TCR vector construction and analysis of constructs
We recently reported that 46% of patients treated by adoptive immunotherapy after nonmyeloablative lymphodepleting chemotherapy exhibited objective tumor regression (Dudley et al., 2002a
). Patient 9 in that trial exhibited a near-complete regression of tumor associated with the oligoclonal expansion of an administered Vβ12+, CD8+
MART-reactive clone. Several different T-cell clones present in the infused TIL were MART1-reactive. The Vβ12+ T cells accounted for 14% of infused cells and 55% of the total PBL 7 days after infusion. Four months after cell transfer (d124) 51% of the PBL were MART1-reactive, and this pool almost completely comprised the Vβ12 expressing clone.
TCR Vβ12 containing clone M1F12 is a highly reactive T-cell clone derived from patient 9 via limiting dilution of post-treatment PBL and was confirmed by DNA sequence, to be the clone infused into the patient (data not shown). The α and the β chain TCR cDNAs were isolated from clone M1F12 and were used to construct four different gene transfer vectors. All constructs used the retroviral MSCV LTR to drive expression of either α or β chains with the second chain either promoted by an internal PGK promoter or the two genes were linked by an IRES. The combinations were thus AIB or BIA (α IRES β or β IRES α) and BPA or APB (β PGK α or α PGK β ) (). The plasmid DNA vectors were transfected into PG13 packaging cells lines, which were tested by intracellular staining with anti-Vβ12 antibody (). Constructs AIB, BIA, and BPA yielded 23% to 53% of cells expressing Vβ12 staining while construct APB yielded only 6% positive cells and thus APB was not analyzed further
FIG. 1 TCR expressing retroviral vectors. A: Diagrams of four biscistonic retroviral vectors used to transfer and express the TCR gene from cytotoxic T lymphocyte (CTL) clone M1F2. The first gene in each vector was driven by the vector long-terminal repeat (LTR), (more ...)
We first tested the activity of these vector producer cell populations by transduction of the human T-cell leukemia cell line, Sup T1 (this line contains chromosomal translocations involving both α and β TCR genes that prevents surface expression of the endogenous TCR complex including CD3). Transduced cells were thus assayed for cell surface expression of the CD3 protein to assess the expression of the TCR α and β chains. After transduction with constructs AIB and BIA using supernatants from the PG13 producer populations, 44% and 21% of Sup T1 cells expressed CD3 () suggesting that these cloned anti-MART-1 α and β chain genes can participate in the formation of a functional TCR complex.
FIG. 2 Transduction of human SupT1 T cell line. A: Fluorescence-activated cell sorter (FACS) histogram of CD3+ cells after transduction of Sup T1 cells by AIB or BIA retroviral vectors using supernatant form producer cell populations. The percentage positive (more ...)
To isolate high-titer vector producing cells, PG13 vector producer cell clones were obtained by limiting dilution and were retested by transduction of SupT1. Sup T1 cells transduced with supernatant from PG13 clones ranged from 15% to 89% MART-1 tetramer positive. Representative data from two of the most active clones (AIB18 and AIB54) is shown in . Vector 18 transduced cells exhibited 69% tetramer staining and AIB 54 engineered cells had 89% tetramer staining. Producer cell clones (AIB 18, AIB 54, BIA 71, and BPA 34) with both high titer and tetramer staining were chosen for further study (data not shown).
Transduction of PBL
In order to engineer PBL with the anti-MART-1 TCR vector, we used a modification of a procedure (Hanenberg et al., 1996
) that involved precoating tissue culture plates with Retronectin (Takara Bio, Inc., Shiga, Japan), a recombinant fibronectin molecule, which colocated the retrovirus with the target cell. This method allowed for high efficiency retroviral vector mediated gene transfer and optimized PBL culture conditions since the vector supernatant is removed prior to adding the target cells (PBL).
