An effective approach to prevent immune rejection of human ESC-derived allografts

Abstract

Human embryonic stem cells (hESCs) hold great promise for cell therapy as a source of diverse differentiated cell types. One key bottleneck to realizing such potential is allogenic immune rejection of hESC-derived cells by recipients. Here, we optimized humanized mice (Hu-mice) reconstituted with a functional human immune system that mounts a vigorous rejection of hESCs and their derivatives. We established knockin hESCs that constitutively express CTLA4-Ig and PD-L1 before and after differentiation, denoted CP hESCs. We then demonstrated that allogenic CP hESC-derived teratomas, fibroblasts, and cardiomyocytes are immune protected in Hu-mice, while cells derived from parental hESCs are effectively rejected. Expression of both CTLA4-Ig, which disrupts T cell costimulatory pathways, and PD-L1, which activates T cell inhibitory pathway, is required to confer immune protection, as neither was sufficient on their own. These findings are instrumental for developing a strategy to protect hESC-derived cells from allogenic immune responses without requiring systemic immune suppression.

Figure 1

Hu-mice can mount robust immune rejection of hESC-derived allografts. (A) Representative FACS analysis of spleen, peripheral blood mononuclear cells (PBMC) and transplanted human thymus derived from Hu-mice. Single cell suspension was stained for the markers of human T cells (CD3, CD4, and CD8) and B cells (CD19). N>10. Single cell suspension derived from the spleen of NSG mice was used as a negative control. (B) Sections of spleens from NSG and Hu-mice were stained with hematoxylin and eosin, or with antibodies against human CD3, CD4, CD8, and human nuclei to show the repopulation of human T cells in spleen. Nuclei were counterstained with DAPI. Scale bar, 100 μm. Representative images are shown. N=4 for NSG group, N>20 for Hu-mice group. (C) NSG and Hu-mice were subcutaneously implanted with allogeneic Hues 3 and Hues 8 hESCs around the hindlegs. Six-to-eight weeks after implantation, teratomas were harvested, sectioned and stained with anti-human CD4 and CD8 antibodies. (D) Extensive necrosis was detected in the teratomas formed by allogeneic Hues 3 and Hues 8 hESCs in Hu-mice, as revealed by hematoxylin and eosin staining. N=4 for Hues 3/Hu-mice group, N=6 for Hues 8/Hu-mice group, N>10 for Hues 3/NSG group, N>10 for Hues 8/NSG group. See also table S1.

Figure 2

Generation of Hues 3 CP hESCs. (A) The endogenous human hypoxanthine phosphoribosyltransferase 1 (HPRT1) locus. Open box indicates the 3′ UTR of HPRT1. Filled boxes indicate part of HPRT1 coding sequence. The stop codon (TAA) and the binding sites of the primers used for identification of targeting clones are indicated. (B) BAC-based targeting vector. The LoxP flanked selection cassette was inserted between the stop codon and the PolyA signal sequence of HPRT1 to block its expression, introducing both positive and negative selections during targeting process. The CAG promoter-driving expression cassette, CAG/CTLA4-Ig/IRES/PD-L1/pA, was inserted about 600 bps downstream of HPRT1 gene. The sizes of homologous arms are indicated. IRES, internal ribosomal entry site. (C) PCR analysis of the targeted clones in human male ESCs (HUES 3). WT, wide type parental HUES 3; 1 & 2, two targeted clones CP-1 & CP-2; Tg, a random integration clone. The primers used are indicated in A and B. (D) Drug sensitivity assay to confirm the expression and function of HPRT1 in the knock-in clones. Cells were seeded onto 12-well plates. At the following day, the media were changed to that containing hypoxanthine/aminopterin/thymidine (HAT), or 6-thioguanine (6-TG), or without a drug. After being treated for three days, the cells were stained with an alkaline phosphatase detection kit. (E) The expression and secretion of CTLA4-Ig was confirmed. Loading buffer with or without the reducing agent β-mercaptoethanol was used to evaluate the dimerization status of CTLA4-Ig. (F) The surface expression of PD-L1 was confirmed by flow cytometry. See also Figure S1

