Author contributions: M.P.: conception and design, collection and/or assembly of data, data analysis and interpretation, and manuscript writing; O.Y.: conception and design, collection and/or assembly of data, and manuscript writing; D.R.: collection and/or assembly of data, data analysis and interpretation, and manuscript writing; N.B.: conception and design, financial support, data analysis and interpretation, manuscript writing, and final approval of manuscript.
Disclosure of potential conflicts of interest is found at the end of this article.
First published online in STEM CELLSEXPRESS September 17, 2012.
The immunogenicity of human pluripotent stem cells plays a major role in their potential use in the clinic. We show that, during their reprogramming, human-induced pluripotent stem (iPS) cells downregulate expression of human leukocyte antigen (HLA)-A/B/C and β2 microglobulin (β2M), the two components of major histocompatibility complex-I (MHC-I). MHC-I expression in iPS cells can be restored by differentiation or treatment with interferon-gamma (IFNγ). To analyze the molecular mechanisms that regulate the expression of the MHC-I molecules in human iPS cells, we searched for correlation between the expression of HLA-A/B/C and β2M and the expression of transcription factors that bind to the promoter of these genes. Our results show a significant positive correlation between MHC-I expression and expression of the nuclear factors, nuclear factor kappa B 1 (NFκB1) and RelA, at the levels of RNA, protein and was confirmed by chromatin binding. Concordantly, we detected robust levels of NFκB1 and RelA proteins in the nucleus of somatic cells but not in the iPS cell derived from them. Overexpression of NFκB1 and RelA in undifferentiated pluripotent stem cells led to induction in expression of MHC-I, whereas silencing NFκB1 and RelA by small hairpin RNA decreased the expression of β2M after IFNγ treatment. Our data point to the critical role of NFκB proteins in regulating the MHC-I expression in human pluripotent stem cells. STEM CELLS 2012;30:2700–2708
Human embryonic stem cells (hESCs) are generated from the inner cell mass (ICM) of blastocysts  and can be maintained in an undifferentiated state in culture indefinitely. When differentiated, they can generate cells of the three embryonic germ layers [1, 2] and have the potential to generate any cell type present in the body. Thus, ESC-based therapy has the potential to change the face of transplantation medicine.
One of the major impediments in the use of hESCs for transplantation purposes is the allogeneic origin of these cells. The major immune response in rejection of transplanted allogeneic cells and organs, involves the classic major histocompatibility complex-I (MHC-I) that is expressed on every nucleated cell in the body. Its function is to display foreign antigens to T cells, which subsequently generate a specific immune response. MHC-I is composed of two polypeptide chains: (a) the alpha polypeptide, which contains three proteins encoded from the human leukocyte antigen (HLA) genes—A, B, and C and (b) the beta two microglobulin (β2M) polypeptide. MHC-II molecules are antigens found mainly on the cells of the immune system which present antigens. Like MHC-I, MHC-II molecules are also heterodimers but in this case consist of two homologous peptides, two alpha and two beta HLA chains. All genes of the HLA family are found on human chromosome 6p21, while β2M gene is found on chromosome 15q21.
To determine whether hESCs might be rejected after transplantation, we have previously examined cell surface expression of the MHC proteins in these cells . Our results showed very low expression levels of MHC-I proteins that increased upon in vitro or in vivo differentiation. An induction of MHC-I proteins was also observed when the cells were treated with interferon-gamma (IFN-γ). Ligands for natural killer cell receptors were either absent or expressed in very low levels in hESCs and in their differentiated derivatives. In accordance, natural killer cytotoxic assays demonstrated only limited lysis of both undifferentiated and differentiated cells . We further examined the in vivo leukocyte response toward hESCs and their derivatives showing that although the cells elicit an immune response, they are less susceptible to immune rejection than adult cells .
Recently, human somatic cells were reprogrammed to a pluripotent state using defined factors [5, 6]. The establishment of these induced pluripotent stem (iPS) cells has allowed for an alternate source of stem cells with similar properties of hESCs. Since iPS cells can be generated from adult cells, patient-specific stem cells can be produced. The ability to manufacture patient-specific pluripotent cells should permit the exclusion of immunosuppression and the need for tissue matching in cells for transplantation, although a recent study with mouse iPS cells suggests may retain some immunogenicity . In spite of major advances in the field, the ability to generate the required cell type for each patient is still far from reality due to restraints in time needed for the reprogramming, its cost, and the difficulty in generating normal iPS cells from patients carrying a genetic disease . Thus, the use of allogeneic iPS cell lines is expected to be the first choice in future clinical trials and thus their expression of the MHC antigens should be taken into account.
