Vγ9Vδ2 T cell cytotoxicity against tumor cells is enhanced by monoclonal antibody drugs—Rituximab and trastuzumab

Authors

Errata

This article is corrected by:

  1. Errata: Erratum: Vγ9Vδ2 T cell cytotoxicity against tumor cells is enhanced by monoclonal antibody drugs—Rituximab and trastuzumab Volume 129, Issue 11, 2761, Article first published online: 29 September 2011

Abstract

Vγ9Vδ2 T cells exert potent cytotoxicity toward various tumor cells and adoptive transfer of Vγ9Vδ2 T cells is an attractive proposition for cell based immunotherapy. Vγ9Vδ2 T cells expanded in the presence of Zoledronate and IL-2 express CD16 (FcγRIII), which raises the possibility that Vγ9Vδ2 T cells could be used in conjunction with tumor targeting monoclonal antibody drugs to increase antitumor cytotoxicity by antibody dependent cellular cytotoxicity (ADCC). Cytotoxic activity against CD20-positive B lineage lymphoma or chronic lymphocytic leukemia (CLL) and HER2-positive breast cancer cells was assessed in the presence of rituximab and trastuzumab, respectively. Cytotoxicity of Vγ9Vδ2 T cells against CD20-positive targets was higher when used in combination with rituximab. Similarly, Vγ9Vδ2 T cells used in combination with trastuzumab resulted in greater cytotoxicity against HER2-positive cells in comparison with either agent alone and this effect was restricted to the CD16+Vγ9Vδ2 T cell population. Our results show that CD16+Vγ9Vδ2 T cells recognize monoclonal antibody coated tumor cells via CD16 and exert ADCC similar to that observed with NK cells, even when target cells are relatively resistant to monoclonal antibodies or Vγ9Vδ2 T cells alone. Combination therapy involving ex vivo expanded CD16+Vγ9Vδ2 T cells and monoclonal antibodies may enhance the clinical outcomes for patients treated with monoclonal antibody therapy. © 2008 Wiley-Liss, Inc.

Gamma delta T cells expressing the Vγ9Vδ2 T-cell receptor (TCR), comprise a small percentage (1–5%) of the lymphocytes that circulate in human blood.1, 2 Evidence that Vγ9Vδ2 T cells have potent cytotoxic activity against tumor cells in a non-MHC-restricted manner in vitro3–6 raise the possibility that Vγ9Vδ2 T cells could be utilized as a component of antitumor therapies.7–11 After their activation by nonpeptidic phosphoantigens via γδ TCR binding, Vγ9Vδ2 T cells can recognize and exert cytotoxicity against tumor cells in various ways. Mechanisms that may be relevant include utilization of MICA/B or UL16 binding protein (ULBP) interactions with NKG2D, antibody coated tumor cells via CD16 (FcγRIII),12–15 and possibly by other unknown receptors.

Vγ9Vδ2 T cells can be divided into 2 populations based on the presence or absence of CD16 expression. CD16+Vγ9Vδ2 cells have potent direct cytotoxicity via a range of mechanisms but have low cytokine production. CD16-Vγ9Vδ2 T cells exert less cytotoxicity but generally have higher cytokine production.15 There is a similar division of functional characteristics between the 2 NK cell populations, CD56dim (CD16bright) NK cells and CD56bright (CD16dim/−) cells.16 CD3+CD16+ T cells (a population that contains a large proportion of Vγ9Vδ2 T cells) or CD16+ γδ T cells can efficiently mediate antibody dependent cellular cytotoxicity (ADCC) via CD16.17, 18 Although CD16 is a low affinity receptor for IgG Fc, IgG-coated cells can bind and activate CD16 positive γδ T cells through CD16.13, 18 Therefore, via ADCC, Vγ9Vδ2 T cells could potentially enhance the antitumor activity of therapeutic monoclonal antibodies that have human Fc portions. It has been speculated that the therapeutic potential of CD16+Vγ9Vδ2 cells may be limited by their poor expansion capabilities.6, 9, 10 However, studies in our laboratory demonstrate that large numbers of CD16+Vγ9Vδ2 T cells can be obtained from peripheral blood of healthy donors and patients with malignancy, after expansion in the presence of Zoledronate and IL-2.

The therapeutic monoclonal antibodies rituximab (active against CD20 positive tumors) and trastuzumab (active against HER2 expressing cancers), have well-established clinical effectiveness alone and in combination with chemotherapy.19–22 Rituximab is a chimeric anti-CD20 IgG1 monoclonal antibody consisting of a human constant region linked to murine variable domains.23 Trastuzumab is a humanized anti-Her2/neu monoclonal antibody. Rituximab and trastuzumab exert cytotoxicity against tumor cells by various mechanisms including ADCC, compliment-dependent cytotoxicity (CDC) and direct induction of apoptosis.24–27 Enhancement of any of these mechanisms may increase antitumor activity, particularly in relapsed or resistant cases.

In our study we show not only that activated CD16+Vγ9Vδ2 T cells can markedly enhance tumor cell killing by rituximab and trastuzumab but that this effector cell population can be expanded for potential therapeutic application in cancer patients.

