Rapidly Generated Multivirus-specific Cytotoxic T Lymphocytes for the Prophylaxis and Treatment of Viral Infections

Introduction

Although hematopoietic stem cell transplant (HSCT) may cure hematological malignancies and genetic disorders, extension to donors other than HLA-matched siblings has resulted in the emergence of viral infections as major contributors to post-transplant morbidity and mortality.^1,2,3,4 With the advent of more intensive viral screening and improved detection, increasing numbers of viral pathogens have been implicated in these complications, expanding from cytomegalovirus (CMV), Epstein–Barr virus (EBV), herpes-simplex virus, Adenovirus (Adv), and BK to include human herpes virus (HHV)-6, respiratory syncytial virus (RSV), parainfluenza, and influenza.² While pharmacological agents are standard therapy for some, they have substantial toxicities, generate resistant variants, are frequently ineffective and do not provide long-term protection.^5,6

Restoration of virus-specific immunity offers an attractive alternative to conventional drugs. We have shown that in vitro expanded virus-specific cytotoxic T lymphocytes (CTL) generated from stem cell donors with specificity for one (EBV), two (EBV and Adv), or three (EBV, CMV, and Adv) viruses are safe and effectively prevent and treat viral infection or disease in the HSCT setting.^7,8,9 More recently, banked, partially HLA-matched virus-specific CTL (3rd party CTLs) are showing promise in allograft recipients with advanced viral disease.¹⁰

Despite these encouraging clinical results broader implementation of T cell therapy is restricted by (i) the limited spectrum of viruses that can be effectively targeted in a single T cell line and (ii) the logistics of manufacture. Antigenic competition between high and low frequency T cells as well as between multiple antigens expressed at different levels and competing for presentation on shared antigen-presenting cells (APCs) may favor generation of lines dominated by responses to a single virus or to a restricted spectrum of viral antigens,^11,12 thus limiting the antiviral coverage provided by a single T cell product. In addition, our current manufacturing process is complex, requiring infectious virus material (EBV/Adv), production of a clinical grade vector, and prolonged (10–12 weeks) in vitro culture.^8,9 To address this latter problem some groups have evaluated more rapid approaches for producing T cell products for adoptive transfer. These include multimer selection to directly isolate virus-specific CD8⁺ T cells from peripheral blood,¹³ as well as the selection of cells based on cytokine production [interferon γ (IFNγ)] or expression of activation markers (e.g., CD154) following antigen exposure.^14,15,16 However, these approaches are expensive, require a large starting blood volume, which is not always available, particularly in the matched unrelated donor setting, and cannot be applied to viruses with low circulating T cell precursor frequencies.

We now describe a mechanism by which we can rapidly generate a single preparation of polyclonal (CD4⁺ and CD8⁺) CTLs, which is consistently specific for 15 immunodominant and subdominant antigens derived from seven viruses (EBV, CMV, Adv, BK, HHV6, RSV, and Influenza) that are frequent causes of post-transplant disease or death. Our approach, which uses standardized (synthetic) peptides as a stimulus and enhancing cytokines to promote the survival and expansion of T cells, is readily adaptable to clinical implementation and may thus be used as a safe and effective broad spectrum antiviral agent for all high risk transplant recipients.

Results

Discussion

We have shown that we can rapidly generate polyclonal, CD4⁺ and CD8⁺ T cells with specificities directed to a wide range of lytic and latent viruses responsible for infection in the immunocompromised host and after HSCT. These cells were Th1-polarized, had high avidity for a multiplicity of individual viral antigens, produced multiple effector cytokines upon stimulation, and killed virus-infected targets without alloreactivity. Because we generated these T cells using combinations of clinically available peptide libraries and prosurvival cytokines, our approach should be well suited to clinical application.

While CMV, EBV, and Adv are the most frequently detected viral infections following allogeneic HSCT, recipients are also susceptible to numerous other viruses, including BK, JC, HHV6, HHV7, influenza, parainfluenza, coronavirus, and RSV, all of which may cause severe morbidity and mortality.^1,2 Several of these viruses are usually seasonally detected (e.g., influenza, RSV) while others, such as HHV7, JC, and coronavirus, are infrequent, so that it is impracticable to cover all these pathogens post-transplant by generating individualized patient and single virus-specific T cell products. Hence, we sought to develop a strategy that would enable the production of a single CTL line with simultaneous specificity for a multiplicity of antigens, which were chosen based on their T cell stimulatory capacity and protective nature in our own studies (refs. 8,9,18,22,23 and U. Gerdemann, L. Keukens, J.H. Keirnan, U.L. Katari, C.T.Q. Ngugen, A.P. de Pagtee et al., manuscript submitted) and in those from other groups.^{13,15,19,24,25,26,27,28,29,30,31,32}

