Naturally selected CD7 CAR-T therapy without genetic editing demonstrates significant antitumour efficacy against relapsed and refractory acute myeloid leukaemia (R/R-AML)

Bone marrow biopsy samples of 3 patients with R/R-AML were obtained from the Department of Haematology, Second Hospital of Hebei Medical University. Bone marrow biopsy samples were fixed in 4% formalin (Sigma‒Aldrich, USA) at room temperature, subsequently decalcified, dehydrated, and embedded in paraffin to generate 3-µm thick tissue sections. The following immunohistochemical antibodies and reagents were obtained from Leica Microsystems (Leica, Germany). Tissue sections (3 µm thick) were subject to CD34, CD117, CD7, and MPO labelling using a Leica Bond MAX Immunostainer (Leica, Germany). The following program was employed: peroxide block for 5 min, marker labelling incubation for 15 min, post primary incubation for 8 min, polymer incubation for 8 min, DAB Refine reagent incubation for 10 min and finally haematoxylin incubation for 5 min. All the above steps were performed at room temperature. A microscope (× 400, × 100) (Olympus, Japan) was used to observe the samples.

The design and production of naturally selected CD7 CAR-T cells was performed as previously described by our group [34]. The DNA sequence of the anti-CD7 single-chain variable fragment derived from the CD7-specific mouse monoclonal antibody TH-69 was synthesized and cloned into a CAR composed of a 12-AA short hinge-only domain, CD28 transmembrane, 4-1BB, CD3ζ, T2A autocleavage sequences, and endodomain-deleted EGFR (tEGFR) [35‐37]. CD7 CAR using lentiviral vectors was produced by transfecting and concentrating 293FT cells with 20,000 g ultracentrifugation after 2 h. The lentivirus loaded with CD7 CAR was stored at – 80 [38].

Healthy human peripheral blood mononuclear cells (PBMCs) were isolated from the peripheral blood by Ficoll density centrifugation. CD3⁺ T cells were obtained from healthy human PBMCs using CD3 microbeads (Miltenyi Biotec, Germany) and then activated for 48 h using CTS CD3/CD28 Dynabeads (Gibco, USA) in TexMACS (Miltenyi Biotec, Germany) supplemented with 200 IU of IL-2 (Sigma‒Aldrich, USA). Two days later, the CD3⁺ T cells were transduced with the CD7 CAR lentivirus. Three days after transduction, the proportion of CAR-positive (CAR⁺) T-cells was measured by flow cytometry. CD7 CAR-T-cells were continuously cultured in TexMACS (Miltenyi Biotec, Germany) with 200 IU/ml IL-2 (Sigma‒Aldrich, USA) until Day 10 post-transduction, and various in vitro functional assays or in vivo injections of CD7 CAR-T cells were performed. Initial culture numbers of CD7 CAR-T cells were increase for injection.

The following antibodies were sourced from BioLegend, USA. Harvested cells were washed twice with DPBS (BI, Israel) with 2% FBS (Oricell, China) and subsequently incubated with labelled fluorescently conjugated antibodies (anti-human CD7-APC, CD4-PE-cy7, CD8-PB, CD3-APC-cy7, PD-1-PB, TIM-3-PE, LAG-3-APC, CD45RO-Percp, CCR7-PE, and ERB-FITC) for 20 min at 4 °C in the dark according to the assay schedule. The percentage of CD7 CAR-T cells was determined by biotinylated Erbitux (ERB) [34, 37]. Finally, all flow cytometry data were obtained in a MACSquant (Miltenyi Biotec, Germany) and analysed with FlowJo software (V10.8.1).

The following experimental flow cytometry antibodies were purchased from BD Bioscience, USA. All bone marrow samples were obtained from the Department of Hematology, Hebei Medical University. Bone marrow blood from patients with R/R-AML was analysed for CD45-KO, CD117-PE, CD34-APC, and CD7-A700 by flow cytometry. When levels greater than those noted the isotype control group were observed, the expression was considered positive. Then, specimens from HDs and R/R-AML patients were labelled with CD45-V500, CD14-APC, CD16-FITC, CD3-FITC, CD19-APC, and CD7-PE, and the proportion cells with CD7 expression in normal cell populations was detected by flow cytometry (Beckman Navios, USA). Finally, Kaluza software (Beckman Coulter, USA) was used to analyse flow cytometry data.

The cell lines KG-1a, MOLM-13, CCRF-CEM, and K562 were purchased from ATCC. These cells were expanded according to ATCC recommendations. A lentiviral vector was employed to overexpress CD7 in K562 cells, and K562-CD7 cells were obtained. First, target cells were added to CFSE working solution (Sigma‒Aldrich, USA) and incubated for 20 min at 37 °C in the dark. Subsequently, a total of 1 × 10⁵ KG-1a, MOLM-13, CCRF-CEM, K562, and CD7-overexpressing K562 (K562-CD7) cells were cocultured with CD7 CAR-T cells or nontransduced T cells in RPMI 1640 media (Gibco, USA) for 4 h according to the various effector and target ratios (E: T). At the end of the four-hour coculture, all cells were stained for 7-AAD-PerCP and annexin-V-APC (BioLegend, USA) and then analysed by flow cytometry (Miltenyi Biotec, Germany).

