Efficient generation of CAR NK cells from human umbilical cord blood CD34+stem and progenitors for democratizing affordable immunotherapy

24 Chimeric antigen receptor (CAR) natural killer cells (CAR NK) cells, leveraging safety and not 25 requiring HLA match in adoptive infusion, have emerged as promising alternative cells to CAR-T cells 26 for immunotherapies. High and multiple doses of CAR NK cell infusions are essential to maintain 27 therapeutic efficacy in clinical trials. This requires efficient methods for generating large-scale CAR 28 NK cells and significantly reducing CAR engineering costs. In this study, we develop a three -step 29 strategy to generate highly high yields of induced NK (iNK) and CAR iNK cells from human umbilical 30 cord blood CD34 + hematopoietic stem and progenitor cells (CD34 + HSPCs). Starting from a single 31 umbilical cord blood CD34+ HSPC, our reliable method efficiently produces 14-83 million mature iNK 32 cells or 7-32 million CAR iNK cells with high expression levels of CD16 and zero T cell contaminations. 33 Introducing CAR expression elements at the HSPC level reduces the quantities of CAR pseudoviruses 34 to 1 / 140.000 - 1 / 600,000 compared to engineering CARs in mature NK cells. The iNK and CAR iNK 35 cells, including fresh cells and thawed cells from cryopreserved conditions, demonstrate remarkable 36 tumoricidal activities against various human cancer cells and significantly prolong the survival of 37 human tumor -bearing animals. The high yields of CAR NK cells and negligible cost s of CAR 38 engineering of our method support the broad applications of CAR NK cells for treating cancer patients. 39 40 41 42 43 44 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted July 30, 2024. ; https://doi.org/10.1101/2024.07.30.605741doi: bioRxiv preprint

45 Cellular immunotherapy, particularly chimeric antigen receptor (CAR) T cell therapy, has 46 revolutionized the treatment of B cell malignancies. Despite the remarkable efficacy of CAR -T cells, 47 their high costs, severe toxicities, and limited cell sources have spurred interest in developing NK cells 48 as an alternative to universal cellular immunotherapy1-3. NK cells have the inherent ability to directly 49 kill neoplastic, virally infected, and certain stressed cells due to abnormal expression of NK cell ligands 50 such as MHC class I and class I-like molecules4. NK cells possess unique advantages of low side effects, 51 such as GvHD and CRS observed in CAR -T cell therapy. CAR NK cell therapy is promising in 52 achieving dual targeting and universal nonspecific killing effects in the treatment of tumors with low 53 health risks5. 54 Human NK cells have a physiological turnover time of about two weeks in blood circulation6. The 55 persistence of allogeneic NK cells in patients ranges from a few days to a few weeks, with an average 56 of 7 days in most clinical trials 7. New methods are needed to generate abundant CAR NK cells and 57 reduce CAR engineering costs, considering the high dose (107 / kg) and multiple dose requirement for 58 CAR NK cell therapy in clinical trials to enhance efficacies8, 9. 59 CD34+ HSPCs can differentiate into multilineage blood and immune cells, including NK cells. A 60 cytokine-based culture system has produced mean output yields of 1,879-4,450 NK cells from a single 61 HSPC. However, at the end of the culture, HSPC-NK cells express deficient levels of CD16 protein 62 (3.0% ± 2.4)10, indicating low ADCC activity in the engagement of adaptive immune cells 11. Direct 63 transplantation of engineered CAR HSPCs results in the generation of CAR NK cells in vivo, with 64 derivations of other lineage CAR cells, including CAR T, CAR Mye, and CAR B cells12, which brings 65 unknown health risks. 66 CD34+ HSPC culture and expansion techniques have advanced over the past decades13, 14. Artificial 67 organoid aggregates combining seed cells with feeder cells significantly increase the generated 68 efficiencies of induced T cells and NK cells from stem and progenitor cells 15, 16Therefore, integrating 69 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted July 30, 2024. ; https://doi.org/10.1101/2024.07.30.605741doi: bioRxiv preprint HSPC expansion and organoid induction can further promote the efficiency of NK cell generation from 70 CD34+ HSPCs. 71 This study introduces a comprehensive method to potentially derive trillions of iNK and CAR iNK 72 cells from a single umbilical cord blood unit of CD34+ HSPCs. This method includes three steps: CD34+ 73 HSPC expansion (Day 0 -Day 14), NK lineage through organoid aggregates (Day 14 -Day 28), and 74 maturation of NK cells and large-scale proliferation in cell culture bags (Day 28-Day 49). Interestingly, 75 a single CD34+ cell ultimately produced 1.4 × 107 ± 0.1 × 107 iNK cells and 7.6 × 106 ± 1.2 × 106 CD19-76 CAR iNK cells on day 42, and much higher yields of iNK cells (8.3 × 107 ± 0.7 × 107) and CD19-CAR 77 iNK cells (3.2 × 107 ± 0.2 × 107) on day 49 using our technique, which significantly increased iNK cell 78 generation efficiency to 6,918-20,224 times over the previous method. In calculation, a single umbilical 79 cord blood unit of CD34 + HSPCs has the potential to produce trillions of CD19-CAR iNK cells using 80 our technique. Interestingly, our method sharply reduces the amounts of CAR pseudoviruses to 81 1/140,000-1.600,000 compared to the conventional approaches of engineering CARs into mature NK 82 cells. NK cells and CD19-CAR iNK cells, including thawed cells from cryopreserved conditions, show 83 ideal immune activities and solid tumor eradication abilities for six months . Our method strongly 84 supports the translational prospects for democratizing affordable CAR iNK cell therapy for the general 85 treatment of patients. 86 87

88 Efficient generation of iNK cells and CD19-CAR iNK cells from CD34+ HSPCs 89 We developed a three-step culture system to generate trillions of human iNK and CAR iNK cells from 90 a single umbilical cord blood unit of CD34 + HSPCs. This system integrates strategies of HSPC 91 expansion (Step I), organoid induction (Step II), and NK cell maturation and proliferation (Step III) 92 (Fig. 1a). To generate iNK CAR cells with meager cost of CAR engineering, we introduced CAR 93 expression elements in the primary CD34+ HSPC stage before any expansion, rather than in the mature 94 NK stage. CD34 + HSPCs were initially stimulated for 48 hours and then transduced with CD19-CAR 95 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted July 30, 2024. ; https://doi.org/10.1101/2024.07.30.605741doi: bioRxiv preprint pseudoviruses at multiplicity of infection (MOI) 10 through spin infection. Twelve hours after infection, 96 the transduced CD34 + HSPCs were immediately used for expansion culture. To enormously expand 97 CD34+ HSPC and CD19-CAR CD34+ HSPC (CD19-CAR HSPC), we established an expansion system 98 using AFT024 cells as feeders and HSC culture formula17-19. In detail, we seeded AFT024 cells (1 × 105 99 cells/well, 20 wells) in a 24-well plate, followed by irradiation (20 Gy) to arrest feeder growth. CD34+ 100 HSPCs or CD19-CAR HSPCs (5 × 104 cells/well, 20 wells) were seeded in irradiated AFT024 cells for 101 the first round of 7-day expansion in the presence of HSC culture medium14, 20. After the first round of 102 expansion, all cells were collected and seeded equally into six flasks (1 × 10 7 cells/flask) containing 103 irradiated AFT024 cells for the second round of 7-day expansion in the presence of HSC culture medium. 