The generation of iPS cells and the reprogramming event have allowed us to study various biological and genetic events that occur to somatic cells when they become iPS cells. Here, we demonstrate that as part of the genetic reprogramming, somatic cells dramatically reduce their levels of MHC-I molecules. The expression of MHC-I in the iPS cells can be induced by IFNγ, or by induction of differentiation. We have also analyzed the mechanism that is involved in the loss of MHC-I expression upon reprogramming and identify nuclear factor kappa B (NFκB) as a major player in this regulation.
MATERIALS AND METHODS
Foreskin fibroblasts (a kind gift from Dr. B. Reubinoff) and hTERT-BJ1 (Clontech) were cultured in Dulbecco's modified Eagle's medium (Sigma) supplemented with 10% fetal calf serum (Biological Industries, Israel). iPS cells were previously generated using the four factors—OCT4, SOX2, c-MYC, and KLF4 in retroviral vectors . All pluripotent stem cell lines were maintained on mitomycin-treated mouse embryonic fibroblast feeder layer as previously described [2, 10]. Embryoid bodies (EBs) were generated as previously described . To induce MHC-I and β2M expression, 50 ng/ml of IFNγ was added to fibroblasts, hESC, or iPS cell lines .
DNA Microarray Analysis
CEL files were downloaded from NCBI Gene Expression Omnibus website, http://www.ncbi.nlm.nih.gov/ sites/entrez?db = geo, from publications that contained DNA microarray analysis of iPS cells using the Affymetrix GeneChip Human Genome U133 Plus 2.0 array (Affymetrix) [11–15]. All raw data files were analyzed together using MAS5 probeset condensation algorithm (Affymetrix expression console software). Resulting data were normalized by dividing the results of each array to the trimmed mean (0.05) and multiplied by a factor of 100.
RNA Isolation and Reverse Transcription
RNA (DNase treated) was extracted using Total RNA Mini Kit (Geneaid). One microgram of RNA was reverse transcribed using ImProm-II reverse transcriptase (Promega). Quantitative real-time polymerase chain reaction (PCR) was performed with 50 ng of cDNA, SYBR Green qPCR Supermix (Applied Biosystems) and analyzed with 7300 real-time PCR system (Applied Biosystems). Primer sequences are as follows: NFκB1—forward “ACTGTGAGGATGGGATCTGCA” and reverse “CCTTCTGCTTGCAATAGGC”; RelA forward “CCACAGTTTCCAGAAC” and reverse “CACTGTCACCTGGAAGCAGA.” β2M—forward “GTGCTCGCGCTACTCTCTCT” and reverse “TTCCATTCTCTGCTGGATGA.” All primers were tested for specificity and accuracy.
Cells were prepared at the concentration of 106 in 100 ml of phosphate buffered saline (PBS). Either HLA-A/B/C conjugated to Phycoerythrin (PE) or β2M conjugated to PE was added for 30 minutes at 4°C, after two washes in PBS, cells were acquired and analyzed on a FACS Calibur (BD Bioscience).
Sixty micrograms of protein was loaded into each well. An 11% acrylamide gel was run at 150 V for 1.5 hours. Transfer was performed using nitrocellulose for 2 hours at 280 A. NFκB1 (Santa Cruz) and RelA (Abcam) antibody were incubated overnight at 4°C, and anti mouse horse raddish perozidase (HRP) was used to detect protein together with the chemiluminescence ECL kit (Beit Haemek). To quantify the band intensity of NFκB1 proteins, Adobe Photoshop software was used. The scanned blots were inverted and the intensity of the bands was measured with the background intensity subtracted. Once all the intensity values were calculated, the intensity values for NFκB1 bands were divided by the intensity of the corresponding endogenous control band. The final step was to divide the value of the IFNγ-treated samples by the value of the control sample giving a value of fold increase in expression.
