Efficient nonviral integration of large transgenes into human T cells using Cas9-CLIPT

Abstract

CRISPR-Cas9 ribonucleoproteins (RNPs) combined with a nucleic acid template encoding a chimeric antigen receptor (CAR) transgene can edit human cells to produce CAR T cells with precise CAR insertion at a single locus. However, many human cells have adverse innate immune responses to foreign nucleic acids, particularly circular double-stranded DNA (dsDNA). Here, we introduce Cleaved, LInearized with Protein Template (Cas9-CLIPT), a circular plasmid containing a single target sequence for the Cas9 RNP, such that during manufacturing, Cas9-RNP binds and cleaves the plasmid to linearize the dsDNA in vitro. Cas9-RNP remains bound to the linearized template and is delivered to cells to promote precise knock-in via homology-directed repair with Cas9-CLIPT. Cas9-CLIPT Nanoplasmids generate up to 1.7-fold higher rates of precise knock-in relative to linearized dsDNA, reaching efficiencies up to 60% with non-homologous end joining inhibition. Cas9-CLIPT-manufactured GD2 TRAC-CAR T cells are potent against GD2⁺ neuroblastoma cells and exhibit an enriched stem cell memory phenotype. On several electroporation instruments and approaching clinically relevant yields, we successfully manufactured TRAC-CAR T cells using Cas9-CLIPT plasmids containing large (2-6 kb) transgenes. Cas9-CLIPT strategies have the potential to simplify donor template production and integrate large transgenes, allowing for more efficient nonviral manufacturing of multifunctional, genome-edited immune cell therapies.

Keywords: CAR T cells; CRISPR; cancer; gene editing; pre-clinical.

Graphical abstract

Figure 1

Robust manufacturing of TRAC-CAR T cells with Cas9-CLIPT (A) An SpCas9 RNP complex targeted to exon 1 of the TRAC locus was used to insert dsDNA HDR donor templates containing a GD2 CAR transgene under the control of the endogenous TRAC promoter: Cas9-CLIPT and Control template. Cas9-CLIPT contains a TRAC gRNA site with PAM to be linearized by the RNP with TRAC sgRNA, while the Control template was linearized via an SspI restriction site. (B) Cas9-CLIPT or Control DNA (±PGA) were incubated with RNP for 10 min and run on a 1% agarose gel (lanes 3–6) with a 1-kb plus ladder (lane 1) and controls (lanes 7–12). (C) Isolated T cells were activated with anti-CD3/CD28 beads in TexMACS for 3 days; electroporated with RNP, PGA, and Nanoplasmid donor templates; recovered in ImmunoCult-XF supplemented with IL-7/IL-15 and M3814; and expanded in ImmunoCult-XF supplemented with IL-7/IL-15 until day 10 to manufacture TRAC-CAR T cells. (D) Representative contour plots of Cas9-CLIPT, Control TRAC-CAR T cells, or non-transfected samples analyzed for CAR and TCR expression on day 8 (day 5 post-EP). (E and F) Bar graphs of Cas9-CLIPT or Control TRAC-CAR T cell (E) CAR or TCR expression on days 8 and 10 of manufacturing and (F) fold expansion and viability on day 10 with controls RNP only (T cells electroporated with RNP but not Cas9-CLIPT) and No electroporation (EP) (T cells that were treated with Cas9-CLIPT but were not electroporated). (G) CAR and TCR expression of Cas9-CLIPT and Control TRAC-CAR T cells manufactured at small- (5e6 cells) and large- (50e6 cells) scales on the Cellares EP device, without treatment of PGA or M3814. CAR, chimeric antigen receptor; Cas9-CLIPT, cleaved, linearized with protein template; dsDNA, double-stranded DNA; RNP, ribonucleoprotein; TRAC, T cell receptor alpha constant. N_Cas9-CLIPT = N_Control = 6 (3 donors), N_{RNP Only} = N_{No EP} = 6 (3 donors), N_{Cas9-CLIPT SmallCEP} = N_{Cas9-CLIPT Large CEP} = 2, N_{Control Small CEP} = 1 (3 separate donors). Error bars represent mean and standard deviation. Statistical significance was determined with paired t-tests; ∗∗∗p < 0.001; ∗∗∗∗p < 0.0001.

