Optimized depletion of chimeric antigen receptor T cells in murine xenograft models of human acute myeloid leukemia

Abstract

We and others previously reported potent antileukemia efficacy of CD123-redirected chimeric antigen receptor (CAR) T cells in preclinical human acute myeloid leukemia (AML) models at the cost of severe hematologic toxicity. This observation raises concern for potential myeloablation in patients with AML treated with CD123-redirected CAR T cells and mandates novel approaches for toxicity mitigation. We hypothesized that CAR T-cell depletion with optimal timing after AML eradication would preserve leukemia remission and allow subsequent hematopoietic stem cell transplantation. To test this hypothesis, we compared 3 CAR T-cell termination strategies: (1) transiently active anti-CD123 messenger RNA-electroporated CART (RNA-CART123); (2) T-cell ablation with alemtuzumab after treatment with lentivirally transduced anti-CD123-4-1BB-CD3ζ T cells (CART123); and (3) T-cell ablation with rituximab after treatment with CD20-coexpressing CART123 (CART123-CD20). All approaches led to rapid leukemia elimination in murine xenograft models of human AML. Subsequent antibody-mediated depletion of CART123 or CART123-CD20 did not impair leukemia remission. Time-course studies demonstrated that durable leukemia remission required CAR T-cell persistence for 4 weeks prior to ablation. Upon CAR T-cell termination, we further demonstrated successful hematopoietic engraftment with a normal human donor to model allogeneic stem cell rescue. Results from these studies will facilitate development of T-cell depletion strategies to augment the feasibility of CAR T-cell therapy for patients with AML.

© 2017 by The American Society of Hematology.

Figures

Figure 1.

Mechanism of rituximab-mediated depletion of CART123-CD20 and alemtuzumab depletion of CART123. (A) Schema of CART123-CD20 construct. (B) Surface expression of CD123-CAR-PE and CD20-PE-Cy7 on CART123-CD20 T cells by FC. (C) In vitro incubation of CD123+ MOLM14 cells with CART123 or CART123-CD20 have similar elevated inflammatory cytokine production as measured by FC, suggesting similar potency of both CD123-redirected CAR T-cell products. Minimal cytokine production occurs with coincubation of CAR T cells and CD123-Jurkat T cells. (D) Similar cytotoxicity is observed with in vitro coculture of MOLM14 cells with CART123 or CART123-CD20 at all effector:target (E:T) ratios, as measured by BLI. (E) Similar cytotoxicity is observed with coculture of MOLM14 cells with RNA-CART123 or CART123-CD20, as measured by BLI. (F) Rituximab depletes CART123-CD20 cells by CDC and by direct cellular cytotoxicity. Coculture of CART123-CD20 with increasing concentrations of rituximab in the absence or presence of 15% complement resulted in CART123-CD20 depletion only in the presence of complement and had direct cytotoxicity at highest concentration. Residual live CAR T cells were quantified by FC after 12 hours of coculture, and percent T-cell depletion was calculated. (G) Rituximab depletes CART123-CD20 by CDC within 4 to 12 hours of coculture. CART123-CD20 cells were incubated with rituximab at indicated concentrations in the presence of complement. Residual live CAR T cells were measured by FC at indicated time points for calculation of T-cell killing. (H) No evidence for ADCC was observed. CART123-CD20 cells (labeled with carboxyfluorescein diacetate succinimidyl ester [CFSE]) were cocultured with Dil-labeled macrophages and rituximab at the indicated concentrations. Phagocytosis after 2 hours (defined as CFSE+Dil+ singlet cells) was quantified by FC. Percent phagocytosis did not differ in the absence or presence of rituximab. (I) Similar to data in panel E, coculture of CART123 cells with alemtuzumab in the absence or presence of complement demonstrates CDC-mediated T-cell killing.

Figure 2.

Effects of alemtuzumab upon human AML. (A) Surface CD52 is minimally expressed on human AML cell lines MOLM13, MOLM14, MV4-11, THP-1, and U937 by FC analysis. Comparably bright CD52 expression is observed on normal human T cells (untransduced) and transduced CART123 cells used in subsequent in vivo experiments. (B) Coculture of MOLM14 cells with alemtuzumab and 15% complement (as in Figure 1) for 72 hours does not impair in vitro leukemia proliferation, as measured by viable cell counting with trypan blue exclusion. Each condition was performed in triplicate. Data points demonstrate mean cell count with standard error of the mean. (C) Histopathologic analyses of NSG mice treated with 1 dose of 1 mg/kg or 5 mg/kg alemtuzumab demonstrate no effect of alemtuzumab upon normal murine hematopoietic tissues (left panels) or upon human AML in NSG mice engrafted with MOLM14 (right panels). Images are hematoxylin and eosin–stained tissue sections from embedded paraffin blocks. Slides were scanned at ×20 magnification with an Aperio Scanscope CS-O slide scanner with visualization via the Aperio Image Analysis Toolkit (Leica Biosystems). ns, not significant by analysis of variance. PBS, phosphate-buffered saline.

