Robust Production of Merkel Cell Polyomavirus Oncogene Specific T Cells From Healthy Donors for Adoptive Transfer

Virus positive Merkel cell carcinoma (VP-MCC) is an aggressive but immunogenic skin malignancy driven by Merkel cell polyomavirus (MCPyV) T antigen (TAg). Since adoptive T cell transfer (ACT) can be effective against virus-driven malignancies, we set out to develop a methodology for generating MCPyV TAg specific T cells. MCPyV is a common, asymptomatic infection and virus-exposed healthy donors represent a potential source of MCPyV TAg specific T cells for ACT. Virus specific T cells were generated using monocyte-derived dendritic cells (moDCs) pulsed with MCPyV TAg peptide libraries and co-cultured with autologous T cells in supplemented with pro-inflammatory and homeostatic cytokines for 14 days. Specific reactivity was observed predominantly within the CD4+ T cell compartment in the cultures generated from 21/46 random healthy donors. Notably, responses were more often seen in donors aged 50 years and older. TAg specific CD4+ T cells specifically secreted Th1 cytokines and upregulated CD137 upon challenge with MCPyV TAg peptide libraries and autologous transduced antigen presenting cells. Expanded T cells from healthy donors recognized epitopes of both TAg splice variants found in VP-MCC tumors, and minimally expressed exhaustion markers. Our data show that MCPyV specific T cells can be expanded from healthy donors using methods appropriate for the manufacture of clinical grade ACT products.


INTRODUCTION
Virus positive Merkel cell carcinoma (VP-MCC) is an aggressive neuroendocrine skin cancer mediated by clonally integrated Merkel cell polyomavirus (MCPyV) (1). Immune checkpoint inhibitors (ICIs) produce frequent and durable responses in patients with advanced VP-MCC (2)(3)(4), highlighting a critical role for T cell immunity in tumor clearance. Therefore, enhancing anti-tumor immunity with adoptive T cell transfer (ACT) of tumor-specific T cells could improve responses to ICIs and offer hope to patients where ICIs are contraindicated or who are refractory to ICIs alone (NCT02584829, NCT03747484) (5,6). Current clinical trials that use VP-MCC tumor specific T cells are human leukocyte antigen (HLA)-restricted and are only applicable to a subset of VP-MCC patients (NCT02584829; NCT03935893; NCT03747484). By using overlapping peptide libraries and enhanced culture conditions, we found that healthy donor peripheral blood can be used to reliably manufacture VP-MCC specific T cells suitable for allogeneic ACT regardless of HLA genotype.
MCPyV is a small, non-enveloped double-stranded DNA virus that typically causes an asymptomatic infection in most individuals during infancy (7). The pathogenesis of VP-MCC requires the clonal integration of MCPyV into the host genome in addition to a large truncating mutation of the T antigen (TAg) gene. MCPyV TAg is a major oncogene of VP-MCC and is expressed homogenously in all VP-MCC tumor cells (8). In VP-MCC, TAg is expressed as two alternatively spliced products: small T antigen (ST) and truncated large T antigen (LTT; Figure 1A) (9,10). Both proteins are required for the proliferation and survival of VP-MCC cells (8). Addiction of VP-MCC cells to MCPyV oncoproteins makes TAg an attractive target for immunotherapy. Moreover, the somatic mutational burden of VP-MCC is extremely low, suggesting the tumors produce few additional neoantigens (11).
T cell immunity directed against MCPyV TAg is associated with longer survival and fewer metastases in VP-MCC patients (12). ACT of autologous CD8 + MHC class I restricted T cells specific for a HLA-A*24:02-restricted MCPyV LT-Ag 92-101 TAg epitope mediated regressions of VP-MCC (13) and this tetramer approach is now in early clinical trials to treat VP-MCC (NCT01758458). HLA-A02restricted engineered TCR is also in phase I/II clinical trials (NCT03747484). However, the universal applicability of this approach is hindered by its HLA-restricted nature, which limits patient eligibility and also risks relapse from HLA down regulation by the tumor (14). Furthermore, relying on autologous cells might exclude many patients with VP-MCC who have dysfunctional T cells, are lymphopenic, or are immunocompromised.
An alternative immunotherapeutic strategy would be to transfer allogeneic MCPyV-specific T cells derived from an HLA matched healthy donor using peptide libraries of the viral oncoproteins. ACT with peptide library-pulsed ex vivo expanded virus or tumor antigen-specific T cells has been used with success in multiple clinical trials to treat viral infections, EBV-driven lymphomas, and solid tumors (15)(16)(17). MCPyV TAg specific CD4 + , but not CD8 + , T cells have been found in healthy donors and have not been evaluated for ACT (12,(18)(19)(20). Here, we show that polyclonal MCPyV TAg specific CD4 + T cells appropriate for ACT treatment of patients with VP-MCC can be generated in vitro from healthy donors using methods easily adaptable to clinical grade manufacture.

