Abstract
Tumor-draining lymph nodes (TD-LNs) are the first site of metastasis of breast cancer. Natural killer (NK) cells that infiltrate TD-LNs [including noninvaded (NI) or metastatic (M)-LNs from breast cancer patients] and NK cells from healthy donor (HD)-LNs were characterized, and their phenotype analyzed by flow cytometry. Low percentages of tumor cells invaded M-LNs, and these cells expressed ULBP2 and HLA class I molecules. Although NK cells from paired NI and M-LNs were similar, they expressed different markers compared with HD-LN NK cells. Compared with HD-LNs, TD-LN NK cells expressed activating DNAM-1, NKG2C and inhibitory NKG2A receptors, and exhibited elevated CXCR3 expression. CD16, NKG2A, and NKp46 expression were shown to be increased in stage IIIA breast cancer patients. TD-LNs contained a large proportion of activated CD56brightCD16+ NK cells with high expression of NKG2A. We also showed that a subset of LN NK cells expressed PD-1, expression of which was correlated with NKp30 and NKG2C expression. LN NK cell activation status was evaluated by degranulation potential and lytic capacity toward breast cancer cells. NK cells from TD-LNs degranulated after coculture with breast cancer cell lines. Cytokine-activated TD-LN NK cells exerted greater lysis of breast cancer cell lines than HD-LN NK cells and preferentially lysed the HLA class Ilow MCF-7 breast cancer cell line. TD-LNs from breast cancer patients, thus, contained activated lytic NK cells. The expression of inhibitory receptor NKG2A and checkpoint PD-1 by NK cells infiltrating breast cancer–draining LNs supports their potential as targets for immunotherapies using anti-NKG2A and/or anti–PD-1.
Introduction
Natural killer (NK) cells are key effectors of the cytotoxic antitumor responses and strong inducers of the adaptive immune response. Hence, NK cells could represent important targets for the immunotherapy of hematologic malignancies and solid tumors. They can be efficiently expanded in vitro and used in autologous or allogeneic NK cell–based immunotherapies (1, 2). Besides their ability to kill tumor cells without prior sensitization, NK cells are also involved in the priming of type 1 T helper-cell (Th1) responses and are considered the major source of IFNγ in vivo (3, 4). They differentiate in the bone marrow and in peripheral lymph nodes (LNs) and reach circulation where they represent 5% to 15% of the blood lymphocytes. Two major NK subpopulations are classically described in humans: CD56dimCD16+ and CD56brightCD16− cells. The CD56dim population predominates in blood (90% of NK cells), as well as at the site of inflammation, and exhibits a high cytotoxic potential, whereas CD56bright NK cells are more abundant in secondary lymphoid tissues, such as LNs (95% NK cells), have the ability to produce cytokines upon activation, and are considered precursors of the mature CD56dim counterpart (5).
NK cell activation depends on a fine balance between activating and inhibitory signals that determines whether the target cell will trigger the NK cell killing program. Activation of NK cells is triggered by numerous receptors such as natural cytotoxicity receptors (NCR): NKp46 and NKp30, expressed by resting NK cells, and NKp44, induced after cytokine activation (6, 7). The NCRs are implicated in the lysis of various tumor cells, including breast cancer (8). The NK group 2 member D (NKG2D) receptor is expressed by a majority of NK cells. It binds MHC class I polypeptide-related sequence (MIC)-A/B molecules and UL16-binding proteins 1-6 (ULBP1-6) induced on stressed cells (9). The DNAX accessory molecule-1 (DNAM-1) is an adhesion molecule belonging to the immunoglobulin superfamily. Its ligands are nectin (CD112) and nectin-like proteins (CD155/PVR), and it has been shown that NKG2D and DNAM-1 also promote tumor cell elimination (10). NK cell activation is also controlled by HLA class I–specific inhibitory NK receptors. The C-type lectin CD94/NKG2A receptor binds to HLA-E molecules sensing the global HLA class I molecules on the target. The killer immunoglobulin-like receptors (KIR), which are expressed on the NK CD56dim subset, bind to particular HLA class I alleles (7, 11). The inhibitory receptors CD96 and TIGIT, belonging to the emerging family of DNAM-1 receptors, are also expressed by certain NK cells. Both compete with CD226/DNAM-1 for CD155 binding and limit NK cell function by direct inhibition (12, 13).
