Precision T cell correction platform for inborn errors of immunity

Abstract

CRISPR-Cas9 gene editing is a promising tool to correct pathogenic variants for autologous cell therapies targeting inborn errors of immunity (IEI). Current strategies, such as gene knockout or cDNA knockin, address many single-gene defects but can disrupt gene expression, highlighting the need for precise correction platforms. While transplanting corrected autologous hematopoietic stem cells is a curative approach, it is unsuitable for patients with advanced disease, inflammation, or acute infections. As correcting T cells is an alternative therapeutic strategy for lymphoid IEIs, we present an efficient T cell single-nucleotide variant (SNV) correction platform based on homology-directed repair (HDR). By using STAT1 gain-of-function, cartilage hair hypoplasia, deficiency of ADA2, and autoimmune polyendocrinopathy-candidiasis-ectodermal dystrophy as IEI models, we demonstrate that our platform achieves up to 80% correction, with resultant functional correction of the disease phenotype in the selected models. Furthermore, we performed safety profiling using GUIDE-seq, single-cell RNA sequencing, long-read genome sequencing, and proteomics analysis and detected no genomic, transcriptomic, or proteomic aberrations. This study establishes HDR-based SNV editing as a portable method for developing clinical autologous T cell therapies and represents a promising step toward a broad-spectrum gene correction platform for treating diverse monogenic immune disorders.

Keywords: CRISPR-Cas9 gene correction; autologous T cell therapy; ex vivo gene editing; gene therapy; homology-directed repair; inborn errors of immunity; non-viral genome editing; platform technology; primary T cell editing; single-nucleotide variant correction.

Graphical abstract

Figure 1

Repair strategy and gRNA screening in patient cells (A) Schematic representation of the gRNA screening strategy, where multiple gRNAs were assessed based on available PAM sites within the 100-bp ssODN area. Forward gRNAs and their PAMs are marked in blue and reverse in yellow. (B) Schematic representation of the repair strategy used in the study, where 100-bp ssODNs with ±50-bp homology arms from the mutation site (red) were used. ssODN design includes correction of the mutation (green) and 3–4 silent SNVs (pink), enabling identical editing strategy in patients and healthy controls and HDR detection by ddPCR. (C) Schematic representation of the ddPCR assay design for HDR and NHEJ detection. (D) ADA2 gRNA screening in HD T cells, fibroblasts and CD34⁺ HSPCs, assessed by ddPCR (n = 4 technical replicates for T cells and fibroblasts, n = 3 for HSPCs). (E) ADA2 gRNA screening in DADA2 patient T cells and fibroblasts, assessed by ddPCR (n = 3 technical replicates). ADA2 gRNA number 4 (asterisk) was not tested in patients due to PAM loss caused by the mutation. (F) AIRE gRNA screening in APECED patient T cells and fibroblasts, assessed by ddPCR (n = 3 technical replicates). (G) RMRP gRNA screening in CHH patient T cells and fibroblasts, assessed by ddPCR (n = 3 technical replicates). One independent experiment was performed for all sets of data. Statistical significance of best-performing gRNAs was assessed by one-way ANOVA with Fisher’s least significant difference (LSD) test, where ∗∗∗p < 0.0002 and ∗∗∗∗p < 0.0001. Bar denotes mean value, error bars represent ±SD.

Figure 2

Establishment and assessment of CRISPR-Cas9 T cell editing platform (A) Schematic representation of the CRISPR-Cas9 T cell editing platform. PBMCs from patient and HD blood samples are first isolated and cryopreserved. PBMCs are thawed on day 1 and stimulated for 3 days with interleukins: IL-2 (120 U/mL), IL-7 (3 ng/μL), and IL-15 (3 ng/μL) and soluble CD3/CD28 (15 μL/mL), which activate and induce expansion of CD3⁺ T cells. Cells are nucleofected on day 4 with custom CRISPR reagents (gRNA, Cas9 nuclease, ssODN). Afterward, cells are cultured for 4 days in IL-2 (250 U/mL), during which Cas9-mediated double-stranded breaks are repaired by HDR/NHEJ. On day 8, cells are harvested for downstream assays, expanded further, or cryopreserved. (B) ADA2 HDR and NHEJ editing in HD T cells with 0.1–1 M nucleofected cells/sample, measured by ddPCR (n = 3 technical replicates). Comparison of different T cell culture media during 11-day cytokine stimulation, assessed by (C) T cell viability (dots represent mean of n = 3 biological replicates), (D) T cell fold change (n = 3 biological replicates), and (E) ADA2, AIRE, and RMRP HDR editing on day 8 (n = 3 technical ddPCR replicates from n = 3 biological replicates). (F) ADA2 HDR editing in HD T cells with Cas9 nuclease at 3.05–15.25 μmol/L/sample, gRNA at 5–25 μmol/L/sample, and ssODN at 5 μmol/L/sample/sample, measured by ddPCR (n = 3 technical replicates). (G) ADA2 HDR editing in HD T cells with selected Cas9-gRNA concentrations and ssODN at 5–25 μmol/L/sample, measured by ddPCR (n = 3 technical replicates). Dashed line indicates mean of Cas9 nuclease at 3.05 μmol/L/sample, gRNA at 5 μmol/L/sample, and ssODN at 5 μmol/L/sample. (H) Frequency of immune cells (CD4⁺, CD8⁺, monocytes, NKT cells, NK cells, B cells) in six HDs on days 1, 4, and 8 (mock, ADA2 edited, or AIRE edited) of the platform, assessed by flow cytometry. Each ring of the doughnut plot represents one HD. HDR editing levels in CD4⁺ and CD8⁺ and the bulk of cells for ADA2 (I) and AIRE (J) on day 8, measured by ddPCR (n = 1 technical replicate). One independent experiment was performed for all sets of data except for (C)−(E), where data from three donors are shown in the graphs, and (F) and (G), where one out of three representative experiments is shown. Bar denotes mean value, error bars represent ±SD.

