Melanoma cell-secreted soluble factor that stimulates ubiquitination and degradation of the interferon alpha receptor and attenuates its signaling

Authors


S. Fuchs, e-mail: syfuchs@vet.upenn.edu

Dear Sir,

Adjuvant therapy of malignant melanoma with recombinant interferon alpha (IFN-α) is one of the few Federal Drug Administration (FDA)-approved treatments (Ascierto and Kirkwood, 2008). Although IFN-α exhibits potent anti-proliferative and anti-survival effects against tumor cells and endothelial cells of tumor vasculature in vitro, its therapeutic efficacy is limited (Mocellin et al., 2010). Overcoming these limitations requires a better understanding of the mechanisms that temper the effects of pharmaceutical IFN-α in the patient settings.

All effects of IFN-α/β on cells are elicited through interaction with a cognate cell surface Type I IFN receptor that consists of IFNAR1 and IFNAR2c chains. This interaction leads to catalytic activation of Janus kinases (JAK), tyrosine phosphorylation of Signal Transducers and Activators of Transcription 1/2 (Stat1/2) proteins and activation of transcription of IFN-stimulated genes whose products restrict cell proliferation and viability (Platanias, 2005). Upon completion of the transcriptional program, the expression of negative regulators of Jak and Stat activities limits the magnitude and duration of this signaling. However, much earlier on, the ability of cells to further react to additional IFN-α/β molecules is rapidly eliminated by the ligand-induced downregulation of Type I IFN receptors from the cell surface (Coccia et al., 2006).

IFN-α/β-induced downregulation of Type I IFN receptor is governed by ubiquitination of its IFNAR1 chain, mediated by the SCFβTrcp E3 ubiquitin ligase (Kumar et al., 2003). This ligase is recruited to IFNAR1 upon IFN-α/β-induced and Jak kinase activity-dependent phosphorylation of the Ser residues (e.g. Ser535) within the degron of IFNAR1 (Kumar et al., 2004). The manner in which levels of βTrcp2 are induced by mitogenic stimuli depends on the mitogen-activated protein kinase Erk1/2 (Spiegelman et al., 2002). A constitutively active BRAF mutant signals via Erk1/2 to increase the levels of βTrcp2, to downregulate IFNAR1, and to decrease cell responses to IFN-α in human melanoma cells (Kumar et al., 2007). Given that melanomas are known to secrete growth factors whose autocrine effect can also contribute to MAPK activation (Shih and Herlyn, 1994), we sought to determine whether pre-incubation of normal human melanocytes with media conditioned by 1205 Lu metastatic melanoma cells alters the levels of βTrcp and cellular responses to IFN-α. Pretreatment for only 2 h was not sufficient to induce βTrcp in melanocytes (Figure 1A). Surprisingly, activation of Stat1 in these cells was markedly inhibited. Similar observations could be made using media conditioned by WM278 and WM793 cells (data not shown). Conversely, when pretreated with media from melanocytes, melanoma cells exhibited a greater extent of response to IFN-α without dramatic changes in βTrcp levels (Figure 1A). These data suggest either that normal melanocytes secrete a soluble factor that stimulates IFN-α signaling, or that melanoma cells produce an inhibitory factor. Furthermore, it is unlikely that either of these putative factors acts via changing the levels of βTrcp.

Figure 1.

 (A) Melanocytes (MC) or 1205 Lu metastatic melanoma cells were pretreated with conditioned media from indicated sources for 2 h and then stimulated with IFN-α (30 IU/ml) for 30 min. Activation of STAT1 was monitored by immunoblot (IB) using pSTAT1 antibody (top panel). Levels of total STAT1, β-Trcp, and β-actin are also shown. (B) Melanocytes were pretreated with 1205 Lu metastatic melanoma conditioned media (MM) for 2 h and then stimulated with human IFN-α (60 IU/ml) or IFN-γ (50 IU/ml) for 30 min. Phosphorylation of STAT1 (top panel) and total levels of STAT1 are shown (bottom panel). (C) Levels of cell surface IFNAR1 analyzed by FACS using monoclonal anti-hIFNAR1 antibody in human melanoma cell lines as indicated. Left line represents the isotype Ig control. 1205 Lu and WM115 melanoma cells were stimulated with IFN-α (30 IU/ml) for 30 min. Phosphorylation and levels of STAT1 were monitored by IB (inset). Relative activation levels of STAT1 are shown as fold-increase. (D) HeLa cells were pretreated with JAK inhibitor I (500 μM, Calbiochem) or vehicle control (DMSO) for 1 h followed by IFN-α (1000 IU/ml) or 1205 Lu melanoma conditioned media (MM) for 30 min. IFNAR1 proteins were immunopurified by IP using IFNAR1 antibody and analyzed by IB using anti-pSer532, anti-pSer535, and anti-IFNAR1 antibodies as indicated. (E) HeLa cells were treated with 1205 Lu conditioned media (MM) for 30 min. Ubiquitination and levels of endogenous IFNAR1 were analyzed by IB using anti-Ub, and anti-IFNAR1 antibodies as indicated (upper panels). Levels of β-actin in the lysates are shown as loading control (bottom panel). (F) HeLa cells were treated with cycloheximide (CHX, 30 μg/ml) alone or in combination with 1205 Lu conditioned media (MM) for indicated times. Levels of endogenous IFNAR1 were analyzed by IB using anti-IFNAR1 antibody (upper panel). Levels of β-actin in the lysates are shown as loading control (bottom panel).

