Article Text
PDF
Abstract
Background Adoptive cell therapy, such as chimeric antigen receptor (CAR)-T cell therapy, has improved patient outcomes for hematological malignancies. Currently, four of the six FDA-approved CAR-T cell products use the FMC63-based αCD19 single-chain variable fragment, derived from a murine monoclonal antibody, as the extracellular binding domain. Clinical studies demonstrate that patients develop humoral and cellular immune responses to the non-self CAR components of autologous CAR-T cells or donor-specific antigens of allogeneic CAR-T cells, which is thought to potentially limit CAR-T cell persistence and the success of repeated dosing.
Methods In this study, we implemented a one-shot approach to prevent rejection of engineered T cells by simultaneously reducing antigen presentation and the surface expression of both Classes of the major histocompatibility complex (MHC) via expression of the viral inhibitors of transporter associated with antigen processing (TAPi) in combination with a transgene coding for shRNA targeting class II MHC transactivator (CIITA). The optimal combination was screened in vitro by flow cytometric analysis and mixed lymphocyte reaction assays and was validated in vivo in mouse models of leukemia and lymphoma. Functionality was assessed in an autologous setting using patient samples and in an allogeneic setting using an allogeneic mouse model.
Results The combination of the Epstein-Barr virus TAPi and an shRNA targeting CIITA was efficient and effective at reducing cell surface MHC classes I and II in αCD19 ‘stealth’ CAR-T cells while retaining in vitro and in vivo antitumor functionality. Mixed lymphocyte reaction assays and IFNγ ELISpot assays performed with T cells from patients previously treated with autologous αCD19 CAR-T cells confirm that CAR T cells expressing the stealth transgenes evade allogeneic and autologous anti-CAR responses, which was further validated in vivo. Importantly, we noted anti-CAR-T cell responses in patients who had received multiple CAR-T cell infusions, and this response was reduced on in vitro restimulation with autologous CARs containing the stealth transgenes.
Conclusions Together, these data suggest that the proposed stealth transgenes may reduce the immunogenicity of autologous and allogeneic cellular therapeutics. Moreover, patient data indicate that repeated doses of autologous FMC63-based αCD19 CAR-T cells significantly increased the anti-CAR T cell responses in these patients.
- Cell Engineering
- Receptors, Chimeric Antigen
- Immunotherapy
Data availability statement
Data are available on reasonable request. All data relevant to the study are included in the article or uploaded as online supplemental information.
This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See http://creativecommons.org/licenses/by-nc/4.0/.
Statistics from Altmetric.com
WHAT IS ALREADY KNOWN ON THIS TOPIC
The immune system of patients develops specific anti-chimeric antigen receptor (CAR) responses after autologous CAR-T cell administration, limiting CAR-T cell persistence and the administration of multiple doses. This is exacerbated for allogeneic CAR-T cell therapies.
WHAT THIS STUDY ADDS
This study proposes a simplified manufacturing strategy to decrease major histocompatibility complex I and II molecule expression, alleviating the immunogenicity of autologous and allogeneic CAR-T cells. This study also reveals an increased anti-CAR response in patients who received multiple doses of CAR-T cells.
HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY
This study highlights an increased anti-CAR response in patients who received multiple CAR-T cell doses. This warrants the consideration of reducing the immunogenicity of autologous CAR-T cells in addition to the current efforts to develop allogeneic products that avoid elimination by the host immune system.
Background
Adoptive cell therapy, such as chimeric antigen receptor (CAR) CAR-T cell therapy, has improved patient outcomes for hematological malignancies.1 2 However, clinical studies demonstrate that autologous CAR-T cell products elicit humoral and cellular responses to the non-self components of the CAR in patients, thus limiting CAR-T cell persistence and the success of administering multiple doses.3 4 Allogeneic or “off-the-self” CAR-T cells originating from healthy donors are in clinical development but are even more prone to host rejection and, thus, may have drastically limited expansion and persistence—particularly in the absence of deep lymphodepleting regimens (such as alemtuzumab or equivalent).4 5
Currently, four of the six U.S. Food and Drug Administration (FDA)-approved CAR-T cell products (all autologous) use the FMC63-based αCD19 single-chain variable fragment (scFv), derived from a murine monoclonal antibody, as the extracellular binding domain.1 2 Clinical trials have indicated that treatment with FMC63-based autologous CAR-T cells elicits an immunological T cell response directed at specifically against the murine portions of scFv within the CAR and can limit CAR-T cell efficacy and persistence.6 Interestingly, human or humanized scFvs can also contain non-self sequences since the variable binding fragments are generated through multiple gene recombination events and somatic hypermutations. Additionally, the expression of proteins encoded by several human genes in a single peptide CAR chain creates fusion sequences at these junctions that do normally not exist and may be immunogenic.3 6–8 Patients with CD19-positive disease relapse could theoretically receive a subsequent infusion of αCD19 CAR-T cells, although clinical responses to second or subsequent infusions tend to be less effective, and cytotoxic T cells with specificity toward the CAR expand in patients after the initial infusion.3 9–12
For allogeneic CAR-T cell therapy, the issue of rejection is even greater due to HLA mismatch.4 13 This is thought to be detrimental to CAR-T cell therapy, as positive clinical outcomes of CAR-T cell therapy are strongly correlated with expansion and persistence of the infused cells.14–17 Solutions have included deep host immunosuppression,18 which unfortunately results in increased infectious complications, and/or complex gene-editing, which increases risks of off-target effects and translocations—especially with multiple genetic edits.19–22 Most of these allogeneic approaches have focused on the elimination of major histocompatibility complex (MHC) class I via gene knockout of beta-2-microglobulin (β2M) (CRISPR therapeutics),23 Precision Biosciences,24 as β2M is required for cell surface expression of MHC I, but this may also result in increased susceptibility of the therapeutic immune effector cells to NK cell-mediated rejection.25
In this study, we propose a simplified one-shot approach to easily generate autologous CAR-T cells with reduced cell surface expression of MHC class I and II by including the gene encoding Epstein-Barr virus (EBV) BNLF2a and an shRNA targeting vlass II transactivator (CIITA) in the CAR-transduction plasmid. This one-shot, additive stealth transgene approach generates functional CAR-T cells with reduced immunogenicity.