Figure 3

HDR enhancement in healthy control and patient T cells (A) Schematic representation of asymmetric ssODN designs with 10- to 90-bp homology arms on either side from target site. HDR editing with asymmetric ssODNs in HD T cells for (B) ADA2, (C) AIRE, and (D) RMRP, measured by ddPCR (n = 3 technical replicates). HDR editing with 3′ LNA- or 3′ PT-modified, position-optimized ssODNs in HD T cells for (E) ADA2, (F) AIRE, and (G) RMRP, measured by ddPCR (n = 3 technical replicates). (H) Validation of HDR-enhancing compounds at three concentrations in increasing order (conc1–3) for ADA2 editing in HD T cells, assessed by ddPCR (n = 2 technical replicates per condition). Compounds were assessed in three HDs, where mean of all donors per condition were compared to the mean of edited DMSO-treated baseline. Statistical significance was assessed by ANOVA. For the heatmap, percentage of HDR fold change from baseline was calculated for each compound concentration. Statistically significant concentrations are indicated with black asterisks. Conc2–3 are marked with a cross for those compounds that were assessed at one concentration. (I) HDR editing for ADA2, AIRE, CTCF1, Enh4-1, RMRP, and RNF2, measured by ddPCR (n = 3 technical replicates) with selected HDR-enhancing compounds (4 μM NU7441, 0.5 μM KU0060648, and 1 μM IDT Alt-R enhancer V2) or DMSO in HD T cells. (J) HDR in DADA2, APECED, and CHH patient T cells with concentration-optimized HDR enhancing compounds (0.5 μM KU0060648 and 0.6 μM IDT Alt-R enhancer V2), where ADA2, AIRE, and RMRP loci, respectively, were corrected. HDR levels were assessed by ddPCR for ADA2 and AIRE (n = 3 technical replicates) and by amplicon sequencing for RMRP (n = 2 technical replicates). (K) ADA2, AIRE, and RMRP HDR editing in HD CD34⁺ HSPCs, measured by ddPCR (n = 3 technical replicates) with concentration-optimized HDR enhancing compounds (0.5 μM KU0060648 and 0.6 μM IDT Alt-R enhancer V2) or DMSO. (L) HDR editing in ADA2, AIRE, and RMRP in HD T cells at different cell passages (p1–p5), measured by ddPCR (n = 3 technical replicates) with concentration-optimized HDR enhancing compounds (0.5 μM KU0060648 and 0.6 μM IDT Alt-R enhancer V2) or DMSO. Three independent experiments were performed for all sets of data where representative experiment is shown, except for (H) where average measurements from three HDs is shown and (J) where all patients are shown in the graph. Bar denotes mean value, error bars represent ±SD. Statistical significance for all sets of data, except (H), was assessed by one-way ANOVA with Fisher’s LSD test, where ∗p < 0.01, ∗∗p < 0.001, ∗∗∗p < 0.0002, and ∗∗∗∗p < 0.0001.

