Toward eliminating HLA class I expression to generate universal cells from allogeneic donors

Abstract

Long-term engraftment of allogeneic cells necessitates eluding immune-mediated rejection, which is currently achieved by matching for human leukocyte antigen (HLA) expression, immunosuppression, and/or delivery of donor-derived cells to sanctuary sites. Genetic engineering provides an alternative approach to avoid clearance of cells that are recognized as "non-self" by the recipient. To this end, we developed designer zinc finger nucleases and employed a "hit-and-run" approach to genetic editing for selective elimination of HLA expression. Electro-transfer of mRNA species coding for these engineered nucleases completely disrupted expression of HLA-A on human T cells, including CD19-specific T cells. The HLA-A(neg) T-cell pools can be enriched and evade lysis by HLA-restricted cytotoxic T-cell clones. Recognition by natural killer cells of cells that had lost HLA expression was circumvented by enforced expression of nonclassical HLA molecules. Furthermore, we demonstrate that zinc finger nucleases can eliminate HLA-A expression from embryonic stem cells, which broadens the applicability of this strategy beyond infusing HLA-disparate immune cells. These findings establish that clinically appealing cell types derived from donors with disparate HLA expression can be genetically edited to evade an immune response and provide a foundation whereby cells from a single donor can be administered to multiple recipients.

Figure 1

Designed ZFNs disrupt HLA-A expression on HEK293. (A) Schematic of the HLA-A genetic loci. Arrows indicate the location of the putative ZFN binding sites. (B) ZFN binding sites within the HLA-A alleles expressed by HEK293. A 46-base pair segment of the HLA-A*03:01 and HLA-A*02:01 alleles is shown, with bold and underlined nucleotides highlighting the anticipated binding sites for the ZFN-L and ZFN-R monomers, respectively. Cleavage occurs in the region between these sites. Note that HLA-A*02:01 differs from and HLA-A*03:01 at 3 positions within this region, which are marked with an asterisk. ZFN-L was designed to ignore these polymorphisms and to cleave both alleles. (C) Levels of HLA-A3 and HLA-A2 genetic disruption assessed by the Surveyor nuclease assay. The lower (fast-moving) bands (arrows) are digestion products indicating ZFN-mediated gene modification. The numbers indicate the percentage of modified HLA-A alleles based on densitometry. GFP represents genetically modified HEK293 expressing GFP as a control. Mock represents no DNA used.

Figure 2

Isolation of HLA-A^negHEK293. (A) Loss of HLA-A2 and HLA-A3 protein expression. Flow cytometry analysis of HLA-A2 and HLA-A3 expression on parental HEK293 and 3 derived genetically modified clones with loss of HLA-A (numbered 18.1, 8.18, and 83). Dotted lines represent isotype (HLA-A2) or SA-PE (HLA-A3) controls; solid line represents HLA-A expression without IFN-γ and TNF-α, and filled lines represent HLA-A expression after culturing with 600 IU/mL IFN-γ and 10 ng/mL TNF-α for 48 hours. Dashed lines in the parental column represent HLA-A2 or HLA-A3 expression on EBV-LCL. (B) Resistance to CTL-mediated lysis. Parental HEK293 and derived HLA-A^neg clones were cultured with IFN-γ and TNF-α for 48 hours and pulsed with serial dilutions of the cognate HLA-A3 peptide RVWDLPGVLK (derived from PANE1 and recognized by CTL clone 7A7) or the HLA-A2 peptide CIPPDSLLFPA (derived from C19ORF48/A2 and recognized by CTL clone GAS2B3-5) and evaluated for recognition by CTL clones in a 4-hour ⁵¹Cr release assay at an effector-to-target ratio of 20:1.HLA-A2⁺ LCL (hatched bar) that expresses PANE1mHAg (not peptide-loaded) were used as a positive control.