Sample Acquisition
Elutriated peripheral monocytes and lymphocytes from healthy donors were obtained from the Department of Transfusion Medicine, NIH, Bethesda, MD, USA under NIH IRB-approved protocol 99-CC-0168 (NCT00001846). Research blood donors provided written informed consent and blood samples were de-identified prior to distribution. Cells were assessed for viability using trypan blue exclusion and cryopreserved in 10% DMSO (Sigma Aldrich), 20% human serum (Gemini Bio-Products), and 0.065 mg/ml gentamycin (Quality Biological) in RPMI (Gibco) for later use. HLA class I and class II sequence-based typing was performed by the NIH Department of Transfusion Medicine's HLA Laboratory.

Culture Media
Primary leukocytes were cultured in cell growth medium in the presence of indicated cytokines. Cell growth medium contains 45% RPMI 1640, 50% AIM V CTS (Gibco by Life Technologies), and 5% non-AB pooled human serum (Gemini Bio-Products) by volume and was supplemented with 65 ug/ml gentamycin (Quality Biological), and 1X penicillin-streptomycin-glutamine solution (Gibco by Life Technologies).

Primary moDC Transduction
On day four of moDC culture, cells were plated at 4 × 10 5 cells/well in a 96 well plate. Cells were transduced using recombinant lentiviral vectors encoding codon optimized genes for ST, LTT, GFP, and MAGE-A3 single infections. Cells were infected at a MOI of 20 and spinoculated at 1800xG at 33C for 2.5 h. Cells were then washed in warm medium, incubated overnight, and co-cultured with autologous expanded T cells. Intracellular cytokine analysis was performed as described above. Aliquots of each donor's transduced cells were assayed by RT-PCR for construct expression. MAGE-A3, ST, and LTT codon optimized lentiviral construct (Cellomics, Pittsburgh, PA, USA) and PCR primer sequences are available in the Supplementary Materials and Methods.

Peptide Libraries
Overlapping peptide libraries were used to activate antigen specific T cells. Each library contained overlapping peptides with a length of 15 amino acids and an overlap of 11 amino acids across the fulllength of the protein sequence.

Generation of T Cell Lines
Antigen specific T cells were generated by stimulation of the lymphocyte population with peptide pulsed moDCs at a T cell to antigen presenting cell (APC) ratio of 5:1. Where indicated, starting lymphocyte populations were enriched for CD4 + T cells by immunomagnetic negative selection (EasySep Stem Cell Technologies). Cells were co-cultured in cell growth medium supplemented with either standard T cell growth cytokines of 10 ng/ml rhIL-7 and rhIL-15 (Peprotech) or enhanced cytokine conditions of 20 ng/ml rhIL-1b, 20 ng/ml rhIL-6, 10 ng/ml rhIL-7, 10 ng/ml rhIL-15, 50 ng/ml rhIL-21, 25 ng/ml rhIL-23, and 5 ng/ml rhTGFb (Peprotech). After 72 h of culture, 30 IU/ml rhIL-2 was added to the culture. Cultures were fed or split every 2-3 days as needed. Cultures were re-stimulated up to 2 times with peptide pulsed moDCs every 10-14 days.