Cells were fixed with PBS containing 4% paraformaldehyde for 5 minutes and permeabilized with 0.5% Triton X-100 for 10 minutes at room temperature. After washing with PBS, the cells were blocked for 1 hour with PBS containing 2% bovine serum albumin (BSA, Sigma) and 0.1% Triton X-100. Staining with primary antibodies was performed for 1 hour at room temperature with antibodies against OCT3/4 (1:100; Santa Cruz) or β2M (1:100; BD Bioscience) or overnight with antibodies against NFκB1 (1:50; Santa Cruz) and RelA (1:200; Abcam). Secondary antibodies used were rabbit anti-mouse-cyanine 3 or cyanine 2 (1:50; Jackson Immunoresearch laboratories), and nuclei were stained with Hoechst 33342 (Invitrogen).
Downregulation of OCT4 Using Small Interfering RNA
Cells were transfected with ON-target siOCT4 small interfering RNA (siRNA) (Thermo Scientific) using Oligofectamine (Invitrogen), 48 hours post-transfection RNA was extracted. Quantitative real-time PCR was then performed using primers to detect NFκB1, RelA, and OCT4. As a control, scrambled siRNA was used.
Downregulation of NFκB1 and RelA Using Small Hairpin RNA
hESCs (H9) were transfected with different small hairpin (shRNA) molecules to silence NFκB1 (6518 and 6519—MISSION shRNA, Sigma) or RelA (4685 and 4684 MISSION shRNA, Sigma) using LT1 reagent (Mirus). hESC clones, which were selected using puromycin to contain sh-NFκB1, were incubated with and without IFNγ for 48 hours at which time RNA was extracted and quantitated real-time-PCR was then performed using primers to detected β2M expression.
Chromatin immunoprecipitation assays were performed according to the standard Upstate protocol  using Bioruptor bath sonicator (Diagenode) to shear DNA to an average size of 500–1,000 bp. One hundred micrograms chromatin was precleared with 40 μl salmon sperm DNA–protein A–agarose beads (Upstate Biotechnology) for 60 minutes, followed by an overnight incubation with 5–10 μg of either RelA (p65; Abcam) antibody or anti HA hybridoma supernatant (kind gift from Dr. Amir Eden) as control. Eluted DNA was isolated using QIAGEN PCR purification kit. For PCR analysis, the bound fraction was corrected relative to unrelated control region on low density lipoprotein (LDL) Receptor. Forward primer for β2M is: GCCGATGTACAGACAGCAAA and reverse primer is: GACCGTCACCTGTCTCCAAG. Forward primer for HLA-B is: GCCAAGACTCAGGGAGACAT and reverse primer is: ACCTGGGACTTCGTCCTGAT. Forward primer for LDL receptor is: TTTTGGCTGTACTTTTGT GAAGA, and reverse primer is: GCCTTTTCCTTTCTAGG AATTGT. The data were analyzed using the using fold change method.
MHC-I Expression Is Suppressed During Reprogramming and can be Induced with Differentiation or IFNγ Treatment
MHC-I expression has been shown to be low in hESCs and the cells were suggested to be less immunogenic than adult cells [3, 17]. However, the level of expression of MHC-1 in iPS cell lines and their parental somatic cells was unknown. We thus examined the expression patterns of HLA-A/B/C and β2M in iPS cell lines and their parental fibroblasts. All eight fibroblast cell lines expressed high levels of both HLA and β2M, while both hESCs and human iPS cell lines, generated from these fibroblasts, expressed significantly lower levels of these molecules (Fig. 1A and supporting information Fig. S1; published lines: HLA-A p < 10−5, HLA-B p < 10−4, HLA-C p < . 10−5, β2M p < . 10−9, our lines: HLA-A p < .05, HLA-B p < .05, HLA-C p < .01, β2M p < .05). Thus, the expression of MHC-I molecules (HLA-A/B/C and β2M) is suppressed during the reprogramming event to mimic levels seen in hESCs (Fig. 1A). Furthermore, this was confirmed in our cell lines where the levels of HLA-B and β2M in undifferentiated hESCs and iPS cells were significantly lower than in fibroblast cells (Fig. 1B, p < .001, p < .01, respectively). MHC-II expression was low in both the somatic and pluripotent cells as expected (supporting information Fig. S2).