Material and methods

Flow cytometry analysis

The following monoclonal antibodies were used for phenotypic analysis and obtained from Beckman Coulter (CA): CD3-FITC (UCHT1), CD3-PE (UCHT1), CD4-PE (13B8.2), CD8-ECD (SFCI21Thy2D3), CD16-FITC (3G8), CD20-PC5 (B9E9), CD27 (1A4CD27), CD45-PC7 (J.33), CD45RA (ALB11), CD56-PC5 (N901), TCR-Vγ9-FITC (IMMU360), TCR-Vδ2 (IMMU389), TCR-pan γ/δ-PC5 (IMMU510). Anti-HER2/neu-FITC (Neu 24.7) was obtained from Becton Dickinson (San Jose, CA). Cell number was assessed by addition of flow-count beads (Beckman Coulter, CA) and cell viability was determined by staining with 7AAD. Cells were stained according to manufacturers' recommendations. All flow cytometric analysis was performed using the Coulter Cytomics FC500 five-color flow cytometer (Beckman Coulter, CA).

Monoclonal antibody drugs

Rituximab and trastuzumab were obtained from Roche (Basel, Switzerland). The concentrations used in our study, 10 μg/ml rituximab and 2 μg/ml trastuzumab, was based on previous clinical studies using standard dosage and schedule,28–31 as well as drug information obtained from pharmaceutical companies.

Culture media

Culture media RPMI 1640 (Cambrex, Walkersville, MD) containing 40 μg/ml of gentamicin (Pharmacia, WA, Australia) and 2 mM of L-glutamine (Cambrex, Walkersville, MD) were used for these experiments. Media was supplemented with 10% human AB serum (Cambrex, Walkersville, MD) for Vγ9Vδ2 T cell culture and 10% FCS (Invitrogen, Australia) for tumor cell line culture.

Proliferation and preparation of Vγ9Vδ2 T cells

Informed consent was obtained from all donors and subjects prior to blood and tissue collection. Our study was approved by Human Research Ethics Committees of the University of Queensland and Greenslopes Private Hospital, Queensland, Australia.

Peripheral blood mononuclear cells (PBMC) were isolated by density gradient centrifugation using Ficoll-Paque (Amersham Biosciences, Sweden) from randomly selected healthy donors or subjects with malignancy. To proliferate Vγ9Vδ2 T cells, PBMC were cultured in media containing 700 IU/ml of recombinant human IL-2 (rhIL-2) (Chiron, Netherlands) and 1 μM of zoledronate (Novartis, Basel, Switzerland). After 10-14 days culture, effector cell populations containing 70–95% of Vγ9Vδ2 T cells were depleted of CD4+cells and CD8+cells using immunomagnetic separation miniMACS (Miltenyi Biotec, Germany) prior to functional assays. When effector cell populations contained more than 1% of CD3CD56+cells (NK cells) CD56+cell depletion was also performed. The resulting purified populations contained less than 1% CD3CD56+cells (NK cells) and CD3+CD8+ T cells, and greater than 95% purity of Vγ9Vδ2 T cells. Greater than 95% of pan γδ+ cells were Vγ9+Vδ2+ T cells after 14 days culture. Where indicated, CD16+Vγ9Vδ2 T cells were sorted from CD16-Vγ9Vδ2 T cells by MoFlo (DakoCytomation, Denmark) and resulting purity of each population was greater than 95%.

Proliferation and preparation of NK cells

To proliferate NK cells, PBMC were cultured in media containing 1,000 IU/ml of rhIL-2. After 14-21 days culture, effector cell populations were depleted of CD3+cells by immunomagnetic separation using miniMACS prior to use. Resultant populations contained greater than 90% CD3CD56+cells.

Cell lines

Daudi, Raji and Ramos were used to represent CD20 positive tumor cells. U937 cell line was used as a CD20 negative control. SK-BR3 and BT474 were used to represent Her2/neu positive tumor cells. MDA-MB231 was included as a Her2/neu negative tumor cell line control. These cells were obtained from the American Type Culture collection (Manassas, VA). All cell lines were cultured in media with 10% FCS and maintained at 37°C in 5% CO2. Adherent cells were detached using 0.05M EDTA.

Tumor cells derived from patients

PBMC were isolated from chronic lymphocytic leukemia (CLL) patients and used as target cells in Vγ9Vδ2 T cell cytotoxicity assays. More than 90% of PBMC from these patients were positive for CD20. Lymph node biopsy tissue derived from a patient with follicular B cell lymphoma was also obtained to collect CD20 positive target cells for use in autologous cytotoxicity assays. To obtain a single cell suspension prior to mononuclear cell (MNC) isolation, lymph node biopsies were minced using a 100 μm nylon cell strainer (BD Biosciences, Bedford, MA) and plunger. Approximately 90% of these MNCs were CD20 positive and viability following MNC extraction was greater than 90%.

Cytotoxicity assessment against nonadherent tumor cell targets (Annexin V/7AAD assay)

To evaluate cytotoxicity of Vγ9Vδ2 T cells against nonadherent target cells, including Daudi, Raji and Ramos, U937 and fresh tumor cells from CLL and NHL patients, the AnnexinV/7AAD flow cytometric assay (BD Biosciences, San Jose, CA) was used. This assay has previously been validated as a non-radioactive alternative to the 51Cr release assay for assessment of effector cell cytotoxicity of nonadherent targets.32, 33

Following PKH26 (Sigma-Aldrich, St. Louis, MO) staining of target cells to allow distinction on the flow cytometer, Vγ9Vδ2 T cells and targets were cocultured at various effector/target ratios (E:T ratio) in the presence or absence of rituximab (10 μg/ml) or in the presence of human IgG (10 μg/ml) (Sigma-Aldrich, St. Louis, MO) for 4 hr at 37°C. All cells were then harvested and stained with AnnexinV and 7-AAD antibodies according to manufacturer's instructions. Early apoptotic (AnnexinV+/7AAD), late apoptotic and necrotic (AnnexinV+/7AAD+) cells were distinguished from viable cells (AnnexinV/7AAD)34 in PKH26 positive target cells. Percent cytotoxicity was determined after subtracting values from appropriate control wells containing targets only.

