig 3A). This attenuated induction of PSMB9 was also evident in Ifnar1^−/−MSCV cells following 6 hrs of IFNγ treatment and importantly, was rescued to wild type equivalent levels in Ifnar1^−/−HA-STAT1 cells (Fig. 3C). Notably, PSMB9 expression reached wild type equivalent levels in all IFNAR1 deficient cell lines by 24 hrs of IFNγ treatment suggesting that, as seen in for multiple genes in the microarray analysis, IFNAR1 expression was required for early, but not late phases responses.

Figure 3. STAT1 reconstitution rescues IFN-induced protein expression of MHC class I antigen presentation proteins in Ifnar1^−/− MEFs. (A-C) wild type, Ifnar1, Ifnar1^−/−MSCV and Ifnar1^−/−HA-STAT1 MEFs were cultured in media alone or treated with 100 U/ml IFNγ for (A) 6 hrs and (B-C) 24 hrs. Protein lysates were separated by SDS-PAGE and probed with antibodies specific for (A) TAP1 (B) TAP2 or (C) PSMB9. Western membranes were re-probed with antibody specific for (A-B) Actin or (C) HSP70 to confirm equivalent loading of proteins. (D-E) wild type, Ifnar1, Ifnar1^−/−MSCV and Ifnar1^−/−HA-STAT1 MEFs were cultured in media alone or were treated with 5 or 10 U/ml IFNγ for 16 hrs. Cells were analyzed by flow cytometry for expression of the MHC class I molecules (D) H2-K^b and (E) H2-D^b. Data represents the IFNγ-induced change in MFI relative to untreated for 11–13 independent experiments and error bars are SEM. Significant changes in MFI relative to untreated (fold increase in MFI) were determined by t-test ^#p < 0.05, ^##p < 0.01, ^###p < 0.001. Significant differences in fold increase in MFI relative to wild type were determined by t-test *p < 0.05, **p < 0.01, ***p < 0.001.

Finally, we assessed the effects of STAT1 on the ability of IFNγ to regulate cell surface expression of the MHC class I heavy chain on Ifnar1^−/− cells. Again, in agreement with the microarray analysis, Ifnar1^−/− and Ifnar1^−/−MSCV MEFs displayed significantly attenuated IFNγ responses, with minimal changes in H2-K^b and H2D^b expression compared to wild type cells (Fig. 3D-E). In contrast, cell surface expression of H2-K^b and H2-D^b following both 5 and 10 U/ml treatments of IFNγ was, like wild type cells, significantly induced on Ifnar1^−/−HA-STAT1 cells (Fig. 3D-E). Together, these data indicated that STAT1-reconstitution was able to rescue the attenuated IFNγ-dependent regulation of mRNA and protein expression of genes involved in protein processing, peptide transport and cell surface antigen presentation in Ifnar1^−/− cells.

Constitutive IFNα/β signaling regulates the expression of multiple signaling molecules

Having found that STAT1 reconstitution of IFNAR1-deficient cells only rescued a proportion (∼15%) of IFNγ regulated genes and that these genes were enriched for involvement in immune and inflammatory responses, we investigated whether constitutive IFNα/β signaling regulated the expression of other transcription factors to modulate IFN crosstalk. We first examined expression of other members of the JAK/STAT signaling pathway known to be regulated by and involved in IFN signaling. In corroboration with our previous qRT-PCR findings, there was attenuated RNA expression of Stat1 and Stat2, but not Socs3, in unstimulated Ifnar1^−/− cells, as determined by microarray and qRT-PCR analysis (Stat1 and Socs3) (Supp. Fig. 1A, Supp. Table 1–2).²³ Similarly, we validated our previously identified decrease in basal STAT3 protein expression using primary C57BL/6 Ifnar1^−/− MEFs and confirmed that the attenuated STAT3 protein expression evident in 129/C3H MEFs lacking IFNAR1 was related to a significant decrease in RNA expression (Supp. Fig. 1A–B).²³ Given that both IFNγ and IL-6 can activate STAT3 and that, IL-6 mediated STAT3 DNA binding is attenuated in the absence of IFNAR1,²⁷ we assessed STAT3 activation following IFNγ or IL-6 treatment of Ifnar1^−/− cells. In contrast to previous findings,²⁷ we found reduced STAT3 phosphorylation following IL-6 stimulation of Ifnar1^−/− cells (Fig. 4A). Similarly, IFNγ-mediated STAT3 activation was also attenuated in IFNAR1-deficient cells (Fig. 4A). These data indicate that constitutive IFNα/β signaling regulates expression of multiple STAT transcription factors and that this may influence crosstalk with cytokines other than IFNs.