Methods
Mice and cell lines
NSG mice were purchased from Jackson Laboratory and bred under pathogen-free conditions at the Center for Comparative Medicine at Massachusetts General Hospital (MGH). HEKT cells, NALM-6 (acute lymphoblastic leukemia), JeKo-1 (mantle cell lymphoma), and K562 (chronic myelogenous leukemia) were purchased from American Type Culture Collection, maintained as outlined by the supplier and, where indicated, transduced to express click beetle green-luciferase/enhanced green fluorescent protein (eGFP). Cell lines were authenticated by short tandem repeat profiling and routinely tested to exclude mycoplasma infection.
(Stealth) CAR-T cell production
In brief, primary human T cells were purified from peripheral blood (Stem Cell Technologies) and activated on day 0 using CD3/CD28-Dynabeads (Life Technologies). Lentiviral transduction was performed on day 1, and on day 5 CD3/CD28-Dynabeads were removed. Where applicable, T cells were electroporated with Cas9 mRNA on day 5. In cases of flow-based sorting, T cells were sorted on day 8 using the eGFP marker and expanded until day 14 to be subsequently cryopreserved. When unsorted CAR-T cells were used, CAR-T cells were normalized for transduction efficiency using untransduced activated T cells from the same donor and expansion.
Cytotoxicity assay
CAR-T cells were incubated with luciferase-expressing tumor targets at indicated effector to target (E:T) ratios for 24 hours. The remaining luciferase activity was measured with a Synergy Neo2 luminescence microplate reader (Biotek). To assess the NK cell-mediated cytotoxicity, NK cells were purified from blood or frozen peripheral blood mononuclear cells (PBMCs) using an NK cell isolation kit (Stem Cell Technologies) and primed with 20 IU/mL hIL-2 and coincubated with CFSE-stained target cells (Life Technologies). After 3 hours, αCD107a antibodies were added. After a total of 4 hour, cells were centrifuged, resuspended with dead/alive marker SYTOXred (Life Technologies), and assessed by flow cytometer for viability and NK cell degranulation.
ELISpot assay
Plates with Immobilon-P membrane (Millipore) were activated and coated with anti-human IFNγ antibody (Clone NIB42, Biolegend). After blocking with 1% BSA, 5×105 PBMCs or 2×105 T cells were co-incubated with respective peptides, antigens, or stimulants. After 24 hours, the plate was washed and incubated with anti-human IFNγ antibody (Clone 4S.B3, Biolegend). After washing, the plate was incubated with avidin-HRP (Biolegend), developed using the BD Elispot AEC Substrate and analyzed with ImmunoSpot-Reader systems. All antibodies were used according to the manufacturers’ recommendation.
ELISA
Interferon ỿ from supernatants was measured following an overnight co-incubation of NLV responder T cells with target at a E:T ratio of 1:5 using Human DuoSet ELISA kits (R&D systems).
Flow cytometry
Cells were stained for 30 min at 4°C and washed twice with RPMI before analysis. SYTOXRed or SYTOXBlue (Life Technologies) was added as dead/alive marker, and singlet discrimination was performed on FSC and SSC detectors. Following antibodies were used according to the manufacturers’ recommendations in combination with their respective isotype control (online supplemental table 1). Antibody binding capacity was measured using Quantum Simply Cellar beads (Bangs laboratories). Analysis was performed by FlowJo software (BD Biosciences).
Supplemental material
Mixed lymphocyte reaction assay
Stealth or CAR-T cells were stained with CFSE (Life Technologies) while autologous or allogeneic T cells were stained with CellTrace Violet (Life Technologies) before being co-incubated at a 4:1 ratio in the presence of 20 IU/mL hIL-2 and either isotype or MHC I (W6/32, Biolegend) or MHC II (Tü39, Biolegend) or both MHC I and II blocking antibodies. Fresh IL-2 was added every other day and the T cells were pulsed with new stealth T cells and blocking antibodies on days 7 and 14. On day 16, T responder cells were stained with SYTOXRed (viability) and assessed by flow cytometry for cell division. Allogeneicity of cells was assessed by PCR (American Red Cross) and a minimum of five out of six mismatched (HLA-A/B/C/DP/DQ/DR) were selected.
In vivo study
Luciferized NALM-6 or JeKo-1 cells were injected (1×106 cells per mouse) in NSG mice by tail vein. Tumor growth was confirmed by bioluminescence, at which time the mice were treated with an injection of 2×106 CAR-T cells in the tail vein. Tumor progression was evaluated by bioluminescence emission using an Ami HT optical imaging system (Spectral Instruments). At day 14 (or as indicated), the blood of the mice was collected and analyzed by flow cytometry for the presence of tumor and CAR-T cells per microliter of blood. For the allogeneic T cell mouse model, ‘activated’ allogeneic T cells were activated with CD3/CD28 beads and ‘Primed’ allogeneic T cells were pulsed twice with irradiated (100 Gy) PBMC originating from the CAR-T-cell donor and then expanded by a rapid expansion protocol.26 The allogeneic T cells were injected in NSG mice by tail vein 1 day prior to NALM-6 tumor cell injection.
