CRISPR-Cas9-mediated knock-in of cytomegalovirus US2 provides an alternative strategy for generating hypoimmunogenic hiPSC lines Authors: Juan J Novoa, Lonneke H Gaykema, Juliette A de Klerk, Janneke H D Peerlings, Rianne Y van Nieuwland, Arno van der Slik, Francijna E van den Hil, Valeria V Orlova, Manuel A F V Gonçalves, Cathelijne W van den Berg, Ton J Rabelink, Arnaud Zaldumbide Published: 2026 ABSTRACT AbstractMismatches in HLA haplotypes between donors and recipients significantly increase the risk of graft failure due to immune rejection. Knockout (KO) of beta-2 microglobulin (B2M) is the current standard for eliminating HLA class I (HLA-I) surface expression and protecting allogeneic products from T-cell–mediated rejection; however, complete HLA-I ablation can trigger natural killer (NK) cell “missing-self” responses and disrupt critical immune-regulatory interactions. To address these limitations, we introduced a cytomegalovirus-derived US2 encoding sequence into the AAVS1 safe-harbor locus of a human induced pluripotent stem cell (hiPSC) line. Flow cytometry showed that US2 selectively abrogates HLA-A2 surface expression while retaining low levels of total HLA-I. In coculture assays, US2 expression abolished HLA-A2 alloreactive T-cell activation without increasing NK cell degranulation, indicating preserved inhibitory signaling. Together, these findings establish US2-mediated immune evasion as a refined single-edit alternative to B2M KO, enabling selective HLA-I modulation while preserving critical immune-regulatory interactions.Supplementary InformationThe online version contains supplementary material available at https://doi.org/10.1038/s41598-026-58651-3.Subject terms: Immunology, Stem cells Introduction In the last ten years, the field of regenerative medicine has entered a new era, providing solutions to the shortage of organ donors and offering alternatives to immunosuppressive regimens. The combination of human induced pluripotent stem cells (hiPSCs), which are generated through the reprogramming of somatic cells, with genome editing technologies (e.g., engineered CRISPR-Cas nucleases) that target the antigen processing machinery, makes the “holy grail” of creating a “stealth” cellular product suitable for all patients a realistic possibility. Current strategies for developing universal “off-the-shelf” hypoimmunogenic hiPSC-derived products focus primarily on disrupting HLA class I (HLA-I) expression to prevent CD8+ T-cell-mediated rejection1,2. The most common approach involves the knockout (KO) of beta-2-microglobulin (B2M), an essential component for the stabilization and surface expression of HLA-I complexes. B2M binds to the HLA-I α-chain, ensuring its transport to the cell surface and enabling antigen presentation3. However, in line with the “missing-self” hypothesis, the complete absence of HLA-I expression renders B2M KO cells highly susceptible to natural killer (NK) cell-mediated lysis4. To mitigate this vulnerability, additional modifications have been envisaged, including the overexpression of nonclassical HLA-I molecules such as HLA-E or HLA-G5,6. These strategies are sometimes combined with the KO of CIITA, the master regulator of MHC class II expression, to prevent CD4+ T-cell priming and the knock-in of immune-inhibitory ligands such as CD47, PD-L1, and CTLA-4 to enhance immune evasion7. Although effective, gene ablation strategies are not without risk. The long-term safety and functional integrity of tissues derived from gene-edited hiPSCs have not been fully evaluated8,9. In addition to its critical role in HLA-I surface expression, B2M can also form heterodimers with HLA-I-like proteins, including the neonatal Fc receptor (FcRn) and homeostatic iron regulator (HFE). Through its association with FcRn, B2M contributes to the recycling and extended half-life of immunoglobulin G (IgG) and albumin. In complex with HFE, it regulates iron uptake and homeostasis. Consistent with these pleiotropic functions, B2M KO mice exhibit systemic iron overload, with increased iron deposition in the liver, spleen, pancreas and kidney10,11. These findings underscore the potential drawbacks of full B2M ablation and support the exploration of alternative strategies for generating a universal source for hiPSC-derived cell grafts suitable for transplantation. Insights from virology offer valuable examples of strategies evolved by viruses to manipulate host gene expression and promote long-term survival of infected cells, thereby maximizing viral replication and spread. Notably, members of the Herpesviridae family have developed highly sophisticated immune evasion mechanisms12,13. This is particularly evident for human cytomegalovirus (HCMV), a beta-herpesvirus with approximately 235 kb genome containing ~ 170 canonical open reading frames (ORFs), many of which are dispensable for viral replication in vitro and implicated in immune modulation14. Moreover, ribosome profiling has revealed a substantially broader translational landscape, with hundreds of translated viral ORFs detected during infection15. Among the best-characterized HCMV immunoevasins is the US2–US11 region, which primarily interferes with MHC class I surface expression. US2 and US11 promote dislocation of newly synthesized MHC class I heavy chains from the endoplasmic reticulum to the cytosol for proteasomal degradation, US3 retains MHC class I molecules in the endoplasmic reticulum, and US6 inhibits TAP-dependent peptide translocation14. Importantly, US2 activity is locus-, allele- and context-dependent: