Loss of type I IFN responsiveness impairs natural killer cell antitumor activity in breast cancer

Abstract

Competent type I IFN signaling is the lynchpin of most immune surveillance mechanisms and has recently proven critical to the efficacy of several anticancer agents. Expression of the type I IFN receptor, IFNAR, underpins type I IFN responsiveness in all cells and facilitates the activation and cytotoxic potential of lymphocytes, while loss of IFNAR on lymphocytes has previously been associated with tumor progression and poor patient survival. This study underscores the importance of intact type I IFN signaling to NK cells in the regulation of tumorigenesis and metastasis, whereby ablation of NK cell IFNAR1 impairs antitumor activity and tumor clearance. Using a preclinical model of triple negative breast cancer, we identified that intact IFNAR on NK cells is required for an effective response to type I IFN-inducing immunotherapeutics that may be mediated by pathways associated with NK cell degranulation. Taken together, these data provide a rationale for considering the IFNAR status on NK cells when devising therapeutic strategies aimed at inducing systemic type I IFN signaling in breast cancer.

Supplementary Information

The online version of this article (10.1007/s00262-021-02857-z).

Introduction

Immune cell regulation of solid tumor progression and metastasis is now recognized as a central mechanism of cancer elimination [1, 2]. The antitumor effect of antigen-experienced T cells has been well characterized [3, 4] and efforts to harness and therapeutically augment T cell activity has intensified over recent years in response to such findings [5–7]. From the advent and increased utilization of checkpoint inhibitors [7, 8] that prolong T cell effector function, to the customization of CAR T cells to improve T cell trafficking, stability and efficacy [9, 10], the exploitation of T lymphocytes has advanced both our understanding and treatment of several malignancies. Similarly, natural killer (NK) cells, which exhibit antigen-independent cytotoxic functions and can induce cancer cell death, have emerged as important mediators of innate antitumor immunity [11–13] at both primary and secondary sites. Indeed, enhanced NK cell primary tumor infiltration and cytotoxic phenotype has been linked to increased survival in melanoma [14] and bone-metastatic prostate cancer [15]. Additionally, NK cells derived from metastatic breast cancer patients have documented impairment in cytolytic function compared to localized tumors [16]. Originating from the bone marrow, NK cells are found throughout the circulation where they facilitate rapid first-line responses to transformed, damaged or infected cells [17]. Several factors aid NK cell detection of cancer cells. These include tumor cell loss or downregulation of major histocompatibility complex (MHC) class I molecules that normally aid in the formation of memory T cell responses and the frequent elevation of stress ligands recognized by NKG2D and DNAX accessory molecule-1 (DNAM-1) on NK cells, such as MHC class I chain-related protein A and B (MICA/B) and UL16 binding protein 1 (ULBP1) in humans and retinoic acid early inducible 1 (Rae-1) and histocompatibility 60 (H60) in mice [18]. Indeed, the effector function of NK cells is largely determined by the culminative target cell expression of inhibitory (i.e. MHC-I, Ly49A/C/D [17–19]) and activatory (i.e. retinoic acid early inducible 1 (Rae-1), histocompatibility 60 (H60) [17, 20]) proteins, along with co-stimulatory molecules that perform dual functions, such as CD155 [21], depending on the simultaneous binding of additional accessory molecules. This balance between stimulatory and inhibitory signals serves to protect healthy cells while enabling the rapid killing of errant cells, primarily through NK cell production of the type II interferon (IFN), IFNγ and cytolytic granules, perforin and granzymes, which directly mediate apoptosis of the targeted cell [13].

The survival, maturation and activation of NK cells is further modulated by the secretion of cytokines from tumor or stromal cells. One such cytokine heavily implicated in the regulation of both adaptive and innate immune responses is type I IFN [22, 23], which directs a range of immune-mediated antitumor effects. Such effects extend to dendritic cell (DC) priming [24], the release of immunosuppression [25], viral mimicry [26]—along with effector cell modulation and the induction of T cell memory [23]. Moreover, endogenous type I IFN has been shown to be required for effective therapeutic intervention in several cancer types [26–29]. Induction of type I IFN signaling occurs via numerous, proximal pathways involving the activation of pattern recognition receptors that converge on IFN regulatory factor (IRF) 7—the master transcriptional inducer of IFNα [30]. Type I IFNs canonically signal through the binding of the type I IFN cognate receptor complex IFNAR1: IFNAR2, which triggers a cascade that results in the transcription of thousands of IFN regulated genes (IRGs), including Irf7, key cytokines integral to NK cell function (interleukin (IL)-15, IL-18), and many of the ligands and receptors that control NK cell status [22, 31, 32]. While early studies largely focused on the antiviral activity of type I IFN [30, 33], several studies exploring the contribution of type I IFN to innate tumor-directed immunosurveillance have revealed that IFNα enhances NK cell activation [34], regulates NK cell response to cytokines and is requisite for effective tumor cell recognition and lysis [35].

In aggressive, solid cancers, including prostate [23] and breast cancer (BC) [36, 37], loss of type I IFN signaling in the tumor has been linked to fatal disease progression and impaired immune surveillance. Reduced type I IFN signaling and response in a cancer setting can occur via several means [27, 37, 38]. One such mechanism is tumor and stromal downregulation of IFNAR [39], which is associated with poor outcome in colorectal cancer (CRC) [40]. Furthermore, during tumorigenesis, loss of IFNAR expression on DCs [41] and CD8⁺ T cells has been demonstrated in murine models of sarcoma and CRC [40], respectively.