Cytotoxicity assessment against adherent tumor cell lines (MTS assay)

Cytotoxicity of Vγ9Vδ2 T cells against adherent target cells, SK-BR3 and MDA-MB231, was determined by the tetrazolium-based colorimetric assay using CellTiter 96 (Promega, Madison, WI) as described previously (MTS assay).33, 35 Cytotoxicity values obtained using the tetrazolium-based colorimetric assay has been shown to strongly correlate with results of the 51Cr release assay.36, 37

SK-BR3 and MDA-MB231 cells were seeded in 96 well plates at 1 × 104 cells/well and allowed to adhere at 37°C overnight. Vγ9Vδ2 T cells were added at different E:T ratios in the presence or absence of trastuzumab (2 μg/ml) or human IgG (2 μg/ml). Effector and target cells were subsequently cocultured at 37°C for 4 or 24 hr as indicated. Wells containing target cells alone were used as negative controls for spontaneous cell death and wells containing target cells with trastuzumab were used to evaluate trastuzumab-induced cell death. Wells containing Vγ9Vδ2 T cells alone were used as Vγ9Vδ2 T cell controls. After coculture, media containing the nonadherent Vγ9Vδ2 T cells was removed from the adherent target cells and replaced with fresh media mixed with the MTS tetrazolium salt. After 3 hr incubation, optical density (OD) was read directly at 490 nm using the Multiskan Ascent microplate reader (Thermo, Finland). Cytotoxicity of Vγ9Vδ2 T cells were measured at each E:T ratio: %Cytotoxicity = 1 − viability of target cells = 1 − {(OD490 of Vγ9Vδ2 T cell and target cells in the presence or absence of trastuzumab or human IgG well) − (OD490 of Vγ9Vδ2 T cell control well)}/OD490 of target cell control well × 100.

Blocking studies

A functional grade anti-CD16 monoclonal antibody, clone CB16 (eBioscience, San Diego, CA) was used at 10 μg/ml to block the CD16-mediated interactions between Vγ9Vδ2 T cells and rituximab treated Ramos targets. Anti-CD16 antibody was used at 2 μg/ml to block binding of Vγ9Vδ2 T cells with trastuzumab treated SK-BR3 targets. The antibody was added to Vγ9Vδ2 T cells 30 minutes prior to coculture.

To inhibit perforin-mediated cytotoxicity, Vγ9Vδ2 T cells were incubated with Concanamycin A (Sigma) at 100 ng/ml for 2 hr at 37°C and washed twice prior to coculture.38

Perforin and IFN-γ release from Vγ9Vδ2 T cells

Secretion of perforin and IFN-γ by Vγ9Vδ2 T cells during 4 hr coculture with or without target cells were assessed by sandwich ELISA. Following 4 hr coculture supernatants were collected and stored at −20°C until ELISA assessment. The secretion of perforin was determined using the Perforin ELISA set (MABTECH AB, Sweden) and the secretion of IFN-γ was assessed using the Human IFN-γ OptEIA ELISA set (BD Biosciences, San Jose, CA), according to the manufacturers' protocols.

Chemokine release from Vγ9Vδ2 T cells

Secretion of chemokines (MCP-1, RANTES, MIG, IP-10 and IL-8) by Vγ9Vδ2 T cells during 4 hr coculture with SK-BR3 were assessed by Cytometric Bead Array (CBA) Human Chemokine Kit (BD Biosciences, San Jose, CA), according to the manufacturers' protocols.

Statistical analysis

All statistical analyses were performed using the Student's t test and results were considered significant if p < 0.05.

Results

Proliferation of CD16+Vγ9Vδ2 T cells

After 14 days culture of MNCs from 5 healthy donors and 9 patients with malignancy, we obtained a high percentage of Vγ9Vδ2 T cells (>70% in CD45+ cells) in all cultures of healthy donors and 5 of 9 cultures from cancer patients (2 colon, 1 duodenal, 1 ovarian and 1 malignant lymphoma). The remaining 4 patient cultures (all malignant melanoma) contained low percentages of Vγ9Vδ2 T cells (<25%). Functional assessment was conducted using the 10 individuals in whom large numbers of highly enriched Vγ9Vδ2 T cells could be generated.

To determine CD16 expression of Vγ9Vδ2 T cells cultured under our conditions and the extent to which CD16+Vγ9Vδ2 T cells could proliferate, we assessed CD16 expression on Vγ9Vδ2 T cells before and after in vitro expansion of healthy donors (n = 5) and subjects with malignancy (n = 5). The total population of Vγ9Vδ2 T cells expanded in the first week, however expansion of the CD16+ subpopulation of Vγ9Vδ2 T cells was consistently slow during this period. There was subsequent rapid proliferation of CD16+Vγ9Vδ2 T cells from day7 to day14 (Fig. 1a). This is in agreement with previous observations that CD16 expression increases slowly after TCR stimulation by phosphoantigens, taking 2–3 weeks to reach a plateau.13

Figure 1.

