Genome editing HLA alleles for a pilot immunocompatible hESC line in a Chinese hESC bank for cell therapies

Abstract Robust allogeneic immune reactions after transplantation impede the translational pace of human embryonic stem cells (hESCs)‐based therapies. Selective genetic editing of human leucocyte antigen (HLA) molecules has been proposed to generate hESCs with immunocompatibility, which, however, has not been specifically designed for the Chinese population yet. Herein, we explored the possibility of customizing immunocompatible hESCs based on Chinese HLA typing characteristics. We generated an immunocompatible hESC line by disrupting HLA‐B, HLA‐C, and CIITA genes while retaining HLA‐A*11:01 (HLA‐A*11:01‐retained, HLA‐A11R), which covers ~21% of the Chinese population. The immunocompatibility of HLA‐A11R hESCs was verified by in vitro co‐culture and confirmed in humanized mice with established human immunity. Moreover, we precisely knocked an inducible caspase‐9 suicide cassette into HLA‐A11R hESCs (iC9‐HLA‐A11R) to promote safety. Compared with wide‐type hESCs, HLA‐A11R hESC‐derived endothelial cells elicited much weaker immune responses to human HLA‐A11+ T cells, while maintaining HLA‐I molecule‐mediated inhibitory signals to natural killer (NK) cells. Additionally, iC9‐HLA‐A11R hESCs could be induced to undergo apoptosis efficiently by AP1903. Both cell lines displayed genomic integrity and low risks of off‐target effects. In conclusion, we customized a pilot immunocompatible hESC cell line based on Chinese HLA typing characteristics with safety insurance. This approach provides a basis for establishment of a universal HLA‐AR bank of hESCs covering broad populations worldwide and may speed up the clinical application of hESC‐based therapies.

or iPSC banks derived from each patient, 5 but these approaches require multiple cell lines with high costs and labor. Using immunosuppressants is an alternative to overcoming immune incompatibility but often comes with a high potential for side effects and toxicity. 6 Therefore, the generation of immunocompatible cells is in reduced immune surveillance 13 and NK cell-mediated killing because of a lack of ligands for the inhibitory receptors of NK cells. 12,14 To overcome these barriers, overexpression of HLA-I (e.g., HLA-E, 15 HLA-G, 16,17 and HLA-A 18 ) or CD47 19 on the surface of hPSCs lacking HLA-I and HLA-II may be a significant attempt.
However, whether transgenes can be expressed consistently and stably after hESC differentiation or transplantation into the body is a question worth considering, not to mention that they may only resist partial NK cell killing if they are expressed. Compared with the approaches mentioned above, a strategy based on selective HLA gene editings, such as retention of one classical HLA-I locus and disruption of the other two while retaining non-classical HLA-I (the single retention strategy) may be more desirable from the standpoints of compatibility and surveillance of the immune system. In accordance with the philosophy of this strategy, seven HLA-C-retained iPSC lines combined with CIITA knockout could cover $95% of the Japanese population. 20,21 An HLA-A*02-retained hESC line has an average prevalence of 27% in 102 countries, 22 but the frequency varies widely by geography, relatively high in white populations. 23 However, the single retention strategy to produce immunocompatible hESCs has not been proposed for the Chinese population. In China, the most highly prevalent allele is HLA-A*11:01, which covers $21% of the Chinese population. In addition, more than 95% of the Chinese population could be covered by just 14 highly prevalent HLA-A-retained hESC lines. 24 Therefore, we hypothesized we could customize an immunocompatible hESC line based on the HLA typing characteristics of the Chinese population using an HLA-A*11:01-retained strategy, which may provide a theoretical basis for establishment of an HLA-A R bank of 14 hESC lines covering 95% of the Chinese population and broad populations around the world.
Here, following the selection of homozygous HLA-A*11:01 hESCs as parental cells, we generated an immunocompatible hESC line based on Chinese HLA typing by deleting HLA-B, HLA-C, and CIITA genes while retaining HLA-A*11:01 and non-classical HLA-I molecules. HLA-A11 R hESC-derived endothelial cells (ECs) escaped HLA-A11-matched T-cell killing but also did not induce activation of NK cells in vitro.
HLA-A11 R hESCs and their derivatives could reliably evade human immune responses in humanized mice with an HLA-A11 + functional human immune system. Furthermore, an inducible caspase-9 system was installed into HLA-A11 R hESCs because of safety concerns, which were induced to commit suicide efficiently by the dimerizer agent rimiducid (AP1903). Moreover, the genomic integrity of engineered hESCs and risks of off-target were assessed. Taken together, we customized an immunocompatible hESC line with safety insurance based on Chinese HLA typing, which could be the basis for establishment of an HLA-A R hESC bank covering broad populations around the world, accelerating the progression of translation-to-clinic of hESCs for a variety of hESC-based therapies.