T Cell Stimulation Assay
T cell lines were plated at 5 × 10 5 cells per well in a 96 well plate and stimulated for 12 h either with anti-CD28 and CD49d antibodies (BD Biosciences) and peptide libraries, peptide pulsed autologous moDCs, or transduced moDCs. An effector to APC ratio of 7:1 was used for moDC re-stimulations. After 1 h, Golgi inhibitors brefeldin A and monensin (GolgiStop ™ and GolgiPlug ™ ; BD Biosciences) were added according to manufacturer's instructions. An antigenspecific response was defined as secretion of IFNg or TNFa by >0.5% of lymphocytes and a frequency of reactive cells at least twice that of negative controls. The CD137 expression was determined by flow cytometry 16 h post antigen stimulation in the absence of Golgi inhibitors. Negative controls were either irrelevant antigen or vehicle where indicated. A "positive donor" was defined as a donor whose expanded cells had at least two successful antigen-specific T cell responses to MCPyV T antigen.

Flow Cytometry
Cells were plated in a 96 well plate at 0.5 × 10 6 cells/well and washed with PBS. Between each wash, plated cells were centrifuged at 500xG for 5 min. Cells were washed again with PBS then stained with fixable viability stain (Invitrogen L34955) in a 96 well plate at 4°C for 30<nbsp/>min. Cells were washed with FACS buffer (2% FBS and 2 uM EDTA in PBS) then stained using the appropriate surface antibodies in FACS buffer for 30 min at room temperature. Cells were washed, permeabilized, and intracellularly stained using either the fixation/permeabilization solution kit (BD Biosciences) kit for cytokine targets or the FOXP3 transcription factor staining buffer set (eBiosciences) for transcription factor targets following respective manufacturer's instructions. Samples were suspended and run on a BD Fortessa flow cytometer using an automated plate reader. Flow cytometry data were analyzed using FlowJo software (Treestar, Ashland, OR, USA). Gating strategies available in the Supplementary Materials and Methods. Polyfunctionality analysis was performed using SPICE v6 software (21).

Flow Cytometry Antibodies
A list of antibodies for flow cytometry is available in the Supplementary Materials and Methods.

Statistics
Statistics were performed using GraphPad Prism v7. Average and standard error of the mean are displayed in bar charts.

MCPyV TAg Specific T Cells are Efficiently Generated from Healthy Donor Blood
We investigated the role of various culturing conditions to improve the production of MCPyV TAg specific cells ex vivo. Cells from healthy donor blood was processed to generate moDCs. These cells were pulsed with TAg overlapping peptide libraries and co-cultured with autologous lymphocytes, allowing for the activation and expansion of reactive T cells. We initially used standard clinical production strategies to expand antigenspecific T cells, whereby lymphocytes were cultured with the cytokines IL-2, IL-7, and IL-15 as depicted in Figure 1B ((22-26); NCT02231853; NCT02694783). Standard cytokine containing medium did not yield detectable TAg specific cells. However, after enriching for CD4 + lymphocytes, standard cytokine medium achieved successful expansion in 2/12 tested donors (Supplementary Figure 1). Testing combinations of various cytokines in T cell culture media revealed that tumor antigen specific CD4 + T cell growth was improved by culturing in IL-1b, IL-2, IL-6, IL-7, IL-15, IL-21, IL-23, and TGFb, and that the addition of TGFb was critical for the improved performance of the cocktail (27). A manuscript characterizing the cytokine cocktail is in preparation. Applying this "enhanced" cytokine cocktail to CD4 + enriched cultures as depicted in Figure 1C lead to more frequent detection of expanded MCPyV TAg specific CD4 + T cells ( Figure 2B). Using these conditions, MCPyV TAg specific T cells were generated from 21/46 (45.6%) random healthy donors.

