A novel full-human CD22-CAR T cell therapy with potent activity against CD22low B-ALL

Dear Editor, CD22-targeted chimeric antigen receptor (CD22-CAR) T cells have been proven to be effective in treating patients with B acute lymphoblastic leukemia (B-ALL) who were unsuitable to receive CD19-CAR T cell therapy. However, a considerable proportion of patients still relapsed after CD22-CAR T cell therapy, with diminished or decreased levels of CD22 expression on blasts. For patients who had not completely lost CD22 expression on blasts, CD22-CAR T cell re-treatment may apply as a salvage regimen. But unfortunately, in a previous trial on a humanized CD22-CAR (termed CD22CAR) T cell therapy, a second CD22-CAR infusion produced no or suboptimal anti-leukemia response and CAR T cell expansion (Table S1 and Fig. S1). Because the immunogenicity of humanized antibodies or CARs has been suggested in previous studies, the poor effect of the second CD22-CAR infusion might be due to the patient’s immune response against CD22-CAR transgene. On the other hand, CD22 downregulation might also represent a mechanism of resistance to a second CAR T-cell therapy. Therefore, a new CD22-CAR with no cross immunogenicity with CD22-CAR, and with strong activity against CD22 cells, may be effective for the second treatment of patients who failed from previous CD22-CAR T cell therapy. Furthermore, since the immunogenicity of full-human antibodies tends to be reduced compared with humanized or chimeric constructs, the usage of fully humanderived CAR constructs might be a better strategy, which can further lower the risk of developing immune responses against the secondarily infused CAR T cells. This study aims to develop a new CD22-CAR construct with low immunogenicity and potent activity for treating B-ALL patients who failed from prior CD22-CAR T cell therapies. Full-human anti-CD22 single chain fragment variant(scFv)s were screened from a full-human scFv yeast display library. The screened anti-CD22 scFvs were fused to the intracellular 4-1BB co-stimulatory and CD3ζ signaling domains to create a panel of CD22-BBz variants (Fig. 1a). The activities of the different CD22-BBz variants were tested with NFAT reporter assay in Jurkat cells in response to CD22 Raji, CD22 JVM-2, and CD22 K562 cells (Fig. 1b), and CD22-BBz 80, 27, and 36 were identified as constructs that could transmit strong antigen-specific activation signals in T cells (Figs. 1c and S2). We then evaluated the effector function of different CD22-CAR variants via CD107a degranulation and cytotoxicity assay and identified CD22-BBz 80 with superior effector activity against CD22 target cells (Fig. 1d, e). The membrane proteome array (MPA) showed that CD22-BBz 80 had a high specificity to the target antigen (Fig. 1f). The in vivo anti-leukemia effect of CD22-BBz 80 was confirmed in NPG mice injected with 1 × 10 Nalm6-Luc cells (Fig. S3). Thus, CD22-BBz 80 (termed CD22-CAR) was identified to have potent and antigen-specific anti-leukemia activity and was used in the subsequent clinical study. A single-center, open-label, phase I clinical trial was conducted to evaluate the safety and efficacy of CD22-CAR T cells in 8 children with CD22 or

A novel full-human CD22-CAR T cell therapy with potent activity against CD22 low B-ALL Yue Tan 1,2 , Haodong Cai 3 , Chuo Li 2 , Biping Deng 4 , Weiliang Song 5 , Zhuojun Ling 5 , Guang Hu 6 , Yongkun Yang 6 , Panpan Niu 6 , Guangrong Meng 6 , Wei Cheng 6 , Jinlong Xu 5 , Jiajia Duan 5 , Zelin Wang 5 , Xinjian Yu 7 , Xiaoming Feng 2,8 , Jianfeng Zhou 3 and Jing Pan 1 Dear Editor, CD22-targeted chimeric antigen receptor (CD22-CAR) T cells have been proven to be effective in treating patients with B acute lymphoblastic leukemia (B-ALL) who were unsuitable to receive CD19-CAR T cell therapy [1][2][3] . However, a considerable proportion of patients still relapsed after CD22-CAR T cell therapy 3,4 , with diminished or decreased levels of CD22 expression on blasts. For patients who had not completely lost CD22 expression on blasts, CD22-CAR T cell re-treatment may apply as a salvage regimen. But unfortunately, in a previous trial on a humanized CD22-CAR (termed CD22-CAR YK002 ) T cell therapy, a second CD22-CAR YK002 infusion produced no or suboptimal anti-leukemia response and CAR T cell expansion (Table S1 and Fig. S1). Because the immunogenicity of humanized antibodies or CARs has been suggested in previous studies [5][6][7][8] , the poor effect of the second CD22-CAR YK002 infusion might be due to the patient's immune response against CD22-CAR YK002 transgene. On the other hand, CD22 downregulation might also represent a mechanism of resistance to a second CAR T-cell therapy 3 . Therefore, a new CD22-CAR with no cross immunogenicity with CD22-CAR YK002 , and with strong activity against CD22 low cells, may be effective for the second treatment of patients who failed from previous CD22-CAR YK002 T cell therapy.