Studies have demonstrated that NK cells infiltrate a variety of healthy and tumor tissues (14, 15), and they can control metastasis in experimental tumor models and human tumors (2, 16–18). However, tumors may escape NK cell immunosurveillance by different mechanisms, and tumor-infiltrating NK cells often display phenotypic alterations such as NCR or NKG2D downregulation (19). Targeting NK immunosuppression is required to restore the control of metastatic breast cancer cells by NK cells (20). In that context, investigating NK cells from metastatic lesions may improve understanding of their implication in breast cancer tumor control and lead to development of new NK cell–based strategies. Because LNs represent the first site of metastasis of breast cancer tumors, we characterized NK cells infiltrating tumor-draining (TD)-LNs from human breast cancer patients for their phenotype and antitumor function.
Materials and Methods
Samples and patients
This study involved 28 breast cancer patients undergoing LN curettage as standard-of-care surgery at Institut Curie Hospital (Paris, France), and were included after obtaining written informed consent. The study protocol was approved by an ethics committee “Ile de France,” and the Declaration of Helsinki protocols were followed. Fragments from TD-LNs selected by the pathologist (to allow the histologic analysis for the validation of the resection quality) were obtained from the operating room. Diagnosis of tumor metastasis in TD-LNs was done by the pathologist following macroscopic examination of the samples, confirmed by the core facility that performed histologic examination and IHC using anticytokeratin to detect isolated cancer cells. Table 1 summarizes the clinical and pathologic characteristics of the patients. Patients were scored by tumor–node–metastasis (TNM) classification (21), and tumors were characterized by IHC for hormone receptors (ESR1 gene, estrogen receptor 1, PGR, progesterone receptor) and tyrosine kinase cell-surface receptor HER2 (gene ERBB2, erbB-2 receptor tyrosine kinase 2) to determine the molecular subtype of the tumor. Nodal invasion by tumor cells was confirmed by CD45/epithelial cell adhesion molecule, EpCAM (clone 9C4, BioLegend) staining by flow cytometry. Most patients did not receive any prior treatments other than surgery. However, 2 patients were treated with neoadjuvant chemotherapy: one patient received 4 cycles of FEC (5-fluororacil, epirubicin, cyclophosphamide) and 4 cycles of TXT (taxans docetaxel) every 3 weeks, the other patient received 6 cycles of EC (epirubicin, cyclophosphamide) every 3 weeks.
Breast cancer patients . | |||
---|---|---|---|
Age . | 56,7 (32–83) . | ||
. | Number . | % . | |
TNM | T1N1M0 | 8 | 27 |
T1N2M0 | 1 | 3 | |
T2N0M0 | 1 | 3 | |
T2N1M0 | 7 | 23 | |
T2N2M0 | 4 | 13 | |
T3N1M0 | 3 | 10 | |
T3N2M0 | 2 | 7 | |
T3N3M1 | 1 | 3 | |
T4N1M0 | 1 | 3 | |
Stage | IIA | 9 | 30 |
IIB | 7 | 23 | |
IIIA | 10 | 33 | |
IIIB | 1 | 3 | |
Molecular subtype | Luminal | 22 | 73 |
Incl. Luminal A | 5 | 17 | |
Luminal B | 15 | 50 | |
HER2 | 3 | 10 | |
Mutational status | ER | 3 | 10 |
HER2 | 1 | 3 | |
ER/PR | 22 | 73 | |
ER/PR/HER2 | 2 | 7 | |
Carcinoma | Ductal | 25 | 83 |
Lobular | 2 | 7 | |
Mixte | 1 | 3 | |
Neoadjuvant treatment | 2 | 7 |
Breast cancer patients . | |||
---|---|---|---|
Age . | 56,7 (32–83) . | ||
. | Number . | % . | |
TNM | T1N1M0 | 8 | 27 |
T1N2M0 | 1 | 3 | |
T2N0M0 | 1 | 3 | |
T2N1M0 | 7 | 23 | |
T2N2M0 | 4 | 13 | |
T3N1M0 | 3 | 10 | |
T3N2M0 | 2 | 7 | |
T3N3M1 | 1 | 3 | |
T4N1M0 | 1 | 3 | |
Stage | IIA | 9 | 30 |
IIB | 7 | 23 | |
IIIA | 10 | 33 | |
IIIB | 1 | 3 | |
Molecular subtype | Luminal | 22 | 73 |
Incl. Luminal A | 5 | 17 | |
Luminal B | 15 | 50 | |
HER2 | 3 | 10 | |
Mutational status | ER | 3 | 10 |
HER2 | 1 | 3 | |
ER/PR | 22 | 73 | |
ER/PR/HER2 | 2 | 7 | |
Carcinoma | Ductal | 25 | 83 |
Lobular | 2 | 7 | |
Mixte | 1 | 3 | |
Neoadjuvant treatment | 2 | 7 |
NOTE: Molecular subtype of each tumor was determined by the pathology department following histologic staining with Luminal A as HER2−, hormone receptor+, and low Ki67, Luminal B as HER2+ or HER2− hormone receptor+ and high level of Ki67. Values in italics are percentages of patients in each category (right column). In molecular subtype values in italics indicate subtype of Luminal.