Proliferation of CD16+Vγ9Vδ2 T cells. (a) PBMCs were cultured with 10% AB serum, Zoledronate and IL-2 for 14 days. The total population of Vγ9Vδ2 T cells (GDT) expanded in the first week, followed by rapid proliferation of CD16+Vγ9Vδ2 T cells from day7 to day14. The data is representative of 10 different donors. (b) Approximately equal numbers of CD16+ and CD16 Vγ9Vδ2 T cells were obtained following 14 days culture. Results show cell number from a total of 1 × 107 PBMC (left panel). Expansion fold of CD16+ Vγ9Vδ2 T cells was significantly higher than CD16- Vγ9Vδ2 T cells after 14 days culture (**p < 0.01) (right panel) (mean ± SEM, n = 10). (c) CD16 expression on patients' Vγ9Vδ2 T cells (n = 5) was not significantly different from healthy donors (n = 5) (p = 0.56).

After 14 days culture, approximately equal numbers of CD16+ and CD16 Vγ9Vδ2 T cells were obtained (mean ± SEM of 4.5 ± 1.1 × 107 CD16+Vγ9Vδ2 T cells and 5.2 ± 1.0 × 107 CD16Vγ9Vδ2 T cells from an original 1.0 × 107 MNCs) (Fig. 1b left panel). Mean ± SEM (n = 10) expansion fold of CD16+ Vγ9Vδ2 T cells (429 ± 85) was significantly higher (p < 0.01) than CD16 Vγ9Vδ2 T cells (172 ± 27) after 14 days culture (Fig. 1b right panel). There was no significant difference in proliferative capability between healthy donors and subjects with malignancy (data not shown).

Previous studies have shown considerable variation in CD16 expression of cultured Vγ9Vδ2 T cells, with some studies showing very low CD16 expression6, 9 and others showing greater than 40% expression.13 With our culture conditions a large proportion of Vγ9Vδ2 T cells generated after 14 days expressed CD16 and those of cancer patients (40.2 ± 7.4%, n = 5) was not significantly different from healthy donors (46.7 ± 7.8%, n = 5) (p = 0.56, mean ± SEM) (Fig. 1c).

CD16 expression and Vγ9Vδ2 T cell memory subset

Unlike NK cells that can be clearly split into CD16+ and CD16 populations, expanded Vγ9Vδ2 T cells exhibited a continuum of expression of CD16, as previously published13 (Fig. 2a). After 14 days culture, most Vγ9Vδ2 T cells were CD45RA CD27 (effector memory) cells with smaller numbers of CD45RACD27+ (central memory) cells. Both populations expressed CD16 (Fig. 2b) in contrast to a previous report stating that most CD16+γδ T cells in freshly obtained peripheral blood were CD45RA+effector memory cells.15

Figure 2.

CD16 expression on Vγ9Vδ2 T cells and memory subset. (a) Expression of CD16 on Vγ9Vδ2 T cells (GDT) is lower than for NK cells. The panel is representative of 10 different donors. (b) Most expanded Vγ9Vδ2 T cells were CD45RACD27 (effector memory) cells but expanded Vγ9Vδ2 T cells from some donors contain CD45RACD27+ (central memory) cells. Both effector populations expressed CD16.

Cytotoxicity of Vγ9Vδ2 T cells and rituximab against CD20+ B lymphoma cell lines

Vγ9Vδ2 T cells cultured for 10–14 days were purified as described in material and methods section before being used as effector cells. The percentage of Vγ9Vδ2 T cells expressing CD16 varied as shown in Figure 1c.

The cytotoxicity of Vγ9Vδ2 T cells against the CD20+ B lymphoma cell targets Daudi, Raji and Ramos, and CD20 promonocytic leukemia cell line (U937), was assessed using the Annexin V/7AAD assay, shown in Figure 3a. Rituximab alone and Vγ9Vδ2 T cells alone exerted relatively little cytotoxicity against CD20+ cell lines. However, rituximab greatly enhanced the cytotoxicity of Vγ9Vδ2 T cells against all CD20+ B lymphoma targets. Rituximab-induced cytotoxicity was not observed against U937. Enhancement of cytotoxic activity was less evident when Vγ9Vδ2 T effector cells expressed very low levels (3.1%) of CD16 (data not shown). Rituximab alone had minimal cytotoxic activity against Daudi cells, (3.4 ± 0.4%). In comparison Vγ9Vδ2 T cells alone, at an E:T ratio of 5:1, were substantially more cytotoxic against Daudi cells (15.1 ± 1.6%) and cytotoxicity of Vγ9Vδ2 T cells in combination with rituximab (32.4 ± 2.3%) was significantly higher (p < 0.01) than Vγ9Vδ2 T cells alone or rituximab alone. Results are expressed as mean ± SEM for 10 independent experiments (Fig. 3b).

Figure 3.

