Background
Triple-negative breast cancers (TNBC), an aggressive form of breast cancer that lacks significant expression of the human epidermal growth factor receptor 2 (HER2), estrogen receptor (ER), and progesterone receptor (PR), accounts for approximately 15~20% of invasive breast cancers. In the absence of obvious targets, patients with TNBC do not benefit from endocrine therapy or other available targeted agents [
1]. To date, the standard treatment still depends on surgery and adjuvant chemotherapy and radiotherapy. Patients with TNBC have a worse outcome after chemotherapy, compared to breast cancers patients with other subtypes [
2], a finding that reflects the intrinsically adverse prognosis associated with the disease. Thus, effective therapeutic strategies are urgently needed for TNBC patients.
Cancer cells including TNBC cells frequently upregulate “stress” induced ligands recognized by the NK cell activating receptors NKG2D (natural-killer group 2, member D) and DNAM-1(CD226) [
3,
4]. Therefore, the adoptive transfer of NK cells may represent a promising treatment strategy for these cancers. Large numbers of autologous NK cells can be infused to patients but often do not mediate tumor regression [
5]. The feasibility of targeting NKG2D ligands (NKG2DLs) utilizing chimeric antigen receptor (CAR) engineered T cell approach was demonstrated by Sentman and colleagues [
6] in 2005, and early clinical trial results now show the significant promise of this approach [
7]. This CAR construct contained the full-length NKG2D fused to the cytoplasmic domain of CD3z with costimulation provided endogenously by Dap10. We and others [
8,
9] have utilized 4-1BB or CD28 signaling platform-based NKG2D CAR T cells, in which the NKG2D extracellular domain (ECD) was connected to a transmembrane portion of the platform in a reverse orientation that maintained the ligand binding specificity and function. In these CARs, the human NKG2D ECD recognize several distinct ligands, including the MIC (MHC class I-related chain) family and six members of the ULBP/RAET (UL16-binding protein, or retinoic acid early transcript) family [
10], which are generally absent or expressed at low levels by healthy tissues but widely expressed on cancer cells. In the present study, we identify for the first time NKG2DL-expressing TNBCs as being sensitive to NKG2D CAR T cell attack, offering a new strategy for effective therapy. Further, we identify an unexpected role for IL-2 in the self-enrichment of first-generation NKG2D CAR T cells as well as the impact of 4-1BB and CD27 costimulatory signaling domains in NKG2D CARs in generating more potent T cells against TNBCs in vitro and in vivo.
Methods
Cell lines
Human cell lines used in immune-based assays include the established human breast cancer cell line MCF7 and TNBC cell lines MDA-MB-231, MDA-MB-436, MDA-MB-468, MDA-MB-453, and BT549. The mouse malignant mesothelioma cell line, AE17 (kindly provided by Steven Albelda, University of Pennsylvania), was used as antigen negative control. For bioluminescence assays, the cancer cell lines were transfected to express firefly luciferase (fluc). Lentivirus packaging was executed using the immortalized normal fetal renal 293T cell line purchased from ATCC. All cell lines were maintained in complete medium: RPMI-1640 supplemented with 10% heat inactivated FBS, 100 U/ml penicillin, and 100 mg/ml streptomycin sulfate.
CAR construction and lentivirus production
The NKG2D CAR constructs are comprised of the extracellular portion of human NKG2D (aa 82–216) linked to a CD8a hinge and transmembrane region, followed by a CD3z signaling moiety alone (NKG2D-z) or in tandem with the 4-1BB or CD27 intracellular signaling motif, which was previously described [
11,
12]. CAR sequences were preceded in frame by a green fluorescent protein (GFP) sequence followed by the 2A ribosomal skipping sequence.