In our present clinical trials of virus-specific T cells, we have used EBV-LCL, adenovectors and/or viral antigen-encoding DNA plasmids to generate virus-directed T cells.^8,9,33 The use of full-length antigen ensures that CTL can be generated from all donors, irrespective of HLA, and that the antigen is physiologically processed by APCs and produces CTLs that recognize multiple CD4⁺ and CD8⁺ T cell epitopes and have sufficient avidity to kill virus-infected targets. The induction of lines that recognize multiple epitopes also minimizes virus escape due to epitope loss and produces potent and sustained antiviral activity in vivo.³⁴ However, the requirements for live virus/vectors are barriers to broader and late phase clinical studies, and also limit the number of pathogens to which a single T cell line can be directed.^8,9 We therefore evaluated whether clinically applicable pepmixes could be used as an alternative. Though clinical studies using minimal epitope peptides as vaccines have resulted in immune tolerance or the activation of low avidity T cells, Melief and colleagues recently demonstrated improved results with long (22-45aa) peptides containing both CD4⁺ and CD8⁺ epitope sequences.³⁵ They observed that these long peptides were processed endogenously, presented to T cells by APCs, and induced both helper and cytotoxic T cells, resulting in robust and effective CTL responses.³⁵ Based on these data, we chose to use a whole antigen source in the form of overlapping peptide libraries, but for optimal induction of polyclonal CTL we compared peptides of different lengths (15mers, 20mers, and 30mers) for stimulation. When using Hexon, which is predominantly recognized by CD4⁺ T cells, as a model antigen we saw no difference in the phenotype, specificity or epitope breadth of our lines. However, Maecker and colleagues found that 20mers were suboptimal for activating pp65-directed CD8⁺ T cells.³⁶ Thus, though still rather controversial, the optimal length of peptide for the activation of T cell lines containing both CD4⁺ and CD8⁺ cells appears to be 15aa. This analysis also highlights the differences between delivering peptides as a vaccine, where one relies on endogenous APCs to take up and process antigen versus in vitro T cell activation using professional APCs within PBMCs at optimal effector:target ratios, where peptides are either endogenously processed or edited by exoproteases following binding to cell surface HLA molecules.^37,38,39,40 Given the ready clinical availability of pepmixes containing 15mer peptides that cover all possible CD8⁺ and the majority of CD4⁺ epitopes, we substituted this antigen source and were able to demonstrate equivalency to “conventionally generated” CTLs with respect to both epitope specificity and avidity.

We next addressed how best to extend the breadth of antigen/epitope specificities that could be accommodated within a single CTL line. Physiologically, T cells are activated when they receive signals from T-cell receptor stimulation (signal 1), costimulation (signal 2), and cytokines (signal 3). Our “conventional CTLs” are activated in the absence of exogenous cytokines, a deficit that appears to adversely affect their proliferative capacity in vitro and also increases their susceptibility to activation-induced cell death, likely resulting in a more restricted repertoire of epitope recognition. Consistent with this possibility, we found that both the frequency and breadth of cells with viral specificity could be increased by supplementing cultures with inflammatory and prosurvival cytokines at initiation. We chose to test cytokines that support cell proliferation in vitro and in vivo (IL2, IL15),^41,42 as well as combinations (IL4+7) that also support the retention of a central memory phenotype, and promote the survival of activated T cells by upregulation of anti-apoptotic molecules e.g., Bcl-2.^43,44,45,46 Only lines supplemented with IL4+7 selectively promoted the expansion and survival of both CD4⁺ and CD8⁺ virus-specific T cells recognizing multiple epitopes, a combination that should favor the subsequent sustained expansion of transferred cells in vivo.³⁴ Of note, the induced cells were Th1-polarized despite exposure to IL4, a prototypic Th2 cytokine. Given the clinical availability of both cytokines and their safety in human clinical trials,^47,48 IL4+7 fulfilled the requirements of the current study, however, other proinflammatory cytokines capable of mimicking the milieu present during viral infection may produce similar benefits. For example, von Rossum and colleagues recently reported that CD3/28-activated CD8⁺ T cells cultured in an inflammatory cocktail consisting of IL1+IL6+IL23 underwent significantly less cell death after activation as compared with cells activated in any of the cytokines alone or activated in the presence of IL12.⁴⁹

