1. Introduction
Breast cancer is the most common cancer in women worldwide and remains a leading cause of death [
1]. While newer therapies have improved survival rates over the past years, they still fail to cure metastatic disease. The effectiveness of standard therapies such as surgery, systemic therapy, and radiotherapy can be limited because tumor cells frequently develop resistance to therapy and subsequently progress [
2]. Over the last years, immunotherapies, including monoclonal antibodies and cellular therapies, have emerged as promising treatment options. Trastuzumab (Herceptin) is a clinically approved antibody that targets human epidermal growth factor receptor 2 (HER2), a receptor that is overexpressed in 15–20% of breast cancer patients [
3]. Next to direct anti-tumor effects, the mechanisms of trastuzumab include antibody-dependent cellular cytotoxicity (ADCC), which is mediated by CD16-expressing immune cells such as natural killer (NK) cells [
4].
As NK cells are a crucial part of the first line of defense against tumors, they have gained increasing interest for cell-based immunotherapies, and ex vivo modifications can help to increase their effector functions in vivo in suppressive tumor microenvironments (TME) [
5]. NK cells selectively kill tumor cells and, unlike T cells, do not require prior sensitization. They get activated when an excess of activation signals, such as stress signals, is received over inhibitory signals. The major inhibitory signals are mediated through human leukocyte antigens (HLA), which are expressed on all nucleated cells and bind to inhibitory killer-immunoglobulin-like receptors (KIRs) and NKG2A on NK cells. HLA ligands can be downregulated on tumor cells to escape T cells, but thereby tumor cells could become more susceptible to NK cells (missing-self hypothesis) [
6]. HLA class I molecules are also critical for NK cell education, a process that is also known as NK cell licensing, and that requires the interaction between HLA ligands and the corresponding KIR or NKG2A receptors on the NK cell [
7]. The more inhibitory receptors that find their ligand are expressed by NK cells, the higher the NK cell responsiveness against target cells, indicating that these receptors play a dual role in regulating NK cell effector functions by licensing NK cells on the one hand and inhibiting effector functions of previously licensed NK cells on the other hand [
7].
In solid tumors, tumor-infiltrating NK cells are generally sparse, and they have been described as less cytotoxic than in healthy individuals [
8]. To increase effectiveness, adoptive transfer of NK cells is tested. For this purpose, NK cells are either derived from the patient (autologous setting) or from healthy donors (allogeneic setting). Our group is focusing on developing effective donor-derived NK cell therapy for cancer. We have previously shown that murine NK cells from an HLA-haploidentical, alloreactive donor can cure mice of 4T1 breast cancer, while NK cells from a syngeneic donor failed to do so [
9]. The concept of alloreactive NK cell donors, leading to an improved outcome compared to non-alloreactive donors, has previously been demonstrated in patients with acute myeloid leukemia that received haploidentical stem cell transplants [
10]. NK cell alloreactive donors expressed licensed KIRs for which the corresponding HLA-ligands were missing in the recipient, similar to endogenous NK cells encountering tumor cells that downregulated HLA; these NK cells are also termed KIR-HLA ligand mismatched NK cells [
11]. Alloreactive NK cells do not attack the recipient tissues, as activating ligands are absent on healthy cells [
11].
Despite much progress with improving their anti-tumor responses, adoptive NK cells are not always effective yet. Tumors themselves can escape immune responses and develop resistance to therapy. In addition, the TME plays a crucial role in suppressing anti-tumor responses because it is frequently an environment with immunosuppressive factors, such as hypoxia, that can mediate inhibitory effects on immune cells including NK cells [
12,
13,
14]. Hypoxic areas with a pO
2 of 2.5 mm Hg (0.3% O
2) or lower were detected in solid malignancies including breast cancer [
15]. In another study, hypoxia, identified by HIF1α expression, was measured in about 40% of breast cancers and associated with poor survival [
16]. Suppressive TME factors such as hypoxia must be overcome to unleash the break in NK cells and unfold their full anti-tumor potential. Such strategies include both potent activation of NK cells and minimizing NK cell inhibition. In a previous study, we observed that hypoxia reduced cytotoxicity and degranulation of unactivated NK cells and demonstrated that the oxygen levels during the kill assay were the most critical influencers of the response [
12]. Importantly, we demonstrated that activation of the NK cells with IL-2 could almost completely restore the NK cell responses, illustrating that IL-2 is a potent activator of NK cells and that NK cells can mediate anti-tumor responses in a hypoxic environment when sufficiently activated [
12]. Another strategy to better activate NK cells is the use of monoclonal antibodies that trigger ADCC via the CD16 NK cell receptor that binds to Fc-fragments of IgG antibodies. We and others showed that NK cells can mediate ADCC in a hypoxic environment against hematological tumors [
17,
18]. Trastuzumab may be a clinically applicable manner to enhance donor NK cell responses against breast cancer. In addition, HLA-mediated NK cell inhibition can be reduced by blocking the interactions of inhibitory receptors and their corresponding HLA ligands through blocking antibodies (e.g., anti-NKG2A antibody Monalizumab or anti-KIR antibody Lirilumab) or by selecting genetically different NK cell donors with a KIR–HLA mismatch [
11,
19].