High-titer replication-defective lentivirus were produced and concentrated as previously described [
13]. Briefly, 293T cells were seeded in 150-cm
2 flask and transfected using
TurboFect (Life Technologies) according to the manufacturer’s instructions. NKG2D CAR transgene plasmid (15 μg) was co-transfected with 18 μg pRSV.REV (Rev expression plasmid), 18 μg pMDLg/p.RRE (Gag/Pol expression plasmid), and 7 μg pVSV-G (VSV glycoprotein expression plasmid) with 174-ul transfection reagent Express In (1 μg/ul) per flask. Supernatants were collected at 24 and 48 h after transfection, concentrated tenfold by ultracentrifugation for 2 h at 28,000 rpm with a Beckman SW32Ti rotor (Beckman Coulter). The viruses were aliquoted and stored at − 80 °C until ready to use for titering or experiments. All lentiviruses used in the experiments were from concentrated stocks.
Human T cells and transfection
Primary human T cells, purchased from the Human Immunology Core at University of Pennsylvania, were isolated from healthy, normal donors following leukapheresis by negative selection. All T cell samples were collected under a protocol approved by a University Institutional Review Board, and written informed consent was obtained from each healthy, normal donor. T cells were cultured in R10 medium and stimulated with anti-CD3 and anti-CD28 monoclonal antibodies (mAb)-coated beads (Invitrogen). Approximately 18 to 24 h after activation, human T cells were transduced. Briefly, 0.5 × 106 T cells were infected with a multiplicity of infection (MOI) 2 of the NKG2D receptor lentiviral vector and expanded for 2 weeks. Human recombinant interleukin-2 (IL-2; Novartis) was added every 2–3 days to a 50-IU/ml final concentration, and a cell density of 0.5 × 106 to 1 × 106 cells/ml was maintained. Engineered CAR T cells were rested in cytokine-free medium for 24 h and were then used for functional analysis.
Flow cytometric analysis
The following fluorochrome-conjugated monoclonal antibodies, purchased from BD Biosciences, were used for T cell phenotypic analysis: APC-Cy7 anti-human CD3, FITC anti-human CD4, APC anti-human CD8, PE anti-human CD45, and APC anti-human NKG2D. PE anti-human CD137, APC anti-human PD-1, and Pacific Blue anti-human TIM-3 were purchased from Biolegend. 7-Aminoactinomycin D (7-AAD) was used for viability staining. Expression of NKG2D CAR was detected by GFP and surface NKG2D expression using anti-NKG2D Ab. For the in vivo experiments, peripheral blood was obtained via retro-orbital bleeding and stained for the presence of human CD45, CD4, and CD8 T cells. Gating specifically on the human CD45+ population, the CD4+ and CD8+ subsets were quantified using TruCount tubes (BD Biosciences) with known numbers of fluorescent beads as described in the manufacturer’s instructions.
NKG2DLs were analyzed using PE anti-MICA/B (clone 6D4, BD Pharmingen), PE anti-ULBP1 (clone 170818, R&D System), PE anti-ULBP2/5/6 (clone 165903, R&D System), anti-ULBP3 (clone 2F9, Santa Cruz), and polyclonal Per-CP-anti-human ULBP4 (R&D System). Recombinant human NKG2D Fc chimera and a control recombinant human folate receptor-alpha (FRA) Fc chimera were purchased from R&D System. Matched secondary and isotype antibodies were used in all analyses. Flow cytometry was performed on BD FACSCanto II flow cytometer, and flow cytometric data were analyzed using FlowJo Version 7.2.5 software.
Cytokine release assays
Cytokine release assays were performed by co-culturing 1 × 105 T cells with 1 × 105 target cells in triplicate in a 96-well flat bottom plate in a total volume of 200-ul R10 media. After 20~24 h, cells free co-culture supernatants were collected and ELISA (Biolegend, San Diego) was performed, according to manufacturer’s instructions, to measure the secretion of IFN-γ. The values shown represent the mean of triplicate wells.
Cytotoxicity assays
For cell-based bioluminescence assays, 5 × 104 firefly luciferase (fLuc)-expressing tumor cells were cultured with R10 media in the presence of different T cell ratios in a 96-well microplate (BD Biosciences). After incubation for ~ 20 h at 37 °C, each well was filled with 50 ul of d-luciferin (0.015 g/ml) resuspended with PBS and imaged with the Xenogen IVIS Spectrum. Tumor cell viability percentage was calculated as the mean luminescence of the experimental sample minus background divided by the mean luminescence of the input number of target cells used in the assay minus background times 100. All data are represented as a mean of triplicate wells.