Stealth CAR design
DNA constructs were synthesized and cloned into a second-generation lentiviral backbone under the regulation of a human EF-1α promoter and/or a human U6 promoter. EBV BNLF2a, herpes simplex virus (HSV) ICP47 and human cytomegalovirus (HCMV) transporter associated with antigen processing inhibitors (TAPi) sequences were synthesized in combination with eGFP by 2A self-cleaving peptide. Similarly, vectors with CRISPR/Cas9 guides for β2M and CIITA were constructed. The shRNA targeting CIITA was designed with software of Dharmacon and the Whitehead institute. The lentiviral vector expressing the combination of shRNA CIITA3, EBV BNLF2a and eGFP was also constructed. For CAR constructs, plasmids expressing the FMC63-based anti-CD19 CAR were synthesized.27
Statistical methods
All statistical analyses were performed with GraphPad Prism V.9 software. Data were presented as means±SEM with statistically significant differences determined by tests as indicated in figure legends.
Results
Expression of viral TAP inhibitors in primary T cells decreases cell surface levels of MHC class I
Herpesviruses have convergently evolved to encode small proteins that inhibit TAP,28 a protein required for transporting cytoplasmic peptides across the endoplasmic reticulum and loading them for presentation on MHC class I molecules at the cell surface. Cells that lack expression of functional TAP complexes show a dramatic reduction in surface MHC I levels, substantially reducing their sensitivity to CD8+ T cells.29 We hypothesized that forced expression of viral TAP inhibitors (TAPi) would reduce MHC I expression in gene-modified cells, thereby preventing cell-mediated immune responses to foreign transgenes. To test if expression of herpesvirus TAPi reduced surface MHC I expression in primary T cells, bicistronic lentiviral constructs were generated to express HSV ICP47, HCMV US6, or EBV BNLF2a TAPi along with eGFP as transduction marker (figure 1A). Lentiviral constructs expressing sgRNA for β-2-microglobulin (β2M), without electroporated with Cas9 mRNA, were used as a positive control (β2M KO) or negative control (β2M−). At similar transduction efficiencies, TAPi-transduced cells had reduced levels of surface MHC I without affecting MHC class II upregulation on activation (figure 1B). Viral TAPi reduced total surface MHC I levels by at least one log-fold, which was maintained on additional stimulation by IFNγ or αCD3-antibody (figure 1C).
Lentivirus transduction of viral ransporter associated with antigen (TAP) inhibitors decreases expression of major histocompatibility complex (MHC) class I on human primary T cells, reducing allogeneic T cell responses, moderately suppressing NK cell-mediated killing, and diminishing pre-existing antiviral T cell response. (A) Schematic overviewing the MHC class I antigen presentation pathway and design of the lentiviral constructs expressing the viral TAPi or a CRISPR-guide for β2M. (B) MHC I and MHC II cell surface expression on primary human T cells transduced with the lentiviral constructs shown in (A) as measured by flow cytometry. Histograms show a representative donor (n=3 donors in total), with the isotype control in light gray, untransduced (UTD) control in dark gray, and transduced cells in colored histograms. Bar graphs represent the mean+SEM of the median fluorescence intensity (MFI) fold change relative to the UTD control of all three donors. Dots represent the fold change in median fluorescence intensity (MFI) for individual donors. (C) MHC class I antibody binding capacity (ABC) of primary human T cells transduced with the lentiviral constructs shown in (A) with or without additional stimulation of αCD3 (OKT3) or IFNγ, quantified by flow cytometry. Bars represent the mean+SEM ABC of three donors. Dots represent the ABC of individual donors. (D) TAPi-expressing or B2M KO primary T cells were incubated with autologous IL2-stimulated NK cells. T cell viability was measured via flow cytometry for CD3+SytoxRed cells. Susceptibility to NK cell lysis is reported as the % of dead cells (CD3+SytoxRed+; labeled as %NK cell-mediated killing). NK cell degranulation was measured by staining for CD107a+cells. Dots represent the mean values of four donors from two technical replicates. Lines connect values from the same donor. (E) TAPi-expressing or β2M KO T cells generated from three donors were mixed with allogeneic T cells from one additional donor or autologous T cells (responder cells) labeled with Celltrace Violet and proliferation was measured by flow cytometry after 16 days. Bars represent the mean+SEM of three donors. Dots represent the mean values of individual donors from three technical replicates with the same allogeneic responder cells. (F) TAPi-transduced or UTD primary T cells from HLA-A2+ donors were transduced with a vector coding for HCMV pp65 (containing the NLV peptide) and incubated with autologous NLV-specific CD8+T cells generated by peptide pulsing isolated CD8+T cells with the NLV peptide. T cell activation was assessed by IFNγ production, measured by ELISA. Bars represent the mean+SEM of two donors. Dots represent the mean values of individual donors from three technical replicates (G) (upper) peripheral blood mononuclear cells (PBMCs) from different donors were assessed in an IFNγ ELISPOT for pre-existing antiviral cellular immunity against herpes simplex virus (HSV), Epstein-Barr virus (EBV), and/or human cytomegalovirus (HCMV) by coincubation with the respective immunogenic peptides. Heat map indicates the mean number of IFNγ-producing cells per million PBMCS measured in triplicate of each donor. Pictures show representative wells from one donor. Donors were considered to have pre-existing immunity if they had ≥100 IFNγ-producing cells per million PBMCs. (lower) CD8+ T cells from donors with a pre-existing antiviral cellular immunity were isolated and incubated with autologous TAPi-transduced T cells in an IFNγ ELISPOT. Heat map indicates the mean number of IFNγ-producing cells per million cells, from three technical replicates of each donor. Pictures show representative wells from one donor. For all panels: statistical significance was measured by Student’s t-test compared with UTD or B2M, as shown; *p≤0.05, **p≤0.01, ***p≤0.001, ****p≤0.0001.