Twenty-four brain-dead, organ donor–derived mesenteric LNs were obtained from the reanimation department at Saint Louis Hospital (Paris, France) following ethical procedures and analyzed as noncancer controls (healthy donor, HD-LNs). Healthy donors included 13 women, and the whole cohort was matched for age with the breast cancer cohort. Blood samples (25 mL) from donors were obtained concomitantly to LN for some donors.
Cell lines
Two breast cancer cell lines, HER2−/HR+ MCF-7 and HER2−/HR− triple-negative MDA-MB-231 purchased from ATCC, were used for in vitro experiments. Cells were authenticated, stored frozen in DMSO, maintained for 2 months in culture. Cells were tested for mycoplasma. K562 cells from the lab were authenticated, stored frozen in DMSO, and used for 2 months. Cells were cultured in RPMI medium (RPMI 1640 GlutaMAX, Life Technologies) supplemented with 10% FCS and 1% penicillin/streptomycin.
LN flow cytometry
LN cell suspensions dissociated from fresh samples (viability always exceeding 90% after trypan blue exclusion), were resuspended in PBS supplemented with 5% AB-human serum (Biowest) and incubated 30 minutes on ice to block nonspecific FcR binding before staining. Cells were then stained for 30 minutes at 4°C with the specific monoclonal antibodies (mAb) at predetermined optimal concentrations. CD45+CD3−CD56+ NK cells, within lymphocyte FSC/SSC gate, were analyzed for the membrane expression of activating receptors, including NCRs (NKp46, NKp30, NKp44); NKG2D; DNAM-1; NKG2C; CD16; inhibitory receptors such as KIRs (CD158a, b, and e) and NKG2A receptors; activation/maturation markers CD117, CD57, HLA-DR, CD69; homing receptors CXCR3 and CCR7; and the adhesion molecule CD62L. The percentage of CD45−/EpCAM+ cells was determined to assess metastatic invasion in each TD-LN.
Expression of the NK receptor ligands MICA/B, ULBP 1-3, CD112, CD115, HLA-A, -B, -C, HLA-E, and PD-L1/L2 were analyzed on tumor cells from 4 M-LNs and breast cancer cell lines. All antibodies used for cytometry experiments are listed in Supplementary Table S1.
Cells were analyzed on a FACSCanto II flow cytometer and FlowJo software (BD Biosciences). The cytometer was calibrated with beads with analytic mAbs and FMO (tube with minus one mAbs) tubes.
CD107a degranulation assays
Cells from LN cell suspensions were stimulated with tumor cell lines (K562, MCF-7, and MDA-MB-231) at a 1:1 effector:target ratio in U-bottom plates (BD Biosciences) in the presence of a protein transport inhibitor (GolgiStop; BD Biosciences) according to the manufacturer's protocol and CD107-FITC antibody (clone H4A3; BD Biosciences) for 5 hours. Cells were then labeled with anti-CD45, anti-CD3, anti-CD56, and anti–CD16, washed, and analyzed on a FACSCanto II flow cytometer and analyzed with FlowJo as described above. Results are expressed as the percentage of FITC+ NK-gated cells. Control baseline NK cell CD107a staining (spontaneous degranulation) was determined in the absence of targets.