Figure 4

gRNA off-target profiling by GUIDE-seq in patient and healthy control T cells DADA2 patient T cell (A) viability (n = 4 technical replicates) and (B) count (n = 4 technical replicates) 24–96 h after nucleofection with 0–2.5 μmol/L/sample dsODN and ADA2 RNPs. (C) dsODN integration 96 h after nucleofection in DADA2 patient T cells, with 0–2.5 μmol/L/sample dsODN and ADA2 RNPs, assessed by ddPCR (n = 3 technical replicates) using forward (gray) and reverse (pink) dsODN probes for detection. (D) Schematic representation of the GUIDE-seq experiment. DADA2, APECED, and CHH patient and HD PBMCs were thawed and stimulated with IL-2 (120 U/mL), IL-7 (3 ng/μL), IL-15 (3 ng/μL), and soluble CD3/CD28 (15 μL/mL) on day 1, diluted on day 4, and nucleofected on day 5 with 1.5 μmol/L/sample dsODN and selected RNPs or mock. Cells were cultured in IL-2 (250 U/mL) until sample collection on day 9, followed by genomic DNA (gDNA) extraction, ddPCR for dsODN integration, and GUIDE-seq library preparation. GUIDE-seq in patient and HD T cells for (E) ADA2 gRNA number 3, (F) AIRE gRNA number 11, (G) RMRP gRNA number 9, and (H) HEK-site 4 gRNA, targeting the endogenous human embryonic kidney HEK-site 4. GUIDE-seq results are shown as mismatch plots, where the on-target sequence is depicted at the first line of the table with sequencing read counts per individual (right). The most abundant off-targets, if applicable, are listed under the target with their corresponding locations in the genome (left) and sequencing read counts (right). One independent experiment was performed for all sets of data. Bar denotes mean value, error bars represent ±SD.

Figure 5

scRNA-seq assessment of CRISPR-Cas9 and HDR-enhancing compounds in DADA2 patient and HD T cells (A) Outline of the experiment. HD and DADA2 patient PBMCs were thawed and stimulated with IL-2 (120 U/mL), IL-7 (3 ng/μL), IL-15 (3 ng/μL), and soluble CD3/CD28 (15 μL/mL) on day 1 and nucleofected on day 4 with ADA2 CRISPR RNPs or mock. Cells were cultured in IL-2 (250 U/mL) and HDR enhancers (0.5 μM KU0060648, 0.6 μM IDT Alt-R enhancer V2) or DMSO for 24 h after nucleofection and IL-2 alone afterward. On day 8, 64 CD4⁺ and 64 CD8⁺ T cells per condition (128 cells in total per condition) were sorted into 384-well plates, and gDNA was extracted from the bulk for ddPCR. Sorted cells were further analyzed with RT-qPCR and scRNA-seq. (B) ADA2 HDR editing in HD and DADA2 patients on day 8, assessed by ddPCR (n = 3 technical replicates). (C) ADA2 editing in HD and DADA2 patients, assessed by RT-qPCR of the scRNA-seq libraries with probes to the corrected and uncorrected nucleotide sequence. For HD, 56, 75, and 77 cells were analyzed for DMSO, KU0060648, and IDT Alt-R enhancer V2-treated cells, respectively. For DADA2 patients, 39, 50, and 50 cells were analyzed for DMSO, KU0060648, and IDT Alt-R enhancer V2-treated cells, respectively. Uniform manifold approximation and projection (UMAP) plots generated from scRNA-seq for ADA2-edited HD treated with (D) DMSO, (E) KU0060648, and (F) IDT Alt-R enhancer V2, compared to unedited HD (DMSO). UMAP plots of corrected DADA2 patient treated with (G) DMSO, (H) KU0060648 and (I) IDT Alt-R enhancer V2, compared to uncorrected DADA2 patient (DMSO). (J) Hallmark gene set enrichment results for ADA2-edited HD (DMSO, KU0060648, and IDT Alt-R enhancer V2) compared to unedited HD (DMSO). (K) Hallmark gene set enrichment results for corrected DADA2 patient (DMSO, KU0060648, and IDT Alt-R enhancer V2) compared to uncorrected DADA2 patient. One independent experiment was performed for all sets of data. Bar denotes mean value, error bars represent ±SD. Statistical significance for HDR editing in (B) was assessed by one-way ANOVA with Fisher’s LSD test, where ∗∗∗∗p < 0.0001. NES, normalized enrichment score.