Transduced T cells were stained for Vβ12 and MART-1 tetramer 48 hr after the final transduction and analyzed via FACS. The percentage of Vβ12 positive cells in the transduced cell populations varied from 43%-86% (background staining for Vβ12 was from 1-3%). Examples of two transductions, AIB18 and AIB54 are shown in . These transductants, which contained 86% and 75% Vβ12-positive cells also exhibit 37% and 29% of cells, respectively, stained with MART-1 tetramer (). Both CD4and CD8-positive cells in the engineered population stained for Vβ12 (data not shown).
FIG. 3 Transduction of primary human lymphocytes. A: Fluorescence-activated cell sorter (FACS) histogram of peripheral blood lymphocytes (PBL) transduced with either AIB 18 or AIB 54 retroviral vectors or untransduced and stained for Vβ12 48 hr posttransduction. (more ...)
Functional tests of the TCR transductants
To determine if TCR-transduced PBL could mediate the release of effector cytokines when exposed to appropriate antigen, coculture experiments were performed. The coculture consisted of transductants and T2 cells pulsed with various peptide antigens (Flu, gp100 and MART-1). PBL were either not transduced (NV), transduced with control vectors encoding GFP (MSGIN), or gp100 specific TCR (APB9) (Morgan et al., 2003
), and various MART-1 TCR retroviral vectors (AIB 18, BPA 34, BIA 71). Positive controls for these assays were the MART-1-reactive T lymphocyte clone JB2F4 or PBL from a patient with metastatic melanoma derived by multiple in vitro
stimulations with MART-1 peptide (PBL-MART).
In these assays, the transduced PBL specifically secreted cytokines when exposed to the relevant peptide stimulus (). Transduced PBL secreted between 48,375 and 75,546 pg/ml of interferon (IFN)- γ compared to control transduced cells that produced 409 pg/ml. The variation in IFN- γ secretion was dependent on which TCR vector clone was used for transduction. For GM-CSF, MART-1 TCR-transduced PBL secreted 5100 pg/ml compared to control transduced cells that secreted 786 pg/ml. The transduced PBL when cocultured with relevant peptide made low but detectable amounts of IL-2 compared to control PBL, which made no detectable IL-2. The PBL modified with the anti-MART-1 TCR vectors (AIB 18 and AIB 54) secreted 2552 and 1941 pg/ml of TNF-α, respectively, compared to 9 pg/ml for the mock-transduced population. One measure of the relative reactivity of a particular TCR is the ability to react to cells pulsed with limiting dilutions of peptides. The antiMART-1 TCR gene-modified PBL assayed in this experiment recognized T2 cells pulsed with as little as 0.1 ng/ml MART-1 peptide, comparable to the CTL clone JB3F4.
Recognition of melanoma cell lines by transduced lymphocytes
To assess the recognition of melanoma cells, the modified lymphocytes (transduced with AIB 18, BIA 71, BPA34 and MSGIN) were cocultured with two HLA-A2 cell lines (526, 624) and two non-HLA-A2 cell lines (888, 938). Specific release of IFN-γ and GM-CSF, was seen when the gene modified PBL were cocultured with HLA-A2-positive but not the HLA-A2-negative melanoma cell lines (). IFNγ secretion by transduced PBL ranged from 66-318 pg/ml when exposed to MEL 526 compared to control secretion of only 12 pg/ml. For MEL 624, transduced PBL secreted between 77-235 pg/ml while control PBL only secreted 1 pg/ml. Combinations of PBL with non-HLA-A2 melanoma lines had insignificant IFNγ . Similarly for GM-CSF, anti-MART-1 TCR-transduced PBL secreted between 195-449 pg/ml when exposed to A2+ melanoma compared to less than 94 pg/ml for A2- melanoma lines.
FIG. 4 Recognition of melanoma cell lines. Shown are cytokine production values (interferon [IFN]-γ top graph, granulocyte macrophage-colony stimulating factor [GM-CSF] bottom graph) of transduced peripheral blood lymphocytes (PBL) after coculture with (more ...)