For some LNs, we performed negative immunoselection of NK cells using the NK cell isolation kit (Nb 130-092-657; Miltenyi Biotec). Purified (>90%) NK cells (0.5 to 1 × 106 cells/mL) were then cultured in the presence of IL2 or IL15 for 6 days (10 ng/mL; Miltenyi Biotec) and then assessed for degranulation as described above.
Cell-mediated lysis assay using label-free and real-time cellular impedance assay
Tumor cell lysis by immunoselected NK cells was assessed using the xCELLigence RTCA SP instrument (ACEA Biosciences, Inc.; ref. 22). This device allows reliable monitoring of cell proliferation and cell death due to the activity of cytotoxic effectors (23). For cell-mediated lysis, 15,000 breast cancer cells/well were seeded into 96-well E-Plates (Ozyme), and their adhesion was monitored for 6 hours. Fifteen thousand cytokine-activated NK cells were then added (effector:target ratio, 1:1). Impedance values (cellular index, CI) were assessed by the system, with a measurement every 15 minutes for up to 10 hours. Results were expressed as the percentage of lysis determined from the CI normalized at the addition of effectors (nCI) with the formula: % lysis = [nCI (no effector) – nCI (effector)]/[nCI (no effector)] ×100.
Statistical analysis
Statistical tests, graphics, and heat map representation were generated by Prism version 7 (GraphPad Software Inc.). A nonparametric Wilcoxon matched-pairs signed-rank test was used to compare the proportions of CD45−/EpCAM+ cells and NK cells between M- and NI-LNs. The same test was used to compare the percentage of NK receptors between NI-LN versus M-LN, as well as CD56brightCD16− and CD56brightCD16+ cells. The Mann–Whitney test was used to compare the expression of NK receptors between donors and patients. Correlations between the different parameters analyzed were assessed by the nonparametric Spearman test. A P value <0.05 was considered significant. Principal component analysis (PCA) was performed to visualize and organize the multivariate data from patient M- and NI-LNs as well as D-LNs.
Results
LN NK cell phenotype was independent of tumor cell phenotype and invasion rate
In the present study, 28 breast cancer patients treated by radical LN dissection were included. One macroscopically noninvaded (NI) and one tumor-invaded (metastatic, M) fragment were obtained and analyzed separately for 11 of these patients. According to the TNM classification, 1 patient had distant metastases, and most patients were N1 or N2 (one N0 and one N3/M1; Table 1).
Phenotypic analyses of cell suspensions from LNs were performed by multiparametric flow cytometry. We compared NK cells from NI and M-LNs independent of the TNM classification of patients. In TD-LNs, the proportion of CD45−EpCAM+ cells was monitored to distinguish NI-LNs (<0.4% of EpCAM+CD45− cells) from M-LNs (>1% EpCAM+). First, we compared the phenotypes of NK cells infiltrating paired invaded (M) and noninvaded (NI) TD-LNs that were obtained from 11 patients. The percentage of CD3−CD56+ NK cells was similar in NI- and M-LNs (Fig. 1A). TD-LN NK cells from NI- and M-LN exhibited similar expression of activating NK receptors and high expression of NKG2A (Fig. 1B). A small but significant reduction in NKp46 expression (%; Fig. 1B), and median fluorescence intensity (MFI; Supplementary Fig. S1) by M-LN NK cells was observed in all but 2 patients. We also found small but significant increase in MFI values for activating markers HLA-DR and CD69 in metastatic TD-LN NK cells (Supplementary Fig. S1), but percentages were similar (Fig. 1C). Chemokine receptor expression was also comparable between NI- and M-LN NK cells (Fig. 1D).
The tumor cells infiltrating the LN were analyzed for their expression of NK ligands from 4 breast cancer patients (Fig. 1E). We found high expression of ULBP2 molecules (expression of ULBP1 and 3 were low) on breast cancer–invading tumor cells, whereas the expression of MICA molecules was present on less than 10% of tumor cells, and MICB, CD112, and CD155 molecules were barely detectable. The tumor cells expressed HLA class I molecules. No correlation between NKG2D and NKG2A on NK cells and their cognate ligands on tumor cells was seen.