Figure 3.

Serial ablation of CART123 with alemtuzumab in AML xenograft model. NSG mice were injected with luciferase-expressing MOLM14 cells at week 0 time point, then treated with saline, UTD, or CART123 (n = 10 mice per cohort) at week 1 time point. Some CART123-treated mice were subsequently treated with 1 dose of 1 mg/kg alemtuzumab (alem) IP at 1, 2, or 3 weeks following CART123 (denoted as colored bars or as “A”) (weeks 2, 3, or 4 time points; n = 5 to 8 mice per cohort). Animals were assessed by (A) BLI with (B) quantification of radiance as a surrogate for AML burden, then euthanized when moribund. Two CART123-treated animals (red line in panel A) were found dead at >3 months post-T cells without evidence of recurrent leukemia. Animals in CART123-induced “remission” without and with alemtuzumab were subsequently rechallenged with MOLM14. (C) AML burden by BLI and CD3-PacificBlue+ CART123 cells in peripheral blood were quantified during treatment. MOLM14 rechallenge of T-cell–ablated mice at the week 12 time point resulted in rapid AML progression and animal death versus sustained remission of rechallenged non–T-cell–ablated mice with CART123-induced AML remission (****P < .0001 by analysis of variance with log-rank test; denoted for alemtuzumab at week 3 [light green] or week 4 [blue] versus saline control [light blue], and for CART123 [red] versus saline at week 2 or week 3).

Figure 4.

In vivo termination of CAR T cells with alemtuzumab in human AML xenograft models. (A) MOLM14-bearing NSG mice (n = 10 mice per cohort) were treated IV with saline (gray bar), 1 × 106 UTD (striped bar), or 1 × 106 CART123 (red bars) at week 1 time point and assessed by weekly BLI. CART123 induced AML remission by week 4. One dose of 1 mg/kg alemtuzumab was administered IP at 4 weeks following CART123 (week 5 time point; blue bar) in designated animals for T-cell depletion. MOLM14 rechallenge at week 9 resulted in rapid leukemia progression and animal death only in mice in which CART123 had been previously ablated with alemtuzumab. (B) Representative peripheral blood FC analysis of MOLM14-bearing, CART123-treated mouse before and after alemtuzumab (alem) administration (week 5 time point). No CD3-PacificBlue+ human T cells are detected at 24 hours (h) following a single dose of alemtuzumab. Blue gate denotes human T-cell count in murine blood normalized to quantitative counting beads. Note that fewer T cells remain in peripheral blood at this time point due to prior CART123-mediated AML clearance. (C) Immunohistochemical analysis of CD8+ T cells and CD33+ AML cells in harvested tissues of MOLM14-bearing mice treated with CART123 (week 2, 1 week after CART123) or CART123 with 1 mg/kg alemtuzumab (week 5, 24 hours after alemtuzumab/4 weeks after CART123). Note that animals euthanized at earlier time points in this subexperiment were not included in the main data analysis shown in panel A. (D) FC analyses of human CD45-APC+ CD33-PE+ CD123-PE-Cy7+ AML and CD3-PacificBlue+ CAR T cells in an AML PDX model (juvenile myelomonocytic leukemia [JMML] 117). Rapid T-cell depletion and sustained AML remission were observed in CART123-treated animals following a single dose of alemtuzumab (administered 24 hours prior to week 5 blood sampling). (E) End-study IHC analysis of murine bone marrow from saline and UTD control mice is hypocellular and fibrotic without evidence of residual AML, suggestive of advanced leukemia in these models. Marrow tissues from CART123-treated mice are more cellular and demonstrate AML clearance even after alemtuzumab administration.

Figure 5.

