Background
Fueled by impressive clinical results, adoptive T-cell therapy (ACT) is gaining more and more momentum in the battle against hematologic malignancies and solid tumors. Conferring specificities for tumor antigens on T cells via antigen-receptor transfer has led to substantial tumor elimination in clinical trials [
1‐
3].
T cells can either be engineered to express an additional α/β T-cell receptor (TCR) recognizing MHC-presented intracellular tumor antigens or they can be equipped with a chimeric antigen receptor (CAR) targeting surface antigens [
4]. A CAR is usually created by merging an antibody-derived single chain variable fragment (scFv), the CD3ζ chain of a TCR and an additional co-stimulatory domain, such as CD28 [
5].
On-target/off-tissue and off-target toxicities induced by receptor-engineered T cells attacking non-malignant host cells still remains a feared side-effect in adoptive T-cell therapy [
6]. This can be caused by various mechanisms: i) due to the (unexpected) expression of tumor antigens on healthy tissue [
7‐
10] or due to the cross reaction of tumor-specific receptors with host molecules [
11,
12], ii) via random formation of new specificities originating from an α/β-chain-mispairing of the introduced α/β TCR with the endogenous α/β TCR, which may create a reactivity for host molecules [
13,
14], and iii) owing to the possible re-activation of anergized self-reactive T cells by signaling through the transfected receptors. Another dreaded side-effect occurs as a result of the T-cell on-target over-activation culminating in a cytokine release syndrome [
15,
16].
Therefore it is crucial for clinical deployment to search for mechanisms enhancing the safe application of engineered T cells. With respect to patient safety, transient receptor transfer, for instance via mRNA electroporation, is superior to permanent DNA-based transfection, as arising on-target/off-tissue and off-target toxicities are transient as well [
17‐
20]. The successful receptor-transfer via mRNA electroporation has been well-established for many years [
19,
21‐
23]. To obviate safety concerns emanating from a potential α/β chain mispairing with the endogenous α/β TCR, or a potential reactivation of dormant self-reactive T cells, the use of γ/δ T cells presents as an ideal solution, because the endogenous γ/δ TCR does not pair with α- or β-chains [
24‐
26]. Equally important, activation of the endogenous γ/δ TCR will not result in autoimmunity.
Contrary to conventional CD8
+ T cells, γ/δ T cells, which constitute up to 10% of peripheral blood T cells, do not recognize MHC-bound peptides [
27,
28]. They are activated by phosphoantigens and aminobisphosphonates, such as isopentylpyrophosphate (IPP), which is upregulated in the context of metabolical disorders [
29]. As tumorigenesis frequently involves disturbing the metabolism of incipient and bona fide malignant cells, γ/δ T cells play a very important role in tumor surveillance [
30]. Moreover, γ/δ T cells were found to express the NKG2D receptor [
31], which is also beneficial in tumor rejection, as tumors, e.g. Ewing’s sarcoma, often upregulate NKG2D ligands [
32]. Importantly, owing to the MHC-independent targeting by γ/δ T cells, tumor cells cannot escape via MHC-downregulation. As for transient receptor expression using mRNA electroporation into γ/δ T cells, it can be hypothesized that the initial strike exerted by the introduced receptor may be prolonged via the endogenous receptor repertoire.
Clinical trials evaluating the intrinsic anti-tumor activity of ex vivo expanded γ/δ T cells showed promising results [
33‐
35]. Other studies focused on boosting the anti-tumor response of γ/δ T cells in vivo with systemically administered zoledronic acid, which also proved to be successful [
36]. Zoledronic acid interferes with cholesterol synthesis by blocking the enzyme farnesyl diphosphate synthase, which leads to the accumulation of the γ/δ TCR ligand IPP [
37]. On top of that, γ/δ T cells, genetically engineered to express a CAR specific for CD19, exhibited anti-tumor activity in vitro and in murine leukemia models [
38]. In addition, γ/δ T cells, lentivirally transfected with an α/β TCR, displayed an antigen-specific lysis of leukemic cells [
39].
