A BAFF ligand-based CAR-T cell targeting three receptors and multiple B cell cancers

B cell-activating factor (BAFF) binds the three receptors BAFF-R, BCMA, and TACI, predominantly expressed on mature B cells. Almost all B cell cancers are reported to express at least one of these receptors. Here we develop a BAFF ligand-based chimeric antigen receptor (CAR) and generate BAFF CAR-T cells using a non-viral gene delivery method. We show that BAFF CAR-T cells bind specifically to each of the three BAFF receptors and are effective at killing multiple B cell cancers, including mantle cell lymphoma (MCL), multiple myeloma (MM), and acute lymphoblastic leukemia (ALL), in vitro and in vivo using different xenograft models. Co-culture of BAFF CAR-T cells with these tumor cells results in induction of activation marker CD69, degranulation marker CD107a, and multiple proinflammatory cytokines. In summary, we report a ligand-based BAFF CAR-T capable of binding three different receptors, minimizing the potential for antigen escape in the treatment of B cell cancers.

Chimeric antigen receptor T cell (CAR-T) immunotherapy has been a great success in the treatment of liquid cancers, providing rapid and durable responses1,2,3,4,5. However, disease relapse often occurs in these patients. The main reasons for relapse are either CAR-specific antigen loss on cancerous cells or poor performance of CAR-T cells due to exhaustion or decreased persistence6,7,8,9,10,11,12. Antigen escape has been reported not only for CD19-directed CARs, but also for BCMA13, EGFRvIII14, and IL13Rα215, highlighting the drawback of CARs targeting a single tumor-associated antigen. Thus, there is a great need to identify alternative targeting strategies and CAR designs16. Targeting multiple markers on malignant cells is a promising method to combat antigen escape6. Due to the inherent binding nature of naturally occurring ligands, CARs composed of a ligand or receptor ectodomain may have the ability to bind multiple proteins. A small number of ligand- or receptor-based CARs have been tested clinically and have shown encouraging results17,18, presenting an intriguing design approach to combat antigen escape.

Most CAR-T therapies are focused on singular targeting of CD19 or other limited antigens19. Such concentrated effort on a few targets may distract future innovation by diluting resources and participation of patients in clinical trials, thus urgently warranting identification and development of new CAR targets. One attractive target is the set of B-cell activating factor (BAFF) receptors. BAFF ligand is a critical B cell survival factor that binds three receptors: BAFF-R, TACI, and BCMA20. These receptors are expressed by mature B cells and in a wide range of B cell neoplasms21,22,23,24,25. The ability of BAFF to bind to multiple receptors may protect against antigen escape due to the decreased likelihood that cancerous B cells can evade cytotoxic CAR-Ts by downregulating BAFF receptors, since they are important for cell survival. Furthermore, in contrast to approved therapies directed against the pan-B cell marker CD19, CAR-Ts designed to target BAFF receptors may be considered a more selective approach to eliminate malignant B cells, due to their more limited expression during B cell development. Currently, the production of CAR-Ts is primarily achieved using viral transduction of transgenes into primary human T cells26.

Despite the relative success of viral transduction, the manufacturing of human CAR-T products is expensive, complex, and associated with notable safety considerations, supporting the need for alternative methods27,28,29. Since the first human application of transposon-mediated gene therapy almost 10 years ago30, a dozen more clinical trials are underway or have been completed, supporting transposon-based systems as a safe and stable gene transfer alternative with comparable or superior efficiencies to viral transduction31.

Here, we report the development and validation of a BAFF ligand-based CAR-T cell product, which can be successfully generated using the non-viral TcBuster (TcB) transposon system. We show that these BAFF CAR-T cells are both functional and specific in targeting the three BAFF receptors (BAFF-R, BCMA, and TACI) expressed by multiple B cell cancers. We demonstrate robust in vitro and in vivo cytotoxicity exerted by BAFF CAR-T cells against mantle cell lymphoma (MCL), multiple myeloma (MM), and acute lymphoblastic leukemia (ALL) xenograft mouse models.

A BAFF ligand-based CAR-T developed using non-viral gene delivery

We designed the BAFF ligand-based CAR construct using intracellular CD28, OX40/CD134 costimulatory domains, and the CD3ζ signaling domain, as well as the CD28 transmembrane domain (Fig. 1a). To effectively bind the BAFF receptors, we used a truncated BAFF sequence that encompasses the majority of the extracellular domain of the natural human BAFF ligand. We also created a construct that lacks BAFF but is identical to the rest of the BAFF-CAR construct to serve as a negative control (no-BAFF control) (Fig. 1a). Due to the homology between soluble endogenous human BAFF and the extracellular domain of our BAFF-CAR, it is reasonable to predict that BAFF CAR-Ts will bind to the classical BAFF receptors BAFF-R, BCMA, and TACI, thus conferring multi-antigen specificity. Importantly, we also utilize a non-viral genetic engineering approach to stably integrate the BAFF-CAR into T cells using the TcB transposon system. Following co-electroporation of TcBtransposase-encoding mRNA and transposon plasmid into T cells, the transposase enzyme excises the BAFF-CAR “cargo” from the transposon plasmid before integrating it into genomic DNA, from which BAFF-CAR protein is stably expressed (Fig. 1b). For traditional lentiviral transduction, a modified pLVX expression vector was used (Supplementary Fig. 1a). Both lentiviral and TcB-based delivery resulted in stable BAFF-CAR expression, as evidenced by the detection of BAFF ligand on the T cell surface via flow cytometry (Fig. 1c). Surface BAFF expression directly correlates with GFP expression in the lentiviral transduced T cells, as the pLVX vector encodes GFP downstream of the same promoter as the BAFF-CAR construct. The CD4/CD8 ratios of donor T cells were also characterized (Fig. 1c). T cells transduced with the no-BAFF control construct express GFP, but not surface BAFF (Supplementary Fig. 1b). BAFF CAR-T cells produced using the TcB transposon method were also assessed for stable expression of BAFF-CAR. After 7 days of expansion, BAFF CAR-T cells produced from different donors were frozen, then thawed and cultured for 21 days. CAR-T cells at 21 days post-thaw expressed a similar amount of cell surface BAFF as those 5 days post-thaw (Supplementary Fig. 1c).