| Estimation of HLA allele frequency
The Chinese HLA data query website (http://cmms.dnaday.cn/index. jsp?local=zh_CN) was used to estimate the frequency of HLA-A*11:01 and the frequency order of HLA-A, -B, -C, -DR, -DQ typing in China (Table S7 and S8). Data for HLA-A*11:01 allele frequency in more populations and regions of the world were extracted from The Allele Frequency Net Database (http://www.allelefrequencies.net; Table S9). The HLA allele sequence was referenced from IPD-IMGT/ HLA database (https://www.ebi.ac.uk/ipd/imgt/hla/alleles/).

| Pluripotency analysis by immunofluorescence staining
HLA haplotype information of humanized mice was listed in Table S5. 2.12 | Teratoma assays 4 Â 10 6 WT hESC or HLA-A11 R hESC were mixed with matrigel at a 1:1 ratio in a 100 μL system and separately injected into the subcutaneous right hindlimb of the NSG and humanized mice. Mice were sacrificed 8 weeks later and analysed by the haematoxylin-eosin (H&E) and immunohistochemistry (IHC) staining of parafcfin-embedded sections.

| Transplantation of ECs from WT or HLA-A11 R hESC
Two million WT ECs or HLA-A11 R ECs in a pro-survival scaffold  and all the slides were observed and photographed with laser scanning confocal microscopy (LSM 880) using a 10Â objective.

| Statistical analyses
Statistical analysis was performed with the GraphPad Prism 8 software, using the two-tailed unpaired or one-tailed paired Student's t-test or Mann-Whitney test analysis. The data were presented as mean values ± SEM. A p-value was considered statistically significant in all types of analyses as follows: *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, ns: not significant.  Table S7). We found that HLA-A*11:01 was the most prevalent allele (accounting for $21% of the Chinese population), and the most popular allele across all regions of China ( Figure S1B and Table S8), in addition to being found in 1% to 22% of different populations worldwide ( Figure S1C and Table S9).
These data suggested the great potential of hESCs, which require only the HLA-A*11:01 match, as valuable resources for stem cell therapy. the genes with a cleavage efficiency >40% (Table S1). After two rounds of co-electroporating gRNA/Cas9-expressing plasmids into WT hESCs, we identified positive PCR-genotyping products (HLA-A11 R hESCs) in 4.5% of derived clones ( Figure S2A and Table S3). Analyses of gel electrophoresis and Sanger sequencing revealed the specific deletion of HLA-B, HLA-C, and CIITA genes on HLA-A11 R hESCs ( Figure S2B,C). After interferon (IFN)-γ stimulation for 48 h, the ablation of the proteins of HLA-B and HLA-C in HLA-A11 R hESCs was confirmed by FCM ( Figure 1D). To demonstrate the destruction of HLA-II molecules, we further differentiated both WT and HLA-A11 R hESCs into ECs 27 (WT ECs and HLA-A11 R ECs), which expressed both HLA-I and HLA-II following IFN-γ stimulation in cells without gene editing. Both types of hESC-derived ECs expressed an equal level of EC markers, CD31 and CD144 (VE-Cadherin), which suggested that genome editing did not affect the differentiation efficiency of HLA-A11 R hESCs ( Figure S3A-D). Deletion of the HLA-II protein was confirmed by staining of HLA-DR antibodies using FCM ( Figure 1C). To confirm that HLA-A11 R hESCs retained pluripotency, expression of OCT4, SOX2, SSEA4, and NANOG was assessed by immunofluorescence staining in HLA-A11 R hESCs, and was found to be equivalent to that of WT hESCs ( Figure 1E). In addition, the pluripotency of HLA-A11 R hESCs was characterized by quantitative realtime PCR (qRT-PCR), FCM, and analysis of teratoma tissue ( Figure S4A,B,D). In summary, we generated a customized immunocompatible hESC (HLA-A11 R ) line with normal pluripotency by deleting HLA-B, HLA-C, and CIITA genes and retaining homozygous HLA-A*11:01, covering $21% of the Chinese population.

