The hESC line NN GMP0050E1C3 was provided by Novo Nordisk as a research cell line and served as the foundation for the subsequent stepwise CRISPR-Cas9-mediated editing process to produce an immune-evasive hESC line. Customized plasmids used for the insertion of transgenes were ordered at GenScript (Piscataway, NJ, United States) and are illustrated in Figure 1. The complete sequence is available upon request. The design included either mCherry or GFP as the fluorescent protein tracer, which allowed for fluorescence-associated single-cell sorting of edited cells. Two LoxP sequences were placed to allow for the removal of the fluorescent protein by adding Cre-recombinase. To facilitate in vivo tracing, hSEAP was introduced into the CLYBL safe harbor as the initial editing step, resulting in the cell line referred to as hESC + hSEAP. The hSEAP tracer was preferred as it can be measured from blood samples, which allowed for frequent sampling without the need for anesthesia and has been published as a functional tracer in vivo in mice (Bao et al., 2000; Nilsson et al., 2002; Hiramatsu et al., 2005). The hESC + hSEAP line was used for all subsequent CRISPR-Cas9 editing steps.

After the confirmation of hSEAP expression in vitro, a new hESC line was generated by inserting splice variant 2 of mCD47 into exon 2 of the B2M loci (GRCh38 chr15:44715509-44715531), thereby disrupting the B2M sequence and creating a knockout via knock-in. Next, exon 2 of CIITA (GRCh38 chr16:10895313-10895335) was targeted with sgRNA to generate a knockout. For additional tracing, in case the hSEAP was untraceable, firefly luciferase was randomly integrated using lentivirus to allow bioluminescent imaging (Contag et al., 1997), resulting in the cell line hESC + hSEAP+2xKO + mCD47.

Lastly, mCD47 was ablated from hESC + hSEAP+2xKO + mCD47 to generate a comparative hESC line, referred to as hESC + hSEAP+2xKO + mCD47KO, to investigate the protective effects mediated exclusively by CD47.

All hESCs were screened for correct genomic insertion, the absence or presence of targeted proteins, pluripotency, the capability of differentiation into all three germ layers, and a normal karyotype. hESCs were assessed after the removal of the fluorescent cassettes by Cre-recombinase.

After the assessment, hESC + hSEAP, hESCs + hSEAP+2xKO + mCD47, and hESC + hSEAP+2xKO + mCD47KO cell lines were used for two in vivo pilot studies to test hSEAP as a marker to follow hESC survival in vivo and assess rejection.

All cells were cultured at 37°C with 5% CO₂. Upon passage of cells, a 10 µM ROCK inhibitor (Tocris, Bristol, United Kingdom) is added to the media for the first 24 h. Reagents were bought from Thermo Fisher Scientific (Waltham, MA, United States) unless stated otherwise.

hESCs were grown on plates coated with 0.25 μL/cm² iMatrix-511 (Nippi, Tokyo, Japan) in the hESC media (NutriStem hPSC XF Medium (Sartorius, Göttingen, Germany) with 0.5% human serum albumin (Akron Bio, Boca Raton, FL, United States), 0.5% penicillin/streptomycin, and 5 µM FGF (PeproTech, Cranbury, NJ, United States)). hESCs were passaged every 3–4 days with Versene (1X) and replated in the hESC media.

