TNFR2/BIRC3-TRAF1 signaling pathway as a novel NK cell immune checkpoint in cancer

Abstract

Natural Killer (NK) cells control metastatic dissemination of murine tumors and are an important prognostic factor in several human malignancies. However, tumor cells hijack many of the NK cell functional features compromising their tumoricidal activity. Here, we show a deleterious role of the TNFα/TNFR2/BIRC3/TRAF1 signaling cascade in NK cells from the tumor microenvironment (TME). TNFα induces BIRC3/cIAP2 transcripts and reduces NKp46/NCR1 transcription and surface expression on NK cells, promoting metastases dissemination in mice and poor prognosis in GIST patients. NKp30 engagement, by promoting the release of TNFα, also contributes to BIRC3 upregulation, and more so in patients expressing predominantly NKp30C isoforms. These findings reveal that in the absence of IL-12 or a Th1-geared TME, TNFα can be considered as a negative regulatory cytokine for innate effectors.

Keywords: BIRC3; NK cells; TNFα; TRAF1; cancer; immune checkpoint; immunity.

Figure 1.

Comprehensive transcriptomics analyses revealing TRAF1 and cIAP2/BIRC3 in NK TILs from GIST. (A-B) Heat map representation of microarray analyses of CD3-CD56^bright NK cells from paired blood and tumors (A) in NKp30 AB versus C profiles (B) on about 20,000 gene products, the most significant hits contrasting the two groups according to the median of the whole cohort being depicted. Of note, for the genes products identified by several distinct probe sets, the most variant was retained in the model. p-values (one-sided differential test) are indicated on the right side for each significant gene product. (C-D) RT-PCR evaluation of the transcription of each gene product pre-selected from the microarray analyses, comparing blood versus tumor expression in at least 3 GIST bearing patients (C) or NKp30AB versus NKp30C profiles (D). (E) Scatter plot, Pearson correlations coefficient and p-value between TRAF1 and BIRC3 transcript abundancy in tumors. Each dot represents one HV or patient. Wilcoxon matched pairs test: * p < 0.05, ** p < 0.01, *** p < 0.001.

Figure 2.

NK cell stimulation by TNFα or NCR3 triggering leads to markedly increased transcription of Traf1 and Birc3. (A-C) Regulation of expression of TRAF1 and BIRC3 in HV and melanoma patients. RT-PCR of both transcripts in blood-derived-NK cells, performed 12 h post-stimulation with 50 ng/mL of TNFα or after cross-linking of NCR3/NKp30 and CD16 (or IgG2a and IgG, respectively, isotype control Abs). Graphs show the relative expression normalized to β2 microglobulin expression in HV (A, left panel for TRAF1 and right panel for BIRC3) and melanoma patients (C), with or without blockade of the TNFR2 by specific neutralizing Abs or IgG1 isotype controls (B) in HV (upper panel) and MM (lower panel). (D-E) Spearman correlations between TRAF1 and BIRC3 transcript abundancy in blood NK cells post-TNFα stimulation (D) or NKp30 engagement (E). Each dot represents one HV or patient. Student paired t-test for A, Wilcoxon matched pairs test for (B-E): * p < 0.05, ** p < 0.01, *** p < 0.001.

Figure 3.

TNFα markedly and selectively downregulates NCR1 expression. (A-C) Regulation of expression of NCR1 in purified blood NK cells from HV (A) and melanoma patients (B) following 12 h-stimulation with rTNFα at 50 ng/mL or NKp30 engagement or rTGFβ at 5 ng/mL (C). Each dot represents one HV or patient. (D-E) NKp46 protein expression on the cell surface of NK cells as assessed in flow cytometry at 48 h post-stimulation at increasing dosing of rTNFα (5 or 50 ng/mL) or rTGFβ (5 ng/mL) in concatenated data from at least 6 HV (left panel) and 6 metastatic melanoma (MM, right panel) (D) and in two representative individuals in overlayed mean fluorescence intensities (E), one treated with rTNFα and TGFβ (upper panel) and another with different concentration of rTNFα (lower panel). Of note, neutralizing anti-TNFR2 Ab inhibits the down regulation of NKp46 at the surface of MM. MFI were superimposable using an isotype control staining in all conditions of stimulation (not shown). Student paired t-test for A, Wilcoxon matched pairs test (B-D): * p < 0.05, ** p < 0.01, *** p < 0.001.

Figure 4.

Anti-correlations between BIRC3 and NCR1 in NKp30C individuals. (A-B) Th1 cytokine release from NK cells stimulated through NKp30 engagement (A) or rTNFα (B) in AB versus C NKp30 profile as determined by ELISA at 12 h (A) and 24 h (B). (C) RT-PCR of all three transcripts (TRAF1, BIRC3, NCR1) in blood –derived-NK cells, performed 12 h post-stimulation with 50 ng/mL of TNFα according to the NKp30 isotype profiles. (D) Spearman correlations between NCR1 and BIRC3 transcript abundancy in blood NK cells post-TNFα stimulation in all patients, separating NKp30 AB from NKp30C individuals. Wilcoxon matched pairs test (A-B) Student paired t-test (C): * p < 0.05, ** p < 0.01, *** p < 0.001.

Figure 5.

TNFα-mediated downregulation of NCR1 in intratumor GIST NK cells. Same experimental settings as in Fig. 4, after sorting of NK cells from fresh GIST tumors. RT-qPCR expression of TRAF1, BIRC3, as well as NCR1 after stimulation with rTNFα at 12 h of ex vivo stimulation in 3 GIST tumors. (B) Membrane expression of NKp46/NCR1 after 48 h incubation of NK TIL with rTGFβ or TNFα in mean fluorescence intensity. MFI were superimposable using an isotype control staining in all conditions of stimulation (not shown).

Figure 6.

Regulation of metastases dissemination by interfering with the TNFα/TNFR2 interaction. (A) Transcription of Traf1, Birc3 and Ncr1 gene products in mouse splenic NK cells post-stimulation with rTNFα as assessed by RT-PCR at 12 h. Similar experimental settings as with human blood or tumor NK cells. (B) Same settings as Fig. 6A but in TNFR2 KO mice. (C-D) B16F10 melanoma lung metastases dissemination post i.v. injection at 3 weeks without treatment (PBS injection) or with daily inoculation of 1 µg of mouse rTNFα for 3 weeks (D, left) or with an inhibitor of TNFα (Etanercept) twice a week for 3 weeks starting at day 1 post-tumor inoculation (D, right). Representative micrograph pictures of metastatic lungs in each group (C) and concatenated data from 2 experiments yielding similar results (C) at mouse sacrifice. Wilcoxon matched pairs test (A), Mann Whitney test (C, D): * p < 0.05.