Additional cytotoxicity of NKG2D CAR-T cells was measured in real time using an xCELLigence (ACEA Bioscience) label-free, impedance-based cell sensing device. AE17, MDA-MB-436, MDA-MB-468, or BT549 cells (2 × 104/well) were left to adhere for 24 h to xCELLigence E-plates. Cell proliferation was measured as a change in relative impedance, termed cell index (CI). After 24 h, the effector cells were added at the ratio of the effector to target cell (E:T = 2:1, 1:1, and 1:2) in a total volume of 200 μl. The cytotoxicity was then monitored by measuring changes in impedance as CI values recorded by the xCELLigence RTCA SP device.
Xenograft model of TNBC
NOD/SCID/γ-chain-/- (NSG) mice were bred, treated, and maintained under pathogen-free conditions in-house under the University of Pennsylvania IACUC-approved protocols. All animals were obtained from the Stem Cell and Xenograft Core (SCXC) of the Abramson Cancer Center, University of Pennsylvania. To establish a TNBC model, 6~10-week-old female NSG mice were inoculated subcutaneously (s.c.) on the flank with 3 × 106 MDA-MB-231 fluc(+) cells on day 0. After the tumors become palpable at about 3 weeks, primary human T cells were activated and transduced. After the primary human T cells were expanded for 2 weeks and the mouse tumor burden was about 200~300 mm3, the mice were treated with the T cells. To investigate the roles of CD27 and 4-1BB costimulated NKG2D CAR in vivo, NKG2D CAR T cells were adjusted from 90 to ~ 30% by adding UNT T cells. The route, dose, and timing of T cell injections are indicated in the individual figure legends. Tumor dimensions were measured with the calipers, and tumor volumes calculated using the formula V = 1/2(length × width2), where length is the greatest longitudinal diameter and width is the greatest transverse diameter. Animals were imaged prior to T cell transfer and 3 weeks thereafter to evaluate tumor growth. Photon emission from fluc+ cells was quantified using the “Living Image” software (Xenogen) for all in vivo experiments. Approximately 50 days after the first T cell injection, the mice were euthanized and the tumors were resected immediately.
Bioluminescence imaging
Tumor growth was also monitored using bioluminescent imaging (BLI). BLI was conducted using Xenogen IVIS imaging system. The photons emitted from fLuc-expressing cells within the animal body were quantified using Living Image software (Xenogen). Briefly, mice bearing MDA-MB-231fLuc tumors were injected intraperitoneally (i.p.) with d-luciferin (150 mg/kg stock, 100 μL of d-luciferin per 10 g of mouse body weight) suspended in PBS and imaged under isoflurane anesthesia after 5~10 min. A pseudocolor image representing light intensity (blue, least intense; red, most intense) was generated using Living Image. BLI findings were confirmed at necropsy.
Statistical analysis
The data are reported as means and standard deviations (SD). Statistical analysis was performed using two-way repeated-measures analysis of variance (ANOVA) for the tumor burden (tumor volume, photon counts). The Student t test was used to evaluate differences in absolute numbers of transferred T cells, cytokine secretion, and specific cytolysis. GraphPad Prism 5.0 (GraphPad Software) was used for the statistical calculations, where a P value of P < 0.05 was considered significant.
Discussion
The expression of NKG2DLs on many primary tumor cells and immunosuppressive cells (e.g., T regulatory cells and myeloid-derived suppressor cells) within the tumor microenvironment makes them attractive targets for the development of novel therapeutics [
17]. Targeting NKG2DLs with NKG2D CAR T cells has been shown to induce tumor elimination and long-term tumor-free survival in various tumor models [
18‐
20]. NKG2DLs are frequently expressed in breast cancer [
21], and we found positive surface expression of NKG2DLs on all (5/5) human TNBC cell lines tested by flow cytometry. Based on these results, we rationalized the extension of our NKG2D CAR T cell-based immunotherapy approach to TNBCs.