| HLA-A11 R hESC-derived ECs inhibited HLA-A11-matched T-cell responses in vitro
To investigate whether HLA-A11 R hESCs enabled to a reduction of immune responses in vitro, we co-cultured WT or HLA-A11 R ECs with allogeneic T cells from PBMCs of healthy donors, who had matched HLA-A*11:01 but other HLA alleles that were mismatched ( Figure 2A). To assess the T-cell proliferation, CFSElabelled allogeneic CD3 + T cells were co-cultured with IFN-γ-pretreated WT or HLA-A11 R ECs for 5 days. The percentage of proliferative CD3 + T cells was lower when incubated with HLA-A11 R ECs compared with WT ECs, consistent with proliferative CD4 + and CD8 + T-cell subpopulations (Figures 2B and S5A).
However, CD8 + cytotoxic T cells exhibited a more pronounced reduction in proliferation, suggesting that HLA-A11 R ECs were more able to suppress CD8 + T-cell proliferation than other T subpopulations. In summary, HLA-A11 R ECs could hinder the proliferation of HLA-A11-matched CD3 + T cells significantly, especially the CD3 + CD8 + T-cell population.
To confirm expression of the markers of T-cell activation after coculture with different hESC-derived ECs, WT and HLA-A11 R ECs were pre-treated with IFN-γ and co-cultured subsequently with CD3 + T cells for 3 or 6 days. We noticed a lower percentage of CD69 + and CD154 + T cells in co-culture with HLA-A11 R ECs compared with WT ECs, the same as CD4 + and the CD8 + T-cell subpopulations ( Figures 2C and S5B,C). To study T-cell cytotoxicity upon co-culture with different types of ECs, LDH release was measured after co-culture with WT or HLA-A11 R ECs for 5 days. T-cell cytotoxicity against HLA-A11 R ECs was $10.31%, which was significantly lower than that of the WT ECs group (40.65%; Figure 2D), thereby suggesting that T-cell cytotoxicity was suppressed by HLA-A11 R ECs, and consistent with the results of the assay for CD8 + T-cell proliferation. Altogether, these data suggested that HLA-A11 R ECs suppressed the response of allogeneic HLA-A11-matched T cells in terms of proliferation, activation, and cytotoxicity.

| HLA-A11 R hESC-derived ECs evaded NK cell responses in vitro
Failure of inhibitory receptors to bind to allogeneic HLA molecules may contribute to NK cell-mediated destruction of allogeneic donor cells. We wished to assess if the strategy of retaining HLA-A11 expression in combination with non-classical HLA-I was sufficient to resist NK cell-mediated activity. Hence, we compared the degranulation ability and cytotoxicity of NK cells against WT or HLA-A11 R ECs using a co-culture system in vitro (Figure 2A). First, we conducted degranulation assays by quantifying cell-surface expression of CD107a, a protein transferred from the intracell to the cell surface when NK cells are activated to release cytotoxic granules. No differential surface expression of CD107a on NK cells was detected when NK cells were exposed to WT or HLA-A11 R ECs ( Figures 2E and S5D), which suggested that retention of HLA-A11 expression in combination with non-classical HLA-I could aid inhibition of NK cell activation. Then, we quantified NK cell cytotoxicity against WT or HLA-A11 R ECs by detecting apoptotic target cells.
NK cell cytotoxicity against HLA-A11 R ECs was not significantly different from WT ECs in terms of the E/T ratio ( Figure 2F), in accordance with NK cell degranulation. In conclusion, HLA-A11 R hESC-derived ECs could evade the activation and cytotoxicity of NK cells in vitro.