Since MHC I expression inhibits targeting by NK cells,30 we investigated the impact of MHC I downregulation on susceptibility to NK cell killing. Like previous reports,25 31 β2M KO T cells were susceptible to autologous NK cell lysis and induced NK cell degranulation, as measured by CD107a expression. Compared with β2M KO T cells, T cells expressing EBV viral TAPi triggered significantly reduced NK cell lysis or degranulation (figure 1D). Similarly, MHC I expression mediates allogeneic T cell responses due to a mismatch between the MHC and TCR. To measure the allogeneic response of TAPi-expressing T cells, a mixed lymphocyte reaction (MLR) was performed. Transduced T cells were incubated with autologous or allogeneic labeled responder T cells in the presence or absence of MHC I and II blocking antibodies. Responder T cell activation was measured by proliferation (figure 1E) and changes in CD69 and CD25 expression (online supplemental figure 1A,B). T cells transduced with viral TAPi, especially EBV BNLF2a, induced less allogeneic responder T cell activation, comparable to MHC I and/or MHC II blockade.
We next tested the ability of TAPi-expressing T cells to present cytoplasmic antigens by assessing the presentation of a peptide derived from the highly immunogenic HCMV pp65 protein. The immunogenic NLV peptide is presented on the HLA-A*02:01 allele and drives NLV-specific CD8+ T cells to secrete IFNγ.32 We first generated lines of NLV-specific ‘responder T cells’ by serial stimulation of PBMC derived from HLA-A*02:01 healthy donors who had evidence of CMV-specific memory responses. We then generated a panel of “stimulator T cells” derived from the same healthy donors, which were untransduced (UTD) or transduced with the constructs as shown (figure 1A), including the three different viral TAPi. Co-cultures of “stimulator cells” with “responder cells” demonstrated that viral TAPi expression, especially when derived from HSV or EBV, reduced antigen presentation, based on a reduction of IFNγ secretion in “responder T cells” (figure 1F). Despite reduced antigen presentation, using a viral protein to knock down the MHC I could lead to an immune response to its sequence. To measure the immunogenicity of viral TAPi transduction in T cells, we identified normal donors with pre-existing cellular immunity to the respective TAPi viruses. PBMCs from normal donors were screened with peptides known to be immunogenic and originating from HCMV, EBV, or HSV in an IFNγ ELISpot assay.33–36 T cells from normal donors with a detectable cellular response those viruses were then transduced with viral TAPi from the same virus and incubated with autologous CD8+ T cells. CD8 T cell activation was measured by IFNγ ELISpot (figure 1G). While T cells from an HCMV-responsive donor were activated in response to transduction with HCMV pp65, they did not respond to transduction with the CMV TAPi. Similarly, HSV-responsive and EBV-responsive donors did not produce IFNγ in response to HSV or EBV TAPi, indicating that these viral TAP inhibitors do not elicit T cell responses, despite the donors being responsive to other known immunogenic sequences from the same viruses.
Expression of shRNA targeting CIITA decreases cell surface levels of MHC class II
Activated human T cells express high levels of MHC class II molecules. In gene-modified cells, high MCH II could trigger rejection via antigen cross-presentation of the genetic modifications.3 37 Similar to MHC class I, direct targeting of MHC class II expression with DNA-editing techniques is highly complex and potentially patient-specific, as these genes are highly polymorphic and harbor significant allelic variation.38 We sought to reduce MHC class II expression by targeting CIITA, the main regulatory factor that controls the transcription of MHC II genes.39 To avoid the use of gene-editing and double-strand breaks, we chose to encode an shRNA targeting CIITA driven by a U6 promoter, which is commonly used to transcribe small shRNA sequences,40 into our lentiviral vectors using a panel of shRNA sequences (figure 2A). We also compared our shRNA vectors to gene knockout of CIITA with CRISPR/Cas9. We noted that primary human T cells maintained an activated profile due to the initial activation by CD3/CD28 beads in the manufacturing process, as observed by a significant MHC II upregulation. Transduction with CIITA-targeting shRNA reduced the cell surface expression of MHC II, comparable to CIITA KO, without affecting MHC I expression (figure 2B). Both CIITA-targeting strategies, CRISPR/Cas9 and shRNA, rendered T cells with less than 20,000 MHC class II molecules on their surface, which was unaffected by additional stimulation with IFNγ or αCD3-antibody (figure 2C). However, only shRNA CIITA3 reduced MHC II expression without compromising T cell proliferation (figure 2D). In an MLR using allogeneic or autologous responder T cells, shRNA-mediated knockdown of CIITA reduced responder T cell proliferation (figure 2E, online supplemental figure 2A,B).
Lentiviral transduction of shRNA targeting class II MHC transactivator (CIITA) decreases expression of MHC class II on human primary T cells, reducing allogeneic responses. (A) Schematic overviewing the major histocompatibility complex (MHC) II antigen presentation pathway and design of the lentiviral constructs expressing shRNA targeting CIITA or CRISPR-guide for CIITA. (B) MHC I and MHC II cell surface expression on primary human T cells transduced with the lentiviral constructs shown in (A) as measured by flow cytometry. Histograms show a representative donor (n=3 donors in total), with the isotype control in light gray, untransduced (UTD) control in dark gray, and transduced cells in colored histograms. Bar graphs represent the mean+SEM of the median fluorescence intensity (MFI) fold change relative to the UTD control of all three donors. Dots represent the fold change in MFI for individual donors. (C) MHC class II antibody binding capacity (ABC) of primary human T cells transduced with the lentiviral constructs shown in (A) with or without additional stimulation of αCD3 (OKT3) or IFNγ, quantified by flow cytometry. Bars represent the mean+SEM of three donors. Dots represent the values of individual donors. (D) Primary T cells transduced with CIITA-targeting shRNA (on day 0) were sorted for GFP+ (vector-expressing) cells via FACS on day nine post-transduction. Proliferation following was assessed by cell count until the end of the manufacturing cycle (day 14). Dots represent the mean+SEM of three donors. Each donor was measured in technical triplicates. (E) CIITA shRNA3-expressing T cells or CIITA KO T cells generated from three donors were mixed with allogeneic T cells from one additional donor or autologous T cells (responder cells) labeled with Celltrace Violet and proliferation was measured by flow cytometry after 16 days. Bars represent the mean+SEM of three donors. Dots represent the mean values of individual donors from three technical replicates with the same allogeneic responder cells. For all panels: statistical significance was measured by Student’s t-test compared with UTD; *p≤0.05, **p≤0.01, ***p≤0.001, ****p≤0.0001.