structural and mutational studies show that US2 engages HLA-A2 near the junction of the peptide-binding groove and α3 domain, with sensitivity determined by subtle sequence variation near the α2/α3-domain junction16. Consequently, US2 preferentially downregulates HLA-A, HLA-G and many HLA-B alleles, whereas HLA-E and selected HLA-B or HLA-C allotypes may be relatively resistant17. This selective modulation is relevant for hypoimmunogenic engineering because preservation of HLA-E and residual HLA-I molecules can maintain inhibitory signaling through NK-cell receptors such as CD94/NKG2A, KIR and or CD85j, potentially limiting missing-self activation18. In this study, we hypothesized that viral-derived immune system modulators (immunoevasins) can serve a natural resource for generating cell products with novel hypoimmunogenic characteristics. To this aim, we generated and characterized a hypoimmunogenic hiPSC line through targeted integration of the viral CMV immunoevasin gene US2 into the AAVS1 locus at chromosome position 19q13.42, a well-established genomic safe harbor that supports stable and homogeneous transgene expression in gene-edited cells. We then compared the immunological properties of these US2-overexpressing hiPSCs and their hiPSC-derived endothelial cells with those of their conventional B2M KO counterparts. Our results indicate that ectopic US2 expression effectively reduces HLA-I surface expression in both undifferentiated hiPSCs and hiPSC-derived endothelial cells, without complete ablation. These findings support the potential use of viral immunoevasin-based strategies to mitigate immune rejection of hiPSC-derived products, particularly when these strategies are applied to parental lines with an HLA-A2-positive phenotype. Results Genetic modification of hiPSCs at the B2M or AAVS1 loci. In hiPSCs, B2M expression was disrupted by the insertion of the coding sequence for EGFP immediately downstream of the B2M start codon (Fig. 1A). To induce US2 expression, its coding sequence was inserted into the AAVS1 locus under the control of a CAG promoter, a strong synthetic promoter composed of the cytomegalovirus immediate-early enhancer, chicken β-actin promoter, and rabbit β-globin splice acceptor sequence (Fig. 1B). This was achieved by introducing the respective homologous recombination (HR) donor DNA together with a matched AAVS1-targeting CRISPR/Cas9 nuclease complex. Correct genomic modification was confirmed by genotyping PCR assays using primers yielding amplicons diagnostic for HR-mediated endogenous/exogenous DNA junctions. A prime pair specific for transgenic DNA served as an internal control for genomic DNA integrity (Fig. 1A-F). Unmodified hiPSCs and a no-template control (H2O) were included as negative controls. Insertion of EGFP at the AAVS1 locus was used as an additional control to account for sustained promoter activity and transgene expression through differentiation and assess potential adverse effects resulting from gene targeting at this locus (Supplementary Figure S1A). This PCR analysis confirmed precise HR-mediated gene targeting in three B2M-KO clones (i.e., B2M-KO#2, #5, #7; Fig. 1C and D), three AAVS1-US2 clones (i.e., AAVS1-US2#2, #10, #32; Fig. 1E and F), and one AAVS1-EGFP control clone (Supplementary Figure S1B and C). Genetic modification of hiPSCs at the B2M or AAVS1 loci. Genomic modification strategy at the (A) B2M locus or (B) AAVS1 locus. The targeted area is indicated by a black triangle and the sequence included in the donor DNA that was used for homologous recombination-mediated gene knock-in is outlined by a gray rectangle. For both loci, 2 primer pairs were used to evaluate correct gene insertion. The sizes of the resulting PCR products are indicated below the gray rectangles. (C-F) Results of genomic PCRs performed with the primer pairs indicated in (A, B). Unmodified (Control) hiPSCs and a sample without cellular DNA (H2O) were used as negative controls. mRNA expression of G) B2M and H) US2, as measured by RT-qPCR in unstimulated and IFN-treated hiPSCs. Data are shown as mean ± SD. Different datapoint shapes represent independent clones (B2M-KO #2, #5 and #7 and AAVS1-US2 #2, #10 and #32). Statistical significance was evaluated via one-way ANOVA with Šidák’s correction for multiple comparison; each sample was compared with the corresponding similarly treated control condition, either unstimulated or IFN-γ-treated (** p < 0.01, **** p < 0.0001). To further validate the gene-edited hiPSC clones, total RNA was isolated from all the lines cultured for 48 h with or without IFN-γ stimulation. Unmodified hiPSCs were used as controls and were compared with B2M-KO clones (#2, #5, #7), AAVS1-US2 clones (#2, #10, #32), and the AAVS1-EGFP control clone. TAP1 expression was comparable across all the cell lines and increased uniformly in response to IFN-γ stimulation (Supplementary Figure S1D), indicating that the gene modifications did not impair IFN-γ responsiveness. As expected, B2M mRNA expression was significantly reduced in B2M-KO clones regardless of IFN-γ stimulation, confirming effective gene disruption (Fig. 1G). US2 transcripts were exclusively detected in the AAVS1-US2 clones and remained unaffected by IFN-γ treatment (Fig. 1H). Similarly, EGFP expression in AAVS1-EGFP clone was robust and unchanged upon IFN-γ stimulation (Supplementary Figure S1E), resembling the expression pattern observed for US2 (Fig. 1H). In contrast, B2M-KO clones presented lower baseline EGFP expression than did the AAVS1-EGFP control but