NKG2D CAR T cells secreted IFN-γ upon stimulation with NKG2DL (+) tumor cells and displayed potent cytolytic capacity in vitro against NKG2DLs+ TNBC cells, even at low
E/
T ratios. Consistently, the production of proinflammatory cytokines and cytolytic capacity by NKG2D-BBz and NKG2D-27z CAR-T cells was substantially increased after co-culture with NKG2DL (+) TNBC cell lines, compared with NKG2D-z T cells. Mechanisms accounting for increased effector function by 4-1BB and CD27 costimulated CAR-T cells in vitro appear linked in part to their ability to resist antigen-induced cell death (AICD) as we reported previously [
11]. These in vitro tumor killing findings further support the notion that NKG2DLs have promise as novel immunotherapy targets for TNBCs, which currently lack effective targeted therapies. Indeed, two injections of 4-1BB or CD27 costimulated NKG2D CAR T cells exhibited in vivo anti-tumor effects in a highly invasive MDA-MB-231 xenograft model of human TNBC, compared to the first-generation NKG2D CAR T cells. Consistent with clinical observations [
22], tumor regression was associated with enhanced T cell persistence in vivo. The greatest number of CAR T cells persisting in the blood 20 days after the first T cell dose was observed in those animals administered NKG2D-BBz and NKG2D-27z CAR-T cells followed by NKG2D-z and untransduced T cells, indicating that simultaneous TCR CD3 signaling and 4-1BB or CD27 costimulation triggered by CAR ligation with antigen improves upon TCR signaling alone, implicating a role for 4-1BB and CD27 costimulation in memory T cell formation in vivo. Results of comparative in vivo studies of CARs containing these various costimulatory domains demonstrated that both 4-1BB and CD27, members of the TNFR superfamily, enhances T cell anti-tumor activity and persistence.
In this study, costimulated NKG2D CAR T cells failed to completely eradicate large, established tumors. Major studies suggest that administration of exogenous IL-2 may be necessary for the anti-tumor activity for the first-generation CAR T cells that lack costimulatory domains [
23,
24]. We have previously found that the administration of IL-2 has little to no anti-tumor effect on human CAR T cells activity in immunodeficient mice, if the CAR contains a costimulatory domain [
25]. Therefore, the omission of exogenous IL-2 is not likely the primary factor accounting for the suboptimal anti-tumor response in this study. The efficacy of NKG2D CAR T cell treatment for TNBC may also be influenced by surface antigen expression level as suggested previously [
26] in the same MDA-MB-231 tumor model. Although MDA-231 cells express various NKG2DLs, a previous study [
27] suggested that NKG2D can bind to only one ligand at a time. In addition, NKG2D displays varying affinities for its ligands that ranges between 600 nM and 1.1 mM [
28] and NKG2D CAR may preferentially bind to the higher affinity form of ligands. Therefore, further investigation of affinity of each ligand’s binding to NKG2D receptor and selecting patients with higher NKG2DL expression may help to enhance the anti-tumor activity of NKG2D CAR T cells in clinical studies. Other antigens such as folate receptor-a [
26], mesothelin [
29], and TEM8 [
30] were also investigated as CAR targets for TNBCs with mixed efficacy results reported. Targeting solid tumors, like TNBC, is still challenging, as it may be greatly hampered clinically by the immunosuppressive microenvironment and the inefficient homing of CAR T cells to tumor sites [
31]. Combining CAR T cells with other therapies like CTLA-4 and PD-1 inhibitor offers the potential to improve anti-tumor effects.