To determine whether MHC-I expression increases with differentiation, pluripotent cell lines were differentiated, either in vitro into EBs or in vivo into teratomas. Expression levels of HLA-A/B/C or β2M were examined either at the RNA or protein levels (Fig. 1B). After 7 days of differentiation, the expression levels of the MHC-I in EBs did not increase significantly; however, by 20 days of differentiation, an increase in expression at the levels of RNA (p ≤ .05 for HLA-B, p < .01 for β2M) and protein was observed (Fig. 1B). Similarly, the expression of HLA-B and β2M was significantly higher in teratomas (Fig. 1B, p < .05, p < .001, respectively).
It has previously been reported that IFNγ can induce the expression of HLA-A/B/C and β2M in hESC lines . Two different iPS cell lines were incubated for 48 hours with 50 ng/ml IFNγ, and the levels of HLA-A/B/C and β2M were analyzed. Induction of expression of both proteins was similar to the induction observed in hESCs. Thus, the MHC-I complex react in iPS cells to the same stimuli observed in ESCs, even though the MHC-I complex has been deactivated by the reprogramming event (Fig. 1C).
NFκB1 and RelA Are Involved in Controlling the Expression of HLA-A/B/C and β2M
In order to find out the molecules that regulate the expression of MHC-I in pluripotent cells, we examined the expression of four families of binding factors known to bind specific sites in the promoters of β2M and HLA-A/B/C: (a) the enhancer element binds factors 1–4 from the upstream stimulatory factor (USF) family [18–20]; (b) the interferon-stimulated response element that binds the interferon regulatory factor family of proteins 1, 2, 3, 4, and 8 [18–20]; (c) the kappa-binding element that binds proteins from the NFκB family [18–20]; (d) S-X-Y module that binds proteins from the regulatory factor X, cAMP response element-binding factor, and the nuclear transcription factor families [18–20] (Fig. 2A). We examined the expression levels of all members of the four families of nuclear factors mentioned above in 36 published cell lines including: five hESC lines, 23 iPS cell lines, and eight of their parental fibroblast cell lines. Analysis of the expression levels of each binding proteins within each family was assessed to exhibit high expression levels in the fibroblast cells while significantly low levels in the pluripotent cells. Only two proteins from the family of NFκB-binding proteins fulfilled these criteria (Fig. 2B). NFκB1 and RelA showed high expression levels in the fibroblast cell lines and low in both the hESC and iPS cell lines, similar to the expression patterns seen in HLA-A/B/C and β2M, and thus suggests that these DNA-binding factors are involved in controlling their expression (Fig. 2B, fibroblasts vs. iPS cells expression, NFκB1 p < .004, RelA p < .0002). All other DNA-binding factors expression patterns did not show this pattern (supporting information Fig. S3).
To further strengthen the correlation between expression of NFκB1 and RelA to that of HLA-A/B/C and β2M, we represented by linear regression, the expression levels of HLA-A/B/C or β2M against the expression levels of nuclear factors that bind to the promoter of HLA/A/B/C or β2M in the pluripotent cell lines (hESCs and iPS cells) (Fig. 2C and supporting information Fig. S4). The coefficient of determination, or R2 values, for β2M versus each binding factor was determined and values higher than 0.65 were consider significant. Only the two binding factors NFκB1 and RelA, from the NFκB family of proteins, showed high correlation with MHC-I expression (left blue columns in Fig. 2C).
Basal Expression of NFκB1 and RelA RNA and Protein in iPS Cells
In order to further prove that NFκB1 and RelA are involved in regulating the MHC-I levels, their expression was assessed in pluripotent cells before and after treatment with IFNγ, given that IFNγ induces MHC-I expression. As expected, the basal levels of NFκB1 and RelA were high in fibroblasts, while significantly lower in hESCs and iPS cells (Fig. 3A [p < .01 for NFκB1 and p < 10−5 for RelA] and Fig. 3B), thus reflecting results seen in the expression microarray analysis (Fig. 2B). We have also demonstrated by immunofluorescence assay nuclear expression of NFκB1 (p50) and RelA (p65) in the BJ fibroblasts, whereas very little expression can be found in the BJ-iPS 94 cells (Fig. 3C).