registered a modest, nonsignificant increase in reporter amounts upon IFN-γ treatment (Supplementary Figure S1E). This finding suggests preservation of IFN-γ responsiveness via the endogenous B2M promotor at the edited locus. Furthermore, the EGFP mRNA levels measured via RT-qPCR correlated well with the protein expression levels assessed via flow cytometry (Supplementary Figure S2). Together, these results confirm that reporter gene targeting at the B2M locus effectively abrogated endogenous B2M mRNA expression, yielding functional B2M knockout, while retaining IFN-γ inducibility from the native promoter. Moreover, AAVS1 gene targeting results in stable, IFN-γ-independent expression of transgenes such as US2 and EGFP. US2 transgenic hiPSC lines maintain a pluripotent phenotype Comparative transcriptomic analysis of B2M knockout and AAVS1-US2 cells revealed a similar degree of overall transcriptional perturbation across the edited clones, as indicated by comparable numbers of up- or downregulated genes relative to the parental line (Fig. 2A). Gene ontology pathway analyses revealed that B2M knockout predominantly affected the class I peptide antigen processing pathway, whereas US2 overexpression primarily influenced genes involved in DNA damage and cellular stress responses (Supplementary Figure S3). Differential gene expression (DGE) analysis comparing edited clones to parental hiPSCs identified modulation of multiple intergenic long non-coding RNAs and downregulation of several immune-related genes, such as CTSF and NLRP2, in AAVS1-US2 hiPSCs (Fig. 2B). However, transcriptional differences across samples were relatively modest and appeared to be driven primarily by the top 10 DEGs contributing to principal components 1 and 2, as illustrated by the PCA loading plot (Fig. 2C). Gene expression profile of genetically modified hiPSCs. (A) Venn diagrams showing differentially expressed genes in B2M KO (pink) and AAVS1-US2 (blue) hiPSCs compared with the unmodified control. The upper panel shows upregulated genes, whereas the lower panel shows downregulated genes. (B) Beta plot representing the log fold expression change of each gene tested between the two conditions: B2M KO vs. parental (x-axis) and US2 vs. parental (y-axis). The gray dots represent genes that were not differentially expressed in either test (i.e., B2M KO vs. parental and US2 vs. parental) (n = 20769). The red dots represent the genes differentially expressed in the B2M KO vs. parental groups (n = 3), and the green dots represent the genes differentially expressed in the US2 vs. parental groups (n = 22). The blue dots represent the genes differentially expressed in both the B2M KO vs. parental and US2 vs. parental groups (n = 9). (C) Principal component analysis (PCA) plot with the top 10 factor loadings for PC1 and PC2. (D) Heatmap of the markers as defined by R-HSA-452,723 generated with gene counts per million (CPM). (E) Representative dot plots from flow cytometry analysis of hiPSC-derived endothelial cells stained with APC-conjugated anti-CD31 and Alexa Fluor 488-conjugated anti-VE-Cadherin antibodies. Percentage of double positive population is represented in each plot. Unstained endothelial cells were used to set the gates and undifferentiated hiPSCs were used as a negative control population. To assess whether gene editing strategies affected the undifferentiated state, we evaluated the expression of pluripotency-associated genes. Similarly to B2M-KO and unmodified parental lines, AAVS1-US2 hiPSCs presented high expression of SMAD2, SMAD4, NANOG, and SOX2, with lower expression of KLF4. Overall, the expression profiles of the core undifferentiated-state markers were comparable across all the cell lines (Fig. 2D). Based on these results, one representative clone per edited line (e.g. B2M-KO #5, and AAVS1-US2 #32), together with the parental line and AAVS1-EGFP control, was selected for further differentiation and immunological characterization. To confirm that US2 expression does not impair the development potential of hiPSCs, we generated endothelial cell from these clones. As displayed on Fig. 2, differentiation into hiPSC-derived endothelial cells (hiECs) was similarly efficient across all lines, yielding in average 90% VE-Cadherin⁺ CD31⁺ cells (Fig. 2E). B2M knockout hiPSCs lack HLA class I surface expression, whereas US2 expressing cells have partially reduced HLA class I expression To evaluate the immunoevasive potential of the US2 strategy in a differentiated cell type, we selected hiECs as our model for immunogenicity testing. Endothelial cells are clinically relevant targets for immune recognition due to their known interactions with both innate and adaptive immune cells19. Following genomic and transcriptomic verification, we assessed the impact of gene-editing strategies on HLA-I surface expression via flow cytometry. hiPSCs and hiECs were cultured with or without IFN-γ stimulation for 48 h. Parental and AAVS1-EGFP cells displayed similar HLA-I surface expression under basal conditions, with an 8- to 10-fold increase registered upon IFN-γ treatment in both hiPSCs and hiECs (Fig. 3A and C), confirming that the gene editing process itself did not impair HLA-I expression. In contrast, B2M-KO cells exhibited a complete absence of HLA-I surface expression, which was not restored by IFN-γ stimulation (Fig. 3A and C). AAVS1-US2 cells presented no detectable HLA-I expression under basal conditions. However, IFN-γ treatment led to a measurable increase in HLA-I levels, although expression remained 10- to 