104 To efficiently drive differentiation of the NK lineage from CD34+ HSPC, we used the organoid 105 aggregate induction method we established previously16. We combined 5 × 105 expanded CD34+ HSPCs 106 or CD19-CAR HSPCs with 2.5 × 107 OP9 cells to form 50 organoids (1 × 104 expanded CD34+ HSPCs 107 or CD19-CAR HSPCs and 5 × 105 OP9 cells per organoid aggregate). These organoids were seeded in 108 transwells followed by a 14-day induction of the NK lineage. On day 28, all organoids on the membrane 109 were digested into single cells, which were immediately transferred to a 1 L cell culture bag (50 110 organoid-derived single CD45+ cells derived from organoids per bag, approximately 1 × 108 cells/bag) 111 for a 7-day culture to achieve maturation and proliferation of iNK cells. On day 35, all cells in the cell 112 culture bag were harvested and distributed into multiple cell culture bags (1 × 108 cells/bag) for another 113 seven days of proliferation of iNK cells or CAR iNK cells. 114 With the three-step culture system in the absence of conventional NK cell expansion feeders21, we 115 successfully harvested mature iNK cells and CD19 -CAR iNK cells on day 42 with more than 99.0% 116 purity (CD45+CD56+CD16+/-) and high expression levels of CD16 (37.4-64.4%) signaled the functional 117 maturation of NK of possessing antibody -dependent cytotoxicity (Fig. 1b, c) 16. The ratios of CD19 -118 CAR positive iNK cells derived from three independent donor CD34 + HSPCs after CAR engineering 119 were 58.5%, 63.9% and 64.4% respectively, with an average of 62.3% ± 3.3% (mean ± SD) (Fig. 1d). 120 In calculation, single CD34+ HSPCs without expansion (Day 0) produce 8.5 × 104 ± 2.1 × 104 iNK cells 121 and 4.9 × 104 ± 1.6 × 104 CD19-CAR iNK cells (mean ± SD, n=3). Single CD34+ HSPCs that undergo 122 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted July 30, 2024. ; https://doi.org/10.1101/2024.07.30.605741doi: bioRxiv preprint 7-day expansion (Day 7) produce 2.1 × 106 ± 0.6 × 106 iNK cells and 1.4 × 106 ± 0.5 × 106 CD19-CAR 123 iNK cells (mean ± SD, n=3). Furthermore, single CD34+ HSPCs undergoing 14-day expansion (Day 124 14) produce 1.4 × 107 ± 0.1 × 107 iNK cells and 7.6 × 10 6 ± 1.2 × 10 6 CD19-CAR iNK cells (mean ± 125 SD, n=3) (Day 7 vs. Day 0, P ＜ 0.01, Day 14 vs. Day 7, P ＜ 0.001) (Fig. 1e, f). A 14-day CD34+ 126 cell expansion step resulted in eventual output efficiencies of iNK cells and CD19-CAR iNK cells more 127 than 160 times compared to fresh HSPCs. Single cell RNA-seq analysis demonstrated that most iNK 128 cells and CD19-CAR iNK cells projected well to activated NK cells from human umbilical cord blood 129 (Fig. 1g). To prepare 1 million CAR NK cells, our method significantly reduced the quantities of CAR 130 pseudoviruses to 1 / 140,000 compared to conventional techniques for engineering CARs into mature 131 NK cells (Fig. 1h)22. Furthermore, iNK and CD19-CAR iNK cells could expand slightly in a bag-based 132 culture system for another seven days . We harvested iNK cells (Supplementary Fig. 1a) and CD19 -133 CAR iNK cells (Supplementary Fig. 1b) on day 49, which exhibited further increases in CD16 134 expression levels (iNK cell group, 67.8% ± 3.4%, CD19-CAR iNK cell group, 53.5% ± 6.2%, mean ± 135 SD, n = 3 in each group), stable expression ratios of CD19-CAR in CD19-CAR iNK cells (62.9% ± 136 5.7%, mean ± SD, n = 3) (Supplementary Fig. 1c), and much higher yields of iNK cells (8.3 × 107 ± 0.7 137 × 107, mean ± SD, n = 3) (Supplementary Fig. 1d) and CD19 -CAR iNK cells (3.2 × 107 ± 0.2 × 107, 138 mean ± SD, n = 3) (Supplementary Fig. 1e). 21 -day bag-based culture further reduced the quantity of 139 quantities of CAR pseudovirus to 1/600,000 when compared to conventional methods of engineering 140 CARs into mature NK cells (Supplementary Fig. 1f). 141 We monitored the immun e phenotypes and summarized the related ratios and cell numbers 142 throughout the process. We first analyzed the ratios of CD34+ cells during the expansion step. On day 143 7, 92.3% ± 3.6% (HSPC group, mean ± SD, n=3) cells retained CD45+CD34+ HSPC phenotypes. Even 144 with engineered CAR expression elements, 86.0% ± 2.3% (CD19-CAR HSPC group, mean ± SD, n=3) 145 cells still maintained the CD45+CD34+ HSPC phenotypes (Fig. 2a). In comparison, we did not observe 146 significant differences in CD34 expression ratios between the HSPC group and the CD19-CAR HSPC 147 group after seven days of expansion (P > 0.05). However, these ratios decreased to 71.9% ± 4.7% 148 (HSPC group, mean ± SD, n=3) and 61.0% ± 2.4% (CD19-CAR HSPC group, mean ± SD, n=3) on day 149 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted July 30, 2024. ; https://doi.org/10.1101/2024.07.30.605741doi: bioRxiv preprint 14, respectively (Fig. 2b). It is noteworthy that the expression of CD19-CAR slightly decreased CD34 150 positive cell rates after 14 days of expansion (HSPC group vs. HSPC group CD19-CAR, P < 0.05) (Fig. 151 2b). Interestingly, 1 × 106 CD34+ HSPC cells resulted in 6.4 × 107 ± 0.3 × 107 (CD34+ HSPC group) and 152 6.1 × 107 ± 0.5 × 107 (CD19-CAR HSPC group) CD34 + cells on day 7. A 14-day expansion further 153 yielded 1.1 × 109 ± 0.2 × 109 (CD34+ HSPC group) and 1.0 × 109 ± 0.1 × 109 (CD19-CAR HSPC group) 154 CD34+ cells, respectively (mean ± SD, n=3) (Fig. 1a and Fig. 2c). 155 To increase the efficiency of NK cell generation, we drove expanded CD34+ HSPC and CD19-156 CAR HSPC toward the NK cell lineage via an efficient organoid induction approach16. From day 14 to 157 day 21, the CD34+ population decreased, while the proportion of cells in the NK lineage that express 158 CD7 and CD56 gradually increased (Fig. 2d, e). On average, 50 organoids produced 1.2 × 108 ± 0.2 × 159 108 CD45+ cells (CD34+ HSPC group) and 1.2 × 108 ± 0.2 × 108 (CD19-CAR HSPC group) CD45+ cells, 160 including precursor cells and NK cells, after 2-week organoid induction (Fig. 1a and Fig. 2f). 161 Subsequently, every 50 organoids as a group were digested in single cell suspensions followed by 162 transferring them to commercial 1-L cell culture bags for 14-day NK cell maturation and proliferation16. 163 After seven days of bag culture, the NK precursor cells matured into the NK cell phenotype 164 (CD45+CD3-CD56+), which reached 97.0% ± 0.2% (iNK cell, mean ± SD, n=3) and 95.6% ± 0.4% 165 (CD19-CAR iNK cell, mean ± SD, n=3) of total CD45+ cells. After 14-day bag culture, NK cell purity 166 reached 99.6% ± 0.0% (iNK cell group, mean ± SD, n=3) and 99.4% ± 0.1% (CD19 -CAR iNK cell 167 group, mean ± SD, n=3). (Fig. 2g, h). On average, single cells of 50 organoids as input produced 6.4 × 168 109 ± 0.6 × 109 (iNK cell group) and 5.6 × 109 ± 0.3 × 109 (CD19-CAR iNK cell group) CD56 + iNK 169 cells after two weeks of maturation and proliferation of NK cells (Fig. 1a and Fig. 2i). To maintain the 170 optimal activity of NK cells, we collected iNK cells and CD19-CAR iNK cells on day 42. 171 Unlike the traditional NK cell expansion method using human tissue -derived NK cells as input, 172 bringing risks of T cell contamination, our method produced over 99.0% pure NK cells with zero T cell 173 contaminations (Fig. 2g). Notably, during the bag culture period, we observed significant increases in 174 CD16 positive iNK cells, rising from 44.2% to 56.7% (iNK cells) and 35.6% to 48.0% (CD19 -CAR 175 iNK cells) (Fig. 2g). Subsequently, we monitored the changes in the dynamic ratio of CD19-CAR 176 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted July 30, 2024. ; https://doi.org/10.1101/2024.07.30.605741doi: bioRxiv preprint expression throughout the CD19 -CAR iNK cell induction process. We observed comparable CD19 -177 CAR expression ratios on day 7 (27.9% ± 3.4%, mean ± SD, n=3), Day 14 (27.2% ± 3.7%, mean ± SD, 178 n=3), and day 21 (28.3% ± 2.2%, mean ± SD, n=3), with a slight increase on day 28 (33.4% ± 3.5%, 179 mean ± SD, n=3) (Fig. 2j, k). Interestingly, CD19-CAR expressing cell ratios increased significantly on 180 day 35 to 59.6% ± 2.9% (mean ± SD, n=3) and reached 62.3% ± 3.3% on day 42 (mean ± SD, n=3), 181 which were comparable to expression ratios in engineered UCB-NK cells (61.3% ± 8.2%, mean ± SD, 182 n=3) (Fig. 1h and Fig. 2j, k). To further ensure CD19-CAR expressing cell ratios of more than 90% in 183 the final collection of CD19-CAR iNK cells on day 42, an additional step of enrichment of CD19-CAR+ 184 CD34+ HSPCs on day seven during the HSPC expansion stage is required (Supplementary Figs. 2a-f). 