a Schematic of the BAFF-CAR construct, which consists of extracellular BAFF ligand, short spacer, hinge from human IgG1, CD28 transmembrane and signaling domains, OX40, and CD3ζ. A construct lacking extracellular BAFF (no-BAFF) was used as a negative control. b Schematic of the non-viral TcBuster (TcB) transposon system, which enables stable expression of a CAR protein. 1. TcBtransposase mRNA and transposon plasmid are introduced into the cell. 2. Protein from TcB mRNA is produced. 3. Transposase cuts the cargo from the transposon plasmid. 4. TcB transposase pastes the transposon cargo into the genomic DNA. 5. Cargo mRNA is stably expressed from genomic DNA. c The BAFF-CAR construct was expressed in primary human T cells via either lentiviral transduction or the TcB transposase system. 5 days after lentiviral transduction, efficiency of transduction was measured based on % GFP expression and correlated exogenous surface expression of BAFF. Efficiency of TcB transposase-mediated expression was measured based solely on % exogenous BAFF expression. CD4+ and CD8+ T cell percentages were also measured. The experiment was repeated with four different T cell donors. d CAR-T cells, no-BAFF control T cells, or unmodified T cells were co-cultured with fluorescently-labeled Jeko-1 MCL cells at 5:1 effector:target (E:T) ratio for 16 h, followed by flow cytometry. Cytotoxicity was measured via propidium iodide (PI) staining and gating on labeled target cells. ns not significant, ***P < 0.001. P = 0.0002 for Unmodified Control vs CAR-T, P = 0.0003 for no-BAFF Control vs CAR-T. Graphs display mean ± SD, n = 3 biologically independent co-cultures, one-way ANOVA with Tukey’s multiple comparisons test. The experiment was repeated with two different T cell donors. Source data are provided as a Source Data file.

Full size image

Next, we conducted preliminary cytotoxicity experiments against the Jeko-1 MCL cell line to confirm BAFF-CAR function. T cells were co-cultured with fluorescently-labeled Jeko-1 cells at a 5:1 effector:target (E:T) ratio for 16 h, then stained with propidium iodide (PI) and gated on the cancer cells to determine cancer cell death. BAFF CAR-T cells exhibited significantly higher cytotoxicity than both unmodified and no-BAFF control T cells, while no-BAFF control T cell cytotoxicity was not significantly different from that of unmodified T cells (Fig. 1d). This confirms that our BAFF-CAR construct is functional and that the presence of BAFF at the N-terminus is necessary for cytotoxic function.

Characterization of BAFF-CAR gene integration into T cells using a non-viral platform

After characterizing the functionality of the CAR-T cells, we wanted to assess the location of transposon integration as well as the presence of any clonal outgrowth of TcB mediated transposon integration events. Historical data of viral integration32,33, as well as 2 other transposase systems34, were compared to transposon integration using TcB (Fig. 2a). After analysis of all integrants, the results indicated that TcB inserted the transposon into transcript-coding DNA regions less frequently than lentivirus (Fig. 2a and Supplementary Table 1). When assessing the distance between integration events and transcriptional start sites, transposon-based methods of integration, including TcB, integrated further away from transcriptional start sites when compared to lentivirus (Fig. 2b). The cell products generated in this study were compared to available lentivirus data and similarly did not exhibit significant clonal outgrowth (Table 1).

a Analysis of the frequency of integration events in exon-coding transcript sites, non-exon-coding transcript sites, and non-transcript sites. Samples were collected from primary human T cells transfected with TcBuster (TcB) or historical data sets generated by Wang et al., Brady et al., and Gogol-Döring et al.32,33,34. Random in silico control sets were also generated and analyzed. SB Sleeping Beauty, LV lentiviral dataset. Graph displays mean ± SD, biologically independent samples. n = 8 for TcB-M, n = 9 for SB, n = 9 for PiggyBac, n = 2 for LV (Wang), n = 4 for LV (Brady), n = 8 for Random. b Median distance measured between transposon integration sites and the transcriptional start site of the nearest gene is presented for transposon and lentiviral integration. SB Sleeping Beauty, LV lentiviral dataset. Graph displays mean ± SD, biologically independent samples. n = 8 for TcB-M, n = 9 for SB, n = 9 for PiggyBac, n = 2 for LV (Wang), n = 4 for LV (Brady), n = 8 for Random. Source data for all graphs are provided as a Source Data file. Sequencing data used to perform integration site analysis are available online (Accession ID: PRJNA779430).