Expanded T Cells are Th1 Phenotype
To evaluate T cell polarization (Th) of the CD4 + T cells generated in the enhanced versus standard conditions, the expression of specific transcription factors and Th defining cytokines in stimulated cells from five donors were evaluated by flow cytometry. T cell cultures stimulated with overlapping peptide libraries and co-stimulated with anti-CD28 and anti-CD49d demonstrated that CD4 + enrichment resulted in the highest yield of TNFa and IFNg-secreting T cells in response to cognate antigen (Figures 2A, B). Because cytokine exposure in vitro can encourage growth of different T cell polarizations, we validated that the cultured CD4 + phenotypes were associated with tumor rejection. We were especially interested in confirming absence of immunosuppressive Tregs that may inhibit anti-tumor efficacy and are promoted by TGFb. As expected, 2 weeks of culture in the standard cytokines produced Th1 polarized CD4 T cells (22,(24)(25)(26). Despite the presence of TGFb, cultureintheenhanced cytokines also produced almost exclusively T-bet + FOXP3 -CD4 + cells, suggesting a Th1 polarization ( Figure 2C, D). Consistent with this, activating the cultured cells with PMA-ionomycin predominately induced expression of the Th1 cytokines TNFa and IFNg in CD4 + T cells. CD4 + enriched cells grown in the enhanced cytokine conditions also produced a statistically significant increase of IL-17A-producing CD4 + T cells ( Figure 2E). Additionally, MCPyV TAg-specific cells gated on CD3 + CD4 + TNFa + cells were polyfunctional for IL-2 and the Th1 effector molecules granzyme B, IFNg, and TNFa ( Figure 2F), a feature associated with immune-mediated tumor clearance (28,29). Cytotoxicity assays demonstrated that expanded MCPyV TAg-specific T cells do not induce killing of peptide pulsed autologous moDCs or phytohemagglutinin induced blasts (data not shown).

MCPyV TAg Peptide-Expanded T Cells Specifically Recognize TAg Presented by Autologous APCs
To investigate the specificities of the expanded T cells, we challenged MCPyV TAg reactive lines with antigen-loaded autologous moDCs. CD4 + T cell lines challenged with peptide-loaded autologous moDCs expressed the polarization independent T cell activation marker CD137 (n = 5; paired t test p = 0.02; Figure 3A) and the Th1 cytokine TNFa (n = 6; paired t test p = 0.02; Figure 3B) specifically in response to MCPyV TAg peptide relative to irrelevant peptide libraries. To confirm that peptide expanded T cells responded to naturally expressed epitopes from natively processed antigen, we challenged six MCPyV TAg expanded lines with TAg transduced autologous moDCs ( Figure 3E). The greatest response to either ST or LTT construct is shown. Stimulation with transduced TAg-expressing moDCs consistently produced more reactive cells than stimulation with moDCs transduced with the irrelevant antigen MAGE-A3 (Paired t test p = 0.02; Figures 3C, D).

All TAg Regions Expressed by VP-MCC Tumors are Immunogenic to Healthy Donors
To identify the relative immunogenicity of MCPyV TAg protein domains in healthy donors, we stimulated TAg specific T cells with three separate TAg peptide pools ( Figures 1A and 4A). MCPyV TAg T cell lines from 12 donors all recognized at least one peptide pool, and 3/12 donors demonstrated a polyclonal response ( Figure 4C). All three TAg regions were immunogenic among positive donors ( Figure 4B). These data suggest that MCPyV TAg is rich in potential epitopes that a diverse healthy donor T cell repertoire can recognize.  Longevity and potency of adoptively transferred T cell products post-transplant is likely a function of immune memory formation and lack of exhaustion. Several studies also indicate that less differentiated T cell products are associated with superior performance in vivo (30)(31)(32). To determine whether the enhanced cytokine cocktail produced terminal effector cells, we determined the memory phenotype of expanded MCPyV TAg cells by the differential expression of the markers CCR7, CD27, CD45RO, and CD95 (33,34). We also assayed exhaustion marker expression of the expanded T cell cultures.  Figure 5A). The majority of cultured CD3 + cells did not express the canonical exhaustion markers PD-1, TIM3, or LAG3, but 18.4 + /− 6.92% expressed PD-1 alone and 0.955 + /− 0.567% expressed TIM3 alone. A minority of cells co-expressed LAG3 + PD-1 + (1.87 + /− 1.75%) or TIM3 + PD-1 + (0.999 + /− 0.775%). Frequency of other exhaustion marker co-expression combinations was <0.5% and no cells expressed all three exhaustion markers ( Figure 5B). Taken together, these data demonstrate that the majority of expanded CD4 + MCPyV specific cells are not terminally differentiated and lack exhaustion markers, suggesting they would be suitable for ACT.