| HLA-A11 R hESCs were compatible with HLA-A11 + human immunity in vivo
To simulate HLA-matched clinical experimental therapy using hESCs and identify the immunogenicity of HLA-A11 R hESCs in mice, we constructed HLA-A11-matched humanized mice by transplanting human foetal thymic tissue and CD34 + haematopoietic stem/progenitor cells. 26,28,29 High levels of human immune cells composed of human T cells (including CD4 + T cells and CD8 + T cells) and B cells were detected 12 weeks post-transplantation, and subsequent experiments were conducted in random groups ( Figure 3A,B). WT and HLA-A11 R hESCs were implanted subcutaneously in NSG mice and humanized mice to assess immunogenicity, and teratoma size was measured periodically. There was no difference in teratoma size or growth curves in NSG mice between WT hESCs and HLA-A11 R hESCs ( Figure 3C).
The teratomas formed by WT and HLA-A11 R hESCs in humanized mice both tended to grow. However, the growth rate of teratomas derived from HLA-A11 R hESCs was faster than those derived from WT hESCs, and macroscopically, teratomas had a big difference in size may be due to the fact that HLA-A11 R hESCs in humanized mice are less attacked by the immune system ( Figure 3D Figure S6C). Then, WT and HLA-A11 R ECs were separately transplanted into subcutaneous bilateral hindlimbs, and Matrigel plugs were observed 3 weeks post-transplantation ( Figure 4A). Markedly, the macroscopic image of the HLA-A11 R EC plug showed a better blood supply than the WT EC plug ( Figure 4B). And, there is no obvious difference in the size of teratomas ( Figure S6D). Analysis of serial sections of the Matrigel plugs formed by WT or HLA-A11 R ECs in humanized mice indicated that human CD4 + ($22%) and CD8 + ($5%) T cells infiltrated with WT ECs. However, the immunogenicity of HLA-A11 R ECs, which infiltrated with human CD4 + ($10%) and CD8 + ($2%) T cells, appeared to be much weaker than that of WT ECs ( Figure 4C,D). Results showing infiltration of CD8 + T cells to be less than that of CD4 + T cells indicated that matched HLA-A might attenuate HLA-I-mediated CD8 + T-cell responses. Furthermore, HLA-A11 R ECs organized into more structures resembling primitive vascular structures, which occasionally contained erythrocytes than WT ECs according to staining of HE and CD31 ( Figure 4E), which suggested the potential of hESC-derived ECs as therapeutic cells to promote vascular regeneration. Collectively, HLA-A11 R ECs had lower immunogenicity than WT ECs in humanized mice with an established HLA-A11 + human immune system.

| Integration of the iC9 suicide cassette into HLA-A11 R hESCs for safety insurance
To improve the safety of HLA-A11 R hESCs in stem cell-based therapy, we precisely installed the inducible caspase-9 suicide system into the AAVS1 safe harbour to generate iC9-HLA-A11 R hESCs ( Figure 5A).
After co-electroporating gRNA/Cas9-expressing and iC9-cassette plasmids into HLA-A11 R hESCs, we identified positive PCRgenotyping products (iC9-HLA-A11 R hESCs) in 53.3% of genotyped clones ( Figure S7A,B and Table S3). First, iC9-HLA-A11 R hESCs were treated in the presence of AP1903 at different concentrations and different times to explore the optimal concentration of AP1903. Treatment with AP1903 at 10 nM for 8 h caused almost all cells to die, but no difference in cell death (all cells died) was observed if AP1903 at 0.001 to 1000 nM was used for 24 h ( Figure S7C). Hence, 10 nM concentration of AP1903 was selected for subsequent apoptosis-induction experiments. Treatment with AP1903 (10 nM) induced obvious apoptosis of iC9-HLA-A11 R hESCs within 2 h; Cells became detached from the culture dish, shrank, and degranulated after 24 h ( Figure 5B). The killing efficiency at different representative time points was measured by counting the cell viability with trypan blue ( Figure 5B): almost no cells were alive after 24 h. In addition, >99% of iC9-HLA-A11 R hESCs were stained positive with annexin V (apoptosis marker) after treatment with AP1903 for 24 h (Figures 5C and S7F), but this effect was not shown in WT hESCs upon AP1903 treatment, or in iC9-HLA-A11 R hESCs not treated with AP1903. This distinguishment was not caused by rates of cell proliferation ( Figure S7D), but possibly by the relatively high expression of iC9 detected in iC9-HLA-A11 R hESCs compared with that in WT and HLA-A11 R hESCs ( Figure S7E). In summary, precise installation of the iC9 suicide cassette on HLA-A11 R hESCs could eliminate these cells efficiently upon AP1903 treatment. We also demonstrated that the pluripotency of iC9-HLA-A11 R hESCs was normal ( Figure S4A-C). Additionally, both engineered hESC lines displayed a normal karyotype, consistent with that of WT hESCs ( Figure 5D and Table S4). To uncover the mutations that occurred during genome editing and single-clone cell culture, we conducted whole genome sequencing (WGS)-based off-target analysis ( Figure S8A). First, the potential off-target sites of each gRNA were identified by Cas-OFFinder in the hg38 human genome allowing for up to 5 bp of mismatches ( Figure S8B). Except for the widely known NGG, the protospacer adjacent motif (PAM) sequence also included NAG and NGA. The aligned sequences obtained from engineered hESC lines were compared with their parental WT hESCs.
Potential off-target sites based on gRNAs were SPATA2P1-RN7SKP6 and Tango6 genes found in HLA-A11 R hESCs, and the SP100 gene in iC9-HLA-A11 R hESCs ( Figure S8C). Potential offtarget sites that are not gRNA-based but related to cancer genes were Metazoa-SRP, RARA, and ACSL6 found in HLA-A11 R hESCs, and PLCG1 found in iC9-HLA-A11 R hESCs ( Figure S8D). These seven potential off-target genes were confirmed by next generation sequencing (NGS), almost no off-target probability was observed 30 (Table S6). Overall, we generated customized immunocompatible hESCs with enhanced safety by installing an effective inducible sui-