Full size image

As vector copy number (VCN) is an important consideration for any future regulatory filing, we evaluated transposon copy number via droplet digital PCR (ddPCR). Seven independent T cell donors were engineered to express the BAFF CAR with TcB. Genomic DNA was harvested and samples were analyzed. No donor achieved a copy number of >10 copies per CAR+ cell (Table 2).

Characterization of BAFF CAR-T cells

We next characterized the functionality and specificity of BAFF CAR-T cells. Since soluble BAFF is normally present in human circulation and may be elevated in patients with B cell cancers24,35, we examined whether it would affect BAFF CAR-T cytotoxicity by competing for receptor binding. Furthermore, soluble forms of all three BAFF receptors have been identified in the blood of healthy adults36,37. These levels can change in correlation with autoimmune diseases and cancer, such as chronic lymphocytic leukemia (CLL)36. Of the three BAFF receptors, levels of sBAFF-R are highest in healthy adults, though BAFF-R has not been shown to function as a decoy receptor, in part because it is not believed to be cleaved from the membrane unless it is ligand-bound36,38. Given the presence of soluble BAFF ligand and receptors in the blood, and given the potential for sTACI and sBCMA to act as decoy receptors, we performed a spike-in cell killing assay to examine whether these soluble recombinant proteins would inhibit BAFF CAR-T-mediated killing. Physiologically relevant concentrations of each soluble protein were selected, as indicated in the caption of Fig. 3a36,37. Results show no diminution of BAFF CAR T cell-mediated cytotoxicity in the presence of any of these soluble BAFF or BAFF receptor family proteins, at any time point or effector:target ratio examined, relative to donor-matched control T cells (Fig. 3a and Supplementary Fig. 2a). We further tested the cytotoxicity of BAFF CAR-T cells in the presence of recombinant BAFF at both pathophysiological (5 ng/mL) and non-physiological, very high (100 ng/mL) concentrations, and we did not observe any changes in cytotoxicity against Jeko-1 cells under either co-culture condition (Supplementary Fig. 2b).

a CAR-T cells or unmodified T cells were co-cultured with luciferase-expressing Jeko-1 cells at 3:1 E:T ratio for 24 h with soluble recombinant BAFF/BAFF receptors (sBAFF = 1 ng/mL; sBAFF-R = 500 ng/mL; sBCMA = 2.5 ng/mL; TACI = 20 pg/mL). Vehicle served as a negative control. Luminescence was measured using a plate reader. Viability was calculated using target-only control, then normalized relative to average luminescence from target cells co-cultured with unmodified T cells. ****P < 0.0001. Mean ± SD, n = 3 biologically independent co-cultures, two-way ANOVA with Šídák’s multiple comparison test. The experiment was repeated with three different T cell donors. b CAR-T cells or unmodified T cells were co-cultured with HEK293T cells expressing BAFF-R, BCMA, or TACI at 5:1 E:T ratio for 16 h. Cytotoxicity was measured via PI staining and gating on labeled target cells. ns not significant, ****P < 0.0001. Mean ± SD, n = 3 biologically independent co-cultures, two-way ANOVA with Šídák’s multiple comparisons test. The experiment was repeated with two different T cell donors. c BAFF CAR-T cells or unmodified T cells were co-cultured with different types of primary human cells at 1:1 E:T ratio for 24 h to evaluate the potential for adverse toxicity. Lactate dehydrogenase (LDH) release upon target cell lysis was measured, then compared with LDH release in lysis controls (0.8% Triton X-100) to calculate cytotoxicity. Percent cytotoxicity = [Experimental LDH release – Effector (spontaneous LDH release) – Target (spontaneous LDH release)]/[Target (maximum LDH release) – Target (spontaneous LDH release)] × 100. CNS central nervous system, PNS peripheral nervous system. ns not significant, *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. The following p-values correspond to Unmodified Control vs CAR-T: P < 0.0001 for airway epithelial cells, P = 0.0003 for aortic smooth muscle cells, P = 0.0377 for cardiac myocytes, P = 0.0027 for CNS neurons, P < 0.0001 for hepatocytes, P = 0.0401 for ovarian surface epithelial cells, P = 0.0094 for PNS neurons, P < 0.0001 for renal epithelial cells, P < 0.0001 for Sertoli cells, P = not significant for trophoblasts. Mean ± SD, n = 8 biologically independent co-cultures, two-way ANOVA with Šídák’s multiple comparisons test. The experiment was repeated with two different T cell donors. Source data for all graphs are provided as a Source Data file.

Full size image

Along with testing the functionality of the BAFF-CAR construct, we also wanted to characterize its specificity. To do this, we used HEK293T cells, which normally do not express any of the BAFF receptors, and engineered them to exogenously express either of the three BAFF receptors (BAFF-R, TACI, or BCMA), verified via flow cytometry (Supplementary Fig. 2c). BAFF CAR-T or unmodified T cells were co-cultured with these modified or parental HEK293T cells for 16 h at 5:1 E:T ratio. We observed that BAFF CAR-T cells exhibited significantly increased cytotoxicity against all three BAFF receptor-expressing HEK293T cell lines but not against the parental cells, compared to unmodified T cells (Fig. 3b).