Successful Expansion of MCPyV TAg Specific Cells is Associated with Advanced Donor Age
To determine if certain donor HLA alleles were associated with successful expansion of peptide derived MCPyV TAg specific T cells, 20 donors used in this study were HLA typed and analyzed. Donors who produced MCPyV TAg reactive T cells in at least two independent assays were classified as positive responders. TAg specific cells were produced with no bias for donor HLA-A, B, Cw, DR, or DQ alleles (Fisher's exact test p > 0.1 Figures 6A, B;  Supplementary Figure 2). We then looked at whether additional donor characteristics such as age, sex, and race correlated with responder status. Of the 46 donors studied, 6 non-responder and 2 responders did not have donor age data available. Of the 38 evaluable donors, responders were older (median age 45.5 + /− 2.97 years) compared to non-responders (34.9 + /− 2.89 years); paired t test p = 0.01) ( Figure 6C). Notably, 75% of donors over 50 years old were responders (n = 12; Figure 6D). No significant trends in sex or race were identified.

DISCUSSION
We have developed a novel method to rapidly and reliably expand healthy donor T cells ex vivo targeting the primary oncogene of VP-MCC using a GMP-adaptable methodology. We produced MCPyV TAg specific T cells with phenotypic characteristics favorable for clinical cell products. Cells were polyfunctional CD4 + memory T cells minimally expressing PD-1. Peptide-expanded TAg specific T cells responded to natively processed MCPyV antigen and not to irrelevant antigens presented by autologous APCs, demonstrating specific on-target activation. Furthermore, multiple epitopes within the two TAg gene products were immunogenic in HLAunrestricted healthy donors, demonstrating a broad diversity of tumor relevant epitopes for VP-MCC immunotherapy.
Overlapping peptide libraries can recapitulate potential epitopes for an antigen of interest without the inefficiency of using whole protein and lack the clinical safety concerns of viral transfected APCs (35). However, synthetic peptides may produce an epitope not readily presented in a native context. Thus, it is critical to validate that peptide reactive cells can also recognize naturally processed antigens. Our TAg overlapping peptide libraries produced T cells that recognized immunogenic epitopes presented across various HLA alleles, and we confirmed that these peptide-generated T cells readily recognize naturally processed antigens presented by transduced autologous APCs. These data strongly suggest that the TAg peptide libraries accurately represent clinically relevant epitopes.
Our approach readily generated TAg specific CD4 + T cells from healthy donors that could be used in ACT. Although tumor eradication has classically been attributed to CD8 + cytotoxic T cells, it is increasingly recognized that CD4 + T cells also play an important role in tumor control. Clinical trials have demonstrated that ACT and tumor infiltration by Th1 CD4 + T cells can mediate regression of advanced solid tumors (36)(37)(38)(39)(40). Recently it was shown that MCPyV TAg-specific CD4 + T cells are enriched in VP-MCC tumors (20), demonstrating that MCPyV MHC class II restricted TAg epitopes are relevant to tumor immunity. Th1 CD4 + cells engage in canonical activation and recruitment of innate immune cells and CD8 + cells to direct anti-tumor immunity (41,42). They also directly kill MHC class II bearing tumor or stromal cells and rescue ICI refractory anti-tumor CD8 + cells in vivo to clear tumors (43,44). Murine models also demonstrated Th17 CD4 + cells are capable of tumor eradication (45,46). Lastly, since VP-MCC immune evasion to MCPyV TAg specific CD8 + can occur in VP-MCC tumors treated with checkpoint inhibitors and HLA class I  restricted transduced TCRs (14), the ACT of VP-MCC-specific CD4 + T cell could be beneficial in reversing such evasion. The hallmarks of successful T cell products for ACT are T cell persistence, polyfunctionality, and on-target specificity. T