T cell editing
Whole blood was obtained with donor consent from the Israeli Blood Bank (Magen David Adom, Sheiba Medical Center) in accordance with Tel Aviv University Review Board. PBMCs were extracted using Lymphocyte Separation Medium (mpbio) and cryopreserved until subsequent use. Following thawing, cells were activated for 24-48hrs with 1ug/ml anti-human CD28 (biogems) and anti-human CD3 (biogems). Cells were cultured in MEM-Alpha (Biological Industries) supplemented with 10% Heat Inactivated FCS (Sigma), 50IU rhIL-2 (Peprotech) and P/S. For RNP electroporation, 18pmol of Alt-R spCas9 Nuclease V3 (IDT) and 66pmol of Alt-R CRISPR-Cas9 gRNA (IDT) per 1E6 cells per target were complexed in Buffer T (IDT). Experiments involving a triple targets (TRAC, TRBC, PD-1), quantities of Cas9:gRNA complexes were tripled for the non-specific control. gRNA sequences can be found in Supplementary Table 4. Cells were harvested, washed and electroporated at 1600v, 10ms, 3pulses in 10ul Buffer T and subsequently grown in culture media devoid of P/S.
Cells were harvested, washed, and resuspended in Cell Staining Buffer (Biolegend) containing 1/100 diluted anti-human CD3 or anti-human TCRα/β (Biolegend), both targeting the TCR complex. Staining was performed for 15mins at room temperature in the dark. Finally, cells were washed and data acquisition was performed on an Attune NxT Flow Cytometer (life Technologies).
Single-cell RNA sequencing
Edited T cells were harvested and live cells were purified using Lymphocyte Separation Medium (mpbio), washed and resuspended in PBS supplemented with 0.5% BSA to achieve optimal concentration of approximately 1000 cells per microliter. Cells were counted and viability assessed manually in Trypan Blue 0.4% (Biological Industries). 17,000 cells were loaded on Next GEM Chip (10x Genomics). Libraries were prepared at the Single-Cell Genomics Core, Faculty of Medicine, Tel Aviv University, using the 10x Genomics Chromium Controller in conjunction with the single-cell 3⍰ v3.1 kit, protocol revision D. The cDNA synthesis, barcoding, and library preparation were then carried out according to the manufacturer’s instructions. Briefly, cDNA amplification was performed for 11 cycles. Sample index PCR was performed for 13 cycles using Chromium i7 Sample Indices. Resulting libraries were quantified and analyzed by Qubit and Tapestation. Libraries were sequenced on the NextSeq 500 platform (Illumina), following the manufacturer’s protocol, using a NextSeq 500/550 High Output Kit v2.5 (75 Cycles) kit (Illumina). Sequencing was performed at the Genomics Research Unit, at the Life Sciences Inter-Departmental Research Facility Unit, Tel-Aviv University.
scRNA-seq gene expression pre-processing
Raw BCL files for the DNA sequencing data were processed using Cellranger DNA (version 5.0.1). Data were aligned to the 10X Genomics GRCh38 genome. Results were visualized in the Loupe scDNA Browser (version 5.0.0). Raw gene expression data were extracted from the Seurat object as recommended in the “Using 10x data” section (inferCNV of the Trinity CTAT Project, https://github.com/broadinstitute/inferCNV).
InferCNV was used to infer copy number changes from the gene expression profiles26,27 The non-targeted T cell population was used as the reference, and the CRISPR/Cas9-targeted population was tested, with the following parameters: “denoise”, default hidden markov model (HMM) settings, and a value of 0.1 for “cutoff”.
Identification of aberrant cells
For each cell, the mean of inferCNV scores was calculated across genes and plotted. The PDCD1 gene resides near the chromosome 2 q-arm telomere and only 15 genes, expressed in T cells, reside between the TCRα gene and the chromosome 14 centromere. Therefore, for chromosome 14 and chromosome 2, all of the expressed genes on the respective chromosomes were used for the analysis. For chromosome 7, the 47 expressed genes that reside distal to TCRβ were used. Cells with a mean lower or higher >2 standard deviations from the mean of the population were determined as cells with a loss or a gain, respectively.
Simulation analysis for downregulated genes
In order to determine whether cells categorized as harboring a chr14 loss of a chr7 distal loss, also had a significant increase in the fraction of zero expression calls (that is, whether these regions are enriched with genes not detected at all by scRNAseq), the ratio between the number of genes from each chromosome with expression = 0 and expression > 0 for each cell population (loss vs. non-loss) was calculated. The fold change of the proportion of zeros calls between the normal and aberrant cells was determined, and 10,000 simulations were then performed, selecting an equivalent number of random genes from other chromosomes. An empirical p-value was determined by comparing the empirical values to the simulated values.