Cytotoxicity of Vγ9Vδ2 T cells with rituximab against CD20+ and CD20 malignancies. (a) Cytotoxicity of Vγ9Vδ2 T cells (GDT) against CD20+ lymphoma cell lines (Daudi, Raji and Ramos) and CD20 promonocytic leukemia cell line (U937) were measured in the presence (○) or absence (□) of rituximab (10 μg/ml) or in the presence of human IgG (Δ). Rituximab enhanced the cytotoxicity of Vγ9Vδ2 T cells against all CD20+ cell lines but not against the CD20 cell line, U937. The data is representative of 10 (Daudi, Raji and Ramos) and 3 (U937) different donors. (b) Cytotoxicity of Vγ9Vδ2 T cells from various donors is significantly higher against Daudi in combination with rituximab than Vγ9Vδ2 T cells alone (**p < 0.01, mean + SEM, n = 10, E:T = 5:1). (c) Perforin-mediated cytotoxicity of Vγ9Vδ2 T cells against Daudi was blocked by Concanamycin A (CMA) both in the absence and presence of rituximab. (d) Anti-CD16 mAb (aCD16) blocked the rituximab enhanced cytotoxicity of Vγ9Vδ2 T cells against Ramos (E:T = 5:1, mean + SEM, triplicate). IgG: mouse IgG. (e) Typical CD20 expression on PBMC from CLL patients used as target cells (left panel). Cytotoxicity of Vγ9Vδ2 T cells (GDT) against CD20+ CLL cells was measured in the presence (○) or absence (□) of rituximab (10 μg/ml) or in the presence of human IgG (Δ). Cytotoxicity of Vγ9Vδ2 T cells alone is low but addition of rituximab greatly enhanced the cytotoxicity of Vγ9Vδ2 T cells, particularly at higher E:T ratios (middle panel). Cytotoxicity of Vγ9Vδ2 T cells in the presence of rituximab was significantly higher than Vγ9Vδ2 T cells alone against CD20+CLL cells (**p < 0.01, mean + SEM, n = 5, E:T = 5:1) (right panel).

Addition of concanamycin A reduced cytotoxicity of Vγ9Vδ2 T cells against Daudi both in the absence and presence of rituximab (Fig. 3c) indicating a major role for perforin in the increased Vγ9Vδ2 T cell mediated killing observed in combination with rituximab. Our observation that perforin was involved in killing by Vγ9Vδ2 T cells alone was consistent with previous data indicating a role for perforin in Vγ9Vδ2 T cell cytotoxicity.9, 39 In addition, rituximab-enhanced Vγ9Vδ2 T cell cytotoxicity against Ramos cells was significantly inhibited in the presence of anti-CD16 mAb (CB16) within 4 hr (Fig. 3d).

Cytotoxicity of Vγ9Vδ2 T cells with rituximab against CD20+CLL cells

Cytotoxic activity of rituximab, Vγ9Vδ2 T cells and the combination of both against CLL cells is shown (Fig. 3e). Results are shown as the mean ± SEM for 5 separate experiments using 3 different Vγ9Vδ2 T cell donors and 3 different CLL patient samples. As was observed for lymphoma cell lines, rituximab alone and Vγ9Vδ2 T cells alone had minimal cytotoxicity against CLL cells. Cytotoxicity of the combination (19.0 ± 3.1%) was significantly higher than rituximab alone (6.0 ± 1.0%, p < 0.01) or Vγ9Vδ2 T cells alone (2.9 ± 0.9%, p < 0.01) using an E:T ratio of 5:1. In this assay more than 50% of Vγ9Vδ2 T cells from all 3 normal donors were CD16+.

Cytotoxicity of Vγ9Vδ2 T cells with trastuzumab against breast cancer cell lines

To evaluate cytotoxicity of Vγ9Vδ2 T cells with trastuzumab, HER2-overexpressing breast cancer cell lines SK-BR3 and BT474, and the HER2-negative MDA-MB231 cell line were used as target cells (Fig. 4a). In this assay more than 50% of Vγ9Vδ2 T cells were CD16+. Cytotoxicity was measured using the MTS assay in 8 independent experiments using Vγ9Vδ2 T cells from 4 different donors at various E:T ratios after 4 hr (Fig. 4b). The addition of trastuzumab significantly increased Vγ9Vδ2 T cells cytotoxicity (56.2 ± 3.8%) against SK-BR3 in comparison with Vγ9Vδ2 T cells alone (5.1 ± 1.0%, p < 0.01) and trastuzumab alone (3.8 ± 1.0, p < 0.01) at an E:T ratio of 5:1 (Fig. 4c, left panel). Similarly, cytotoxicity against BT474 cell line was higher for the combination than with Vγ9Vδ2 T cells alone or trastuzumab alone (data not shown). The addition of trastuzumab had no effect on Vγ9Vδ2 T cell mediated cytotoxicity of the HER2 negative MDA-MB231 cells (Fig. 4c, right panel).

Figure 4.

Cytotoxicity of Vγ9Vδ2 T cells with trastuzumab against HER2 positive and negative cell lines. (a) HER2-overexpressing breast cancer cell line SK-BR3 (left panel) and HER2-negative MDA-MB231 (right panel) were used as target cells. (b) Cytotoxicity of Vγ9Vδ2 T cells (GDT) against SK-BR3 and MDA-MB231 was measured in the presence (○) or absence (□) of trastuzumab (2 μg/ml) or in the presence of human IgG (Δ) by 4 hr MTS assay. The combination of Vγ9Vδ2 T cells and trastuzumab greatly enhanced cytotoxicity against SK-BR3 (left panel), but not against HER2 negative MDA-MB231 (right panel). The data is representative of 8 different experiment using 4 different donors. (c) Cytotoxicity of Vγ9Vδ2 T cells with trastuzumab was significantly higher than Vγ9Vδ2 T cells alone against SK-BR3 (left panel) but not against MDA-MB231 (right panel) (**p < 0.01, mean + SEM, n = 8, E:T = 5:1). (d) Perforin-mediated cytotoxicity of Vγ9Vδ2 T cells against SK-BR3 was blocked by Concanamycin A (CMA) both in the absence and presence of trastuzumab. (e) Comparison of Vγ9Vδ2 T cell and NK cell cytotoxicity against SK-BR3 in the presence and absence of trastuzumab measured by 24 hr MTS assay (E:T = 5:1). Trastuzumab enhanced cytotoxicity of both Vγ9Vδ2 T cells and NK cells, with NK cells exhibiting greater cytotoxicity. The data shows representative of 3 different donors (E:T = 5:1, mean + SEM, triplicate).