Despite accumulating evidence suggesting that intact type I IFN conveyance may be required for effective control of tumor progression by both adaptive and innate arms of immunity, few studies have explored the contribution of direct NK cell responsiveness to type I IFN on either tumor growth or therapeutic efficacy. Here, we sought to investigate the effect of NK cell loss of IFNAR1 (referred to herein as IFNAR) on tumor cell modulation and how this impacts antitumor responses post-IFN-based therapy in triple negative breast cancer (TNBC), a subtype known to generate poor adaptive immune cell memory responses [42, 43] and in which innate responses may be critical. Using the weakly metastatic, syngeneic model of TNBC, E0771.LMB [44], in hosts harboring a complete or NK cell-specific deletion of Ifnar1 to ablate the effects type I IFN, we show that the cytotoxic function of NK cells is indeed compromised by the loss IFNAR expression. Lack of type I IFN responsiveness not only impairs NK cell killing of TNBC cells but also specifically contributes to the effectiveness of known type I IFN-inducer, poly I:C [23, 28, 45] to deter tumorigenesis. The findings of the current study reveal a role for intact type I IFN signaling in NK cells for effective immunotherapeutic response in a murine model of TNBC and highlights the importance of innate immune populations in therapeutic targeting of cancer.

Materials and methods

Mouse models and cell lines

Animal studies were conducted in accordance with the NHMRC Australian code for the use and care of experimental animals and approved by the La Trobe Animal Experimentation Ethics Committee (AEEC) and the Peter MacCallum Cancer Centre AEEC. Wildtype (WT) female C57BL/6 mice (~ 8 weeks) were obtained from the Walter and Eliza Hall Institute of Medical Research (Melbourne, Vic., Australia). C57BL/6 Ifnar1^−/− mice (herein referred as Ifnar^−/−) were bred inhouse as previously described [36]. C57BL/6 Ifnar1^fl/fl mice (derived by Jörg Kirberg, herein referred to as Ifnar^fl/fl) were crossed with C57BL/6 NKp46^iCre (control^icre) mice (derived by Eric Vivier) and bred inhouse to generate the Ifnar1^−/−NKp46^iCre model (herein referred to as Ifnar^−/−NKp46) in which IFNAR1 is selectively knocked out on NK cells. C57BL/6-Tg (CMV-cre) were also utilized as controls. All in vivo and ex vivo studies were performed with female age-matched mice.

The mCherry⁺ EO771.LMB cell line was derived [44] in the laboratory of Prof. Robin Anderson’s laboratory (Olivia Newton-John Cancer Research Institute, Vic., Australia). B16F10 cell lines were sourced from the ATCC. Murine cell lines were cultured in DMEM (10% FBS) at 37 °C (5% CO₂) and passaged for no more than 2 weeks per experiment using EDTA (0.01% w/v in PBS). The YAC-1 murine leukemia cell line is a prototype NK cell target, previously described [45]. Tumor lines were verified to be mycoplasma negative by the Victorian Infectious Diseases References Lab (Melbourne, Vic, Australia).

In vivo analysis of primary tumor and lung-metastatic burden

For primary tumor experiments, 1 × 10⁵ EO771.LMB cells resuspended in 20 μl PBS were injected into the fourth mammary fat pad (IMFP) of WT, Ifnar^−/−, Ifnar^fl/fl, control^iCre or Ifnar^−/−/NKp46 mice on day 0 under isoflurane anesthesia. Growth was measured using electronic calipers and tumor volume (mm³) calculated as L (mm) x W (mm)²/2. Mice were sacrificed at ~ 14 days post-IMFP injection for primary tumor growth assessment or at ~ 400mm³ for tumor composition analysis. For therapeutic primary tumor studies, Ifnar^fl/fl, control^iCre and Ifnar^−/−NKp46 mice were injected as described above, treated thrice weekly with saline (100 μl) or 25 μg of poly I:C in saline (100 μl) by intraperitoneal (IP) injection and sacrificed at day 24. For lung metastasis experiments, 1 × 10⁶ EO771.LMB cells were injected in 50 μl of PBS into the tail vein (intravenous; IV) of WT, control^iCre or Ifnar^−/−NKp46 mice on day 0. Mice were sacrificed at day 24 post-IV injection due to signs of metastasis. Lung tumor burden was confirmed at endpoint by fluorescence imaging of mCherry on the Lumina XR-III detection (Caliper Life Science). Living image v4.4 software was used for normalization and luminescence quantitation (Caliper Life Science).

Flow cytometry analysis

For analysis of peripheral blood (PB) lymphocytes, tail vein blood (< 100 μl) was collected and subject to RBC lysis, as previously described [28]. For primary tumor analysis, a single cell suspension was obtained using methods previously described [42]. Cells were stained with panels of antibodies: IFNAR-CD8α-PE-Cy7 (53–6.7), CD4-APC-Cy7 (GK1.5), CD69-APC (H1.2F3), NK1.1-BV421 (PK136), TCR-β-FITC (H57-597), IFN-γ-PE (XMG 1.1) all from BD Biosciences; IgG2a-APC/PE/PE-Cy7 (eBM2a), IgG2a-FITC (eB149/10H5), CD27- PE-Cy7 (LG.7F9), NKG2D-PE-Cy7 (CX5), CD107a-PE (eBio1D4B), CD11b-FITC (M1/70) all from eBioscience. For characterization of tumor cell PD-L1, IFNAR, MHC-I and NK cell ligand expression, EO771 and B16F10 cells were stained with PD-L1-PE (MIH5), H2-K^b-FITC (AF6-88.5), KLRG1-FITC (2F1) all from BD Biosciences; IFNAR1-APC/PE (MAR1-5A3), IgG1-PE (P3.6.2.8.1), CD155-PE (TX56), all from Biolegend; and RAE1-PE (Pan Specific) and H60-APC from R&D Systems. Data are represented as lymphocyte percentage unless absolute counts provided. All analysis performed on the BD FACSCanto II and FACSymphony (BD Biosciences) and data analyzed using FlowJo 10.5.0 software (Tree star).