| DISCUSSION
In this study, we have generated an individualized immunocompatible hESC line with safety guarantee using the HLA-A*11:01-retained strategy, which covers roughly 21% of the Chinese population ( Figure 6). According to data showing that 95% of the Chinese population could be covered by 14 HLA-A R or 37 HLA-B R or 18 HLA-C R hESC lines ( Figure S1A and Table S7) Figure S1D and Table S9). Next, based on compliant hESC resources established by the National Stem Cell Resource Bank and the cell preparation platform that complies with GMP regulations, we aim to manipulate the same process on clinical-grade hESC lines with high-frequency HLA-A alleles under GMP conditions. In this way, we could achieve establishment of an HLA-A R bank of 14 hESC lines covering 95% of the Chinese population and other broad populations worldwide ( Figure 6). Establishment of such a universal stem cell bank is essential to support the development of allogeneic hESC-based therapy. The approach will reduce the cost of preparation but also enable advanced preparation for rapid transplantation for patients with acute diseases. Similarly, Yamanaka and colleagues constructed a clinical-grade haplobank of 27 iPSC lines from seven donors in accordance with GMP regulations, covering approximately 40% of the Japanese population. 31 In contrast to haplobanks of iPSCs, which requires multiple strains of cells to cover a broader population, our anticipated HLA-A R bank of hESCs could be achieved more readily using fewer cell lines.
To generate individualized immunocompatible donor cells, Li and colleagues knocked out B2M and transferred the exogenous HLA-A*11:01 gene to 70% iPSCs by lentiviral infection. 18 The potential risks associated with this strategy have been considered in our work, including retention of non-classical HLA-I, deletion of HLA-II, and a better approach for integration. We chose Homology directed repair (HDR)-mediated site-specific integration to knockin our genes of interest, promoted by the CAG promoter. Interestingly, we observed that the exogenous genes promoted by the EF1α promoter were silenced in HLA-A11 R hESCs (data not shown) in accordance with the results generated by Carman and co-workers. 32 Further studies are required to compare the expression of integrated genes promoted by different promoters in hPSCs.
Besides incompatibility with the immune system, hPSC-based therapy also faces the challenge of tumorigenicity, which can be caused by undifferentiated and/or immature cells, genetic mutations, 33 or infections. The presentation of tumour antigens by HLA-I is critical to the success of clearance aimed at stimulating antitumour CD8 + T-cell responses. 13,34 HLA-A11 R hESCs retained the capacity to present internal 'danger signals' by the HLA-A11 molecule with immune surveillance. Immune surveillance should acquire more attention in stem cell-based applications, whereas cancer cells are transmissible in other species (e.g., Tasmanian devil). 35,36 To further ensure safety, the suicide gene approach is an additional safeguard. We installed a drug-inducible suicide system into HLA-A11 R hESCs. We showed that iC9-HLA-A11 R hESCs could be induced to commit suicide efficiently in vitro by AP1903, used in clinical trials. [37][38][39] Although the drug-inducible suicide system was not conducted in a mouse model here, previous studies have indicated that this system is effective in vivo. 32,40 Suicide switches may reduce risks, but also abrogate the potential for functional effects. Hence, a drug-regulatable system may provide a more attractive safety guarantee for functional effects without ablating the transplanted cells permanently. 41 In general, we generated a pilot immunocompatible hESC line with safety insurance that could be used to cover $21% of the Chinese population. These hypoimmunogenic hESCs can be protected from allogeneic immune cells while also maintaining immune surveillance competence. We aim to establish an HLA-A R bank of 14 hESC lines as a source of next-generation donor cells for regenerative medicine to cover 95% of the Chinese population and other broad populations around the world.