We elected to employ a proven clinical development strategy already in use for CAR T-cell production for patients at City of Hope [11,12,13]. To create the clinical-grade vector BAFF-R:4-1BB:ζ/EGFRt, the BAFF-R-targeting single-chain variable fragment (scFv) [10] was cloned into a second-generation pHIV7 clinical lentiviral vector backbone (Fig. 1a), containing the 4-1BB and CD3ζ motifs, a mutant human IgG4 Fc hinge and CD3 extracellular motif and a truncated EGFR (EGFRt) extracellular motif (see Supplementary Table 1). The latter replaces the GFP tracker from the prototype vector (BAFF-R:4-1BB:ζ/GFP in a pLenti7.3/v5-DEST lentiviral vector backbone), and can be used as a suicide switch to mitigate cytokine release syndrome (CRS) caused by over-activated CAR T cells [14]. Following the research-grade CAR production protocol (Fig. 1b), the prototype and clinic-ready (clinical vector used in research-grade production) BAFF-R CAR T cells were produced as previously described [15] for a head-to-head in vitro and in vivo comparison to verify that CAR T cells produced using the two vectors were equivalent. The research-grade production run yielded ≥ 90% enriched naïve T cell (TN)-derived prototype or clinic-ready CAR T cells, measured by FACS analysis of CD3 and tracker (GFP or EGFRt), respectively, and equivalent expansion rates were observed (Supplementary Figure S1a).
We compared specific cytotoxicity of both CD4 and CD8 TN-enriched CAR T cells produced using clinical vector BAFF-R:4-1BB:ζ/EGFRt and prototype vector BAFF-R:4-1BB:ζ/GFP using an vitro cytotoxic T lymphocyte (CTL) assay [15] in a panel of chromium-51 labeled target malignant B-cell lines including Nalm-6 B-ALL (Fig. 1c). In all tumor models and at varying effector to target ratios, BAFF-R:4-1BB:ζ/EGFRt and BAFF-R:4-1BB:ζ/GFP CAR T cells showed similar in vitro activity. Similarly, CAR T cells generated using the two BAFF-R vectors were compared in an established, benchmark B-cell lymphoma tumor model, Z-138 (Fig. 1d). Tumor bearing mice, randomized prior to a single CAR T-cell infusion, were monitored by bioluminescent imaging of luciferase-expressing Z-138 tumors as previously described [15]. Rapid tumor clearance was observed in both the prototype and clinic-ready BAFF-R CAR treated mice. Mice in these treated groups also demonstrate comparable and significant tumor free survival compared to controls (Fig. 1e). Together, these results show that the BAFF-R:4-1BB:ζ/EGFRt clinical-grade vector-generated CAR T cells are functionally equivalent to those produced by the BAFF-R:4-1BB:ζ/GFP prototype vector. We could then reasonably proceed to further preclinical development.
We next evaluated cGMP protocols for lentiviral vector production and CAR T cell manufacture with the aim of completing FDA-mandated investigational new drug (IND)-enabling studies following the procedure outlined in Fig. 2a. We therefore set out to optimize BAFF-R CAR T-cell production under several key criteria including: minimum viral load; maximized transduction efficiency, functional potency; and passing FDA required release testing.
The City of Hope Biologics & Cellular GMP Manufacturing Center produced BAFF-R:4-1BB:ζ/EGFRt clinical vector under cGMP conditions. Transfection of TN/MEM cells obtained from healthy donors [Michael Amini Transfusion Medicine Center at the City of Hope (IRB: 15283)] was evaluated at three multiplicities of infection (MOI), 0.5, 1, and 2 (Fig. 2b). All three MOIs produced viable CAR T cells with equivalent growth and expansion rates (Supplementary Figure S1b), and transduction efficiency measured by FACS analysis of EGFRt were equivalent for all MOIs. Maximized functional potency was confirmed by BAFF-R specific activation of BAFF-R CAR T cells, measured by CD107a degranulation marker following incubation with target cell lines (Fig. 2c, Supplementary Figure S2) as previously described [15]. Functional potency at MOI = 0.5 was similar to MOI = 1 or 2. Furthermore, the BAFF-R-induced CAR T cell INF-γ release did not vary significantly between MOIs (Supplemental Figure S3). Subsequent batches of BAFF-R CAR T cells were produced with MOI = 0.5 transfection and underwent FDA required release testing.
Three donor leukapheresis products were successfully used to generate BAFF-R CAR T cells following the GMP protocol derived from a successful production strategy that implements CliniMACS automation for T-cell isolation [11]. The original manufacturing strategy was developed and implemented at City of Hope and produces CAR T cells from early stage T cells (TN/MEM) consisting of naïve, memory-like stem, and central memory T cells. Each batch of cells were closely monitored during the transfection, expansion, and activation stages. Each exceeded the minimum requirements of qualification release testing set forth by the FDA, which included viability (≥ 70%), identity measured by CD3 (≥ 80%), transduction efficiency and CAR expression measured by EGFRt (≥ 10%), WPRE copy insertion (≤ 5 copies/cell), and VSVG copy insertion (≤ 2.5 copies/50 ng genomic DNA) (Table 1).
Table 1 Qualification release testing Lastly, we repeated an in vivo study to confirm the efficacy of the BAFF-R:4-1BB:ζ/EGFRt CAR T cells produced under the GMP protocol, as the research and GMP production strategies differed in the specific T-cell isolation and expansion steps. Specifically, the GMP strategy isolates all early stage T cells (naïve, memory-like stem, and central memory), whereas the research product was exclusively produced from naïve T cells. Secondly, the CAR T-cell enrichment was omitted in the GMP procedures, yielding a more heterogeneous product consisting of 25–35% CAR T cells. Therefore, experimental group dosing in the in vivo Nalm-6 ALL tumor model study is denoted as low dose (2.8 × 106 T cell yielding an effective 1 × 106 BAFF-R CAR T cells) or high dose (5.6 × 106 T cell yielding an effective 2 × 106 BAFF-R CAR T cells) (Fig. 2d). Remarkably, mice treated with either a low or high dose of cGMP BAFF-R CAR T cells demonstrated rapid tumor clearance and tumor-free survival up to 140 days after tumor challenge (Fig. 2e).