A novel and efficient tandem CD19- and CD22-directed CAR for B cell ALL

Abstract

CD19-directed chimeric antigen receptor (CAR) T cells have yielded impressive response rates in refractory/relapse B cell acute lymphoblastic leukemia (B-ALL); however, most patients ultimately relapse due to poor CAR T cell persistence or resistance of either CD19⁺ or CD19^- B-ALL clones. CD22 is a pan-B marker whose expression is maintained in both CD19⁺ and CD19^- relapses. CD22-CAR T cells have been clinically used in B-ALL patients, although relapse also occurs. T cells engineered with a tandem CAR (Tan-CAR) containing in a single construct both CD19 and CD22 scFvs may be advantageous in achieving higher remission rates and/or preventing antigen loss. We have generated and functionally validated using cutting-edge assays a 4-1BB-based CD22/CD19 Tan-CAR using in-house-developed novel CD19 and CD22 scFvs. Tan-CAR-expressing T cells showed similar in vitro expansion to CD19-CAR T cells with no increase in tonic signaling. CRISPR-Cas9-edited B-ALL cells confirmed the bispecificity of the Tan-CAR. Tan-CAR was as efficient as CD19-CAR in vitro and in vivo using B-ALL cell lines, patient samples, and patient-derived xenografts (PDXs). Strikingly, the robust antileukemic activity of the Tan-CAR was slightly more effective in controlling the disease in long-term follow-up PDX models. This Tan-CAR construct warrants a clinical appraisal to test whether simultaneous targeting of CD19 and CD22 enhances leukemia eradication and reduces/delays relapse rates and antigen loss.

Keywords: B-ALL; CD19; CD22; patient-derived xenografts; relapse; tandem CAR T cells.

Graphical abstract

Figure 1

Generation, transduction, expansion, and detection of Tan-CAR T cells (A) Scheme of the CAR constructs used: (S) and (L) denote short -(G₄S)₄- and long -(G₄S)₇- size for the inter-scFv linker, respectively. (B) T cell activation after 48-h exposure to anti-CD3/CD28, plus IL-7 and IL-15 evaluated by CD25 and CD69 expression by FACS (left panel) and by light microscopy analysis of activated T cell clusters (right panel, magnification ×40) (n = 5). (C) Transduction efficiency (left panel) and expansion (right panel) of activated T cells transduced with the indicated CARs (n = 5). The arrow represents the time of CAR T cell harvesting for CAR detection in the surface of T cells. (D) Representative flow cytometry plots of CAR expression on human T cells detected as GFP⁺ (top panels), anti-scFv (second row), CD19-Fc/anti-Fc-PE (third row), and anti-HisTag-APC or CD22-HisTag/anti-HisTag-APC (bottom row). CAR-transduced T cells are shown in green. (E) Representative CD19 and CD22 mean fluorescence intensity (MFI) quantification by FACS of the GFP⁺ CAR T cells shown in (D). MFI values are indicated in the upper right corner. (F) Representative CAR detection on human CD4⁺ and CD8⁺ T cells. (G) VCN representing the number of integrated copies of the provirus (CAR vector) per transduced genome. Each symbol represents a different donor (n = 3). See also Figure S2.

Figure 2

Robust antileukemic efficacy and specificity of both Tan(S)- and Tan(L)-CAR T cells in vitro (A and B) Absolute number (A) and percentage (B) of alive target cells (SEM or NALM6) after 48-h incubation with the indicated CAR T cells and E:T ratios. Results in (B) are normalized to Mock-CAR data. PBMCs from n = 3 independent HDs. (C) Production of the pro-inflammatory cytokines IL-2, IFN-γ, and TNF-α by CAR T cells after 48-h exposure to SEM or NALM6 target cells at 1:1 E:T ratio. PBMCs from n = 3 independent HDs. (D) Different CD22/CD19 combinatorial phenotypes of CRISPR/Cas9-edited SEM cells. (E and F) Absolute number (E) and percentage (F) of alive target cells after 48-h incubation with the indicated CAR T cells and E:T ratios. Results in (F) are normalized to Mock-CAR data. PBMCs from n = 5 independent HDs. (G) Production of IL-2, IFN-γ, and TNF-α by the indicated CAR T cells after 48-h exposure to the indicated phenotypes of SEM cells (n = 5). Data are shown as means ± SEMs. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001; 1-way ANOVA with the Tukey post hoc test. See also Figures S1 and S3.

Figure 3

Tan(S)-CAR T cells are as efficient as CD19-CAR T cells in vivo using both NALM6 and SEM B-ALL cell lines (A) Scheme of the experimental design. NSG mice (n = 6/group) were intra-BM transplanted with 1 × 10⁵ Luc-expressing NALM6 or SEM cells. Four days later, 4 × 10⁶ Mock-CAR, CD19-CAR, or Tan(S)-CAR T cells were i.v. injected. Leukemic burden was monitored weekly by BLI. Mice were sacrificed and FACS analyzed for leukemic burden and T cell persistence when Mock-treated mice were fully leukemic by BLI. (B) IVIS imaging of NALM6 leukemic burden monitored by BLI at the indicated time points. (C) Average radiance quantification (p/sec/cm²/sr) for NALM6 at the indicated time points. (D) NALM6 leukemic burden quantified by FACS in BM and PB at sacrifice of mice treated with Mock-CAR, CD19-CAR, and Tan(S)-CAR, respectively. Each dot represents a mouse. (E) Representative FACS plots showing NALM6 cells (blue) and T cell persistence (red) in BM at sacrifice of mice treated with Mock-CAR, CD19-CAR, and Tan(S)-CAR, respectively. (F–I) Identical analysis to (B)–(E) for SEM target cells. Data are shown as means ± SEMs. ∗p < 0.05, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001; 1-way ANOVA with the Tukey post hoc test.