185 Collectively, using our three-step induction strategy, a single CD34+ cell ultimately produced 1.4 186 × 107 ± 0.1 × 107 iNK cells and 7.6 × 106 ± 1.2 × 106 CD19-CAR iNK cells on day 42, and much higher 187 yields of iNK cells (8.3 × 107 ± 0.7 × 107) and CD19-CAR iNK cells (3.2 × 107 ± 0.2 × 107) on day 49. 188 A single umbilical cord blood unit of CD34+ cells (>1 × 106) roughly has approximately the capacity to 189 deliver 1.4 × 1013 - 8.3 × 1013 mature iNK or 7.6 × 1012 – 3.2 × 1013 CAR iNK cells, which prospectively 190 ensures 760 to 83,000 doses (109-1010 cells per dose) for treating patients. 191 192 Molecular features and immune activities of CD34+ HSPC-derived iNK cells 193 The activation state and functionalities of NK cells are determined by the expression patterns of 194 activating and inhibitory receptors (Fig. 3a). As expected, iNK cells expressed classical NK activating 195 receptors, including CD319, NKp30, NKp44, NKG2D and CD69, and NK inhibitory receptors, 196 including CD94, NKG2A and CD96 23, 24. Furthermore, iNK cells also highly expressed critical NK 197 effector molecules, including apoptosis-related ligands TRAIL and FasL25 (Fig. 3b). Subsequently, we 198 analyzed CD107a 26, a typical membrane protein associated with the cytotoxic activity of NK cells, 199 along with tumor necrosis factor alpha (TNF-α)27, interferon gamma (IFN-γ), Perforin, and Granzyme 200 B (GZMB) 28. After exposure to PMA/ionomycin, CD107a, TNF -α, and IFN -γ were upregulated in 201 UCB-NK and iNK cells. Phenotypic analysis revealed that both UCB-NK cells and iNK cells exhibited 202 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted July 30, 2024. ; https://doi.org/10.1101/2024.07.30.605741doi: bioRxiv preprint significant expression of cytotoxic granules, such as perforin and GZMB (Fig. 3c). Furthermore, 203 statistical analysis did not show significant differences in protein expression patterns between UCB -204 NK cells and iNK cells (Fig. 3d). 205 Unlike adaptive T cells, NK cells possess an innate ability to recognize and eliminate tumor cells 206 exhibiting reduced MHC-I expression29. Therefore, we first evaluated the tumoricidal capacity of iNK 207 cells using various cancer cell lines, including the human erythroleukemia cell line K562, the human 208 Caucasian promyelocytic leukemia cell line HL60, the human B lymphoblastic leukemia cell line Nalm-209 6, the human ovarian cancer cell line A1847, the human acute myeloid leukemia cell line MOLM-13, 210 and the Epstein-Barr virus-negative B cell lymphoma cell line THP1 (Fig. 3e). The iNK cells showed 211 similar tumor killing activities to those of UCB-NK cells at different E: T ratios (Fig. 3f-i). At lower E: 212 In the T ratios, iNK cells exhibited superior cytotoxicity against MOLM -13 and THP1 (Fig. 3j,k). To 213 evaluate the persistent cytotoxic activity of iNK cells, we performed three rounds of tumor-killing 214 assays using K562 tumor cells at an effector -to-target (E: T) ratio of 1:1 (Fig 3l). iNK cells 215 consistently demonstrated robust cytotoxicity, similar to the observation of UCB-NK cells in the first- 216 and second-round killing assays. Intriguingly, iNK cells , but not UCB -NK cells, showed superior 217 cytotoxicity in the third round killing assay (Fig. 3m, P < 0.05). Furthermore, we analyzed the 218 expression patterns of the NK cell classical activation markers (2B4, DNAM-1, and NKp46) before and 219 after three rounds of tumor cell killing tests (Fig. 3n) 30The data revealed that both UCB-NK cells and 220 iNK cells maintained high expression levels of these three activity receptors. It is noteworthy that iNK 221 cells exhibited higher mean fluorescence intensities (MFI) for DNAM -1 regardless of the absence or 222 presence of tumor cells (Fig. 3o). Collectively, CD34+ HSPC-derived iNK cells express typical markers 223 related to NK cell function and possess the ability to kill broad-spectrum tumors. 224 225 CD34+ HSPC-derived iNK cells suppressed tumor growth in A1847 tumor-bearing mice 226 To evaluate the in vivo therapeutic efficacy of HSPC-derived iNK cells, we established a human tumor 227 cell line -derived xenograft model by intraperitoneal injection of luciferase -expressing A1847 cells 228 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted July 30, 2024. ; https://doi.org/10.1101/2024.07.30.605741doi: bioRxiv preprint (A1847-luci+, 2.0 × 10 5 cells /mouse) into NCG mice (NOD/ShiLtJGpt -Prkdcem26Cd52Il2rgem26Cd22/Gpt 229 strain) on day -1. We infused iNK cells (1.5 × 107 cells / dose, i.p.) twice into the tumor-bearing animals 230 on day 0 and day 7. In parallel, we used UCB-NK cells as a treatment control. Weekly bioluminescence 231 imaging (BLI) was performed to capture the dynamics of tumor burden (Fig. 4a). As expected, iNK 232 cells effectively suppressed tumor growth in vivo, demonstrating tumor-killing efficiencies comparable 233 to those of UCB-NK cells (Fig. 4b, c). On the contrary, tumor-only animals exhibited progressive 234 intensification of tumor burden, evidenced by increased radiance and total flux (Fig. 4b, c). iNK and 235 UCB-NK cell-treated animals maintained stable body weight for 28 days. However, the tumor -only 236 group showed significant body loss (Fig. 4d). We further investigated the dynamics of UCB-NK cells 237 and iNK cells (CD45+CD56+) in the peritoneum after cell infusion (Fig. 4e). We detected 4.5% ± 4.0% 238 (mean ± SD, n=4) of UCB -NK cells or 4.4% ± 3.2% (mean ± SD, n=4) of iNK cells in total nuclear 239 cells of peritoneal dropsy from treated mice on day 7. As previously reported, CD56 expression 240 decreased in vivo in tumor-bearing mice over time, indicating partially impaired activation of iNK cells 241 (Fig. 4f)31. CD45+ CD56+ cells in the the peritoneum decreased to much lower levels in UCB-NK cell-242 treated mice (2.2% ± 1.7%, mean ± SD, n=4) and iNK cell -treated mice (1.6% ± 1.4%, mean ± SD, 243 n=4) on day 14 (Fig. 4f, g). However, the kinetics of UCB-NK and iNK cells exhibited similar patterns 244 (Fig. 4f, g). In particular, both the iNK cell and the UCB-NK cell significantly prolonged the survival 245 of treated tumor-bearing animals (Tumor only: 35 days; Tumor + UCB-NK: 83 days; Tumor + iNK: 86 246 days; P ＜ 0.001) (Fig. 4h). In conclusion, these results show that the CD34+ HSPC-derived iNK cells 247 extend suppress tumor cell growth and markedly extend the survival of A1847-tumour-bearing animals. 