PBMC samples from 3 R/R-AML patients from the Department of Hematology, Second Hospital of Hebei Medical University were used for the primary AML cell cytotoxicity assay. Naturally selected CD7 CAR-T or nontransduced T cells were cocultured with PBMCs of R/R-AML patients according to E:T (1:1, 2:1, 4:1) for 4 h at 37 °C. Then, all cells were harvested and washed once with DPBS (BI, Israel) and labelled with fluorescent antibodies against CD45-V510, CD34-PE, CD7-PE-cy7, CD117-BV421, 7-AAD-Percp, and Annexin-V-APC (Biolegend, USA) for 10 min at 4 °C in the dark. Finally, flow cytometry was used for detection (Miltenyi Biotec, Germany).

The AML cell line KG-1a was coincubated in RPMI 1640 with NTR and naturally selected CD7 CAR-T cells at E: T = 1:1. At the end of overnight coculture, the cells were harvested by centrifugation to retain the supernatant. The LEGENDplex Multianalyte flow assay kit (BioLegend, USA) was used according to the instructions, and the data were analysed using LEGENDplex software (BioLegend, USA) as previously described [38].

Based on previously reported studies, we designed an AML xenograft model to study the in vivo effects of naturally selected CD7 CAR-T cells [15, 16]. Six- to 8-week-old nonobese diabetic (NOD)–Prkdcscid-Il2rgem1 (NTG) female mice were purchased from SPF (Beijing) Biotechnology Co. (China) and cared for at the Animal Center of the Fourth Hospital of Hebei Medical University. All in vivo studies were performed in compliance with the Animal Ethics Committee of the Fourth Hospital of Hebei Medical University. The MOLM-13-GFP-luciferase reporter cell line (MOLM-13-GFP-Luc) was constructed to assess in vivo tumour kinetics. Mice were injected with 5 × 10⁵ MOLM-13-GFP-Luc cells through the caudal vein, and the monitoring was performed every 5 days. The IVIS imaging system was used to monitor the construction of acute myeloid leukaemia models. The AML model was successfully constructed. Then, the CD7 CAR group was injected with 1 × 10⁶ CAR⁺ T-cells. The NTR group was injected with the same number of nontransduced T cells, and the vehicle group was injected with the same dose of DPBS (BI, Israel). The tumour burden of mice injected intraperitoneally with 150 mg/kg of D-luciferase at the indicated time was monitored using the IVIS imaging system (Berthold LB983 NC100, Germany) by recording bioluminescence.

The data are presented as the means ± SEMs. Statistically significant differences between samples was determined by an unpaired two-tailed Student’s t test and in multiple comparisons by a one-way ANOVA with postTukey’s tests. If the variance was not equal, then the unpaired two-tailed Student’s t test with Welch’s correction was employed. All P values were calculated using Prism 8 software (GraphPad). The significance of the results are defined as follows: ns, not significant, p > 0.05; *p < 0.05; **p < 0.01; and ***p < 0.001 and ****p < 0.0001.

The distribution of CD7 expression determines the antitumour effect and off-target toxicity of CD7 CAR-T cells. We examined CD7 expression in T cells, B cells, monocytes, and neutrophils from normal cell subsets from HDs. CD7 was highly expressed in T-cell subsets but expressed at low levels or absent in other myeloid subsets. CD7 expression was further determined in normal cell subsets from R/R AML patients and showed similar expression to HDs (Fig. 1A–E). We summarized the expression of CD7 in the MRD-positive reports of R/R-AML patients from July 1, 2022 to January 1, 2021 in the Department of Hematology, the Second Hospital of Hebei Medical University. The results suggested that 5 of the 16 MRD-positive patients (31.25%) exhibited different proportions of CD7 expression, and this finding was also demonstrated by immunohistochemistry (Fig. 1F–G) (Table 1). These results support the use of CD7 as a therapeutic target.

To determine whether normal T cells can redirect, specifically recognize and attack CD7-expressing AML malignant cells upon loading CD7 CAR. We designed a CAR structure including CD7-specific single-chain variable fragments from CD7 hybridoma antibody TH-69, CD28 transmembrane region, 4–1BB, CD3ζ, and T2A autocleavage sequences as well as endodomain-deleted EGFR (tEGFR) as a CAR expression detection marker for CD7 CAR-T cells [39, 40] (Fig. 2A). We measured the expansion rate of CD7 CAR-T cells at a variety of time points in vitro and found that CD7 CAR-T cells had a lower proliferation ratio than NTR cells (Fig. 2B). Reduced viability of CD7 CAR-T cells was noted compared with nontransduced T cells, but the lowest viability recorded was greater than 78.3% on post transduction Day 7 and 80.7% on post transduction Day 12 (Fig. 2C). We used lentivirus-loaded CD7 CARs to infect normal human T cells and assayed the percentage of CAR⁺ T cells by flow cytometry on post transduction Days 4 and 12. The results indicated that CAR⁺ T cells included a high proportion of CD3⁺ T cells, with the lowest proportion recorded at 83.6% on post transduction Day 4 and 90.8% on Day 12. These values were considerably increased compared with those noted in the control group (Fig. 2D–F). To continuously monitor CD7 expression of CAR-T cells, we measured the percentage of CD7 by flow cytometry on Days 4 and 12 after transduction, and the results showed that CD7 CAR-T cells were mainly dominated by the CD7⁻ T-cell subpopulation either Day 4 or Day 12 after transduction (Fig. 2G–I). As described in previous studies, this finding may imply specific binding of CD7 CAR-T cells to the CD7⁺ cell population, thereby eliminating the CD7⁺ T-cell subpopulation from the CD7 CAR + T-cell population[34].