Finally, we tested whether BAFF CAR-T cells display any adverse toxicity towards different normal cell types from a number of organ systems. BAFF CAR-T cells or unmodified control T cells were co-cultured with different isolated primary human cell types, including airway epithelial cells, aortic smooth muscle cells, cardiac myocytes, central and peripheral nervous system neurons, hepatocytes, renal epithelial cells, ovarian surface epithelial cells, Sertoli cells, and placental trophoblasts. Adverse toxicity was benchmarked against cytotoxicity against Jeko-1 cells, which was repeated for each type of primary cells. We did not observe any remarkable adverse toxicity against these cell types when compared to unmodified T cells (Fig. 3c), further highlighting the clinical potential of BAFF CAR-T cells.

In vitro cytotoxicity of BAFF CAR-T cells against multiple B cell cancers

Next, we comprehensively tested the in vitro cytotoxicity of the BAFF CAR-T cells against several different cell lines that represent different B cell malignancies: Jeko-1 (MCL), rs4;11 (ALL), RPMI-8226 (MM), U266 (MM), and MM.1s (MM) cell lines. These cell lines express one or more of the three BAFF receptors, as evident from flow cytometry analysis (Fig. 4a). Jeko-1 cells are known to express high levels of BAFF-R and low levels of BCMA, whereas the opposite is true for the MM cell lines (Fig. 4a). Both the MCL and MM cell lines express TACI. The rs4;11 cells express relatively high BAFF-R, a moderate amount of TACI, and virtually no BCMA. The MM cells stained negative for CD19 and thus serve as a CD19-negative cancer model for BAFF CAR-T testing.

a Cell lines of different B cell malignancies were stained for BAFF-R, BCMA, TACI, and CD19 expression. Red = Jeko-1 (MCL); blue = rs4;11 (ALL); cyan = RPMI-8226 (MM); orange = U266 (MM); purple = MM.1s (MM); black = unstained cells. b CAR-T cells or unmodified T cells were co-cultured with fluorescently-labeled cancer cells at indicated E:T ratios for 16 h (Jeko-1, rs4;11, MM.1s) or 40 h (RPMI-8226, U266). Cytotoxicity was measured via PI staining and gating on labeled target cells. Circle = Unmodified Control, triangle = CAR-T. **P < 0.01, ***P < 0.001, ****P < 0.0001. Jeko-1: P = 3.34e-4 (10:1), P = 1.92e-4 (5:1), P = 6.72e-4 (1:1). rs4;11: P = 9e-6 (10:1), P = 3e-6 (5:1), P = 8.8e-5 (1:1). RPMI-8226: P = 2.1e-5 (1:5), P = 2.85e-4 (1:10), P = 2.919e-3 (1:20). U266: P = 7.2e-5 (1:5), P = 3e-6 (1:10), P = 2.179e-3 (1:20). MM.1s: P = 6.6e-5 (5:1), P = 3e-6 (3:1), P = 1.29e-4 (1:1). Mean ± SD, n = 3 biologically independent co-cultures, multiple unpaired two-tailed t-tests with Holm-Šídák correction for multiple comparisons. c CRISPR was used to knock out expression of either BAFF-R, TACI, or both BAFF-R and TACI in Jeko-1 cells. Fluorescently-labeled cancer cells were co-cultured with BAFF CAR-T cells or unmodified T cells 16 h at 3:1 E:T ratio. Parental Jeko-1 cells served as negative control. Cytotoxicity was measured as above. ns not significant, ****P < 0.0001. Mean ± SD, n = 3 biologically independent co-cultures, two-way ANOVA with Tukey’s multiple comparisons test. d CRISPR was used to knock out expression of either BCMA, TACI, or both BCMA and TACI in RPMI-8226 cells. Fluorescently-labeled cancer cells were co-cultured with BAFF CAR-T cells or unmodified T cells 36 h at 1:10 E:T ratio. Parental RPMI-8226 cells served as negative control. Cytotoxicity was measured as above. ns not significant, *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. P = 0.0015 (Parental vs TACI KO), P < 0.0001 (Parental vs BCMA/TACI Dual KO), P = 0.0006 (BCMA KO vs TACI KO), P < 0.0001 (BCMA KO vs BCMA/TACI Dual KO), P = 0.0137 (TACI KO vs BCMA/TACI Dual KO). Mean ± SD, n = 3 biologically independent co-cultures, two-way ANOVA with Tukey’s multiple comparisons test. Source data for all graphs are provided as a Source Data file. All cytotoxicity experiments were repeated with two different T cell donors.