For editing, 0.5 µL of sgRNA (44 µM ribonucleoprotein) (IDT, Coralville, IA, United States) was mixed with 0.5 µL Cas9 (33 µM ribonucleoprotein) (IDT) and incubated for 15 min at room temperature to form the CRISPR-Cas9 complex. hESCs were harvested using TrypLE for 5 min and washed once in PBS. A total of 4*10⁵ hESCs were collected by centrifugation for 3 min at 300 G and re-suspended in 9 μL R buffer and 2 µL Cas9 electroporation enhancer (10,8 µM) (IDT). For the insertion of transgenes, plasmids were added to the CRISPR-Cas9 complex during the last centrifugation step. Re-suspended hESCs were mixed with the CRISPR-Cas9 complex and electroporated with the 10 µL Neon^TM Transfection System (pulse width: 20, pulse: 2, 1,100 V). hESCs were plated in the hESC media with the ROCK inhibitor and 10% KO-serum. The next day, the media was changed to the hESC media. hESCs receiving only plasmid were included as a control to assess the degradation of unintegrated plasmid. After 10 days, plasmid controls were checked for the degradation of the plasmid visible by the lack of fluorescence, and the edited cells were checked for the fluorescent signal. Subsequently, the edited cells were FACS-sorted as single cells in wells of a 96-well plate. After 14 days, the sorted hESCs were screened for the knockout of CIITA, and the hESC + hSEAP+2xKO + mCD47KO line was screened for CD47 knockout by indel detection by amplicon analysis (IDAA), which assesses the formation of indels by comparing the amplicon length of the edited cells with those of the wildtype control. hESCs were additionally screened for the insertion of hSEAP and murine CD47 by PCR amplification of the 3′ and 5′ ends of the insert, including part of the DNA backbone. Two clones from each cell line had successful editing confirmed by Sanger sequencing. Based on the sequence and growth rate, one clone from each line was selected. To determine the mono- or bi-allelic insertion, long-range PCR, covering the plasmid, homolog arms, and part of the DNA backbone, was made using the Platinum™ SuperFi II PCR Master Mix and the manufacturers’ protocol with an extension time of 3.5 min. After the assessment of plasmid insertion, the fluorescent cassettes were removed by electroporation of 100 ng Cre mRNA using the same set-up as that for editing. hESCs were single-sorted and assessed for the correct removal of the cassette by PCR amplification of the 3’ end of the plasmid. Primers used for the various PCR experiments and sequencing are listed in Table 1. All primers were ordered from Eurofins Genomics (Luxembourg).

hESCs were harvested with Versene for 15 min and differentiated cells with Accutase for 5 min, washed with PBS, and either fixated and permeabilized using the transcription factor buffer set from BD Bioscience (cat# 562574) before staining or directly stained by resuspension in flow buffer (PBS + 1% BSA (Sigma)) + antibody (Table 2). The cells were incubated with antibodies for 15–30 min at 4°C, followed by 3 min of centrifugation at 300 G. Subsequently, the cells were washed twice in flow buffer and finally re-suspended in 200 µL flow buffer. The cells were analyzed on the cytoFLEX S (Beckman Coulter, Brea, CA, United States), and the data were processed in FlowJo.

hESCs were washed with PBS, after which they were fixated in 4% PFA for 15 min. The fixated hESCs were washed three times in PBS before permeabilizing with 0.2% Triton X (Sigma) for 20 min, followed by 30 min of blocking with 3% BSA. Primary antibody (Table 2) was diluted in 3% BSA and added and incubated O/N at 4°C. The next day, hESCs were washed three times, and the secondary antibody was added for 1 h RT in the dark. hESCs were washed three times, after which DAPI (Table 2) was added for 7 min in the dark. Finally, hESCs were washed four times, PBS was added to the well, and images were obtained on the same day using a fluorescent microscope. The plates were stored in the dark at 4°C until use.

Spontaneous differentiation was achieved by the formation of embryoid bodies for 7 days. hESCs were collected in colonies by Accutase treatment for 1 min, cell scraping, and 1 min centrifugation of the collected cells. hESC clusters were carefully re-suspended in hESC media and plated in low-attachment plates with media change every second day. After 7 days, embryoid bodies were transferred to Matrigel-covered plates with fibroblast media (DMEM+10%FBS + FGF + P/S). The media was changed every second day. After 14 days, the cells were immunocytochemically stained for the trilineage markers alpha-fetoprotein (AFP), smooth muscle actin (SMA), and beta-III tubulin (TUBIII) (Table 2).