In this study, we investigated whether the combination of ADDC-triggering and the selection of KIR–ligand mismatched NK cells can enhance the NK cell anti-tumor response to human breast cancer in clinically relevant settings. To address our research question, we used the anti-HER2 antibody trastuzumab and determined the cytotoxic potential of IL-2-activated, donor-derived NK cells in breast cancer models in the presence of hypoxia, an immunosuppressive factor frequently present in solid tumors. In addition, we evaluated the degranulation potency of KIR–ligand mismatched NK cells in this setting.
2. Materials and Methods
2.1. Cell Culture and Animals
The breast cancer cell line MCF7, purchased from ATCC, was cultured in an EMEM medium (ATCC, Manassas, VA, USA) supplemented with 10 µg/mL insulin, 10% fetal calf serum (FCS), 100 U/mL penicillin, and 100 µg/mL streptomycin (1% Pen/Strep, Thermo Fisher Scientific, Waltham, MA, USA). The breast cancer cell line SKBR3, purchased from DSMZ, was cultured in McCoy’s 5A medium (Gibco), supplemented with 20% FCS and 1% Pen/Strep. The HLA class I-negative cell line K562, purchased from ATCC, was used as a control cell line and cultured in IMDM medium (Gibco), supplemented with 10% FCS and 1% Pen/Strep. The cells were cultured at 37 °C in an incubator containing 21% O2 and 5% CO2. For hypoxia exposure, the cells were cultured at 37 °C in a hypoxic chamber containing 0.2% O2 and 5% CO2 (InvivO2 1000, Ruskinn Technology Ltd., Bridgend, UK). NOD SCID gamma (NSG) mice were injected with 1 × 106 MCF7 cells subcutaneously into the flank. The local animal ethical committee had approved the experiments. Primary tumors were harvested and dissociated into single-cell suspension using the Tumor Dissociation Kit human (Miltenyi Biotec, Bergisch Gladbach, Germany) together with gentleMACS Dissociator (Miltenyi). Dissociated tumor cells were frozen until assays were performed.
2.2. NK Cell Culture
NK cells were isolated from healthy anonymous buffy coats (Sanquin blood bank, Maastricht, The Netherlands). The use of buffy coats does not need ethical approval in the Netherlands under the Dutch Code for Proper Secondary Use of Human Tissue. NK cell donors with an HLA C1+ C2+ Bw4+ genotype and expression of KIR2DL1, KIR2DL2/3, and KIR3DL1 receptors were used to obtain NK cells licensed for all three KIRs. Peripheral blood mononuclear cells (PBMCs) were isolated by density gradient centrifugation using Lymphoprep (Axis-Shield, Dundee, Scotland). From the PBMCs, NK cells were obtained by negative selection, utilizing an NK cell isolation kit and the MACS separation column system (Miltenyi Biotec) according to the manufacturer’s protocol. The NK cells were subsequently activated with 1000 U/mL IL-2 (Proleukin, Novartis, Basel, Switzerland) and cultured overnight in RPMI-1640 medium (Gibco), supplemented with 10% FCS, 100 U/mL penicillin, and 100 µg/mL streptomycin at 37 °C in an incubator containing 21% O2 and 5% CO2. For some of the primary breast cancer cells, expanded NK cells were used as effector cells. NK cells were expanded from CD3-depleted PBMCs in SGCM medium supplemented with 10% FCS, 1% Pen/Strep, and 1000 U/mL IL-2. After 17 days of expansion, NK cells were frozen and, prior to the experiment, thawed and recovered overnight in the presence of IL-2.
