Fax: (011) 39-02-2643-4760
Killer cell Ig-like receptors ligand-mismatched, alloreactive natural killer cells lyse primary solid tumors
Article first published online: 27 JUN 2006
Copyright © 2006 American Cancer Society
Volume 107, Issue 3, pages 640–648, 1 August 2006
How to Cite
Re, F., Staudacher, C., Zamai, L., Vecchio, V. and Bregni, M. (2006), Killer cell Ig-like receptors ligand-mismatched, alloreactive natural killer cells lyse primary solid tumors. Cancer, 107: 640–648. doi: 10.1002/cncr.22002
- Issue published online: 18 JUL 2006
- Article first published online: 27 JUN 2006
- Manuscript Accepted: 10 MAR 2006
- Manuscript Revised: 20 FEB 2006
- Manuscript Received: 13 DEC 2005
- Associazione Italiana Ricerca sul Cancro (AIRC)
- killer cell immunoglobulin (Ig)-like receptors (KIR) cells;
- natural killer (NK) cells
Donor alloreactive natural killer (NK) cells have a potent antileukemic effect in haploidentical stem cell transplantation. Whether alloreactive NK cells are able to specifically kill fresh tumor cells from primary solid tumors was analyzed.
NK cells were purified from healthy donors for the expression of inhibitory killer cell immunoglobulin (Ig)-like receptors (KIRs), ex vivo expanded, and used as effector cells. Their cytotoxic effect on tumor cells freshly obtained from surgical specimens was assessed by means of a single-cell cytotoxic assay (SCCA).
Tumor cells from 1 ovarian, 1 gastric, 3 colon, and 4 renal cell cancers were analyzed and found susceptible to alloreactive NK cell killing (>20% lysis at an effector cell to target cell [E:T] ratio of 10:1 for tumor cells not expressing at least 1 human lymphocyte antigen [HLA] class I KIR-ligand group). Remarkably, NK cells that recognized specific HLA-C group mismatches were able to kill HLA-C KIR ligand-mismatched tumor cells, whereas no lysis of target cells occurred with KIR ligand-matched tumor targets.
Alloreactive NK-cell mediated antitumor effects might provide useful insights for designing new cell therapy approaches against solid tumors. Cancer 2006. © 2006 American Cancer Society.
Natural killer (NK) cells, a first-line of defense against primary viral infections and tumor cells,1–8 are regulated by inhibitory and activatory signals.9–13 Negative regulation is performed by inhibitory receptors that are specific for self-major histocompatability complex (MHC) class I molecules.14 In humans, as clonally distributed inhibitory killer cell immunoglobulin (Ig)-like receptors (KIRs) are specific for determinants shared by certain human leukocyte antigen (HLA) class I alleles, NK subsets in the repertoire are blocked by specific class I allele groups. Three major NK inhibitory receptors recognize 3 specific class I allele groups.15–18 The KIR2DL2/3 receptors (CD158b1/b2) recognize alleles expressing Asn80 (i.e., HLA-C group 1 alleles such as Cw1, 3, 7, 8, and 9). KIR2DL1 (CD158a) recognizes alleles sharing Lys80 (i.e., HLA-C group 2 alleles [Cw2, 4, 5, and 6]). Finally, KIR3DL1 (CD158e) is specific for HLA-B alleles sharing the Bw4 supertypic specificity. Consequently, NK cells mediate alloreactions when faced with mismatched target cells that do not express the class I alleles blocking NK cells in the repertoire.
Building on these premises, the Perugia group observed that full-haplotype mismatched hematopoietic stem cell transplantation was associated with a transient wave of reconstituting NK cells, whose repertoire was identical to that originally displayed by the donor, including high-frequency donor-versus-recipient alloreactive NK clones.7 Alloreactive NK cells exerted a potent antileukemia effect against acute myeloid leukemia (AML).8
Whether NK cell alloreactions, such as those occurring after KIR-ligand mismatched haplotransplants, are effective against solid tumors remains to be established. Indications that cancer is susceptible to immune attack come from reports showing that HLA-matched allogeneic transplantation induces a graft-versus-tumor effect.19–22 Interestingly, Igarashi et al.23 demonstrated in vitro that human KIR-ligand mismatched, alloreactive NK cells lyse melanoma and renal cell carcinoma cell lines. Because cell line killing may not be predictive of in vivo efficacy against primary tumor cells, in this study we assessed alloreactive NK cell killing of freshly excised, primary solid tumor cells of different histotypes.
