The PSC lines A3 and A6 have been previously described and characterized as pluripotent stem cells (31). They had been generated from oocytes of (C57BL/6J × DBA/2J) F1 mice (B6D2F1) by activation with strontium chloride and inhibition of the second polar body extrusion with cytochalasin B. The blastocysts were transferred onto mitomycin C-inactivated mouse embryonic fibroblasts (MEFs) and inner cell mass outgrowths were manually selected at day 5 and sequentially further cultured and dispersed to establish cell lines. Since PSCs are largely homozygous, the line A3 carries the H2^d haplotype on both chromosomes 17. Line A6, in contrast, is heterozygous for the MHC region due to a crossing over during meiosis I and carries both the H2^d and H2^b haplotypes (31). Both PSC lines were cultured on inactivated MEFs in Dulbecco’s modified Eagle’s medium (DMEM) (Gibco, Thermo Fisher Scientific, Waltham, MA, USA) with high glucose, no pyruvate, and 25 mmol/l N-2-hydroxyethylpiperazine-N′-2-ethanesulfonic acid. The stem cell medium was supplemented with 20% fetal calf serum (FCS), 1,000 U/ml leukemia inhibitory factor (Merck, Darmstadt, Germany), 2 mmol/l glutamine, 1× non-essential amino acids, 50 U/ml penicillin, 50 μg/ml streptomycin, 1 mmol/l Na-pyruvate, 30 μmol/l adenosine, 30 μmol/l guanosine, 30 μmol/l cytidine, 30 μmol/l uridine, 10 μmol/l thymidine, and 100 μmol/l 2-mercaptoethanol. The PSC colonies were dissociated every 48–72 h in 0.25% trypsin/EDTA and split at a ratio of 1:3 or 1:6. Before being used for experiments, the cell suspensions were preplated for 30 min on tissue culture plates to reduce the content of MEFs. The mouse cell lines YAC-1 (H2^a), RMA (H2^b), and X63-Ag8.653 (H2^d) were cultured as described previously (36, 39).

For in vitro differentiation of PSCs, the hanging drop method was applied as described previously (31). Briefly, hanging drops containing 500 cells in 20 μl Iscove’s medium (Gibco) supplemented with 20% FCS, 2 mmol/l glutamine, 1× non-essential amino acids, 50 U/ml penicillin, 50 μg/ml streptomycin, and 100 μmol/l 2-mercaptoethanol were cultured for 2 days. In this time embryoid bodies had formed and were subsequently transferred into Petri dishes for a 5-day suspension culture period before final transfer to cell culture dishes for further 5 days. At day 12, the in vitro-differentiated cells were harvested with 0.25% trypsin/EDTA and used for experiments.

Lymphokine-activated killer (LAK) or NK cells were obtained from spleens of 129Sv, Balb/c, C57BL/6, DAB/2J, and FVB.N mice. CTLs were derived from spleens of TCR-transgenic OT-I mice (40) that recognize the peptide SIINFEKL in a H2K^b-restricted manner. All mice were bred in the central animal facility of the Medical University Center Göttingen. They were narcotized in CO₂ and killed by cervical dislocation before spleens were taken aseptically. Lymphocytes were obtained from the spleens using a Tenbroeck homogenizer. Erythrocytes were removed from splenocytes by incubation for 5 min in lysis buffer (155 mM NH₄Cl, 10 mM KHCO₃, 0.1 mM EDTA, pH 7.4–7.8). To obtain LAK cells, the splenocytes were then cultured for 4 days in DMEM supplemented with 10% FCS, 50 μM 2-mercaptoethanol, and 1,000 U/ml IL-2 (Immunotools, Friesoythe, Germany). Untouched mouse NK cells were isolated by magnetic-activated cell sorting (MACS) using a mouse NK cell kit (Miltenyi Biotec, Bergisch Gladbach, Germany) for negative selection according to the manufacturer’s instruction. The purity of the isolated CD49b⁺CD3⁻ NK cells was always determined by flow cytometry and found to be >90% in all experiments. For activation, the purified NK cells were then cultured for 4 days similarly as the LAK cells. Splenocytes from naïve OT-I mice were stimulated in vitro in the presence of 1 nM SIINFEKL peptide (Ovalbumin 257–264; Bachem Biochemica, Heidelberg, Germany) as previously described (41) before being used as effector cells.