2.3. Cytotoxicity Assay
The cytotoxic potential of NK cells against breast cancer cell lines was determined in flow-cytometry based assays. The target cells SKBR3, MCF7, and K562 were either labeled with CellTrackerTM CM-DiI Dye or with CellTrackerTM Deep Red Dye (both Thermo Fisher Scientific) and were incubated at 37 °C either with 21% O2 or with 0.2% O2. After 16 h incubation, the target cells were harvested and counted and 2 × 104 cells were plated per well in a 96-well plate. The target cells were pre-incubated with 1 µg/mL trastuzumab (Roche, Basel, Switzerland) or with a culture medium, as a control, for 30 min. The hypoxia-exposed cells were kept at 0.2% O2 for all steps. IL-2-activated NK cells were harvested and washed before they were co-cultured with the target cells in a 1:1 or a 5:1 Effector:Target (E:T) ratio at 37 °C either with 21% O2 or with 0.2% O2. After 4 h of co-culture, plates were put on ice to stop the reaction. The cells were washed with PBS (Sigma-Aldrich, Munich, Germany) and stained for dead cells with Live/Dead® Fixable Aqua Dead Cell Stain Kit (Thermo Fisher Scientific) for 30 min on ice. The assay was analyzed by flow cytometry. Specific cytotoxicity was calculated as follows: (% dead tumor cells − % spontaneous tumor cell death)/(100% − % spontaneous tumor cell death) × 100. Spontaneous tumor cell death in the presence of trastuzumab was used to calculate specific cytotoxicity in the conditions with trastuzumab.
2.4. CD107a Degranulation Assay
The target cell lines SKBR3, MCF7, and K562 were incubated for 16 h at 37 °C either with 21% O2 or with 0.2% O2 and subsequently harvested for the CD107a degranulation assay. In each well, 105 target cells were plated and pre-incubated with 1 µg/mL trastuzumab (Roche) or, as a control, with a culture medium for 30 min. IL-2-activated NK cells were harvested, washed, and subsequently co-cultured with the target cells in a 1:1 E:T ratio at 37 °C either with 21% O2 or with 0.2% O2. To each well, 5 µL of CD107a-Horizon V450 antibody (Miltenyi) was added. After 1 h of co-culture, Monensin (BD Biosciences, San Jose, CA, USA) was added to prevent reinternalization of CD107a, and after another 3 h of co-culture, the plates were put on ice to stop the reaction. The cells were washed with PBS and first stained with Live/Dead® Fixable Aqua Dead Cell Stain Kit (Thermo Fisher Scientific) for 30 min on ice before surface staining with the following antibodies was performed for 30 min on ice: anti-CD3-APC-Vio770 (BW264/56), anti-CD56-PerCP-Vio700 (REA196), anti-KIR2DL1-APC (143211), anti-KIR2DL2/3-PE (DX27), anti-KIR3DL1-FITC (DX9), and anti-NKG2A-PE-Vio770 (REA110). The assay was analyzed by flow cytometry.
2.5. KIR-Ligand Mismatched and Matched NK Cells
The HLA class I genotype of SKBR3 and MCF7 was determined by Luminex-SSO. The genotype for SKBR3 cells was HLA C1+ C2− Bw4−, and the genotype for MCF7 cells was HLA C1− C2+ Bw4+. Flow cytometry was performed to determine the phenotypic expression of HLA-C and Bw4 as described below. The matched and mismatched NK cell populations were identified based on the genotypic expression of HLA C1, HLA C2, and HLA Bw4, as well as the phenotypic expression of Bw4. For SKBR3, KIR-ligand-matched NK cells were KIR2DL2/3+ NK cells, and KIR-ligand-mismatched NK cells were KIR2DL1+, KIR3DL1+, and KIR2DL1+ KIR3DL1+ double-positive NK cells. For MCF7, KIR-ligand-matched NK cells were KIR2DL1+ NK cells, and KIR-ligand-mismatched NK cells were KIR2DL2/3+, KIR3DL1+, as well as KIR2DL2/3+ KIR3DL1+ double-positive NK cells.
