Mitigation of chromosome loss in clinical CRISPR-Cas9-engineered T cells

Abstract

CRISPR-Cas9 genome editing has enabled advanced T cell therapies, but occasional loss of the targeted chromosome remains a safety concern. To investigate whether Cas9-induced chromosome loss is a universal phenomenon and evaluate its clinical significance, we conducted a systematic analysis in primary human T cells. Arrayed and pooled CRISPR screens revealed that chromosome loss was generalizable across the genome and resulted in partial and entire loss of the chromosome, including in pre-clinical chimeric antigen receptor T cells. T cells with chromosome loss persisted for weeks in culture, implying the potential to interfere with clinical use. A modified cell manufacturing process, employed in our first-in-human clinical trial of Cas9-engineered T cells, ¹ dramatically reduced chromosome loss while largely preserving genome editing efficacy. Expression of p53 correlated with protection from chromosome loss observed in this protocol, suggesting both a mechanism and strategy for T cell engineering that mitigates this genotoxicity in the clinic.

Figure 1:. CRISPR-Cas9 genome editing of TRAC results in whole and partial chromosome loss

(A) Cas9 gRNA target sites tiled across the first exon of TRAC on chromosome 14. (B) Gene dosage from transcriptome-wide scRNA-seq of T cells treated with Cas9 and a non-targeting gRNA (top heatmap) or TRAC-targeting gRNA (bottom heatmap). Each individual row corresponds to a single cell and each column corresponds to a specific gene and its genomic position, grouped by chromosome (outlined in black). Red represents increase in gene dosage while blue represents decrease in gene dosage. Rows outlined in black represent cells treated with different TRAC-targeting gRNAs. Blue arrows highlight chromosome 14, where TRAC is located. (C) Distribution of computationally predicted chromosome 14 breakpoints in cells predicted to have a chromosomal loss event. The distribution is an aggregate of 11 different TRAC-targeting gRNAs (all within ~300 bp) in cells with partial chromosome loss. (D) Representative single cell chromosome 14 gene dosage plots illustrating a cell with no chromosome loss (left), whole chromosome loss (middle), or partial chromosome loss (right). Gene dosage was normalized to non-targeting samples. Gray shaded area (gene dosage of 0.95–1.05) represents normal gene dosage (2 copies). Blue shaded area (gene dosage of <0.95) represents reduction in gene dosage (1 copy). Dotted lines represent the centromere, black lines represent the Cas9 target site, and the red line represents the computationally predicted breakpoint. (E) Quantification of whole and partial chromosome 14 loss across all gRNAs from scRNA-seq. NT indicates non-targeting gRNA. (F) Schematic of ddPCR assay to measure chromosome loss. The yellow lightning bolt represents the Cas9 target site. The detection of both HEX and FAM probes indicates no DSB or repaired DSB (top illustration). The detection of the HEX probe but not the FAM probe indicates an unresolved DSB that represents chromosome loss (bottom illustration). (G) Quantification of chromosome loss at the Cas9 target site across all gRNAs from ddPCR (n = 3, n = 2 biological donors). Numbers next to each point represent the TRAC-targeting gRNA. NT indicates non-targeting gRNA and represents samples from four different ddPCR amplicons. Error bars represent the standard deviation from the mean. Dashed line represents linear regression line of best fit and shaded region represent 95% confidence intervals (Slope = 1.082, R² = 0.9853). (H) Comparison of chromosome 14 loss between TRAC-targeting conditions where the PAM is distal (Centromere-PAM distal) or proximal (Centromere-PAM proximal) to the centromere relative to the gRNA spacer sequence. Chromosome 14 loss was measured by scRNA-seq (n = 1 biological donor) or ddPCR (n = 3, n = 2 biological donors). P-values are from Welch’s unpaired t-tests and are from left to right 0.8689 and 0.7338. ns = not significant.

Figure 2:. Genome-scale CRISPR-Cas9 screen reveals target-specific chromosome loss

(A) Workflow for a CRISPR-Cas9 screen to estimate chromosome loss in T cells. Primary human T cells were transduced with a CROP-seq lentiviral library expressing one of 384 gRNAs. Cells were then nucleofected with Cas9 protein, before GFP+ cells (co-expressed on the CROP-seq gRNA vector) were enriched via fluorescence-activated cell sorting. Enriched cells were subject to scRNA-seq and downstream analysis. (B) Quantification of targeted chromosome loss enrichment for each target gene. Each of the 92 bars represents the combination of four unique gRNAs targeting the same gene. Chromosome loss enrichment was calculated relative to the baseline loss per chromosome in cells containing a gRNA targeting a different chromosome. Error bars represent 95% confidence intervals. (C) Chromosomal loss enrichment at each somatic chromosome across all gRNAs. Rows represent the chromosome targeted by the Cas9 gRNA. Columns represent the chromosome analyzed for chromosome loss. (D) Chromosome loss measured by ddPCR at 15 different Cas9 target sites across the genome. Rows titles indicate the identity of the gRNA used. Column titles indicate the site in the genome that was analyzed via ddPCR. Heatmap values represent the mean of replicates (n = 3, except n = 2 for B2M target column). (E) Correlation between chromosome loss from 25 gRNAs as measured by scRNA-seq and ddPCR. Spearman’s correlation = 0.59, **P = 0.0017 (two-tailed).

