Activation of STAT1 and the IFN-γ response are thought to be mediated exclusively through the Y440 motif of the human IFNGR1 receptor subunit. Contrary to this accepted dogma, here it is shown that IFNGR1 with a mutant (Y440F) motif, when stably expressed in IFNGR1-negative human fibroblasts at levels similar to wild type, can sustain a substantial IFN-γ response. The mutant receptor supports selective induction of IFN-γ-inducible genes but is notably defective in the CIITA, class II HLA, suppressor of cytokine signaling and antiviral responses. Remarkably, similar selective defects are observed in human fibrosarcoma cells expressing a mutant JAK1. The phenotypes are novel and appear distinct from those observed in response to the inhibition of known additional pathways. Data from different cell types further emphasizes the importance of cellular background in determining the response.
Experiments with mutant cell lines and knockout mice have established an essential role for the JAK/STAT1 pathway in the IFN-γ response (reviewed in Ref.1). There are four members of the mammalian JAK family of protein tyrosine kinases, JAKs 1–3 and Tyk2, and seven mammalian STAT genes. For IFN-γ, JAK/STAT1 signaling is through JAK1 and JAK2 and, in human cell systems at least, predominantly STAT1. The IFN-γR has two subunits: HuIFNGR1 and HuIFNGR2. HuIFNGR1 is the signal transduction subunit and recruits JAK1. HuIFNGR2 has a short 66-aa cytoplasmic domain for which the only known function is the recruitment of JAK2 (2, 3). The activation of JAK2 triggers the response (4). STAT1 is recruited to a phosphorylated tyrosine motif (Y440) toward the C terminus of HuIFNGR1. For HuIFNGR1 only the membrane proximal JAK binding domain and Y440 motif appear essential for an IFN-γ response ((5, 6, 7, 8); reviewed in Ref.9). On ligand binding there is receptor rearrangement with auto- and trans-phosphorylation/activation of preassociated JAKs, phosphorylation of the HuIFNGR1 Y440 motif, and recruitment and phosphorylation of STAT1. The phosphorylated STAT1 is released to migrate to the nucleus where, with or without additional factors, it activates transcription (reviewed in Ref.1). At some point in the activation cascade the ability of STAT1 to induce transcription is increased by enhanced phosphorylation of serine 727, through an as-yet-to-be-identified STAT1 kinase(s) (reviewed in Ref.10).
Although essential JAK/STAT1 signaling is not necessarily sufficient for a full IFN-γ response. A kinase-negative JAK1, for example, can sustain JAK/STAT1 signaling but not an antiviral response: additional signal(s) are likely required (4). Indeed, there is steadily increasing evidence for such a requirement (reviewed in Ref.11 and Discussion). Recent data have also emphasized the likely importance of cellular background in determining the response (e.g., Ref.12 ; J. F. Schlaak, A. P. Costa-Pereira, and I. M. Kerr, unpublished observations).
Here it is shown that a mutant HuIFNGR1 lacking a phosphorylatable Y440 motif, when stably expressed in IFNGR-null (IFNGR−) human diploid fibroblasts (HDFs)3 to yield “F440” cells, can sustain a substantial IFN-γ response. STAT1 activation although reduced is prolonged and IFN-inducible genes (ISGs) are selectively expressed. The induction of guanylate-binding protein (GBP) and IFN regulatory factor-1 mRNAs, for example, is relatively unaffected, whereas that of suppressor of cytokine signaling (SOCS)1 and 3, CIITA, and class II HLAs is substantially reduced and the antiviral response is minimal for the mutant receptor. A similar phenotype is observed in human fibrosarcoma cells stably overexpressing a deletion mutant of JAK1. To date we have failed to obtain any convincing evidence for a role for a defect in serine phosphorylation of STAT1, or in the PI3K/Akt or ERK1/2 and p38 MAPK additional signaling pathways in the phenotypes observed. The contrasting results obtained in different human cell systems and in comparison with those reported for mouse cells (9, 13) emphasize the importance of cellular background in determining the signaling response.
