CRISPR / Cas-mediated non-viral genome specific targeted CAR T 1 cells achieve high safety and efficacy in relapsed / refractory B-cell 2 non-Hodgkin lymphoma 3

30 In recent years, chimeric antigen receptor (CAR) T cell therapy has shown great 31 promise in treating hematological malignancies. However, CAR T cell therapy 32 currently has several limitations. Here we successfully developed a two-in-one 33 approach to generate non-viral genome specific targeted CAR T cells through 34 CRISPR/Cas9. Based on the optimized protocol, the feasibility was preliminarily 35 demonstrated by a preclinical study inserting an anti-CD19 CAR cassette into the 36 AAVS1 safe harbor locus. We found that non-viral AAVS1-knockin CAR T cells 37 behave comparably to those conventionally produced by lentivirus. Furthermore, an 38 innovative type of anti-CD19 CAR T cells with PD1-integration was constructed and 39 shows a superior ability to eradicate tumor cells with high PD-L1 expression. In 40 adoptive therapy for relapsed/refractory (r/r) aggressive B-cell non-Hodgkin 41 lymphoma (B-NHL), we observed a high rate (87.5%) of complete remission (CR) 42 and durable responses without serious adverse events in eight patients after treatment. 43 Notably, these enhanced CAR T cells were effective even at a low infusion dose and 44 with a low CAR percentage, which indicated that they have higher potency. No 45 off-target events were found in the infusion product. Single-cell RNA sequencing 46 analysis further validated the advantage of PD1 interference that results in fewer 47 dysfunctional CAR T cells through this treatment. Collectively, our results 48 demonstrate the outstanding safety and efficacy of non-viral genome specific 49 integrated CAR T cells, thus providing a revolutionary technology for CAR T cell 50 therapy. 51 52 Introduction 53 In recent years, CAR T cell therapy has rapidly developed and shows a great potential 54 in cancer therapy, which is exemplified by the FDA approval of four anti-CD19 CAR 55 T cell treatments. Nevertheless, there still remain some limitations, including the 56 complicated manufacturing process, high production cost, long preparation time and 57 potential safety concerns of current therapies. The use of virus in CAR T cell 58 production is one area of concern, as the disadvantages include that insertional 59 mutagenesis increases the risk of tumor development. Furthermore, specific 60 responses to virus-derived DNA tend to impede CAR expression and virus 61 manufacture frequently incurs high costs. Although some strategies, such as using 62 transposon systems and mRNA transduction, are being exploited to generate 63 CAR T cells without virus, the low homogeneity of final products caused by random 64 integration and discontinued CAR expression become additional problems. Recently, 65 several studies have shown that CRISPR/Cas9 technology can be applied to generate 66 locus specific integrated CAR T cells by using an adeno-associated virus (AAV) 67 vector as a template. Furthermore, one preferential non-viral strategy was 68 proposed to produce T cell products with point mutation correction and precise 69 insertion of the TCR element. Thus, in order to simultaneously solve the 70 disadvantages of virus usage and random integration, here we further optimized the 71 conditions and developed non-viral genome specific targeted CAR T cells through 72 CRISPR/Cas9. The feasibility was preliminarily demonstrated by preclinical 73 experiments using AAVS1-targeted anti-CD19 CAR T cells. Given that blockage of 74 the PD1/PD-L1 pathway by inhibitors or gene editing has been reported to improve 75 the antitumor activity of CAR T cells, we generated enhanced PD1-integrated 76 anti-CD19 CAR T cells and demonstrated their safety and effectiveness in treating 77 patients with r/r B-NHL. 78


53
In recent years, CAR T cell therapy has rapidly developed and shows a great potential 54 in cancer therapy, which is exemplified by the FDA approval of four anti-CD19 CAR 55 T cell treatments 1-5 . Nevertheless, there still remain some limitations, including the 56 complicated manufacturing process, high production cost, long preparation time and 57 potential safety concerns of current therapies. The use of virus in CAR T cell 58 production is one area of concern, as the disadvantages include that insertional 59 mutagenesis increases the risk of tumor development 6,7 . Furthermore, specific 60 responses to virus-derived DNA tend to impede CAR expression 8,9 and virus 61 manufacture frequently incurs high costs 10 . Although some strategies, such as using 62 transposon systems [11][12][13][14] and mRNA transduction 15,16 , are being exploited to generate 63 CAR T cells without virus, the low homogeneity of final products caused by random 64 integration and discontinued CAR expression become additional problems. Recently,65 several studies have shown that CRISPR/Cas9 technology can be applied to generate 66 locus specific integrated CAR T cells by using an adeno-associated virus (AAV) 67 vector as a template 17,18 . Furthermore, one preferential non-viral strategy was 68 proposed to produce T cell products with point mutation correction and precise