15-fold lower than that in the stimulated parental and AAVS1-EGFP cells (Fig. 3A and C). Thus, B2M disruption resulted in complete HLA-I ablation, whereas US2 expression produced partial HLA-I downregulation that was maintained after endothelial differentiation. HLA surface expression analysis of hiPSCs and hiECs. Surface expression of (A) HLA-ABC and (B) HLA-A2 was measured in untreated and IFN-γ-treated hiPSCs via flow cytometry (n = 3 biological replicates). Surface expression of C) HLA-ABC and D) HLA-A2 was similarly assessed in hiECs (n = 3 biological replicates). The results are shown as mean ± SD, and significance was evaluated via one-way ANOVA with Šidák’s correction for multiple testing; each sample was compared with a similarly treated control (unstimulated or IFN-γ-treated) sample (** p < 0.01, **** p < 0.0001). Given the allele-dependent activity of US2, we next assessed HLA-A2 expression20. HLA-A2 was induced by IFN-γ in the parental and AAVS1-EGFP lines but remained undetectable in both B2M-KO and AAVS1-US2 cells (Fig. 3B and D). These findings indicate that US2 expression selectively eliminates HLA-A2 from the cell surface while allowing partial retention of total HLA-I. Of note, all edited hiPSC clones show similar HLA-ABC and HLA-A2 modulation (Supplementary Figure S4A). US2 provides comparable immune protection to that of B2M KO against HLA-A2-reactive T-cells without increasing the risk of NK cells activation To evaluate the impact of HLA-I modulating strategies on immune recognition, hiECs were cocultured with either HLA-A2-reactive CD8⁺ T-cells or preactivated primary NK cells. Prior to coculture, hiECs were stimulated with 1000 IU mL− 1 IFN-γ for 24 h to induce HLA expression. In the CD8⁺ T-cell assay, T-cells cocultured with parental or AAVS1-EGFP hiECs produced detectable levels of MIP-1β, which was consistent with antigen-specific recognition and activation. In contrast, B2M-KO and AAVS1-US2 hiECs failed to induce MIP-1β secretion, indicating that both B2M knockout and US2 overexpression effectively prevented recognition and activation by HLA-A2-restricted CD8⁺ T-cells (Fig. 4A). Consistently, all B2M-KO and AAVS1-US2 hiPSC clones showed reduced T-cell activation compared with parental controls (Supplementary Figure S4B). T-cell and NK cell cocultures with hiECs. (A) MIP-1β secretion quantified via ELISA to assess the activation of HLA-A2-reactive CD8+ T cells after 18 h of coculture with IFN-γ-pretreated hiECs (1:4 E: T). T cells cultured in the absence of target cells served as the no-target control. Data are shown as mean ± SEM (n = 3). Statistical significance was determined by one-way ANOVA with Dunnett’s post hoc test (* p < 0.05, ** p < 0.01, **** p < 0.0001). (B) NK cell degranulation assessed by measuring CD107a expression in the CD3−CD56+ population following coculture with hiECs (1:1 NK: Target). Each symbol represents the average normalized data from each donor (n = 5). Degranulation data were normalized to the parental hiECs condition for each donor and expressed as percent change in CD107a⁺ NK cells. Each symbol represents the mean normalized value for an individual donor (n = 5). Data are shown as mean ± SD. Statistical significance was evaluated using a paired t test (*p < 0.05). (C, D) Surface expression of (C) HLA-C, (D) HLA-Bw6 measured by flowcytometry in IFN-γ-treated parental hiECs (n = 1), AAVS1-EGFP hiECs (n = 1), B2M-KO hiECs (n = 3 technical replicates) and AAVS1-US2 hiECs (n = 3 technical replicates). To evaluate the consequences on NK cell reactivity, primary NK cells were freshly isolated from five independently donated buffy coats. NK cell purity (> 90%) was assessed by flow cytometry, and NK cell functionality was confirmed for each donor using the NK-sensitive K562 cell line. Supplementary Fig. 5S A-G shows a representative flow cytometry characterization of isolated NK cells, including the gating strategy used to quantify CD107a⁺ NK cells. Supplementary Figure S5H shows NK cell degranulation in the absence of target cells (negative control) and following coculture with K562 cells (positive control). NK cell coculture data are shown as the change in the frequency of CD107a⁺ NK cells relative to parental hiECs across all donors. Coculture with AAVS1-US2 hiECs did not result in a significant increase in NK cell degranulation. In contrast, NK cells cocultured with B2M-KO hiECs exhibited a mean increase of 64.7% in degranulation when compared with that of the parental line (Fig. 4B). To better define the residual HLA-I signal in hiECs and the potential molecular mechanisms involved in NK evasion, we additionally assessed HLA-C and HLA-Bw6 expression in the selected differentiated clones (e.g. B2M-KO #5, AAVS1-US2 #32, AAVS1-GFP and parental). Flow cytometry analysis showed no significant reduction in HLA-C or HLA-Bw6 surface expression in AAVS1-US2 hiECs compared with parental and AAVS1-EGFP hiECs, whereas B2M-KO hiECs showed fluorescence intensities comparable to the corresponding isotype controls (Fig. 4C and D). Consistent with the HLA allotype of the parental cell line (HLA-Bw6 positive; HLA-Bw4 negative), HLA-C-dependent inhibitory KIR engagement is likely to represent the main HLA-I-dependent mechanism contributing to NK cell inhibition in this system21. Of note, HLA-E surface expression could not be detected under the conditions tested, indicating that HLA-E/CD94-NKG2A-mediated inhibition is unlikely to explain the preserved resistance of AAVS1-US2 hiECs to NK cell activation22. Collectively, these results demonstrate