Addition of concanamycin A reduced cytotoxicity of Vγ9Vδ2 T cells against SK-BR3 targets both in the presence and absence of trastuzumab (Fig. 4d), indicating a major role for perforin in the increased Vγ9Vδ2 T cell-mediated killing observed in combination with trastuzumab.

Similar cytotoxic trends were observed in 24 hr coculture assays. Cytotoxicity of Vγ9Vδ2 T cells with trastuzumab was substantially greater than Vγ9Vδ2 T cells alone and trastuzumab alone against SK-BR3. However cytotoxicity of Vγ9Vδ2 T cells with trastuzumab was slightly less than that of NK cells with trastuzumab (Fig. 4e).

Cytotoxicity of CD16+ and CD16-Vγ9Vδ2 T cells with trastuzumab against SK-BR3

To confirm the role of CD16 expression on Vγ9Vδ2 T cells for enhanced cytotoxicity of SK-BR3 targets in combination with trastuzumab, we sorted effector Vγ9Vδ2 T cells (Fig. 5a) into CD16+ and CD16 populations and compared cytotoxic activity. Trastuzumab in combination with CD16+Vγ9Vδ2 T cells, but not CD16 Vγ9Vδ2 T cells, greatly enhanced cytotoxicity against SK-BR3 targets (Fig. 5b).

Figure 5.

Cytotoxicity of CD16+ and CD16 Vγ9Vδ2 T cells against SK-BR3. (a) Purification of CD16+ and CD16 Vγ9Vδ2 T cells (GDT). After sorting, we obtained more than 95% pure CD16+ and CD16Vγ9Vδ2 T cells. (b) Cytotoxicity of CD16+ and CD16Vγ9Vδ2 T cells against SK-BR3 in the presence (○) or absence (□) of trastuzumab (2 μg/ml) or human IgG (Δ) in a 4 hr MTS assay. Trastuzumab in combination with CD16+Vγ9Vδ2 T cells, but not CD16Vγ9Vδ2 T cells, greatly enhanced cytotoxicity against SK-BR3 targets. The data is representative of 3 different patient samples. (c) Anti-CD16 mAb (aCD16) blocked the trastuzumab (Tras) enhanced cytotoxicity of Vγ9Vδ2 T cells against SK-BR3 in 4 hr coculture (E:T = 5:1, mean + SEM, triplicate). IgG: mouse IgG

In support of this, enhanced cytotoxicity of Vγ9Vδ2 T cells by trastuzumab against SK-BR3 shown in Figure 4 was significantly inhibited in the presence of anti-CD16 mAb (CB16) within 4 hr (Fig. 5c).

Perforin, IFN-γ and chemokine release from Vγ9Vδ2 T cells

The release of perforin and IFN-γ from Vγ9Vδ2 T cells, measured by ELISA of culture supernatant, was not influenced by the presence of trastuzumab or the HER2 positive cell line SK-BR3 alone but was significantly increased (p < 0.05 and p < 0.01 respectively) in the presence of both SK-BR3 and trastuzumab (Figs. 6a and 6b, left panel). Perforin release (mean ± SEM, n = 3 from 3 separate donors) was 652.3±83.9 pg/ml in the presence of both trastuzumab and SK-BR3 target cells compared to 388.7±17.4 pg/ml in the presence of target cells alone (Fig. 6a, left panel). IFN-γ release (mean ± SEM, n = 3 from 3 separate donors) was 139.9±6.0 pg/ml in the presence of trastuzumab and SK-BR3 target cells compared to 46.1±8.8 pg/ml in the presence of target cells alone (Fig. 6b, left panel). Similar increases of perforin and IFN-γ secretion of Vγ9Vδ2 T cells were observed against Daudi in the presence of rituximab (data not shown). In contrast to that observed with SK-BR3 and Daudi, increased secretion of perforin and IFN-γ was not observed when trastuzumab was added to co-cultures of Vγ9Vδ2 T cells and the HER2 negative cell line MDA-MB231 (perforin secretion 465.3 ± 15.6 pg/ml with trastuzumab versus 424.0 ± 5.7 pg/ml without trastuzumab and IFN-γ secretion 69.3 ± 2.9 pg/ml with trastuzumab versus 61.3 ± 4.1 pg/ml without trastuzumab) (Figs. 6a and 6b, right panel). The increases in perforin and IFN-γ secretion observed with the addition of either trastuzumab or rituximab to the Vγ9Vδ2 T cell/SK-BR3 or Vγ9Vδ2 T cell/Daudi cultures corresponded to the observed increase in lysis of each target.

Figure 6.