One intriguing finding from this study is that in the presence of exogenous IL-2, a potent cytokine stimulator of activated effector lymphocyte expansion and proliferation, NKG2D CAR T cells, undergoes substantial long-term expansion without evidence of prolonged detriment from fratricide in vitro. In addition, the NKG2D CAR (+) T cell population was enriched during prolonged culture, suggesting a possible auto-stimulatory effect through endogenous expression of induced NKG2DLs after T cell stimulation. These findings are similar to a CD28 costimulated CD5-specific CAR [
32], which kills both CD5(+) autologous T cells and malignant T cell lines, yet results in limited and transient T cell fratricide. However, by incorporating 4-1BB, CD27, OX40, CD30, CD28, ICOS, or HVEM into the CD5 CAR, a more recent study suggested that only CD5 CARs costimulated with non-TNFR domains (CD28, ICOS, or non-costimulated) overcome impaired the expansion of CAR T cells and resulted in complete or partial downregulate CAR expression [
33]. Unlike the CD5 CAR platform, which differs significantly in expansion and transduction methodology and antigen specificity, our NKG2D CAR T cells undergo dramatic expansion and preferentially enrich for CAR expressing T cells during culture. Here, exogenous IL-2 promoted the expansion and enrichment of NKG2D-z, NKG2D-BBz, and NKG2D-27z CAR (+) T cells, which was dependent on a high expression level of the IL-2 receptor-alpha subunit, CD25. Without IL-2 support, T cells expressing any of the NKG2D constructs did not expand well and only NKG2D CAR (+) T cells containing 4-1BB or CD27 costimulatory domains enriched for CAR-positivity, demonstrating a benefit from 4-1BB and CD27 costimulation in vitro. These results suggest that incorporation of costimulatory domains into CAR constructs promotes preferential T cell survival and resistance to AICD by upregulating anti-apoptotic proteins as suggested previously [
11]. These data also support the three-signal hypothesis that optimal T cell activation and expansion involves T cell receptor (TCR) activation (signal 1) in addition to costimulatory receptor engagement (signal 2) and cytokine receptor engagement (signal 3) [
34]. Although NKG2DLs can be expressed on active T lymphocytes, our NKG2D CARs do not elicit tonic signaling, as seen with other CARs reported previously [
16,
35]. In one report [
35], tonic signaling CAR T cells displayed sustained proliferation for up to 3 months, resulting from autocrine CAR crosslinking that is independent of cognate antigen and did not require the addition of exogenous cytokines or feeder cells after a single stimulation of the TCR and CD28. NKG2D-z, NKG2D-BBz, and NKG2D-27z CAR (+) T cells do not undergo prolonged and unfettered proliferation. Whether regulated proliferation in NKG2D CAR T cells is impacted by the selection of the 4-1BB and CD27 TNFR costimulatory domains, and not CD28, is not known. However, in the setting of tonic CAR signaling, 4-1BB costimulation reduces exhaustion induced by persistent CAR signaling and augments anti-tumor activity in vivo, while CD28 costimulation does not [
16].
In light of the therapeutic potential of NKG2D CAR T cells for TNBC and because of the potential for expression of NKG2DLs on healthy tissues, the concerns about potential “on-target, off-tumor” toxicity must be considered. VanSeggelen et al. [
36] described lethal toxicity in mice treated with murine NKG2D CAR T cells. This toxicity was both CAR construct and strain dependent and was exacerbated upon the use of a lymphodepleting conditioning regimen. These results suggested that predicting the toxicity of NKG2D CAR approach becomes especially problematic because of the diversity of target ligands. Moreover, both efficacy and toxicity will likely be affected by the ligand tissue distribution, density, and affinity with which NKG2D binds the respective ligand. None of these parameters are very well characterized and likely vary greatly between mice and humans [
37]. However, results from a phase I clinical study [
38] testing the safety of NKG2D CAR-T cells in patients with AML/MDS and multiple myeloma (
ClinicalTrials.gov Identifier: NCT03018405) are promising with no reported cases of cytokine-release syndrome, CAR T cell-related neurotoxicity, autoimmunity, or patient death. It is important to have adequate safety data in multiple clinical studies using different NKG2D CAR constructs. In the current ongoing clinical trial [
38], the full-length NKG2D is used in a CAR design that incorporates Dap10 signaling, which was showed the enhanced CAR expression and toxicity [
36]. Notably, one recent case study reports that NKG2D-based chimeric antigen receptor therapy safely induced remission in a patient with relapsed/refractory acute myeloid leukemia [
7]. Our NKG2D CAR construct contains 4-1BB and lacks DAP10, yet still possesses potent anti-tumor capacity against TNBC. Therefore, future clinical studies are warranted.