Materials and Methods
Cell lines and reagents
The normal and HuIFNGR1− HDFs were gifts from Drs. S. Dupuis and J.-L. Casanova (Institut National de la Santé et de la Recherche Médicale, Necker-Enfants Malades Medical School, Paris, France). All cells were cultured in DMEM supplemented with 10% heat-inactivated FCS, 2 mM l-glutamine, 50 U/ml penicillin and 50 μg/ml streptomycin. Resistant cells were maintained in medium containing 250 μg/ml hygromycin (Calbiochem) and/or 250 μg/ml G418 (Invitrogen Life Technologies) and/or Puromycin 1.25 μg/ml (Sigma-Aldrich). PI3K inhibitor (LY294002) (Sigma-Aldrich) was used at 20 μM and was added 30 min before IFN-γ treatment. Highly purified recombinant IFN-γ (4 × 107 IU/mg protein) was a gift from G. Adolf (Ernst Boehringer Institut fûr Arzneimittelforschung, Vienna, Austria).
Generation of stable populations of transfected human fibroblasts by retroviral infection
Production of viral stock.
The day before transfection, 2 × 106 BOSC virus-packaging cells were plated in 10-cm dishes. Five minutes before transfection, the medium was changed to 5 ml of DMEM/10% FCS containing 25 μM chloroquine (diphosphate salt) to inhibit degradation of DNA by lysosomal hydrolases. Ten micrograms of plasmid DNA containing the wild-type or mutant HuIFNGR1 cDNAs in the vector pBabe was diluted in sterile H2O to give a final volume of 440 μl. Sixty microliters of 2 M calcium chloride 2-hydrate were then added to the DNA. Five hundred microliters of 2× HBS (280 mM NaCl, 10 mM KCl, 1.5 mM Na2HPO4, 12 mM d-glucose, and 50 mM HEPES (pH 7.05)) was added dropwise to the DNA/calcium chloride solution. The resulting solution was immediately added to the BOSC cells, swirling the plate to mix. Following incubation at 37°C for 6–9 h, the medium was aspirated from the BOSC cells and replaced with 10 ml of DMEM/10% FCS. The following morning, the medium was changed again to 5 ml of DMEM/10% FCS. After additional incubation at 37°C for 8 h, the virus stock was harvested and filtered through a 0.45-μm filter.
Infection of target cells.
The day before infection, the human fibroblasts were plated at 25% confluency. On the day of infection, the medium was aspirated from the target cells and replaced with 4 ml of virus stock (for a 10-cm plate). Polybrene (hexadimethrine bromide) was added to give a final concentration of 4 μg/ml to facilitate entry of virus into cells. After incubation overnight at 37°C, the virus stock was replaced with 10 ml of DMEM/10% FCS. Twenty-four hours later, the medium was changed again, and cells were put under puromycin (1 μg/ml) selection. If cells were already confluent, they were split into selection medium. Cells were maintained under selection until a stable drug-resistant population was obtained.
Abs used included anti-IFNGR1 (GIR-94; BD Pharmingen), PE-conjugated anti-HLA DRα (L243; BD Biosciences), and FITC-conjugated anti-class I (HLA-A, -B, -C) (W6/32; Harlan) for cell surface staining and anti-P-tyrosine (4G10; Upstate Biotechnology), anti-P-Ser473-Akt (Cell Signaling Technology), Akt (Cell Signaling Technology), anti-JAK1 (HR-785; Santa Cruz Biotechnology), anti-P-Ser727-STAT1 (Upstate Biotechnology), anti-P-Tyr701-STAT1 (Cell Signaling Technology), anti-N-term-STAT1 (BD Transduction Laboratories), and anti-C-term-STAT1 (C-111; Santa Cruz Biotechnology) for Western blotting.
Cells treated with medium only or 1000 IU/ml IFN-γ for 24–72 h were removed from the plates with buffered EDTA, washed in ice-cold PBS, and incubated with anti-HLA DRα-PE, anti-HLA-A, -B, -C-FITC, or isotypic control Ab for 45 min on ice in the dark. Cells were then washed three times with ice-cold PBS, fixed in 1% p-formaldehyde and analyzed on a FACS.
Cell lysis, Western blotting, and EMSA
Cells were lysed on ice using 50 mM Tris (pH 8.0), 0.5% Nonidet P-40, 10% glycerol, 150 mM NaCl, 1 mM DTT, 0.1 mM EDTA, 0.2 mM sodium orthovanadate, 0.5 mM PMSF, 3 μg/ml aprotinin, and 1 μg/ml leupeptin, cell debris were removed by centrifugation, and whole cells extracts used for EMSA, immunoprecipitations, or Western blots, essentially as described previously (14). To detect P-STAT1 and STAT1, cells were lysed in 25 mM Tris (pH 7.6), 10% glycerol, and 2% SDS, sonicated, and cell debris were removed by centrifugation before assay. Protein concentration was determined using the Bio-Rad protein assay (Bio-Rad) according to the manufacturer’s instructions.