IFNγ Treatment and OCT4 Inhibition Induce the Expression of NFκB1 and RelA in Pluripotent Stem Cells
Upon stimulation with IFNγ, NFκB1, and RelA, RNA levels significantly increased (p < .01) over fourfolds in hESC and iPS cell lines with relative little increase in the fibroblast cells (Fig. 4A). Moreover, protein levels for NFκB1 increased at least twofolds in iPS cells, and in most cases significantly, with the treatment of IFNγ, while no such induction was observed in the fibroblast cells (*, p < .01, Fig. 4B, 4C). These results further strengthen the possibility that NFκB1 and RelA regulate the expression of the MHC-I complex in pluripotent stem cells. Accordingly, following chromatin immunoprecipitation analysis, RelA protein was indeed found to bind to the promoters of β2M and HLA-B in BJ fibroblasts (supporting information Fig. S5).
OCT4 is a crucial protein in pluripotent stem cells and its expression is central to maintaining the pluripotent state seen in hESCs and iPS cells. Downregulation of its expression can be performed using specific siRNA. OCT4 siRNA was transfected into cultures of hESCs and iPS cells; 48 hours later, RNA was extracted and expression levels of NFκB1, RelA, and OCT 4 were quantified using real-time PCR. Upregulation of both NFκB1 and RelA transcripts could be detected in cells that received siOCT4 while levels of OCT4 was reduced in the treated cells as compared to cultures that contained either scrambled siRNA or no siRNA (Fig. 4D).
Manipulation of NFκB1 and RelA Expression Increases MHC-I Expression
To verify the direct relationship between NFκB1 and RelA for MHC-I expression, NFκB1 and RelA were transiently overexpressed in hESC and iPS cell lines, and the cells were subsequently analyzed for expression of HLA-A/B/C and β2M. Analysis by flow cytometry showed an increase in expression of HLA-A/B/C and β2M following 48 hours of overexpression of NFκB1 and RelA (Fig. 5). Transfecting both plasmids increased the expression of the MHC-I complex in pluripotent stem cells even further (Fig. 5). To examine whether suppressing the transcription of NFκB1 will decrease the expression of β2M, plasmids containing shRNAs targeted to silence RelA and NFκB1 were transfected into hESCs. Stable clones showed decrease in RelA and NFκB1 (Supporting Information Fig. S6A) transcription. hESC clones containing the Sh-RelA and sh-NFκB1 plasmids were incubated with IFNγ for 48 hours to induce MHC-1 expression. Indeed Sh-RelA and sh-NFκB1 stable clones showed marked decrease in relative expression of β2M (supporting information Fig. S6B).
Allogeneic transplant recipients need continuous immunosuppressive treatment to prevent graft rejection. However, life-long suppression of the immune system reduces the appropriate response to infectious resulting in a substantial increase in morbidity for the host. Thus, research in the field of transplantation medicine has focused on ground-breaking strategies to increase tolerance to donor tissue. Patient cells immunologically recognize the transplant tissue as foreign mainly through the MHC, which are expressed on every nucleated cell of the body. However, due to the genetic variability in humans, the chances of finding full MHC match are very low.
Fundamentally, the more mismatch in MHC the greater chances of rejection by the patient to the donor tissue. However, there is an important exception to this rule, wherein pregnant women tolerate their unborn fetuses, although the fetus expresses a full set of nonmaternal antigens inherited by the father. It has been shown that the nonclassic MHC-I molecule, HLA-G, is selectively expressed on cytotrophoblasts at the feto-maternal interface and plays an important role in maternal tolerance of the fetus [21–23]. Expression levels of MHC-I proteins in the ICM of human blastocysts from in vitro fertilization (IVF) clinics have shown variable results [23, 24]. However, once hESCs, isolated from the ICM of blastocyst stage embryos, could be grown into cell lines, extensive studies showed that the hESC lines expressed very low levels of the MHC-I complex  and this increased only after differentiation of hESCs. Thus, the basal level of immunogenicity in hESC lines increases following differentiation [3, 4, 25–27]; although they are less immunogenic than adult cells . It was unclear whether the low levels of MHC-I in hESCs are related to their unique developmental origin or to their pluripotent phenotype. The generation of iPS cells from somatic cells may allow us to shed light on the origin of their low immunogenicity. Human iPS cells are truly pluripotent cells; however, they are clearly derived from somatic cells. Low levels of MHC-I in iPS cells, and similar regulation of these molecules in ESCs and iPS cells, suggest that the low immunogenicity is related to the pluripotent phenotype.