We have already demonstrated the feasibility of transferring a virus-specific α/β TCR into γ/δ T cells via mRNA electroporation to combat adenovirus infections ensuing hematopoietic stem cell transplantations [
40] and of an NKT-TCR to render the γ/δ T cells responsive for the NKT ligand α-galactosylceramide [
41]. As endogenous γ/δ TCRs do not recognize MHC molecules [
27,
28], they do not evoke graft-versus-host disease after transferring γ/δ T cells into HLA-mismatched recipients [
42]. With regard to adoptive T-cell therapy, this implies that γ/δ T cells can be obtained from multiple sources, including healthy donors, whose T cells are not compromised by tumor- or therapy-related immunosuppression [
43,
44]. Thus, adoptive T-cell therapy could be applied to a multitude of patients, irrespective of their T-cell numbers and HLA-type.
Regarding malignant melanoma, which still ranks high among tumors with bad prognoses [
45], there are various tumor antigens that can be exploited in adoptive cell therapy. The melanosomal membrane-protein glycoprotein 100 (gp100), which is enriched in melanocytes and melanoma cells [
46,
47], can be targeted with an α/β TCR [
19]. Besides, 90% of melanoma lesions express the surface protein melanoma-associated-chondroitin-sulfate-proteoglycan (MCSP), also known as chondroitin sulfate proteoglycan 4 (CSPG4) or high molecular weight-melanoma-associated antigen (HMW-MAA) [
48], which can be attacked with CARs [
49]. Of note, MCSP is also present on other tumor entities such as gliomas, [
50] sarcomas, [
51], and triple-negative breast cancer, [
52] and on other cells within the tumor, like activated pericytes [
53,
54]. Some healthy tissues, e.g. smooth muscle cells, express MCSP, but to a much lower extent [
55].
In this study we aimed at establishing a protocol for the expansion and transfection of γ/δ T cells to generate clinically applicable numbers, which can easily be adapted to GMP-compliant production. We investigated the functionality of γ/δ T cells, transfected with either an α/β TCR specific for the melanoma-related antigen gp100 [
19] or a second generation CAR directed against the membrane-bound melanoma antigen MCSP [
49] using mRNA electroporation, in direct comparison to CD8
+ T cells. To our knowledge, this is the first study to evaluate the functional transfer of a CAR and a melanoma-specific α/β TCR into γ/δ T cells by means of mRNA electroporation [
56,
57].
Methods
Cells and reagents
At first healthy blood donors (aged: 19–63 years) willing to voluntarily participate in this study were selected from a donor database pre-existent in our group by availability and willingness to donate blood at the required date and time. The possibility to serve as a blood donor was publicly announced and each eligible candidate was educated and approved by a medical doctor after basic blood examinations.
Second, peripheral blood mononuclear cells (PBMC) were extracted from whole-blood, procured from those healthy donors following written informed consent and approved by the institutional review board (reference number: 166_14 B), via density centrifugation using lymphoprep (Axis-Shield, Oslo, Norway). Approximately 80% of the obtained PBMC were subjected to a two-step magnetic-activated cell sorting (MACS) according to the manufacturer’s instructions (Miltenyi, Bergisch-Gladbach, Germany) to successively isolate γ/δ+ T cells and CD8+ T cells. Purified T cells and remaining PBMC were resuspended at 106 cells/ml before expansion (see below) in R10 medium consisting of RPMI 1640 (Lonza, Basel, Switzerland) supplemented with 2 mM L-glutamine (Lonza), 100 IU/ml penicillin (Lonza), 100 mg/ml streptomycin (Lonza), 10% (v/v) heat-inactivated fetal calf serum (PAA, GE healthcare, Piscataway, NY, USA), 2 mM HEPES (PAA, GE healthcare), and 2 mM β-mercaptoethanol (Gibco, Life Technologies, Carlsbad, CA, USA). Note that the fetal calf serum would have to be replaced by human serum for full GMP compliance.
Target cell lines incorporated the TxB cell hybridoma
T2.A1 (HLA-A2
+, gp100
−, MCSP
−; kind gift from Prof. Dr. Schulz, Nuremberg), and the melanoma cell lines
Mel526 (HLA-A2
+, gp100
+, MCSP
+; kind gift from Prof. Dr. Hinrich Abken, Köln) and
A375M (HLA-A2
+, gp100
−, MCSP
+; kind gift from Dr. Aarnoudse, Leiden, Netherlands; ATCC CRL-3223). The human lymphoma cell line Daudi (ATCC CCL-213) was a kind gift from Dr. Manfred Smetak (Nuremberg). Target cells were cultured in R10 medium, before undergoing co-incubation with effector cells.