Furthermore, since the immunogenicity of full-human antibodies tends to be reduced compared with humanized or chimeric constructs [9][10][11] , the usage of fully humanderived CAR constructs might be a better strategy, which can further lower the risk of developing immune responses against the secondarily infused CAR T cells.
This study aims to develop a new CD22-CAR construct with low immunogenicity and potent activity for treating B-ALL patients who failed from prior CD22-CAR T cell therapies. Full-human anti-CD22 single chain fragment variant(scFv)s were screened from a full-human scFv yeast display library. The screened anti-CD22 scFvs were fused to the intracellular 4-1BB co-stimulatory and CD3ζ signaling domains to create a panel of CD22-BBz variants (Fig. 1a). The activities of the different CD22-BBz variants were tested with NFAT reporter assay in Jurkat cells in response to CD22 high Raji, CD22 low JVM-2, and CD22 -K562 cells (Fig. 1b), and CD22-BBz 80, 27, and 36 were identified as constructs that could transmit strong antigen-specific activation signals in T cells (Figs. 1c and S2). We then evaluated the effector function of different CD22-CAR variants via CD107a degranulation and cytotoxicity assay and identified CD22-BBz 80 with superior effector activity against CD22 low target cells (Fig. 1d, e). The membrane proteome array (MPA) showed that CD22-BBz 80 had a high specificity to the target antigen (Fig. 1f). The in vivo anti-leukemia effect of CD22-BBz 80 was confirmed in NPG mice injected with 1 × 10 6 Nalm6-Luc cells (Fig. S3). Thus, CD22-BBz 80 (termed CD22-CAR FH80 ) was identified to have potent and antigen-specific anti-leukemia activity and was used in the subsequent clinical study. CD22 low B-ALL who failed in prior CD19 and CD22-CAR therapies. The median age was 9 (range, 5-16) years. Six patients (75%) had hematological relapses as confirmed by bone marrow morphology. One patient (pt 01) had persistent positive measurable residual disease (MRD + ) detected by flow cytometry and the other patient (Pt 06) was MRD − in bone marrow but had diffused extramedullary disease (Table S2). Four patients (50%; Pt 01, 04, 06, and 08) had previously undergone allogeneic hematopoietic stem cell transplantation (allo-HSCT 3 of 4 once, 1of 4 twice). All patients had received at least two lines of treatments including chemotherapies, allo-HSCT, CD19, or CD22 CAR T therapies. All patients (100%) failed from prior versions of CD22-CAR therapies (two refractory, six relapsed Fig. S4 and Table S2). The CD22 expression level on blasts was not obviously reduced in six patients after prior CD22-CAR YK002 therapies, while patients 03 and 04 displayed lower CD22 expression on blasts than that before the CD22-CAR YK002 treatments (Fig. 1g, h). After receiving lymphodepleting chemotherapy with fludarabine (30 mg/m 2 /day) and cyclophosphamide (250 mg/m 2 /day), all patients received 1 (range, 0.68 to 9.4) × 10 6 CD22-CAR FH80 T cells per kilogram body weight (/kg) (Table S3). Seven (87.5%) of eight patients had a response to CD22-CAR FH80 T cells. Six patients (75%) achieved MRDcomplete remission, including two (Pt 03, 04) with a low level of CD22 expression at enrollment; one patient (Pt 06) achieved partial remission in his extramedullary disease, but unfortunately, she succumbed to the infection on day 42 after infusion (Fig. 2a, b and Fig. S5). One nonresponse patient achieved remission after receiving inotuzumab, but died due to transplantation complication at 5 months. The six remission patients were followed up with a median time of 6 months. Of the three patients (Pt 02, 04, and 08) who received no further treatment, two (Pt 04 and 08) remained in remission for 9 and 5 months, while one (Pt 02) had a CD22 − relapse and died of tumor progression 6 months after infusion. Three patients (Pt 01, 03, and 07) were bridged to HSCT as consolidation after CAR T cell infusion (the donors and preconditioning regimen are detailed in Table S4), and among them, two (Pt 01 and 07) remained in remission 10 and 5 months after infusion, and one (Pt 03) had a relapse with a mixture of CD22 low and CD22 − blasts 3 months after HSCT (Fig. S6).
B cell aplasia and hypogammaglobulinemia could be used as a measurement of the active surveillance of CAR T cells 12 . The three patients who achieved remission and received no further treatments all exhibited B cell aplasia and hypogammaglobulinemia until the observation end point (Fig. 2e), suggesting a prolonged persistence of CAR T cells.