MATERIALS AND METHODS
Analysis of the NK-cell population was performed using the following anti-human monoclonal antibodies (MoAbs): anti-CD56-fluorescein isothiocyanate (FITC), anti CD3-peridinin chlorophyll protein (PerCP), anti-CD158e-phycoerythrin (PE) (DX9), anti-CD158a-PE (HP-3E4), and anti-CD158b1/b2-FITC (DX27 and CH-L) (Becton Dickinson, San Jose, CA). Double fluorescence analysis was performed as follows: CD3−/CD56+, CD56+/CD158e+, CD158e+/CD158b1/b2+, CD158a+/CD158b1/b2+.
Primary tumor cells were analyzed for the expression of adhesion molecules known to be involved in effector-to-target binding and/or NK activation, namely: anti-LFA-1-FITC, anti-CD58-FITC, anti-CD11c-PE, anti-CD54-PE, anti-CD49d-PE, anti-CD29-PE, and anti-CD11b-PE (Becton Dickinson).
Cell surface antigen expression was detected on cell populations as follows: incubation for 10 minutes with different MoAbs or with irrelevant isotype-matched MoAbs (Becton Dickinson); cells were washed once, resuspended, and analyzed on a FACScan analyzer (Becton Dickinson). A minimum of 10,000 events were acquired.
The chronic myeloid leukemia cell line K-562 was maintained in RPMI-1640 with 10% fetal calf serum (FCS). Epstein–Barr virus (EBV)-transformed lymphoblastoid cell lines (EBV-LCL) were ex vivo established from tumor-bearing patients according to the following method: peripheral blood mononuclear cells (PBMC) were isolated on a Ficoll-Hypaque gradient. After washing, 107 PBMC was resuspended with 1 mL of concentrated supernatant from the B95-8 EBV producer cell line (a human type I EBV-transformed marmoset B-cell line)24 for 1 hour at 37°C. The cell mixture was transferred into vented 25 cm3 flasks and cultured at 37°C and 5% carbon dioxide in RPMI-1640 media (Cambrex Bioscience, Verviers, Belgium), 20% heat-inactivated FCS, 10 mM HEPES, 2 mM/L L-glutamine, 50 U/mL penicillin/streptomycin, and PHA-P (20% stock) (Difco Laboratories, Detroit, MI). After a 24-hour incubation, cyclosporine A, at a dose of 100 μg/mL (CSA, Sandoz Pharmaceuticals, Washington, DC) was added to the flask. EBV-LCL were established after 4 weeks.
The EBV-LCL BAT-3 cell line transfected with Cw03 (HLA-CAsn80), Cw05 (HLA-CLys80), and Bw44 (HLA-BBw4) was utilized as a target for functional NK analysis (a kind gift of Dr. Andrea Velardi of the University of Perugia, Perugia, Italy).
The ovarian cancer cell line A2780 expressing 3 major KIR ligands, Cw07 (HLA-CAsn80), Cw02 (HLA-CLys80), and Bw4901 (HLA-BBw4) was utilized as a target for functional NK analysis (kindly provided by Dr. Zunino of the National Cancer Institute, Milan, Italy).
Genomic low resolution HLA-B and -C typing was performed on tumor cells and peripheral blood leukocytes by commercial Dynal Reli SSOTM, according to the manufacturer's recommendations.25
NK Cell Selection and Expansion
A 40-mL blood sample from healthy donors was drawn in a heparinized test tube. Mononuclear cells (MNCs) were isolated by Ficoll-Hypaque density gradient sedimentation and depleted of monocytes by plastic adherence for 2 hours at 37°C.