RNase protection assays (RPAs)
Cytoplasmic RNA was isolated using RNeasy columns (Qiagen), as directed by the manufacturer and protection assays conducted as described previously (15). The SOCS probes were purchased from BD Pharmingen, the CIITA probe was generated from a 350 bp HindIII fragment (nt 2935–3285) from a human CIITA cDNA (from Dr. R. Flavell, Howard Hughes Medical Institute, New Haven, CT) cloned into pGEM4 and the other ISG probes were generated as described previously (15).
Luciferase activity was measured using the assay system purchased from Promega. Cells were lysed in reporter lysis buffer (as supplied), frozen on dry ice and thawed, vortexed briefly and then samples centrifuged for 2 min at 12,000 × g, and the supernatant assayed for Luciferase activity or stored at −70°C. To assay the luciferase activity, a 20 μl aliquot of the extract was mixed with 100 μl of luciferase assay reagent using an injection luminometer. Light intensity was measured over a 10-s period and averaged to give the relative light intensity for each sample.
Cells seeded in 24-well plates at 2 × 105 cells/well were incubated overnight at 37°C, treated with serial dilutions of IFN-γ for 18 h and challenged with encephalomyocarditis (EMC) virus (0.3–1 pfu/cell). Twenty hours postinfection cells were frozen, thawed, centrifuged to remove cell debris, and the virus yield assayed by serial dilution of the supernatants onto just-subconfluent mouse L cells (INsensitive to “carryover” of human IFNs) in 96-well plates, which 20 h postinfection were fixed with formol saline and stained with Giemsa for live cells
HDFs expressing wild-type and mutant HuIFNGR1
The human IFNGR1-null fibroblasts (HuGR1− cells) were derived from a patient homozygous for a mutation that introduces a stop codon in exon 2 (S. Dupuis and J. L. Casanova, unpublished observations; cf Ref.16). IFNGR1 protein is not detectably expressed in these cells or in a population of these cells transfected with a vector-only construct (pBabe cells). Populations of the HuIFNGR1− cells were derived, which stably express wild-type HuIFNGR1 (HuGR1 cells) or a Y440F mutant HuIFNGR1 (F440 cells) to comparable levels slightly in excess of the levels of the endogenous receptor in wild-type cells from a normal volunteer (HDFs) (Fig. 1). Consistent with the absence of a receptor, no signal was obtained in the HuGR1− or pBabe cells in assays for receptor phosphorylation in marked contrast to the massive IFN-γ-inducible IFNGR1 tyrosine phosphorylation observed in the HuGR1 cells or HDFs (Ref.17 and data not presented). Indeed, the HuGR1− and pBabe cells showed no detectable response to IFN-γ in any of the assays used throughout. The Y to F mutation in the IFNGR1 was confirmed by sequencing before transfection and re-sequencing of the transfected receptor “recovered” by PCR amplification from the F440 cells. No other mutation was found. It is reasonable to conclude that the HuGR1− cells do indeed lack IFNGR1 and that the F440 cells express only the Y440F mutant receptor.
Activation of JAK/STAT signaling in response to IFN-γ
Rapid and prolonged tyrosine phosphorylation/activation of STAT1 occurred in response to IFN-γ in both the HuGR1 and F440 cells. The relative amounts were, however, somewhat variable. This was particularly true at early (15 min and 1 h) time points for which the data for the F440s ranged from paralleling those for the HuGR1s (but at a lower level) to showing, as in Fig. 2,A, a much reduced level without the sharp peak of activation seen by EMSAs for the HuGR1 cells. Irrespective, at later time points (>3 h, e.g., 6 and 24 h; Fig. 2,A) prolonged activation to between 5 and 30% of HuGR1 levels, sustained through 48–72 h, was routinely observed for the F440s. JAK1 phosphorylation which peaks early in the HuGR1 cells, is sustained at peak levels for at least 24 h in the F440s (Fig. 2,B). It is reasonable to conclude that in the F440s the JAK/receptor complex is activated normally, but that JAK1 activation is prolonged in the absence of high levels of induced SOCS1 (cf Fig. 3,B). This in turn may in the F440s contribute to the sustained activation of STAT1 independent of a phosphorylatable Y440 motif. Such activation could a priori be through alternative receptor motifs or the JAKs. Consistent with the latter a minimal chimeric receptor (EgΔB; Fig. 2,C) can mediate JAK/STAT1 activation. EgΔB (Fig. 2,C) comprises the external region of the erythropoietin (Epo) receptor and only the transmembrane and juxtamembrane JAK1 binding domains of a truncated gp130, the signal transduction subunit of IL-6-type cytokines’ receptors. It can, when stably expressed in 293T cells, support Epo-stimulated activation of STAT1 (Fig. 2 D) and the induction of, for example, 9–27 mRNA, a well-established γ-ISG (Ref.17 and data not shown).