Activation of TD-LN NK cells from breast cancer patients compared with HD-LN NK cells
We then analyzed and compared the phenotype of TD-LN NK cells from 28 patients to the mesenteric LNs from 21 organ donor (HD-LNs) NK cells. No difference in the percentages of NK cells infiltrating HD-LN and TD-LN was observed (Fig. 2A, graph). NK cells infiltrating LNs were CD56bright NK cells in comparison with blood NK cells and constituted 0.3% to 10% of CD45+ cells (Fig. 2A, dot plots). TD-LN NK cells had high expression of NKp46 (median 38%), NKp30 (median 31%), and NKG2D (median 56%), and 15% of LN-NK cells expressed NKp44. The expression of HLA-E–specific inhibitory receptor NKG2A was elevated, and KIR expression was low in TD-LN NK cells. NKG2C was present in 25% of TD-LN NK cells. The coreceptor DNAM-1 was detected in 62% of patients' NK cells (Fig. 2B). We analyzed HD-LN NK cells from organ donors age-matched to patients. In comparison, HD-LN NK cells exhibited significantly reduced percentages of NKG2C, DNAM-1, and NKG2A receptors, whereas percentages of KIR receptors were higher compared with patients (Fig. 2B; Supplementary Fig. S2). HD-LN NK cells had significantly higher expression of CD69 and CD57 (Fig. 2C; Supplementary Fig. S2). CXCR3 was expressed by 40% to 60% of TD-LN NK cells, whereas less than 15% of HD-LN NK cells expressed it, indicating an inflammatory profile of NK cells in breast cancer–draining LNs (Fig. 2D). No difference was observed when analyses were restricted to female controls. We found that PD-1 was expressed by more than 10% of LN NK cells from donors and patients. The proportions of PD-1+ NK cells exhibited large interindividual variation and even higher for patients (Fig. 2B). PD-1 expression was correlated with NKp30 and NKG2C expression (Supplementary Fig. S3A). We also found that inhibitory receptors CD96, TIGIT, and TIM-3 were expressed by HD-LN NK cells with preferential expression of TIGIT and TIM3 by CD16+ NK cells (Supplementary Fig. S3B and S3C).
Data analyses by PCA showed that NK cells from NI- and M-LNs clustered in the same region, whereas HD-LN NK cells clustered in a distinct region (Fig. 2E). PCA identified the most robust parameters involved in the distinction of the donor and patient samples, which were markers relating to NK regulation (NKp46 and NKG2A) and migration (CXCR3 and CD62L; Supplementary Table S2). Overall, the comparison of donor and TD-LNs showed the presence of increased proportions of NKG2A+ NK cells. Despite higher expression of CXCR3 by TD-LN NK cells, their recruitment to the LN cannot be ascertained in the absence of time-based studies.
Correlation of TD-LN NK profile and disease stage
Most patients included in the study had luminal B or A subtype breast cancer, ductal carcinoma, and were positive for ER and PR by IHC. The number of metastatic LNs was up to 3 metastatic LNs in stage IIA and IIB patients and between 4 and 9 metastatic LNs in stage IIIA patients. Hierarchical clustering of the phenotypic data obtained from the patients showed that some modulation of activating NK receptors associated with the stage of the disease (Fig. 3A) and with the number of tumor-invaded LNs (N1 <4 vs. N2: 4–9 invaded LNs, according to the TNM classification). The CD16+ subset was prominent in stage IIIA, T3–4, and N2 patients (Fig. 3B–D). The expression of activating receptors NKp46 and CD69 also positively correlated to the stage of the patients. Elevated NKG2A correlated to disease stage and N2–3 status. Altogether, these data indicated that increased proportions of CD16+NKG2A+ NK cells were found in the TD-LN of breast cancer patients and were associated with locally advanced disease (N2M0 and N3M0).