Further studies revealed that simply replacing conditioned media on melanoma cells with fresh media increased the extent of IFN-α signaling (data not shown). Furthermore, pretreatment with 1205 Lu metastatic melanoma cell-conditioned media (MM) robustly reduced the extent of Stat1 activation in response to IFN-α but not to IFN-γ in normal human melanocytes (Figure 1B) and early stage radial growth phase WM115 melanoma cells or HeLa cells (data not shown). Given that, similarly to IFN-α, IFN-γ activates Janus kinases but acts via a completely different receptor (Platanias, 2005), it is plausible that a melanoma-secreted soluble factor acts by altering the levels of Type I IFN receptor available to interact with the ligand. Indeed, the cell surface levels of the IFNAR1 chain of this receptor were noticeably lower in 1205 Lu metastatic melanoma than in normal melanocytes or early WM115 cells that exhibited a robust response to IFN-α (Figure 1C). Low levels of IFNAR1 were reported to correlate with poor prognosis for patients with metastatic melanoma (Messina et al., 2008). These data suggest that downregulation of IFNAR1 may occur in metastatic melanomas.

Downregulation of IFNAR1 is mediated by its βTrcp-dependent ubiquitination, which requires phosphorylation of IFNAR1 degron on Ser535 (Kumar et al., 2004). This phosphorylation was indeed increased in endogenous IFNAR1 from either HeLa or 293T cells treated with IFN-α or with melanoma cell-conditioned media (Figure 1D and data not shown). Pretreatment of these cells with Jak inhibitor I noticeably decreased Ser535 the phosphorylation of IFNAR1 induced by its natural ligand (i.e. IFN-α) but not by the conditioned melanoma media (Figure 1D). This indicates that soluble factor(s) secreted by melanoma cells can activate the ligand/ JAK-independent pathway of IFNAR1 phosphorylation. We previously reported that this pathway involves priming phosphorylation of IFNAR1 on Ser532 (Bhattacharya et al., 2010). Indeed, a robust stimulation of this priming phosphorylation was seen in cells treated with melanoma-conditioned media (Figure 1D).

We then sought to investigate whether these soluble factors affect ubiquitination and degradation of IFNAR1. Treatment of HeLa cells with melanoma-conditioned media robustly increased the extent of ubiquitination of endogenous IFNAR1 (Figure 1E). A similar result was obtained when ubiquitination of exogenously expressed Flag-IFNAR1 was analyzed (data not shown). Furthermore, consistent with a key role of IFNAR1 ubiquitination in the degradation of this receptor (Kumar et al., 2003), we observed that conditioned media from malignant melanoma cells when added to HeLa cells led to a robust stimulation of the turnover of endogenous IFNAR1 (Figure 1F). Similar accelerated degradation was observed on exogenously expressed receptor (data not shown).

Our results suggest that melanoma-secreted factor(s) accelerate IFNAR1 phosphorylation, ubiquitination, degradation and downregulation through a novel ligand/ JAK-independent pathway. It is plausible that this mechanism might be at least in part responsible for impeding the cellular responses to IFN-α. Future identification of these soluble factors and subsequent delineation of the signaling pathways through which these factors elicit these changes may point to novel targets for improving the efficacy of IFN-α therapy in melanoma patients.

Acknowledgements

We thank Dr. M. Herlyn for cell lines. This work was supported by NIH grant CA92900 (to S.Y.F.).

Ancillary