The generation of iPS cells from adult cells enables the production of patient-specific pluripotent stem cells, and thus the potential to generate any cell in the body without the risk of rejection. However, time constraints and cost prevent the actualization of this possibility, and thus allogeneic transplantation of iPS cells may prove to be a more practical solution. It has been suggested that if a hESC line bank was generated with approximately 50 blood-group 0 donors, it would provide MHC compatible cells for approximately 75% of the recipient pool . However, the law of diminishing returns applies so that if more hESC lines were added to the bank, the contribution to increasing the recipient pool would become less and less significant . Similar to the case with ESCs, iPS cell repositories are in the process of being established.
Our aim is to study the basis of the low immunogenicity of pluripotent stem cells and thus to enable safer transplantation. Specifically, we have tried to unravel the pathways that might maintain the low levels of MHC-I in iPS cells (Fig. 1). Both the promoter of HLA-A/B/C and β2M contain four binding sites: (a) the USF element, (b) the interferon-stimulated response element, (c) the kappa-binding element, and (d) the S-X-Y module . After examining the expression levels of every binding factor and assessed their correlation to the expression levels of either HLA-A/B/C or β2M, we found only two proteins whose expression levels showed significant correlation—NFκB1 and RelA (Fig. 2). Furthermore, we were able to show low expression of NFκB1 and RelA both at the RNA and protein levels in both iPS and ESCs, and that the expression can be increased with treatment of IFNγ (Figs. 3, 4). This is supported by the nuclear localization of both NFκB1 and RelA in adult fibroblast but not iPS cells (Fig. 3).
The nuclear factor kappa light chain enhancer of activated B cell (NFκB) family of proteins that bind to the kappa element is a central component in the regulation of the immune response. These proteins are usually found inactive in the cytoplasm, but upon activation, emanating from antigen receptors TNF or IL-1, translocate to the nucleus . The NFκB family of transcription factors, consist of five proteins, is found in almost all animal cell types and characterized by a Rel homology domain at the N terminus, which is responsible for their homodimerization and heterodimerization . The most abundant form of NFκB heterodimers found in cells contains the NFκB1and RelA types of NFκB. NFκB1 has an active 50 kDa form and a precursor 105 kDa form that is cleaved upon activation of the antigen receptors . In addition, RelA (or p65) contains a transcriptional activation domain on its C-terminal that allows its binding to DNA, thus NFκB1 creates a heterodimer with RelA that binds to the promoter of HLA-A, B, and C and β2M  (supporting information Fig. S5). Furthermore, mice knockout studies assessing the effects of the various proteins in the NFκB family observed lethality at E15 in RelA−/− and lethality at E13 in NFκB1−/− RelA−/− embryos [31, 32].
To further substantiate the role that NFκB1 and RelA play in inducing the expression of HLA-A/B/C and β2M, we have shown that overexpression of NFκB1 or RelA induces MHC-I expression (Fig. 5) and silencing RelA and NFκB1 decreased the expression of β2M (supporting information Fig. S6). We have thus unraveled the molecular mechanism that control HLA-A/B/C and β2M expression in human pluripotent stem cells and allows for the silencing of MHC-I during the reprogramming of somatic cells to human iPS cells.
In conclusion, our data demonstrate that NFκB is involved in the regulation of expression of both HLA-A/B/C and β2M proteins. During the reprogramming of somatic cells to iPS cells, the expression levels of NFκB1 and RelA are reduced, thus decreasing the expression of the MHC-I to levels seen in human embryonic stem cells and potentially decreasing the immunogenicity of these cells.
This research was partially supported by funds from the European Community (ESTOOLS, Grant Number 018739), a research fellowship from the Israel Cancer Research Fund (NY, USA), and the Lady Davis Fellowship Trust and by a grant from Roche-Yissum collaboration. We would like to thank Ela Elyada for her technical help with the immunofluorescent staining of NFκB1 and RelA.
DISCLOSURE OF POTENTIAL C ONFLICTS OF INTEREST
The authors indicate no potential conflicts of interest.