Anti-AML efficacy of RNA-CART123 and of CART123/CD20 with rituximab depletion. (A) MOLM14-bearing NSG mice (n = 10 mice per cohort) were treated IV with saline, mock-RNA T cells × 3 doses, RNA-CART123 × 3 doses, CART123, or CART123-CD20 and assessed by BLI. Some mice were injected IP with 60 mg/kg cyclophosphamide diluted in 200 μL saline at 24 hours prior to second and third doses of mock-RNA T cells or RNA-CART123 to deplete previously administered T cells. RNA-CART123 and CART123-CD20 eradicated leukemia by 4 weeks post-T cells (week 5 time point), with similar kinetics to those of CART123, resulting in long-term animal survival. Ablation of CART123/CD20 with 1 mg/kg rituximab IP administered at week 5 did not alter leukemia remission status. MOLM14 rechallenge at week 9 resulted in rapid disease progression and animal death in mice previously treated with RNA-CART123, confirming lack of longer-term persistence of these T cells in vivo. MOLM14 rechallenge of CART123-CD20–treated animals subsequently treated with rituximab also resulted in rapid AML relapse, confirming successful prior ablation of CART123-CD20 in vivo. (B) Treatment of AML290-PDX mice (established as in Figure 4D; n = 8 mice per cohort) with 3 doses of RNA-CART123, as in panel A, induced human leukemia remission measured in murine tissues harvested at 6 weeks after the third dose of T cells (week 8 time point). (C) Treatment of juvenile myelomonocytic leukemia [JMML] 117 PDX mice with CART123 (123) or CART123-CD20 (123-CD20) at week 1 (n = 5 to 8 mice per cohort) resulted in marked reduction in CD45-APC+ CD33-PE+ CD123-PE-Cy7+ AML burden in murine tissues at 8 weeks post-T cells (week 9 time point). PDX mice treated with 1 dose of 1 mg/kg rituximab at the week 5 time point demonstrated sustained AML remission at the week 9 time point (8 weeks after CART123-CD20, 4 weeks after rituximab). (D) FC analysis of CD45-APC+ CD3-PacificBlue+ CAR T cells in peripheral blood of JMML117 PDX mice treated with CART123-CD20 without and with subsequent rituximab. Total CAR T-cell counts (blue gate) in murine blood normalized by quantitative counting beads are markedly reduced following rituximab treatment, although T-cell elimination is less rapid than with alemtuzumab treatment, as in Figure 3. Note that detectable CART123-CD20 decreased over time due to ongoing AML clearance from peripheral blood. (E) Immunohistochemical (IHC) staining of harvested murine spleen tissues, as in panel C, for CD33+ human AML and CD3+ CAR T cells at week 9 time point (8 weeks post-T cells) in JMML117 PDX model. Slides were scanned at ×20 magnification with an Aperio Scanscope CS-O slide scanner with visualization via the Aperio Image Analysis Toolkit (Leica Biosystems).

Figure 6.

Effective termination of CART123 allows subsequent engraftment of human bone marrow cells. (A) Experimental schema: NSGS mice were conditioned with busulfan 30 mg/kg IP prior to IV injection of 2.5 × 106 T-cell–depleted (TCD) BMMCs from a healthy normal male donor (ND). Human hematopoietic engraftment was confirmed 2 weeks later by FC analysis of murine peripheral blood (PB). CD123-redirected CAR T cells were generated from the same donor. Xenografted mice were treated with 1 dose of saline, UTD, CART123, or CART123-CD20 IV (n = 5 mice per cohort). Two weeks following T cells, mice were bled to confirm reduction/ablation of myeloid cells by CART123-CD20. Mice subsequently received 1 dose of 10 μg rituximab (RTX) IP to deplete CAR T cells and were assessed by serial peripheral blood analysis to confirm eradication of T cells. Mice were then injected with 2.5 × 106 T-cell–depleted BMMCs IV from a healthy female donor. Four weeks after infusion of the female-origin bone marrow, mice were euthanized and bone marrow was harvested for analysis of tissues. (B) Analysis of murine peripheral blood 2 weeks after T-cell treatment demonstrates significant expansion of CD3+ T cells in mice treated with CART123, CART123-CD20, or UTD, but not with saline. (C) Analysis of murine peripheral blood 2 weeks after T-cell treatment demonstrates significant reduction of human CD3-PE-Cy7/CD19-APC double-negative myeloid cells in mice treated with CART123-CD20 or CART123 versus UTD or saline controls. (D) At study end point (3 weeks following injection of female BMMCs), xenograft mice treated with CART123-CD20 and rituximab ablation demonstrate peripheral blood engraftment of myeloid cells and differentiation into monocytes (E), whereas mice treated with CART123 cells rejected engraftment of female donor BMMCs. (F) At study end point, FISH analysis of bone marrow from xenograft mice treated with CART123-CD20 shows engraftment of second donor female-origin (XX) hematopoietic engraftment. Conversely, mixed male-origin (XY) and female-origin hematopoietic engraftment is observed in saline-treated control mice, and previously, CART123-treated animals rejected second transplantation with female BMMCs. Data are representative of 2 independent experiments.