Figure 4

Tan(S)-CAR T cells eliminate primary and PDX B-ALL cells in vitro (A and B) CD22 and CD19 FACS expression in primary B-ALL blasts (n = 3) (A) and B-ALL PDX cells (n = 3) (B). (C and D) Absolute number of live primary B-ALL blasts (C) and B-ALL PDX cells (D) after 24-h incubation with the indicated CAR T cells at 2:1 E:T ratio. (E and F) Production of the pro-inflammatory cytokines IL-2, IFN-γ, and TNF-α by the indicated CAR T cells after 24-h exposure to either primary B-ALL blasts (E) or B-ALL PDX cells (F) at 2:1 E:T ratio. PBMCs from n = 3 independent HDs. (G) Absolute number of live T cells after 24-h incubation with the indicated primary B-ALL blasts or B-ALL PDX cells at 2:1 E:T ratio. (H and I) Autologous cytotoxicity experiment. Representative FACS plots of CAR T cells produced from MACS-sorted T cells from the PB of a B-ALL patient (Pt #4) showing the transduction efficiency at day 6 (H). Absolute number of live SEM cells (left panel) or live primary B-ALL blasts from the same patient (patient 4) after 24-h incubation with the Tan(S)-CAR T cells at 2:1 E:T ratio (I). Data are shown as means ± SEMs. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001; 2-way ANOVA with the Tukey post hoc test.

Figure 5

Tan(S)-CAR is very effective in controlling the disease in a long-term follow-up B-ALL PDX model (A) Scheme showing the experimental design. NSG mice (n = 5–9/group) were i.v. transplanted with 0.5 × 10⁶ or 1 × 10⁶ of B-ALL cells from PDX#3 or PDX#4, respectively. Upon B-ALL engraftment detectable in BM, mice were randomized, and 5 × 10⁶ Mock-CAR, CD19-CAR, or Tan(S)-CAR T cells were i.v. injected. Leukemic burden and CAR T cell persistence was monitored in PB biweekly by FACS. BM aspirates were FACS analyzed when Mock-treated mice were sacrificed (week 4) and at the endpoint (week 13). (B) Upper panels, leukemic burden in PB and BM at the indicated time points after CAR T cell infusion. Bottom panels, representative BM FACS analysis showing CD19 and CD22 expression for both PDXs before CAR T cell infusion (1 day before CAR T cell infusion [−0.1]) and at the time Mock-treated mice were sacrificed (week 4). The gating strategy is shown on the left. Indicated percentages are referred to the total live single cells in each sample. The complete gating strategy is shown in Figure S1B. (C) Follow-up at the indicated time points after CAR T cell infusion of leukemic progression/relapse (left panel) and persistent T cells (right panel) of CD19-CAR-treated versus Tan(S)-CAR-treated mice transplanted with PDX#4 (n = 7 mice/group). (D) DFS curves for CD19-CAR-treated versus Tan(S)-CAR-treated mice transplanted with PDX#4. The log-rank (Mantel-Cox) test was used to calculate significance. (E) Leukemic burden at sacrifice/endpoint (week 13 after CAR T cell infusion) in BM from CD19-CAR-treated versus Tan(S)-CAR-treated mice transplanted with PDX#4 (n = 7 mice/group). A mouse is considered in relapse when the percentage of blasts in BM is >1% (horizontal dotted line) or >0.1% in PB. Each dot represents a mouse. The bottom panels show the expression of CD19 and CD22 by FACS analysis of B-ALL cells for each independent CD19-CAR-treated and Tan(S)-CAR-treated mouse. Indicated percentages are referred to the total live single cells in each sample. The complete gating strategy is shown in Figure S1B. Data are shown as means ± SEMs. ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001; 1-way ANOVA with the Tukey post hoc test. See also Figure S1.

Figure 6

In vivo bispecificity of Tan(S)-CAR T cells (A) Average radiance quantification (p/sec/cm²/sr) from mice transplanted with SEM CD22-KO (n = 4–5/group) or SEM CD19-KO (n = 9–10/group) at the indicated time points after Mock- or Tan(S)-CAR T cell infusion. Data from n = 2 independent experiments. (B) Representative BLIs of SEM CD22-KO- and SEM CD19-KO-transplanted mice from week 2 to week 4 after CAR T cell infusion. (C) CD19 and CD22 expression of the CD19⁻CD22⁺ blasts from the B-ALL patient (Pt#5) used for the in vivo bispecificity experiment. (D) NSG mice (n = 6/group) were i.v. transplanted with 1 × 10⁶ of CD19⁻CD22⁺ B-ALL cells (Pt#5). Upon detectable B-ALL engraftment in BM, mice were randomized to receive 5 × 10⁶ Mock-CAR, CD19-CAR, CD22-CAR, or Tan(S)-CAR T cells. The leukemic burden in BM before CAR T cell infusion (day 0) and at endpoint of the experiment (day 49) is shown. Data are shown as means ± SEMs. ∗p < 0.05, ∗∗p < 0.01, 2-way ANOVA (mixed-effects model) with Šídáks multiple comparisons test.