Chimeric Antigen Receptor Modified T cells That Target Chemokine Receptor CCR4 as a Therapeutic Modality for T-cell Malignancies

Introduction

The chemokine receptor CCR4 is a seven trans-membrane G protein-coupled cell surface receptor molecule (alternatively designated CD194) with selective expression on cells of the hematopoietic system (reviewed in ^{1, 2}). In the peripheral blood of healthy individuals, CD4⁺CD25⁺Foxp3⁺ T regulatory (Treg) cells, T_H2 and T_H17 T cells, as well as platelets display a preponderance of CCR4 expression (3–8). Although initial reports have suggested some level of promiscuity in CCR4 ligand specificity, it is now validated that the C-C chemokines CCL17 and CCL22 are the only, high affinity ligands of this receptor (9–11). The aberrant over-expression of CCL17 and CCL22 especially by stromal cells that is seen in a variety of neoplasms including, breast, ovarian, gastric, and esophageal cancers and in lymphoid malignancies such as Hodgkin's disease (12–17) is the primary driver for the extensive influx of CCR4 bearing Tregs into the tumor micro-environment of such malignancies that not only impede effective anti-tumor immune responses, but also promote tumor trafficking and metastasis as well (18). In adult T cell leukemia (ATL), an HTLV-1 associated T cell malignancy that some consider to be a CD4⁺ Treg disorder, the cell-surface expression of CCR4 on leukemic cells is near universal (19, 20). Moreover, in a majority of cutaneous T cell lymphomas [peripheral T-cell lymphoma (PTCL)/cutaneous T-cell lymphoma (CTCL), Mycosis Fungoides (MF)/Sezary syndrome (SS)], the homing of malignant cells to the skin is primarily dictated by the cell-surface-expressed CCR4 (21, 22). A significant proportion of anaplastic large cell lymphomas (ALCL), and other peripheral T-cell lymphomas also express CCR4 receptors perhaps as a consequence of the aberrant expression of the T_H2 master transcriptional regulator GATA-3 in these tumors and the presence of CCR4 in these tumors has shown to be an independent prognostic factor of poor survival in both peripheral T-cell lymphoma-not otherwise specified (PTCL-NOS) and anaplastic lymphoma kinase (ALK) negative ALCL by a multivariate analysis (23–26).

Mogamulizumab is a defucosylated humanized antibody engineered to exert enhanced antibody-dependent cytotoxicity that targets CCR4 and is approved in Japan for the treatment of relapsed or refractory adult T-cell leukemia/lymphoma (27) and accumulating evidence reveals considerable efficacy of this modality even in heavily pretreated patients (28). The observed adverse events are primarily of hematologic nature in addition to pyrexia and skin disorders that are manageable and are also reversible (28, 29). Equally important, mogamulizumab therapy also selectively depletes effector type CD4⁺Foxp3⁺ Tregs resulting in the induction of heightened anti-tumor responses in the treated patients (30, 31).

Taking a different approach, we have developed a lentiviral gene transfer system to genetically engineer autologous T cells to express a chimeric antigen receptor (CAR) that targets CCR4 using humanized variable heavy (VH) and kappa light (VL) chain moieties derived from an anti-CCR4 antibody that is different from mogamulizumab (32). Donor T cells modified ex-vivo with the CCR4 targeting CAR efficiently lysed patient-derived tumor cell lines that expressed CCR4 in an antigen-specific manner and demonstrated in vivo efficacy in a murine xenograft model of adult T cell leukemia illustrating the potential feasibility of translating this approach to treat T-cell malignancies that express CCR4.

Methods

Results

The structural configuration of our CAR molecule is depicted in Fig 1A which consists of a transmembrane module derived from the human CD8 molecule along with two intracellular signaling modules derived from 4-1BB (CD137) and CD3æ in addition to the extracellular CCR4- binding domains derived from a humanized anti-CCR4 antibody. We transduced donor derived, activated CD3⁺ T cells once and routinely achieved over 80% transduction efficiency as determined by GFP positivity and Protein L binding activity by flow cytometry (Fig.1, panels B and C). The representation of CD4⁺ and CD8⁺ subsets in the transduced cell population is depicted in panels D and E of Fig. 1, after a period of 72 hours post transduction in culture. Having confirmed the robust expression of CCR4 targeting CAR in the transduced T cells, the functional activity of these CCR4- directed CAR transduced CD3⁺T cells was then evaluated. The cell line ATL 41214 was derived from a patient with HTLV-1 associated acute adult T cell leukemia and expresses cell-surface CCR4 whereas the BJAB cell line was derived from a patient with Burkitt lymphoma and does not express CCR4 (see Fig. 2, panel A). The cell line BJAB/CCR4 is a derivative of BJAB parental cell line created by transfecting a human CCR4 expression plasmid to stably express human CCR4. Previously, we created stable derivatives of these cell lines that express a luciferase reporter gene and the availability of these luciferase-tagged cell lines enabled us to use a sensitive bioluminescence-based cytolytic assay. As shown in Fig. 2B, cytolytic activity was evident by 120 minutes of co-culture of target (ATL 41214) and effector cells that continued to increase throughout the observation period of 20 hours. Moreover, the ability of these CAR T cells to engage in multiple rounds of killing was evident by the data, especially when one considers the percentage of killing activity at low effector:target ratios that continued to increase over time thus validating the robust functionality of our CAR.