Target cells were labeled by incubating 1 × 10⁶ cells in a suspension of 200 μl DMEM, 100 μl FCS, and 50 μCi Na₂⁵¹CrO₄ (Hartmann Analytic, Braunschweig, Germany) for 1 h at 37°C and then washed three times with DMEM. These ⁵¹chromium-labeled target cells (5 × 10³ per well of round-bottomed microtiter plates) were cocultured with LAK or NK cells for 4 h at 37°C in triplicates at several effector to target (E:T) ratios in 200 μl of DMEM containing 10% FCS. To allow for a peptide-specific killing by CTLs derived from OT-I mice, the respective wells were supplemented with 0.5 μg/ml SIINFEKL peptide and to determine the calcium dependency of killing, 2 mM ethyleneglycol-bis(β-aminoethyl ester)-N,N,N′,N′-tetraacetic acid (EGTA) and 4 mM MgCl₂ were added to additional wells. The microtiter plates were centrifuged for 5 min at 40 × g before supernatants and sediments were separately taken to determine radioactivity in each well using a Wallac MicroBeta Trilux counter (PerkinElmer Life Sciences, Rodgau Jügesheim, Germany). Spontaneous release was determined by incubating target cells without the effector cells. The percentage of specific lysis of the target cells was calculated as described previously (42).

Flow cytometry was performed on a FACS Calibur flow cytometer (BD Biosciences, Heidelberg, Germany) using CellQuestPro data acquisition and analysis software. The antibodies (Abs) used for flow cytometry are described in Table S1 in Supplementary Material. Isotype controls were used for directly labeled monoclonal antibodies (mAbs). For staining, 2 × 10⁵ cells were incubated in 100 μl phosphate-buffered saline (PBS) with 1 μg of the respective primary mAb for 30 min at 4°C before washing with PBS. To detect the unlabeled mAbs, the cells were incubated subsequently in 100 μl PBS with 1 μl of a fluorescein isothiocyanate (FITC)-labeled goat anti-rat IgG Ab (for anti-CD112, anti-CD155, and anti-RAE-1) or FITC-labeled goat anti-mouse IgM Ab (for anti-SSEA-1) for 30 min. The biotinylated anti-Qa-1 mAb was detected with PE/Cy5-labeled streptavidin (Biolegend). In these experiments, cells stained with the secondary reagent only served as control. Recombinant mouse NKG2D-human IgG₁-Fc (139-NK; R&D Systems, Wiesbaden, Germany), mouse 2B4-human IgG₁-Fc (3514–2B; R&D Systems), and mouse NKp46-human IgG₁-Fc (2225-NK; R&D Systems) chimeric proteins were used to detect cell surface expression of NKG2D, 2B4, and NKp46 ligands, respectively. A polyclonal FITC-conjugated goat anti-human IgG served as a secondary reagent. A recombinant mouse DNAM-1-mouse IgG_2a-Fc chimeric protein (4436-DN; R&D Systems) was used to detect DNAM-1 ligands. A polyclonal FITC-conjugated goat anti-mouse IgG was used as a secondary reagent. The percentage of positive cells expressing the analyzed molecule and their mean fluorescence intensity (MFI) was determined by subtracting the values of the appropriate controls.