RT-qPCR

For gene expression analysis, RNA was extracted from fresh cell pellets using the QIAGEN RNeasy Plus Mini Kit (Qiagen; 74136) according to manufacturer’s instructions. When required, cells were transfected with poly I:C (10 μg/ml) overnight, prior to RNA extraction. cDNA was generated using the iScript Reverse Transcriptase Supermix cDNA for RT-qPCR kit (BioRad). Quantitative real-time PCR (RT-qPCR) was performed using SsoAdvanced Universal SYBR Green Supermix (BioRad) to quantify murine Irf7 and Mx1 transcript expression on the CFX96 (BioRad) cycler per manufacturer’s guidelines. Gene expression (arbitrary units) relative to housekeeper gene Hprt was calculated as mean relative transcript abundance (RTA) by methods previously outlined [36]. Primers sourced from IDT were used as follows: Irf7 fwd: 5′-CCACACCCCCATCTTCGA-3′; Irf7 rev: 5′-CCTCCGAGCCCGAAACTC-3′; Mx1 fwd: 5′-GATCCGACTTCACTTCCAGATGG-3′; Mx1 rev: 5′-CATCTCAGTGGTAGTCAACCC-3′; Hprt fwd: 5′-GGCCAGACTTTGTTGGATTT-3′; Hprt rev: 5′-ACTGGCAACATCAACAGGACT-3’.

For quantitation of tumor burden in metastatic tissues, multiplex qRT-PCR was performed on purified genomic DNA using SsoAdvanced Universal Probes Supermix (BioRad) to quantify the ratio of mCherry DNA present in tumor cells in comparison to vimentin present in all cells. Gene expression (arbitrary units; AU) relative to housekeeper gene Vimentin was calculated as mean relative transcript abundance (RTA). Primers/probes sourced from IDT were used as follows: mCherry fwd: 5′-GACCACCTACAAGGCCAAGAAG-3′; mCherry rev: 5′-AGGTGATGTCCAACTTGATGTTGA-3′; mCherry probe: 5′-FAM/CAGCTGCCC/ZEN/GGCGCCTACA/3IABkFQ-3′; mVimentin fwd: 5′- AGCTGCTAACTACCAGGACACTATTG-3′; mVimentin rev: 5′-CGAAGGTGACGAGCCATCTC-3′; mVimentin probe: 5′-HEX/CTTTCATGTTTTGGATCTCATCCTGCAGG/TAMRA-3’.

Gene abundance or expression (AU) was calculated as relative tumor burden (RTB) or relative transcript abundance (RTA) by: AU = 10,000/2^ΔCT where ΔCT = Mean Cq (mCherry or gene of interest) − Mean Cq (Vimentin or Hprt), respectively.

IFNα enzyme-linked immunosorbent assay (ELISA)

IFNα ELISA was performed using standard molecular biology techniques. Capture antibody, clone RMMA-1 was used at 1/500 (0.16 μg/ml; PBL Interferon source) prior to detection antibody (1/500 rabbit polyclonal mouse IFNα 32100–1; 4 μg/ml; PBL Interferon source) and tertiary antibody (1/1000 anti-rabbit-HRP; AP182P; Chemicon). In-house generated recombinant murine IFNα gifted from the laboratory of Paul Hertzog. For IFNα induction, cells were stimulated with poly I:C (10 μg/ml) for 24 h.

Interferome

IRG binding sites in RAE-1 were determined through INTERFEROME [46] transcription factor (TF) analysis.

Intracellular cytokine staining (ICS)

Single cell suspensions of primary tumor and peripheral blood samples from WT, Ifnar^−/−, Cre-control or Ifnar^−/−NKp46 mice were generated as described above. ICS for T cell and NK cell IFNγ was performed as previously described [47]. Briefly, ex vivo cell suspensions were incubated with 5 × 10⁴ EO771.LMB cells for 5 h at 37 °C (5% CO₂) in the presence of 10 µg/ml Brefeldin A and stained for cell surface markers before fixation with 1% paraformaldehyde and permeabilization with 0.4% Saponin. Incubation with intracellular IFNγ stain was performed for 30 min, 4ºC prior to analysis by flow cytometry, as described above.

NK cell isolation and cytotoxicity assays

Standard 4 h cytotoxicity assays were completed as previously described [45]. Briefly, splenic NK cells from naïve or poly I:C-activated (25 μg IP; 24 h prior to isolation) control^iCre and Ifnar^−/−NKp46 animals were enriched (NK cell enrichment kit; #19,755; Stem Cell Technologies) according to the manufacturer's instructions and suspended in complete NK cell medium. The assay was performed according to methods previously outlined [45]. Subsequent cytotoxicity data were expressed as percent (%) lysis relative to the spontaneous (target cells alone) and maximum release (1% triton X-100 treated cells) for the target in question with the equation: % lysis ¼ ((sample − spontaneous)/(maximum − spontaneous)) × 100. NK cells were recovered and subject to flow cytometry analysis, described above.