The memory phenotype of T cells is critical for CAR-T antitumour efficacy and persistence [41, 42]. According to previous studies, T_CM cells were defined by CD45RO and CCR7 double positivity [43, 44]. We analysed the cell subpopulations of CD4⁺ T and CD8⁺ T cells of naturally selected CD7 CAR-T and NTR cells, respectively. The results showed that the CD7 CAR and NTR groups did not exhibit differences in the proportion of naïve T cells (T_N) (CD45RO⁻CCR7⁺) and effector memory T-cells (T_EM) (CD45RO⁺CCR7⁻) among CD4⁺ and CD8⁺ T-cell subpopulations. However, a significant increase in the proportion of central memory phenotypes (T_CM) (CD45RO⁺CCR7⁺) was noted in the CD7 CAR group compared to the NTR group. We also failed to observe accelerated terminal differentiation due to the death of naturally selected CD7 CAR-T cells, as there was no difference in effector-like T cells (T_Eff)(CD45RO⁻CCR7⁻) in the CD4⁺ T-cell subpopulation. Even in the CD8⁺ T-cell subpopulation, the proportion of T_Eff was lower in the CD7 CAR group than in the NTR group (Fig. 3A, B).

In addition, we assessed depletion markers in naturally selected CD7 CAR-T cells, and the results suggested a trend of increased programmed cell death protein 1 (PD-1) expression in the CD7 CAR group but, no significant difference was noted compared with the NTR group. Increased T-cell immunoglobulin and mucin domain containing-3 (TIM-3) expression was noted and significantly different from that noted in the NTR group. LAG-3 expression in the CD7 CAR group was not significantly different from that noted in the NTR group, but a decreasing tendency was observed (Fig. 3C–F).

The target specificity of CAR-T allows it to kill target-positive cell populations [45]. We chose to measure the CD7 expression levels in the acute myeloid leukaemia cell lines MOLM-13, KG-1a, and K562; CD7-overexpressing K562 (K562-CD7) cells; and the CD7-positive human T-lymphocyte leukaemia cell line CCRF-CEM. The results showed that all cell lines expressed CD7, except K562 (Fig. 4A). To verify the antileukaemic effect of CD7 CAR-T cells in vitro, we used the above cell lines as target cells and set different E:T ratios. We found that CD7 CAR-T cells could rapidly eliminate CD7⁺ cell lines at various dose ratios (Fig. 4B, D–H).

To further validate the cytotoxicity of naturally selected CD7 CAR-T cells against R/R-AML, CD7 CAR-T cells or nontransduced T cells were cocultured with 3 samples from R/R-AML patients. We found that after 4 h of coculture, CD7⁺ AML blasts were significantly reduced in the CD7 CAR-T group compared to the NTR group at any E:T ratio (Fig. 4C, I).

In addition, the naturally selected CD7 CAR-T cells exhibited robust IL-2 and IFN-γ secretion in response to coculture with CD7⁺ AML cell line KG-1a compared to the NTR group (Fig. 4J).

These results demonstrate the potent cytotoxic effect of naturally selected CD7 CAR-T cells on CD7-positive AML cells in vitro.

Previous in vitro studies have demonstrated the potent antitumour ability of CD7 CAR-T cells. To further validate the clearance of acute myeloid leukaemia cells by CD7 CAR-T cells in vivo. We injected 5 × 10⁵ MOLM-13-GFP-Luc cells into the tail vein of nonobese diabetic (NOD)–Prkdcscid-Il2rgem1 (NTG) mice to generate a xenograft model. IVIS imaging was performed every 5 days after tumour cell infusion to detect tumour implantation, and successful AML implantation was detected at the end of the first 5-day period. Subsequently, 1 × 10⁶ CD7 CAR-T cells or NTR cells were injected, whereas the vehicle group was injected with an equal dose of DPBS buffer (Fig. 5A, B). No mice died during the 34 day monitoring period, and no evidence of graft-versus-host disease (GVHD) was observed. Compared to the NTR group, bioluminescence imaging (BLI) data indicated suppression of leukaemic cells on Day 7 after CD7 CAR-T injection in the CD7 CAR group followed by a rapid decrease in tumour burden, which could not be detected in mice of the CD7 CAR group at 22 days (Fig. 5B, C).