To investigate the presence of promoter shutdown, hESCs were differentiated into forebrain neural precursors following a protocol adapted from Gantner et al., 2021. Before differentiation, hESCs were adapted to iPS-brew XF basal media (Miltenyi Biotec, Bergisch Gladbach, Germany) and laminin-521 (BioLamina, Sundbyberg, Sweden) coating. On day −1, hESCs were split with Accutase and plated as a confluent monolayer. The next day, hESCs were washed with PBS, and the media was changed to neural media (96% 1:1 DMEM and Neurobasal, 2% B27, 1% N2, 1% NNEA, 0.5% P/S, 0.5% GlutaMAX, and 0.09% β-mercaptoethanol) with the addition of SMAD inhibitors (100 nM LDN193189 and 10 µM SB431542) to induce neuroectodermal induction. Media was changed daily, and after 11 days, cells were passaged 1:2.5 with EDTA. The next day, the media was changed to neural media with 20 ng/mL FGF, with daily media change until day 18. On day 18, cells were passaged with Accutase and seeded at a 1:5 ratio. The media was changed the following day to neural media with FGF2. On day 21, the media was changed to neural media supplemented with 200 nM ascorbic acid, 40 ng/mL BDNF, 40 ng/mL GDNF, 50 µM dcAMP, and 1 μg/mL mouse laminin. After this, the media was changed every second day until day 35, at which hSEAP expression was assessed. On day 18, some of the cells were harvested and analyzed for neural differentiation markers and mCD47 expression (Table 2).

The amount of hSEAP in the cell media and serum was measured using the kit and the protocol from the Phospha-Light™ SEAP Reporter Gene Assay System. In short, 50 μL of cell media was diluted with a 50 µL dilution buffer and heated for 30 min at 65°C. The diluted sample (50 µL) was incubated at RT with a 50 µL assay buffer, after which a 50 µL reaction buffer was added for 20 min. The reaction was analyzed in a black-well 96-well plate using a luminometer for 0.1 s.

For serum samples, 12–25 µL of serum was mixed with 38–25 µL of the dilution buffer and processed as described above.

The in vivo studies were conducted according to the European legislation on animal experimentation (directive 2010/63/EU) and were approved by the Danish Animal Experiment Inspectorate and the Novo Nordisk Animal Welfare Council. All the studies were reported according to the ARRIVE guidelines (Percie du Sert et al., 2020). BALB/c mice were housed in groups of 4–6 animals per cage under the standard conditions (12/12 h light–dark cycle with ad libitum access to standard chow and water). Animals were acclimatized for 11 days, after which they were weighed once a week (Figure 2) and observed daily with a focus on the size of the subcutaneous transplant.

For the in vivo studies, edited hESCs had been single-cell sorted based on the presence of the fluorescent molecules GFP or mCherry expressed by the insert (Supplementary Figure S2). Single-sorted cells were expanded, genotyped, and sequenced as previously described to ensure the correct insertion and a pure population of edited cells. hESCs were collected by Versene for 15 min and re-suspended in 100 µL of ice-cold Matrigel, kept in separate marked tubes, and stored on ice until transplantation. Transplantations were performed on anesthetized mice. Anesthesia was induced in an induction chamber (4% isoflurane and 1L/min O₂) and maintained through a nose cone (2% isoflurane and 1L/min O₂). The injection site was shaved and cleaned with ethanol, and the negative toe-pinch reflex was verified before initiating injection.

hESCs in Matrigel were mixed once by pipetting and collected using a G30 insulin needle. All equipment was kept on ice to prevent the solidification of the Matrigel. Using the insulin needle, hESCs were subcutaneously injected into the left dorsal flank region.

During the studies, blood samples were obtained to follow the survival of the graft. At each time point, approximately 100 µL of blood was sampled from the sublingual vein into a 100 µL lithium-heparinized Microvette (Sarstedt, Nümbrecht, Germany). Blood was centrifuged for 10 min at 1.500 x g at 4°C to collect serum. Serum was transferred to freezing tubes placed on dry ice and stored at −80°C.

For the collection of tissue, the animals were terminally anesthetized in an induction chamber (4%–5% isoflurane, 1L/min O2), and up to 1 mL of intra-orbital blood was obtained and processed as described above. After blood sampling, the animals were euthanized by cervical dislocation, and various tissue samples were collected: the injection site was dissected from the surrounding tissue, and the regional lymph node draining the implantation site was macroscopically evaluated and collected. Both were placed in 4% PFA. Additionally, the spleen, kidney, and liver were dissected, placed in 4% PFA, and kept for potential future studies. Injections and euthanasia were conducted in a blinded manner by the person carrying out the procedures. No animals were excluded from the studies, and no signs of distress or pain were observed. Humane endpoints included general humane endpoints as well as any signs that the graft size affected normal body function.