Quantification and statistical analysis

Unless otherwise described in the figure legends, Student’s two-tailed t tests were used to determine significance between groups. MFI calculated as target (mean) − isotype (mean) and normalized MFI calculated as target (median) − isotype (median)/target (robust standard deviation) + isotype (robust standard deviation). Dot plots and histograms are means and all error bars ± SEM with exact n described. GraphPad Prism software (v8.0.2) was used for analyses and p values were deemed significant as follows, * p < 0.05, ** p < 0.005, *** p < 0.0005. **** p < 0.0001.

Results

Intact host IFNAR is critical to the regulation of tumorigenesis and immune cell antitumor function

To assess the contribution of host lymphocyte type I IFN responsiveness on tumorigenesis, wild type (WT) C57BL/6 mice and aged-matched mice lacking IFNAR1 (Ifnar^−/−) were inoculated with the p53 mutant, triple-negative EO771.LMB BC cell line [44] into the fourth mammary fat pad. An increase in primary growth (Fig. 1a) and weight (Fig. 1b) was observed in Ifnar^−/− animals, as has previously been reported in C57BL/6 syngeneic models of CRC [48] and melanoma [49], suggesting a role for intact host IFN signaling in suppressing tumorigenesis. In a separate cohort of mice, immune cell status was assessed in peripheral blood (PB) and the primary tumor, respectively. No difference in peripheral NK cell frequency was observed between WT and Ifnar^−/− EO771.LMB-tumor bearing mice (Fig. 1c), which has also been reported in cancer-free animals [50]. Yet, a decrease in NK cell activation, indicated by NK cell expression of the early activation marker, CD69 [45], was observed in hosts unable to respond to type I IFN (Fig. 1d). Likewise, at the primary tumor level, while NK cell numbers were unchanged between tumors from WT and Ifnar^−/− mice (Fig. 1e), NK cell activation (Fig. 1f, Fig. S1a) and IFNγ production (Fig. 1g, h), were severely compromised in mice deficient in IFNAR (Fig. 1i). Moreover, we observed that greater NK cell cytotoxicity capacity corresponded with decreases in intratumoral naïve NK cells (Fig. S1b), more effector NK cells (Fig. S1c) and more pre-effector NK cells (CD27⁺/CD11b⁻; Fig. S1d), while the most mature NK cell population (CD27⁻/CD11b⁺; Fig. S1e,f) remained unchanged. These data indicate that in a cancer-specific context, the ability of NK cells to respond to transformed cells is influenced by the effect type I IFNs impart on innate cell development. Interestingly, enhanced tumorigenesis and reduced NK cell function in Ifnar^−/− mice occurred concurrently with an increased in tumor-intrinsic expression of the NKG2D ligand and IRG, Rae-1 (Fig. 1j), normally implicated in NK cell activation yet previously associated with enhanced metastasis and poor outcome in BC [51].

Fig. 1.

Loss of host IFNAR1 increases tumorigenesis and impairs host antitumor immunity. 1 × 10⁵ EO771.LMB cells were injected into the fourth mammary fat pad (IMFP) of wild type (WT) and Ifnar^−/− C57BL/6 mice (in which host IFNAR1 is deleted) at day 0 and subsequent alterations to circulating and primary tumor cells assessed. Peripheral blood (PB) was taken at day 7 for FACS analysis and for intracellular staining (ICS) of monocytes was performed at endpoint. Primary tumor a volume (mm³) from WT (n = 4); Ifnar^−/− (n = 5) mice and b weights (mg) of WT (n = 7); Ifnar^−/− (n = 6) mice at experimental endpoint. Flow cytometric analysis of PB c NK (NK1.1⁺) cell and d CD69⁺ NK cell frequency (%); WT (n = 7); Ifnar^−/− (n = 6) e Absolute number (cell #) of primary tumor NK cells (n = 5/group). Flow cytometric analysis of primary tumor (n = 5/group) f CD69⁺ and g IFNγ⁺ NK cell frequency (%) (h representative flow cytometry plots shown), i NK cell IFNAR1 expression and j mCherry⁺ EO771.LMB Rae-1 expression, represented as MFI (mean fluorescence intensity). Flow cytometric analysis of primary tumor (n = 5/group) k. CD4⁺ T cells IFNAR1 expression, l CD4⁺ IFNγ⁺ T cells, m CD8⁺ T cells IFNAR1 expression and n CD8⁺ IFNγ⁺ T cells, represented as frequency (%). p values * < 0.05, ** < 0.005, *** < 0.0005. **** < 0.0001 determined by Student’s t test. Errors bars, SEM

As expected given the known importance of type I IFN on T lymphocyte function and activation [40], loss of host IFNAR1 was associated with decreased intratumoral CD4⁺ and CD8⁺ T cell production of IFNγ (Figs. 1k–n). Altered intratumoral activation in Ifnar^−/− hosts was not accompanied by a change in tumor infiltrating or circulating T cell numbers (Figs. S1g–i,k) or CD4⁺ peripheral activation status (Fig. S1j) in Ifnar^−/− hosts with the exception of a modest increase in peripheral CD8⁺ T cell activation in the Ifnar^−/− group (Fig. S1l). Together, these data suggest that retained type I IFN responsiveness in innate and systemic immune populations at the tumor site is critical to restrict tumorigenesis and promote tumor immune surveillance mechanisms and function.