Total RNA was extracted from cell lines or tissues using the TRIZOL reagent (Invitrogen, Carlsbad, CA, USA) according to the manufacturer’s instructions. The RNA was then treated with RNase free DNase (RQ1, Promega, Madison, WI, USA) for 20 min at 37°C and purified by phenol–chloroform–isoamyl alcohol (25:24:1) extraction and precipitated with 1/10 volume 300 mM sodium acetate (pH 4.8) and 1 volume 2-propanol before washing with 70% ethanol and solving in RNase free water. The quantity of the extracted RNA was determined with a ND-1000 spectrophotometer (NanoDrop Technologies, Wilmington, DE, USA). For synthesis of cDNA random oligo primers (Promega, Madison, WI, USA) were used. The reverse transcription of RNA was performed for 60 min at 37°C with M-MLV RT polymerase (Promega, Madison, WI, USA) in a total volume of 25 μl. Gene expression levels were analyzed by quantitative real-time polymerase chain reaction assays using the forward and reverse primers given in Table S2 in Supplementary Material. The mRNA expression of the housekeeping gene hypoxanthine guanine phosphoribosyl transferase 1 (Hprt1) was always monitored as internal control. Amplification reactions were carried out in 96-well plates in 25 μl reaction volumes with the Power SYBR green PCR master mix (Applied Biosystems, Foster City, CA, USA). The PCR plates were preheated for 2 min at 50°C and for 10 min at 95°C followed by 40 cycles of denaturation (15 s at 95°C), and amplification (1 min at 60°C). All reactions were performed in technical triplicates using an ABI 7500 Real Time PCR System. For the data analysis, the ABI 7500 SDS software (Applied Biosystems) was used. The variations in cDNA concentration in different samples were normalized to the housekeeping gene Hprt1. The relative level of transcripts was then indicated as the delta cycle threshold (Δct) value (ct for the gene of interest minus ct for Hprt1).

Results are shown as mean with SD or as mean with SEM for cytotoxicity assays. The data were evaluated with the SPSS software (IBM, Armonk, NY, USA). Excel or the GraphPad Prism (GraphPad Software, La Jolla, CA, USA) programs were used to generate graphs. Analyses of variance (ANOVA) adjusted for repeated measures at different E:T ratios were used to evaluate the cytotoxicity data for differences with respect to target and/or effector cell types and Bonferroni tests were selected as post hoc tests. Non-parametric Kruskal–Wallis (H) tests were used to compare flow cytometric data sets that included non-normally distributed target variables or variables with few data points. Dunn’s or Wilcoxon (U) tests were used as post hoc test to compare two sets of data. A P-value of ≤0.05 in two-sided tests was considered significant. Significance levels are shown in the following way, if not the exact P-value is given: *P ≤ 0.05; **P ≤ 0.01.

We have previously shown that several murine pluripotent stem cell types, including ESCs, iPSCs, and maGSCs, are targets for LAK as well as purified IL-2-activated NK cells (36). Therefore, we now determined the susceptibility of PSCs, as a further pluripotent stem cell type, to LAK and NK cells. Two PSC lines A3 (H2^d) and A6 (H2^b/d) were investigated and compared to the typical NK target cell line YAC-1 (H2^a) in ⁵¹Cr-release assays as shown exemplarily in Figure S1A in Supplementary Material with LAK effector cells from 129Sv mice. Independent experiments were done at different days with effector cells obtained from different mice. To allow for a better comparison with our previously published results on other pluripotent stem cell targets (36), the specific lysis of YAC-1 cells at the highest E:T ratio (200:1) was set to 100% in an additional analysis and the relative lysis at lower E:T ratios and of other target cells was calculated accordingly (Figure S1B in Supplementary Material). In general, all targets (PSC A3, PSC A6 and YAC-1 cells) were killed by LAK cells but their specific lysis varied (P < 0.0001, ANOVA). Overall, PSC A3 cells were killed better than PSC A6 cells (P < 0.0001) and YAC-1 cells (P < 0.0001, Bonferroni post hoc test). The specific lysis of PSC A6 cells was in similar range as YAC-1 control cells (Figure 1A). The relative lysis of the PSC targets in relation to YAC-1 cells is shown in Figure 1B to allow for comparison with the killing of other pluripotent stem cell types (36).