Study I was a pilot study designed to test if the inserted hSEAP could be monitored in vivo and provide information on the survival of immunogenic hESCs (cell line E1C3) in immunocompetent mice (Figure 3). For this, 24 male 10-week-old BALB/c mice with an average weight of 22.4 ± 0.9 g (mean ± SD) from Janvier Labs (France) were divided into three groups (N = 8) and stratified by weight. Group 1 received subcutaneous transplantation of low-dose (1*10⁵) hESC + hSEAP. Group 2 received subcutaneous transplantation of high-dose (2.5*10⁶) hESC + hSEAP, and group 3 received subcutaneous transplantation of high-dose (2.5*10⁶) non-edited hESCs.

To uphold animal welfare legislation and not exceed the recommendations of blood sample volume per animal during the course of the study, animals from each group were sub-grouped into groups A and B, respectively. Animals from group A were sampled on day 0 before anesthesia and on day 7. Group B was sampled on days 1 and 10.

On day 14, the animals were euthanized, and the study was terminated.

The second in vivo study was designed as a pilot study to test whether the knockout of B2M and CIITA in combination with mCD47 overexpression could protect transplanted hESCs from rejection in an immunocompetent host. To assess the effect of mCD47, a mCD47 knockout cell line hESC + hSEAP+2xKO + mCD47KO was included.

The design of study II was based on findings from study I. Using the hSEAP means from day 1 of study I, a power of 0.8, and an alpha value of 0.05, the group size for blood sampling was calculated to be n = 4 (Figure 4). The sample size for the collection of tissue for histology was set to three, as the study set out to confirm the findings from the hSEAP measurement and give an indication for immune response by either showing the presence or lack of stained cells (binomial endpoint).

Eight-week-old male BALB/c mice with an average weight of 25.5 ± 1.3 g (mean ± SD) from Charles River (Germany) were randomized into three groups stratified by weight (N = 18) using a computer-based random number generator. All groups had 1.25*10⁶ hESCs subcutaneously transplanted. Group 1 received hESC + hSEAP, group 2 received hESC + hSEAP+2xKO + mCD47, and group 3 received hESC + hSEAP+2xKO + mCD47KO.

Animals from the three groups were randomly sub-grouped into a and b for blood sampling. Blood samples from sub-group a were collected on days 1, 8, and 15. Blood samples from sub-group b were collected on days 3 and 11. On day 22, samples from groups a and b were pooled for the remaining study blood collected. To determine hESC survival during the study, serum was extracted, as illustrated in Figure 4, and analyzed within 24 h of collection.

To assess the injection site at different time points, three animals were euthanized from each of the three groups on days 1, 3, 8, and 15, as described above. For days 1, 3, and 8, a non-treated animal was euthanized and included as a control. On day 25, the 5 remaining animals in each group were euthanized, and tissue and blood were collected as previously described.

One animal had a bite mark in the injection area, which could be signs of itch and irritation or a bite from another mouse.

The tissue was fixed in 4% PFA for at least 72 h, after which the injection site was divided into halves. Both halves of the implantation site and the regional lymph node were dehydrated in ethanol and cleared in Clearene Solvent (Leica, Wetzlar, Germany), after which they were embedded in paraffin. The blocks were sectioned into 3–5 µm slices and arranged on glass slides, with each slide containing a section each from the halves of the injection site and the lymph node.

The slides were H&E stained by removing paraffin with 3 × 5 min xylene treatment, followed by 3 × 5 min 99% ethanol, 2 × 5 min 96% ethanol, 1 × 5 min 70% ethanol, and 5 min deionized H₂O treatment. Afterward, the slides were treated with Mayer’s hematoxylin (Table 2) for 3 min and washed in tap water for 5 min. The slides were stained with 0.1% eosin (Table 2) for 1 min, shortly washed in deionized H₂O, and dehydrated in increasing concentrations of ethanol (70%–99%).