Loss of IFNAR exclusively on NK cells suppresses activity but does not impact immune-mediated regulation of tumorigenesis

Given the demonstrated role of host IFNAR expression on antitumor immunity, we sought to investigate whether the specific loss of Ifnar1 on NK cells was sufficient to impact tumor development. EO771.LMB cells were injected into the fourth mammary fat pad of Ifnar^−/−NKp46 mice and control^iCre mice and immune and tumor cell readouts taken at timepoints reflected in Fig. 1. In contrast to full host Ifnar1 deletion (Fig. 1a, b), no difference was observed in primary tumor size (Fig. 2a) between groups, suggesting that type I IFN signaling to NK cells is not sufficient to alter tumorigenesis in this model. Reflecting current findings in Ifnar^−/− hosts (Fig. 1c,d), no difference in NK cell frequency was evidenced in circulation (Fig. S2a) or intratumorally (Fig. 2b) despite clear decreases in both peripheral and primary tumor NK cell activation (Figs. 2c-e) and NK cell cytotoxicity (Fig. 2f) in animals in which NK cell IFNAR1 was diminished (Fig. 2g). Evidence of impaired NK cell maturation was demonstrated in both non-tumor-bearing (Fig. S2b-d) and tumor-bearing (Fig. S2e-i) Ifnar^−/−NKp46 hosts, with increases in naïve NK cells and predominant decreased mature NK cell subsets observed on the whole, with the exception of pre-effector NK cells. Contrary to results in Ifnar^−/− hosts, primary tumor cell expression of Rae-1, an IRG that contains both STAT1 and Irf7 binding sites (Fig. S2j), remained unchanged (Fig. S2k).

Fig. 2.

Loss of NK cell IFNAR1 impairs NK cell function, may impact adaptive immune regulation of tumorigenesis and increases lung metastasis. 1 × 10⁵ EO771.LMB cells were injected into the fourth mammary fat pad (IMFP) of control^iCre and Ifnar^−/−NKp46 C57BL/6 mice at day 0 and subsequent alterations to circulating and primary tumor cells assessed. Peripheral blood (PB) was taken at day 7 for FACS analysis and for intracellular staining (ICS) of monocytes was performed at endpoint. a Primary tumor weights (mg) of control (n = 7) and Ifnar^−/−NKp46 mice (n = 7) mice at experimental endpoint. Flow cytometric analysis of b absolute number (cell #) of primary tumor NK cells (n = 6/group). c PB CD69⁺ NK (NK1.1⁺) cell frequency (%). Primary tumor d CD69⁺ (e. representative plots shown) and f IFNγ⁺ NK cell frequency (%) and g IFNAR1 expression, represented as MFI (n = 5/group). Flow cytometric analysis of h CD8⁺ T cell IFNAR1 (MFI; n = 5/group) and i. PB CD69⁺ CD8⁺ T cells, j primary tumor CD69⁺ CD8⁺ T cells, k CD8⁺ IFNγ⁺ T cells, l. PB CD69⁺ CD4⁺ T cells and m primary tumor CD4⁺ IFNγ⁺ T cells, expressed as frequency (%). n RT-qPCR of mCherry (EO771.LMB) DNA expression in the lungs post-intravenous (IV) injection of EO771.LMB cells into WT (n = 9), control^iCre (n = 5) and Ifnar^−/−NKp46 (n = 8) mice at day 24. Representative fluorescence imaging of mCherry (EO771.LMB) in lungs shown for each group. All PB analysis (n = 7/group) and primary tumor analysis (n = 6/group). p values * < 0.05, ** < 0.005, *** < 0.0005. **** < 0.0001 determined by Student’s t test. Errors bars, SEM

While IFNAR on T lymphocytes in Ifnar^−/−NKp46 mice (Fig. 2h; Fig. S2l) remained comparable to controls and peripheral CD8⁺ T cell frequency remained unchanged (Fig. S2m), a significant increase in circulating CD8⁺ T cell activation (Fig. 2i) was observed. This was also reflected at the primary tumor level (Fig. 2j) along with increased IFNγ production by intratumoral CD8⁺ T cells in Ifnar^−/−NKp46 hosts (Fig. 2k) yet was not sufficient to deter tumorigenesis arising from the EO771.LMB model. Conversely, no difference in circulating CD4⁺ T cell frequency (Fig. S2n), activation (Fig. 2l) or primary tumor IFNγ production (Fig. 2m) was observed in Ifnar^−/−NKp46 mice. Beyond the primary tumor, loss of NK cell-specific loss of IFNAR was associated with an increase in metastatic burden in lung compared to control^iCre animals (Fig. 2n), suggesting a critical role for intact type I IFN signaling in BC spread, as has previously been identified in BC [45] and sarcoma [52] models. Surprisingly, no difference in lung metastasis was observed between WT and Ifnar^−/−NKp46 groups, perhaps resulting from alterations incurred during the introduction of iCre, which warrants further exploration and may provide cautionary evidence for investigations in which WT mice rather matched iCre controls are used in Cre/LoxP systems. Overall, these results suggest innate populations are not the sole mediators of IFN-driven immune-regulated tumor rejection but are likely to play a key role in IFN-induced metastasis suppression.