Mel526 and
A375M were additionally pulsed with the HLA-A2-restricted peptide gp100
280–288 (YLEPGPVTA) as previously described [
58] where indicated. Peptide-pulsing was performed in DC-medium, which consists of RPMI 1640 (Lonza), 1% human serum (Sigma-Aldrich, Taufkirchen, Germany)(heat-inactivated, 30 min, 56 °C), 2 mM L-glutamine (Lonza), and 0.04% 20 mg/l gentamycin (Lonza).
T-cell expansion
PBMC were directly activated (on the day of isolation) with a single dose of zoledronic acid (Zoledronsäure HEXAL®
, HEXAL, Germany) applied at a final concentration of 5 μM [
59] or with 0.1 μg/ml anti-CD3 antibody OKT3 (Orthoclone OKT3; Jannsen-Cilag, Neuss, Germany). Concomitantly, MACS-isolated γ/δ
+ T cells and CD8
+ T cells were stimulated with 0.1 μg/ml OKT3 directly after isolation (on the same day). Ensuing T-cell expansion was performed in alignment with a GMP-compliant protocol devised by our group [
60]. In brief, 1000 IU/ml interleukin-2 (Proleukin; Novartis, Nuremberg, Germany) was administered on days 0, 2, 3, 5, and 7. On day 3, cells were counted and re-adjusted to 0.2 × 10
6 cells/ml by adding fresh medium. On day 7, the total cell culture volume was first doubled, and subsequently split by transferring half of the volume to a second culture flask. After 10–11 days, cells were counted and prepared for further experiments.
Flow cytometric analyses of phenotypic parameters
A FITC-labeled pan TCR γ/δ IgG1 antibody (Thermo Fisher Scientific, USA) was used in combination with PE-conjugated anti-CD3 IgG1 (ImmunoTools, Germany) and PE-conjugated anti-CD8 IgG1 (BD Biosciences, USA) antibodies to analyze the cellular composition of cell populations pre- and post-expansion. Unstained and isotype-stained cells served as controls. Immunofluorescence was measured using a FACScan cytofluorometer (BD Biosciences, Heidelberg, Germany) equipped with CellQuest software (BD Biosciences). Data were analyzed using FCS Express 5 (De Novo Software, USA).
In vitro transcription of RNA
A TCR specific for the HLA-A2-restricted peptide consisting of amino acids 280–288 (YLEPGPVTA) of the melanosomal glycoprotein 100 (gp100) and a second generation CAR (MCSP
HL-CD28/CD3ζ-CAR) directed against MCSP (melanoma-associated chondroitin sulfate proteoglycan) were used for transfer into T cells. The molecular compositions of both receptors were specified previously [
19,
51]. In vitro transcription of receptor-encoding mRNA was performed with T7 RNA polymerase (mMESSAGE mMACHINE T7 Ultra kit; Life Technologies, Carlsbad, CA, USA) according to the manufacturer’s instructions. Afterwards, RNA was purified on RNeasy columns (Qiagen GmbH, Hilden, Germany) according to the manufacturer’s instructions. RNA quality was assessed by agarose gel electrophoresis.
RNA electroporation
RNA transfection was executed as detailed elsewhere [
19,
61]. In short, following 10–11 days of expansion, T cells were washed in OptiMem (Life technologies, Carlsbad, CA, USA,) resuspended at 10
6 cells/ml and transferred to 4-mm gap electroporation cuvettes (Biolabproducts GmbH, Bebensee, Germany). Cells were either mock-electroporated (no RNA), transfected with 15 μg of RNA coding for the gp100/A2-specific TCR α- and β-chains [
19], with 15 μg RNA encoding the MCSP-specific CAR (MCSP
HL CD28-CD3ζ), or with 15 μg RNA encoding the carcinoembryonic antigen (CEA)-specific CAR (CEA CD28-CD3ζ) [
62] using a Gene Pulser Xcell (Bio-Rad, Hercules, CA, USA) at 500 V (square wave pulse) for 5 ms. After transfection, T cells were immediately transferred to R10 medium.
Surface expression of transfected receptors
Surface expression of the introduced receptors was analyzed flow-cytometrically 1, 2, 4, 6, and 20 h (Fig.