2.6. Generation of F(ab’)2 Fragment
The Pierce F(ab’)2 Preparation Kit (Thermo Fisher Scientific) was used according to the manufacturer’s protocol to generate an F(ab’)2 fragment of the trastuzumab antibody. The following secondary antibody was used to stain for the F(ab’)2 fragment or trastuzumab: Alexa Fluor 647 AffiniPure F(ab’)2 Fragment Goat Anti-Human IgG, F(ab’)2 fragment specific (Jackson ImmunoResearch, Cambridgeshire, UK).
2.7. Primary Human Breast Cancer Cells
Primary human breast cancer tissue was obtained from the Maastricht Pathology Tissue Collection. Collection, storage, and use of tissue and patient data were performed in agreement with the “Code for Proper Secondary Use of Human Tissue in the Netherlands” and have been approved by the local ethics committee. The tissue was immediately stored in MACS Tissue Storage Solution (Miltenyi) until processing. To dissociate single cells, the Tumor Dissociation Kit human (Miltenyi) was used together with a gentleMACS Dissociator (Miltenyi) according to the manufacturer’s instructions. Subsequently, the cell suspension was enriched for tumor cells by negative selection using the Tumor Cell Isolation Kit, human (Miltenyi). Tumor cells were identified by PanCytokeratin-AF488 (C11, ThermoFisher) and the purity of tumor cells was at least 70% PanCK+ cells, with one exception of 44% PanCK+ cells. For cytotoxicity and CD107a assays, either freshly isolated or IL-2-expanded NK cells were used, and co-cultures were performed for a duration of 16 h.
2.8. Flow Cytometry
To determine HER2 and HLA surface expression, SKBR3 and MCF7 cells were stained with Live/Dead® Fixable Aqua Dead Cell Stain Kit (Thermo Fisher Scientific) for 30 min on ice, followed by staining with HER2-APC (Neu24.7, BD), HLA-C-PE (DT9, BD), HLA-Bw4-PEVio770 (REA274, Miltenyi), HLA-E-PE (3D12, Thermo Fisher Scientific), HLA-ABC-APC (G46-2.6, BD), or HLA-ABC-PE (REA230, Miltenyi), or matched isotype controls for 30 min on ice. All analysis by flow cytometry was performed with BD FACS Canto II. Data were analyzed with FlowJo v10.6.1 64-bit software, (TreeStar, Ashland, OR, USA).
2.9. Statistics
The statistical analysis was performed with GraphPad Prism 8.4.3 software (Graphpad Software, San Diego, CA, USA) using paired, non-parametric t-tests (Wilcoxon matched-pairs signed-rank test).
4. Discussion
With the aim to develop effective NK-cell-based therapies against breast cancer, we investigated the combination of the monoclonal antibody trastuzumab and KIR-ligand-mismatched donor NK cells to improve the responses against breast cancer in an immunosuppressive environment. We found that KIR-ligand-mismatched NK cell subsets degranulated stronger against breast cancer than their matched subsets and that trastuzumab activated all NK subsets when HER2 was overexpressed. Importantly, our observations were consistent in a hypoxic environment, emphasizing that the combination of reducing the activation threshold for NK activation by the selection of KIR-ligand-mismatched donors and maximizing NK cell activation with an ADCC-inducing antibody can potentiate the NK cell anti-breast cancer response.
In our experiments with cell lines, we used hypoxia to mimic one of the important factors in the TME. Severe hypoxia has namely been observed in the core of tumors from breast cancer patients and has been associated with metastasis formation and thereby with severity of disease [
15]. Moreover, we and others showed that hypoxia can reduce effector functions of unactivated NK cells [
20] and that NK cell activation with high-dose IL-2 could restore NK cell cytotoxicity against multiple myeloma [
12]. Here, we report that the IL-2-activated NK cell also remained functional against breast cancer.