248 249 CD19-CAR iNK cells efficiently eliminated B cell lymphoma and leukemia tumor cells 250 To confirm the specific cytotoxicity of CD19 -CAR iNK cells, we performed in vitro tumor-killing 251 assays by coculture of CD19-CAR iNK cells with CD19-positive Nalm-6 cells, primary tumor cells of 252 CD19-positive human B cell lymphoma and leukemia, respectively. We first selected the Nalm-6 tumor 253 cell line for the tumor-killing effect of CD19-CAR iNK cells. Initially, Nalm-6 cells (Targets, T) were 254 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted July 30, 2024. ; https://doi.org/10.1101/2024.07.30.605741doi: bioRxiv preprint incubated with UCB-NK cells, iNK cells, and CD19-CAR iNK cells (Effectors, E) respectively at E: T 255 = 0:1, 0.2:1, 0.4:1, 0.8:1, 1.6:1, and 5: 1 for 4 hours (Fig. 5a). As expected, CD19-CAR iNK cells 256 showed superior cytotoxicity against tumor targets than UCB-NK cells and iNK cells (Fig. 5b, P ＜ 257 0.001). We then analyzed the expression of CD107a, a typical membrane protein associated with NK 258 cell cytotoxicity 26. Our result showed that CD19 -CAR iNK cells exhibited much higher levels of 259 CD107a expression than UCB-NK cells and iNK cells (Fig. 5c, d, P ＜ 0.001). Furthermore, we also 260 evaluated the nonspecific tumoricidal capacity of CD19-CAR iNK cells using K562 tumor cells (Fig. 261 5e). Our results indeed demonstrated that these CD19-CAR iNK cells retained nonspecific cytotoxicity 262 similar to that of UCB-NK and iNK cells (Fig. 5f). To assess the sustained cytotoxic activity of CD19-263 CAR iNK cells, we performed three rounds of tumor -killing assays with Nalm -6 tumor cells at an 264 effector-to-target (E: T) ratio of 1:1 (Fig. 5g). Throughout three rounds, CD19 -CAR iNK cells 265 maintained constant cytotoxicity. It is noteworthy that the serial killing feature was only observed in 266 CAR iNK cells but not in UCB -NK and iNK cells (Fig. 5h, i, P ＜ 0.001). Therefore, CD19-CAR 267 HSPC-derived CD19-CAR iNK cells possess specific cytotoxicity against CD19 positive tumor cells 268 and maintain the nonspecific tumoricidal capacity of NK cells. Next, we tested the particular 269 cytotoxicity of CD19-CAR iNK against CD19-positive primary tumor cells from patients with B cell 270 lymphoma and B cell leukemia (Fig. 5j). Our results showed that CD19-CAR iNK cells exhibited 271 increased cytotoxicity against patient-derived B cell lymphoma (Patient 1) and B cell leukemia cells 272 (Patient 2 and 3) ( P ＜ 0.001). However, iNK cells showed limited cytotoxicity against these three 273 patient tumor cells even at higher E: T ratios (Fig. 5k-n). 274 Collectively, CD19-CAR iNK cells exhibit superior specific cytotoxicity against CD19 -positive 275 tumor cells and still possess serial killing capacity and maintain nonspecific cytotoxicity of NK cells. 276 277 CD19-CAR iNK cells suppressed tumor growth in Nalm6 tumor-bearing mice 278 To evaluate the therapeutic efficacy in vivo of CD19-CAR iNK cells, we established B-ALL xenograft 279 animal models using Nalm-6 cells expressing luciferase (Nalm-6 luci+). On day 1, B-NDG hIL15 mice 280 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted July 30, 2024. ; https://doi.org/10.1101/2024.07.30.605741doi: bioRxiv preprint (NOD.CB17-PrkdcscidIl2rgtm1Il15tm1(IL15)/Bcgen background) were injected intravenously with 1 × 10 5 281 Nalm-6 luci+ cells. Tumor-bearing mice were injected with control iNK cells or CD19-CAR iNK cells 282 (the equivalent of 1.0 × 107 CD19-CAR iNK cells/dose, twice) through the tail veins. Tumor progression 283 was monitored weekly using bioluminescent imaging (BLI) (Fig. 6a). iNK cell treatment alone did not 284 suppress Nalm-6 cell growth in vivo, consistent with the observations in UCB-NK cell-treated Nalm-6 285 tumor-bearing animals22. As expected, CD19-CAR iNK cells effectively suppressed tumor growth in 286 vivo, demonstrating superior tumor-killing efficiencies over iNK cells (Fig. 6b, c). We further 287 investigated the in vivo kinetics of iNK cells and CD19-CAR iNK cells in tumor-bearing mice (Fig. 6d). 288 We observed apparent human CD45 +CD56+ iNK cells that circulate in the peripheral blood of the 289 animals treated with iNK cells (7.4% ± 2.2%, mean ± SD, n=5)- and iNK cells treated with CD19-CAR 290 (10.0% ± 1.7%, mean ± SD, n=5) -treated animals on day 7. However, circulated CD45 +CD56+ 291 decreased too much lower levels in mice treated with iNK cells (1.2% ± 0.8%, mean ± SD, n=5) and 292 mice treated with CD19-CAR iNK cells (1.9% ± 1.4%, mean ± SD, n=5) on day 14 (Fig. 6e, f). In 293 particular, the CD19 -CAR iNK cell therapy significantly extended the survival of Nalm -6 tumor -294 bearing mice (Tumor only: 25 days; Tumor + iNK: 27 days; Tumor + CD19 -CAR iNK: 49 days; P < 295 0.001) (Fig. 6g). 296 Our data demonstrate that CD19-CAR iNK cells can efficiently kill CD19-positive tumor cells in 297 vivo and prolong the survival of tumor-bearing animals. 298 299 Cryopreserved CD19-CAR iNK cells retained tumor-killing efficacies in vitro and in vivo. 300 To mimic the potential ‘off -the-shelf’ products of CD19 -CAR iNK cells, we harvested and 301 cryopreserved CD19 -CAR iNK cells using commercial GMP grade CryoStor® CS10 cell freezing 302 medium32. After six months, cryopreserved cells were thawed and revived for recovery tests for 24, 48, 303 and 72 hours to assess the dynamic variations of their viability, cell numbers, and tumor-killing activities 304 (Fig. 7a). After 72 hours of in vitro revival, the viabilities of NK cells reached 87.8% ± 1.1% (mean ± 305 SD, n = 3) . The CD19-CAR iNK cell count was 90.0% ± 1.1% (mean ± SD, n = 3) (Fig. 7b, c). 306 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted July 30, 2024. ; https://doi.org/10.1101/2024.07.30.605741doi: bioRxiv preprint Furthermore, after 72 hours of in vitro culture, the iNK cell and CD19 -CAR iNK cell counts were 307 restored to 89.2% ± 2.8% (mean ± SD, n = 3) and 91.2% ± 3.8% (mean ± SD, n = 3) of pre-cryo-308 preservation input counts, respectively (Fig. 7d, e). Furthermore, Nalm-6 tumor-killing assays showed 309 that revived CD19-CAR iNK cells exhibited similar cytotoxicities to fresh CD19-CAR iNK cells (Fig. 310 7f, g). 311 Furthermore, we evaluate the in vivo tumor killing efficacy of cryopreserved CD19-CAR iNK cells 312 (Fig. 7h). As expected, thawed CD19-CAR iNK cells without extended culture revival still retained 313 enhanced antitumor activity over NK cells in vivo (Fig. 7i, j). We also observed apparent human 314 CD45+CD56+ iNK cells circulating in the peripheral blood of the animals treated with iNK cells (11.4% 315 ± 2.8%, mean ± SD, n=4)- and CD19-CAR iNK cells (12.7% ± 2.6%, mean ± SD, n=4)-treated animals 316 on day 7. However, circulating CD45+CD56+ decreased too much lower levels in mice treated with iNK 317 cells (1.1% ± 0.4%, mean ± SD, n=4) and mice treated with CD19-CAR iNK cells (1.7% ± 0.7%, mean 318 ± SD, n=5) on day 14 (Fig. 7k, l). In particular, thawed CD19-CAR iNK cell therapy significantly 319 extended the survival of Nalm -6 tumor-bearing mice ( Tumor only: 24 days; Tumor + iNK: 26 days; 320 Tumor + CD19-CAR iNK: 43 days; P < 0.001) (Fig. 7m). 321 In summary, our data demonstrate that cryopreserved CD19-CAR iNK cells maintain antitumor 322 activity both in vitro and in vivo. 323 324