2c-f), and 4, 24, 48, 72, 96, and 120 h (Fig.
2g-j) after electroporation. The TCR was stained with a PE-conjugated anti-Vbeta 14 antibody (clone CAS 1.1.3; Beckman Coulter, Immunotech, Marseille, France) and the CAR was stained with a PE-conjugated goat-F(ab’)2 anti-human IgG antibody (Southern Biotech, Birmingham, AL, USA) directed against the extracellular
IgG1 CH2CH3 CAR-domain. Additionally, an aliquot of TCR-transfected cells was cryopreserved one day after electroporation. After thawing, a PE-conjugated MHC-Dextramer HLA-A*0201/YLEPGPVTV was employed to detect the transferred TCR.
Immunofluorescence was measured using the FACScan cytofluorometer (BD Biosciences) equipped with CellQuest software (BD Biosciences). Data were analyzed using FCS Express 5.
Cytokine secretion
Cytokine secretion by transfected T cells was assayed as described before [
58]. In short, 4 h after electroporation T cells were harvested and stimulated over night at a 1:1 ratio with UV-irradiated (0.005 J/cm
2) human lymphoma cell lines
T2.A1 and
Daudi, as well as with human melanoma cell lines
Mel526 and
A375M, which were employed either loaded with the HLA-A2-restricted peptide gp100
280–288 at a concentration of 10 μg/ml or nonloaded. Cytokine concentrations in the supernatants were determined using the Th1/Th2 Cytometric Bead Array Kit II (BD Biosciences) according to the manufacturer’s instructions. Immunofluorescence was detected using the FACS Canto II (BD Biosciences, Franklin Lakes, NJ, USA) equipped with FACSDiva software (BD Biosciences). Data were analyzed using FCS Express 5.
Additionally, intracellular cytokine staining was performed as follows: 4 h after electroporation, T cells were incubated over night at a 1:1 ratio with T2.A1 cells and gp100-pulsed A375M cells. Cytokine secretion was blocked by adding Brefeldin A and Monensin (eBioscience, San Diego, CA, USA) one hour after starting the co-incubation. For FACS analysis, cells were labeled with the viability marker LIVE/DEAD (Invitrogen, Carlsbad, CA, USA). Thereafter, fixation and permeabilization was performed using the Foxp3/Transcription Factor Fixation/Permeabilization Concentrate and Diluent kit (eBioscience, San Diego, USA). Next, APC-H7-CD3 (BD Biosciences), PE-TCRγ/δ (Life Technologies, Carlsbad, CA, USA), APC-IL-2 (BD Biosciences), PE-Cy7-TNF (BD Biosciences), and Alexa Fluor 700 IFNγ (BD Biosciences) antibodies were employed to stain surface markers as well as intracellular cytokines. Immunofluorescence was detected using the LSRFortessa (BD Biosciences) equipped with FACSDiva software. The gating strategy was applied as follows: After doublet exclusion, a gate was drawn around lymphocytes and finally dead cells were excluded by viability staining. Data were analyzed using FCS Express 5.
Cytotoxicity
Specific cytotoxicity of transfected T cells was examined with a standard 4–6 h
51chromium-release assay 24 h after electroporation, as previously described [
58]. In brief, human tumor cell lines
T2.A1,
Mel526,
A375M, and
Daudi were labeled with 20 μCi of Na
2
51CrO
4/10
6 cells (PerkinElmer, Waltham, MA, USA) for 1 h. An aliquot of A375M cells was additionally loaded with the HLA-A2-restricted peptide gp100
280–288 at a concentration of 5 μg/ml for 1 h. Subsequently, target cells were transferred to 96-well plates and co-incubated with effector cells at decrementing effector to target ratios. Chromium-release in the supernatants was measured after 4 h with the Wallac 1450 MicroBeta plus Scintillation Counter (Wallac, Turku, Finnland). Percentage of cytolysis was calculated as follows: [(measured release – background release)]/[(maximum release – background release)] × 100%.
Graphs were created and statistical analysis was performed using GraphPad Prism, Version 6 (GraphPad Software, USA). P-values were analyzed using the Students t-test. * indicates p ≤ 0.05, ** indicates p ≤ 0.01, and *** indicates p ≤ 0.001.