CRS occurred in 7/8 (87.5%) patients, including six with grade 1 CRS (75%) and 1 with grade 3 CRS (12.5%). The median time to onset was 3 days (range, 1-12 days), and the median duration was 10 days (range, 2-19 days). The non-responding patient exhibited no signs of CRS, in accordance with his minimal CAR T cell expansion. Neurologic toxicities occurred in 2/8 (25%) patients, including 1 with grade 2 (12.5%) and 1 with grade 3 (12.5%). The detailed manifestations and managements (see figure on previous page) Fig. 1 Development of a novel full-human CD22-CAR FH80 construct and leukemic CD19/CD22 expression in patients before enrolling in CD22-CAR FH80 therapy. a Schematic of the recombinant lentiviral vectors encoding the full-human CD22-BBz variants. The expression of CAR transgene is under the control of the elongation factor 1α promoter. CD8 (H/TM), CD8α hinge and transmembrane domains; EGFR, epidermal growth factor receptor; T2A, Thoseaasigna virus 2A. b Histogram of CD22 expression on Raji, JVM-2, and K562 cells, determined by flow cytometry. c Nuclear factor of activated T cells (NFAT) reporter assay in T cells transduced with different CD22-BBz variants, after co-culturing with CD22 high Raji, CD22 low JVM-2, and CD22 -K562 cells, or cultured in medium alone in triplicates. d Bar chart showing CD107a expression in specific CD22-BBz T cell clones after co-culturing with CD22 high Raji, Nalm6, Reh cells, CD22 low JVM-2, CD22 − K562, Jurkat, and U266 cells, as determined by flow cytometry. CD107a degranulation rates were calculated as the percentages of CD107a + cells among CAR + CD8 + T cells and indicated above each bar. e Cytolytic effects of specific CD22-BBz T cell clones against Nalm6 and Reh cells during co-culturing in triplicates at E:T ratios of 2:1, 1:1, 0.5:1, and 0:1. % specific lysis = (spontaneous relative light unit (RLU) − test RLU)/(spontaneous death RLU) × 100. A two-tailed, unpaired two-sample t-test was used for statistical analysis. The brown, green, and blue asterisks indicate the comparison of the cytolytic effects of CD22-BBz 80 with those of CD22-BBz 36, CD22-BBz 27, or control T cells, respectively. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. f Membrane proteome array for CD22-BBz.  of CRS, ICANS, and other toxicities suspected to be related to CAR T cells are shown in Fig. 2f, g and Table S6. We further compared the severity of CRS and ICANS between CD22-CAR FH80 and the prior CD22-CAR YK002 therapies in the same individual patients but found no significant difference (P = 0.414 and 0.285, Fig. S7). Serum cytokines indicative of systemic inflammation were detailed in Fig. S8. Most patients had dramatic increases in IL-6, ferritin, and sCD25, while only a small proportion of the patients had an obvious increase in TNF-α and IL-10. The patient (Pt 07) who developed grade 3 CRS and ICANS showed the highest peak levels of IL-6, ferritin, and sCD25. In this study, we developed a novel full-human CD22-CAR FH80 construct with superior activity against CD22 low target cells. In seven of eight B-ALL patients who were refractory or relapsed after previous CD19-and CD22-CAR T cell therapies, CD22-CAR FH80 T cell therapy exhibited potent anti-tumor activity with a manageable safety profile. The high response rate with CD22-CAR FH80 therapy implicates that there is no overt cross immunogenicity between CD22-CAR FH80 and prior infused CD22-CAR YK002 transgenes. Markedly, CD22-CAR FH80 therapy was even effective in two patients with CD22 low blasts, while in the two patients who relapsed after CD22-CAR FH80 therapy CD22 − blasts were majorly present, indicating that CD22 low leukemia cells can be efficiently eliminated by CD22-CAR FH80 T cells. Three patients without HSCT consolidation remained in B cell aplasia and hypogammaglobulinemia until the cut-off date, indicating the continued presence of CD22-CAR FH80 T cells, but the follow-up time is not long enough, and the immunogenicity and long-term persistent capability of CD22-CAR FH80 need to be further investigated. Our previous version of CD22-CAR YK002 was also very efficient in inducing remission in B-ALL 2 . Whether CD22-CAR FH80 outperforms CD22-CAR YK002 in persistence or long-term efficacy still warrants future study. In addition, a CD22-CAR containing a full-human scFv, termed m971, generated from a human phage library has been reported to produce convincing activity in B-ALL patients in a phase I trial 3 . The specific property and value of different CD22-CAR versions need to be further addressed in future studies. Although future phase 2/3 trials are needed to verify and optimize CD22-CAR FH80 therapy, our study indicates an important value of CD22-CAR FH80 as a salvage regimen for advanced B-ALL cases that are refractory to prior versions of CD22-CAR therapies.

Conflict of interest
Adimab, LLC, a global company that contributes to the discovery and optimization of fully human monoclonal and bispecific antibodies, holds a patent on full-human non-immune monoclonal antibody yeast display library. Y.Y., P.N., G.M., and W.C. are also employees of Nanjing Iaso Biotherapeutics Co. Ltd., whose potential product was studied in this work. The remaining authors declare no competing interests.
Publisher's note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.