To determine the functional reactivity of transduced T-cells, we performed a CD107a mobilization assay (). In this experiment PBL were transduced with the AIB 18 vector and then the population cocultured with HLA-A2-negative/MART-1-positive melanoma cell line 888 or HLA-A2+/MART-1 + melanoma cell line 624. After coculture, transduced T cells (i.e., Vβ12-positive cells) were analyzed for CD3 and CD107a antigen expression. Data in demonstrate that while few (1.6%) of transduced cells mobilized CD107a upon coculture with melanoma line 888, 29% of the TCR vector-transduced cells became positive for CD107a expression upon coculture with melanoma line 624.
FIG. 5 CD107a mobilization by TCR transduced peripheral blood lymphocytes (PBL). Fluoresence-activated cell sorter (FACS) analysis for degranulation marker protein CD107a was determined after coculture of AIB 18 vector-transduced PBL with no melanoma cells (No (more ...)
As an additional test of T-cell effector cell function, we next determined the ability of the transduced PBL to lyse melanoma tumor cell lines. TCR-transduced PBL were tested against melanoma cell lines in a 4-hr 51Cr release assay () using anti-MART-1 CTL clone JB2F4 as positive control. The TCR-engineered cells (using vector AIB 18) readily lysed HLA-A2-positive melanoma cell lines (526, 624) but not the HLA-A2-negative 938 melanoma, while mock transduced PBL (NV) had little to no lytic capability.
FIG. 6 Lysis of melanoma cells by transduced peripheral blood lymphocytes (PBL). 51Cr release assay was performed on PBL either transduced with AIB18 (anti-MART-1 TCR) or a no vector (NV) control population. After 4-hr incubation with 51Cr-labeled melanoma cell (more ...)
Conversion of non-reactive TIL cultures to MART-1 reactivity
To be effective in adoptive immunotherapy, TIL cultures must be reactive to shared tumor antigens or the patient's autologous tumor. We next determined whether nonreactive TIL could be converted, via transduction with the anti-MART-1 TCR, to antigen-reactive TIL. Approximately 17-24% of TIL after transduction were Vβ12-positive compared to less than 1% for the mock-transduced TIL controls (data not shown). Patient's TIL was transduced with either the AIB18 anti-MART1 TCR vector or the APB9 anti-gp100 TCR vector and then analyzed for CD8 expression and appropriate tetramer binding. The cultures were almost entirely CD8 + (96%) and each had 7-8% tetramer staining versus less than 1% for controls.
TIL transduced with the AIB 18 anti-MART-1 TCR vector, the anti-gp100 TCR vector, no vector, or GFP exhibited specific release of IFNγ and GM-CSF as shown in . The anti-MART-1 TCR-transduced cells secreted 14,695 pg/mL of IFNγ when the TIL were exposed to T2 cells pulsed with the MART-1 peptide (versus <500 pg/ml in control-engineered cultures). Similarly, the transduced TIL specifically secreted large amounts of the two other cytokines GM-CSF (79,969 pg/ml) and IFNα (1568 pg/ml).
The TCR-engineered TIL also recognized MART-1 antigens processed on HLA-A2-positive melanoma cells but not HLAA2-negative lines assessed both by cytokine release and in four-hour 51Cr release assays ( and ). The MART-1 TCR vector-transduced cells produced 1674 and 2139 pg/ml of IFNγ in the presence of A2+ melanoma cell lines (526 and 624) compared to 18 and 14 pg/ml in the presence of A2- cell lines. Similar results were observed for GM-CSF (). The 51Cr release assay demonstrated lysis of tumor by the MART1 TCR transduced TIL compared to the near-complete lack of lysis in the mock transduced TIL in HLA-A2-positive melanoma cell lines. The anti-MART-1 TCR transduced TIL lysed between 50-70% of the targets at the highest effector to target ratio. In addition to the lysis of the two established melanoma cell lines (526 and 624), the transduced TIL also specifically recognized an HLA-A2-positive primary fresh melanoma digest ().