NK cell populations were purified by negative depletion using the MACS system according to the manufacturer's recommendations (Miltenyi Biotec, Auburn CA). Briefly, monocyte-depleted PBMCs were incubated with 20 μL of anti-CD3 microbeads and 80 μL of MACS buffer (phosphate-buffered saline [PBS] [pH 7.2], supplemented with 0.5% bovine serum albumin and 2 mM of ethylenediamine tetraacetic acid) for 15 minutes at 4°C. After incubation, cells were washed twice with buffer. Negative depletion was performed on 2 consecutive LD columns on MidiMACS apparatus. NK cell purity was determined by flow cytometry. T-cell contamination was determined to be less then 2% by double staining technique, as previously described.
NK cells were subsequently isolated for the expression of KIRs (CD158b1/b2, CD158a, CD158e) on their surfaces from CD3-negative cells by MACS using anti-fluorochrome microbeads (anti-FITC, anti-PE) according to the manufacturer's protocol (Miltenyi Biotec). NK cells were cultured with RPMI-1640, supplemented with 10% heat-inactivated human AB Serum (ABS), 50 U/mL penicillin, 50 μg/mL streptomycin, 1% PHA, and irradiated (100 Gy) allogeneic EBV-LCL feeder cells (15:1 EBV-LCL/NK cell ratio) in a humidified 37°C, 5% CO2 incubator, as described.23
Recombinant human interleukin 2 (IL-2) (Hoffmann-LaRoche, Nutley, NJ) was added at a final concentration of 500 IU/mL on days 1 and 4 of culture. NK cell lines were restimulated weekly with irradiated allogeneic EBV-LCL feeders in the presence of IL-2 (500 IU/mL). On Day 14, immunofluorescence analysis by flow-cytometry was performed utilizing a gating strategy to exclude CD3 contamination.
Preparation of Primary Tumor Cells
Human cancer samples were obtained during surgical tumor resection after written informed consent. Single cell suspensions were obtained by mechanical disaggregation. Tumors that did not disaggregate well and yielded poor cell suspensions were subjected to overnight digestion with collagenase (type IV; Sigma Chemical Company, St. Louis, MO). Cell suspensions were filtered, washed, and resuspended in RPMI 1640 supplemented with 10% FCS, 50 U/mL penicillin, and 50 μg/mL streptomycin.
Suspensions were placed on top of a double gradient solution of Ficoll-Hypaque containing 75% plus 100% Ficoll, respectively, and centrifuged at 700g for 25 minutes at room temperature to separate tumor cells from the mononuclear cells. Tumor cells were found mostly on the top of 75% Ficoll-Hypaque. The tumor cell layer was collected and washed in RPMI 1640 supplemented with 10% FCS.
Tumor cells were retained and further analyzed for viability, as assessed by the Trypan blue exclusion assay, and tumor cell content of the suspension, as assessed by light microscopy on cytospin preparations. In 2 cases, very early-passage tumor cells isolated from biopsies were maintained in RPMI-1640 with 10% FCS for later use as previously described.23
Single-Cell Cytotoxicity Assay
For target cell staining, primary tumor cells were stained with 10 μL/106 cells DiOC18 (3 mM solution) for 20 minutes at 37°C in 5% CO2 according to the manufacturer's instructions (LIVE/DEAD Cell-Mediated Cytotoxicity Kit, Molecular Probes, Eugene, OR). After labeling, cells were washed twice with PBS, counted, and adjusted to 1 × 106/mL.
Effector cells were washed, counted, and adjusted to 0.8 × 105 in 260 μL final volume. A mixture of stained tumor cells (1 × 104) and effector cells at different effector cell to target cell (E:T) ratios (40:1; 20:1; 10:1) was added to sterile Falcon polystyrene tubes in a final volume of 140 μL. Tubes were then incubated at 37°C in an humidified 5% carbon dioxide incubator for 2 hours. After incubation, 130 μL of the counterstaining solution containing propidium iodide (PI) (5 μg/mL) was added to each cell mixture and gently mixed. Finally, the cells were suspended in 0.5 mL PBS and placed in an ice bath. Flow cytometric data acquisition was performed consecutively. Several tubes were assayed in parallel to serve as controls for the target cells: unstained target cells to assess autofluorescence, single-stained target cells as green fluorescence control with DiOC18, single-stained target cells killed with 0.2% Triton-X as PI-positive red fluorescence control, double-stained PI/DiOC18 positive target cells as a control for nonspecific target cell death (the latter being equivalent to “spontaneous” 51Cr release). Effector controls consisted of unlabeled cells and cells stained with PI to estimate viability. The analysis was performed by gating on DioC18-labelled, green fluorescent target cells.