The direct stimulation of PBMCs with pepmixes and culture in cytokine-supplemented conditions also allowed us to overcome a second major barrier to increasing the spectrum of viruses targeted in a single CTL line, namely antigenic competition resulting from the use of a common APC to simultaneously present multiple antigenic components from different viruses.^8,9 Antigenic competition results both from limited access of peptides to HLA molecules and physical constraints on the simultaneous stimulation of both high and low frequency T cells.^11,12 To overcome these issues, investigators have used artificial APCs that are engineered with molecules to provide the necessary T-cell receptor and co-stimulatory events required for immune synapse formation.⁵⁰ However, to avoid the inevitable complexities and costs of introducing a gene-modified cellular product into the manufacturing process, we evaluated whether patient PBMCs themselves could act as both a source of antigen-presenting and responding cells. B cells, monocytes and macrophages may all have the capacity to present antigen to T cells and these APCs can utilize endo- and exopeptidases to liberate class I or class II epitopes from 15mer peptides. By taking advantage of these properties, we can avoid reliance on a single APC endogenously expressing multiple antigens at different levels as a shared T cell stimulator, and instead have a diverse group of APCs in which each cell has the potential to display a diverse repertoire of peptides, allowing sufficient access for both high and low frequency T cells. Thus, antigenic competition both within the APC and between T cells could be alleviated. As proof of principal we generated a single culture of T cells with reactivity for 15 antigens derived from 7 latent and lytic viruses (EBV, CMV, BK, HHV6, Adv, Flu, and RSV) using pooled pepmixes as a stimulus and saw no evidence of competition. Additional pathogens may be included in this platform, although ultimately APC numbers will likely eventually become limiting, thus additions must be performed in a stepwise manner and one must evaluate changes in the frequency and breadth of T cell recognition of all peptides in the mix.

Critically for clinical feasibility, our approach was able to produce large numbers of virus-specific T cells. By seeding just 1.5 × 10⁷ PBMCs in the G-Rex and a single in vitro stimulation we could regularly manufacture 1 × 10⁸ CTLs within 10 days, with a >10-fold enrichment in virus-specific cells and a corresponding reduction in alloreactive T cells to levels observed in repetitively stimulated conventional CTLs, which have a proven safety record in vivo. Thus, using our new manufacturing technology we predict that multivirus-specific CTL will be safe for infusion after a single exposure to pepmixes and will provide broad spectrum antiviral protection without GvHD. Should additional cells be required, for example if banked virus-specific CTLs are established for 3rd party recipients, a second stimulation using pepmix-pulsed PHA blasts can expand the total number of CTLs without impairing their epitope specificity or breadth.

Materials and Methods

Donors and cell lines. PBMCs were obtained from healthy volunteers with informed consent using a Baylor College of Medicine institutional review board-approved protocol. The donors tested all had pre-existing immunity to the target viruses. PBMCs were used to generate DCs, CTL lines, and PHA blasts. PHA blasts were generated from PBMC (2 × 10⁶/ml) using PHA (5 µg/ml) and maintained in CTL media [RPMI 1640, 45% Click's (Irvine Scientific, Santa Ana, CA), 2 mmol/l GlutaMAX TM-I, and 5% Human AB Serum] supplemented with IL2 (100 U/ml; NIH, Bethesda, VA), which was replenished every 3 days.

CTL generation—peptide stimulation

Peptides/pepmixes. For PBMC stimulation we used commercially available pepmixes (15mers overlapping by 11aa) spanning EBV-LMP2, BZLF1, EBNA1; Adv-Penton, Hexon; CMV-pp65, IE-1; BKV-VP1, large T; Influenza A-MP1 (H3N2), NP (H3N2); RSV-F, N; JPT Technology, Berlin, Germany. Pepmixes spanning HHV6 U14 and U90 were synthesized by Genemed Synthesis, San Antonio, TX. Peptide libraries spanning the 414aa C-terminus of Adv-Hexon were synthesized by Proimmune, Oxford, UK or Alta Bioscience, University of Birmingham, Edgbaston, Birmingham, UK. Lyophilized peptides were reconstituted at 5 mg/ml in DMSO.