cell persistence is a function of memory status, exhaustion, and proliferative capacity. In clinical and murine models, ACT of central and effector memory cells are superior to terminal effector T cells (32,47). In addition, expression and engagement of exhaustion  immune checkpoints prevent anti-tumor T cell activation and subsequent proliferation in vivo. Our expanded TAg specific T cells lack terminal effector phenotypic markers and infrequently express canonical immune checkpoints, demonstrating a favorable persistence phenotype that is suitable for ACT. Furthermore, the expanded MCPyV TAg T cells are polyfunctional for Th1 effector cytokines as well as IL-2. IL-2 is a key cytokine associated with early polyfunctional memory T cells and naïve T cells.
In our work, we successfully generated CD4 + T cells specific for MCPyV TAg and little response in the CD8 + compartment, possibly due to inherent properties of our system. This preference for CD4 + T cells may be due to the 15mer length of the peptide libraries, however this peptide library format has readily produced antigen-specific CD8 + T cells for other viral targets, such as adenovirus and Epstein-Barr virus (48). Shorter peptide libraries may efficiently stimulate some CD8 + restricted epitopes that longer peptides fail to induce (49). It is therefore possible that cryptic HLA class I epitopes exist in the T cell repertoire of heathy individuals. Electroporated APCs and several rounds of stimulation produced MCPyV TAg-specific CD8 + T cells from healthy donors (50). However, several studies using HLA class I tetramers and multimers also failed to isolate CD8 + MCPyV TAg restricted epitopes from healthy donors, suggesting CD8 + TAg specific T cells may be low frequency in the healthy donor T cell repertoire (12,(18)(19)(20).
Clinical implementation of MCPyV directed ACT could be combined with additional immunotherapies to tailor the needs of each patient. In this proof of concept study, we used blood from a diverse set of healthy donors to produce T cells ex vivo that are specific to antigens present in all VP-MCC tumors. By selecting HLA-matched donors, MCPyV antigens expressed through the entire class II HLA array could be targeted by the T cell product. Off-the-shelf donor derived MCPyV CD4 + T cells sharing at least one HLA allele with the recipient could also be used. The possibility of graft-versus-host-disease remains a concern for allogeneic T cell transplant studies, and so current trials take a conservative approach in lymphodepletion preconditioning regimens, dosage, and initial T cell selection (51)(52)(53). However, this approach could also be adapted for autologous material from VP-MCC patients, who often maintain a higher MCPyV antigenic burden (54). Potency of these products targeting TAg VP-MCC could be further enhanced by including CD8 + T cells with alternative tumor specificities such as MAGE-A3, survivin, and SPA17 expressed by VP-MCC (55)(56)(57). Although MCPyV TAg cells minimally express the exhaustion markers PD-1 and TIM3 after production, co-administration with ICIs could potentially improve potency and persistence of the adoptively transferred cells.
In conclusion, we have discovered that heathy donors frequently have circulating CD4 + T cells specific for MCPyV TAg, that these cells are present across most HLA types, and that they can be efficiently expanded using peptide libraries under enhanced cytokine culture conditions. This study identifies a practical strategy for treating patients with adoptively transferred TAg specific CD4 + T cells for immunotherapy of VP-MCC.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/Supplementary Material. Further inquiries can be directed to the corresponding authors.

ETHICS STATEMENT
The studies involving human participants were reviewed and approved by National Institutes of Health Institutional Review Board. Written informed consent to participate in this study was provided by the participants' legal guardian/next of kin.