Intact type I IFN signaling in NK cells is required for effective IFN-based immunotherapeutic response

Based on reports that the loss of IFNAR on cytotoxic lymphocytes limits the efficacy of immune-targeted therapies against cancer, we next asked whether the NK cell-specific loss of IFNAR would impact primary tumor growth in response to a systemic immune-activating treatment known to modulate the type I IFN pathway. The TLR3 agonist, poly I:C—previously demonstrated to have anticancer activity [23, 28]—is a potent inducer of type I IFN signaling in many cells including tumor cells, evidenced at both the transcriptional (Figs. S3a-b) and protein level (Fig. 3a) in the EO771.LMB model, without impacting IFNAR expression (Fig. S3c). This extended to the induction of IRGs, namely MHC-I (H2-K^b; Fig. 3b) and the immune checkpoint protein, programmed death-ligand 1 (PD-L1; Fig. S3d), however, did not alter NK cell ligand expression on EO771.LMB cells (Fig. 3c), measured 24 h post-poly I:C treatment. In vivo treatment of Ifnar^fl/fl mice bearing EO771.LMB mammary tumors with poly I:C decreased primary tumor burden compared to saline-treated controls (Fig. 3d). While no associated alterations were observed in NK cell frequency (Fig. 3e) or the expression of late-NK cell activation marker, NKG2D (Fig. 3f), an increase in early NK cell activation was observed (Fig. 3g), independent to changes in CD8⁺ T cell frequency and status (Fig. S3f-g). A similar trend was observed in CMV-cre-control mice, in which IFNAR was intact on all lymphocyte populations (Fig. S3e,h–k), where poly I:C treatment enhanced NK cell activation which correlated with a reduction in primary tumor size (Fig. S3l). Conversely, poly I:C failed to impact primary growth in Ifnar^−/−NKp46 hosts (Fig. 3h), with no associated changes in NK cell or effector T cell number or activation status evidenced at the timepoint assessed (Figs. 3i-k and S3m-n, respectively). These data suggest that type I IFN responsiveness in NK cells, specifically, contribute to the therapeutic efficacy of immune-targeting agents and may impact the treatment regimens of BC patients in which loss of IFNAR is reported.

Fig. 3.

NK cell IFNAR1 is required for effective immunotherapeutic control of tumorigenesis using a systemic IFN-inducing agent. IFNα production (IU/ml) by EO771.LMB cells (a) ± 24 h of poly I:C (10 μg/ml) treatment, as measured by ELISA (n = 4/group). Flow cytometry analysis of EO771.LMB cell b H2-K^b and c NK-cell ligands (CD155, KLRG1, H60, Rae-1) ± 24 h of poly I:C (10 μg/ml) treatment (n = 3/group). d Primary tumor weight (mg) at day 16 following IMFP injection of EO771.LMB and thrice weekly treatment of poly I:C (25 μg delivered IP) in WT (n = 7 saline; n = 7 poly I:C) mice. Flow cytometric analysis of PB e NK (NK1.1⁺) cells, f NKG2D⁺ NK cells and g CD69⁺ NK cells at day 12, expressed as frequency (%). h Primary tumor weight (mg) at day 16 following IMFP injection of EO771.LMB and thrice weekly treatment of poly I:C (25 μg delivered IP) in Ifnar^−/−NKp46 mice (n = 8 saline; n = 6 poly I:C). Flow cytometric analysis of PB i NK (NK1.1⁺) cells, j NKG2D⁺ NK cells and k CD69⁺ NK cells at day 12, expressed as frequency (%). p values ** < 0.005, *** < 0.0005. **** < 0.0001 determined by Student’s t test. Errors bars, SEM

NK cell IFNAR is critical to activated NK cell lysis of cancer cells

To understand how the loss of IFNAR on NK cells interferes with innate regulation of tumor growth in the presence of a type I IFN-inducing agent in vivo, we assessed the in vitro killing potential of NK cells from WT and Ifnar^−/−NKp46 hosts against EO771.LMB cells in addition to B16F10 (melanoma) cells and prototypic NK cell targets, YAC-1 (lymphoma) cells. Despite tumor-intrinsic differences in type I IFN signaling capacity and production (Fig. S4a)—known to directly impact NK cell responses in vivo—along with IFNAR1 (Fig. S4b), MHC-I expression (Fig. S4c) and NK modulating ligands (Fig. S4d-f), both the EO771.LMB and B16F10 models showed comparable susceptibility to NK cell lysis. Under poly I:C-activated conditions, NK cells from treated control^iCre mice, on which IFNAR is intact, could efficiently eliminate up to 85% of EO771.LMB cells (Fig. 4a) while those derived from treated Ifnar^−/−NKp46 hosts exhibited impaired lytic function. Correspondingly, analysis of Ifnar^−/−NKp46-derived NK cells recovered from the cytotoxicity assay revealed a robust decrease in activation (Fig. 4b) and degranulation (Fig. 4c-d), as measured by NK cell expression of CD107a [53], previously shown to be inducible by IFNα [54]. These data were largely mirrored using both B16F10 (Figs. 4e–g) and YAC-1 targets (Fig. S4g-j), with the exception of reduced degranulation, suggesting that the activation of cytolytic pathways associated with perforin and granzyme-B release may be the primary mode of NK cell elimination of EO771.LMB cells in the presence of systemic type I IFN induction. This is further supported by the finding that, in the absence of in vivo poly I:C activation, only 6% of NK cells showed evidence of degranulation against EO771.LMB cells (Fig. 4h) compared to up 15% in the pre-treatment setting (Fig. 4c). Lower percentages were also observed in unstimulated NK cell cultures containing B16F10 (Fig. 4i) and YAC-1 (Fig. S4j) targets, with greater degranulation of NK cells observed in the presence of all cancer cell lines compared to NK cells alone (Fig. 4j), regardless of NK phenotype. Unsurprisingly, poly I:C-stimulated NK cells from control mice demonstrated evidence of activation in the absence of target cells compared to those derived from Ifnar^−/−NKp46 hosts (Fig. S4k). Yet, early activatory responses were diminished to as low 2% in NK cells from unstimulated control^iCre animals (Fig. S4l-m) compared to an average of 15% under poly I:C-stimulated conditions (Fig. 4b, f; Fig. S4h), regardless of target type. This, coupled to the complete lack of activation in NK cells derived from Ifnar^−/−NKp46 hosts, with or without poly I:C stimulation and irrespective of target type, compared to basal NK cell activity (no targets; Fig. S4n), suggests that type I IFN or another IRG-induced factor may be critical for an effective an NK-driven antitumor response, underpinned by NK cell retention of IFNAR.