Gene induction in response to IFN-γ
By RPA the induction of individual γ-ISGs was selectively impaired in the F440 cells. The induction of GBP-1 mRNA, for example, approached wild-type levels, that of IRF-1 was slightly reduced whereas that of CIITA and secondarily DRα (Fig. 3,A, compare lanes 3–6 with 7–10) and SOCS3 and 1, in particular, were more profoundly affected (Fig. 3,B, compare lanes 6–10 with 11–15). In addition, the induction of class II, but not class I, HLA was much reduced in the F440 cells when assayed by FACS for cell surface expression (Fig. 3 C), or for protein in whole cell extracts (data not shown).
The antiviral response to EMC virus through IFN-γ is defective in F440 cells
For HuGR1 cells an antiviral response is seen with as little as 0.3 IU/ml IFN-γ and the maximal ∼100-fold reduction of virus yield (typical of IFN-γ in many cell types) is reached at 300 IU/ml (Table I). Accordingly, it is remarkable that in the F440 cells no antiviral response is seen at 103 IU/ml, a concentration at which prolonged STAT1 activation to 5–30% of wild-type levels is observed (e.g., Figs. 2 and 3). Even at 5 × 103 IU/ml only a marginal antiviral response, if any, is observed (Table I).
|IFN-γ (IU/ml) .||Virus Yielda .||.||.||.||.||.||.|
|.||0 .||0.3 .||3 .||30 .||300 .||1000 .||5000 .|
|IFN-γ (IU/ml) .||Virus Yielda .||.||.||.||.||.||.|
|.||0 .||0.3 .||3 .||30 .||300 .||1000 .||5000 .|
Arbitrary units: EMC virus dilution for 50% cell death.
STAT1 Ser727 phosphorylation in the IFN-γ response
STAT1 activation is through phosphorylation of tyrosine residue 701. STAT1 transcriptional activity, however, is enhanced by phosphorylation of serine residue 727 (18). IFN-γ-inducible Ser727 phosphorylation is clear in HuGR1 cells. In contrast it is not convincingly detectable in F440 cells (Fig. 4, A and B). Accepting the up to 20-fold lower level of P-Tyr STAT1 at early time points in F440 vs HuGR1 cells (Figs. 2,A and 4,B), however, even if the tyrosine-phosphorylated STAT1 were Ser727-phosphorylated it would predictably be at or below the limits of detection of the assay. A variety of approaches including affinity concentration of the STAT1-P-Tyr701 have so far failed to yield a clear result for the F440 cells. Accordingly the effects of 1) potential inhibitors of STAT1-Ser727 phosphorylation and 2) mutation of STAT1-Ser727 were investigated. For HuGR1 cells exposure to accepted inhibitory concentrations of LY294002 (20 μM) resulted in complete inhibition of PI3K-dependent phosphorylation/activation of Akt (Fig. 4,B). However, in marked contrast to the results with fibrosarcoma cells (Ref.19 see below and Fig. 5,D), it was without significant effect on STAT1-Ser727 phosphorylation in the HuGR1 cells (Fig. 4,B). In the alternative approach the IFN-γ response was analyzed in STAT1-null U3A cells stably expressing a nonphosphorylatable serine 727 to alanine mutant of STAT1 (STAT1.S727A). The mutation did not mimic the F440 phenotype. It did not inhibit the induction of CIITA, DRα, invariant chain (Ii), IRF-1 or GBP (Fig. 5 C, lanes 1–3 and 10–15). Therefore, it is unlikely that the differential defect in the induction of these genes in the F440s reflects a defect in STAT1-Ser727 phosphorylation.