Phenotypic comparison of CD16− and CD16+ NK cells infiltrating TD-LNs and HD-LNs
We previously described high proportions of activated CD56brightCD16+ NK cells in M-LNs from melanoma patients (24). Significantly higher percentages of CD56brightCD16+ were also found in LNs from breast cancer patients, compared with healthy donors, on CD45+CD3−CD56+ gated cells (Fig. 4A). In contrast, HD-LNs contained a small CD16+ NK subset (20%, range, 10%–39%) compared with breast cancer TD-LNs (35% of CD16+, range, 14%–76%), suggesting regulation of the CD56brightCD16+ LN NK cells in patients with cancer (Fig. 4A). We then compared the percentages and MFIs for NK receptors expressed by both CD16− and CD16+ NK subsets in donors and breast cancer patients (Supplementary Fig. S3). We determined the deviation in the expression of each receptor expressed by CD16+ compared with CD16− NK cells and calculated the ratios of CD16+ versus CD16− (Fig. 4B–D). When the ratio was greater than 1 (dotted line), the marker was more expressed by the CD16+ subset, whereas it was reduced in the CD16+ subset when the ratio was below the dotted line (Fig. 4B–D). CD16+ LN NK cells exhibited significantly higher expression of activating receptors NKp46, NKp30, and NKG2D, as well as elevated expression of NKG2A, in both donors and patients. Expression of KIRs (CD158a, b, e), CD57, and CXCR3 were increased in a CD16+ subset of TD-LN NK cells. In contrast, HD-LN CD16+ NK cells displayed decreased expression of NKp44 and DNAM-1 (Fig. 4B–D). PD-1 expression by CD16− and CD16+ NK cells was comparable for both donors and breast cancer–derived LNs. The activated, mature CD16+ NK cell subset was prominent in TD-LNs, and these NK cells overexpressed NCRs, NKG2D, NKG2A, CXCR3, and CD57. These CD16+ LN NK cells may perform ADCC and may be regulated by HLA-I expression.
Functional properties of LN-NK cells from breast cancer patients
Finally, we assessed the lytic potential/capacity of LN NK cells against MCF-7 (HER2−/HR+) and MDA-MB-231 (HER2−/HR−, triple-negative) breast cancer cell lines. First, we showed that the cell lines differentially expressed MIC-A and MIC-B, ULBP2-3, and HLA-A, B, C, E molecules, depicted in radar chart diagram (Fig. 5A). MDA-MB-231 cells had high expression of ULBP2-3, HLA class I and HLA-E molecules, and PD-L1/L2, whereas MCF-7 cells were characterized by low expression of NKG2D ligands and low HLA class I and HLA-E molecules (Fig. 5A and B). Degranulation assays were performed to assess the activation of LN NK cells. LN NK cells efficiently degranulated following stimulation with K562 and breast cancer cell lines. A trend for higher degranulation in TD-LN NK cells toward K562 compared with HD-LN NK cells was observed (Fig. 5C).
To further determine the lysis of breast cancer cell lines by LN NK cells, isolated LN NK cells were activated with IL2 or IL15 before the xCELLigence lysis assay (24). Cytokine activation maintained the same proportion of CD16+ NK cells (Supplementary Fig. S4A). Cytokine-activated LN NK cells exhibited increased expression of NKp46, NKp30, NKp44, NKG2D, DNAM-1, CD69, HLA-DR, and CD117 (Supplementary Fig. S4B-C). NK cells from 5 TD-LNs and 3 HD-LNs were included for xCELLigence assays. IL2-activated TD-LN NK cells lysed breast cancer cell lines efficiently, and lysis of HLA class Ilow MCF-7 cells was always at least 50% greater than or superior to that of HLA class Ihigh MDA-MB-231 cells (Fig. 5D), a response compatible with high NKG2A/HLA-E engagement. Lysis of breast cancer cells by HD-LN NK cells was low compared with that of TD-LN NK cells (Fig. 5D), in agreement with the activated phenotype of TD-LN NK cells. Finally, IL15-activated LN NK cell–mediated lysis was 20% superior to that of IL2-activated LN-NK cells (Fig. 5E), which may be related to the higher expression of activating NK receptors under this condition (Supplementary Fig. S4).