TRANSDUCED TIL MELANOMA REACTIVITY
FIG. 7 Lysis of melanoma cells by transduced tumor-infiltrating lymphocytes (TIL) TIL transduced with the anti-MART-1 AIB vector or mock transduced (NV) were tested for their ability to lyse 51Cr-labeled melanoma cells. After a 4-hr incubation with 51Cr-labeled (more ...)
Antigen-specific proliferation of TCR-engineered lymphocytes
Last, we sought to examine whether these engineered cells would proliferate in vitro when stimulated by an appropriate antigen. Mock-transduced PBL and anti-MART-1 TCR-transduced PBL were labeled with CFSE dye and then were cocultured with MEL 526 (A2+) or MEL 888 (A2-) in concentrations of IL-2 ranging from 0-10 IU/ml. Four days after stimulation, the proliferation of CD3 + lymphocytes was determined by FACS analysis (), where dilution of the CFSE peak was indicative of cell proliferation. FACS analysis of melanoma/lymphocyte cocultures in the presence of no exogenous IL-2, demonstrated 25% of the cells had undergone cell division versus 2-5% of control cultures. When minimal amounts of exogenous IL-2 were added (1-10 IU/ml), up to 55% of the lymphocytes in the MEL 526 coculture had demonstrable proliferation, compared to 7-13% of control cultures.
FIG. 8 Proliferation of TCR-transduced peripheral blood lymphocytes (PBL) in melanoma cocultures. Shown are resultant fluorescence-activated cell sorter (FACS) histograms of carboxy-fluorescein diacetate, succinimidyl ester (CFSE)-labeled PBL cocultured with (more ...)
There are potential safety concerns regarding the infusion of large numbers of MART-1-reactiv, but many of these theoretical concerns have, by and large, been addressed in previous clinical studies. The gene used in this report is a naturally occurring anti-MART-1 TCR derived from the TIL of patient 9 in the report by Dudley et al. (2002a)
. This exact gene, in the context of TIL, was administered to this patient (1.2 × 1010
cells) along with high-dose IL-2 after nonmyleoablative but lymphocyte-depleting chemotherapy. No toxicities associated with the infusions of this highly reactive T cell were observed in this patient (other than vitiligo). In addition, we have treated 15 patients using infusions of large numbers of antigp100 T-cell clones in the setting of nonmyeloablative chemotherapy with no side effects (Dudley et al., 2002b
). While autoimmunity (e.g., vitiligo or uveitis) may be a possible consequence of anti-MART-1-reactive cell infusions, this response may correlate with antitumor responses (Dudley et al., 2002a
; Phan et al., 2003
). If severe autoimmunity is observed, it can be controlled by chemotherapy or steroid treatment.
A second potential safety concern is the induction of novel or diverse functional activity and the possibility of the selective expansion of novel reactivity driven by the tumor antigen and its potential consequences. The expansion of tumor reactive cells is a desirable outcome following the infusion of antigen-reactive T cells and any associated autoimmunity (or other novel reactivity) can be managed if it becomes medically necessary. We have administered over 3 × 1011
TIL with widely heterogeneous reactivity including CD4, CD8, and natural killer (NK) cells without difficulty. Finally, in regard to the malignant potential of these cells, we do not believe this is a significant risk for this patient population (i.e., patients with malignant melanoma). While the risk of insertional mutagenesis is a known possibility using retroviral vectors (Hacein-Bey-Abina et al., 2003
) this has only been observed in the setting of infants treated for X-linked severe combined immunodeficiency (X-SCID) using retroviral vector-mediated gene transfer in to CD34+
bone marrow cells. In the case of retroviral vector-mediated gene transfer into mature T cells, there has been no evidence of long-term toxicities associated with these procedures since the first National Cancer Institute sponsored gene transfer study in 1989. Although continued follow-up of all gene therapy patients will be required, data suggest that the introduction of retroviral vector transduced mature T cells is an acceptable risk for appropriately informed adult patients with metastatic cancer.