Figure 4. Aberrant expression and activation of signaling intermediaries in unstimulated Ifnar1^−/− MEFs. (A) wild type and Ifnar1^−/− MEFs were cultured in media alone or treated with 100 U/ml IFNγ or 100 ng/ml IL-6 and 100ng/ml IL-6R for 30mins. Protein lysates were separated by SDS-PAGE and probed with an antibody specific for pSTAT3. Western membrane was re-probed with an antibodies specific for Actin to confirm equivalent loading of proteins. (B) RNA was extracted from untreated wild type and Ifnar1^−/− 129/C3H MEFs and mRNA expression of c-Jun, Irf1, Myd88, Nfkbiα and Nfkbie was determined by qRT-PCR. Expression of each gene was normalized to L32 and untreated wild type MEFs. Data represents the mean of 4–6 independent experiments and error bars are SEM. Significance determined by t-test *p < 0.05, **p < 0.01, ***p < 0.001. (C-D) Protein lysates from untreated wild type and Ifnar1^−/−MEFs were separated by SDS-PAGE and probed with antibodies specific for (C) c-JUN, and (D) ERK1/2. Western membranes were re-probed with antibodies specific for HSP70 to confirm equivalent loading of proteins.

To determine the effect that loss of constitutive IFNα/β signaling has on the global transcriptional profile of resting cells, we assessed transcription factor expression in unstimulated wild type and Ifnar1^−/− MEFs by microarray analysis. 1264 genes had ≥1.2-fold (p≤0.05) differential basal expression in Ifnar1^−/− MEFs compared to wild type MEFs (Supp. Table. 1). Of the 1264 genes with differences in basal expression between wild type and Ifnar1^−/− cells, 106 were known transcription factors (Supp. Table. 2). Furthermore, qRT-PCR analysis of intracellular signaling molecules known to crosstalk with the type I IFN signaling pathway confirmed the reduced expression of genes involved in TLR and NF-κB signaling, Myd88, Nfkbia and Nfkbie (Fig. 4B) in Ifnar1^−/− cells (Fig. 4.b and Supp. Table. 2).

We also assessed RNA and protein expression of several transcription factors known to regulate IFNγ responses that were not identified by the microarray analysis as having attenuated expression in Ifnar1^−/− MEFs. We found significantly reduced RNA expression of the transcription factors c-Jun and Irf1 by qRT-PCR analysis (Fig. 4B). As shown in Figure. 4C, Ifnar1^−/− MEFs also expressed less c-Jun protein than wild type MEFs and thus the decrease in c-Jun mRNA expression observed correlated with a similar decrease in total protein expression. Similarly, although we found no reduction in extracellular signal-regulated kinase (ERK)1/2 expression by microarray analysis, basal expression of ERK1, but not ERK2 was reduced in Ifnar1^−/− MEFs (Fig. 4D). Overall, these data revealed that the absence of IFNAR1 results in reduced aberrant expression of multiple transcription factors which are involved in a range of cytokine and inflammatory signaling responses.

DISCUSSION

The canonical signal transduction pathways activated by cytokines to modulate cellular responses and gene expression have been extensively investigated. These pathways are comprised of signaling intermediaries such as kinases, G proteins, adaptor proteins and transcription factors that get turned ‘on’ and ‘off’ in the presence of a stimulus. In contrast, the physiological scenario of concomitant exposure to several cytokines and the consequences and mechanisms regulating these interactions are relatively undefined.²⁸ Cytokine crosstalk can occur at the level of the receptor, by regulation of the expression of receptor components or the formation of receptor complexes.^19,27,29-31 Crosstalk can also occur at the level of gene promoters as a result of multiple promoter binding sites or chromatin priming.^32,33 In this study, we examined the impact that the regulation of the basal levels of intracellular signaling molecules has on cytokine responses.