Fig. 4.

Loss of NK cell IFNAR1 impairs poly I:C-induced tumor cell killing due to decreased NK cell activation and degranulation. In vivo poly I:C-activated (25 μg, IP, 24 h prior to spleen excision) NK cells were purified from control^iCre or Ifnar^−/−NKp46 mice (n = 3/group) and cocultured with calcein-AM-labelled a EO771.LMB cells for 4 h, after which the percentage of target cell lysis was quantitated and NK cells recovered for flow cytometric analysis of b CD69⁺ and c CD107a⁺ NK1.1⁺ cell frequency (%) (d representative flow cytometry plots shown); or e B16F10 cells for 4 h, after which the percentage of target cell lysis was quantitated and NK cells recovered for flow cytometric analysis of f CD69⁺ and g CD107a⁺ NK1.1⁺ cell frequency (%). NK cell degranulation (CD107a⁺) was assessed post-coculture of NK cells purified from naïve control^iCre or Ifnar^−/−NKp46 mice (n = 3/group) with h EO771.LMB cells and i B16F10 cells. j Baseline poly I:C-activated NK cell degranulation (no targets) also shown. p values * < 0.05, ** < 0.005, determined by Student’s t test. Errors bars, SEM

Discussion

Several recent studies have highlighted the importance of type I IFN signaling in antitumor immune surveillance, both at the primary tumor level and in metastasis [23, 28, 42]. Critical to IFN-driven immune cell responsiveness and the subsequent elimination of cancer cells is the expression of IFNAR1, reported to be lost in aggressive, solid cancers [40, 48, 55] and associated with immune suppression and poor immunotherapeutic efficacy in a tumor setting. While previous studies have largely focused on the incidence and impact of IFNAR loss on T cells, the current study reveals a requirement for intact IFNAR on NK cells for optimal antitumor activity in a model of TNBC. It has previously been reported [50] that no notable differences in naïve NK cell frequency occurs between WT and IFNAR-deficient hosts, and that loss of NK cell IFNAR can actually increase effector populations in non-tumor-bearing animals. However, in a tumor-bearing setting, we reveal that effector NK cell expansion and the associated cytotoxic potential of NK cells relies on intact IFNAR expression, illustrating that the loss of type I IFN responsiveness exclusively in NK cells interferes with NK cell status and function in a cancer context, and impairs NK-driven lysis of tumor cells irrespective of cancer type. Importantly, we also provide evidence that NK cell-specific deletion of IFNAR can limit the anticancer effectiveness of type I IFN-modulating agents, which has implications for the implementation of immune-activating therapies in patients in which IFNAR downregulation may have occurred.

A role for NK cells in the regulation of tumor growth has been supported through numerous studies exploring the contribution of innate populations to successful anticancer immunosurveillance [15, 56, 57] and, upon failure, the establishment of an immunocompromised tumor niche [58–60]. Competent type I IFN signaling has been shown to be the crux of an effective and durable antitumor response at both the tumor-intrinsic [23, 36] and host-cell level [24, 41, 45]. Conversely, the loss of type I IFN in the tumor microenvironment (TME) has been linked to the promotion of immune evasion, exclusion and suppression by tumor cells and the concomitant recruitment and expansion of cells with known protumor functions in the TME [23, 61, 62]. Indeed, our finding that molecular ablation of host type I IFN responsiveness diminishes the ability of effector immune cells to combat tumorigenesis in the EO771.LMB model reflects existing literature in other cancers [35]. However, whether this is the direct result of the loss of direct paracrine signaling by IFNα/β in the TME due to IFNAR deletion in effector cells or the suppression of lymphocyte-activating cytokines (i.e. IL-18) from innate populations such as DCs, known to be directly stimulated by type I IFN, remains to be shown. Interestingly, the specific deletion of IFNAR on NK cells did not accelerate primary tumor growth. While not investigated in the current study, this finding may reflect a concomitant downregulation of intratumoral NK cell IFNAR in control mice, previously reported to occur on T cells [55] in tumor-bearing hosts and potentially clarified through the future molecular stabilization of IFNAR on NK cells to block in vivo loss. However, the concurrent loss of NK cell activation and cytotoxic function due to depressed type I IFN sensitivity corresponded with an increase in T cell activity, which has been previously reported in models of infection [63, 64] and cancer [65] when NK cells are dysfunctional or depleted. Yet, the demonstrated increase in CD8⁺ T cell activation in Ifnar^−/−NKp46 mice was not sufficient to impact tumorigenesis in hosts in which type I IFN signaling to NK cells is impaired.