Figure 2

Long-read sequencing of on-target TRAC edits in Cas9-CLIPT CAR T cells Genomic DNA was extracted on day 10 of manufacturing from Cas9-CLIPT CAR T cells, Control TRAC-CAR T cells, and non-transfected T cells. Regions surrounding the cut site were amplified via PCR, and amplicons were sequenced on the PromethION 24. (A) Indel editing efficiency in genomic DNA isolated from the CAR T products as measured by long-read sequencing. The modification frequency of each nucleotide around the TRAC cut site was calculated relative to the human genome TRAC reference sequence. The protospacer (P) and PAM are underlined, and a vertical dotted line indicates where Cas9 nuclease should create a dsDNA DSB. (B) Alignment of long-reads from sequencing the genomic DNA of Cas9-CLIPT CAR T cells from donor A to the expected transgene amplicon sequence (3,152 bp). Gray histogram indicates the relative number of reads at each base pair aligned to the insert and homology arms. Red indicates the negative strand; blue indicates the positive strand. (C) Allelic knock-in efficiency as measured by long-read sequencing. Knock-in efficiency is defined as the number of reads aligned to the transgene divided by the total number of reads aligned to both the transgene and the reference genome. DSB, double-stranded break. Three donors, N_Cas9-CLIPT = N_Control = N_{Non-transfected} = 6. Error bars represent mean and standard deviation. Statistical significance was calculated with paired t-tests; ∗p < 0.05; ∗∗∗∗p < 0.0001.

Figure 3

Whole-genome sequencing of Cas9-CLIPT TRAC-CAR T cells indicates minimal off-target effects Genomic DNA was extracted from Cas9-CLIPT TRAC-CAR T cells and sequenced via Oxford Nanopore Technologies Sequencing for unbiased whole-genome sequencing (WGS) to determine off-target sites. (A) Schematic of unbiased WGS alignment. Sequencing reads were aligned to the transgene, and hits were then mapped to the human genome. After filtering with an overlap window threshold, reads that aligned to both the transgene and human genome were considered an off-target integration site. (B) Percentage of reads that were considered an off-target integration site on each chromosome on a log scale. Chromosome 14 contains the on-target TRAC locus for Cas9-CLIPT TRAC-CAR T cells. (C) The number of integration site reads on each chromosome, and the corresponding average alignment length of the read to the transgene (14g2a) and human genome (chromosomal). The average MAPQ of each alignment is also reported. One donor, N_Cas9-CLIPT = 1.

Figure 4

Cas9-CLIPT TRAC-CAR T cells are stem cell memory-like and potent against solid tumor cells in vitro (A) Cas9-CLIPT TRAC-CAR T cells were analyzed for expression of stem cell memory T cell surface markers by spectral flow cytometry on day 7 post-EP. (B) Representative contour plots of CD62L vs. CCR7 expression in CD45RA ± populations of Cas9-CLIPT TRAC-CAR T cells. (C) Percentage of naive (CD45RA⁺/CD62L⁺/CCR7⁺), naive-central memory (CD45RA⁺/CD62L⁺/CCR7⁻), central memory (CD45RA⁻/CD62L⁺/CCR7⁺), central-effector memory (CD45RA⁻/CD62L⁺/CCR7⁻), effector memory (CD45RA⁻/CD62L⁻/CCR7⁻), and terminal effector (CD45RA⁺/CD62L⁻/CCR7⁻) T cells. (D) GD2⁺ neuroblastoma CHLA-20 cells were plated in 96-well plates 24 h before TRAC-CAR T cell addition. The potency was measured continuously for up to 72 h. (E) Percent change in GFP fluorescence from GD2⁺ CHLA-20 neuroblastoma cells vs. time in cancer/Cas9-CLIPT or Control TRAC-CAR T cell co-cultures for a 1.25:1 E:T ratio. (F) Percent cytotoxicity at 72 h for cancer/Cas9-CLIPT and Control TRAC-CAR T cells at an E:T ratio of 1.25:1 compared to a cancer only control. GFP, green fluorescent protein; E:T, effector:target ratio. Three donors, N_Cas9-CLIPT = N_Control = 6. Error bars represent mean and standard deviation. Statistical significance was determined with one-way ANOVA; ∗p < 0.05; ∗∗p < 0.01.