We had previous established that STAT1 reconstitution of Ifnar1^−/− cells rescued IFNγ-dependent STAT1 activation and DNA binding and partially rescued IFNγ-mediated protection from the cytopathic effects of ECMV.²³ Using microarray analysis to ascertain the effects of STAT1 reconstitution of IFNAR1-defient cells on a genome wide scale we found that although STAT1 is the canonical transcription factor activated by IFNγ, only up to 15.4% and 3.2% of IFNγ-dependent gene induction and repression were rescued by STAT1 respectively. Moreover, the ability of STAT1 to rescue IFNγ responses diminished over time. The relatively minor effect of STAT1 expression of IFNγ mediated gene repression was unsurprising however, as gene repression by IFNγ is purported to be mediated via STAT1 independent pathways³⁴ such as CIITA mediated histone deacetylation.^35-37 Also, as STAT1 is induced by IFNγ signaling,^38-40 it is plausible that at later time points de novo STAT1 production in Ifnar1^−/− cells in response to IFNγ treatment was able to rescue the IFNγ responsiveness of genes at later time points.

Treatment of primary human fibroblasts with both IFNβ and IFNγ synergistically induces expression of genes involved in apoptosis and inflammatory responses.⁴¹ Similarly, the biological response pathways which showed attenuated IFNγ responses in Ifnar1^−/− cells and for which STAT1 reconstitution significantly rescued IFNγ responsiveness were both enriched for immune responses, apoptosis and alternate cytokine signaling pathways. As expected, our microarray and protein analysis indicated that IFNγ-dependent regulation of the JAK/STAT signaling pathway was significantly affected by the loss of tonic IFNα/β however, interestingly, the MHC class I antigen presentation pathway was the most sensitive to loss of IFNAR1 and was also significantly rescued by STAT1 reconstitution. Expression of a number of genes involved in MHC class I antigen presentation, including Tap1, Psmb9 and MHC class I heavy chain, are known to be mediated by type I and type II IFNs in a STAT1-dependent manner.^12,42-45 It was however surprising, given that this regulation is co-operatively mediated by other factors including IRF1, IRF2 and the ISGF3 complex for which basal expression was also reduced (Fig. 4 and Supp. Table 2), that STAT1 alone was able to rescue the IFNγ-dependent regulation of these genes.

Although reconstitution of STAT1 in Ifnar1^−/− cells had a major effect on several biological pathways, ∼85% of IFNγ-mediated gene regulation was not rescued. Mediation of cytokine crosstalk by the regulating the expression of signaling intermediaries has also been suggested for STAT4,^46,47 IκB,⁴⁸ IRF9⁴⁹ and IRF2.⁵⁰ Our studies, which found that expression of over 100 transcription factors and numerous other signaling intermediaries, including Myd88, IKKγ and MAPK kinases, were decreased in cells that lacked tonic IFNα/β, suggests that rather than being regulated by a single transcription factor or signaling complex, cytokine crosstalk likely involves regulation of the expression of a multitude of signaling molecules.

This level of regulation may enhance specific biological responses when particular sets of transcription factors are involved. For example, although expression of STAT1 had a profound effect on IFNγ dependent regulation of genes involved in antigen presentation, other IFNγ-dependent responses, including regulation of genes involved in development and TLR-dependent immune responses, were not rescued by STAT1 reconstitution of Ifnar1^−/- MEFs. As RNA expression of the signaling intermediary Myd88, a fundamental component of the TLR signaling pathway,^51,52 was, like STAT1, attenuated in cells lacking constitutive IFNβ signaling, it is plausible that tonic IFNα/β augments TLR responsiveness by regulating the expression of Myd88. Moreover, as expression or activation of transcription factors involved in signaling via other cytokines known to engage in crosstalk with IFNα/β such as IL-6, IL-1 and TNFα^{6,27,28,53-55} were also revealed by our studies to be attenuated in IFNAR1-deficient cells, it suggest that tonic IFNα/β also regulates responsiveness to these cytokines by this mechanism.

Overall we have demonstrated that, in addition to maintenance of STAT1 expression, constitutive IFNα/β-signaling mediated by IFNAR1 regulates the expression of multiple transcription factors and signaling intermediaries. Moreover, based on the sub-set of biological responses pathways affected by STAT1 expression, we suggest that, the regulation of signaling molecules involved in specific biological pathways or activated by specific cytokines may be a mechanism whereby synergistic biological responses by multiple cytokines are fine-tuned.

MATERIALS AND METHODS

Mice

C57BL/6 J wild-type mice were purchased from the Walter and Eliza Hall Institute of Medical Research or bred in house at the Peter MacCallum Cancer Center. C57BL/6 IFNAR1-deficient (Ifnar1^−/−)⁵⁶ mice were bred in house at the Peter MacCallum Cancer Center. All experiments were performed in accordance with the animal ethics guidelines ascribed by the National Health and Medical Research Council of Australia. All experiments were approved by the Peter MacCallum Cancer Center Animal Ethics Committee.