Lymphokine-activated killer cells from four mouse strains (129Sv, C57BL/6, DBA/2J, and FVB.N) were used as effector cells in these experiments (Figures 1C–F). These strains have been selected because LAK cells from 129Sv, C57BL/6, and FVB.N mice had been used previously to assess the susceptibility of ESCs, iPSCs, and maGSCs to LAK cell-mediated killing (36) and DBA/2J mice were included in addition because they are the second parental strain of the two PSC lines. The inbred strains represent together some of the genetic variability of mice. We compared the susceptibility of the target cell lines to LAK cells derived from the four strains and indeed found overall significant differences (P < 0.0001, ANOVA). The LAK cells differed only slightly in their ability to kill YAC-1 control cells (P = 0.012, ANOVA) with DBA/2J being less efficient than C57BL/6 and FVB.N LAK cells (P < 0.05, Bonferroni post hoc tests) (Figure S2A in Supplementary Material). Differences of the ability of LAK cells of the four mouse strains to kill the PSC lines were more pronounced (PSC A3: P < 0.0001, PSC A6: P < 0.0001, ANOVA). PSC A3 cells were more susceptible to LAK cells from 129Sv mice than C57BL/6 (P = 0.004) and FVB.N mice (P < 0.0001, Bonferroni post hoc test). PSC A6 cells were more resistant to LAK cells from FVB.N mice than from the other strains (P < 0.05, Bonferroni post hoc tests). Moreover, the PSC A3 cells were lysed more efficiently than PSC A6 cells by LAK cell from 129Sv (P < 0.0001, ANOVA) and FVB.N mice (P < 0.0001), whereas the specific lysis of the two PSC lines did not differ with C57BL/6 (P = 0.1332) and DBA/2J (P = 0.7324) effector cells. The relative lysis of the targets by effector cells of the four strains is shown in Figures S2B–E in Supplementary Material to allow for comparison with previously published data on other pluripotent stem cell types (36). The 129Sv and C57BL/6 mice share the MHC haplotype H2^b with A6 cells, DBA/2J mice share the H2^d haplotype with A3 and A6 cells, and FVB.N mice share the H2^q haplotype with none of the targets. However, the patterns of susceptibility of the PSCs to LAK cells of the different strains are not explainable by a partial inhibition of NK cell activity by self-MHC class I molecules.

Next, we purified NK cells by MACS and used them as effectors against the PSC and YAC-1 target cells to confirm that indeed NK cells among LAK cells have been the cytotoxic effector cells against the PSCs. In these experiments, NK cells from the two parental strains of the PSC lines, C57BL/6 and DBA/2J, were used. In addition, NK cells from Balb/c mice were included because PSC A3 cells had formed teratomas not only in DBA/2J but also in Balb/c mice, whereas they were rejected in C57BL/6 mice in our previous study (31). Purified resting NK cells failed to kill the PSC lines efficiently at low E:T ratios (Figure 2A). Even at this low level of killing, the specific lysis of PSC A3, PSC A6, and YAC-1 target cells varied (P < 0.0001, ANOVA) and PSC A6 cells were more resistant than PSC A3 and YAC-1 cells (P < 0.01, Bonferroni post hoc tests). Moreover, also the cytotoxic activity of NK cells from the three strains was different (P < 0.0001, ANOVA; with C57BL/6 NK cells > DBA/2J NK cells > Balb/c NK cells, P < 0.05, Bonferroni post hoc tests). In a stratified analysis with respect to the NK cell donor strains, the specific lysis of the two PSC lines A3 and A6 by resting NK cells from DBA/2J mice did not differ (Figure 2C). However, PSC A3 cells were more susceptible than PSC A6 cells to killing by Balb/c (P = 0.0135) (Figure 2E) and C57BL/6 NK cells (P = 0.020, ANOVA) (Figure 2G).