Figure 5

Large knock-in using Cas9-CLIPT enables fluorescent reporter of CAR activation (A) Schematic of Cas9-CLIPT NFAT-mCh DNA containing a GD2 CAR transgene under the control of the endogenous TRAC promoter and an mCherry reporter gene conditionally expressed with NFAT binding domains and the minimal IL-2 promoter targeted to the TRAC locus. Cas9-CLIPT templates contain a TRAC gRNA site with PAM to be linearized by the RNP. (B) NFAT-mCh CAR T cells contain an inducible mCherry reporter gene dependent on CAR activation. A representative contour plot and bar graph depict NFAT-mCh CAR T cell TCR and CAR expression. (C) One million NFAT-mCh CAR T cells were stimulated with either PMA/ionomycin or 200,000 CHLA-20 neuroblastoma cells for 24 h. Stimulated and unstimulated Cas9-CLIPT NFAT-mCh or Cas9-CLIPT TRAC-CAR T cell samples were then assayed for NFAT-driven mCh expression. Representative contour plots detail CAR vs. mCh expression. (D) Bar graphs of percentage of CAR⁺/mCh High populations and mCh MFI. MFI, mean fluorescence intensity. Three donors, N_Cas9-CLIPT = N_NFAT-mCh = 6. Error bars represent mean and standard deviation. Statistical significance was determined with one-way ANOVA; ∗∗∗∗p < 0.0001.

Figure 6

Long-read sequencing of on-target TRAC edits in potent NFAT-mCh CAR T cells Genomic DNA was extracted on day 10 of manufacturing from Cas9-CLIPT, NFAT-mCh, and Control TRAC-CAR T cells, as well as non-transfected T cells. Regions surrounding the cut site were amplified via PCR, and amplicons were sequenced on the PromethION 24. (A) Indel editing efficiency in genomic DNA isolated from the CAR T products as measured by long-read sequencing. The modification frequency of each nucleotide around the TRAC cut site was calculated relative to the human genome TRAC reference sequence. The protospacer (P) and PAM are underlined, and a vertical dotted line indicates where Cas9 nuclease should create a dsDNA DSB. (B) Allelic knock-in efficiency as measured by long-read sequencing. Knock-in efficiency is defined as the number of reads aligned to the transgene divided by the total number of reads aligned to both the transgene and the reference genome. (C) GD2⁺ neuroblastoma GFP⁺ CHLA-20 cells were plated in 96-well plates 24 h before TRAC-CAR T cell addition. GFP expression was measured continuously for up to 72 h. (D) Percent change in GFP fluorescence from GD2⁺ CHLA-20 neuroblastoma cells vs. time in cancer/NFAT-mCh CAR T cell co-cultures for E:T ratios of 10:1, 5:1, 2.5:1, or 1.25:1 and (E) bar graph of extent of cytotoxicity at 72 h for 1.25:1 E:T ratios. Error bars represent mean and standard deviation. Three donors, N_{Cas9-CLIPT NFAT-mCh} = 6. Error bars represent mean and standard deviation. Statistical significance was determined with paired t-tests; ∗p < 0.05; ∗∗p < 0.01; ∗∗∗∗p < 0.0001.