In most if not all clinical trials reported to date, the emphasis has been to limit the ex vivo CD3/CD28-based expansion phase to a period of 10–12 days to ensure an optimal state of differentiation with minimal telomere degradation and the avoidance of possible senescence or exhaustion of the cells that are being transfused. In exploring this issue further, after the initial 12-day, CD28-based expansion, we continued expansion of CAR T cells with irradiated ED41214C(-) cells in the presence of IL-2 (300 IU/ml). Surprisingly, we observed our CAR T cells continued to expand robustly even up to 8 weeks and still retained equal potency in lysing target cells (ATL 41214) bearing CCR4 (Fig. 2 C). In contrast, no cytolytic activity was observed when CCR4 CAR T cells were co-cultured with BJAB cells. However, BJAB/CCR4 cells that stably express human CCR4 manifested sensitivity to CCR4 CAR T cell mediated cytolysis thus validating the robust functionality and specificity of our CCR4 CAR (Fig. 2D).

To assess the broader utility of the CCR4-directed CAR approach, we used a panel of patient derived malignant cell lines representing a spectrum of lymphoid malignancies to assess their sensitivity to CCR4 CAR transduced donor derived CD3⁺ T cells (Fig. 3). The cell lines ED40515(+), ATL55T(+), KOB, LM-Y1, KK1, ATL43T(+) and ED41214(+) represent chronic/smoldering adult T cell leukemia and are IL-2 dependent for their growth (panel A), while ED40515(-), ST1, ATL43Tb(-), Su9T01, ATN1, ED41214C(-) and MT1 are from acute adult T cell leukemia patients and are not dependent on IL-2 for growth (panel B). The cell lines HH, HuT78, MJ, HuT102 represent cutaneous T cell lymphomas (panel C). The cell lines Mac-1, Mac-2A, and Mac2B cell lines were derived from anaplastic large cell lymphoma without chromosome t(2;5) translocations and are thus ALK (anaplastic lymphoma kinase) negative whereas JB-6, Karpas299, SUDHL-1, SR-786, SUP-M2 and DEL have been derived from ALK-positive ALCL (panel D). Hodgkin cell lines L428, HDLM-2, KM-H2, and L1236 were also included in the analysis (panel E), as well as SUDHL-4 derived from a diffuse large B cell lymphoma.

Almost all cell lines derived from T-cell malignancies were susceptible to lysis by CCR4 CAR T cells although the level of sensitivity varied somewhat among the cell lines tested. In the panel of ALCL cell lines tested, there was clear separation of ALK negative cell lines (Mac-1, Mac-2A, and Mac-2B) as a group being more sensitive than the ALK positive cell lines (JB-6, Karpas299, SUDHL-1, SR-786, SUP-M2 and DEL). With HDL cell lines, HDLM-2 which displays T cell antigens (CD25, CD4) and L428 devoid of any T cell antigens both displayed significant sensitivity while the other HDL cell lines were refractory as was SUDHL-4, a DLBCL-derived cell line. While all CTCL and ATL cell lines were sensitive, the exception was ED40515(-) that was refractory, prompting us to look at the cell surface CCR4 levels in the tested cell lines as a way of explaining the differences in sensitivity. We determined the surface expression of CCR4 on these cell lines by flow cytometry and found most of them to have detectable levels of CCR4 on their cell surface (see Supplemental Figure S1). The cell surface expression of CCR4 as detected by flow cytometry did not appear to be a very sensitive measure for quantitative analysis although the trend of cell surface CCR4 levels generally paralleled their sensitivity to CCR4 CAR T cell mediated cytolysis. To exclude the possibility of sub-optimal surface staining of CCR4, we tested a range of anti-CCR4 antibody concentrations (see Supplemental Figure S2) and the most optimal antibody concentration was used in the experiment shown in Supplemental Fig. S1. It is noteworthy that certain cell lines such as SUDHL-1, Karpas299, and DEL, all of which are derived from ALK+ ALCL, despite lacking measurable cell surface CCR4 expression by flow cytometry, were not totally refractory to CCR4 CAR T cells. In contrast, concordant with the high sensitivity to CCR4 CAR T cell mediated cytolysis, cell lines derived from ALK negative ALCL (Mac-1, Mac2A and Mac-2B) displayed abundant cell surface CCR4 as shown in Supplemental Fig. S1, panel B. We also measured the CCR4 mRNA levels in these cell lines by quantitative PCR (TaqMan). CCR4 mRNA levels in some cell lines were extremely high while in others, the levels were relatively low (Supplemental Table 1) and we observed a better correlation between lytic susceptibility and CCR4 mRNA levels than with cell surface CCR4 expression by flow cytometry. For example, SUDHL-4 cell line that was devoid of CCR4 mRNA were also refractory to CCR4 CAR T cell mediated cytolysis while ALK+ ALCL cell lines such as DEL, Karpas299 and SUDHL-1 with low levels of CCR4 mRNA were only modestly responsive to CCR4 CAR T cells.