Expression of the viral TAP inhibitor EBV BNLF2a and an shRNA targeting CIITA can be combined in primary T cells to decrease cell surface levels of both MHC class I and class II.
Both strategies to downregulate the cell surface expression of MHC I or II were effective separately, but the question remained as to whether these strategies could be combined. The TAPi EBV BNLF2a was selected to be combined with the shRNA CIITA3. This TAPi reduced sufficient MHC I at the cell surface to suppress antigen presentation while the remaining MHC I at the cell surface can potentially suppress NK cell activation. These MHC I and II downregulation strategies were combined by including both EBV TAPi and shRNA CIITA3 into one lentiviral vector (figure 3A). When transduced into primary human T cells, the combined EBV-TAPi/shRNA-CIITA3 vector reduced MHC I and II expression (figure 3B,C) and reduced proliferative responses in MLRs (figure 3D, online supplemental figure 3A,B). This demonstrates that gene-modified primary T cells can successfully evade cellular immune responses by our proposed MHC class I and II downregulation strategies, creating “stealth” T cells.
Combining Epstein-Barr virus (EBV) transporter associated with antigen processing (TAPi) with class II MHC transactivator (CIITA)-targeting shRNA decreases major histocompatibility complex (MHC) I and II expression on primary human T cells, reducing allogeneic T cell responses. (A) Schematic overviewing the lentiviral construct combining EBV TAPi with shRNA CIITA3 and the control, single-expression constructs. (B) MHC I and MHC II expression on primary human T cells transduced with the lentiviral constructs shown in (A) as measured by flow cytometry. Histograms show a representative donor (n=3 donors in total), with the isotype control in light gray, untransduced (UTD) control in dark gray, and transduced cells in colored histograms. Bar graphs represent the mean+SEM of the median fluorescence intensity (MFI) fold change relative to the UTD control of all three donors. Dots represent the fold change in MFI for individual donors. (C) MHC class I (upper graph) and II (lower graph) antibody binding capacity (ABC) of human primary T cells transduced with the lentiviral constructs shown in (A) with or without additional stimulation of αCD3 (OKT3) or IFNγ, quantified by flow cytometry. Bars represent the mean+SEM of three donors. Dots represent the values of individual donors. (D) T cells expressing EBV TAPi and/or shRNA targeting CIITA cells generated from three donors were mixed with allogeneic T cells from one additional donor or autologous T cells (responder cells) labeled with Celltrace Violet and proliferation was measured by flow cytometry after 16 days. Bars represent the mean+SEM of three donors. Dots represent the mean values of individual donors from three technical replicates with the same allogeneic responder cells. For all panels: statistical significance was measured by Student’s t-test compared with UTD; *p≤0.05, **p≤0.01, ***p≤0.001, ****p≤0.0001.
Stealth-enabled αCD19 CAR-T cells are functional in vitro and in vivo
The murine scFv FMC63, recognizing CD19 and used in four of the six FDA-approved CAR-T cell products, has been reported to elicit autologous T cell responses in patients.3 6 Thus, we tested our stealth strategy in the context of FMC63 CARs to verify they retain function and avoid eliciting cellular immunity. The stealth FMC63-based αCD19 CAR was generated by incorporating both the EBV TAPi and shRNA CIITA3 (figure 4A). The transduction efficiency of the stealth CAR-T cells was reduced compared with the αCD19 CAR-T cells (as measured by the %GFP+cells), likely due to the increased construct size. However, there was no significant difference in GFP MFI between the stealth and αCD19 CAR-T cells, suggesting similar CAR transgene expression between the vectors (online supplemental figure 4A-C). The stealth αCD19 CAR-T cells had reduced MHC I and II molecules on their cell surface compared with the T cells transduced with the αCD19 CAR alone and had robust expression of EBV TAPi and reduced CIITA mRNA expression compared with the αCD19 CAR alone by qPCR (figure 4B). Interestingly, this reduction of MHC I molecules at the cell surface did not increase NK cell cytotoxicity, and proliferation of the CAR-T cells was unchanged compared with the untransduced T cells (figure 4C, D). Additionally, phenotypic analysis by CD4, CD8, CCR7, and CD45RA further showed no differences in CD4/CD8 ratios and memory phenotypes comparing the αCD19 CAR-T cells without the stealth technology (figure 4E). The stealth αCD19 CAR-T cells also maintained their ability to target tumor cells in vitro. When co-incubated with luciferase-expressing acute lymphoblastic leukemia (ALL) NALM6 cells or mantle cell lymphoma JeKo-1 cells, stealth αCD19 CAR-T cells reduced tumor cell viability to the same extent as αCD19 CAR-T cells (figure 4F). The in vivo functionality was also investigated. After tumor engraftment with NALM6 cells or JeKo-1, mice were left untreated or injected with αCD19 CAR-T cells with or without stealth technology. CAR-T cell expansion in the blood was assessed by flow cytometry, and tumor clearance was measured by bioluminescence imaging (BLI) (figure 5A,E). Mice treated with αCD19 CAR-T cells with or without stealth technology showed comparable tumor clearance while tumors vastly expanded in untreated mice by BLI (figure 5B,F). Both αCD19 CAR-T cells and stealth αCD19 CAR-T cells expanded similarly in the blood, as observed at day 14 by the presence of GFP+CD3+ cells (figure 5C,G). Tumor cells (GFP+CD3 cells) were absent or minimally present in the blood of CAR-T cell-treated groups, while a large expansion was found in the untreated group, similar to the BLI imaging. Kaplan-Meier survival curves demonstrated no difference in the survival of mice treated with αCD19 CAR-T cells with or without the additional stealth technology (figure 5D,H). In summary, stealth αCD19 CAR-T cells retained their ability to recognize and clear CD19-expressing cells both in vitro and in vivo.