A PI3K/Akt inhibitor does not mimic the F440 phenotype
More generally, irrespective of the question of STAT1-Ser727 phosphorylation, despite good inhibition of the PI3K in HuGR1 cells (above, Fig. 4,B), the LY294002 inhibitor did not mimic the F440 phenotype. It did not inhibit the induction of CIITA and invariant chain (Ii) (Fig. 4,C, compare CIITA, in particular, at 6 and 18 h, lanes 2 and 3—no inhibitor, vs lanes 5 and 6—plus LY294002 inhibitor). It similarly failed to mimic the F440 phenotype with respect to the inhibition of SOCS mRNAs (Fig. 3 B), being completely without effect on the induction of SOCS1, 3 and cis at 3, 6, or 18 h (data not presented). It can be concluded that the PI3K/Akt inhibitor, although highly effective in the inhibition of Akt phosphorylation/activation, is without major effect on the IFN-γ response in human fibroblast assayed either by STAT1 P-Tyr701 or P-Ser727 phosphorylation, or the induction of typical γ-ISGs.
Differential induction of γ-ISGs in human fibrosarcoma cells stably expressing a mutant JAK1
The F440 phenotype was obtained with a population and not with a clone of transfected cells. The concern remains that this phenotype could be peculiar to the patient from whom the HuGR1− cells were isolated. However, a similar phenotype has been observed in human fibrosarcoma (2fTGH) cells stably expressing a deletion mutant (Fig. 5,A) of JAK1 (2fJAK1ΔB cells). Two clones, which expressed the mutant JAK1 at levels 5- to 10-fold greater than that of the endogenous wild-type JAK1, showed the phenotype described. The phosphorylation/activation of STAT1 in response to IFN-γ in the 2fJAK1ΔB cells was indistinguishable from that observed in the parental 2fTGH cells (Fig. 5,B). Induction of γ-ISGs was, however, differentially affected (Fig. 5,C, lanes 1–3 vs 7–9) in a manner similar to that observed for the F440 cells (Fig. 3,A, lanes 3–6 vs 7–10). The induction of GBP-1, for example, was essentially identical to that in the parental 2fTGH cells, that of IRF-1 was slightly reduced, whereas that of CIITA and secondarily, invariant chain (Ii) and DRα were more profoundly affected. These differentials were maintained throughout a more extensive kinetic analysis from 6 to 30 h (data not presented) and, as would be expected from this, cell surface expression of class II but not class I HLAs in response to IFN-γ was greatly reduced in the 2fJAK1ΔB cells (Fig. 6, middle panel). Consistent with the defect in CIITA expression indicative of an early defect in the IFN-γ response, class II HLA expression was restored in 2fJAK1ΔB cells stably expressing transfected CIITA (Fig. 6, lower panel). A defect in the transcription rather than the stability of CIITA mRNA seems likely as IFN-γ-inducible expression from a CIITA promoter IV (the IFN-γ-responsive element, Ref.20)-luciferase reporter construct, but not an IRF-1 promoter-luciferase reporter construct, was inhibited in the 2fJAK1ΔB cells (Fig. 7). Similar results were obtained with a CIITA promoter IV-globin construct (data not presented). It is reasonable to conclude that in the 2fJAKΔB cells there is a differential defect in the transcription of a subset(s) of γ-ISGs that cannot be attributed to a defect in STAT1 activation (P-Tyr701) or P-Ser727 phosphorylation (Fig. 5,B). In accord with the latter, the inhibition of STAT1-Ser727 phosphorylation by 1) the PI3K inhibitor LY294002 (Fig. 5,D) or 2) a Ser727Ala mutation of STAT1 did not inhibit the induction of CIITA, invariant chain (Ii) or DRα (Fig. 5 C).
In wild-type 2fTGH cells the IFN-γ antiviral response vs EMC virus is first observed between 10 and 30 IU/ml and a maximal 100-fold reduction in virus yield is obtained at 103 IU/ml. For the 2fJAK1ΔB cells a much reduced response is observed with only partial protection at 103-104 IU/ml. In contrast, both cell lines showed the normal much greater response to IFN-α with protection first being observed between 10 and 100 IU/ml with a maximal 105-fold reduction in virus yield at 103 to 104 IU/ml (data not shown).