The slides were fluorescently stained after the removal of paraffin, as described above. The slides were treated with TEG and microwaved for 15 min, followed by 5 min of rinsing in tap water. Then, 1% H₂O₂ was added for 15 min and washed for 2 min in tap water. To reduce non-specific binding, the slides were treated with 0.05% Tween 20 in TBS for 3 × 2 min and blocked with 0.5% TNB buffer for 30 min. Primary antibodies were diluted in TNB according to Table 2 and incubated for 1 h at room temperature or overnight at 4°C. Afterward, the slides were washed for 3 × 2 min in 0.05% Tween 20 and treated with the secondary antibody (Table 2) and DAPI for 30 min at RT, followed by 3 × 2 min wash, after which fluorescent dye was diluted (Table 2) and added for 15–30 min. Lastly, the slides were washed in 0.05% Tween 20 and rinsed in deionized H₂O, after which the slides were mounted in fluorescent mounting. The quantification of fluorescent signals was blinded and carried out using ImageJ software.

The guides (sgRNAs) used in this study have previously been shown by IDAA to have a cutting efficiency above 50% (Supplementary Figure S1). For the insertion of transgenes, the length of the homology arms gave varying integration efficiencies. Insertion of the transgene encoding hSEAP with 300 bp arms showed ∼1% integration efficiency, and the transgene encoding for mCD47 with 800 bp arms showed integration efficiency of 0.4–3.8% (Supplementary Figure S2). Correct insertion of the transgenes was confirmed by long-range PCR, which showed mono-allelic insertion of hSEAP and bi-allelic insertion of mCD47 (Figure 5A). Digital droplet PCR confirmed the mono- and bi-allelic insertion and showed no random integration of the transgenes (Supplementary Figure S3). The selected edited hESCs showed the expression of the pluripotency markers OCT4, NANOG, and SSEA3 by immunocytochemical staining (Figure 5B), and more than 93% of the hESCs stained positive for OCT4 and SOX2 in flow analysis (Figure 5C). Pluripotency was further confirmed by the hESCs’ abilities to differentiate into cell types from all three germ layers upon spontaneous differentiation (Figure 5D). All the results were comparable to the pluripotency state of the non-edited wildtype hESCs (Supplementary Figure S4).

Several other tests were conducted that confirmed the quality of the hESCs after editing, including PluriTest, KaryoStatTM, STR, and off-target sequencing (Supplementary Figures S5–S8).

The expression of the inserted hSEAP was assessed in vitro and showed a significantly increased signal compared to wildtype hESCs (Figure 6A). Additionally, luciferase expression was assessed and showed luminescence upon treatment with luciferin (Supplementary Figure S9). Murine CD47 was detected by flow analysis in hESC + hSEAP+2xKO + mCD47 but could not be detected in the knockout line hESC + hSEAP+2xKO + mCD47KO (Figure 6B). The functionality of the expressed mCD47 was demonstrated by a binding assay to murine SIRPα, in which 100% binding was seen for hESC + hSEAP+2xKO + mCD47, but no binding was seen for the knockout line hESC + hSEAP+2xKO + mCD47KO (Figure 6C). These results strongly indicate that the inserted mCD47 transgene in hESC + hSEAP+2xKO + mCD47 can mediate an anti-phagocytic signal by interacting with the SIRPα receptor, which could not be seen for hESC + hSEAP+2xKO + mCD47KO. Flow cytometry for MHC-I revealed no expression, which confirmed the knockout because of the transgene insertion in B2M (Figure 6D). Sequencing of the target site in CIITA showed the generation of frameshift (Supplementary Figure S10), disrupting the MHC-II expression.

Sequencing of the 5’-end of the transgenes showed correct removal of the fluorescent cassette for both hSEAP and mCD47 (Supplementary Figure S11). For mCD47, sgRNA cut 81 bp downstream from where the homology arms were designed. This results in an 81 bp deletion in the 5′ and an 81 bp insertion in the 3’ end. As the study aimed to knock out the B2M gene, the 81 bp swap is of no consequence and may increase the chance of an efficient knockout.

The full length of the inserted mCD47 transgene was sequenced to check for unexpected genomic alterations. As both hESC-hSEAP+2xKO + mCD47 and hESC-hSEAP+2xKO + mCD47KO reside in the same clone transfected with the mCD47 transgene, the sequencing results are identical, except for the knockout in hESC-hSEAP+2xKO + mCD47KO (Supplementary Figure S12). In addition to the 81 bp shift from the homology arms previously described, a heterozygous deletion from 2,709 bp to 3,916 bp was seen, covering the last part of the UbC promoter and the first half of the CD47 coding region (Supplementary Figure S12). This deletion causes mono-allelic expression of mCD47, despite a bi-allelic insertion. To investigate whether the deletion was seen due to folding of the genomic DNA used for PCR, increased denaturing temperature, time, and GC-enhancer were applied in the PCR, but it showed no change in amplicon, indicating the hairpin formation of the plasmid during editing (Supplementary Figure S13).