After activation with IL-2 for 4 days, the NK cells, similarly as LAK cells, killed all target cell lines rather efficiently. The killing of all target cells increased significantly (YAC-1 P < 0.001, PSC A3 P < 0.0001, PSC A6 P < 0.0001, ANOVA) (Figures 2A,B). Increased killing by IL-2-activated NK cells was also observed for all target cell lines in separate analyses for the three mouse strains (P < 0.001, ANOVA). The specific lysis of PSC A3, PSC A6, and YAC-1 cells by IL-2-activated NK cells was different (P < 0.0001, ANOVA). Overall, PSC A6 cells were more resistant than PSC A3 cells (P < 0.0001) and YAC-1 cells (P < 0.0001, Bonferroni post hoc test), whereas the specific lysis of PSC A3 cells was in similar range as YAC-1 control cells and not significantly different (Figure 2B). IL-2-activated NK cells from DBA/2J (Figure 2D), Balb/c (Figure 2F), and C57BL/6 mice (Figure 2H) differed in their efficacy to kill the three target cell lines (P < 0.0001, ANOVA). The NK cells did not significantly vary in their ability to kill YAC-1 cells (P = 0.144, ANOVA) (Supplementary Figure 3). However, they differed in killing of PSC A3 (P = 0.0006) and PSC A6 cells (P = 0.004, ANOVA) (Figures 2D,F,H). DBA/2J NK cells were more efficient than Balb/c NK cells (PSC A3: P = 0.0004; PSC A6: P = 0.0028, Bonferroni post hoc tests). The PSC A3 cells were lysed more efficiently than PSC A6 cells by NK cells from DBA/2J (P = 0.0011) and C57BL/6 mice (P = 0.0049, ANOVA). In the experiments with Balb/c NK cells, a trend toward this difference was observed (P = 0.0695). Thus, NK cells derived from mice carrying the H2^d haplotype (DBA/2J and Balb/c) were not inhibited by the PSCs. Although, the C57BL/6 NK cells were killing the PSC A3 cells (H2^d) better than the PSC A6 cells (H2^b/d) this does not suggest a specific inhibition by MHC class I molecules of the H2^b haplotype as this pattern was observed also with NK cells of the other strains.

The cytotoxic activity of NK cells is controlled by inhibitory and activating NK receptors. Ligands for inhibitory NK receptors are mainly MHC class I molecules. Murine pluripotent stem cells, including ESCs, iPSCs, and maGSCs, have been shown to lack MHC class I molecules at least at a level detectable by flow cytometry (39). Both PSC lines were also largely lacking H2K and H2D molecules, as exemplified in Figure S4 in Supplementary Material, although especially H2K^d molecules were found at low levels on average on 10.6% of the PSC A3 and on 7.0% of the PSC A6 cells (Figure 3A). RMA (H2^b) or Ag8.653 (H2^d) served as positive control cells in these experiments. Recombinant Fc fusion proteins of the activating NK receptors NKG2D, DNAM-1, 2B4, and NKp46 were used to detect the respective ligands on the PSCs and YAC-1 cells (Figure 3B). Both PSC lines expressed ligands for the activating NK receptors NKG2D and DNAM-1 as demonstrated by binding of recombinant receptor proteins. Ligands for the 2B4 and NKp46 receptors were not detected. The NKG2D ligands present on the PSC lines belonged to the RAE-1 family (Figure 3C). MULT-1 and H60 proteins, by contrast, were hardly expressed. Both DNAM-1 ligands, CD112 and CD155, were found on the PSC lines (Figure 3D). However, the PSC line A6 largely lacked the functional HLA-E homolog Qa-1^b (Figure 3E). For most molecules tested by flow cytometry (with exception of DNAM-1 and NKp46 ligands as well as Qa-1 molecules), significant expression differences between the analyzed cells were found (ANOVA, see Figures 3A–E). However, the post hoc tests indicated differences only between the control cells and the PSC lines but none between the two PSC lines. In addition to the percentage of positive cells for the various molecules, their MFI has been determined and showed the same patterns (Figure S5 in Supplementary Material). In summary, the PSC lines express ligands for activating NK receptors and largely lack MHC class I molecules that serve as ligands for inhibitory receptors. Consequently, they are targets for LAK and NK cells.