Discussion
The breakthrough of immunotherapies with immune-checkpoint blockade has revolutionized the treatment of cancers. However, not all cancer patients benefit from these treatments, and additional knowledge on immune effector regulation is required for wider application. The correlation between immune profile and better survival in breast cancer patients incites investigation of the interactions between tumor and antitumor immune effectors, in order to identify new targets to enhance antitumor efficiency (25). Reports indicate that NK cells are involved in the course of breast cancer in mouse models, as well as in human disease. In experimental models, NK cells suppress breast cancer metastases to multiple organs in tumor-bearing mice (26), and NK cell deficiencies accelerate metastases in SCID mice (27, 28). NK cell suppression or NK cell depletion induced by surgery accelerates metastases in immunocompetent mice (20, 29). In contrast with these findings, various alterations of tumor-infiltrating NK cells have been described in patients with breast cancer (30), as well as in blood NK cells from patients with advanced disease (31).
Investigating NK cells from TD-LNs, we showed that NK cells from axillary armpit TD-LNs expressed chemokine receptors CXCR3 and CD62L, regardless of tumor cell invasion, indicating that NK cell activation precedes invasion by tumor cells. Tumor-draining sentinel LNs showed enhanced lymphangiogenesis preceding metastasis and may function as a permissive “lymphovascular niche” for the survival of metastatic cells (32). However, the increased blood flow and lymphangiogenesis may induce DC activation (33) and, in turn, favors NK cell recruitment through CXCR3 before invasion by tumor cells. The production of CXCL9 and CXCL10, ligands of CXCR3, was reported in breast cancer tumors, and NK cell accumulation in tumors was dependent on IFNγ and CXCR3 ligands (34), suggesting trafficking of tumor-infiltrating NK cells to the LNs. Alternatively, these TD-LN NK cells may be activated from LN-residing NK cells.
TD-LN NK cells were characterized by high expression of activating NK receptors, NKG2D and DNAM-1, and the inhibitory receptor NKG2A. CD62L expression positively correlated with the activating receptor NKp46 and may confer antitumor activity to TD-LN NK cells, as it was reported to suppress metastatic formation in LNs by L-selectin–mediated NK cell recruitment (35). We confirmed the presence of activated CD56brightCD16+ NK cells in the TD-LNs from breast cancer patients, as we previously reported in metastatic TD-LN from melanoma (23). These CD16+ NK cells overexpressed activating NCRs, NKG2D, and CD57, as well as increased inhibitory NKG2A.
In TD-LNs invaded by tumor cells from melanoma patients, NK cell degranulation was inversely correlated to the percentages of invading melanoma cells (24). Tumors cells are reported to induce signaling defects in NK cells, altering the Jak/STAT pathway, decreasing NK cell functions in breast cancer tumors (36), and decreasing IRF1 and DAP10 in metastatic melanoma blood NK cells (37, 38). In contrast, the low percentages of invading tumor cells and the high lytic potential of activated TD-LN NK cells from breast cancer patients suggest that such defects may not be present in TD-LNs. We also found that IL15 potentiates the lytic capacities of LN-NK cells. The production of IL15 by sinusoidal CD163+ macrophages, associated with favorable nodal status in breast cancer patients, may account for the activation of LN-NK cells of breast cancer patients through cross-talk with these myeloid cells (39).
Studies indicate the role of HLA class I expression in breast cancer immunosurveillance. The total loss of HLA class I is an independent indicator of good prognosis in breast cancer (40). Patients with mixed HLA class I–expressing tumors had a higher probability of disease recurrence after 5 years than patients with either HLA class I–positive or –negative tumors, and in patients who received adjuvant therapy, the mixed phenotype was not associated with disease recurrence (41). Further investigation of the impact of HLA class I expression showed that expression of HLA class Ib molecules (HLA-E and HLA-G) resulted in a worse relapse-free period only in patients with loss of HLA class Ia (HLA-A, -B, and -C molecules) tumor expression. Immunogenic breast cancer tumors may escape CTL immune responses by downregulation of HLA class Ia molecules, and in HLA-Iow/− tumors, upregulation of HLA class Ib molecules favors tumor escape from NK cells (42, 43). In agreement with this regulation, TD-LN NK cells preferentially lysed HLA-Ilow breast cancer targets. Monoclonal antibodies targeting the inhibitory receptors involved in HLA class I molecule recognition may restore NK cell activities and, in turn, boost the antitumor adaptive immune response.