325 In this study, we develop an efficient three -step strategy that can generate trillions of iNK cells and 326 CD19-CAR iNK cells from a single umbilical cord blood unit of CD34 + HSPCs. The generating 327 efficiencies of iNK cells and CD19-CAR iNK cells are 6,918-20,224 times over traditional methods10. 328 Interestingly, our method sharply reduces the engineering cost of CD19-CAR to negligible. iNK cells 329 derived from CD34 + HSPC and CD19 -CAR iNK cells possess ideal tumoricidal activities against 330 human tumors. Our study provides profound insight into the use of CD34 + HSPCs as cell sources to 331 generate CAR NK cells and expand their accessibility and affordability for patients. 332 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted July 30, 2024. ; https://doi.org/10.1101/2024.07.30.605741doi: bioRxiv preprint Our CD19-CAR iNK cells showed CD19 -CAR positivity rates ranging from 58.5% to 64.4%, 333 comparable to the positivity rates of CD19 -CAR NK cells obtained by engineering mature UCB-NK 334 cells (47.8% to 87.4%, median = 66.6%)33. Transduction of CAR into CD34+ HSPCs does not alter the 335 differentiation efficiencies of the generation of CAR iNK cells or results in silencing of CAR expression 336 in derived iNK cells. Furthermore, in a mouse model with Nalm-6 xenograft, CD19-CAR iNK cells 337 showed superior specific killing activity compared to unmodified iNK cells, consistent with previous 338 studies in which cryopreserved and thawed CD19 -CAR NK cells (injected with 5 × 10 6 CD19-CAR 339 NK cells per dose, four times) without any culture revival significantly inhibited the growth of Nalm-6 340 tumors in mice 34. By administering 1 × 10 7 CD19-CAR iNK cells twice, we achieved similar 341 suppression effects on tumor growth, confirming the therapeutic potential and application value of 342 CD19-CAR iNK cells in antitumor therapy. However, unlike long-term therapeutic efficacies observed 343 via single-dose CD19 CAR-T treatment in animal models and in certain patients, two doses of CD19 -344 CAR iNK cell treatment still lack persistent therapeutic efficacies in tumor-bearing animals treated, 345 indicating that multiple doses and high doses of CAR -iNK cell therapy are critical for persistent 346 suppression of tumors. However, our method produces reliable large-scale NK and CAR NK cells from 347 CD34+ HSPCs to prospectively treat human cancers. 348 Over decades, the AFT024 cell line has been shown to be a feeder cell priority for the expansion 349 of human CD34+ HSPCs, which ideally maintains the stemness and curbs the differentiation of human 350 long-term hematopoietic stem cells (HSCs)17-19. In our approach, AFT024 cells further expanded CD34+ 351 HSPCs without losing their differentiation potential from the NK lineage. The 14-day expansion based 352 on AFT024 feeder cell-based 14-day expansion in the presence of HSC culture medium20 significantly 353 increases the iNK and CAR iNK cell yields more than 160 times, indicating that the potential of the NK 354 lineage is preserved primarily in expanded CD34 + HSPCs. Our method reliably produces large -scale 355 homogeneous iNK cells and CAR iNK cells from a single donor umbilical cord blood unit of CD34 + 356 HSPCs, which prospectively ensures the need for 760 to 83,000 doses (10 9 - 1010 cells per dose)8, 9 to 357 treat tumor patients. Our result shows that the thawed CD19-CAR iNK cells retain anti-tumor efficacy, 358 providing obvious leverages for long-term safety inspections of CAR NK products and comprehensive 359 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted July 30, 2024. ; https://doi.org/10.1101/2024.07.30.605741doi: bioRxiv preprint clinical applications. Thus, in clinical settings, the tumorigenic risks of CAR iNK cells can be well 360 evaluated in animal models before large-scale clinical applications, as autogenous CAR-T cell therapies 361 reportedly show a few cases of CAR-T-derived secondary T cell tumors 35. Our method shows that 362 engineering CAR at the CD34+ HSPC stage barely harms iNK induction efficiencies. Consequently, we 363 have reduced the use of CAR engineering materials to 1 / 140,000 –1 / 600,000, the resulting cost of 364 which is negligible for the manufacturing of CAR NK cells. Interestingly, the CD19-CAR expression 365 rates in our method maintain consistently high levels (58.5% to 64.4%), indicating that there is no 366 obvious silencing of CAR expression during the entire three -step process. Thus, reducing the 367 manufacturing cost of CAR iNK cells will promote the accessibility of this type of cell therapy. 368 In conclusion, we have developed a comprehensive technique for efficiently generating massive 369 CD34+ HSPC-derived iNK cells and CAR iNK cells, with characteristics of trillion-scale yields and 370 negligible cost of CAR engineering from a single donor umbilical cord blood unit of CD34+ HSPCs. 371 Our study provides profound insight into the use of CD34+ HSPCs as cell sources to generate CAR NK 372 cells and expand their accessibility and affordability for patients. 373 374

375 1. Larson, R.C. & Maus, M.V . Recent advances and discoveries in the mechanisms and 376 functions of CAR T cells. Nat Rev Cancer 21, 145-161 (2021). 377 2. Laskowski, T.J., Biederstadt, A. & Rezvani, K. Natural killer cells in antitumour 378 adoptive cell immunotherapy. Nat Rev Cancer 22, 557-575 (2022). 379 3. Vivier, E. et al. Natural killer cell therapies. Nature 626, 727-736 (2024). 380 4. Cichocki, F. & Miller, J.S. In vitro development of human Killer -Immunoglobulin 381 Receptor-positive NK cells. Methods Mol Biol 612, 15-26 (2010). 382 5. Gong, Y ., Klein Wolterink, R.G.J., Wang, J., Bos, G.M.J. & Germeraad, W.T.V . 383 Chimeric antigen receptor natural killer (CAR -NK) cell design and engineering for 384 cancer therapy. J Hematol Oncol 14, 73 (2021). 385 6. Zhang, Y . et al. In vivo kinetics of human natural killer cells: the effects of ageing and 386 acute and chronic viral infection. Immunology 121, 258-265 (2007). 387 7. Page, A., Chuvin, N., Valladeau-Guilemond, J. & Depil, S. Development of NK cell -388 based cancer immunotherapies through receptor engineering. Cell Mol Immunol 21, 389 315-331 (2024). 390 8. Liu, E. et al. Use of CAR-Transduced Natural Killer Cells in CD19-Positive Lymphoid 391 Tumors. N Engl J Med 382, 545-553 (2020). 392 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted July 30, 2024. ; https://doi.org/10.1101/2024.07.30.605741doi: bioRxiv preprint 9. Marin, D. et al. Safety, efficacy and determinants of response of allogeneic CD19 -393 specific CAR-NK cells in CD19(+) B cell tumors: a phase 1/2 trial. Nat Med 30, 772-394 784 (2024). 395 10. de Jonge, P. et al. Good manufacturing practice production of CD34(+) progenitor -396 derived NK cells for adoptive immunotherapy in acute myeloid leukemia. Cancer 397 Immunol Immunother 72, 3323-3335 (2023). 398 11. Myers, J.A. & Miller, J.S. Exploring the NK cell platform for cancer immunotherapy. 399 Nat Rev Clin Oncol 18, 85-100 (2021). 400 12. De Oliveira, S.N. et al. Modification of hematopoietic stem/progenitor cells with 401 CD19-specific chimeric antigen receptors as a novel approach for cancer 402 immunotherapy. Hum Gene Ther 24, 824-839 (2013). 403 13. Moore, K.A., Ema, H. & Lemischka, I.R. In vitro maintenance of highly purified, 404 transplantable hematopoietic stem cells. Blood 89, 4337-4347 (1997). 405 14. Fares, I. et al. Cord blood expansion. Pyrimidoindole derivatives are agonists of human 406 hematopoietic stem cell self-renewal. Science 345, 1509-1512 (2014). 407 15. Montel-Hagen, A. et al. Organoid -Induced Differentiation of Conventional T Cells 408 from Human Pluripotent Stem Cells. Cell Stem Cell 24, 376-389 e378 (2019). 409 16. Huang, D. et al. Lateral plate mesoderm cell -based organoid system for NK cell 410 regeneration from human pluripotent stem cells. Cell Discov 8, 121 (2022). 411 17. Thiemann, F.T., Moore, K.A., Smogorzewska, E.M., Lemischka, I.R. & Crooks, G.M. 412 The murine stromal cell line AFT024 acts specifically on human CD34+CD38 - 413 progenitors to maintain primitive function and immunophenotype in vitro. Exp Hematol 414 26, 612-619 (1998). 415 18. Punzel, M., Gupta, P., Roodell, M., Mortari, F. & V erfaillie, C.M. Factor(s) secreted by 416 AFT024 fetal liver cells following stimulation with human cytokines are important for 417 human LTC-IC growth. Leukemia 13, 1079-1084 (1999). 