For SKBR3, we observed a small reduction in NK cell killing potential under hypoxia. This reduction could suggest an adaption to hypoxia in the SKBR3 target cells contributing to resistance to NK cells, e.g., by a reduction in activating ligands or enhanced expression of inhibitory ligands. Resistance could be acquired by the NK cells, e.g., via altered receptor expression, which would result in less efficient NK cell activation. In our previous study on hypoxia, we observed a minor decrease in expression of CD16 and NKG2D but not in any of the other common NK cell receptors [
12]. Moreover, on multiple myeloma cell lines, we did not see a change in the expression of stress-induced activating ligands MICA/B and ULBP1/2 [
12]. In the present study, we observed that expression of HLA class I was not altered by hypoxia and that IL-2-activated NK cells were potently degranulating under hypoxia. We therefore anticipate that NK cells were activated under hypoxia in this breast cancer setting and that resistance would mainly occur inside SKBR3 target cells. Baginska et al. also demonstrated that hypoxia-induced resistance was not caused by defective recognition of targets cells but by autophagy, leading to the breakdown of NK cells’ cytotoxic granules in hypoxic breast cancer cells [
14]. To develop future strategies to improve the response in hypoxic tumors, it could be interesting to combine NK-cell-based strategies as proposed in this study with autophagy-reducing strategies such as chloroquine.
Although hypoxia is an important TME factor, our reductionist approach with cell lines did not take the full complexity of the TME into account, which may lead to an underestimation of the impact that hypoxia may have in combination with other TME factors. Our model with primary breast cancer samples showed considerable variability in sensitivity to NK cells between the different patients, illustrating the potential importance of other TME factors. The exposure of the primary breast cancer cells to the TME in patients is therefore an important advantage of our model, as it enabled us to assess the effects of TME-induced resistance mechanisms to NK cells more comprehensively. Examples of such cell TME resistance mechanisms are changes in the level of autophagy or expression levels of activating or inhibitory ligands on the tumor cells. However, a limitation of the model could be the relatively harsh digestion procedure, which could affect tumor cell viability as well as surface expression of activating ligands MICA and MICB [
21], leading to underestimation of the contribution of these ligands. Moreover, it does not predict the direct effects that soluble TME factors and other tumor-associated cells can have on NK cells, which illustrates the necessity to develop more complex in vitro or in vivo models mimicking the multifactorial TME in patients to further evaluate the impact of the TME on NK cell efficiency.
In our study, trastuzumab enhanced NK cell degranulation much more vigorously against HER2-amplified targets compared to non-amplified targets in vitro, and it did not enhance NK cell functions against non-amplified primary breast cancer cells, suggesting that the HER2 expression level is important for the potential of the ADCC response. Our observations are in agreement with previous studies reporting that the effect of monoclonal antibodies such as daratumumab and trastuzumab is specific to target cells expressing high levels of antigen [
17,
22,
23]. HER2 expression can be modulated through receptor internalization [
24]. Recently, a study elegantly showed that HER2 internalization could be prevented by endocytosis inhibitors, resulting in an improved ADCC response [
25]. In our study, HER2 could be detected after 4 h incubation with trastuzumab, and NK cells were potently degranulating in response to HER2 amplified cells, indicating that HER2 endocytosis cannot be the only factor limiting the ADCC response we observed.
We report a large discrepancy between a potent increase in NK cell degranulation against HER2-amplified target cells but a less pronounced effect on the actual killing of the target cells induced by trastuzumab. We have not previously noticed such a large discrepancy between degranulation and cytotoxicity. When using the same experimental setup with multiple myeloma cells as targets, we observed an ADCC effect with the anti-CD38 antibody daratumumab that was more similar to the degranulation effect [
17]. Although further studies are required to unravel the mechanisms behind this observation, the strong degranulation with trastuzumab could indicate a very potent cytokine release by the NK cells. A profound release of cytokines, such as IFN-γ and TNF-α, by NK cells can boost the overall anti-tumor response by stimulating antigen presentation, Th1 polarization, and CD8 effector functions [
26]. Studies evaluating the relation between cytokine production and degranulation (CD107a) on a single-cell level demonstrated that NK cells either produce cytokines, express CD107a, or do both upon activation with target cells [
27]. Cytokine production has also been shown to occur in an HLA-dependent manner, as KIR-ligand-mismatched NK cells had higher intracellular IFN-γ levels than KIR-ligand-matched cells upon activation with L721.221 target cells [
28]. The trastuzumab-induced degranulation of NK cells described in our study suggests that the combination of trastuzumab and IL-2-activated KIR-ligand-mismatched NK cells may also trigger a stronger production of cytokines, which could contribute to improved adaptive anti-tumor immunity.