The percentage of specific lysis was calculated using the following formula: percentage of DioC18-labeled target cells stained with PI after incubation with effector cells (%) − percentage of stained PI/DiOC18-positive target cells equivalent to “spontaneous” 51Cr release (%).
51Chromium (51Cr)-Release Assay for Evaluation of Alloreactivity of NK Cells
For the 51Cr-cytotoxicity assay, primary tumor cells, EBV-LCL, and human tumor cell lines were utilized as targets. Target cells were incubated with 51Cr (Amersham Pharmacia Biotech, Piscataway, NJ) 100 mCi per 106 cells for 2 hours. 51Cr-labeled targets were aliquoted at a concentration of 1 × 104 cells/well. Samples were plated in triplicate at a 20:1 E:T ratio into 96-well U-bottom plates (NUNC, Naperville, IL, Brand Products). Chromium release was assayed after a 4-hour incubation at 37°C. The percentage of specific lysis was calculated using the standard formula: 100 × (cpm experimental release − cpm spontaneous release)/(cpm maximal release − cpm spontaneous release).
Statistical analysis of the results was performed using the Pearson correlation coefficient.
Validation of Single-Cell Cytotoxicity Assay vs. Standard 51Cr Release Assay
To assess tumor cell killing, single-cell cytotoxicity assay (SCCA) was used and preliminarily validated against a standard 51Cr release assay.
Effector cells were bulk KIR-selected NK cell preparations from healthy donors who expressed the 3 major KIR-ligands (HLA-CAsn80, HLA-CLys80, HLA-Bw4). In the standard 51Cr release assay, the HLA class I-deficient, NK-sensitive human cell line K-562 exhibited 60% NK-mediated specific lysis. The human cell lines BAT-3 and A2780 transfected with, or constitutively expressing, alleles belonging to the 3 major class I groups recognized by KIRs were resistant to NK killing (specific lysis <20% by 51Cr release assay at 20:1 E:T ratio) (Fig. 1A). The same cell lines were analyzed for NK-cell-mediated killing by the SCCA assay and superimposable results were obtained (Fig. 1B).
Regression analysis demonstrated a correlation coefficient of R = 0.93 between sets of data obtained with the 2 techniques. Moreover, we performed independent experiments in which the killing of the same cell line, K-562, mediated by NK cells from 5 different donors tested both in the SCCA assay and in 51Cr release assay showed a correlation coefficient of R2 = 0.42 (Fig. 1C).
Solid Tumor Cells of Different Histotypes Are Susceptible to Alloreactive NK Cell Killing
Of 51 processed samples, only 9 samples contained >90% viable tumor cells (Fig. 2). HLA typing was performed on both tumor cells and in PB leukocytes in 3 patients (Patients 1, 6, and 8) to rule out the possibility of MHC class I down-regulation in tumor cells, obtaining superimposable results; in the other patients, due to the low number of fresh tumor cells obtained from biopsies, we could not perform the HLA typing on all tumor samples and we analyzed the HLA expression only on PBLs. Cancer cells were used as targets in the SCCA. Cytotoxicity assays were performed on 1 gastric, 1 ovarian, 3 colon, 2 renal freshly isolated cancer samples, and on 2 very early passage renal cell carcinoma (Patients 8 and 9) cells. In Table 1 the HLA typing and the expression of KIR ligands on tumor cells are reported. Primary tumor cells consistently expressed adhesion molecules known to be involved in effector-to-target binding and/or in NK activation, such as members of the beta1-integrin family (CD49d, CD29), the beta2-integrin family (LFA-1, CD11b, CD11c) and the Ig super-family (CD58/LFA-3, CD54/ICAM-1) (data not shown).