Stealth αCD19 chimeric antigen receptor (CAR) T cells have a similar phenotype and function compared with control αCD19 CAR T cells in vitro. (A) Schematic overviewing the control (αCD19 CAR alone) and stealth (coexpression of Epstein-Barr virus (EBV) transporter associated with antigen processing (TAPi) and CIITA3 shRNA) αCD19 CAR lentiviral constructs. (B) (Left) Major histocompatibility complex (MHC) I and MHC II expression on primary human T cells transduced with the lentiviral constructs shown in (A) as measured by flow cytometry. Bar graphs represent the mean+SEM of the median fluorescence intensity (MFI) fold change relative to the UTD control of all three donors. Dots represent the fold change in MFI for individual donors. (right) EBV TAPi and CIITA expression in transduced T cells measured by qRT-PCR. Bars represent the mean+SEM of three donors. Dots represent the mean values of individual donors from three technical replicates. (C) UTD, control, or stealth CAR T cells were incubated with autologous IL2-stimulated NK cells. T cell viability was measured via flow cytometry for CD3+SytoxRed cells. Susceptibility to NK cell lysis is reported as the % of dead cells (CD3+SytoxRed+; labeled as % NK cell-mediated killing). K562 cells (lacking MHC I expression) were used as a positive control. Bars represent the mean+SEM of three donors. Dots represent the mean values of individual donors from three technical replicates. (D) Primary T cells transduced with the control or stealth constructs (on day 0) were sorted for GFP+ (vector-expressing) cells via FACS on day nine post-transduction. Proliferation following was assessed by cell count until the end of the manufacturing cycle (day 14). Dots represent the mean+SEM of three donors. Each donor was measured in technical triplicates. (E) (upper) CD4:CD8 ratios of UTD and αCD19 CAR-T cells measured by flow cytometry. Dots represent the mean values of four donors. Lines connect values from the same donor. (lower) Pie charts of T cell memory phenotypes measured by flow cytometry according to CD45RA and CCR7 expression. Pie sections represent the mean percentage of each population from three donors. Naïve, CD45RA+CCR7+; EMRA, terminally differentiated effector memory cells re-expressing CD45RA, CD45RA+CCR7−; TCM, central memory, CD45RA-CCR7+; TEM, effector memory, CD45RA-CCR7−. (F) The ALL cell line NALM-6 or the mantle cell lymphoma cell line JeKo-1 were mixed with αCD19 CAR T cells with or without stealth technology and cytotoxicity was measured by luciferase expression retaining in live cells (reported as % viability). Dots represent the mean±SEM of two donors from three technical replicates. For all panels: statistical significance was measured by Student’s t-test compared with UTD or control αCD19 CAR, as shown; *p≤0.05, **p≤0.01, ***p≤0.001, ****p≤0.0001.
Stealth αCD19 CAR T cells have similar tumor-clearing efficacy and expansion compared with control αCD19 CAR T cells in vivo. (A, E) NSG mice were engrafted with NALM6 (A) or Jeko-1 (E) tumor cells and left untreated or treated with control or steath αCD19 CAR T cells according to the timeline. (B, F) Tumor burden was measured by in vivo bioluminescence imaging (BLI). All mice (five per group) from the experiment are shown. (C, G) On day 14 after treatment, blood was collected to assess the number of tumor and CAR T cells per mL of blood via flow cytometry. Tumor cells were identified eGFP expression in the CD3- gate. CAR T Cells were detected by eGFP expression in the CD3+gate. (D, H) Survival as measured by Kaplain-Meier curve. Experiments were repeated twice per cell line, using different donors to manufacture the CAR T cells. One representative donor/experiment is shown per tumor cell line.
Stealth-enabled αCD19 CAR-T cells evade CAR-mediated immune recognition by T cells from patients who received a single or second infusion of αCD19 CAR-T cells
Next, we tested the ability of the stealth technology to avoid antigen presentation of immunogenic CAR sequences. We identified 11 patients who had received autologous FMC63-based CARs either once or twice (figure 6A). To assess whether T cells from these patients could be activated by their autologous T cells expressing the FMC63-based αCD19 CAR, we made fresh αCD19 CAR-T cells (without stealth technology) and untransduced (UTD) control cells from T cells in their PBMC that were absent of CAR. These newly produced CAR And UTD T cells were used as “stimulators” and co-cultured with fresh autologous, unactivated T cells as “responders” in an IFNγ ELISpot assay (figure 6B–E). Responder T cells became activated in the presence of FMC63-based αCD19 CAR-T cell products, but not UTD cells or stealth αCD19 CAR-T cells. Activation of responder T cells was particularly high in subjects who had received two infusions of FMC63-based CAR-T cells and in three of the four non-responders. These data suggest that multiple infusions may increase anti-CAR immunity in patients and that a fraction of non-responders may robustly reject their autologous αCD19 CAR-T cells when reinfused. However, larger patient numbers would be required to establish a correlation between a lack of response and CAR-T cell rejection.