Figure 3

Expression of PD-L1 and CTLA4-Ig protects CP hESC-derived teratomas from allogeneic immune rejection. (A) Images of teratomas derived from WT and CP hESCs formed in Hu-mice and NSG mice. Mice were subcutaneously injected with WT hESCs and CP hESCs around the left and right hindlegs, respectively. Six-to-eight weeks after implantation, the mice were euthanized and teratomas examined. Representative images are shown. (B) Extensive tissue necrosis was detected in the teratomas formed by WT hESCs in Hu-mice. Significant T cell infiltration was detected in the teratomas formed by WT hESCs but not those formed by CP hESCs in Hu-mice as shown by immunohistochemistry (C) and flow cytometry (D). (E) Summary of teratoma formation, immune rejection and CD4⁺ T cell infiltration. Teratomas with apparent regressing phenotype or containing only liquid-filled cysts without cell mass were classified as rejection. (F) Relative mRNA levels of IL-10, TGFβ1 and IL-2 in T cells isolated from CP hESC- and WT hESC-derived teratomas formed in the same Hu-mouse were determined by real-time PCR. Mean values are presented with SD (N=3). See also Figure S2 and S3.

Figure 4

Various lineages of cells derived from CP hESCs are protected from allogeneic immune rejection. (A) Various cell lineages were present in the teratomas formed by CP hESCs in Hu-mice. SE, squamous epithelium; R, rosette; B, bone; C, cartilage; H, hepatocyte-like cells; GE, gut-like epithelium. (B) Established teratomas formed by CP hESCs in Hu-mice contained various cell lineages indicated by immunohistochemical staining. NeuN, neuronal marker; C-peptide, pancreatic β cell marker; Keratin, pan-epidermis marker. (C) The relative expression levels of the three germ layer specific genes in teratomas formed by CP hESCs derived from NSG and Hu-mice. Mean values are presented with SD (for NSG mice, N=7; for Hu-mice, N=12).

Figure 5

Fibroblasts and cardiomyocytes derived from CP hESCs are protected from allogeneic immune rejection in Hu-mice. Fibroblasts (A) or cardiomyocytes (B) derived from WT and CP hESCs were transplanted into the same Hu-mice or NSG mice. Two weeks later, the grafts were recovered and stained with indicated antibodies. T cells were identified by anti-CD3 antibody, and human cells in the grafts identified by a human nuclei-specific antibody (Hu-Nuclei). Representative images are shown. Parental WT and CP hESC-derived grafts in NSG mice were stained as negative controls. Nuclei were counterstained with DAPI. Scale bar, 50 μm. N=3 per group. (C) hESC-derived cardiomyocyte allografts transplanted in Hu-mice were sectioned and stained for T cells (CD3+) and human cardiomyocytes (cTnl+), indicating the presence of cardiomyocytes but lack of infiltrating T cells in the CP hESC-derived cardiomyocyte allografts. (D) Quantification of T cells infiltrated into fibroblast and cardiomyocyte allografts transplanted into Hu-mice. CD3⁺ and HuNu⁺ cells in three randomly selected view fields were counted. The ratio of CD3⁺ to HuNu⁺ cells was used to quantify for T cell infiltration. Mean values are presented with SD (N=3). See also Figure S4.

Figure 6

Expression of PD-L1 or CTLA4-Ig alone cannot protect the derivatives of hESCs from allogeneic immune rejection. (A) Extensive T cell infiltration was detected in the teratomas formed by WT hESCs, PD-L1-KI-hESCs and CTLA4-Ig-KI-hESCs in Hu-mice. T cells were identified by anti-CD4, anti-CD8 and anti-CD3 antibodies. (B) Extensive necrosis was detected in the teratomas derived from WT hESCs, PD-L1-KI-hESCs and CTLA4-Ig-KI-hESCs in allogeneic Hu-mice, as revealed by hematoxylin and eosin staining. (C) The frequency of CD4⁺ T cell infiltration and teratoma rejection in Hu-mice transplanted with parental and various knock-in hESCs. See also Figure S5.