At the mRNA level, H2K and B2m as well as transcripts of the chaperones involved in antigen presentation (Calr, Canx, Erp57) were found in the PSC lines although they were mostly less abundant than in RMA control cells or DBA/2J splenocytes (Figure S6A in Supplementary Material). Tap1 and Tapbp mRNAs were found at low level in the PSC lines. H2D, Tap2, Qa1b, and the genes Psmb9 and Psmb8 encoding the alternative subunits of the immunoproteasome (LMP2 and LMP7), by contrast, were hardly expressed in the PSCs. Transcripts of Nectin2 (encoding CD112), Pvr (encoding CD155), as well as Raet1 (encoding RAE-1) and at lower levels Ulpb1 (encoding MULT-1) were detected in PSCs in contrast to H60 mRNA. Interestingly, transcripts of the NK receptor genes Cd226 (encoding DNAM-1) and Klrk1 (encoding NKG2D) were detectable at very low level. The genes encoding the co-stimulatory molecules CD80 and CD86 were basically not expressed in the PSC lines.

Differentiated cells are expected to be able to inhibit at least autologous NK cells to ensure self-tolerance. Therefore, we differentiated the PSC lines in vitro for 12 days in an undirected manner and determined the susceptibility of the differentiated cells to LAK cells in ⁵¹Cr-release assays. The differentiation significantly reduced the expression of the stem cell marker SSEA-1 indicating that the in vitro differentiation process was successful (Figures S7A–C in Supplementary Material).

The differentiated cells and YAC-1 cells were used as targets for LAK cells derived from the same four strains that were used in the experiments with undifferentiated PSCs. The specific lysis of the target cells (differentiated A3, differentiated A6, and YAC-1 cells) by LAK cell from all strains varied (P < 0.0001, ANOVA). Both differentiated cell types A3 (P < 0.0001) and A6 (P < 0.0001) were more resistant than YAC-1 cells and A6 cells were more resistant than A3 cells (P < 0.0001) as indicated by Bonferroni post hoc testing (Figure 4A). The relative lysis of the targets is illustrated in Figure 4B to allow for comparison with previously published data of other pluripotent stem cell types (36). In addition, LAK cells derived from the four strains (129Sv, C57BL/6, DBA/2J, and FVB.N) varied in their efficacy to kill the targets (P < 0.0001, ANOVA) (Figures 4C–F). LAK cells from C57BL/6 mice were more potent than the others (P < 0.001, Bonferroni post hoc tests). When comparing the specific lysis of the two differentiated cell types only, C57BL/6 and DBA/2J LAK cells were more efficient killers than LAK cells derived from 129Sv and FVB.N mice (P < 0.001, Bonferroni post hoc tests). Notably, the specific lysis of the two differentiated target cell lines A3 and A6 was significantly different when effector cells were used derived from 129Sv (P < 0.0001, ANOVA) and C57BL/6 (P = 0.0017) but not DBA/2J (P = 0.1182) or FVB.N mice (P = 0.0989). This could indicate that differentiated A6 cells expressing H2^b MHC class I molecules better inhibit NK cells of the haplotype H2^b (129Sv and C57BL/6) than the A3 cells carrying the H2^d haplotype only. However, A6 cells did not efficiently inhibit NK cells from DBA/2J (H2^d) mice. The relative lysis of the targets by effector cells of the four strains is shown in Figures S8A–D in Supplementary Material.