Secretion of Perforin, IFN-γ and chemokines from Vγ9Vδ2 T cells. Supernatant from coculture in cytotoxicity assays were used to evaluate secretion of both perforin and IFN-γ by Vγ9Vδ2 T cells (GDT) (E:T = 5:1, 5 × 104 Vγ9Vδ2 T cells, 4 hours). (a, b) Secretion of both perforin and IFN-γ was significantly higher against SK-BR3 but not MDA-MB231 in the presence of trastuzumab (Tras). (c) Secretion of RANTES, but not MCP-1, MIG and IP-10, by Vγ9Vδ2 T cells was significantly higher against SK-BR3 in the presence of trastuzumab. (*p < 0.05, **p < 0.01, mean + SEM, n = 3 different donors). IgG: mouse IgG

Secretion of RANTES from Vγ9Vδ2 T cells in culture with SK-BR3 was significantly greater in the presence of trastuzumab (813.3 ± 23.6 pg/ml, p < 0.01) than without (310.6 ± 44.8 pg/ml) but no difference was observed for other chemokines, MCP-1, MIG, IP-10 (Fig. 6c) and IL-8 (data not shown). Secretion of these chemokines was not observed in SK-BR3 alone, SK-BR3 + mouse IgG, and SK-BR3 + trastuzumab (data not shown), suggesting that chemokine readouts were a product of Vγ9Vδ2 T cells.

Cytotoxicity of Vγ9Vδ2 T cells against autologous B lymphoma cells

To evaluate the potential role of autologous Vγ9Vδ2 T cells in lymphoma patients we determined the cytotoxicity of peripheral blood Vγ9Vδ2 T cells from a patient against autologous follicular lymphoma cells (Fig. 7). Vγ9Vδ2 T cells were proliferated from patient peripheral blood for 14 days. The cytotoxicity of these autologous Vγ9Vδ2 T cells (>90% pure) against CD20+ lymphoma cells (expressed as mean ± SEM from three separate experiments) was low with Vγ9Vδ2 T cells alone (13.7 ± 1.5%) but significantly higher when rituximab was added (27.5 ± 2.2%, p < 0.01). Rituximab alone, in the absence of effector cells had minimal direct cytotoxicity (3.8 ± 0.4%) (Fig. 7c). The CD16 expression of the patient Vγ9Vδ2 T cells was relatively low (25%) (Fig. 7a, right panel) and this is likely to have contributed to the relatively low cytotoxicity observed.

Figure 7.

Autologous cytotoxic assay of Vγ9Vδ2 T cell. (a) CD20 positive B-lymphoma cells derived from a patient lymph node was used as target cells. Vγ9Vδ2 T cells (GDT) were proliferated from peripheral blood of the same patient and 25% of Vγ9Vδ2 T cells expressed CD16. (b) Cytotoxicity of Vγ9Vδ2 T cells was measured in the presence (○) or absence (□) of rituximab (10 μg/ml) or in the presence of human IgG (Δ). Rituximab enhanced the cytotoxicity of Vγ9Vδ2 T cells against B-lymphoma cells. The data is representative of 3 independent experiments from one patient. (c) Cytotoxicity of Vγ9Vδ2 T cells is significantly higher in combination with Rituximab than Vγ9Vδ2 T cells alone (**p < 0.01, mean + SEM, n = 3, E:T = 25:1).

Discussion

Monoclonal antibody based therapies against malignancy are clearly effective in situations where there is a cell surface target on the malignant cells. Although there are excellent response rates and prolongation of survival in some studies, particularly when used in combination with chemotherapy,19–22 these responses are often incomplete or not durable. Additional therapies with synergistic activities but non overlapping toxicities are required to improve clinical outcomes.

In our study we showed that CD16+Vγ9Vδ2 T cells with the potential for ADCC can be markedly expanded in vitro in response to zoledronate stimulation and these cells bind to tumor cells through CD16 interactions in the presence of monoclonal antibody drugs. This leads to activation of these cells accompanied by secretion of perforin, IFN-γ and chemokine (RANTES) and cytotoxicity against tumor cells. Our study demonstrates the feasibility of using CD16+Vγ9Vδ2 T cells in combination with tumor targeting monoclonal antibodies.

Unlike vaccine based immune therapy strategies for B cell lymphoma or CLL, in which developing T cell and antibody based immune responses may be hindered by coadministration of immune suppressing chemotherapy and rituximab, respectively, ex vivo expanded CD16+ cytotoxic T cells can be administered by adoptive transfer simultaneously or soon after conventional therapy providing the opportunity for additive antitumor activity. The same is true for vaccine strategies with respect to T cell responses except that therapeutic monoclonal antibodies against solid tumor, such as trastuzumab, do not impair antibody responses.

The observation that our results for the rituximab/Vγ9Vδ2 T cell combination are mirrored for trastuzumab/Vγ9Vδ2 T cell combination indicates that our observations are not unique to rituximab and B lineage malignancy and suggests that the combination of monoclonal antibodies and adoptively transferred CD16+ cells may be broadly applicable to a range of malignancies. There are an expanding number of therapeutic monoclonal antibodies in clinical trials and routine use for a range of hematological and solid tumors40 providing many potential partnerships with CD16+Vγ9Vδ2 T cells or other CD16+ cells.

In our cultured Vγ9Vδ2 T cells, a CD16+ population was observed in CD45RA-CD27- (effector memory) cells and occasionally in CD45RA-CD27+ (central memory) cells. This finding contrasts with what has been previously observed for fresh, uncultured peripheral blood in which CD16+γδ T cells were exclusively CD45RA+CD27- (CD45RA+effector memory) cells.15 It is known that phosphoantigen stimulated CD45RA+ effector memory cells have low expansion capacity41 and it is difficult to generate CD16+γδ T cells through this method of stimulation. However, our findings show that CD16+γδ effector memory T cells with significant cytotoxicity in combination with monoclonal antibody drugs can be generated. The expression of CD16 on Vγ9Vδ2 T cells is less than that of NK cells, however, CD16+Vγ9Vδ2 T cells have substantial activity as ADCC effectors, and this is only slightly less than that of NK cells in our systems.