Stealth αCD19 chimeric antigen receptor (CAR) cells evade CAR-mediated immune recognition by T cells from patients who received a single or double infusion of αCD19 CAR T cells. (A) Swimmers plot of the analyzed patients showing the αCD19 CAR T cell product they received and their response to treatment. (CR, complete response; PD, progressive disease; PR, partial response). (B) Schematic overviewing the expected outcomes of the ELISPOT assay if the patients have T cell immunity to the CAR following treatment (left) and if the stealth technology suppresses this immunity (right). (C, D) αCD19 CAR T cells with or without stealth technology were made from the T cells of patients who had received one of the FMC63-based αCD19 CAR T cells (Yescarta or Kymriah) products. The αCD19 CAR T cells were then mixed with autologous T cells and anti-CAR T cell responses were measured by IFNγ ELISPOT. (C) Heat map indicates the mean number of IFNγ-producing cells per million T cells from technical triplicates of each donor. Patients are grouped by responders and non-responders who received one CAR T cell infusion and patients who received two CAR T cell infusions. (D) Pictures show representative wells from the ELISPOT for one donor in each group. (E) Histograms depicting eGFP expression levels after sorting and graphs indicating the CAR-mediated T cell activation and anti-CAR responses from the ELISpot assay. Statistical significance was measured by Student’s t-test compared with UTD or control αCD19 CAR, as shown; *p≤0.05, ****p≤0.0001. eGFP, enhanced green fluorescence protein.
Stealth αCD19 CAR-T cells reduce allogeneic responses in vitro and in vivo
Finally, we investigated if stealth αCD19 CAR-T cells could evade targeting from allogeneic T cells. When αCD19 CAR-T cells or stealth αCD19 CAR-T cells were co-incubated with expanded allogeneic T cells (expanded by αCD3/αCD28 beads) in vitro, the stealth technology reduced both IFNγ-secretion and cytotoxicity towards the CAR-T cells (figure 7A). We also implemented a previously reported in vivo mouse model,41 where αCD3/αCD28-expanded allogeneic T cells were injected before NALM6 inoculation and subsequent treatment with CAR-T cells (figure 7B). Stealth CAR-T cells had expanded significantly more in the blood on day 14 (via flow cytometry) compared with αCD19 CAR-T cells without the stealth technology, despite the presence of similar levels of allogeneic T cells and tumor burden (figure 7C,D). However, because our stealth system did not eliminate the large numbers of untransduced, activated T cells, the incidence and severity of xenogeneic graft-versus-host disease (GvHD; as indicated by fur loss and sclerosis) was early and high, resulting in no change in survival (online supplemental figure 5). Therefore, a second allogeneic model was implemented using allogeneic T cells that were primed by pulsing twice with irradiated PBMC originating from the CAR-T-cell donor to boost the allogeneic response. These primed cells were then expanded according to the rapid expansion protocol26 before injecting into the mice (figure 7E). In this model, the stealth αCD19 CAR-T cells robustly expanded over course of 4 weeks compared with control αCD19 CAR-T cells (figure 7F) and had comparable antitumor activity (figure 7G,H). The primed allogeneic T cells allowed for longer monitoring of the mice before the onset of severe xenogeneic GvHD. Importantly, the stealth αCD19 CAR-T cells expanded more robustly than the αCD19 CAR-T cells.
Stealth αCD19 chimeric antigen receptor (CAR) T cells evade in vitro allogeneic response and expand more in an allogeneic model in vivo. (A) Stealth or control αCD19 CAR T cells were mixed with allogeneic Tcells. IFNγ-producing cells were measured via ELISPOT and cytotoxicity was measured by flow cytometry. Bars represent the mean+SEM of two donors and technical duplicates. (B) NSG mice were engrafted with αCD3/αCD28-expanded allogeneic T cells (from normal donor 2, ND2), inoculated with NALM6 tumor cells, and treated with αCD19 CAR T cells (from ND1) with or without stealth technology or left untreated. (C) On day 14 after treatment, blood was collected to assess the number of tumor and CAR T cells per mL of blood via flow cytometry. Tumor cells were identified by eGFP expression in the CD3− gate. CAR T Cells were detected by green fluorescent protein (GFP) expression in the CD3+ gate. (D) Tumor burden was measured via in vivo bioluminescence imaging (BLI). All mice (five per group) from the experiment are shown. The experiments were repeated twice, using different donors to manufacture the CAR T cells. One representative donor/experiment is shown. (E) Allogeneic T cells (from ND2) were pulsed twice with irradiated PBMCs from the CAR T cell donor (ND1) and expanded by REP protocol. NSG mice were injected with the allogeneic T cells, inoculated with NALM6 tumor cells, and treated with αCD19 CAR T cells (from ND1) with or without stealth technology or left untreated. CAR T cell expansion was assessed by weekly blood draws from day 7 to day 28. (F) CAR T cell numbers in the peripheral blood as measured by flow cytometry for GFP+cells. Bars represent the mean+SEM of the five mice per experiment. Dots represent the values of individual mice. (G) Tumor burden was measured by BLI. Spider graph shows the total emission per mouse over time. (H) Survival as measured by Kaplain-Meier curve. Experiments were repeated twice, using different donors to manufacture the CAR T cells. One representative donor/experiment is shown. Statistical significance was measured by Student’s t-test compared with control αCD19 CAR; *p≤0.05, ***p≤0.001.
Discussion
Here, we show that the inclusion of stealth transgenes, EBV TAPi BNLF2a and shRNA targeting CIITA, effectively reduced MHC cell surface molecules to evade autologous and allogeneic T cell responses. We incorporated these stealth transgenes within the CAR-transduction vector to develop a one-shot transduction to produce CAR-T cells with T cell-evasive properties. This simplified approach is particularly valuable as it does not rely on CRISPR/Cas9 gene-editing technique to ablate MHC I/II from the cell surface.4 42 CRISPR/Cas9 can introduce off-target effects through INDELs that promote aberrant mRNA or protein products, which are increased on introducing multiple targets.19 43 Since CRISPR/Cas9 is also being investigated for a variety of other targets in CAR-T cells, such as targets to increase CAR-T cell fitness and persistence,21 42 alternate solutions to reduce HLA from the CAR-T cell surface would enable CRISPR/Cas9 to still be used for these purposes.