Full size image

To measure cytotoxicity, BAFF CAR-T cells were co-cultured with fluorescently-labeled cancer cells at different E:T ratios for either 16 h (Jeko-1, rs4;11, and MM.1s) or 40 h (RPMI-8226 and U266), followed by PI staining to measure cell death via flow cytometry. BAFF CAR-T cells exhibited significantly enhanced cytotoxicity against all of these B cell malignancies compared to unmodified T cells (Fig. 4b). This was true for all tested cell lines and at all tested E:T ratios. At a 5:1 E:T ratio, Jeko-1 cells (69%) and rs4;11 cells (57%) displayed significant cell death mediated by BAFF CAR-T cells. At a 3:1 E:T ratio, BAFF CAR-T cells displayed the greatest increase in cytotoxicity against MM.1s cells (63%). Similar enhanced cytotoxicity was observed at a 1:5 E:T ratio against RPMI-8226 cells (57%) and U266 cells (37%) (Fig. 4b). At these same ratios, unmodified T cells displayed much lower cytotoxicity against target cell lines. Thus, we have shown that BAFF CAR-T cells produced using non-viral genetic engineering are cytotoxic against multiple B cell cancers.

To confirm the involvement of BAFF/BAFF receptor binding in mediating cytotoxicity and also to mimic the antigen escape scenario in patients, we generated Jeko-1 cells in which either BAFF-R, TACI, or both BAFF-R and TACI receptor expression were knocked out using CRISPR/Cas9. Next, we measured the cytotoxicity of BAFF CAR-T cells against these cells. Jeko-1 cells express high levels of BAFF-R and moderate levels of TACI, although they express virtually no BCMA on the surface (Fig. 4a). Thus, by knocking out both BAFF-R and TACI, we generated Jeko-1 cells with no detectable BAFF receptor expression, serving as a tool to confirm the specificity of the BAFF-CAR construct. Successful knockout cells were verified and separated from untransfected cells via fluorescence-activated cell sorting (FACS) (Supplementary Fig. 3a). As expected, BAFF CAR-T cells displayed significantly reduced cytotoxicity towards Jeko-1 cells with dual BAFF-R/TACI receptor knockout, compared to parental Jeko-1 cells (Fig. 4c). Interestingly, BAFF CAR-T cells were able to kill TACI or BAFF-R single knockout cells effectively, showing that either TACI or BAFF-R expression is enough for BAFF-CAR-mediated killing. Dual BAFF-R/TACI receptor knockout caused a sharp decrease in cytotoxicity compared to TACI knockout alone. This shows that the targeting of both receptors contributes significantly to BAFF-CAR functionality, but Jeko-1 cell expression of TACI alone does not impact BAFF CAR-T activity. Furthermore, loss of either BAFF-R or TACI alone still leaves BAFF CAR-T cells with significant cytotoxic capability.

Similarly, we used CRISPR/Cas9 to knock out either BCMA, TACI, or both BCMA and TACI in RPMI-8226 cells since they express high levels of both BCMA and TACI but no detectable levels BAFF-R. Knockout cells were verified and separated from untransfected cells via FACS (Supplementary Fig. 3b). BAFF CAR-T cells exhibited significantly decreased cytotoxicity against dual BCMA/TACI knockout cells (Fig. 4d). Single TACI knockout RPMI-8226 cells also showed some inhibition in cytotoxicity, while BCMA knockout cells behaved similar to parental cells. Thus, we have shown that we are able to successfully produce BAFF CAR-T cells that are both functional and specific against target cells that express any of the three BAFF receptors, even when expression of one of the natural receptors is abrogated experimentally.

BAFF CAR-T cell activation after co-culture with cancer cells

We next characterized T cell activation by staining for CD69, a T cell activation marker that is expressed on the surface following antigen engagement. BAFF CAR-T cells displayed much higher levels of CD69 expression than unmodified cells when co-cultured with all target cell lines (Fig. 5a). This can be compared to CD69 expression at baseline, without the presence of target cells, in which a much smaller proportion of BAFF CAR-T cells express surface CD69 (~16%), albeit still somewhat higher than with unmodified T cells (~4%) (Supplementary Fig. 4a). We also stained for CD107a, a degranulation marker expressed on the surface of cytotoxic immune cells during the release of lytic enzymes such as perforin and granzyme B. BAFF CAR-T cells exhibited significantly higher levels of CD107a, whereas unmodified cells displayed negligible levels of CD107a on their surface (Fig. 5b). Without the presence of target cells, neither BAFF CAR-T cells nor unmodified T cells exhibit appreciable surface expression of CD107a (Supplementary Fig. 4b).

Next, we used a multiplex, bead-based flow cytometry assay to measure the release of various pro-inflammatory cytokines and lytic enzymes from T cells after co-culture with target cells. Against all tested cancer cell lines, the BAFF CAR-T cells released significantly higher levels of the pro-inflammatory cytokines TNF-α and IFN-γ, lytic enzymes granzymes A and B and perforin, soluble Fas ligand (sFasL), and granulysin (Fig. 5c). In order to measure the baseline release of these cytokines and lytic enzymes, we co-cultured BAFF CAR-T cells and unmodified control T cells with and without Jeko-1 cells. The multiplex assay revealed that BAFF CAR-T cells do not express significantly different levels of TNF-α, IFN-γ, sFasL, or granzyme A compared to control T cells at baseline, but when co-cultured with Jeko-1 cells, the release of these cytokines increased dramatically for BAFF CAR-T cells (Supplementary Fig. 4c). To measure IL-2 secretion from these T cells, we co-cultured BAFF CAR-T cells and unmodified control T cells with Jeko-1 cells in IL-2-free media, and we observed that BAFF CAR-T cells secrete significantly higher levels of IL-2 compared to control (Supplementary Fig. 4d).