|1||Gastric cancer||08||07||HLA- CAsn80|
|2||Ovarian cancer||08, 18||07||HLA- CAsn80|
|3||Colon cancer||37, 51||06, 15||HLA-CLys80, Bw4|
|5||Colon cancer||27, 55||02, 03||HLA-CAsn80, HLA-CLys80, Bw4|
|6||Renal cell cancer||18, 44||0701, 0704||HLA-CAsn80, Bw4|
|7||Renal cell cancer||18, 58||07||HLA-CAsn80, Bw4|
|8||Renal cell cancer||15, 5801||0102, 07||HLA-CAsn80, Bw4|
|9||Renal cell cancer||08, 51||07, 15||HLA-CAsn80, HLA-CLys80, Bw4|
In the first set of experiments, effector cells were allogeneic bulk NK cell preparations from healthy donors who expressed the 3 major KIR-ligands in their HLA type (HLA-CAsn80, HLA-CLys80, HLA-Bw4) (Fig. 3A).
Cancer cells from all patients who failed to express at least 1 KIR ligand were killed efficiently by NK cells, as assessed by SCCA assay (>20% specific lysis at a 10:1 E:T ratio). Representative results of efficient tumor cell killing are shown for a gastric cancer (Patient 1) and a renal cell cancer (Patient 8) (Fig. 3B). Almost absent lysis (<20%) was observed when cancer cells were obtained from a patient who possessed all 3 major KIR ligands in their HLA type (Patient 9; Fig. 3B). A summary of specific lysis results for all tumor specimens tested is shown in Table 2. As a positive control, these NK cell preparations killed allogeneic targets (EBV-LCL blasts) from healthy individuals that did not possess 1 of the ligands (not shown).
|Patient sample no.||E:T Ratio at 40:1||E:T Ratio at 20:1||E:T Ratio at 10:1|
In a subsequent set of experiments, effector cells were obtained from a healthy donor who was homozygous for HLA-CAsn80 alleles. NK cells specific for HLA-CAsn80 alleles enriched by immunomagnetic sorting with anti-CD158b1/b2 monoclonal antibodies were used as effector cells. Target cells were analyzed by gating on DioC18-positive cells. Tumor cells derived from a KIR ligand-matched (HLA-CAsn80-positive) renal cell cancer patient were largely resistant to lysis at a 20:1 E:T ratio (Fig. 4A, left). EBV-LCL derived from the same patient was not susceptible to NK cells (Fig. 4A, right). In contrast, tumor cells from a colon cancer patient who did not express HLA-CAsn80 alleles (and was homozygous for HLA-CLys80 alleles) were lysed efficiently at a 20:1 E:T ratio (Fig. 4B, left). The same result was obtained with EBV-LCL derived from the same patient (Fig. 4B, right). This is representative of 3 other experiments in which purified NK cells obtained from HLA-CLys80-positive individuals were tested against KIR ligand-mismatched and KIR ligand-matched tumor cell samples. KIR ligand-mismatched (HLA-CAsn80-positive) tumors were killed, whereas KIR ligand-matched (HLA-CLys80) tumors were resistant (not shown).
Although it has been demonstrated that human tumor cell lines may be susceptible to lysis by NK cells,23 this might not be representative of the in vivo situation, because tumor cell lines could have undergone evolution in culture and/or chromosomal rearrangements, which modify the primary tumor antigenic profile. The present study shows for the first time, to our knowledge, that tumor cells of different histotypes freshly obtained after surgical resection are susceptible to NK cell-mediated lysis.
Unfortunately, only 18% of tumor samples were suitable for the in vitro assay, based on viability and tumor cell content. Furthermore, we experienced several difficulties with the standard 51Cr release assay for different reasons. First, viable tumor cells are often too few for 51Cr labeling.26 Second, cells undergoing apoptosis during the incubation time cause high spontaneous 51Cr release in a standard assay. Finally, tumor lesions also contain noncancer cells, such as tumor infiltrating lymphocytes, stromal, and endothelial cells.
We sought to circumvent these problems by means of a flow-cytometry-based SCCA, which selectively identify the tumor cells at a single-cell level by virtue of a fluorescent dye that binds to cellular membrane and cytoplasm. After the target cells were incubated with effector cells, the membrane-impermeant nucleic acid counterstain propidium iodide was added to label any cells with compromised cellular membranes. We confirmed previous reports27–29 of a positive correlation between results of the SCCA and the 51Cr release assay.