As the source of the genes encoding the anti-TAA TCR, we selected T-cell clone M1F12, which possessed a highly reactive MART-1 TCR associated with an objective antimelanoma response in one of our adoptive immunotherapy patients (Dudley et al., 2002a
). At the height of this patient's antitumor response, this particular Vβ12 clone comprised 50% of the patient's peripheral lymphocytes and could be detected in tumor biopsy samples. We chose as retroviral vector backbone to express these genes, the retroviral vector MSGV1. The design of MSGV1 incorporates two key elements that were intended to optimize TCR gene expression. First, the LTR promoter derives from the MSCV retrovirus that we had previously shown to promote high levels of gene expression in a variety of cell types, including primary hematopoietic cells (Cheng et al., 1998
). Next, the inclusion of the naturally occurring MLV envelope gene splicing acceptor site, as exemplified in the MFG-class of retroviral vectors, was used to facilitate mRNA splicing and translation via a Kozak consensus translation initiation signal (Onodera et al., 1998
). The vector BPA had the α and β chains expressed from independent promoters while the AIB and BIA vectors had the expression of the chains coupled by an IRES. In multiple FACS analysis of transduced Sup T1, PBL, and TIL, staining with MART-1 tetramer was comparable, suggesting that any of these vector designs could mediate high levels of gene expression and transfer a biologically active TCR gene into transduced human cells.
The results reported here demonstrate significantly enhanced antimelanoma activity from anti-MART-1 TCR gene-transduced T cells than our previous observations with anti-MART1 TCR gene transfer (Clay et al., 1999
). The main difference in these two reports were different retroviral vector backbone designs and the specific anti-MART-1 TCR used. While we believe that our new vector designs contributed to the enhanced antitumor activity in our current report, the main determinant of this superior activity is likely the specific TCR used herein. While many factors contribute to the overall reactivity of the TCR gene-transduced T lymphocytes, the specific TCR used has been demonstrated to be the major contributor to antitumor activity (Roszkowski et al., 2003
; Rubinstein et al., 2003
; Schaft et al., 2003
). Our previous anti-MART-1 TCR was obtained from a tumor-reactive CTL derived in vitro
from a TIL culture by limiting dilution (Clay et al., 1999
). While the in vivo
activity of the previous anti-MART-1 clone is unknown, the current anti-MART-1 TCR was derived directly from a CTL clone associated with a pronounced antitumor response derived from in vivo
samples obtained from a patient effectively treated with adoptive cell therapy (Dudley et al., 2002a
The transfer of these highly reactive TCR genes into primary lymphocytes had efficiencies always greater than 30% and occasionally as high as 75-80% via Vβ12 staining (). The anti-MART-1 engineered primary lymphocytes secreted large amounts of proinflammatory cytokines, IFN-α , GM-CSF, and IFNα in an antigen-specific manner. These transduced PBL exhibited activity comparable to MART-1-specific PBL () and while they do not produce the same magnitude of cytokine production as do control CTL clones (Tables and ), their affinity for peptide observed in limiting dilutions experiments () appears to be equivalent to CTL clones.
