Modulation of Target Antigen Density Improves CAR T-cell Functionality and Persistence

Abstract

Purpose: Chimeric antigen receptor T-cell (CART) therapy targeting CD22 induces remission in 70% of patients with relapsed/refractory acute lymphoblastic leukemia (ALL). However, the majority of post-CD22 CART remissions are short and associated with reduction in CD22 expression. We evaluate the implications of low antigen density on the activity of CD22 CART and propose mechanisms to overcome antigen escape.

Experimental design: Using ALL cell lines with variable CD22 expression, we evaluate the cytokine profile, cytotoxicity, and in vivo CART functionality in the setting of low CD22 expression. We develop a high-affinity CD22 chimeric antigen receptor (CAR) as an approach to improve CAR sensitivity. We also assess Bryostatin1, a therapeutically relevant agent, to upregulate CD22 and improve CAR functionality.

Results: We demonstrate that low CD22 expression negatively impacts in vitro and in vivo CD22 CART functionality and impairs in vivo CART persistence. Moreover, low antigen expression on leukemic cells increases naïve phenotype of persisting CART. Increasing CAR affinity does not improve response to low-antigen leukemia. Bryostatin1 upregulates CD22 on leukemia and lymphoma cell lines for 1 week following single-dose exposure, and improves CART functionality and in vivo persistence. While Bryostatin1 attenuates IFNγ production by CART, overall in vitro and in vivo CART cytotoxicity is not adversely affected. Finally, administration of Bryostain1 with CD22 CAR results in longer duration of in vivo response.

Conclusions: We demonstrate that target antigen modulation is a promising strategy to improve CD22 CAR efficacy and remission durability in patients with leukemia and lymphoma.See related commentary by Guedan and Delgado, p. 5188.

Figure 1:. CD22 expression decreases following CD22 CART, and decreased CD22 expression attenuates CART expansion and persistence.

(A) Patient samples pre- and post-CD22 CAR therapy were evaluated for CD22 expression using Quantibrite-PE bead evaluation. Statistics were performed using Wilcox test. (B) Four separate CD22 CART patient samples – two at Dose level 1 and two at Dose level 2 – were evaluated for site density using Quantibrite-PE analysis and CD22 CAR using flow cytometry. (CR – MRD-negative Complete Response; SD – Stable Disease) (C) 1×10⁵ tumor cells were co-cultured with 1×10⁵ CD22 CAR cells from CD22 CART patient samples for 18hrs. Supernatant was evaluated by Meso Scale Multiplex pro-inflammatory cytokine panel. Statistics were calculated using unpaired t-test (**** p<0.0001, *** p<0.0002, ** p<0.0021). (D) Kaplan-Meier curves comparing CD22^neg to CD22^lo and Nalm6 leukemia-bearing mice with different CART doses. (E) Peripheral blood was collected from mice at interval timepoints and assessed for CAR expansion using flow cytometry and analyzed on Fortessa flow machine. (F) CAR expansion was evaluated from bone marrow of mice 16 days after tumor injection. Cells were stained for flow cytometry and analyzed on Fortessa flow machine. Statistics were calculated using Mann-Whitney test (**** p < 0.0001, * p < 0.0332).

Figure 2:. Site density affects early activation and memory phenotype of CD22 CART cells.

(A) Site density cell lines were co-incubated with CD22 CART for 24 hours, and PD1 expression was stained flow cytometry and analyzed on Fortessa flow machine. (B) CD22 CAR was co-cultured in vitro with CD22^lo or Nalm6 leukemia for 8 days and cells were evaluated using flow cytometry and analyzed on Fortessa flow machine. (C) Cells were extracted from CD22 CAR treated CD22^lo or Nalm6 NSG mice 16 or 30 days after tumor injection. Cells were stained for flow cytometry and analyzed on Fortessa flow machine. (D) Cells were extracted from CD22 CAR treated CD22^lo or Nalm6 NSG mice 16 days following tumor injection. Cells were stained for flow cytometry and analyzed on Fortessa flow machine. This data was reproducible across two separate experiments. Statistics were calculated using Mann-Whitney test (**** p < 0.0001, * p < 0.0332).

Figure 3:

(A-B) 1×10⁵ tumor cells were co-cultured with 1×10⁵ Mock, CD22, or CD22^V1 CAR and assessed for (A) IFN-γ and IL-2 cytokines by ELISA from cell culture supernatants (statistics were calculated using paired t test (* p<0.0332)) or (B) AnnexinV staining was assessed over time using IncuCyte ZOOM. This data is representative of two separate experiments and was consistent across 2 different E:T ratios. (C) NSG mice were injected with 1×10⁶ GPF-positive CD22^neg, CD22^lo, or Nalm6 tumor cells on Day 0. On Day 3, 5×10⁶ CD22 or CD22^V1 CAR were injected for treatment. Mice were imaged using IVIS technology and luciferin-D IP injections. Luminescence quantification is shown on the right. This data is representative of two separate experiments. (D) CD22 or CD22^V1 CART were co-incubated with tumor cells. On Days 1 and 8, CAR was harvested, stained for flow cytometry, and analyzed on Fortessa Flow machine. Statistics were calculated using unpaired t-test (* p < 0.0332). (E-F) Mice were injected with CD22^lo or Nalm6 leukemia on Day 0. 5×10⁶ CD22 CAR T cells were administered on Day 3 and mice were sacrificed on Day 16. Bone marrow cells were stained for flow cytometry and analyzed on Fortessa Flow machine. Statistics were calculated using unpaired t-test (**** p<0.0001, * p<0.0332).