In vivo cytotoxicity of BAFF CAR-T cells against MCL liquid cancer

We analyzed the effect of BAFF CAR-T against a liquid MCL xenograft model by using Jeko-1 cells that were modified to express the firefly luciferase gene (Jeko-1-luc). Male or female NSG mice were injected intravenously (i.v.) with Jeko-1-luc cells (1.5e6) and on Day 9 post-tumor inoculation, mice were injected i.v. with BAFF CAR-T (10e6) or unmodified Control-T cells (10e6) (Supplementary Fig. 5a). Bioluminescent imaging was conducted weekly until Day 36 (Fig. 6a and Supplementary Fig. 5b). Mice treated with Control-T cells were either dead or sick due to extensive tumor burden by Day 36, while BAFF CAR-T-treated mice were all alive and did not show any disease symptoms (Fig. 6a). Bioluminescence signal intensity for BAFF CAR-T-treated mice remained significantly lower compared to control mice over the course of the experiment (Supplementary Fig. 5b). We observed these mice until Day 60 and a survival curve was plotted. BAFF CAR-T-treated mice displayed significantly prolonged survival in contrast to the Control-T-treated mice (Fig. 6b). We wanted to gauge T cell persistence in these mice. Therefore, we collected peripheral blood from several mice in each treatment group on Days 23, 37, 51, and 58 post-innoculation to quantify the number of human CD3+/CD45+ T cells in circulation via flow cytometry (Fig. 6c). After gating on the T cells, we also measured the percentage of cytotoxic CD8+ T cells (Fig. 6d).

There were 2 mice in the CAR-T treatment groups that relapsed, and we wanted to determine whether their relapsed disease was due to antigen loss of any of the BAFF receptors from the surface of the Jeko-1 cells. We processed spleen and bone marrow samples of these mice at the time of euthanasia. Using flow cytometry and gating on human CD19, we did not observe any significant change in the receptor expression profile of Jeko-1 cells from the spleen of 2 relapsed CAR-T-treated mice, as compared to Jeko-1 cells from Control-T-treated mice (Fig. 6e). In addition to assessing the possibility of antigen loss, at the experimental endpoint, spleen and bone marrow samples were collected from BAFF CAR-T-treated mice to compare the percentage of Jeko-1 cells and human T cells in these compartments with previously collected samples from Control-T-treated mice. Representative samples from BAFF CAR-T-treated mice show virtually undetectable Jeko-1 cells in both spleen and bone marrow, as determined by staining for human CD19, whereas the cell population of the spleen and bone marrow of Control-T-treated mice are dominated by Jeko-1 cells (Supplementary Fig. 6). Analogously, BAFF CAR-T-treated mice show high percentages of T cells in the spleen and some bone marrow samples, while Control-T-treated mice tended to have low or undetectable amounts of T cells.

We conducted another experiment using the same i.v. Jeko-1 xenograft model to determine whether a lower dose of CAR-T cells would exhibit similar efficacy. Male or female NSG mice were injected i.v. with Jeko-1-luc cells (1.5e6) and after 1 week, mice were injected i.v. with BAFF CAR-T cells (3e6) or Control-T cells (3e6) or PBS (Supplementary Fig. 7a). Bioluminescent imaging was conducted weekly until Day 70. The PBS and Control-T treatment group demonstrated aggressive progression of the disease; female mice were killed by Day 28 and most male mice by Day 35 (Supplementary Fig. 7b). However, mice that received the BAFF CAR-T cells showed significantly reduced tumor burden up to Day 70. BAFF CAR-T-treated mice displayed significantly prolonged survival in contrast to the PBS- and Control-T-treated mice (Supplementary Fig. 7c). As expected, the BAFF CAR-T-treated mice had a lower average spleen weight compared to mice treated with PBS or Control-T cells (Supplementary Fig. 7d). Furthermore, BAFF CAR-T-treated mice maintained a relatively steady and healthy weight (Supplementary Fig. 7e), and they did not display any obvious signs of distress or morbidity. We also included a group of mice without tumor inoculation and only injected with BAFF CAR-T cells in order to assess BAFF CAR-T related toxicity. NSG mice were injected i.v. with BAFF CAR-T cells (3e6), and during the 70-day period, they did not exhibit any signs of morbidity from the BAFF CAR-T cells alone, nor did they experience any concerning changes in body weight (Supplementary Fig. 7f). After 70 days, we assessed serum markers for liver toxicity and also collected organs including kidney, liver, lung, and spleen for histological analysis. We measured common liver injury markers alanine aminotransferase (ALT) and aspartate aminotransferase (AST) activity in serum from healthy NSG mice and BAFF CAR-T cell injected mice. We observed no significant differences in ALT or AST activity between the two cohorts (Supplementary Fig. 7g). There were no significant lesions, tissue damage, or notable changes in tissue architecture in BAFF CAR-T injected mice (Supplementary Fig. 8).