Through differentiation using a neural differentiation protocol, hESC + hSEAp+2xKO + mCD47 showed the expression of neural markers (Supplementary Figure S14) and continued to show significant expression of hSEAP (Figure 7A) and mCD47 (Figure 7B) in vitro, which supports the notion that no promoter shutdown is taking place as a result of the differentiation. As hESC + hSEAP and hESC + hSEAP+2xKO + mCD47KO carry the identical hSEAP transgene as hESC + hSEAp+2xKO + mCD47, the lack of promotor shutdown in this line is expected to be representative for all the lines. Since both hSEAP and mCD47 were expressed continuously, the generated hESCs were further tested in an in vivo setting.

To test whether the hSEAP signal was strong enough for in vivo monitoring and obtain a timeline for the rejection of wildtype hESCs, an in vivo pilot study I was conducted (Figure 3). The successful detection of hSEAP would provide a timeframe for the rejection of wildtype hESCs in BALB/c mice. hSEAP analyses of serum from days 0, 1, 7, 10, and 15 showed a hSEAP signal for days 1 and 7 for both groups injected with low and high numbers of hSEAP-expressing hESCs (Figure 8A). Both samples from day 0 and wildtype hESCs showed no detectable signal. On days 1 and 7, the signal was significantly higher for the group injected with high numbers of hESC + hSEAP compared to that of the wildtype hESCs, according to Kruskal–Wallis analysis. The high number of hESC + hSEAP, furthermore, showed higher signals than the group injected with low numbers of cells. This confirms that differences in the numbers of hESCs are detectable in vivo and hereby highlights the capability of hSEAP as an in vivo biomarker. On day 10, the signal had decreased greatly, and it was depleted by day 14, indicating that the rejection of non-edited hESCs was initiated before day 10 in immune-competent mice.

The second in vivo study (In vivo study II) assessing the survival of gene-edited hESCs (Figure 4) showed no significant difference in hSEAP between the hESC + hSEAP, hESC + hSEAp+2xKO + mCD47, and hESC + hSEAp+2xKO + mCD47KO at the different time points, as determined by Kruskal–Wallis analysis (Figure 8B). On days 8 and 11, hSEAP levels decreased but were still above the detection limit. On day 15, the signal was no longer detectable, but the animals were kept in case a few hESCs survived and proliferated. On day 22, there was still no sign of growth, and the remaining animals were euthanized on day 25.

To examine the survival and immune response of hESCs, we performed immunohistochemistry to characterize the implantation site and regional lymph nodes at various time points. Figure 9 shows representative immunohistology for the transplant site of mice injected with hESC + hSEAp+2xKO + mCD47. The injection sites were readily visible with H&E staining. Cluster formation of cells with larger nuclei is present inside the injection site on days 1 and 3 (Figure 9A). On days 8 and 15, the clustered nuclei are no longer detectable in the injection site and randomly distributed smaller nuclei are present instead, indicating immune infiltration. Staining for the human nuclear protein KU80 confirms that the cluster formations seen in H&E are human cells (Figure 9B). No KU80-positive cells could be detected on days 15 or 25, and only a few positive cells were present in the injection sites on day 8. No human cells could be detected in the lymph node at any of the time points. To evaluate the contribution of leukocytes, the injection sites were stained with the general leukocyte marker CD45. Staining for the immune cells was only carried out up until day 15 since no cell clusters of hESCs were found after that point. All time points showed CD45-positive cells; however, the injection site showed obvious signs of immune infiltration on days 8 and 15 (Figure 9C).

CD3 staining to detect T cells indicated targeted rejection (Figure 9D). Furthermore, few CD3-positive cells should be present if the hESCs are cleared by apoptosis. However, on days 1 and 3, T-cell counts were similar to those of the control samples (Figure 9E). On days 8 and 15, T-cell counts significantly increased compared to control samples and samples from days 1 and 3, as determined by Kruskal–Wallis analysis (Figures 9D, E), suggesting rejection by an adaptive immune response.