In comparison with the standard 51Cr release assay, which measures the 51Cr released by the whole population of necrotic cells, the SCCA analyses cells at a single level. In particular, the viable tumor cell population can be distinguished from dead cells present in the biopsies. Furthermore, tumor cells can easily be identified on the cytometric assay based on physical parameters, thus excluding noncancer contaminating cells from the analysis. Finally, target cells were labeled with DioC18 to exclude the noise of the effector cells in the target region, thus allowing a better analysis of the specific killing.
Solid tumors from patients who do not possess at least 1 of the major HLA class I allele groups, recognized by inhibitory KIRs, were susceptible to alloreactive NK cell killing. The results from our in vitro assay suggest that there is no correlation between the percentage of specific lysis and number of missing HLA alleles on tumor cells. This finding leads to the speculation that the NK cell-mediated cytotoxicity activity can also be affected by the presence and the intensity of expression of both activatory molecules and KIRs on NK cells. Moreover, the intensity of expression of HLA class I molecules and NK cell activatory ligands on tumor cells is essential for NK cell recognition.30, 31 Further studies will be necessary to investigate the expression of these molecules on NK cells and tumor cell surfaces. To investigate whether killing of tumor cells was mediated by NK cell-recognition of missing KIR ligands on target cells, we used cells from homozygous donors for HLA-C allele groups as effector cells. Targets were tumors from individuals failing to express the donor HLA class I allele group recognized by the KIR-selected NK cells. These experiments indicated that NK cell-mediated lysis of tumor cells was consistent with killing through KIR ligand incompatibility. In contrast, KIR ligand-matched tumors exhibited little, if any, susceptibility to NK lysis.
The NK cell ability to recognize allogeneic targets failing to express class I molecules has been exploited in haploidentical hematopoietic transplantation for hematologic malignancies.32 After transplantation from donors who are KIR ligand-mismatched in the graft-versus-host direction, donor-versus-recipient alloreactive NK cells mediate a graft-versus-leukemia effect and improve engraftment, while not causing graft-versus-host disease.7, 8 Although HLA-matched allogeneic transplantation has recently been shown to induce a graft-versus-host effect in some solid tumors,19–22 to our knowledge HLA-mismatched transplantation in this setting has not been explored to date.
Our data show enhanced susceptibility of fresh tumor cells to NK cells with KIR incompatibility, and suggest that the NK cell-mediated antitumor effects after KIR-incompatible mismatched allografting for patients with AML might be induced also against selected solid tumors. This strategy could be clinically exploited in at least 2 ways (i.e., as a cellular therapy in patients with refractory tumors, by infusing ex vivo expanded NK cells from a haploidentical donor to patients that have received an immunosuppressive conditioning). The successful in vivo persistence and expansion of haploidentical donor NK cells adoptively transferred in patients with tumors has been recently published.33 Alternatively, patients with refractory solid tumors could be offered a haploidentical, nonmyeloablative transplant from a KIR-incompatible sibling donor, as has been reported in 1 patient with renal cancer.34 The infusion of donor NK cells after haploidentical transplantation for adoptive immunotherapy is safe, without inducing graft-versus-host disease, as demonstrated by other investigators.35 Indeed, we are starting at our institution a pilot study of haploidentical sibling donor transplant in patients with refractory, advanced renal cell cancer without HLA-identical donors. Our present observations on the susceptibility of freshly isolated solid tumors to alloreactive NK cell-mediated killing opens new perspectives for the development of NK-based immune interventions for the treatment of cancer.
We wish to thank Dr. Tito Roccie and the colleagues of the Hematology Unit (Scientific Institute San Raffaele); Dr. Isebelle Sessi and the colleagues of the Pathology Unit (Scientific Institute San Raffaele), and the surgeons and staff of the operating room (Scientific Institute San Raffaele).
We also thank Dr. Andrea Velerol (University of Perugia, Italy) for his invaluable advice.
- 26Quantitative assay of the lytic action of immune lymphoid cells on 51-Cr-labelled allogeneic target cells in vitro; inhibition by isoantibody and by drugs. Immunology. 1968; 14: 18–22., , , et al.