Reagents and cell culture

Generation of wild type and Ifnar1^−/− 129/C3H MEFs and retroviral transduction of Ifnar1^−/− 129/C3H MEFs was described previously.²³ Briefly, Ifnar1^−/− 129/C3H MEFs were transduced with viral supernatants using murine stem cell virus (MSCV) encoding GFP alone or cDNA encoding HA-tagged STAT1 (generous gift from Thomas Decker). Primary MEFs were derived from day 13.5–14.5 wild type or Ifnar1^−/− C57BL/6 embryos and used during early passages (passage 2–6). For all culture conditions, cells were maintained at 37°C in Dulbecco's Modified Eagle Medium (DMEM) (JRH Biosciences) supplemented with 10% foetal bovine serum (FBS) (Thermotrace), 2 mM L-glutamine (JRH Biosciences), 50µM 2-mercaptoethanol (Sigma Chemical Co.) and penicillin/streptomycin (Invitrogen Life Technologies Corp.). Recombinant cytokines and receptors utilised in these experiments were IFNγ (Roche, 11276905910), IL-6 (Peprotec, 200-06) and IL-6 receptor (IL-6R) (R & D systems Inc., 227-SR).

Microarray analysis

RNA was extracted from wild type, Ifnar1^−/−, Ifnar1^−/−MSCV and Ifnar1^−/−HA-STAT1 transduced 129/C3H MEFs using Trizol (Invitrogen) according to the manufacturer's instructions. RNA quality was assessed by bioanalyzer using RNA Nano Chip (Agilent technologies) as per the manufacturer's instructions. RNA samples were considered degraded if RNA integrity number (RIN) was below 8.

All reagents (except 100% and 80%v/v Ethanol) were purchased from Affymetrix (Millennium Science) as part of the One-Cycle Target Labeling and Control Reagents (P/N 900493) package and GeneChip® Hybridization control kit (P/N 900454). cDNA preparation, sample hybridization and scanning of the oligonucleotide microarray GeneChip Arrays (Affymetrix) was performed as per manufacturer's instructions. Briefly 1µg RNA was reverse transcribed using Superscript II and T-7 Oligo (dT). Second strand cDNA synthesis, mediated by RNase H was followed by an in vitro transcription (IVT) reaction with T7 RNA polymerase, biotin-11-CTP and biotin-16-UTP to generate biotin labeled cRNA. Purified cRNA was fragmented by mild alkaline treatment, then hybridized to the GeneChip® Mouse Genome 430 2.0 Array Mouse chips (probe arrays) (Affymetrix) and washed using the GeneChip® Fluidics Station 450. Array chips were then stained with Streptavidin Phycoethrin (SAPE) (Molecular Probes) Biotinylated antibody (Vector Labs). Cartridges were loaded into autoloader of the GeneChip® Scanner 3000 7 G laser scanner.

The data (.dat) file, was automatically converted to a cell intensity (.cel) file and the quality of the individual arrays was assessed using GCOS. Microarray CEL files (Affymetrix probe identities and intensities) were converted to annotation files using Affy software (Affy).⁵⁷ All microarray data was assessed for degradation (degradation plots, Affy) normalized by Robust multi-array (RMA) normalization (Affy), p-values determined by Students t-test using Linear Models for Microarray Analysis (LIMMA) and multiple test corrections for each sample set comparison was performed using Benjamini-Hochberg multiple test correction method.⁵⁸ Significance cut offs of p ≤ 0.05 and fold change ≥ 1.2 for each comparison were implemented.

Quantitative real time PCR

RNA was extracted using Trizol (Invitrogen) according to the manufacturer's instructions. Alternatively, RNA was extracted using the RNeasy Plus Mini Kit (QIAGEN) as per manufacturer's instructions. cDNA was synthesized from 1 µg RNA using superscript III (Invitrogen) as per the manufacturer's instructions. The abundance of specific genes in the samples was quantitated using the SYBR Green dye detection method (Applied Biosystems, CA, USA). Primers to murine genes were as follows.