The mechanisms underlying CAR T cell engagement with their targets and ensuing cytolysis of target cells remain yet to be fully elucidated. Both CD4⁺ and CD8⁺ CAR T cells display cytolytic activities and are likely to utilize multiple mechanisms including perforin/granzyme exocytosis, death receptor signaling involving Fas/FasL and/or TNF/TNF-R pathways. Any genetic aberrations in these pathways in neoplastic target cells could impact their susceptibility. Furthermore, threshold antigen densities that are optimal for cell killing are currently unknown (reviewed in refs ^{37, 38}). The expression of co-inhibitory ligands on target cells could also impact CAR T cell cytolytic activities. When we assessed the expression of PDL-1 in our panel of ATL cell lines only ST1 displayed appreciable amount of PDL-1. In contrast, as has been reported previously (39), all ALCL cell lines displayed abundant cell surface expression of PDL-1 including the ALK negative Mac-1, Mac-2A and Mac-2B cell lines that were highly susceptible to CCR4 CAR T cell mediated cytolysis (see Fig. 4, panels A, B and C). Therefore, the expression of PDL-1 by itself may not explain the discrepancies between CCR4 expression and the effectiveness of cytolysis observed in Fig. 3 and Supplemental Fig. S1.

To further delineate the functionality of the CCR4 CAR, we measured the strength of the proliferative signal it transduces by co-culture of CAR T cells with irradiated CCR4-bearing target cells ATL55T(+) or CCR4 negative SUDHL-1 cells. We found, as shown in Fig.5, panel D, an increase over 3-fold in H³ thymidine incorporation within a period of 72 hours with ATL55T(+) cells in comparison to SUDHL-4 cells (mean counts = 129100±1586 SD versus 42610±1786 SD, p=0.000). Similar increases in proliferative responses with CCR4 CAR T cells were seen in co-culture experiments with ST1, ED41214(-) and BJAB/CCR4 cells but not with the parental BJAB cells (data not shown).

In a parallel experiment, we also measured the cytokines and chemokines that were elicited by co-culture of target cells with CCR4 CAR T cells for a period of 24 hours (Supplemental Table 2). Despite the induction of several inflammatory cytokines such as TNF-alpha, TNF-beta, and IFN-gamma, a cytokine that is a principal driver of cytokine release syndrome often associated with CAR T cell therapy, IL-6 was not induced implying that the source of IL-6 seen during cytokine release syndrome is not the CAR T cells. The basal levels of MIP-1á and MIP-1â expression in CCR4 CAR T cells were quite high (19 ng/ml and 31 ng/ml respectively) and these levels were further increased upon co-culture with target cells somewhat (36% and 25% respectively). Furthermore, we failed to detect any meaningful induction of the following cytokines namely, IL-12/23p40, IL-12p70, IL-15, IL-16, IL-1á, IL-1â, IL-7, VEGF, IL-10, Eotaxin, Eotaxin-3, IP-10, MCP-1, MCP-4, CCL-17 and IL-17.

Finally, we assessed in vivo efficacy of our CCR4 CAR T cells in a murine xenograft model of adult T cell leukemia. Mice that were transfused with 10 million CCR4 CAR T cells intraperitoneally 3 days after inoculating 10 million patient derived ATL cells (luciferase tagged ATL41214 cells) eradicated tumor cells and remained tumor free (panel F of Fig. 4) at the conclusion of the study, while all of the control mice inoculated with lentivector expressing GFP transduced T cells displayed continuing tumor growth 12 weeks after transplanting ATL cells (Fig. 4, panel E) where leukemic tumor growth was quantitated using bioluminescence imaging flux (photons/second) as a surrogate measurement of leukemic tumor burden (panel G).