Ex vivo activation and expansion of CD16+Vγ9Vδ2 T cells may provide an effective alternative or addition to previously explored strategies to enhance ADCC by in vivo or ex vivo proliferation and activation of NK cells.42–46 One commonly employed method for inducing NK cell proliferation and activation in vivo is low-dose IL-2 administration but this has a more pronounced effect on CD56bright (CD16dim/−) cells, which express high-affinity IL-2 receptors, than on CD56dim (CD16bright) cells.42, 44, 47 Even if NK cell numbers in patients' peripheral blood doubles following low-dose IL-2 administration, ADCC ability of PBMC was shown to not significantly increase in one study42 and to increase only 30% in another.48 The ease with which CD16+Vγ9Vδ2 T cells can be expanded and their potential for antitumor cytotoxicity through multiple, additional mechanisms suggests they warrant further in vivo clinical evaluation in parallel with studies evaluating other immune effector cell populations.

There are distinct advantages of pursuing Vγ9Vδ2 T cells for adoptive immune therapy. Vγ9Vδ2 T cells can be readily generated from cryopreserved MNCs collected by leukapherasis for use as effector cells against solid tumors (Nicol et al., unpublished observation). This means that autologous MNC containing Vγ9Vδ2 T cells can be cryopreserved from cancer patients prior to treatment with chemotherapy or immunosuppressive agents, both of which may suppress the capacity to generate clinically useful numbers of Vγ9Vδ2 T cells. This also makes scheduling adoptive transfer of Vγ9Vδ2 T cells in combination with monoclonal antibodies and/or chemotherapy relatively easy.

In the absence of monoclonal antibodies the range of tumor cells recognized by Vγ9Vδ2 T cells differs from that recognized by NK cells.49, 50 Vγ9Vδ2 T cells and NK cells may therefore exert complementary cytotoxic activity against tumor cells. The recent finding that tumor infiltrating lymphocytes contain a high percentage of γδ T cells51 is encouraging evidence that adoptively transferred γδ T cells have the ability to infiltrate tumor sites. These previous observations are strengthened by our recent data that adoptively transferred indium111-labeled Vγ9Vδ2 T cells migrate to tumor sites (Nicol. et al, unpublished observation).

An interesting possibility is that “marking” the tumor cells with tumor specific monoclonal antibodies may facilitate binding of CD16+Vγ9Vδ2 T cells to the tumor cells resulting in local chemokine release that may facilitate recruitment of other immune effector cells to tumor sites. The chemokine release profile of CD16+Vγ9Vδ2 T cells activated by antibody-coated tumor cells might be different from that of NK cells, providing additional impetus to exploring adoptive transfer of Vγ9Vδ2 T cells in parallel with ongoing studies involving NK cells. Secretion of RANTES from NK cells in the presence of IL-2 or IL-12 did not increase on costimulation with trastuzumab-coated HER2-overexpressing tumor cells, though secretion of some other chemokines increased.52 In contrast we show that secretion of RANTES from CD16+Vγ9Vδ2 T cells was increased by stimulation of trastuzumab-coated HER2-overexpressing tumor cells.

Recent data suggesting potential antigen presenting cell characteristics of γδ T cells53 provides an additional possible role of Vγ9Vδ2 T cell therapy. We speculate that combination of monoclonal antibodies and Vγ9Vδ2 T cells may lead to tumor cell death in an environment containing various cytokines and chemokines, including RANTES and IFN-γ observed in our study to be released by Vγ9Vδ2 T cells, conducive to the generation or enhancement of antigen specific immune responses. In addition, the direct binding of Vγ9Vδ2 T cells to tumor cells via CD16 may facilitate transfer of antigens from apoptotic tumor cells to Vγ9Vδ2 T cells, for example by receptor-mediated endocytosis via Fcγ-receptors.54

Impairment of γδ T cell number and function in patients with malignancy is one potential hurdle to the broader use of these cells in adoptive immune therapy.10 For example, in one study about 50% of patients with myeloma did not respond to pamidronate and IL-2 in vitro8, 10 and in another study of patients with renal cell carcinoma, 4 of 15 cultures stimulated by synthetic phosphoantigen and IL-2 contained less than 20% of Vγ9Vδ2 T cells.9 We have made similar observations using samples from subjects with a range of malignancies (data not shown). We are uncertain why some but not all patient Vγ9Vδ2 T cells respond to in vitro culture but we speculate one reason of poor proliferation of Vγ9Vδ2 T cells is the effect of regulatory T cells (Treg). A previous study showed that the percentage of Treg was high in circulating blood of malignant melanoma patients55 and suggested that Treg have a negative effect upon Vγ9Vδ2 T cell proliferation in vitro.56 We are continuing to investigate this issue. Until methods to generate larger numbers of Vγ9Vδ2 T cells in poor expanders are developed, trials of adoptive immune therapy with Vγ9Vδ2 T cells will need to be restricted to the population of patients who do expand well in vitro.

In conclusion, we show in vitro that expanded CD16+Vγ9Vδ2 T cells have greater antitumor activity through ADCC when used in combination with tumor targeting monoclonal antibody drugs. Our data demonstrating that large numbers of CD16+Vγ9Vδ2 T cells with significant antitumor activity via ADCC can be generated from healthy donors and a proportion of patients with active malignancy supports the possibility of their clinical application in allogeneic and autologous settings.

Ancillary