418 19. Nolta, J.A. et al. The AFT024 stromal cell line supports long-term ex vivo maintenance 419 of engrafting multipotent human hematopoietic progenitors. Leukemia 16, 352 -361 420 (2002). 421 20. Chen, Y . et al. ADGRG1 enriches for functional human hematopoietic stem cells 422 following ex vivo expansion-induced mitochondrial oxidative stress. J Clin Invest 131 423 (2021). 424 21. Denman, C.J. et al. Membrane -bound IL-21 promotes sustained ex vivo proliferation 425 of human natural killer cells. PLoS One 7, e30264 (2012). 426 22. Wang, Y . et al. Comparison of seven CD19 CAR designs in engineering NK cells for 427 enhancing anti-tumour activity. Cell Prolif, e13683 (2024). 428 23. Bryceson, Y .T., March, M.E., Ljunggren, H.G. & Long, E.O. Activation, coactivation, 429 and costimulation of resting human natural killer cells. Immunol Rev 214, 73-91 (2006). 430 24. Quatrini, L. et al. Human NK cells, their receptors and function. Eur J Immunol 51, 431 1566-1579 (2021). 432 25. Prager, I. & Watzl, C. Mechanisms of natural killer cell-mediated cellular cytotoxicity. 433 J Leukoc Biol 105, 1319-1329 (2019). 434 26. Alter, G., Malenfant, J.M. & Altfeld, M. CD107a as a functional marker for the 435 identification of natural killer cell activity. J Immunol Methods 294, 15-22 (2004). 436 27. Di Santo, J.P. Functionally distinct NK -cell subsets: developmental origins and 437 biological implications. Eur J Immunol 38, 2948-2951 (2008). 438 28. V oskoboinik, I., Whisstock, J.C. & Trapani, J.A. Perforin and granzymes: function, 439 dysfunction and human pathology. Nat Rev Immunol 15, 388-400 (2015). 440 29. Franks, S.E., Wolfson, B. & Hodge, J.W. Natural Born Killers: NK Cells in Cancer 441 Therapy. Cancers (Basel) 12 (2020). 442 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted July 30, 2024. ; https://doi.org/10.1101/2024.07.30.605741doi: bioRxiv preprint 30. Vankayalapati, R. et al. Role of NK cell -activating receptors and their ligands in the 443 lysis of mononuclear phagocytes infected with an intracellular bacterium. J Immunol 444 175, 4611-4617 (2005). 445 31. Van Acker, H.H., Capsomidis, A., Smits, E.L. & Van Tendeloo, V .F. CD56 in the 446 Immune System: More Than a Marker for Cytotoxicity? Front Immunol 8, 892 (2017). 447 32. Li, Y ., Mateu, E. & Diaz, I. Impact of Cryopreservation on Viability, Phenotype, and 448 Functionality of Porcine PBMC. Front Immunol 12, 765667 (2021). 449 33. Liu, E. et al. Cord blood NK cells engineered to express IL -15 and a CD19 -targeted 450 CAR show long-term persistence and potent antitumor activity. Leukemia 32, 520-531 451 (2018). 452 34. Golubovskaya, V . et al. CAR -NK Cells Generated with mRNA -LNPs Kill Tumor 453 Target Cells In Vitro and In Vivo. Int J Mol Sci 24 (2023). 454 35. Levine, B.L. et al. Unanswered questions following reports of secondary malignancies 455 after CAR-T cell therapy. Nat Med 30, 338-341 (2024). 456 457

458 Ethics statement 459 NCG mice and B -NDG hIL15 mice were housed in SPF-grade animal facilities at the Guangzhou 460 Institutes of Biomedicine and Health, Chinese Academy of Sciences. All animal-related procedures in 461 this study received approval from the Institutional Animal Care and Use Committee of the Guangzhou 462 Institutes of Biomedicine and Health. NK cell antitumor activity assessments in animals received 463 approval from the Biomedical Research Ethics Committee of the Guangzhou Institutes of Biomedicine 464 and Health, Chinese Academy of Sciences. The use of patient samples was carried out in accordance 465 with the provisions of the Declaration of Helsinki. All patient samples were collected with prior consent 466 signatures from the patient and reviewed and approved by the ethics committee of the Jinan University 467 First Affiliated Hospital. This informed consent included allowing patients to publish the data arising 468 from the tissues they obtained. 469 470 Statistics 471 All quantitative analyses were performed with SPSS (version 23, IBM Corp., Armonk, NY , USA). The 472 Shapiro-Wilk normality test in SPSS was used to evaluate the normal distribution of the data. The two-473 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted July 30, 2024. ; https://doi.org/10.1101/2024.07.30.605741doi: bioRxiv preprint tailed independent t test (for normally distributed data) and the Mann-Whitney U test (for nonnormally 474 distributed data) were applied for comparison of two groups of data. For three or more groups: the one-475 way ANOV A test (for normally distributed data) and Kruskal-Wallis tests (for nonnormally distributed 476 data) were used. Survival curves were plotted using the Kaplan-Meier method. Differences in survival 477 rates between groups were evaluated using the Log rank test (Mantel -Cox). Statistical analyses were 478 performed using GraphPad Prism (8.0.2, GraphPad Software). 479 480 Data availability 481 Single-cell RNA sequencing data (fastq files) have been uploaded to the public database of the Genome 482 Sequence Archive (HRA007978). Raw flow cytometry data files and bioluminescent imaging data are 483 available upon request. Other relevant information or data are available from the corresponding authors 484 upon reasonable request. 485 486

487 This work was supported by grants from the National Natural Science Foundation of China (81925002) 488 and the National Key R&D Program of China (2020YFA0112404). 489 490 Author contributions 491 J.L., Y .W., and X.Z. performed the core experiments and contributed equally to this work. J.L. wrote 492 the manuscript. Y .L. and Q.W. analyzed the RNA-seq data. X.L., Y .G., H.W., L.L., H.P., B.W., D.H., 493 C.X., T.W., and M.Z. participated in multiple experiments. X.D., H.Z., F.D., Y .Z. and X.Z. discussed 494 the data and manuscript. J.W. and F.H. designed the project and wrote the manuscript. J.W. provided 495 the competing interests. 496 497 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted July 30, 2024. ; https://doi.org/10.1101/2024.07.30.605741doi: bioRxiv preprint 498 Fig.1 A three-step strategy for generating iNK cells and CD19-CAR iNK cells from CD34+ HSPCs. 499 a, Schematic diagram showing the generation of iNK cells and CD19 -CAR iNK cells from CD34 + 500 HSPCs or CD19 -CAR HSPCs. Step I was for the expansion of CD34 + HSPC or CD19 -CAR CD34+ 501 HSPC. AFT024 cells (1 × 105 cells /well) were seeded in the 24-well plate and then irradiated (20 Gy). 502 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted July 30, 2024. ; https://doi.org/10.1101/2024.07.30.605741doi: bioRxiv preprint CD34+ HSPCs or CD19-CAR HSPCs (5 × 10 4 cells/well) were seeded on the irradiated AFT024 cells 503 and cocultured for the 1 st round. On day 7, CD34+ HSPC or CD19-CAR HSPC from the first round 504 expansion were collected and seeded in the new irradiated AFT024 cells (1 × 10 7 cells /flasks) for the 505 second round expansion. Step II was for the differentiation of the NK lineage using the organoid 506 aggregate induction method. CD34 + HSPCs or CD19-CAR HSPCs (1 × 10 4 cells/organoid) collected 507 from Step I were combined with OP9 cells (5 × 10 5 cells/organoid) to prepare organoid aggregates, 508 which were seeded in transwell and cultured for 14 days for induction of iNK cells or CD19-CAR iNK 509 cells. Step III was for the maturation and proliferation of NK cells. CD45+ cells collected from Step II 510 were transferred to cell culture bags and cultured for 14 days. b-c, flow cytometric analysis of the 511 immune phenotypes of iNK cells ( b) and CD19 -CAR iNK cells ( c) (CD56+ CD16+/-). Data were 512 collected from three donor umbilical cord blood units. d, flow cytometric analysis of CD19 -CAR 513 expression (CD56 +CD19-CAR+). e-f, Statistical analysis of the output efficiencies of iNK cells 514 (CD45+CD3-CD56+CD16+/-) or CD19 -CAR iNK cells (CD45 +CD3-CD56+CD16+/-CD19-CAR+) 515 derived from single CD34 + HSPCs or CD19-CAR HSPCs. Day 0, fresh CD34+ HSPCs. Day 7, 7-day 516 expanded CD34+ HSPCs. Day 14, 14 -day expanded CD34+ HSPCs. g, UMAP visualization of UCB-517 NK cells, iNK cells, and CD19 -CAR iNK cells. h, Table showing the calculated quantities of CD19 -518 CAR retroviruses (TU). The viral particles required to generate 1.0 × 10 6 CD19-CAR iNK cells and 519 CD19-CAR NK cells were compared. Data were presented as means ± SD. Independent two-tailed t 520 test (e and f). NS, not significant, **P < 0.01, ***P < 0.001. 