The engineered cells were effective at detecting antigen and releasing effector cytokines in response to MART-1 in the context of melanoma cell lines ( and ). In addition, the anti-MART-1 TCR-engineered PBL were able to mobilize CD107a and effectively lyse melanoma cell lines in an HLAA2-restricted manner (Figs. and ). Together these data suggest that TCR gene-transduced T cells can be an alternative to highly active CTL clones in adoptive immunotherapy. The engineering of polyclonal T-lymphocyte populations such as PBL or TIL has a practical advantage over CTL clones, in that, these populations can generally be expanded more than 1000-fold in vitro
while it is difficult to expand CTL clones by greater than 200-fold in vitro
. In addition, while infusion of large numbers of CTL have not been associated with tumor regression (Dudley et al., 2001
), polyclonal TIL populations can mediate the regression of large established tumors in melanoma patients (Dudley et al., 2002a
Given the random assortment of TCR α and β chains in the transduced PBL, (likely reflected by the lower percentage of tetramer-positive cells compared to Vβ12-positive cells, ) it was possible that insufficient anti-MART-1 activity could have been observed. Our results demonstrate that despite lower percentages of TCR-transduced cells in the engineered populations, and a Gaussian distribution of TCR Vβ expression (), activity comparable to that of highly active CTL clones was observed (Tables and , and ). The ability of our TCR gene-transduced polyclonal T cells to exhibit antitumor reactivity is likely a function of starting with a highly reactive TCR gene and the fact that only a few TCR-antigen/major histocompatibility contacts are required for full effector cell function (Padovan et al., 1993
; Labrecque et al., 2001
). Using a native TCR with high reactivity (such as the TCR from patient 9's CTL clone M1F12) may be the critical determinant for successful transfer of antitumor properties to nontumor-reactive lymphocytes, because the TCR is the main determinant of avidity, and high avidity is correlated with in vivo
antitumor activity (Zeh et al., 1999
Adoptive transfer of tumor-reactive TIL has been shown to mediate cancer regression in vivo
in several clinical trials (Rosenberg et al., 1988
; Papadopoulos et al., 1994
; Mackinnon et al., 1995
; Walter et al., 1995
; Dudley et al., 2002a
). Sometimes it has not been possible to obtain TIL from patients with metastatic melanoma, either because of the inability to resect tumor, or even with sufficient tumor available, some cultures do not yield tumor-reactive TIL. In our hands, approximately 39% of TIL cultures are not reactive to either shared melanoma antigens or autologous tumor (Dudley et al., 2003
). These nonreactive TIL may retain the cell surface molecules for homing but lack the anti-TAA activity, which is necessary for tumor lysis. In our experiments, we studied TIL that were nonreactive. The efficiency of transduction of TIL was not as high as was shown for PBL, which we speculate was because of the slower replication of the TIL. Effector cytokines IFNγ and GM-CSF were secreted in large quantities compared to untransduced TIL when exposed to peptide-pulsed targets (). Also, effector cytokines could be generated in a specific manner when TCR-engineered TIL was exposed to HLAA2-positive melanoma cell lines (). Transduced TIL had improved cytolytic activity versus HLA-A2-positive melanoma cell lines, and primary cultures of A2-positive tumor were lysed effectively by TCR-transduced but not untransduced TIL (). This is the first time that lysis of a primary human tumor has been reported by an anti-MART-1 TCR gene-transduced human T lymphocytes. Clinically, it may be possible to apply this transfer of TCR genes to TIL that lack tumor reactivity, potentially increasing the number of patients with metastatic melanoma who can be treated by immunotherapy.
Last, in vitro
transduced PBL showed the ability to proliferate in response to MART-1 in the context of a melanoma cell line (). Growth occurred only when the antigen was present in the context of the HLA-A2 molecule. Also, growth occurred despite suboptimal levels of IL-2 in the culture media. This was consistent with the ability of these T cells to synthesize IL-2 after coculture with MART-1-expressing targets (). The ability of TCR gene-transduced human T cells to proliferate in response to tumor antigens has not been previously reported. In a mouse model of this approach (Kessels et al., 2001
), antigen-driven expansion of TCR gene-modified T-lymphocytes was essential to the antitumor response. In future clinical application of this technology, expansion of transduced cells following antigen exposure in vivo
will be critical as proliferation of adoptively transferred TIL appears to be associated with the ability of TIL to mediate tumor regression (Dudley et al., 2002a