PBMC stimulation. 15 × 10⁶ fresh/frozen PBMCs were pelleted in a 15-ml tube and pulsed for 30–60 minutes at 37 °C with peptide libraries/pepmixes, either singly or pooled, at a concentration of 100 ng/peptide/15 × 10⁶ PBMCs. After incubation cells were resuspended in CTL media alone or supplemented with cytokines (as outlined below) and transferred to a G-Rex10 (Wilson Wolf Manufacturing, New Brighton, MN) (15 × 10⁶/G-Rex10) or plated out in a 24-well plate (2 × 10⁶/well). Media and cytokines were replenished on day 5, and cultures were split when they reached a density >50 × 10⁶/G-Rex10 or >3 × 10⁶ cell/24-well. On day 9–12 CTLs were harvested, counted and used for phenotypic and functional studies.

Cytokines for promoting CTL activation and expansion. We compared four conditions; (i) no cytokine, (ii) IL7 (10 ng/ml) + IL4 (1,666 U/ml), (iii) IL15 (5 ng/ml) (R&D Systems, Minneapolis, MN), and (iv) IL2 (20 U/ml). Cytokines were added to CTLs at day 0 and replenished on day 5.

CTL expansion. For expansion CTLs were restimulated at a S:R ratio of 1:1 with irradiated (30Gy) pepmix-pulsed autologous PHA blasts in CTL media with IL4+7 and IL15 (5 ng/ml) on the day of restimulation and fed with IL15 twice weekly. Seven days later CTLs were harvested, and used for further studies.

CTL generation using plasmid-nucleofected DCs

Plasmids. DNA plasmids expressing the viral antigens (i) Hexon and Penton, (ii) IE-1 and pp65, and (iii) EBNA1ΔLMP2 and BZLF1 were generated by NTC (Lincoln, NE).³³

CTL activation. Monocyte-derived DCs were matured on day 5 using a cytokine cocktail containing 100 ng IL6, 10 ng/ml IL-1β, 10 ng/ml TNF-α (R&D Systems), 1 µg/ml PGE2 (Sigma, St Louis, MO), 800 U/ml granulocyte-macrophage colony-stimulating factor (Immunex, Seattle, WA) and 1,000 U/ml IL4 for 48 hours. Twenty four hours after maturation 0.5–1 × 10⁶ DCs were nucleofected with 5 µg/plasmid using program U2 (Lonza Nucleofector IV) and subsequently incubated overnight in DC media (CellGenix, 2 mmol/l GlutaMAX TM-I) (CellGenix, Antioch, IL) with the cytokine maturation cocktail and then used for T cell stimulation. For CTL initiation CD14-ve 15 × 10⁶ PBMCs were stimulated with nucleofected DCs at a S:R ratio of 1:20 in the G-Rex10 in CTL media supplemented with IL4+7.³³

Flow cytometry

Immunophenotyping. CTLs were surface-stained with monoclonal antibodies to: CD3, CD4, CD8, CD16, CD56, CD28, CD45RO, and CD62L (Becton Dickinson BD, Franklin Lakes, NJ). Cells were washed once with phosphate-buffered saline (PBS) (Sigma) containing 2% fetal bovine serum (FBS) (HyClone, Thermo Fisher Scientific, Logan, UT), pelleted, and antibodies added in saturating amounts (10 µl). After 15 minutes at 4 °C in the dark, cells were washed twice and analyzed. Approximately 20,000 live cells were acquired using a FACS Calibur equipped with Cell Quest software.

CFSE. To measure cell proliferation PBMCs were isolated, pelleted and pulsed with pp65 pepmix (100 ng/15 × 10⁶ PBMC) for 30–60 minutes. Next PBMCs were washed twice using PBS+0.1% FBS and incubated for 10 minutes with 150 µl/20 × 10⁶ PBMC 10 µmol/l CSFE. Subsequently FBS was added at a 1:1 ratio and incubated for 10 minutes at 37 °C. After CFSE labeling PBMCs were washed twice using PBS+2% FBS and plated at a concentration of 1 × 10⁶/ml in CTL media with cytokines. Dilution of CFSE was examined every 2–3 days by flow after surface staining with CD3, CD4, CD8, and CD56.