Figure 3:. Genetic and epigenetic factors influence Cas9-induced chromosome loss

(A) Heatmap of differentially expressed genes in cells with chromosome loss compared to cells without chromosome loss. Cells with chromosome loss were divided into 22 groups depending on which somatic chromosome was lost (rows), and differentially expressed genes were individually investigated (columns). Upregulated genes are shown in red while downregulated genes are shown in blue. Genes were given a score of 1 (upregulated), −1 (downregulated), or 0 (no difference) for each chromosome loss group. Summed gene scores across all chromosome loss groups is shown below; genes with a score >|13| are displayed. (B) Gene ontology analysis based on differential gene expression. The most significantly upregulated modules are displayed. (C) Cell cycle analysis based on expression profiles. The percentage of cells in each cell cycle phase were quantified for cells with no chromosome loss or cells with chromosome loss. (D) Influence of epigenetic marks on chromosome loss. The gRNA sequence for cells with or without chromosome loss was analyzed for localization within ± 75 bp of an epigenetic marker peak. P-values were calculated using a two-sided Fisher’s Exact Test and are from left to right (ATAC) 0.365496, (H3K36me3) 0.789824, and (H3K9me3) 0.305706. ns = not significant.

Figure 4:. Cas9-induced chromosome loss persists for weeks but results in reduced fitness and proliferation

(A) ddPCR measurements of chromosome loss at the Cas9 TRAC target site over 14 days. Error bars represent the standard deviation from the mean (n = 3). Day 3 results were additional used as the donor 2 (female) results shown in Fig. 1g. (B) ddPCR measurements of chromosome loss for 15 different gRNAs targeted to sites across the genome over 14 days. Error bars represent the standard deviation from the mean (n = 3, except n = 2 for B2M). Day 3 results were additionally used for the diagonal values in the heatmaps of Fig. 2d. (C) Measurement of chromosome loss across T cells of varying proliferative capacity. T cells were stained with CellTrace Violet (CTV) and cultured for five days before sorting the top and bottom quartile (top panel). ddPCR was used to measure chromosome loss in lowly proliferative (CTV high) and highly proliferative (CTV low) populations (bottom panel). NT = non-targeting gRNA. Non-targeted samples evaluated for chromosome loss at the gRNA 2 or gRNA 8 amplicon were combined into a single column (n = 3 for each of the two different ddPCR amplicons). P-values are from Welch’s unpaired t-tests and from left to right are 0.002970, 0.002970, and 0.275572.

Figure 5:. Pre-clinical CAR T cell production via homology-directed repair results in chromosome loss

(A) Strategy to generate CAR T cells via HDR with Cas9. AAV6 encoding a 1928ζ CAR between left and right homology arms (LHA and RHA, respectively) serves as a template for HDR after Cas9 cleavage (yellow lightning bolt) of TRAC. (B) Three potential genomic outcomes after Cas9 HDR: indels that disrupt TCR expression (top), insertion of the CAR transgene that simultaneously disrupts TCR expression (middle), and chromosome loss that disrupts TCR expression (bottom). (C) Quantification of chromosome 14 loss enrichment across two TRAC-targeting gRNAs with or without an AAV HDR template from scRNA-seq (n = 2 biological donors). Two separate batches of CAR T cells were manufactured, before being subjected to scRNA-seq four or seven days after generation. Chromosome 14 loss enrichment was calculated relative to T cells treated with Cas9 and a non-targeting gRNA. (D) Chromosome 14 loss enrichment over time, normalized to Cas9 editing efficacy (n = 2 biological donors). Editing efficacy was determined by the percentage of TCR negative cells as measured via flow cytometry (see Extended Data Fig. 13c).

Figure 6:. Clinical CRISPR-Cas9 genome editing protocol in patient T cells mitigates chromosome loss

(A) Strategy to investigate chromosome loss in two clinical trial patients with CRISPR-edited T cells. Two patients with refractory cancer had T cells isolated, nucleofected with TRAC, TRBC, and PDCD1-targeting Cas9 RNPs, and transduced with a lentivirus encoding an NY-ESO-1 TCR (NYCE cells). Cells were subjected to scRNA-seq prior to infusion (Day 0) and as well as at different time points post-infusion (Days 10, 28, and/or 113). (B) Chromosome loss enrichment on chromosome 14 (TRAC), chromosome 7 (TRBC), or chromosome 2 (PDCD1) at different timepoints for both patients. Enrichment was calculated relative to non-targeted chromosomes (all chromosomes but 2, 7, and 14). Day 0 represents NYCE cells prior to infusion, while other later timepoints represent NYCE cells that were collected after circulation in vivo. (C) Diagram of two different protocols for Cas9 genome editing of primary human T cells. The laboratory protocol (top) consisted of activating/stimulating cells prior to Cas9 nucleofection, and was used throughout this study. The clinical protocol (bottom) consisted of nucleofecting cells with Cas9 prior to activating/stimulating and is representative of our clinical trial. (D) Relative chromosome loss with 11 different TRAC-targeting gRNAs using the laboratory or clinical protocol in primary human T cells. Chromosome loss was normalized to the indel efficacy (see Extended Data Fig. 14b). P-values are from Welch’s unpaired t-tests and from left to right are 0.008320, 0.111695, 0.000052, 0.000076, 0.000159, 0.000073, 0.000125, 0.000087, 0.000073, 0.000050, and 0.001416. (E) Fold TP53 mRNA expression during the laboratory or clinical protocols for Cas9 genome editing of primary human T cells (n = 5 biological donors). Data points are the mean of two technical replicates. X-axis letters correspond to timepoints in Fig. 6c. “Cas9 DSBs” represents the timepoints in the laboratory or clinical protocols where Cas9 was nucleofected into T cells to generate DSBs.