Fig 2: Completely immune-evasive human HIP iECsCD64.a, The HIP immune phenotype of HLA class I and II deficiency and CD47 overexpression was confirmed in flow cytometry (representative graph of two independent experiments). b, Single antigen bead assay results for HLA class I and II antibodies in transplant recipients 1 and 2 experiencing AMR. c, Human iECs and HIP iECs were incubated with serum from both patients in impedance CDC assays (mean ± s.d.; three independent replicates per group and timepoint). d, Flow cytometry histogram for MICA expression on human HIP iECs (representative graph of two independent experiments). e, Flow cytometry histograms for CD64 expression on human HIP iECs and HIP iECsCD64 (representative graph of two independent experiments). f, Flow cytometry histograms for the binding of free IgG1 Fc (anti-CD52, alemtuzumab; representative graph of two independent experiments). g,h, Human HIP iECs and HIP iECsCD64 were challenged in impedance NK cell ADCC (g) and CDC (h) assays with different concentrations of an anti-MICA IgG1 antibody (mean ± s.d.; three independent replicates per group and timepoint).

Fig 3: CD64 expression protects human HIP iECsCD52,CD64 from non-HLA antibody killing in vitro and in vivo.a, Flow cytometry histograms for CD52 and CD64 expression on human HIP iECsCD52 (representative graphs of two independent experiments). b, Flow cytometry histograms for CD52 and CD64 expression on human HIP iECsCD52,CD64 (representative graphs of two independent experiments). c,d, Human HIP iECsCD52 and HIP iECsCD52,CD64 were challenged in impedance NK cell ADCC (c) and CDC (d) assays with different concentrations of an anti-CD52 IgG1 antibody (mean ± s.d.; three independent replicates per group and timepoint). e, 5 × 104 HIP iECsCD52 or HIP iECsCD52,CD64 were injected subcutaneously into NSG mice with 106 human NK cells. Some groups received three subcutaneous doses of alemtuzumab 1 mg on days 0, 1 and 2. f,g, BLI signals of HIP iECsCD52 (f) and HIP iECsCD52,CD64 (g) were followed (all individual mice were plotted, and BLI pictures of one representative mouse per group are shown). s.c., subcutaneous.

Fig 4: Human CAR TCD64t cells are protected from antibody-mediated killing.a, Flow cytometry histograms for anti-CD19 scFv and CD64t expression on human CAR T and CAR TCD64t cells (representative graphs of two independent experiments). b, Flow cytometry histograms for the binding of free IgG1 Fc (anti-TPO IgG1; representative graph of two independent experiments). c, The kinetics of Nalm6 target cell killing by T cells, CAR T cells and CAR TCD64t cells is expressed as hours that it takes for the cell index to drop from 1 to 0.5. Different T cell-to-Nalm6 ratios are shown (mean ± s.d.; three independent replicates per group and timepoint). d, The kinetics of Nalm6 target cell killing by CAR TCD64t in the presence and absence of 1 µg ml-1 of anti-CD52 is expressed as hours that it takes for the cell index to drop from 1 to 0.5. Different CAR T cell-to-Nalm6 ratios are shown (mean ± s.d.; three independent replicates per group and timepoint). e,f, Human CAR T and CAR TCD64t cells were challenged in impedance NK cell ADCC (e) and CDC (f) assays with antibodies against HLA (HLA-A2), non-HLA (CD52 and CD3), rhesus blood type antigen D (Rh(D)) and the CAR (anti-CD19 scFv) at 1 µg ml-1 (mean ± s.d.; three independent replicates per group and timepoint).

Fig 5: Human HIP iECsCD52,CD64 evade all cellular and antibody-mediated immune attacks.a, Human HIP iECsCD52 and HIP iECsCD52,CD64 were challenged with isolated allogeneic innate immune cells. Allogeneic NK cells with or without IL-2, allogeneic macrophages or allogeneic PMN cells with or without IL-2 were used. b, Human HIP iECsCD52 and HIP iECsCD52,CD64 were challenged with primed allogeneic PBMCs after they underwent 10 days of in vitro stimulation with the target cells. Then, PMNs and serum were added to combine all immune compartments of the peripheral blood in this assay. c, Human HIP iECsCD52 and HIP iECsCD52,CD64 were challenged with all immune compartments of the peripheral blood while adding anti-CD52 to enable ADCC and CDC. All graphs show mean ± s.d. and three independent replicates per group and timepoint. d, 5 × 104 HIP iECsCD52 or HIP iECsCD52,CD64 were injected subcutaneously into humanized mice with 106 human NK cells. Both groups received three subcutaneous doses of anti-CD52 IgG1 (alemtuzumab) 1 mg on days 0, 1 and 2. e,f, BLI signals of HIP iECsCD52 (f) and HIP iECsCD52,CD64 (g) were followed (all individual mice were plotted, and BLI pictures of one representative mouse per group are shown). s.c., subcutaneous.