Discussion
In the current study we present a novel strategy for the adoptive T-cell therapy of melanoma. So far, the majority of adoptively transferred T cells has been based on conventional CD4
+ or CD8
+ T cells, which were transduced with DNA encoding either an additional α/β TCR or a CAR specific for tumor-antigens [
1‐
4]. To mitigate concerns about permanent off-target toxicities [
7‐
12] arising from the use of stably transduced α/β T cells, e.g., α/β chain mispairing [
13,
14], we explored the transient RNA-based introduction of melanoma-specific antigen receptors into γ/δ T cells. This is the first study to report on the GMP-compliant transfer of a TCR and a CAR into γ/δ T cells using mRNA electroporation for the use in adoptive T-cell therapy against melanoma.
We electroporated γ/δ T cells with mRNA encoding a gp100/HLA-A2-specific α/β TCR [
19] or a CAR specific for MCSP [
49]. Over the course of many years, mRNA electroporation has proven to be a reliable procedure to equip T cells with TCRs or CARs for a limited time-span [
19,
21‐
23]. Contrary to the permanent DNA-based transfection, electroporating receptor-encoding RNA is far safer, as potential off-target toxicities cease after the receptor expression has diminished. Therefore, RNA-transfection represents an ideal avenue for testing engineered T cells with respect to auto-reactivity [
19,
21]. This additional level of safety comes at the expense of a timely limited anti-tumor activity, which ceases while the receptor expression declines. This issue can be addressed by a variety of mechanisms: i) Repetitive injections of transfected T cells, which favors the use of γ/δ T cells, as they can be obtained in large quantities from healthy donors. ii) Direct application of receptor-transfected T cells into tumor lesions minimizes the time required for locating and accessing the tumor and maximizes the period, in which the tumor can be antigen-specifically targeted with the transiently-expressed receptors. As for melanoma, direct injection into cutaneous lesions is conceivable. But some locations of tumors are ill-suited for direct injection, such as the liver, which is a prime location of melanoma metastases [
67]. To maximize the concentration of effector T cells in the liver, we propose a direct application of transfected T cells into the metastases via a transarterial catheter. Regarding the burden this intervention imposes on the patient, up to two successive intrahepatic injections of transfected T cells would be feasible. This procedure of directly attacking liver metastases has been successfully applied in clinical practice for delivering radioactive isotopes to the tumor [
68,
69]. iii) The initial strike exerted by the transiently introduced receptors may be prolonged by the intrinsic anti-tumor activity of γ/δ T cells. One of the major traits of these cells is their ability to kill tumor cells via their endogenous receptor repertoire recognizing metabolic disorder and stress ligands [
29,
31]. This was experimentally validated by the detection of an increased lysis of melanoma cells by mock-electroporated γ/δ T cells in contrast to MACS-isolated CD8
+ T cells. Moreover, we hypothesize that the first strike exerted by the transfected receptors will further disrupt the metabolic homeostasis of tumor cells and result in an enhanced recognition by γ/δ T cells and thus maintain the anti-tumor response. In sum, these mechanisms provide attractive strategies to compensate the temporally confined anti-tumor activity associated with RNA-transfection.
Our functional characterizations concerning cytokine secretion and cytotoxic capacity toward melanoma cell lines showed that receptor-transfer via mRNA electroporation has conferred an additional specificity for melanoma antigens on γ/δ T cells. Hence, these cells can attack melanoma cells in a variety of ways: i) The α/β TCR allows for a MHC-dependent targeting of intracellularly localized tumor antigens, such as gp100, which constitute the majority of tumor antigens [
19,
70]. In general, α/β TCRs can mediate a destabilization of the tumor micro-environment by killing tumor-associated myeloid-derived stromal cells, which cross-present phagocytosed tumor antigens [
71,
72]. ii) The CAR targets surface antigens, e.g. MCSP, in an MHC-independent manner [
49]. Apart from being expressed on tumor cells, MCSP is also present on the surface of activated pericytes involved in tumor neo-angiogenesis [
53,
54].This enables a direct targeting of tumor cells coupled with an indirect targeting via the destruction of tumor-associated vasculature. iii) As described above, the endogenous receptor repertoire of γ/δ T cells [
29,
31], recognizes tumors in an MHC-independent fashion [
27,
28] and may become more and more active as the tumor cells and their micro-environment is pressurized by the transfected receptors, potentially supported by a radio-chemotherapy prior to the adoptive cell transfer [
73]. MHC-independent targeting is also beneficial in case of MHC-downregulation or antigen-loss, which represent major strategies of immune evasion [
64,
74]. iv) MHC-downregulation is further addressed by the γ/δ T-cell-inherent capacity to eliminate MHC-deficient cells, such as Daudi cells [
66]. In sum, γ/δ T cells can mount a tumor-specific attack via introduced and endogenous receptors and sustain this attack even after the tumor has lost its antigen or shut down antigen presentation.
