A High-Throughput Method for Characterizing Novel Chimeric Antigen Receptors in Jurkat Cells

Abstract

Chimeric antigen receptor (CAR) development involves extensive empirical characterization of antigen-binding domain (ABD)/CAR constructs for clinical suitability. Here, we present a cost-efficient and rapid method for evaluating CARs in human Jurkat T cells. Using a modular CAR plasmid, a highly efficient ABD cloning strategy, plasmid electroporation, short-term co-culture, and flow-cytometric detection of CD69, this assay (referred to as CAR-J) evaluates sensitivity and specificity for ABDs. Assessing 16 novel anti-CD22 single-chain variable fragments derived from mouse monoclonal antibodies, CAR-J stratified constructs by response magnitude to CD22-expressing target cells. We also characterized 5 novel anti-EGFRvIII CARs for preclinical development, identifying candidates with varying tonic and target-specific activation characteristics. When evaluated in primary human T cells, tonic/auto-activating (without target cells) EGFRvIII-CARs induced target-independent proliferation, differentiation toward an effector phenotype, elevated activity against EGFRvIII-negative cells, and progressive loss of target-specific response upon in vitro re-challenge. These EGFRvIII CAR-T cells also showed anti-tumor activity in xenografted mice. In summary, CAR-J represents a straightforward method for high-throughput assessment of CAR constructs as genuine cell-associated antigen receptors that is particularly useful for generating large specificity datasets as well as potential downstream CAR optimization.

Keywords: CAR-T; CD69; EGFRvIII; Jurkat; T cell; high-throughput; live imaging; plasmid; pre-clinical; screening.

Graphical abstract

Figure 1

Development of pSLCAR and CAR-J: A Modular CAR Plasmid and Screening Strategy (A) Typical CAR structure with examples of commonly used domains. Bold type indicates specific domains in the CAR construct used here. (B) Single-chain variable fragment (scFv) structure, as utilized in many CAR constructs. (C) Functional regions of simple lentiviral CAR plasmid (pSLCAR) and CD19-targeting pSLCAR. (D–F) Proof-of-concept CAR-J assay. (D) Illustration of flow cytometry gating strategy, permitting detection of CAR-expressing cells with GFP. (E) CD69-APC expression on GFP-positive Jurkat cells electroporated with pSLCAR-CD19 incubated with Nalm6 cells at increasing ratios. (F) Dose-response of CAR-J response to Nalm6. (G) Depiction of scFv swapping strategy and BpiI restriction cassettes contained in pSLCAR-CD19 with gene design for DNA fragments. (H and I) CAR-J conducted with Jurkat cells electroporated with anti-CD19 and anti-HER2 CARs. Targets were Raji (H) and SKOV3 (I). (J) CAR-J conducted using CARs derived from 16 antibodies previously determined to specifically bind CD22. Results in (F), (H), (I), and (J) represent means ± SEM of 3 independent experiments (n = 3). Asterisks in (F), (H), and (I) represent significant difference (p < 0.05) between the indicated group and all other groups at individual ratios, calculated using one-way ANOVAs. Pound signs in (J) represent significant difference (p < 0.05) between the indicated group and pSLCAR control, calculated using one-way ANOVA.

Figure 2

CAR-J Allows Short-Listing of Candidate scFv Sequences (A) Overview of high-throughput CAR-J setup. (B) CAR-J tonic signaling (auto-activation) as indicated using flow cytometry assessment of surface CD69-APC expression when cultured alone (no target cells) or with EGFRvIII-negative U87 cells. (C) CAR-J dose-response activation when co-cultured alongside increasing numbers of U87-vIII cells. (D–F) Gene fragments encoding different anti-EGFRvIII scFvs were cloned into pSLCAR-CD19-28-3ζ, -BB-3ζ, or -3ζ. (D and E) CAR-J at a 1:1 Jurkat-to-target cell ratio using (D) cloned and sequenced (complete protocol) or (E) the polyclonal (rapid protocol) plasmid preparation. (F) These two approaches were compared using a CAR-J magnitude/specificity score (see Materials and Methods). Results represent means ± SEM of 3 independent experiments (n = 3). Pound signs in (B) represent significant difference (p < 0.05) between the indicated group and CD19 control, calculated using one-way ANOVA. In (F), protocols were not different (t test) when considered for individual construct-scFv combinations or when data were combined. A two-way ANOVA analysis of combined complete and rapid protocol data indicated main effects of construct and scFv, indicated with lowercase letters where groups with different letters are significantly different from each other.