In vivo cytotoxicity of BAFF CAR-T cells against MCL solid, MM liquid, and ALL liquid cancers

We next wanted to assess the in vivo efficacy of BAFF CAR-T cells using B cell cancer xenograft models. We used a solid MCL model where Jeko-1 cells were injected subcutaneously (s.c.) into immunodeficient NSG mice. After visible tumor development, mice were treated with a single intratumoral injection of either BAFF CAR-T cells, unmodified Control-T cells, or PBS alone (Supplementary Fig. 9a). We observed that while tumor volume continued to increase in mice treated with PBS or Control-T cells, tumor volume decreased rapidly after treatment with BAFF CAR-T cells (Fig. 7a). Mice treated with BAFF CAR-T cells also experienced significantly prolonged survival compared to controls (Fig. 7b).

Next, we tested BAFF CAR-T efficacy using a bioluminescent MM.1s myeloma cell line (MM.1s-luc) to generate an intravenous MM xenograft model. MM.1s-luc cells (5e6) were injected i.v. into NSG mice via tail vein, followed by treatment with CAR-T cells (2e6) or Control-T cells (2e6) 8 days and 16 days later (Fig. 7c and Supplementary Fig. 9b). Imaging was performed every 3–5 days until Day 29, when tumor burden was clearly visible in the majority of Control-T-treated mice (Fig. 7c). Bioluminescence intensity was quantified and plotted, showing a clear and significant inhibition of tumor burden in BAFF CAR-T-treated mice compared to Control-T-treated mice (Fig. 7d). BAFF CAR-T-treated mice experienced much slower tumor progression than Control-T-treated mice.

We next tested the BAFF CAR-T cells against two different ALL xenograft models. In one model, rs4;11 cells were injected i.v. into NSG mice, followed by infusion of CAR-T cells or PBS in control mice 6 days later (Supplementary Fig. 9c). Mice were killed 45 days post-tumor inoculation and CD19+ tumor load in blood, spleen, and bone marrow was quantified using flow cytometry. Mice treated with BAFF CAR-T cells showed significantly lower percentages of CD19+ rs4;11 cells (Fig. 7e) compared to control mice. Additionally, the average weight of harvested spleens from these mice was significantly lower, compared to mice treated with PBS alone (Fig. 7f). In addition, we developed a patient-derived xenograft (PDX) model using ALL blasts collected from a drug-resistant ALL patient (Pt2 cells). Pt2 cells were injected i.v. into NSG mice and then treated 6 days later (Supplementary Fig. 9d). Mice were killed 31 days post-tumor inoculation. Similar to the results from the rs4;11 model, mice treated with BAFF CAR-T cells had almost negligible counts of CD19 + Pt2 cells remaining compared to mice treated with PBS (Fig. 7g). CAR-T-treated mice also had significantly lower average spleen weight compared to mice treated with PBS alone (Fig. 7h).

Of the CAR-T cell therapies in clinical development, a third still target CD19, the same target as the currently approved CD19-CAR T therapies19. Beyond CD19, 70% of these clinically investigated CAR-T therapies are directed against just 10 antigens, reflecting a significant duplication of targets19. We have developed a BAFF ligand-based CAR construct and have successfully delivered it into T cells using a non-viral transposon system. BAFF CAR-T targets BAFF-R, BCMA, and TACI expressed on surface of almost all B cell cancers. CD19 CAR-T cell therapy adverse effects include severe B cell aplasia, due to pan-B cell expression of the CD19 marker. Unlike CD19 CAR-T cells, BAFF CAR-T cells designed to target BAFF receptors may produce less severe B cell aplasia, as the BAFF receptors are not expressed by early-stage B cells. Unaffected early B cells could easily replenish the mature B cell population.

Experiments using HEK293T cells as target cells confirmed that the BAFF CAR-T induced cytotoxicity is indeed mediated by BAFF/BAFF receptors interaction. The parental HEK293T cell line does not express any of the BAFF receptors, and as expected, there was no cytotoxicity by BAFF CAR-T cells. Once we transduced these HEK293T cells to overexpress each of the BAFF receptors, these cells were efficiently killed by BAFF CAR-T cells. These data show that the BAFF-CAR construct specifically targets three BAFF receptors and even expression of one of the receptors by target cells is enough to be killed by BAFF CAR-T. This also supports our hypothesis that treatment failure in patients as a result of antigen escape is expected to be much less likely in response to BAFF CAR-T therapy. Another level of confirmation came from the double knockout of BAFF-R and TACI in Jeko-1 cells. Jeko-1 parental cells express both BAFF-R and TACI, and knocking out both of these receptors using CRISPR significantly reduced BAFF CAR-T cell-mediated cytotoxicity. Similarly, double knockout of BCMA and TACI in RPMI-8226 cells also greatly reduced BAFF CAR-T cell-mediated killing. There is some residual cytotoxicity observed in double knockout RPMI-8226 cells and we attribute this to the low level of BAFF-R expression seen in these cells. Either one of the three BAFF receptor’s expression by cancer cells was sufficient to evoke a cytotoxic response by BAFF CAR-T cells. Some primary cancer cells are reported to express all three BAFF receptors5,39,40, and in those rare cases, BAFF CAR-T will be able to target all three receptors. Most of the B cell cancers express at least two of the three BAFF receptors41,42,43,44, and in those cases it will be dual-targeting. Even if one receptor is lost, the other one is sufficient to evoke a cytotoxic response by BAFF CAR-T cells.