FoxP3 staining. To measure regulatory T cells Foxp3 staining was performed using the e-Bioscience FoxP3 staining kit. Briefly, CTLs were rested in CTL media for 48 hours, then 1 × 10⁶ CTLs were resuspended in PBS+2% FBS and surface stained for CD3, CD25 and CD4. After washing the cells were resuspended in 1 ml Fixation/Permeabilization solution and incubated for 1 hour at 4 °C, then washed, resuspended in permeabilization buffer and incubated with 0.2 µl isotype or 10 µl FoxP3 antibody (Clone PCH101) for 30 minutes at 4 °C. After a final wash cells were acquired using a FACSCalibur equipped with Cell Quest software.

Intracellular cytokine staining. CTLs were harvested, resuspended at a concentration of 5 × 10⁶/ml in CTL media and plated at 200 µl/well in a 96 well plate. The cells were then stimulated with 100 ng of test or control pepmix in the presence of Brefeldin A (1 µg/ml), (BD) CD28 and CD49d (1 µg/ml) for 5–7 hours. Subsequently, CTLs were washed with PBS+2% FBS, pelleted, and surface stained with CD8, CD4, and CD3 (10 µl/antibody/tube). After 15 minutes, cells were washed twice, pelleted, fixed, and permeabilized with Cytofix/Cytoperm solution (BD) for 20 minutes at 4 °C in the dark. After washing twice with PBS/2%FBS containing 0.1% saponin (Calbiochem, EMD Chemicals, Gibbstown, NJ) cells were incubated with 20 µl IFNγ and/or TNFα antibodies (BD) for 30 minutes at 4 °C in the dark. Cells were then washed twice with cold PBS/2%FBS containing 0.1% saponin and at least 200,000 live cells from each population were analyzed with a FACSCalibur equipped with Cell Quest software (BD).

Functional studies

Multiplex assay. To assess cytokine production we used a multiplex assays. 1 × 10⁵ pp65-CTLs were restimulated using 500 ng/ml pp65 or control pepmix. After 16 hours supernatant was collected and the cytokine profile assessed using the MILLIPLEX High Sensitivity Human Cytokine Magnetic Bead Panel (Millipore, Billerica, MA). Specifically, 50 µl supernatant was incubated overnight at 4 °C with cytokine antibody beads. After incubation, samples were washed and incubated for 1 hour at room temperature with the biotinylated detection antibody. Finally Streptavidin–Phycoerythrin was added for 30 minutes at room temperature, then samples were washed and analyzed using the Luminex 200 instrument. Samples were run in duplicate.

Enzyme-linked immunospot assay. We used ELISpot to quantify IFNγ -producing T cells and assess the breadth of reactivity in our CTL lines. The populations were serially diluted from 4-1 × 10⁵ cells/well, and antigen-specific activity measured after direct pepmix or peptide minipool stimulation. Each condition was run in triplicate. After 20 hours, plates were developed, dried overnight at room temperature, then sent to Zellnet Consulting, New York, NY for quantification. SFC and input cell numbers were plotted, and a linear regression calculated after excluding plateau data points.

T-cell receptor avidity assessment. T-cell receptor avidity was assessed by IFNγ ELISpot. 2 × 10⁵ CTLs were stimulated with serial dilutions of pepmixes (pp65, Hexon) or 9mer peptides (NLV-pp65: NLVPMVATV HLA-A2 restricted, QYD-pp65: QYDPVAALF HLA-A24 restricted; TDL-Hexon: TDLGQNLLY HLA-A1 restricted). The frequency of T cells specific for each antigen/peptide was expressed as a percentage of the maximal SFC/input cell number.

Chromium release assay. We measured the cytotoxic specificity in a standard 4 hours Cr⁵¹ release assay, using E:T ratios of 40:1, 20:1, 10:1, and 5:1. CTLs were used as effectors and the targets were PHA blasts pulsed with pepmixes. Autologous and allogeneic PHA blasts alone or loaded with an irrelevant pepmix were used as specificity and alloreactivity controls. The percentage of specific lysis was calculated as [(experimental release – spontaneous release)/(maximum release – spontaneous release)] × 100.

SUPPLEMENTARY MATERIAL Figure S1. Phenotype and specificity of Penton and LMP2-specific CTLs generated in the presence of different growth-promoting cytokines. Figure S2. CD3⁺ T cell expansion after addition of growth-promoting cytokines. Figure S3. TCR avidity is comparable in Hexon DC plasmid-activated and pepmix-stimulated PBMCs. Figure S4. Comparable expansion of CTLs stimulated with pooled versus single pepmixes. Figure S5. Lack of alloreactivity in pepmix-stimulated PBMCs.

Supplementary Material