that US2 expression allows a selective HLA-I modulation providing protection from CD8⁺ T-cell-mediated recognition comparable to that conferred by B2M KO. In contrast, however, US2 expression uniquely preserves resistance to NK cell-mediated cytotoxicity, via, at least partially, maintenance of HLA-C as described previously17. This dual advantage underscores the potential of US2 transgene knock-in as a promising alternative to conventional immune evasion strategies for developing universal, allogeneic hiPSC-derived cell therapies. Discussion In this study, we have shown that hiPSCs can be successfully engineered via CRISPR-Cas9-mediated gene editing to stably express the CMV-derived immunoevasin US2. Unlike B2M knockout, which eliminates all HLA-I molecules, US2 expression resulted in selective HLA-I modulation, most notably complete HLA-A2 removal, while preserving low levels of total HLA-I surface expression. Rather than being a limitation, this partial HLA-I preservation may represent a strategic advantage for generating HLA-A2 hiPSC-derived cell products that are protected from CD8+ T-cell-mediated killing and, simultaneously, NK cell-triggered degranulation. Under inflammatory conditions involving IFN-γ treatments, US2 expression was sufficient to prevent the activation of HLA-A2-reactive CD8+ T-cells against both hiPSCs and hiECs, with an efficacy comparable to the “classical” B2M knockout approach. These findings align with our previous work demonstrating that lentiviral vector-mediated US2 expression in primary human islets confers full protection from alloreactive T-cell-mediated killing in vivo23. Together, our results thus support the potential of CMV US2-based strategies to shield transplanted cells from adaptive immune responses. Moreover, beyond T-cell evasion, US2 expression also mitigated NK-cell degranulation compared with B2M KO. This contrasts with B2M KO, which fully ablates HLA-I surface display and thereby promotes NK-cell “missing-self” activation, as evidenced by the increased NK-cell degranulation observed in this study. Mechanistically, US2 facilitates the translocation of selected HLA-I molecules from the endoplasmic reticulum to the cytosol for proteasomal degradation, with susceptibility determined by specific HLA sequence features24. In our system, this selectivity was reflected by complete HLA-A2 removal while maintaining residual HLA-I expression, including detectable HLA-C and HLA-Bw6 in AAVS1-US2 hiECs. This pattern is consistent with the concept that US2 does not act as a pan-HLA-I ablator, but rather reshapes the HLA-I repertoire in an allele-dependent manner. Llano et al. similarly showed that US2 can differentially affect class I allotypes while preserving inhibitory signals relevant for NK-cell recognition, including HLA-E and CD94/NKG2A-dependent protection in their model18. In our study, the parental line lacks HLA-Bw4 and did not show detectable HLA-E surface expression under the conditions tested, indicating that the preserved NK resistance is unlikely to be mediated through HLA-Bw4-dependent KIR signaling or HLA-E/CD94-NKG2A. Instead, preserved HLA-C may contribute to this phenotype through killer immunoglobulin-like receptors (KIRs) mediated inhibitory signaling21. In addition, US2 is not exclusively an HLA-I-targeting protein. Plasma membrane profiling has shown that US2 can downregulate additional cell-surface proteins beyond MHC-I, including molecules involved in NK-cell activation25. Overall, the allele-dependent activity of US2 implies that its effectiveness as a hypoimmunogenic engineering strategy will depend on the HLA background of the donor line and the immune specificity of the recipient. Combining multiple viral immunoevasins or targeting shared components of antigen presentation may therefore be required to achieve broader immune protection across different HLA contexts. In line with gene pleiotropism, complete gene deletion is rarely an inert process, as illustrated by the systemic iron overload, altered transferrin receptor expression, and increased oxidative stress, particularly in renal cells reported upon B2M ablation10,11,26,27. The use of a modulator, as opposed to a full ablator of immune recognition, may offer a more nuanced and potentially safer alternative for hypoimmunogenic cell product design. Nonetheless, even targeted integration at the AAVS1 “safe harbor” locus can result in transcriptomic alterations28. Our transcriptomic analysis of the AAVS1-US2 hiPSCs revealed differential expression of several genes compared with the parental line. As anticipated, we observed reduced expression of PPP1R12C ( gene encoding the protein phosphatase 1 regulatory subunit 12c harboring the AAVS1 target site within its first intron), and modest expression changes in several of the previously reported off-targets of the AAVS1 single guide RNA used in the hiPSC engineering29–31. The observed upregulation of DNA damage and stress response pathways in AAVS1-US2 cells warrants further investigation to ensure the genomic stability and long-term safety of these engineered lines. Importantly, none of the alterations appeared to impair the expression of pluripotency markers nor differentiation capacity along distinct cell lineages (i.e., X and Y). Nevertheless, to further minimize unintended genomic and transcriptomic changes resulting from on-target and/or off-target DNA cleavage, future applications may benefit from applying more precise genome-editing strategies such as those involving seamless DNA break-free gene targeting