By using zoledronic acid (ZA) instead of anti-CD3 antibody in conjunction with a GMP-compliant expansion protocol, originally designed for conventional T cells [
60], the selective expansion of γ/δ T cells to clinically applicable numbers was achieved. In theory, these receptor-transfected γ/δ T cells could also be obtained from a healthy donor and infused into an HLA-mismatched patient (in an immunosuppressed setting), since γ/δ T cells are not allo-reactive [
27,
28,
42]. After 10–11 days of expansion, a moderate percentage of propagated cells remained CD3
+/γδ
−. In contrast to the predominating γ/δ T cells, these contaminating T cells might lead to graft-vs-host disease [
75] in such an allogenic setting. Therefore, further adjustments are required to minimize the percentage of γ/δ
− cells, to employ this therapy not only in autologous settings but also in allogenic settings. Possible options involve extending the expansion course [
59] and administering repeated doses of ZA. Contrary to previous work on ZA-expanded γ/δ T cells [
59,
76], we additionally performed functional testing after depleting the γ/δ
− cells post expansion via MACS-untouched isolation to validate that the functional activity originates from γ/δ T cells. We detected a considerable reduction in cytokine production and cytolytic capacity. This would imply that a part of secreted cytokines and displayed cytotoxicity is exerted by the remaining γ/δ
− cells. This, however, collides with the fact that the lytic capacity of purified MACS CD8
+ T cells and OKT3-stimulated PBMC is similar to ZA-expanded γ/δ T cells. Thus, we think that the depletion procedure per se impaired the functionality of the receptor transfected cells by interfering with cell physiology. In addition, the use of antibody-coated magnetic beads and their alike is difficult to perform under GMP compliance. Therefore, the future focus should be on rendering depletion unnecessary by refining the process of ZA-expansion to reduce the percentage of contaminating conventional T cells.
Unlike the majority of previous studies on γ/δ T cells [
38,
39,
77], we evaluated our results in direct comparison to conventional α/β T cells, notably CD8
+ T cells, which represent the current gold standard in adoptive T-cell therapy. After receptor transfer, ZA-expanded γ/δ T cells exhibited a similar lytic capacity compared to CD8
+ T cells but a lower cytokine production. In contrast to ZA-expanded CAR-transfected γ/δ T cells, CAR transfection induced an unspecific background cytokine secretion in conventional α/β T cells, manifesting itself against the MCSP
− cell line T2.A1. This poses a potential danger of off-target toxicities and reframes the possible superiority of ZA-expanded γ/δ T cells. Generally, the fact that the lytic capacity of γ/δ T cells clearly exceeded their capacity to produce cytokines may also add to the safety of this approach, because a severe and potentially deadly [
16] side-effect of adoptive T-cell transfer is the so called cytokine release syndrome [
15], which is caused by a massive systemic release of pro-inflammatory cytokines from the transplanted cells.
The overall anti-tumor effect of adoptive T-cell therapy does not only rely on the individual functionality of the infused cells, but also on maintaining a sufficient concentration of transfected T cells in the blood [
78,
79]. Due to their absent allo-reactivity [
27,
28,
42], γ/δ T cells could be obtained from healthy donors, if used in immunodeficient recipients. Therefore, injecting higher numbers of γ/δ T cells in shorter intervals over a longer period of time may be possible and is a clear advantage over conventional T cells, which have to be obtained in an autologous fashion from the patient [
80]. The latter may be difficult anyway, since T cells of tumor-bearing patients, especially at an advanced stage, often exhibit dysfunctionalities and exhaustion phenotypes, reflected in the upregulation of inhibitory receptors, such as PD-1 [
81‐
84] and in functional impairments [
85,
86].