Figure 4:. Bryostatin1 upregulates CD22, but not CD19, and CD22 increased expression is durable for 1 week after drug exposure.

(A) Four cell lines were co-incubated with 1nM Bryostatin1 for 24 hours and analyzed using flow cytometry one day after Bryostatin1 exposure. Site density was analyzed through use of standardized Quantibrite-PE beads. Statistics were calculated using unpaired t-test (*** <0.0002, ** p<0.0021, * p<0.0332). (B) Cell lines were co-incubated with 1nM Bryostatin1 for 24 hours, washed, and analyzed using flow cytometry at 1 and 7 days after Bryostatin1 exposure. MFI fold change = CD22 MFI^Bryostatin1/CD22 MFI^DMSO. (C) NSG mice were injected with 1×10⁶ GPF-positive Nalm6 leukemia cells on Day 0. Bryostatin1 was administered at 0.8ug/kg on Day 3. Mice were sacrificed 7 and 12 days after Bryostatin1 injection and CD22 was evaluated through flow cytometry. MFI fold change = CD22 MFI^Bryostatin1/CD22 MFI^DMSO. (D) CD22-low relapse patient-derived xenograft cell line was cultured in vitro with 1nM Bryostatin1 for 24 hours and measured for CD22 expression using CD22 antibody.

Figure 5:. RNAseq analysis shows no substantive difference in CD22 expression; Bryostatin1-mediated increased CD22 augments CART potency; and CART exposure to Bryostatin1 decreases IFN-gamma production, but increases Granzyme B production, and does not adversely affect tumor clearance in vitro or in vivo.

(A-B) KOPN8 or SEM cell lines were exposed to 1nM Bryostatin1 for 16, 24, 48, or 72 hours. RNA was extracted and analyzed by RNAseq. (C-E) Cell lines were co-incubated with 1nM of Bryostatin1 for 24 hours. Then 1×10⁵ target tumor cells were co-cultured with 1×10⁵ CD22 CAR for 16hrs. IL-2 (C), IFN-γ (D), and granzyme B (E) were measured by ELISA from cell culture supernatants. Statistics were calculated using unpaired t-test (**** p<0.0001, *** p<0.0002, ** p<0.0021, * p<0.0332). (F-G) CD22 CART cells were co-incubated with 1nM of Bryostatin1 for 24 hours and washed. Then 1×10⁵ target tumor cells were co-cultured with 1×10⁵ CD22 CAR for 18 hrs. IFN-γ (F), and granzyme B (G) were measured by ELISA from cell culture supernatants. Statistics were calculated using unpaired t-test (**** p<0.0001, *** p<0.0002, ** p<0.0021). (H) Mock or CD22 CAR T cells were co-incubated at 1:1 effector-to-target ratio with either CD22^neg, CD22^lo or Nalm6 cells. Cell death was monitored by loss of GFP-positive cells using IncuCyte ZOOM. (I) NSG mice were injected with CD22^neg, CD22^lo, or Nalm6 on Day 0, CD22 CAR T cells on Day 3, and then were given either DMSO control or Bryostatin1 at 40ug/kg once weekly for 2 weeks. Leukemia progression was monitored using IVIS technology and luciferin-D IP injections.

Figure 6:. Bryostatin1 treatment pre-CAR infusion alters T cell phenotype without T cell exhaustion, and post-CAR infusion improves durability of remission in vivo.

(A) Nalm6 cells were exposed to 1nM Bryostatin1 for 24 hours, then injected into mice on Day 0. CD22 CAR T cells were administered on Day 3 and mice were sacrificed on Day 10. Bone marrow cells were stained for flow cytometry and analyzed on Fortessa Flow machine. Statistics were calculated using unpaired t-test (* p < 0.0332). (B) NSG mice were injected with Nalm6 on Day 0, CD22 CART on Day 3, and then were given either DMSO control or Bryostatin1 at 40ug/kg once weekly for 2 weeks. Mice were sacrificed 30 days after tumor injection. Cells were stained for flow cytometry and analyzed on Fortessa Flow machine. (C) NSG mice were injected with 1×10⁶ GPF-positive Nalm6 or SEM tumor cells on Day 0. On Day 3, either 3×10⁶ (Nalm6) or 2×10⁶ (SEM) Mock or CD22 CAR were injected for treatment. Mice were given 40ug/kg of Bryostatin1 or DMSO once weekly for 2 weeks. Mice were imaged using IVIS technology and luciferin-D IP injections. (D) NSG mice were injected with 1×10⁶ GPF-positive PDX tumor cells on Day 0. On Day 42, 3×10⁶ Mock or CD22 CAR were injected for treatment. Mice were given 40ug/kg of Bryostatin1 or DMSO once weekly for 2 weeks starting on Day 45. Mice were imaged using IVIS technology and luciferin-D IP injections.