using CRISPR-Cas nickases32,33. While further studies are needed to fully assess the long-term safety and functional consequences of integrating a virus-derived immunoevasin into the genome, our findings serve as a proof-of-concept for the ectopic expression of immunoevasins as an alternative to conventional hypoimmunogenic cell generation approaches based on targeted genome-editing or otherwise. In conclusion, by leveraging the evolutionarily fine-tuned repertoire of viral immune evasion mechanisms, we propose a targeted approach for generating HLA-matched, hypoimmunogenic, hiPSC lines that may increase the survival, engraftment and, ultimately, the performance of allogeneic cell therapies. Materials and methods Ethics statement All methods were carried out in accordance with relevant guidelines and regulations in place at the time the donor material was taken (2011). The hiPSC line used in this study, LUMC0072iCTRL01, was generated by the Leiden hiPSC Center from anonymized surgical waste material obtained by LUMC Department of Dermatology. hiPSCs were generated under the Umbrella protocol P13080 (Creation of disease model systems to understand and correct genetic diseases through gene or other therapy using iPS cells derived from somatic cells: “the iPS study”) approved by the Medical Ethics Committee Leiden The Hague Delft (METC). The approved protocol covered the generation of hiPSCs from residual surgical material and included genetic modification of derived hiPSC lines. Peripheral blood buffy coats were purchased from Sanquin as residual donor material collected under written informed consent permitting the use of remaining material for research purposes. No additional human participants were recruited for this study. hiPSC culture and genetic modification by CRISPR-Cas9 The hiPSC line LUMC0072iCTRL01 (HLA-A2, HLA-A32, HLA-A19, HLA-B71, HLA-B72, HLA-B70, HLA-Bw6, HLA-Cw7, HLA-C*12) was generated via the RNA Simplicon reprogramming kit (Millipore, # SCR550); detailed information on this line is available at the Human Pluripotent Stem Cell Registry, https://hpscreg.eu/. The hiPSCs were cultured under sterile conditions on recombinant human vitronectin coated plates in Essential 8 (E8) medium (Thermo Fisher Scientific, #A1517001) and passaged every 3–4 days using 0.5 mM UltraPure EDTA (Thermo Fisher Scientific, #15575020). For cryopreservation, cell aggregates equivalent to half a well of a 6-well plate were frozen in 1 ml CryoStor CS10 (StemCell Technologies, #07952). Cryopreserved hiPSCs were stored in the vapor phase of liquid nitrogen at <-150 °C. Before experimental use, hiPSCs were confirmed to express the undifferentiated-state markers NANOG, OCT-4 and SSEA-4, to demonstrate spontaneous differentiation capacity towards endoderm, mesoderm and ectoderm lineage, and to be free of mycoplasma contaminations. Genetic modification of the hiPSCs was performed at passage 39 by using the P3 Primary Cell 4D-Nucleofector™ X Kit (Lonza, #V4XP-3032) program CA137. A total of 2.2 × 106 hiPSCs were mixed with 6–8 mg of a plasmid DNA mixture containing Cas9, gRNA, and donor DNA plasmids as described in the plasmid production section. RevitaCell supplement (Thermo Fisher Scientific, #A2644501) was added to the culture medium for 1 day after nucleofection. After recovery, 1.5 × 106 cells were passaged per 10-cm dish as single cells using TrypLE select (Thermo Fisher Scientific, #12563011), and antibiotic selection was started by the addition of 0.2 µg mL− 1 puromycin (InvivoGen, #ant-pr-1). The culture medium was changed every 2 days and selection continued for 6–7 days, after which plain E8 medium was added. Single colonies remaining after selection were allowed to grow to 1–3 mm and then transferred to separate culture wells for expansion. For the transfer of the colonies, the cells were incubated with Gentle Cell Dissociation Reagent (GCDR) (Stemcell Technologies, #100–0485), and single colonies were scraped off with a pipet tip and repeatedly pipetted to obtain clusters of approximately 5–20 cells. Multiple clones were expanded, and material was collected to assess the genomic modification. Expanded hiPSC clones were cryopreserved at passage 47 for banking using CryoStor CS10 and stored in the vapor phase of liquid nitrogen, as described above. Plasmid production For the genetic modification of hiPSCs, 3 types of DNA plasmids were used per sample: a Cas9 plasmid, a gRNA plasmid, and a donor DNA plasmid. The Cas9 plasmid, expresses the Streptococcus pyogenes Cas9 endonuclease from a CAG promoter (AV62_pCAG.Cas9.rBGpA)30. The gRNA plasmids for targeting the AAVS1 and B2M loci were created by ligating the complementary oligonucleotide pairs: 5’-ACCGGCCACGGAGCGAGACATCT-3’/5’-AAACAGATGTCTCGCTCCGTGGC-3’ and 5’-ACCGGGGCCACTAGGGACAGGAT-3’/5’-AAACATCCTGTCCCTAGTGGCCC-3’, respectively, into the BveI-digested gRNA spacer acceptor plasmid S7_pUC.U6.sgRNA.Bvel-stuffer as described in Chen et al.34. The former plasmid expresses an AAVS1-targeting gRNA (AAVS1-T2) initially assembled by Mali et al.35. The donor DNA plasmids were assembled through gene synthesis (BaseClear) and contained homology arms for either the B2M or AAVS1 locus flanking the sequences for targeted chromosomal gene insertion. For the modification of the B2M locus, the sequence within the homology arms contains the coding sequence for EGFP followed by a PGK1 gene promoter and a puromycin resistance gene to allow for antibiotic selection of genetically modified cells. Via homology-directed repair, the EGFP gene is expected