Loss of type I IFN pathway signaling in NK cells has been reported in primary BC patients, irrespective of stage [37], compared to healthy controls, indicating that type I IFN responsiveness in NK cells does, in fact, impact tumor formation and progression. Yet, as our study shows, the exact role that NK cells play in primary tumor development remains to be elucidated, as others have suggested [66]. In contrast, the importance of NK cells in metastasis has been better defined with several studies linking NK cell dysfunction to decreased metastasis-free survival [15, 67]. Our observation that NK-specific deletion of Ifnar1 increases EO771.LMB tumor burden in lung demonstrates that type I IFN responsiveness in NK cells may contribute to the regulation of metastasis. Compared to WT animals, loss of host type I IFN signaling has been previously shown to accelerate metastasis in association with innate and adaptive immune cell dysfunction [45] in two independent models of TNBC. Moreover, increased lung metastasis in EO771.LMB-inoculated mice has previously been reported in animals in which NKLAM (Natural Killer lytic-associated molecule) has been deleted [68], confirming a specific requirement for efficient NK cell activation to successfully deter lung-metastatic outgrowth. Interestingly, the loss of NKLAM on NK cells, which impairs NK cell cytolytic activity, has been associated with a concurrent decrease in type I IFN signaling in NK cells [69]. As such, further dissecting the unique role of type I IFN conduction in NK cells in a metastatic setting alongside determination of the role of NK cell IFNAR in the regulation of the lung-metastatic cascade is worthy of future investigation.

In advanced TNBC, one of the major challenges has been improving the efficacy of immunotherapeutic agents that have shown promise in melanoma [70]. Several recent reports have suggested that the conservation of type I IFN signaling underpins therapeutic success in numerous cancer types [23, 26, 28, 71, 72]. Loss of IFNAR is one mechanism through which IFN signaling is downregulated in cancer, which may occur due to receptor instability or degradation resulting from the dysregulation of proinflammatory cytokines and growth factors, stress and hypoxia [27, 55]. The downregulation of IFNAR on CD8⁺ T cells has been linked to multiple mechanisms of immune suppression and therapeutic resistance, including NK-mediated cytolysis of effector T cells [73] and the promotion of T cell death by suppression IL-2-driven pro-survival pathways [40]. Moreover, loss of T cell IFNAR has been previously evidenced to compromise the effectiveness of type I IFN-inducing anticancer therapies, specifically [40]. Given that many immune-targeted anticancer strategies gaining traction are aimed at enhancing at T cell endurance and memory, either directly or via increased DC priming [74], loss of T cell IFNAR may explain the general lack of therapeutic response in several solid cancers, particularly those in which moderate T cell infiltration occurs. However, data herein reveals that such immunotherapeutic interference also occurs when IFNAR is depleted on NK cells in response to type I IFN-targeted treatment, which may be particularly relevant in cancers with poor T cell profiles, low immunogenicity or T cell resistance. Indeed, a recent study delineating the importance of type I IFN-driven NK cell-mediated tumor clearance in response to cyclic dinucleotide STING (stimulator of interferon genes) pathway activation [75] is consistent with our observations. We cannot comment on the role of T lymphocytes in tumor rejection in poly I:C-treated animals with intact type I IFN signaling at the timepoint assessed, which is an area of future study. However, our data did emphasize a specific role for NK-IFNAR retention in the effective cytolysis of BC cells, as mediated by enhanced degranulation when NK cells are stimulated by a type I IFN-inducing agent. Indeed, several studies have identified degranulation as the primary means of direct NK-mediated tumor cell killing [76, 77] and linked defective cytolytic mechanisms in association with a loss of type I IRGs [72] to cancer progression [78]. Interestingly, we report that more effective degranulation occurred in type I IFN stimulated IFNAR⁺ NK cells against tumor cells with high intrinsic IFNα. Yet, given that the loss of NK cell IFNAR was sufficient to significantly impair tumor cell lysis in two cancer models, irrespective of tumor-inherent differences in both type I IFN and NK cell ligand expression, further studies may better delineate the impact of tumor-driven changes on NK cell degranulation and cancer clearance in response to therapeutic stimulation of type I IFN signaling.

The current study highlights that loss type I IFN responsiveness in NK cells significantly impacts NK cell function and impairs primary and metastatic tumor clearance. Furthermore, that IFNAR deletion on NK cells impairs the efficacy of type I IFN-inducing anticancer therapy. Our findings argue for the employment of NK cell profiling in breast cancer patients who may be at increased risk of progression due to type I IFN dysregulation in innate cells and to stratify individuals for more strategic immunotherapeutic regimens. Future studies incorporating the use of agents that may stabilize IFNAR, including inhibitors of kinases thought to promote IFNAR degradation, may present new methods of overcoming IFNAR downregulation on NK cells in order to improve the application of type I IFN-targeting anticancer therapeutics.

Supplementary Information

Below is the link to the Supplementary Information.

Authors’ Contributions

BSP and NB conceived the study. BSP, KLO and DZ designed the experiments. DZ, NB, KLO and AS performed the experiments. KLO and DZ analyzed and interpreted the data. BSP supervised the overall research. KLO and BSP wrote the paper. KLO, DZ, NB and BSP reviewed and/or edited the paper.

Data availability

The authors declare that the data supporting the findings of this study are available within the paper (and its supplementary information files).

Compliance with ethical standards

Conflict of interest

The authors have no competing interests to declare.