521 522 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted July 30, 2024. ; https://doi.org/10.1101/2024.07.30.605741doi: bioRxiv preprint 523 Fig.2 Cellular kinetics in the three-step strategy to generate iNK cells and CD19-CAR iNK cells. 524 a, Representative FACS plots showing the ratios of CD34 + HSPCs (CD45+CD34+) during expansion 525 (Step I) of CD34 + HSPCs (Left panel) and CD19 -CAR HSPCs (CD45 +CD34+) (Right panel) (Day 0, 526 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted July 30, 2024. ; https://doi.org/10.1101/2024.07.30.605741doi: bioRxiv preprint Day 7, and Day 14). b, Statistical analysis of the ratios of CD34 + HSPCs and CD19-CAR HSPCs 527 (CD45+CD34+) during step I. (n = 3 in each group, n indicated three donor umbilical cord blood units). 528 c, Statistical analysis of the output of CD34+ HSPCs and CD19-CAR HSPCs derived from 1.0 × 10 6 529 CD34+ HSPCs in Step I. (n = 3 in each group, n indicated three donor umbilical cord blood units). d, 530 Representative FACS plots showing the ratios of CD45 +CD34+, CD45+CD7+, and CD45 +CD56+ cells 531 of the organoid aggregates derived from CD34 + HSPCs and CD19-CAR CD34+ HSPCs in Step II. e, 532 Statistical analysis of the ratios of CD45+CD56+ cells in step II. (n = 3 in each group, n indicated three 533 donor umbilical cord blood units). f, Statistic analysis of CD45+ cell outputs derived from 50 organoids 534 in Step II. (n = 3 in each group, n indicated three donor umbilical cord blood units). g, Representative 535 FACS plots showing the phenotypes (CD45 +CD3-CD56+CD16+/-) of total cells on day 35 and day 42 536 during cell maturation and proliferation of iNK (Left panel) or CD19 -CAR iNK (Right panel) cell 537 maturation and proliferation (Step III). h, Statistical analysis of the ratios of CD45+CD3-CD56+cells on 538 day 35 and day 42 in Step III. (n = 3 in each group, n indicated three donor umbilical cord blood units). 539 i, Statistical analysis of the output of CD45+CD3-CD56+cells derived from 50 organoids on day 35 and 540 day 42 in step III. (n = 3 in each group, n indicated three donor umbilical cord blood units). j, 541 Representative FACS plots showing the expression of CD19-CAR of total cells at the indicated time 542 points. k, Statistical analysis of the ratios of CD45+CD19-CAR+cells at the indicated time points. (n = 543 3 in each group, n indicated three donor umbilical cord blood units). Data were presented as means ± 544 SD. Independent two-tailed t test (b, c, e, f, h, and i). NS, not significant, *P < 0.05. 545 546 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted July 30, 2024. ; https://doi.org/10.1101/2024.07.30.605741doi: bioRxiv preprint 547 Fig.3 Molecular features and immune activities of CD34+ HSPC derived iNK cells. a, Schematic 548 diagram showing the mechanisms of iNK cells that recognize and kill tumor cells. b, Flow cytometric 549 analysis of typical NK receptors and effectors (CD319, NKp30, NKp44, NKG2D, CD69, CD94, 550 NKG2A, CD96, TARIL, FasL). c, Representative flow cytometry histograms showing the expression 551 levels of the CD107a, TNFα, IFN-γ, Perforin, and GZMB proteins in iNK cells and UCB -NK cells 552 (CD45+CD56+). d, Statistical analysis of expression levels of CD107a, TNF α, IFN-γ, Perforin, and 553 GZMB proteins (n = 3 repeats in each group). e, Schematic diagram of measuring the nonspecific 554 cytotoxicities of iNK cells. f-k, Statistical analysis of nonspecific cytotoxicities of iNK cells and UCB-555 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted July 30, 2024. ; https://doi.org/10.1101/2024.07.30.605741doi: bioRxiv preprint NK cells. (n = 3 repeats in each group). Cytotoxicity was calculated using a formula: (percentage of 556 tumor cell death – percentage of tumor cell spontaneous death) / (1 - percentage of tumor cell 557 spontaneous death) × 100. l Schematic diagram of measuring consecutive killing cytotoxicity of iNK 558 cells. m, Statistical analysis of the consecutive cytotoxicity of iNK cells and UCB -NK cells. (n = 3 559 repeats in each group). n Representative flow cytometry histograms showing the expression levels of 560 NK cell activation proteins (2B4, DNAM-1 and NKp46) before and after the consecutive killing assay 561 in m. o, Statistical analysis of mean fluorescence intensity (MFI) of UCB-NK cells and iNK cells before 562 and after the consecutive killing assay. (n = 3 repeats in each group). Data were presented as means ± 563 SD. Independent two-tailed t test (d, f-k, m and o) or Mann-Whitney U test (d and j). NS, not significant, 564 *P < 0.05, **P < 0.01, ***P < 0.001. 565 566 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted July 30, 2024. ; https://doi.org/10.1101/2024.07.30.605741doi: bioRxiv preprint 567 Fig.4 The iNK cell suppressed human tumor cell growth in xenograft animals. a, Schematic 568 diagram of evaluation of nonspecific cytotoxicities of iNK cells in the tumor-bearing mouse. b, BLI 569 analysis of the xenograft models (n = 4 in each group). c, Statistical analysis of total flux 570 (photons/second, p/s) in xenograft models (n = 4 in each group). d, Statistical analysis of body weights 571 of the xenograft models on day 28 after iNK cell injection (n = 4 in each group). e, Schematic diagram 572 of analyzing the proportions of iNK or UCB -NK cells (CD45 +CD56+) in NK cell -treated xenograft 573 models. f, Representative FACS plots showing the proportions of iNK cells or UCB -NK cells 574 (CD45+CD56+) in the peritoneum on day 7. g, Statistics analysis of the proportions of the iNK cells or 575 UCB-NK cells (n = 4 in each group). h, Kaplan-Meier survival curves for xenograft models ( P ＜ 576 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted July 30, 2024. ; https://doi.org/10.1101/2024.07.30.605741doi: bioRxiv preprint 0.001, Log-rank test). Data were presented as means ± SD. One-way ANOV A test (c), Kruskal–Wallis 577 tests (c), or two-tailed independent t test (c, d, and g). NS, not significant, *P < 0.05, ***P < 0.001. 578 579 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted July 30, 2024. ; https://doi.org/10.1101/2024.07.30.605741doi: bioRxiv preprint 580 Fig.5 CD19-CAR iNK cells exhibited specific tumor-killing activities against B-cell lymphoma and 581 leukemia in vitro. a, Schematic diagram of evaluation of the in vitro specific cytotoxicities of CD19-582 CAR iNK cells. b, Statistic analysis of the specific cytotoxicities of CD19 -CAR iNK cells. c, 583 Representative FACS plots showing the expression levels of the CD107a protein of CD19-CAR-iNK 584 cells. d, Statistical analysis of the expression levels of the CD107a protein (n = 3 repeats in each group). 