Discussion

T-cell malignancies represent a heterogeneous group of lymphoid neoplasms often with a poor clinical outcome (reviewed in ⁴⁰). Their relative lack of well-defined genetic aberrations perhaps with the exception of ALK(+) ALCL with t(2;5) translocations, along with the lack of definitive diagnostic markers have been a major impediment in unraveling the pathogenesis as well as devising novel efficacious therapeutic strategies for most T-cell malignancies (40). With gene expression profiling, it has been recently established that nearly half of PTCL-NOS, a diverse group that accounts for about 50% of PTCL, lack any classifiable markers but over-express the T_H2 master transcriptional regulator GATA-3 and often carry a poor prognosis (24, 25). As in normal T_H2 CD4⁺ T lymphocytes, GATA-3 drives the transcription of CCR4 in these PTCL-NOS and consequently they over-express cell surface CCR4 (41). Moreover, recent assessments of mutational landscapes in CTCL, Sezary syndrome and ATL have revealed genetic aberrations of the chromosome 10p to be a shared feature in these T cell malignancies with the inactivation of ZEB1 resulting in the enhanced expression of GATA3 (42–45). In addition to GATA3 driven transcription of CCR4 gene, frequent mutations in the C-terminus of CCR4 polypeptide in PTCL-NOS, CTCL and ATL impair the ligand induced receptor internalization of CCR4 thereby contributing to its cell surface over-expression in these malignant cells as well (33, 43, 46, 47). The oncogenic transcription factor FRA-2 has also been implicated in the over-expression of CCR4 in both CTCL and ATL (48, 49). As almost all CTCL and ATL also overexpress CCR4, as well as a large percentage of ALCL especially the more aggressive ALK(-) subtype, the overexpression of surface CCR4 appears to be a target that can be exploited to develop therapeutic strategies with a broader applicability in the treatment of T-cell malignancies. In addition, because Treg cells express CCR4 in abundance (3), the CCR4-directed therapy is likely to eliminate Tregs that in the tumor microenvironment constitute a formidable barrier against tumor-eradicating host immune responses.

Therefore, our CCR4 CAR represents a novel adoptive cellular therapy approach for T cell malignancies in which ex-vivo engineered T cells can be utilized to ablate a spectrum of neoplastic T-cell diseases in patients. Although both accumulating clinical experience with mogamulizumab and the documented unremarkable phenotypic features of CCR4 gene disrupted mice are reassuring (28, 29, 50), it is of paramount importance that we have a thorough understanding of what the toxicity profile of a CCR4-targeted CAR T cell therapy is likely to be, before we initiate patient treatment studies. Unlike CD19 directed CAR therapies that result in long-term or permanent aplasia of B cells (51), the CCR4 directed CAR therapy is unlikely to cause major perturbation of T cell subsets in the treated individuals. As seen in data from Fig. 1, panels D and E, it is clear that both CD4⁺ and CD8⁺ T cells can express our CAR and remain viable long term without undergoing self-ablation or fratricidal death in culture. However, it is important to note that with multiple donor-derived T cell preparations (n=4), the post-transduction CD4:CD8 ratio in culture was reproducibly reversed revealing that a certain proportion of CD4⁺ T cells undergo ablation due to their cell surface CCR4 expression in the presence of CCR4 CAR T cells. Another serious concern that arises is whether infusion of ex vivo modified CCR4 targeted CAR T cells could detrimentally impact platelets leading to a profound and persistent thrombocytopenia (7, 8). In a Phase II trial, Ishida et al. (52), reported hematologic toxicities that included lymphopenia (96%), neutropenia (52% but with 19% being Grade 3 without any febrile episodes), and thrombocytopenia (52% but with 19% being Grade 3–4 with no associated bleeding). Moreover, during the initial 4-month post marketing surveillance in ATL patients, dermal toxicities including four cases of Steven-Johnson syndrome/toxic epidermal necrolysis and one fatality have been observed (53). However, as mentioned earlier, in the trials conducted by Duvic et al (29) and Ogura et al (28), the hematologic adverse effects were minimal and/or manageable. Collectively, despite these concerns the tantalizing promise that CCR4 CAR could be exploited to effectively treat T-cell malignancies, have convinced us that a non-human primate study is essential prior to undertaking clinical studies in patients. The extracellular domains of rhesus macaque CCR4 is 100% identical to that of human CCR4 (54) and the antibody from which our CCR4 CAR was derived reacts with non-human primate CCR4 (55) thereby allowing us to conduct an evaluative safety in rhesus macaques with ex-vivo engineering of monkey CD3⁺ T cells with our CCR4 CAR and reinfusion of the autologous modified T cells back into the animals to uncover any potential on target but off tumor toxicities that might arise in a clinical trial of CCR4-positive T cell malignancies.

In conclusion, we are cautiously optimistic that the adoptive cellular therapeutic modality we have developed using a chemokine receptor CCR4 directed CAR, can be exploited to effectively treat a spectrum of T-cell malignancies for which the existing treatment regimens are largely inadequate and do not confer long-term survival.

Supplementary Material