to be inserted such that the B2M start codon is used to drive the synthesis of EGFP from the endogenous B2M promoter. For the modification of the AAVS1 locus, the sequence for insertion contains a CAG promoter followed by the coding sequence for EGFP or US2 and a PGK1 promoter followed by the puromycin resistance gene. The transformation of the plasmids was performed in chemoresistant E. Coli bacteria, and the plasmids were isolated via Maxiprep kits (Genomed) according to the manufacturer’s instructions. The correct plasmid sequences were validated via Sanger sequencing at the Leiden Genome Technology Center (LGTC). Genomic PCR DNA isolation was performed on the cell pellets with the DNeasy kit (Qiagen), and the DNA concentrations were measured with the NanoDrop system (Thermo Fisher Scientific). PCRs were performed via Phusion High-Fidelity DNA Polymerase (New England Biolabs) under the following conditions: 100 ng DNA, 1× GC rich buffer, 0.5 units Phusion High-Fidelity DNA Polymerase, 10 pmol of each primer, and 200 mM dNTPs in a 25 µL volume for 40 cycles in a C1000 Touch Thermal Cycler (Bio-Rad, Hercules, USA). The primers used for genotyping screening are described in Table S1. PCR products were loaded on a 1% agarose gel together with a GeneRuler 1 kb Plus DNA Ladder (Thermo Fisher Scientific) for size estimation. DNA bands were visualized by ethidium bromide detection in a Molecular Imager Gel Doc XR+ (Bio-Rad). RT-qPCR Total RNA was isolated with the Nucleospin RNA/protein kit (Macherey-Nagel) from pellets of approximately 0.5–1 × 106 hiPSCs cultured with or without stimulation with 500 IU mL− 1 IFN-γ (R&D Systems) for 48 h. The RNA concentrations were measured with the NanoDrop system. cDNA was produced via SuperScript III Reverse Transcriptase (Invitrogen) and subsequently added as a template for real-time polymerase chain reaction (RT-qPCR). RT-qPCR was performed using SYBR Green Supermix (Bio-Rad) in a CFX Connect Real-Time System (Bio-Rad) with the primers described in Table S1. The expression of each gene was normalized to the level of glyceraldehyde-3-phosphate dehydrogenase (GAPDH) expression and relative mRNA levels were determined based on the comparative Ct method (2-ΔΔCt) to the respective controls. RNA sequencing Total RNA samples isolated from parental hiPSCs (n = 2 independent passages), B2M-KO hiPSCs (n = 3 clones), and AAVS1-US2 hiPSCs (n = 3 clones) were isolated using Nucleospin miRNA Kit (Bioke, product code 740971) according to the manufacturer’s guidelines. RNA quality was characterized via the Experion RNA StdSens 1 K Analysis Kit (Bio-Rad, product number 7007103) on an Experion Automated Electrophoresis System (Bio-Rad) following the manufacturer’s protocol. Samples were sequenced on an Illumina NovaSeq6000 (20 M reads/sample). The quality of the raw data was evaluated with FastQC. STAR was used for mapping to the hg38 reference genome v109. Library orientation, library composition and coverage along transcripts were checked with Picard tools. The following analyses were conducted with R software. Lowly expressed RNAs were filtered out (threshold of mean ≥ 5 copies). The data were normalized with the edgeR (v4.3.1) bioconductor package with the Trimmed Mean of the M values (TMM) prior to differential expression analysis with quasi-likelihood F-tests from edgeR package. Pathway analysis was performed with ReactomePA (v1.16.2). Data was visualized with R-package ggplot2 (v4.0.0). Endothelial cell differentiation hiPSCs were differentiated into hiECs according to a previously described protocol36, with the addition of cyclosporine A (CsA) treatment reported by Tiemeier et al.37. Briefly, hiPSCs maintained on vitronectin-coated plates in StemFlex medium (ThermoFisher, #A3349401) were routinely passaged once per week using Versene solution (ThermoFisher, #15040066); differentiation was initiated one day after passaging (designated day 0). Mesoderm specification was induced by supplementing B(P)EL medium (Table S2) with CHIR99021 (1:500; Tocris Bioscience, #4423). On day 3, mesoderm-inducing factors were removed and replaced with vascular specification medium: B(P)EL medium containing human vascular endothelial growth factor (VEGF165, 1:1000; Miltenyi Biotec, #130-109-386) and the small-molecule inhibitor SB431542 (1:2000; Tocris Bioscience, #1614). The vascular specification medium was changed on days 5, 7 and 9 of differentiation, and 500 nM CsA (Sigma, #30024) was added in each medium change. On day 10, hiECs were isolated via magnetic-activated cell sorting with anti-CD31 antibody-coupled Dynabeads (Invitrogen, #11155D). The isolated hiECs were then replated on 0.1% gelatin-coated plates (Sigma-Aldrich, #G1890) in endothelial cell serum-free medium (EC-SFM; Invitrogen, #11111-044) supplemented with VEGF-165 and basic fibroblast growth factor bFGF (Miltenyi Biotech, #130-093-842), hereafter referred to as EC-CGM. The cells were cultured at 37 °C with 5% CO2. Flow cytometry hiPSCs and hiECs cultured with or without stimulation with 1000 IU mL− 1 IFN-γ (R&D Systems, #285-IF-100) for 24 h were dissociated using TrypLE select to obtain a single-cell suspension. The antibodies used for flow cytometry are shown in Supplementary Table S3. All the antibody incubations were performed for 30 min at 4 °C after which the samples were washed in PBS supplemented with 0.1% BSA. Flow cytometry was performed on cell suspensions in FACS buffer consisting of PBS containing 0.1% BSA and 0.5 mM EDTA on a FACSCanto (BD), and the results were analyzed with FlowJo (BD). T-cell coculture assay For the T-cell coculture assay, an alloreactive CD8+ T-cell clone specific for HLA-A2 (JS132) was used. The generation of the JS132 clone has been described previously38. HLA-A2-reactive T cells were retrieved from cryopreservation and cocultured with 20,000 IFN-γ-stimulated hiECs per well in a 96-well round bottom plate for 18 h. The medium used during coculture consisted of IMDM (Gibco, #12440053) supplemented with 10% FBS (Gibco, #A5256701), 1% PenStrep (Gibco, #15070063) and 200 UmL− 1 IL-2 (LUMC pharmacy). The supernatants from the T-cell coculture experiments were used to quantify the concentration of the chemokine MIP-1β with a CCL4 (MIP-1β) human uncoated ELISA kit (Invitrogen, #88-7034-88). The concentration of MIP-1β detected in the coculture was compared to the spontaneous release by T-cells in the absence of target cells. NK cell coculture assay NK cells were freshly isolated from peripheral blood mononuclear cells (PBMCs) obtained from five independently donated buffy coats (Sanquin). Isolation was performed with an NK Cell Isolation Kit (Miltenyi Biotec, #130-092-657) following the manufacturer’s instructions and a previously described protocol39. NK cells were preactivated overnight with 500 UmL− 1 IL-2 in RPMI-1640 medium (Gibco, #11875093) supplemented with 10% FBS and 1% PenStrep (hereafter referred to as NK cell medium). To assess NK cell reactivity, preactivated NK cells were cocultured with IFN-γ-stimulated hiECs. Cocultures were performed in a 96-well flat-bottom plate with 20,000 hiECs per well, at NK to target cell ratios of 0.5:1, 1:1 and 5:1. NK-sensitive K562 cells were used as a positive control and NK cells alone as negative control. The cells were incubated for 4 h in NK cell medium supplemented with a FITC-conjugated CD107a (LAMP-1) monoclonal antibody (1:20; eBioscience, #11-1079-42). Following coculture, the NK cells were harvested and stained with a CD56 antibody and a CD3 antibody. Flow cytometry analysis was performed as described in the flow cytometry section. NK cell degranulation was quantified as the percentage of CD107a+ cells within the CD56+ CD3− population. To account for donor-to-donor variability in NK cell responsiveness, degranulation data were normalized to the response observed in cocultures with parental hiPSC-ECs. Specifically, degranulation in experimental conditions was expressed as the percent change relative to the parental control within the same donor, calculated as: This normalization enabled direct comparison of NK cell degranulation responses across donors by reducing variability arising from intrinsic differences in NK cell activation capacity. Statistical analysis The results were statistically analyzed and visualized via Graphpad Prism (Version 10.2.3). Each sample was compared to the respective control of the assay using one-way ANOVA with Dunnett’s or Šidák’s correction for multiple testing as indicated. A ratio-paired t test was used for NK cell assays to account for donor variability. Experimental outcomes are shown as mean ± standard deviation (SD) and a p-value < 0.05 was considered statistically significant. For RNA-seq a significance threshold of FDR < 0.05 was used. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 2 Supplementary Material 3 Supplementary Material 4 Acknowledgements We thank Christian Freund (hiPSC core facility, LUMC, Leiden, the Netherlands) for providing the hiPSC line LUMC0072iCTRL01. We acknowledge the support of Ellen Lievers, Anneloes Verwey and anyone who has helped with the experiments but is not included as an author. This study was supported by the LUMC Light and Electron Microscopy Facility, the LUMC Flow Cytometry Core Facility, and Cure One, Type 1 Diabetes Research Center. This study was funded by LUF/Stichting Prof. Jaap de Graeff-Lingling Wiyadharma Fonds 2020-01 and the Novo Nordisk Foundation Center for Stem Cell Medicine, reNEW, supported by Novo Nordisk Foundation grant NNF21CC0073729. Author contributions Conceptualization of the study: JN, LG and AZ. Collection and/or assembly of data: JN, LG, JK, JP, RvN, AvdS, FvdH. Design of gene editing tools: MG. Analysis and interpretation of data: JN, LG, VO, CvdB, AZ. Funding acquisition TR. Drafting and revision of the manuscript: JN, LG, MG, AZ and TR. Data availability Analyzed transcriptomic data can be found in the supplemental tables S4-S6. The datasets generated during the current study are available in the European Genome-Phenome Archive (EGA) repository under the accession number EGAD50000002321. Materials availability Plasmids generated in this study can be made available on request, but we may require a completed materials transfer agreement. Declarations Competing interests Authors AZ, TR, LG, and JN are inventors on a patent application related to the work described in this manuscript. The application is currently pending. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Juan J. Novoa and Lonneke H. Gaykema contributed equally to this work. References Associated Data Supplementary Materials Supplementary Material 1 Supplementary Material 2 Supplementary Material 3 Supplementary Material 4 Data Availability Statement Analyzed transcriptomic data can be found in the supplemental tables S4-S6. The datasets generated during the current study are available in the European Genome-Phenome Archive (EGA) repository under the accession number EGAD50000002321. Plasmids generated in this study can be made available on request, but we may require a completed materials transfer agreement.