585 e, Schematic diagram of evaluation of nonspecific cytotoxicities of CD19-CAR iNK cells. f, Statistical 586 analysis of the nonspecific cytotoxicities of CD19-CAR iNK cells. (n = 3 repeats in each group). g, 587 Schematic diagram of the evaluation of the consecutive killing cytotoxicities of CD19-CAR iNK cells. 588 h-i, Statistical analysis of consecutive specific cytotoxicities of CD19-CAR iNK cells. (n = 3 repeats in 589 each group). j, Schematic diagram showing the specific cytotoxicities of CD19-CAR iNK cells against 590 CD19-expressing tumor cells isolated from patients with B cell leukemia or B cell lymphoma. BM, 591 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted July 30, 2024. ; https://doi.org/10.1101/2024.07.30.605741doi: bioRxiv preprint bone marrow. k, Representative FACS plots showing the expression levels of the CD19 protein in 592 patient-derived B-cell lymphoma or leukemia cells (Gated from CD45 +). l, Statistical analysis of the 593 specific cytotoxicities of CD19-CAR iNK cells against patient -derived B cell lymphoma cells. (n = 3 594 repeats in each group). m-n, Statistical analysis of the specific cytotoxicities of CD19-CAR iNK cells 595 against patient-derived B cell leukemia cells. (n = 3 for each group). Cytotoxicity was calculated using 596 the formula: (percentage of tumor cell death – percentage of tumor cell spontaneous death) / (1 – 597 percentage of tumor cell spontaneous death) × 100. Data were presented as means ± SD. One-way 598 ANOV A test (b and f) or two-tailed independent t test (d, i, l, m and n). NS, not significant, **P < 0.01, 599 ***P < 0.001. 600 601 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted July 30, 2024. ; https://doi.org/10.1101/2024.07.30.605741doi: bioRxiv preprint 602 Fig.6 The CD19-CAR iNK cell suppressed the growth of human B leukemia cells in xenograft 603 animals. a, Schematic diagram of the evaluation of the cytotoxicities of CD19-CAR iNK cells in vivo. 604 b, BLI analysis of the xenograft models (n = 5 in each group). c, Statistical analysis of total flux 605 (photons/second, p/s) in xenograft models (n = 5 in each group). d, Schematic diagram for analyzing 606 the proportion of human CD45 +CD56+ cells in peripheral blood (PB). e, Representative FACS plots 607 showing the proportion of human CD45 +CD56+ cells from PB in tumor-bearing mice on day 7. f, 608 Statistical analysis of the ratios of CD45+ CD56+ peripheral blood NK cells in tumor-bearing mice on 609 day 7 and day 14 (n = 5 in each group). g, Kaplan-Meier survival curves for xenograft models ( P ＜ 610 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted July 30, 2024. ; https://doi.org/10.1101/2024.07.30.605741doi: bioRxiv preprint 0.001, Log-rank test). Data are presented as means ± SD. One-way ANOV A test (c), Kruskal–Wallis 611 tests (c), or two-tailed independent t-test (f). NS, not significant, **P < 0.01. 612 613 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted July 30, 2024. ; https://doi.org/10.1101/2024.07.30.605741doi: bioRxiv preprint 614 Fig.7 The cryopreserved CD19-CAR iNK cells retained tumor-killing efficacies in vitro and in vivo. 615 a, Schematic diagram for evaluating the specific cytotoxicities of the cryopreserved CD19 -CAR iNK 616 cells in vitro and in vivo. b-c, Statistical analysis of the viabilities of iNK cells and CD19 -CAR iNK 617 cells after thawing for 24 hours, 48 hours, and 72 ho urs. (n = 3 repeats in each group). d-e, Statistical 618 analysis of cell yield recovery rates of iNK cells and CD19-CAR iNK cells after thawing for 24 hours, 619 48 hours, and 72 hours. (n = 3 for each group). f-g, Statistical analysis of the specific cytotoxicities of 620 iNK cells and CD19-CAR iNK cells after thawing for 24 hours, 48 hours, and 72 hours. (n = 3 repeats 621 in each group). h, Schematic diagram of the evaluation of the specific cytotoxicities of the 622 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted July 30, 2024. ; https://doi.org/10.1101/2024.07.30.605741doi: bioRxiv preprint cryopreserved CD19-CAR iNK cells in vivo. i, BLI analysis of the xenograft models (n = 4 in each 623 group). j, Statistical analysis of total flux (photons/second, p/s) in xenograft models (n = 4 in each 624 group). k, Representative FACS plots showing the proportion of iNK cells and CD19 -CAR iNK cells 625 (CD45+CD56+) from peripheral blood on day 7. l, Statistics analysis of the ratios of iNK cells and 626 CD19-CAR iNK cells (CD45+CD56+) from peripheral blood on day 7 and day 14 (n = 4 in each group). 627 m, Kaplan-Meier survival curves for xenograft models (P ＜ 0.001, Log-rank test). Cytotoxicity was 628 calculated using the formula: (percentage of tumor cell death – percentage of tumor cell spontaneous 629 death) / (1 – percentage of tumor cell spontaneous death) × 100. Data were presented as means ± SD. 630 Independent two-tailed t test (b, c, d, e, f, and l), one-way ANOV A test (g and j), or Kruskal–Wallis 631 tests (j). NS, not significant, *P < 0.05, **P < 0.01, ***P < 0.001. 632 633 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted July 30, 2024. ; https://doi.org/10.1101/2024.07.30.605741doi: bioRxiv preprint 634 Supplementary Fig.1 Immune phenotypes and yields of iNK cells and CD19-CAR iNK cells after 635 3-week bag-based culture. a-b, flow cytometric analysis of the immune phenotypes of iNK cells (a) 636 and CD19-CAR iNK cells (b) (CD56+ CD16+/-). Data were collected from three donor umbilical cord 637 blood units. c, Flow cytometric analysis of CD19-CAR expression (CD56+CD19-CAR+). d-e, Statistical 638 analysis of the output efficiencies of iNK cells (CD45 +CD3-CD56+CD16+/-) or CD19-CAR iNK cells 639 (CD45+CD3-CD56+CD16+/-CD19-CAR+) derived from single CD34 + HSPCs or CD19 -CAR HSPCs. 640 Day 0, fresh CD34+ HSPCs. Day 7, 7-day expanded CD34+ HSPCs. Day 14, 14-day expanded CD34+ 641 HSPCs. f, Table showing the calculated quantities of CD19-CAR retroviruses (TU). The viral particles 642 required to generate 1.0 × 10 6 CD19-CAR iNK cells and CD19 -CAR NK cells were compared. Data 643 were presented as means ± SD. Independent two-tailed t test (d and e). NS, not significant, **P < 0.01, 644 ***P < 0.001. 645 646 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted July 30, 2024. ; https://doi.org/10.1101/2024.07.30.605741doi: bioRxiv preprint 647 Supplementary Fig.2 Enrichment of CD19 -CAR+ CD34+ HSPCs on day 7 further increased the 648 positive CD19-CAR ratios of CD19 -CAR iNK cells on day 42 by more than 90%. a, Schematic 649 diagram of the generation of CD19 -CAR+ HSPC cells from expanded 7-day CD19-CAR HSPCs. b, 650 Flow cytometric analysis of the CD19-CAR expression in CD34+ HSPCs before and after enrichment 651 using a CD19 -CAR based microbead enrichment process on Day 7. c, Statistical analysis of CD19-652 CAR expression levels in CD34+ HSPC before and after enrichment. Data were collected from three 653 donor umbilical cord blood units. d, Representative FACS plots showing the expression of CD19-CAR 654 of total cells at the indicated time points. e, Statistical analysis of the CD45+ CD19-CAR+ cell ratios at 655 the indicated time points. (n = 3 in each group, n indicated three donor umbilical cord blood units). f, 656 Statistical analysis of the yields of CD19 -CAR iNK cells from single 7 -day expanded CD19 -CAR 657 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted July 30, 2024. ; https://doi.org/10.1101/2024.07.30.605741doi: bioRxiv preprint HSPCs. Data were presented as means ± SD. Independent two-tailed t test (c and f). NS, not significant, 658 *P < 0.05. 659 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted July 30, 2024. ; https://doi.org/10.1101/2024.07.30.605741doi: bioRxiv preprint