Differential gene expression analysis
The “FindMarkers” package from the Seurat library43 was used to detect the differentially expressed (DE) genes between two groups of cells. The function receives two identities of clusters in the data set and a value for the minimum percentage that is required for a feature to be detected in either of the two groups of cells. The minimum percentage value that we used is 0.25.
The first comparison was made to detect DE genes between the cells that have undergone loss in the TRAC gRNA and the cells that did not show loss or gain in the same treatment group. The second was to detect DE genes between the cell that have undergone gain in the TRAC gRNA and the cells that did not show gain or loss in the same treatment group.
Gene Set Enrichment Analysis
The lists of differentially expressed genes between each two conditions were determined using Seurat as described above. These lists were subjected to gene set enrichment analysis (GSEA) using the GSEA-MSigDB portal (https://www.gsea-msigdb.org/gsea/msigdb/). The analysis was run using the following curated gene sets: “Hallmark”, “KEGG”, “GO biological process” and “positional” gene sets from MSigDB44–46.
For digital droplet PCR, whole genomic DNA was extracted from cells using Gentra PureGene Tissue Kit (Qiagen). In order to remove sheared genomic fragments, resulting eluates were further purified using AmpureXP beads (Beckman Coulter) at a 0.5:1 ratio. DNA fragmentation by digestion was performed in reaction, using 66ng of purified genomic DNA and 10U HindIII-HF (NEB) in ddPCR Supermix for Probes (BioRad). Thermo-cycling reaction was performed as per manufacturer recommendation. Sequences for the primers and probes can be found in Supplementary Table 4. Reactions were performed using a QX200 Droplet Digital PCR System (Bio-Rad). To analyze for dislinkage, we used the following equation as per the resulting Quantasoft (BioRad) Linkage (Linkage) and Concentration (CHEX, CFAM) values:
For dislinkage followup, multiple electroporations of treated cells were pooled and then divided in separate wells for collection at each time point. Cells were seeded at 1E6 cells/ml in MEM-Alpha (Biological Industries) supplemented with 10% Heat Inactivated FCS (Sigma) and 50IU rhIL-2 (Peprotech).
Nucleic acid manipulations
For T7E1 assays, PCR amplification was performed on Gentra PureGene Tissue Kit (Qiagen) extracted genomic DNA. 200-500ng of genomic DNA was amplified using PrimeStar MAX (Takara) for 35 cycles. Primers for these reactions can be found in Supplementary Table 4. Resulting amplicons were denatured and reannealed in a thermocycler before nuclease reaction using T7 Endonuclease 1 (New England Biolabs) at 37C for 20min. Resulting fragments were analyzed by agarose gel electrophoresis and quantified using Biovision (Vilber Lourmat) using a rolling ball for background subtraction. Efficiency was calculated using the following equation:
For TIDE analysis, PCR amplicons were subjected to purification by AmpureXP beads (Beckman Coulter) at a 1:1 ratio. Sanger sequencing was performed at the DNA Sequencing Unit, Tel Aviv University. Samples were compared using TIDE (https://tide.nki.nl/)47.
Fluorescence in-situ hybridization
Fluorescence in situ hybridization (FISH) analysis was performed following the manufacturer’s instructions (Cytocell) on interphase human T cells from peripheral blood spreads’ using the TRACD breakapart probe. Images were captured using GenASIs imaging system.
For FISH, genomic DNA cleavage efficiency and flow cytometry knock out efficiency, statistical analyses were performed on Prism (GraphPad). For t-tests on dislinkage by ddPCR, for each donor, technical replicates were averaged and t-test was performed on the averaged values. Each figure legend denotes the statistic used, central tendency and error bars.
Multiple ongoing clinical trials use site-specific nucleases to disrupt T cell receptor (TCR) genes in order to allow for allogeneic T cell therapy1–5. In particular, the first U.S. clinical trial using CRISPR-Cas9 entailed the targeted disruption of the TCR chains and programmed cell death protein 1 (PDCD1) in T cells of refractory cancer patients6. Here, we used the same guide RNA sequences and applied single-cell RNA sequencing (scRNAseq) to more than 7000 primary human T cells, transfected with CRISPR-Cas9. Four days post-transfection, we found a loss of chromosome 14, harboring the TCRα locus, in up to 9% of the cells, and a chromosome 14 gain in up to 1.4% of the cells. We further identified truncations of chromosome 7, harboring the TCRβ locus, in 9.9% of the cells. Loss of heterozygosity (LOH) was further validated using fluorescence in situ hybridization (FISH) and the temporal dynamics of cleavage and incomplete repair were monitored using digital droplet PCR (ddPCR). Aneuploidy was found among all T cell subsets and was associated with transcriptional signatures of reduced proliferation and metabolism as well as with induced p53 activation and cell death. We conclude that aneuploidy and chromosomal truncations are frequent outcomes of CRISPR-Cas9 cleavage in clinical protocols. Monitoring and minimizing these aberrant products is crucial for future applications of genome editing in T cell engineering and beyond.