Figure 3

Loss of HLA-A expression on primary OKT3-propagated T cells after genetic editing with ZFNs. (A). Loss of cell surface expression of HLA-A2 after electro-transfer of mRNA species encoding ZFN-L and ZFN-R targeting HLA-A2 (top). T cells were harvested 6 days after initial stimulation with γ-irradiated aAPC. Five million T cells were premixed with ZFN mRNA in 100 μL Human T-cell Nucleofector solution and electroporated in a cuvette using a Nucleofector II device with program T-20. Coexpressions of HLA-A2, CD4, and CD8 were analyzed 4 days after electro-transfer of graded doses of the mRNA species encoding ZFN-L and ZFN-R. Flow cytometry data were gated on the propidium iodide-negative, live cell population. Numbers in the lower-right quadrant indicate the percentage of CD4 and CD8⁺ T cells that are HLA-A^neg. Improved disruption of HLA-A expression by cold shock (bottom). Data were collected 4 days after electro-transfer of graded doses of the mRNA species encoding ZFN-L and ZFN-R. Cells were cultured at 30°C from days 1 to 3 after electro-transfer of ZFNs, returned to 37°C, and cultured for 1 additional day before analysis. (B) Improved efficiency of HLA-A disruption by ZFN-L and ZFN-R fused to the heterodimeric FokI domain variants. mRNA species encoding the ZFN-L and ZFN-R heterodimeric FokI mutants EL:KK targeting HLA-A were electro-transferred into primary T cells. HLA-A2 expression was analyzed after culturing the cells for 4 days at 37°C or 3 days at 30°C followed by 37°C for 1 day. X-axis represents CD4 and CD8 expression and y-axis represents HLA-A2 expression.

Figure 4

Enrichment of HLA-A^neg primary T cells after genetic editing with ZFNs. (A) Generation of an HLA-A2^neg T-cell population. HLA-A2^neg T cells were enriched by magnetic bead-based selection. Input dose of mRNA coding for ZFN and 3-day culture conditions (37°C vs 30°C) after electro-transfer of mRNA are indicated. The numbers represent HLA-A2 negative population within the CD4- and CD8-positive population. EL:KK, obligate heterodimer mutant FokI cleavage domain; wt, wild type FokI cleavage domain. (B) Surveyor nuclease assay of the enriched HLA-A2^neg T cells. Analysis of T cells enriched for loss of HLA-A2 expression demonstrates disruption in the HLA-A2 locus by the appearance of a fast-moving band (arrow). (C) Sequencing of the HLA^neg T cells. PCR products using HLA-A2-specific primers from enriched cell (2.5 μg ZFNs, EL:KK FokI domain, 30°C treatment) were cloned into a TOPO vector (Invitrogen), and plasmid products were sequenced. The wild-type sequence is listed at the top, with the expected ZFN binding sites underlined. Shown below are the sequences obtained from the ZFN-treated and enriched cells. Deletions are indicated by hyphens, and sequence changes are highlighted in bold. All 18 sequence changes result in frame shifts predicted to prevent protein translation.

Figure 5

Loss of HLA-A expression on primary CD19-specific CAR⁺T cells genetically edited with ZFNs. (A) Disruption of HLA-A2 in CAR⁺T cells by electro-transfer of mRNA encoding ZFNs. T cells from a HLA-A2⁺ donor were electroporated and propagated to express CD19-specific CAR (CD19RCD28). These T cells were re-electroporated with 2.5 μg of each mRNA encoding the heterodimeric FokI domain variants of the HLA-A-specific ZFNs (ZFN-L-EL and ZFN-R-KK). HLA-A2 expression was analyzed after culturing at 30°C for 3 days, followed by 37°C for 1 day. Enrichment of the HLA-A2^neg population was performed by paramagnetic selection. (B) HLA-A^neg CAR⁺T cells evade lysis by HLA-A2-restricted CTL. Pools of the indicated CAR⁺T cells were pulsed with serial dilutions of cognate peptide before being used as targets in a CRA. CTL clone GAS2B3-5, which is specific for C19ORF48/A2, was added at an effector-to-target ratio of 20:1. (C) ZFN-modified HLA^neg CAR⁺ T cells maintain desired antigen-specific cytotoxicity. Redirected specificity for CD19 by HLA-A^neg T cells expressing CD19RCD28CAR was demonstrated using the mouse T-cell line EL4 genetically modified to express a truncated variant of human CD19. Expression of introduced human CD19 on EL4 was 100%. (D) HLA^neg CAR⁺ T cells maintain cytotoxicity against CD19⁺malignant cells. Cytotoxicity of HLA-A^neg CD19-specificCAR⁺ T cells was evaluated against CD19⁺ cell lines (NALM-6 and Daudi) and primary lymphoma cells derived from patients. Data shown are from an effector-to-target ratio of 10:1. Primary cells from DLBCL are diffuse large B-cell lymphoma, and those from MCL are mantle cell lymphoma.

Figure 6

ZFN-mediated elimination of HLA-A expression on human ESC. The HLA-A2⁺HLA-24⁺hES parental cell line WIBR3 was modified by ZFN and donor plasmid coding for antibiotic resistance. Clones (5230, 5255, 5258) were chosen with loss of HLA-A expression and differentiated into fibroblasts. Expression of HLA-A2 and HLA-A24 on derived fibroblasts was assessed by flow cytometry after culturing with 600 IU/mL IFN-γ and 10 ng/mL TNF-α for 48 hours. Dashed line in parental panel represents isotype control.