80
First, we sought to optimize the protocol for producing non-viral genome 81 specific integrated T cells. It was found that a homology directed repair (HDR) 82 template, in the form of linear double-stranded DNA (dsDNA), could achieve high 83 recombination efficiency and cell viability (Figure 1a, S1a-c). More viable integrated 84 cells were acquired when electroporation was carried out in stimulated T cells by 85 applying 800bp homology arms (Figure 1b, c, S1d-g, S2). After confirmation of an 86 optimal protocol, for proof of concept, we first chose to introduce the CAR targeting 87 construct into the AAVS1 safe harbor, which excludes the influence caused by 88 functional endogenous genes, to evaluate whether this approach would affect the 89 properties of CAR T cells. An anti-CD19 CAR sequence was constructed, which was 90 comprised of the intracellular domain of 4-1BB and CD3ζ(named as 19bbz). The 91 integration efficiency of 19bbz into AAVS1 was about 10% (up to 19.80%) and the 92 indel percentage ranged from 67% to 87% in healthy donor cells (Figure 1d, e, S3a). 93 Also, the integration was unbiased between bulk CD3 + , CD4 + and CD8 + T cells  Figure S4). Taken together, these results demonstrate that the strategy to 111 produce non-viral genome specific targeted CAR T cells is feasible.

112
Due to the well-known inhibition of T-cell effector function through the 113 PD1/PD-L1 pathway, we set out to develop an enhanced type of CAR T cells by 114 integrating an anti-CD19 CAR sequence into the PD1 gene (named as .  Table S1). Indel events were also not detected 139 at 29 top-ranked potential off-target sites predicted by the Benchling CRISPR tool, by 140 using deep sequencing analysis (Table S2).

141
Eight patients were given a lymphodepleting chemotherapy regimen using 142 combined cyclophosphamide and fludarabine, followed by one infusion of PD1-19bbz 143 cells with a dose of 0.56×10 6 -2.35×10 6 cells/kg body weight (Table 1, Table 1). Partial remission (PR) was observed in the remaining (1/8) 155 patient, thus the best objective response rate reached 100% in all the patients. Of note, 156 PD1-19bbz cells effectively functioned even at a low infusion dose and with a low 157 CAR percentage, thereby indicating high potency of these PD1 knockout CAR T cells.

158
Together, these data demonstrate that non-viral PD1-integrated CAR T cells have 159 high safety and efficacy for patients with r/r B-NHL.  suggesting that PD1-19bbz cells had a lower tendency to become exhausted in vivo. 181 The activities of different pathways were also analyzed in the samples ( Figure S14).

182
Altogether, these scRNA-seq data reveal more memory and fewer dysfunctional 183 CAR + cells in pre-infused and post-infused PD1-19bbz cells, thus giving a 184 mechanistic explanation for their superior efficacy in the clinical trial. have not, indeed, found any indel events using WGS and deep sequencing analyses, 200 thus mitigating the safety concern of genome editing. During the process from bench 201 to production, an unexpected lower CAR recombination efficiency in two infusion 202 products (patient-1, patient-4) and low PD1 indel percentage in one infusion product 203 (patient-1) were detected. The reason is attributed to the early premature 204 manufacturing process, which has been solved, rather than individual variance or low 205 reproducibility of method. Taken together, we demonstrate the feasibility of formal 206 large-scale production of non-viral genome specific targeted CAR T cells for clinical 207 application. 208 We are the first to demonstrate the safety and efficacy of non-viral genome 209 specific targeted CAR T cells in a clinical trial. Relative to conventional CAR T cell 210 therapies 37-39 , we found superior safety for patients with r/r B-NHL by using non-viral  In this study, we describe an innovative strategy to develop non-viral genome 238 specific targeted CAR T cells by CRISPR/Cas9. This technology is advanced due to 239 combining the advantages of both non-viral manufacturing processes and precise 240 genome editing. As a two-in-one approach without using virus, the manufacturing 241 procedure is simplified, with shortened preparation time, reduced production expenses,     Table S5. During hospitalization, any AEs that occurred after CAR T cell infusion were recorded.   and then isolated by density gradient centrifugation using Ficoll (Sigma-Aldrich).

540
The scRNA-seq libraries were generated using the 10X Genomics Chromium