As expected, differentiated cells were overall more resistant to LAK cells from all strains than undifferentiated PSCs (P < 0.0001, ANOVA) (Figures S9A in Supplementary Material). The PSC A3 cells were more susceptible to killing than PSC A6 cells (P < 0.0001, Bonferroni post hoc tests), differentiated A3 cells (P < 0.0001), and differentiated A6 cells (P < 0.0001). Differentiated A6 cells were significantly more resistant than all other targets (P < 0.0001, Bonferroni post hoc tests). However, the specific lysis of PSC A6 cells and differentiated A3 cells did not differ significantly. This pattern was also observed (P < 0.0001, Bonferroni post hoc tests) when the specific lysis elicited by 129Sv or FVB.N LAK cells was analyzed separately (Figures S9B,E in Supplementary Material). When the killing elicited by LAK cells derived from C57BL/6 or DBA/2J was separately analyzed, only differentiated A6 cells were more resistant than the other cell types (Figures S9C,D in Supplementary Material). Overall, both differentiated A3 (P < 0.0001) and differentiated A6 cells (P < 0.0001, ANOVA) were more resistant to LAK cells from all strains than the respective undifferentiated PSCs. Both differentiated cell types A3 (P < 0.0001) and A6 (P < 0.0001, ANOVA) were more resistant than the respective undifferentiated PSCs to 129Sv LAK cells (Figure S9B in Supplementary Material). The same result was observed for FVB.N effectors (A3: P < 0.0001; A6: P = 0.0308) (Figure S9E in Supplementary Material). When C57BL/6 or DBA/2J effector cells were used, only the differentiation of A6 cells decreased the susceptibility to LAK cells (C57BL/6: P = 0.0002; DBA/2J: P = 0.0216) (Figures S9C,D in Supplementary Material). The killing of PSC A3 and differentiated A3 cells, by contrast, was not significantly different (C57BL/6: P = 0.6651; DBA/2J: P = 0.1216).

At the mRNA level, the differentiation of the PSCs partly increased the expression of MHC class I transcripts, such as H2D and H2K, in A6 cells (Figure S6A in Supplementary Material). However, expression of most genes encoding proteins involved in antigen expression was not markedly increased or even decreased. Notably, transcripts of most NKG2D and DNAM-1 ligands were less abundant in both differentiated cell types. At the protein level, the differentiated cells also largely lacked MHC class I molecules as determined by flow cytometry and no significant differences to the PSCs were detected (Figure 5A). However, NKG2D ligands and specifically RAE-1 as well as the DNAM-1 ligands CD112 and CD155 were differently expressed and downregulated on differentiated cells although not always all molecules were downregulated on both cell types (Figure 5B).

In previous experiments, several murine pluripotent stem cell types, including ESCs, iPSCs, and maGSCs, were targets for peptide-specific CTLs derived from TCR-transgenic OT-I mice after pulsing with the appropriate peptide SIINFEKL despite being negative for MHC class I molecules in flow cytometry (39). Of the two PSC cell lines analyzed here, only one, i.e., A6, carried the required MHC molecule H2K^b. Thus, the line A3 carrying the H2^d haplotype served as negative control, whereas RMA cells (H2^b) were used as positive control. The specific lysis of the three target cell lines (RMA, PSC A3, and PSC A6) varied (P < 0.0001, ANOVA). However, compared to the positive control target cell line RMA, both PSC lines were largely resistant to the CTLs. The PSC A3 target cells showed a minimal background killing of 5% at the highest E:T ratio but the specific lysis of PSC A6 targets was only slightly higher with 13% on average at the highest E:T ratio (Figure 6). The low killing of PSC A6 cells and differentiated A6 cells was peptide and Ca²⁺-dependent (Figure S10 in Supplementary Material), indicating a lysis by CTLs using the granule exocytosis pathway of killing that involves perforin and granzymes. Post hoc tests indicated that the specific lysis of both PSC lines did differ significantly from RMA cells (P < 0.0001). Moreover, PSC A6 cells were also more susceptible to the peptide-specific CTLs than PSC A3 cells (P < 0.05). The in vitro differentiation of PSC A6 cells over 12 days was not sufficient to significantly increase the susceptibility of the cells to CTLs (P = 0.066) (Figure 6). At the mRNA level, we did not observe a strong expression of Ido1, Arg1, or Tgfb1, which encode proteins that have a role in the suppression of CTL activity, neither in the PSCs nor in the differentiated cells. Similarly, Serpinb9 that encodes a granzyme B inhibitor was not strongly expressed in any of the CTL-resistant targets (Figure S6B in Supplementary Material). However, the CTL resistance is in agreement with the low expression of H2K^b that was found on the differentiated cells (see Figure 6). Thus, the differentiation protocol applied here does not result in a cell population with a normal susceptibility to CTL-mediated killing.