A Y440F mutant of human IFNGR1 stably expressed at similar-to-normal levels in IFNGR1− HDFs mediates a substantial IFN-γ response. P-Tyr phosphorylation/activation of STAT1 was variable but at later time points (6–48 h) values from 5 to 30% of wild type were usually observed. JAK activation appeared normal consistent with normal JAK/receptor function but peak activation levels were prolonged for the mutant receptor (e.g., Fig. 2,B). Of the γ-ISGs routinely monitored by RPA assays, induction through the mutant receptor varied from approaching wild type for GBP-1 mRNA to much reduced for CIITA (and secondarily, DRα) and particularly SOCS1 (Fig. 3, A and B). A similar spectrum of levels of inducibility was observed for the top 20 most highly inducible genes in a limited series of “custom” macroarray (12) analyses targeting 90 known γ-ISGs (J. F. Schlaak, A. P. Costa-Pereira, and I. M. Kerr, data not presented). The reduced level of induction of SOCS1 likely contributes to the prolonged level of JAK activation (Fig. 2 B).
Despite prolonged, substantial activation of STAT1 (Fig. 2,A) the CIITA, class II HLA, SOCS and antiviral responses are differentially defective in the F440 cells (Fig. 3). For example, an antiviral response is first detectable at as little as 0.1 IU/ml IFN-γ in wild-type cells whereas essentially no antiviral response was obtained in the F440s at up to 5 × 103 IU/ml (Table I), a concentration well in excess of that required (e.g., Fig. 2,A) to give substantial STAT1 activation throughout the period of the assay. For the F440 cells STAT1-Ser727 phosphorylation is not convincingly detectable. The IFN-γ response mediated by a non-Ser727-phosphorylatable mutant STAT1 (STAT1S727A) did not, however, mimic the F440 phenotype (Fig. 5,C). Therefore, it is unlikely that the phenotype reflects a defect in STAT1-Ser727 phosphorylation. Inhibition of the PI3K pathway (Fig. 4, B and C), or the ERK1/2 or p38 MAPK pathways (data not presented) did not mimic the F440 phenotype.
It is worth remembering both that the vast majority of previous work defining the requirement for the Y440 motif for the human IFNGR1 was conducted in a mouse cell background and that, a priori, there is no reason to exclude the possibility that the Y440 motif may be required for other signals in addition to STAT1. The results here are in accord with the original work establishing a requirement for the Y440 motif of the HuIFNGR1 for an IFN-γ response (7). They are, however, at odds with convention and the results of a recent detailed analysis of the corresponding murine IFNGR1 receptor mutant in IFNGR1− MEFs (13) in the lack of an absolute requirement for this motif for STAT1 activation and substantial induction of γ-ISGs. The mechanism of activation of STAT1 through the JAK/receptor complex in the F440 cells is not known. It is well established that an intact SH2 domain is required for all ligand-mediated STAT activation and it is reasonable to assume that this is likely to be the case here. Activation may be through the JAKs as was reported some years ago by Hirano and colleagues (21) for STAT5 and JAKs 1, 2, and 3 and as may well be the case for the EgΔB/293T cell system mentioned here (Fig. 2 D). Alternatively, it could, a priori, be through cross-recruitment of another receptor or, contrary to dogma the F440 motif, or through additional tyrosine or other as-yet-to-be-defined motifs in HuIFNGR1. A very substantial amount of additional work would be required both to determine which of these possibilities is correct and, more importantly, to what extent, if at all, the alternative mechanism(s) of activation are used under normal circumstances through wild-type receptors in a variety of human cell types. A requirement for IFN-γ translocation to the nucleus cannot formally be excluded (e.g., Ref.22).
The F440 cells used throughout were populations. Therefore, the F440 phenotype is not unique to a particular clone of cells. Nevertheless, the concern remains that this phenotype could be peculiar to the patient from whom the HuGR1− cells were isolated. A similar phenotype has, however, been observed in HT1080-based human fibrosarcoma cells stably expressing a mutant JAK1 (2fJAK1ΔB; Figs. 5–7) Interestingly, a selective defect in the IFN-γ inducibility of CIITA (and, secondary to it, of class II HLA and invariant chain, but not GBP) has been reported to be a common feature of human tumor cell lines of potential importance in the evasion of immune surveillance (23). In the 2fJAK1ΔB system despite apparently identical activation (P-Tyr701 sustained >48 h) and P-Ser727 phosphorylation of STAT1 (Fig. 5,B), the induction of CIITA, invariant chain and DRα, for example, are differentially defective (compared with GBP and IRF-1, Fig. 5,C). The cell surface expression of class II HLAs (Fig. 6) and the antiviral response (data not shown) are similarly defective. There is no detectable defect in STAT1-Ser727 phosphorylation in response to IFN-γ in these cells (Fig. 5,B). Moreover neither inhibition of such phosphorylation using the LY294002 inhibitor nor the replacement of wild-type STAT1 with a nonphosphorylatable Ser727Ala mutant STAT1 yielded a 2fJAK1ΔB/F440-like phenotype (Fig. 5 C).