Evasion of T cell immunity can be especially valuable in αCD19 CAR-T cell therapy, which efficiently eliminates normal B cells in addition to the intended tumor cells, thereby naturally limiting the humoral immune response to non-self CAR components. Since anti-CAR or donor-specific antibodies and their potential interference with αCD19 CAR-T cell therapy is very limited,3 43 equipping αCD19 CAR-T cells or αCD19 NK cells with a mechanism to prevent T cell immunity could have a major impact. Clinical trials with autologous CAR-T cells have shown that patients treated with CAR-T cells develop a CAR-reactive T cell response.3 6 9 We demonstrate that our stealth CAR-T cells evade anti-CAR responses originating from the FMC63-based αCD19 CAR and obtain increased proliferation in an allogeneic model. Furthermore, an increased CAR-reactive T cell response was found in patients who received multiple FMC63-based αCD19 CAR-T cell infusions. Although CAR-T cells with a non-FMC63 CAR were not assessed as control, this does warrant further investigation of stealth technology in autologous CAR-T cell therapy since the induction of an anti-CAR immune memory could cause CAR-T cells to be rejected by the patient—leading to decreased persistence and therapy failure.
We did not perform an exhaustive comparison of all the ways that can be used to evade immunogenicity. Indeed, CRISPR/Cas9 and TALEN gene knockouts are frequently employed to eliminate the T cell receptor and/or B2M in allogeneic T cell products. It may also be possible to use shRNA to B2M,44 or base-editing technologies to mutate B2M.45 An advantage of our technology is that it could be easily combined with other gene-editing strategies, such as CRISPR/Cas9 while economizing on the number of double-strand breaks or possible translocation events. Furthermore, the incorporation of stealth transgenes into autologous, “simple” lentiviral-transduced autologous products could be implemented quickly, without the need to develop exhaustive sequencing-based strategies to measure off-target gene editing effects or additional release assays. Our technology does not eliminate allogeneic, untransduced T cells (unlike the anti-BB-z CAR by Baylor),41 and thus could not be suitably tested as an immunoevasive strategy in xenograft models. Unfortunately, the models in our field are quite limited, and there are often multidirectional immune reactions occurring when injecting mice with human tumors, untransduced human T cells from one donor (xenogeneic GvHD and allogeneic rejection), and CAR-T cells from a second tumor (xenogeneic GvHD and allogeneic rejection and susceptibility to allogeneic rejection). Thus, we could not extend conclusions from our data as to whether the stealth technology we have developed is sufficient to overcome the high bar of donor acceptance of allogeneic, off-the-shelf cellular products.
Besides the potential of the stealth transgenes in CAR-T cell therapy, this stealth technology may be useful in additional settings that employ gene-modified cells, where either the transgene, junctional sequences, or the cell types are not autologous and where avoidance of early rejection can enhance the desired therapeutic effects.3 8 46
Data availability statement
Data are available on reasonable request. All data relevant to the study are included in the article or uploaded as online supplemental information.
Ethics statements
Patient consent for publication
Ethics approval
All animal experiments performed were approved by the MGH IACUC, protocol #2020N000114. Human T cells were purified from deidentified healthy donor leukapheresis products purchased from the MGH blood bank under an Institutional Review Board (IRB)-exempt protocol. T cells from patients treated with axicabtagene ciloleucel or tisagenlecleucel at MGH were collected on an IRB-approved protocol (16-206) with written informed consent; PMBC from one subject treated at Seattle Cancer Care Alliance with two infusions of autologous FMC63-based CAR-T cells were provided by CJT and collected with written informed consent.
Acknowledgments
We thank the following core facilities of the MGH Cancer Center: MGH Blood Bank and Flow Cytometry core.
References
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Footnotes
X @MJFzeta
Contributors KG, MJF and MVM designed the experiments and KG, MJF, MVM, RL, MBL, and SRB analyzed the data, which were performed by KG, MCK and HS. AAB performed the animal experiments, and DM and KG provided reagents. The manuscript was written by KG, TB, and MVM. MVM serves as the guarantor for this study.
Funding This project was supported by the NIH and the NCI under award numbers R01CA238268 (MVM), T32AI007529 (RL), T32GM007306 (RL) and T32CA009216 (SRB).
Competing interests KG, MJF, and MVM are inventors of patents related to the stealth technologies described here. CJT has received research funding from Juno Therapeutics, BMS, and Nektar Therapeutics and has the right to receive royalties from Fred Hutch as an inventor on patents related to CAR-T cell therapy that are licensed to Juno Therapeutics/BMS. CJT holds equity in Precision Biosciences, Eureka Therapeutics, Caribou Biosciences, Myeloid Therapeutics, and ArsenalBio, serves on Scientific Advisory Boards of Precision Biosciences, Eureka Therapeutics, Caribou Biosciences, T-CURX, Myeloid Therapeutics, ArsenalBio and Century Therapeutics, and serves on ad hoc advisory boards (last 12 months) of Genentech, GlaxoSmithKline and Decheng Capital. MVM is an inventor on patents related to adoptive cell therapies, held by Massachusetts General Hospital and the University of Pennsylvania (some licensed to Novartis). MVM holds equity in TCR2, Century Therapeutics, Genocea, Oncternal, and Neximmune, serves on the Board of Directors of 2Seventy Bio, and has served as a consultant for multiple companies involved in cell therapies. MVM’s interests were reviewed and are managed by Massachusetts General Hospital, and Mass General Brigham in accordance with their conflict-of-interest policies.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.