Dysregulation of anti-tumor immunity by the matrix metalloproteinase-2

Abstract

The matrix metalloproteinase-2 (MMP-2), overexpressed in most cancers, induces TH2 polarization by conditioning dendritic cells to over-express OX40L and downregulate IL-12p70 through the degradation of the type-I IFN receptor IFNAR1. Elucidating mechanisms underlying detrimental tumor-associated type-2 responses represent a crucial step in designing effective immune therapies to treat cancer patients.

As CD4 help appears crucial for generating effective anti-tumor immunity, we investigated the existence of MMP-2-specific CD4⁺ T cells. We detected such cells in healthy donors and more importantly at a much higher frequency in over 40% of melanoma patients tested. We cloned CD4⁺ T cells specific for 11 novel MMP-2-derived epitopes. Strikingly, these cells displayed an inflammatory T_H2 phenotype, i.e., mainly secreting TNFα, IL-4 and IL-13 and expressing GATA3. Whereas IFNγ-secreting T_H1 cells appear to exhibit effective anti-tumor properties, T_H2 cells are often described as detrimental in this function.^4-6 Despite a lack of statistical significance, we also observed a trend toward a poorer clinical outcome/survival for patients with detectable MMP-2-specific T_H2 cells, suggesting the significance of these cells and/or MMP-2 itself in melanoma progression.

We investigated the mechanism underlying the observed T_H2 skewing and found that MMP-2-conditioned dendritic cells (DCs) preferentially prime inflammatory T_H2 cells. Of note, MMP-2 enzyme in its active conformation was the strongest inducer of T_H2 responses. While, according to the literature, T_H1 differentiation reliably depends on IL-12, the generation of T_H2 cells is not fully understood and/or may differ from one model to another. IL-4, through GATA3 induction, is described as critical for T_H2 differentiation in most studies. However, IL-4 did not seem to be involved in our system. A default mechanism could also explain T_H2 differentiation, where the lack of T_H1-polarizing signal, namely IL-12, would be sufficient and/or necessary.⁷ Strikingly, we found that DCs exposed to active MMP-2 lost their ability to produce IL-12p70. Moreover, we demonstrated that this lack of IL-12p70 actually played a major role in the MMP-2-dependent T_H2 polarization. Indeed, DCs exposed to active MMP-2 lost their ability to prime T_H2 cells when rhIL-12p70 was supplemented in the culture. Rather, CD4⁺ T cells differentiated into T_H1-like cells. We characterized the mechanism behind MMP-2-dependent inhibition of IL-12p70: by degrading the type I IFN receptor (IFNAR1) on DCs, and subsequently preventing STAT1 phosphorylation, MMP-2 inhibits IL-12p35 subunit transcription (Fig. 1).

Figure 1.

MMP-2-dependent mechanism licensing DCs to prime inflammatory TH2 cells. The left panel (orange) represents « healthy conditions »: activated DCs produce IL-12p35 through IFNAR1 and STAT1 signaling. DCs secrete high levels of IL-12p70 necessary for priming naïve CD4+ T cells into TH1 cells. In the context of melanoma (right panel, green), tumor cells produce and activate MMP-2. Active MMP-2 degrades IFNAR1 on DCs, leading to low levels of STAT1 phosphorylation preventing transcription of IL-12p35. Furthermore, MMP-2 in both active and inactive conformations induces OX40L overexpression on the surface of DCs through triggering of a receptor yet to be determined. Both lack of IL-12p70 and OX40L overexpression by MMP-2-exposed DCs are responsible for priming naïve CD4+ T cells into inflammatory TH2 cells.

Furthermore, we established that MMP-2 induced DCs to overexpress OX40L, which was also involved in the observed T_H2 skewing, as shown by performing the same priming experiments in the presence of a blocking antibody for OX40L: naïve CD4⁺ T cells primed against MMP-2 when OX40L is being blocked, differentiated into IFNγ-secreting T_H1-like cells (Fig. 1). Despite evidence suggesting that OX40 signaling in CD4⁺ T cells can directly induce type-2 lineage commitment by inducing NFATc1, which triggers IL-4 production and subsequent GATA3 expression,⁸ the role of OX40L in T_H2 differentiation is not yet clearly understood. MMP-2-induced type-2 differentiation likely works differently as we showed that IL-4 was not a major determinant in this model. Thymic stromal lymphopoietin (TSLP) can also promote T_H2 differentiation through OX40L-expressing DCs. Interestingly, MMP-2 did not induce DCs to produce TSLP, implying that MMP-2-induced expression of OX40L does not depend on TSLP. TSLP-dependent T_H2 differentiation is believed to involve basophils in several models,^9,10 but we could not detect any basophil activation by MMP-2. We are now exploring how MMP-2 induces OX40L expression and are trying to identify the responsible receptor(s). Surface molecules such as αvβ3, CD91 or MT1-MMP are known to bind MMP-2. However, we excluded any role for these receptors in OX40L induction.

Together, our priming results suggest that both the lack of IL-12 and OX40L overexpression play distinct and additive roles in our model. Moreover, OX40L blockade had an effect even in the presence of IL-12p70, indicating that the absence of IL-12 is not strictly required for T_H2 polarization in this model.

While OX40L expression is induced after exposure to MMP-2 in both active and inactive conformations, IL-12p70 blockade only occurs in the presence of active MMP-2. Aggressive tumors display high levels of activated MMP-2 and therefore should license in situ DCs to express OX40L in the absence of IL-12, creating, an ideal environment to generate T_H2 cells, according to our model. Whether T_H2 polarization in tumors is only a consequence of the environment or actually plays a part in tumor aggressiveness by restricting immune control still needs to be determined.

An additional major finding of our study is that MMP-2 acts as a type-2 conditioner for CD4⁺ T cells specific for other tumor antigens, as we showed for Melan-A/MART-1 and NY-ESO-1. One could imagine that MMP-2-exposed DCs would turn naïve CD4⁺ T cells of any specificity into type-2 cells. This would imply the generation/accumulation of a local type-2 tumor micro-environment, potentially reinforcing itself through second-hand IL-4 secreted by resident T cells.

Unraveling the mechanisms underlying tumor-associated T_H2 polarization, including the one we characterized in this study, opens the way to developing new therapeutic strategies for cancer patients able to induce effective type-1 responses and limit tumor cells escaping the immune system.