In the 2fJAK1ΔB cells the defect in class II HLA expression (but not the antiviral response) is “complemented” on expression of stably transfected CIITA (Fig. 6). For CIITA the defect in the IFN-γ response appears to be at the level of transcription. Promoter IV is the IFN-γ-inducible element of the CIITA promoter (20). Transcription from CIITA promoter IV, but not IRF-1 promoter, reporter constructs is defective in the 2fJAK1ΔB cells (Fig. 7). We have, however, been unable to detect an obvious defect in promoter interaction for the factors known (24) to be involved in mediating the IFN-γ response (STAT1, IRF-1 and USF1; V. Arulampalam, T. M. Williams, and I. M. Kerr, data not presented). Accordingly, it remains possible that the defect(s) are in signals governing promoter availability, for example, in higher transcription complexes.
There is substantial evidence for variable involvement of three major additional signaling pathways, the PI3K/Akt and the ERK1/2 and p38 MAPK pathways in the IFN-γ response, depending on cell type (reviewed in Ref.11). In the present study, with or without serum starvation, substantial constitutive activity for each of the three major pathways above was routinely observed. Against this background only minimal additional signals, if any, were obtained in response to IFN-γ in the wild-type cells precluding any convincing detection of a defect in the 2fJAK1ΔB cells. In an alternative approach the effects of the established LY294002, PD98059, and SB202190/203580 inhibitors of these pathways were assayed in the appropriate wild-type cells. In no case was a 2fJAK1ΔB-like phenotype observed (e.g., Fig. 5 C for the LY294002 inhibitor). Furthermore, in a limited series of “custom” macroarrays monitoring 90 known γ-ISGs only minor (predominantly <2-fold) pan effects similarly affecting all of the genes represented with no striking differential subsets, were observed (A. P. Costa-Pereira, J. F. Schlaak, and I. M. Kerr, unpublished results). To date, therefore, we have failed to detect a 2fJAK1ΔB-like phenotype in experiments of this type. Moreover, to our knowledge a similar phenotype has not been observed for any of the other systems (reviewed in Ref.11) in which additional signaling has been reported. Accordingly, here novel signaling pathways or usage are likely involved.
The differing results obtained in the different human cell systems and in comparison with previous and current work in mouse cells are also of interest with respect to the role of cellular background in the IFN-γ response. Although not mentioned to date, a comment concerning STAT3 is perhaps appropriate. It is generally accepted that the activation of STAT3 in response to IFN-γ is more clearly seen in murine than human cell systems. In our experience there is, in addition, for human systems, considerable variation between cell types. Here, low-level activation of STAT3 was variably seen in both the HuGR1 and F440 cells but crucially no consistent difference between the two was observed. Additional examples of variation in the response with cellular background arising out of this work include 1) the contrasting requirements for the Y440 motif for STAT1 activation in the HDFs (here) vs MEFs (13); 2) STAT1 activation through the minimal EgΔB chimera in stably transfected 293T (Fig. 2,C) but not 2fTGH (25) cells; and 3) the contrasting sensitivity of STAT1-Ser727 phosphorylation to the LY294002 inhibitor in the human fibrosarcoma (Ref.19 and Fig. 5,D) vs the HDFs (Fig. 4 B). Together these provide evidence for differences in the IFN-γ response not only between normal and transformed cells but also between human and mouse cells. They add significantly to the increasing evidence for the importance of cellular background in determining the nature of the signaling response.
We are heavily indebted to Dr. G. Adolf for the IFN-γ and to Dr. Stephanie Dupuis and Dr. Jean-Laurent Casanova (Institut National de la Santé et de la Recherche Médicale) for the IFNGR1-deficient HDFs used in this study.
The authors have no financial conflict of interest.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Abbreviations used in this paper: HDF, human diploid fibroblast; ISG, IFN-inducible gene; EMC, encephalomyocarditis; IRF-1, IFN regulatory factor-1; SOCS, suppressor of cytokine signaling; RPA, RNase protection assay; Epo, erythropoietin; GBP, guanylate-binding protein.