In our study, we confirmed the functional relevance of HLA class I as an important inhibitory immune checkpoint for NK cell effector functions in breast cancer. As for many cancers, HLA class I expression can be partially or completely downregulated in breast cancer [
29,
30,
31]. In our study, we mimicked the absence of HLA class I by using KIR-ligand-mismatched NK cells. In the presence of trastuzumab, KIR-ligand-mismatched NK cells remained the stronger degranulating subset against low HER2-expressing MCF7 cells, and when HER2 expression was high as in SKBR3, trastuzumab led to vigorous degranulation in all subsets, suggesting that the inhibitory ligand HLA class I matters less for the NK cell responses when trastuzumab is present. Muntasell et al. investigated predictive biomarkers for the response to trastuzumab treatment and found that patient stratification based on high HLA class I expression together with infiltrating NK cells improved prediction of better responses [
32]. These data are in line with our results showing that NK cells can be effective against HLA class I
+ tumors in combination with trastuzumab. Muntasell’s study also supports that both NK cells and T cells are major contributors to the anti-breast cancer response of HER2
+ patients.
Our results emphasize that selection of NK cell donors based on their KIR expression and HLA genotype can be an effective way to reduce inhibition in a setting with adoptive transfer of donor NK cells, which could be particularly useful for HER2-negative patients. However, since HER2 expression can be heterogeneously expressed within one tumor or be downregulated in response to trastuzumab treatment [
33], KIR-ligand-mismatched NK cells may also be advantageous for HER2-positive patients treated with trastuzumab. Selection of KIR-ligand-mismatched donors is possible for patients that lack at least one of the three HLA epitope groups binding to inhibitory KIRs, which is the case for circa 70% of the population [
34]. Although the here-evaluated KIR-ligand-mismatched subsets comprise a rather small percentage of total NK cells, KIR-ligand mismatching helps to reduce inhibition in these subsets, which can nonetheless be beneficial in a TME where many factors can limit NK cell anti-tumor responses. Our analysis indicated that the NKG2A
+ KIR-ligand-mismatched NK cells performed as well as or slightly better than their NKG2A
− counterpart against HLA-E negative target cells. This observation implies that the NKG2A
+ KIR-ligand-mismatched subsets can also be considered fully mismatched against HLA-E negative targets, which can significantly enlarge the mismatched NK cell population since NKG2A is expressed on 20–80% of NK cells. In breast cancer patients, HLA-E expression was detected in 20–50% of samples [
30,
35]. High HLA-E expression can, however, inhibit NK cell responses [
36]. We did not have breast cancer cells available that expressed HLA-E. However, in a previous study with the same experimental setup, we showed that high HLA-E levels in multiple myeloma cells inhibited NK cells, while low HLA-E levels were not sufficient to do so [
36]. The inhibitory potential of HLA-E has been established in multiple tumor models [
37]. Therefore, it seems likely that high HLA-E expression will also limit NK cell anti-breast-cancer responses, and in those cases, blocking antibodies such as the anti-NKG2A antibody monalizumab might need to be considered. It would be relevant to further evaluate the additive effect of this approach on primary breast cancer with high levels of HLA-E.
Based on our results, we envision that alloreactive donors should be selected for NK cell-based therapies against HLA class I+ breast cancer. Multiple clinical trials showed that infusion of alloreactive NK cells is well tolerated when combined with lymphodepleting chemotherapy to suppress the host’s immune response. It needs to be assessed whether KIR-ligand mismatching can further enhance trastuzumab-induced NK cell degranulation in a setup that better represents the complex TME where many factors can limit NK cell anti-tumor responses. Genetic manipulation of NK cells could be a second strategy to limit inhibitory signaling via HLA, which could be done by CRISPR/CAS9-mediated knock-out of inhibitory receptors such as NKG2A and KIR. Another attractive strategy could be temporarily reducing expression levels of the inhibitory receptors via silencing RNAs. Given the critical role of NKG2A and KIR in NK cell licensing, transient reduction of receptor expression may be especially relevant when full receptor knock-outs negatively influence NK cell potency.