c-Jun ACTCCGAGCTGGCATCCA CCCACTGTTAACGTGGTTCATG

Irf1 TGGAGATGTTAGCCCGGACACTTT ACGGTGACAGTGCTGGAGTTATGT

L32 TTCCTGGTCCACAACGTCAAG TGTGAGCGATCTCGGCAC

Myd88 AGCAAGGAATGTGACTTCCAGACCA TGGGAAAGTCCTTCTTCATCGCCT

Nfkbia TGGCCTTCCTCAACTTCCAGAACA TCAGGATCACAGCCAGCTTTCAGA

Nfkbie ACATTGATGTACAGGAGGGCACGA ACAGAGTGGATGAGATGCTGTTGAGG

Socs3 CCTTCAGCTCCAAAAGCGAG GCTCTCCTGCAGCTTGCG

Stat1 CGCGCATGCAACTGGCATATAACT AAGCTCGAACCACTGTGACATCCT

Stat3 AGTTCCTGGCACCTTGGATTGAGA TTGGACTCTTGCACCAATCGGCTA

Threshold cycle numbers (Ct) were measured in the exponential phase for all samples. Relative abundance of sample genes was calculated using the “DDCT method” relative to the L32 control gene. mRNA abundance was normalized to the untreated samples of each genotype or relative to wild type as indicated.

Flow cytometry

Samples were first incubated in FC block in 2% FBS/PBS for 20 minutes. Antibodies against H2D^b (28-14-8) and H2K^b (AF6-88.5) were obtained from BD Pharmingen. Viability was determined by absence of Fluoro-gold uptake.

SDS-PAGE and protein gel blotting

Cells were washed once in PBS and lysed in whole cell lysis buffer (0.05% Triton X-100, 50 mM Tris, pH 7.5, 150 mM NaCl, 0.1 mM ethylene diamine tetra-acetic acid, 0.1 mM ethylene glycol tetra-acetic acid and 10% glycerol), containing protease and phosphatase inhibitors (1mg/mL aprotinin,0.5 mg/mL leupeptin, 0.2 mM PMSF, 10 mM NaF, 1 mM Na₃VO₄, 1 mM β glycero-phosphate) at 4°C for 10 min. Lysates were then cleared by centrifugation. Proteins were separated by SDS-PAGE and transferred to PVDF membranes. Membranes were probed with specific antibodies against Actin (Sigma), HSP70 (kind gift from R. Anderson), STAT1α/β (C terminus) (BD Bioscience), p-STAT3, STAT3, c-JUN, ERK1/2 (Cell Signaling Technology), PSMB9 (Abcam), TAP1 or TAP2 (Santa Cruz Biotechnology). Secondary antibodies were conjugated to horseradish peroxidase and images were visualized by chemiluminescence (ECL, GE Healthcare).

Statistical analysis

Statistical significance was assessed using Microsoft Excel software. For comparisons of means, Student's t-Test (t-test) was used and p-values <0.05 were considered significant. For cluster analysis of microarray probes, k-means organized arrays via Euclidean distance were displayed as logbase2. For cluster analysis of antigen presentation genes, hierarchical organized arrays via Euclidean distance were displayed as logbase2. Cluster analysis was performed using Cluster 3.0⁵⁹ and visualised using Java Tree View.⁶⁰ Pathway enrichment analysis and identification of transcription factors was performed using MetaCore from Thomson Reuters (MetaCore Genego, Thomson Reuters, https://portal.genego.com/Genego).

ABBREVIATIONS

B2M	=	Beta 2 microgloblin
ERK	=	Extracellular signal-regulated kinase
Hrs	=	Hours
IFN	=	Interferon
IFNAR	=	IFNα/β receptor
IFNAR1	=	IFNα/β receptor α chain
IL	=	Interleukin
IRF	=	Interferon regulatory factor
JAK	=	Janus activated kinase
MAPK	=	Mitogen activated protein kinase
MEF	=	Mouse embryonic fibroblast
MSCV	=	Murine stem cell virus
NFκB	=	Nuclear factor κB
PSMB	=	Proteasome subunit,β type
STAT	=	Signal transducer and activator of transcription
TAP	=	Transporter associated with antigen processing
TGF	=	Transforming growth factor
TNF	=	Tumor necrosis factor
U/ml	=	Units per ml

DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST

No potential conflicts of interest were disclosed.

ACKNOWLEDGMENTS

We thank Natalie Thomson for bioinformatics analysis and QC of microarray data. We also thank Dr Edwin Hawkins, Dr Sabine Hoves and Linda Hii for technical help.

Funding

The research reported in this publication was supported by the Cancer Research Institute (CRI scholarship), a Program Grant from the National Health and Medical Research Council of Australia (NHMRC) and a project grant from the Cancer Council Victoria. RWJ is an NHMRC Senior Principal Research Fellow.