We also found that >10% of LN NK cells expressed PD-1. Upregulation of PD-1 ligands is reported in different types of breast cancer cancers, including basal type (44, 45), and it has been associated with prognosis (46, 47). In TD-LNs, the positive correlation of PD-1, NKG2C, and NKp30 indicated a profile compatible with functional cytotoxic NK cells bearing some features corresponding to memory-like NK cells (48). There are few studies of PD-1+ NK cells. In renal cell carcinoma patients, blood PD-1+ NK cells correlated with disease stage, exhibited an activated effector phenotype, and high expression of perforin and granzyme B. The expression of all 3 biomarkers declined rapidly after surgery (49). A subpopulation of fully mature PD-1+ NK cells was increased in patients with ovarian carcinoma, whose impaired antitumor activity was partially restored by disruption of PD-1/PD-L1 interaction (50). Additional analyses of HD-LNs indicated that inhibitory receptors CD96, TIGIT, and Tim3+ were present on NK cells. Further analyses will be required for assessing these receptors in NK cells from TD-LN cells.
The presence of CD16+NKG2A+ cytotoxic TD-LN NK cells was associated with the stage, tumor size, and lymph node invasion (stage IIIA, T and N status). Tumor-reactive NK cells in nondisseminated tumors in breast cancer patients are encouraging arguments for the development of NK cell–based therapies. High expression of NKG2A and correlation of PD-1 and NKp30 suggest the possibility of controlling inhibition. NKG2A monoclonal antibodies may be a therapeutic option for breast malignancy in the adjuvant setting. Combined treatment with anti–PD-1 may also improve NK cell antitumor function and adaptive immune responses. Finally, combinational therapies with novel antibody reagents such as bi- and trispecific killer engagers (BiKEs and TriKEs) in combination with IL15 against tumor-specific antigens to enhance NK cell–mediated tumor rejection may be promising (51).
Disclosure of Potential Conflicts of Interest
N. Dulphy is a consultant/advisory board member for Celyad and Celgene. L. Zitvogel reports receiving commercial research funding from Bristol-Myers Squibb, Incyte, Transgene, and Lytix Biopharma. No potential conflicts of interest were disclosed by the other authors.
Authors' Contributions
Conception and design: A. Frazao, M. Messaoudene, N. Nunez, E. Piaggio, A. Toubert, A. Caignard
Development of methodology: A. Frazao, M. Messaoudene, E. Piaggio, A. Caignard
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): A. Frazao, M. Messaoudene, C. Sedlik, E. Piaggio, F. Roussin
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): A. Frazao, M. Messaoudene, E. Piaggio, A. Toubert, A. Caignard
Writing, review, and/or revision of the manuscript: A. Frazao, M. Messaoudene, N. Dulphy, L. Zitvogel, E. Piaggio, A. Toubert, A. Caignard
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): A. Frazao, N. Nunez, C. Sedlik, E. Piaggio
Study supervision: E. Piaggio, A. Caignard
Acknowledgments
This work was supported by grants from l'Institut National du Cancer (PLBIO-2011-6) and PAIR Mélanome (2013-0662013), l'Association pour la Recherche contre le Cancer (ARC, 3964 to A. Caignard), la Ligue Nationale contre le Cancer (Comité Ile de France to A. Caignard, PhD grant for M. Messaoudene), l'Assistance Publique des Hôpitaux de Paris (APHP; DRCD, Immumela program), the Siric (site de recherche intégrée sur le cancer). N. Nunez was supported by a fellowship from Institut Curie and from la Ligue Nationale Contre le Cancer, and A. Frazao by grants from Cancéropôle Ile-de-France and Association Robert Debré pour la Recherche Médicale.
We particularly thank Sophie Viel (Inserm U932), Delphine Loirat (INSERM U932 and Département d'Oncologie Médicale), Maud Milder (Clinical Immunology Laboratory), and Anne Vincent-Salomon (Center of Clinical Investigations CICBT507 IGR/Curie and Department of Pathology) for providing clinical samples and associated clinical data and for organizing the logistics of these samples. We thank Marie-Françoise Avril (Hôpital Tarnier/Cochin) for helpful discussion. Thanks to all the participating patients.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.