Figure 3

CAR-J Tonic Signaling Predicts Human Target-Independent CAR-T Cell Activity and Differentiation in Primary T Cells Human CAR-T cells were generated from donor PBMCs via transduction with lentivirus produced from pSLCAR-EGFRvIII-CD28-CD3ζ. (A and B) Representative flow cytometry plots of T cell/CAR-T cell CD25 expression 7 days after lentiviral transduction (A) and other T cell phenotype markers on donor 1 15 days after transduction (B). (C) Quantification of CD25-positive cells in total (mock/untransduced T cells) or gated GFP-positive CAR-T cells (EGFRvIII-CAR-transduced cells) 7 days after transduction. (D) Correlation between day 7 CD25 and tonic CAR-J activation data from (C). (E) Similarly, the relationship between day 7 CD25 and the CAR-J magnitude/specificity score (to maintain positive relationship, presented as 1/score) as calculated on 28-3ζ constructs (Figure 2F). Results in (C) represent means ± SEM of 4 separate donors (n = 4), and results in (D) and (E) represent results from 4 separate donors (n = 4) and 3 independent CAR-J assays (n = 3). Asterisk in (C) represents significant difference (p < 0.05) between the indicated group and all other groups, calculated using one-way ANOVA.

Figure 4

CAR-J Predicts Human CAR-T Cell Function Anti-EGFRvIII CAR-T cell effector function was assessed using continuous live-cell imaging of co-cultures between human CAR-T cells (GFP, green) and various cancer cell lines (stably expressing mKate2, red). (A) Representative images taken on day 6 from donor 2. (B) Autonomous T cell proliferation when cultured alone, as measured by assessing phase contrast. (C and D) Antigen-driven CAR-T cell proliferation, as measured by assessing GFP (C), and growth of target-specific cells, as measured by assessing mKate2 (D), during U87-vIII co-culture. (E–G) Growth of various EGFRvIII-negative cancer cell lines during co-culture with mock or EGFRvIII-CAR T cells. Targets were SKOV3 (E), Nalm6 (F), and MCF7 (G). (H) CAR-T cell proliferation during SKOV3 co-culture. (I) Average expansion of U87-vIII and MCF7 cells across 3 donors. (J) Area under the curve (AUC) measurements performed on data in (I). Scale bars represent 400 μm. Results shown in (B)–(H) are derived from donor 2 only but are representative of repeated experiments (see Figures S3–S5), while (I) and (J) represent donors 1, 2, and 4 (n = 3). Asterisks in (J) represent significant difference (p < 0.05) between the indicated group and all other groups, calculated using one-way repeated-measures ANOVA. Note that a technical issue was encountered during the experiment with donor 3, and in vitro killing assay was not possible.

Figure 5

Repeated Co-culture Re-challenge Leads to Loss of Effector Function for Auto-activating CARs After 7 days of co-culture with U87-vIII cells (first challenge), EGFRvIII CAR-T cells generated from donor 4 were isolated, diluted, and used to conduct a similar co-culture assay with U87-vIII cells (second challenge). CAR-T cells cultured alongside U87-vIII cells for 7 days in the second challenge were then used to conduct a similar tertiary re-challenge assay (third challenge). (A) U87-vIII cell growth was assessed by measuring the change in mKate2 (red) area from the first two time points (0 and 1 h). (B) CAR-T cell expansion was assessed by the total GFP (green) area. (C) Representative images from the tertiary re-challenge experiment. Scale bars represent 400 μm.

Figure 6

Anti-EGFRvIII CAR-T Cells Demonstrate Therapeutic Efficacy In Vivo NSG mice were subcutaneously xenografted with U87-vIII tumor cells. One week later, the indicated CAR-T cells were prepared and injected intra-tumorally, after which tumor size and survival were measured. (A) Tumor growth as determined by thrice weekly caliper measurements. (B) Tumor size 24 days after tumor injection. (C) Kaplan-Meyer survival curve, with median survival numbers indicated. (D) Fluorescent in vivo imaging performed to visualize tumor growth of mKate-expressing U87-vIII cells. Asterisks in (B) and (C) represent significant difference (p < 0.05) between the indicated group and all other groups, calculated using one-way ANOVAs. Median survival was calculated using GraphPad Prism. n = 5 mice per group.

Figure 7

Overview of Complete CAR-J Screening Protocol See Materials and Methods for a complete description of this protocol. See Figure S6 for a similar cartoon flow diagram outlining the rapid CAR-J protocol.