We also observed various degrees of cytotoxicity at the tested E:T ratios against the different cancer cell lines, which express varying levels of the three BAFF receptors. Understanding the binding affinities of BAFF to each of its receptors can also help in estimating the binding affinity of BAFF-CAR towards these endogenous receptors, though the fusion of the CAR backbone may modify it to a certain extent. BAFF has been reported to bind its receptors with varying affinity, with highest affinity towards BAFF-R, followed by TACI and then BCMA35. Therefore, even as our multi-antigen targeting BAFF-CAR-T cells may benefit from the stronger binding interactions of BAFF to BAFF-R, they also take advantage of the slightly lower affinity interactions between BAFF and TACI or BCMA, which may confer enhanced efficacy and persistence against a number of different B cell cancers that simultaneously express two or more of these BAFF receptors41,42,43,44.

We have demonstrated the cytotoxicity of BAFF CAR-T cells against MCL, ALL, and MM cells in vitro and in vivo using xenograft models. Each of these in vivo models was done using BAFF CAR-T cells generated from T cells isolated from different donors, thus validating the efficacy of BAFF CAR-T cells generated from multiple donors. It is highly likely that BAFF CAR-T cytotoxicity is not limited to only these cancer cells and that it will serve as a potential therapy for other B cell cancers as well, including those that are not effective using CD19 CAR-T therapy. MM cells do not express CD19, but they express TACI and BCMA; this is a perfect example to show that BAFF CAR-T cells can display significant cytotoxicity against cancer cells that do not express CD19. Though we noticed some minor increases in LDH release after co-culturing BAFF CAR-T cells with normal cells, there were no signs of toxicity noticed in vivo even after 70 days of injecting BAFF CAR-T cells, as evident from body weight, AST/ALT levels, and histology analysis of different organs.

Biologics targeting BAFF/BAFF-R signaling have also been effective for the treatment of autoimmune diseases45,46, but are still being investigated in clinical trials for cancer (for example, NCT03400176). To our knowledge, using the BAFF ligand as a ligand-based CAR as we describe here has not been previously reported. Relative to a scFv-based BAFF-Receptor CAR47, which targets only BAFF-R, our preliminary data supports that our non-viral engineered BAFF-CAR T cells possess the added benefits of multi-marker targeting and decreased manufacture complexity. Beyond these BAFF-R-CARs, related ligand-based CAR approaches have been reported utilizing a “proliferation-inducing ligand” (APRIL) as an alternative means to BCMA-targeting CAR-T cells for the treatment of MM, which has traditionally been plagued by antigen escape and low persistence48. One of the limitations of our study is that we did not perform a comparison study using our ligand-based BAFF CAR-T v/s existing APRIL CAR-T or scFv CARs such as BCMA CAR-T and BAFF-R CAR-T cells. Hence, we are unable to comment on an efficacy comparison of these CAR-T cells in killing B cell cancers.

In the literature, for some context, there is evidence that the presence of soluble target antigens may reduce CAR efficacy49,50,51 and that this may be especially true for sTACI and sBCMA, which can act as decoy receptors for BAFF52. Here, our data shows that this is not a concern, as soluble BAFF or BAFF receptors have no effect on BAFF CAR-T cell cytotoxic activity. In our in vivo experiments, MCL cells relapsing after BAFF CAR-T treatment is not due to antigen escape as the MCL cells still express all three BAFF receptors. We hypothesize that it may be due to a decreased population of CAR-T cells in these mice. We did not observe any gender-specific influence on BAFF CAR-T efficacy as both male and female mice treated with BAFF CAR-T showed significantly prolonged survival.

CAR-T cell production using viral transduction is a very common procedure, but there are many disadvantages, including its labor-intensive production and expensive clinical applications. Another serious drawback of using viral vectors is lingering uncertainty over their safety profile. Although viruses have been genetically engineered to reduce their safety risks, there are still concerns of insertional mutagenesis and variable transgene expression. Our analysis of the TcB transposon system yielded noteworthy comparisons to both lentiviral vectors and other transposon systems. Overall, TcB tended to insert transposons into non-transcript regions much more frequently than lentivirus but somewhat less frequently than the PiggyBac (PB) or Sleeping Beauty (SB) transposon systems. TcB generally inserted transposons into exonic regions less frequently than lentivirus but somewhat more frequently than PB and SB. When comparing TcB, PB and SB, SB had the highest number of insertions outside of transcribed regions. Finally, when assessing the location of the insertions, it was found that TcB inserted DNA farther than transcriptional start sites when compared to lentivirus, but closer to transcriptional start sites when compared to PB and SB. Additionally, BAFF CAR-T cells created with TcB exhibited lower clonal outgrowth than CAR-T cells created using lentiviral, PB, or SB methods. Thus, TcB appears to exhibit an insertion profile that is comparable to existing transposon systems and is potentially advantageous over lentiviral transduction.

Overall, we have demonstrated the effectiveness of a BAFF ligand-based CAR-T cells, generated using non-viral methods, in killing ALL, MCL, and MM cells in vitro and in vivo. We believe these BAFF CAR-T cells have the potential to enter into clinical trials soon for the treatment of multiple B cell cancers.