Figure 6

Mass spectrometry analysis of corrected and uncorrected DADA2 patient T cells (A) Outline of the experiment. DADA2 patient and HD PBMCs were thawed and stimulated with IL-2 (120 U/mL), IL-7 (3 ng/μL), IL-15 (3 ng/μL), and soluble CD3/CD28 (15 μL/mL) on day 1 and diluted on day 4 for further expansion. Cells were nucleofected with ADA2 CRISPR RNPs or mock on day 5 and cultured in IL-2 (250 U/mL) and HDR enhancers (0.5 μM KU0060648 and 0.6 μM IDT Alt-R enhancer V2) or DMSO for 24 h. Afterward, cells were cultured in IL-2 (250 U/mL) until sample collection on day 12. (B) ADA2 HDR editing in three DADA2 patients (DADA2 1–3) treated with HDR enhancers or DMSO, assessed by ddPCR (n = 3 technical replicates). (C) Abundance of ADA2 protein in DADA2 patients, reported as intensities (n = 4 technical replicates). Comparison of protein expression levels in (D) ADA2-corrected (DMSO) DADA2 patients to uncorrected DADA2 patients, (E) ADA2-corrected (KU0060648) DADA2 patients to uncorrected DADA2 patients, (F) ADA2-corrected (IDT Alt-R enhancer V2) DADA2 patients to uncorrected DADA2 patients, and (G) uncorrected DADA2 patients to unedited HDs, assessed by mass spectrometry. For (D)–(G), volcano plots were created by reporting protein expression fold change from mean of three DADA2 patients and three HDs on the x axis and –log10 p value on the y axis. One independent experiment was performed for all sets of data. Statistical significance was assessed by one-way ANOVA with Fisher’s LSD test, where ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.0002, and∗∗∗∗p < 0.0001. Bar denotes mean value, error bars represent ±SD. DDX3, ATP-dependent RNA helicase; DNAJB, DnaJ homolog subfamily B; NUDT, U8 snoRNA-decapping enzyme; PARL, presenilin-associated rhomboid-like protein; RNABP2, E3 SUMO-protein ligase RanBP2; TUBB, tubulin beta.

Figure 7

T cell proliferation assay in cartilage-hair hypoplasia patients (A) RMRP HDR editing in three cartilage-hair hypoplasia (CHH) patients (CHH 1–3) 4, 7, and 14 days after nucleofection with concentration-optimized HDR enhancing compounds (0.5 μM KU0060648 and 0.6 μM IDT Alt-R enhancer V2) or DMSO. HDR was assessed by amplicon sequencing (n = 2 technical replicates). (B) Outline of the CFSE-based T cell proliferation experiment. CHH patient PBMCs were thawed and stimulated with IL-2 (120 U/mL), IL-7 (3 ng/μL), IL-15 (3 ng/μL), and soluble CD3/CD28 (15 μL/mL) on day 1 and diluted on day 4 for further expansion. Cells were nucleofected with CRISPR RNPs for RMRP correction or mock on day 6 and cultured in IL-2 (250 U/mL) and 0.5 μM KU0060648 for 24 h after nucleofection. Afterward, cells were cultured in IL-2 (250 U/mL) until re-stimulation on day 13 with the same setup as on day 1. Cells were stained with CFSE on day 20 and cultured in IL-2 (250 U/mL) for 4 days. On day 24, cells were stained for flow cytometry. T cell proliferation in corrected and uncorrected CHH patients for (C) CD4⁺ and (D) CD8⁺ T cells, assessed by flow cytometry. One independent experiment was performed for all sets of data. The patient number corresponds to patient information in Table S15. Bar denotes mean value, error bars represent ±SD. Statistical significance was assessed by one-way ANOVA with Fisher’s LSD test, where ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.0002, and ∗∗∗∗p < 0.0001.

Figure 8

Assessment of STAT1 phosphorylation in corrected and uncorrected STAT1-GOF patients (A) Schematic representation STAT1 gRNA design and repair strategy. Correction of pathogenic mutation is marked with green and silent SNVs in pink. Three gRNAs were designed (green), where the PAM site is represented as an arrow (purple). (B) STAT1 gRNA screening in STAT1-GOF patient T cells assessed by measuring HDR editing using ddPCR (n = 3 technical replicates) in two independent experiments, indicated in light and dark blue. (C) 3′ PT modified asymmetric ssODNs screening with best-performing guide (g number 2) in STAT1-GOF patient T cells assessed by measuring HDR editing using ddPCR (n = 3 technical replicates) in two independent experiments, indicated in light and dark blue. (D) STAT1 HDR editing in T cells from STAT1-GOF patient 7 days after nucleofection with optimized RNP was assessed by ddPCR (n = 2 technical replicates) in two independent experiments, indicated in light and dark blue. The ddPCR readouts in (B)–(D) are reported as twice the measured value as the mutation is heterozygous and uncorrected allele is present at the time of assessment. (E) The cells obtained on 7 days post-nucleofection from the second experiment in (D) were also stimulated with IFN-α, followed by assessment of phosphorylated STAT1 levels in CD3⁺ cells using flow cytometry. Mock electroporated patient T cells and healthy donor T cells that did not receive CRISPR RNPs were used as controls. The dotted line and the solid line show unstimulated and stimulated samples, respectively.