Detection of chromosomal alteration after infusion of gene-edited allogeneic CAR T cells.
Free full text in Europe PMC
Abstract
Free full text
Detection of chromosomal alteration after infusion of gene-edited allogeneic CAR T cells
Abstract
A chromosome 14 inversion was found in a patient who developed bone marrow aplasia following treatment with allogeneic chimeric antigen receptor (CAR) Tcells containing gene edits made with transcription activator-like effector nucleases (TALEN). TALEN editing sites were not involved at either breakpoint. Recombination signal sequences (RSSs) were found suggesting recombination-activating gene (RAG)-mediated activity. The inversion represented a dominant clone detected in the context of decreasing absolute CAR Tcell and overall lymphocyte counts. The inversion was not associated with clinical consequences and wasnot detected in the drug product administered to this patient or in any drug product used in this or other trials using the same manufacturing processes. Neither was the inversion detected in this patient at earlier time points or in any other patient enrolled in this or other trials treated with this or other product lots. This case illustrates that spontaneous, possibly RAG-mediated, recombination events unrelated to gene editing can occur in adoptive cell therapy studies, emphasizes the need for ruling out off-target gene editing sites, and illustrates that other processes, such as spontaneous V(D)J recombination, can lead to chromosomal alterations in infused cells independent of gene editing.
Introduction
Autologous chimeric antigen receptor (CAR) T cell therapy has shown impressive activity clinically and has changed the treatment landscape for B cell malignancies.1,2 Despite these impressive advances, autologous CAR T have several drawbacks that limit their activity and clinical deployment. Patient T cell health can be affected by previous lines of therapy, by underlying disease, and by patient age.3,4,5,6,7 Due to the time necessary to manufacture products, many patients progress before their cell product is ready. Therapy is dependent on success of T cell production and, in the event of manufacturing failure, no treatment is possible. In registrational trials, up to 18% of patients with ALL and 33% of patients with diffuse large B cell lymphoma (DLBCL) did not receive their therapy for these two reasons alone.8,9,10 In addition to these obstacles, complex supply chain logistics limit treatment to cell therapy centers of excellence, and the fact that each patient requires individualized good manufacturing practice (GMP) product limits supply. For these reasons, an off-the-shelf or allogeneic CAR T therapy made from healthy donor T cells holds great promise.
In order to allow cells from an unrelated host to be used without causing graft-versus-host disease (GvHD), allogeneic CAR T are modified to prevent expression of the T cell receptor (TCR) complex by knocking out the TRAC gene.11,12 To generate ALLO-501A, T cells were transduced with a lentiviral vector (LVV), to express the CD19 CAR, and gene edited with TALEN technology11 at the TRAC locus on chromosome 14 to prevent TCR expression to avoid GvHD, and at the CD52 locus on chromosome 1 to allow lymphodepletion with an anti-CD52 antibody. The present study describes ALLO-501A as a treatment for patients with relapsed/refractory non-Hodgkin’s lymphoma. ALLO-501A has demonstrated similar efficacy and safety to autologous CAR T cells13,14 in Phase I studies and is being studied in a pivotal trial in large B cell lymphoma. During follow-up of a cytopenic patient who had been treated with ALLO-501A in Phase I, a chromosomal rearrangement was detected in the infused cells. The rearrangement was found to be a chromosome 14 inversion, warranting investigation since TRAC disruption by TALEN was conducted on chromosome 14. This finding caused a clinical hold on the ALLO-501A trial and all other trials with alternative ALLO CAR T cells using a similar manufacturing process to establish the cause and clinical significance of the inversion. Investigation revealed that the inversion was not present at the site of TALEN gene editing and was instead most likely caused by RAG-mediated recombination, an event occurring post-thymically in normal T cells. Cytopenias preceded the presence of the inversion, no clinical significance was attributed to the inversion, and the clinical hold was released after characterization of the breakpoints and sequence analysis of patient and product samples.
Results
Patient presentation
A female patient with treatment-refractory transformed follicular lymphoma who was unable to have cells manufactured for an autologous CAR T cell product was enrolled in an allogeneic anti-CD19 CAR T cell study (NCT04416984: ALPHA-2; safety and efficacy of ALLO-501A anti-CD19 allogeneic CAR T cells in adults with relapsed/refractory large B-cell lymphoma). The CAR T cells were gene edited at the TRAC locus on chromosome 14 to prevent TCR expression to avoid GvHD, and at the CD52 locus on chromosome 1 to allow lymphodepletion with an anti-CD52 antibody.
Following lymphodepletion with fludarabine, cyclophosphamide, and an anti-CD52 antibody, the patient received one infusion of CAR T cells from a male donor on day 0 with a resulting partial response and another dose of CAR T cells on day 36 (planned as part of a consolidation regimen) from a second male donor without additional lymphodepletion. Prior to the second infusion, the patient underwent bone marrow biopsy on day 26 due to cytopenias, a common side effect observed with CAR T cell therapy.15,16 The patient was diagnosed with human herpesvirus 6 (HHV-6) reactivation on day 30, potentially contributing to the decrease in cell counts, and underwent anti-viral therapy.
Trilineage hematopoiesis with a cellularity of 30% was observed in the biopsy with some normal female lymphocytes in addition to male donor lymphocytes. CAR T cell expansion was observed following the first infusion, peaking at day 36, followed by contraction and a second comparable peak following the second dose of CAR T cells, again followed by subsequent secondary contraction (Figure 1). This profile is expected and typical of CAR T cell expansion/contraction observed in other patients treated with similar cell products.
Longitudinal measurement of CAR T cell VCN, ALC, and TCRβ clones in the patient
The primary y axis (left) captures CAR T cells measured by VCN (teal line), and the secondary y axis (right) captures levels of ALC (purple line). (Bottom) The five most abundant TCRβ clones at day 56. The VCN rose after the first infusion at day 0, peaked first at day 36, and then again at day 43. The rearrangement was first detected at day 47, in the context of falling VCN and ALC. The proportion of cells positive for the rearrangement by FISH rose through day 69 as the VCN continued to decline. The TCRβ clone detected at days 50 and 56 with frequencies of 29% and 41%, respectively, is assumed to be the clone carrying the inversion given its similarity in abundance and trajectory to the event identified by the FISH assay. The inversion junction was measured by ddPCR relative to a control gene to estimate relative abundance of the total lymphocyte population and is presented in red and shown in the third axis.
Detection of a chromosome 14 inversion in infused cells
A repeat bone marrow evaluation at day 47 showed a chromosome 14 (q11.2q32) inversion in all donor-derived T cells (19 out of the total 20 cells assayed; Figure 2). The cells carrying the chromosome 14 inversion in the peripheral blood were traced back to the first infused allogeneic drug product lot via human leukocyte antigen (HLA) and short tandem repeat (STR) multiplex genomic typing, to uniquely identify recipient and donor alleles. This showed that 98% of the T cells were from the first donor infusion, and the remaining 2% of the cells present were of host patient origin, without any cells from the second donor infusion. Repeat karyotyping was conducted on bone marrow from day 62 but failed due to inadequate mitotic activity (no metaphase cells to analyze). A peripheral blood analysis on day 69 yielded only four mitotic cells that showed the presence of the inversion. The difficulty in obtaining metaphase cells at both time points indicates that these remaining cells had a low proliferative index. Although it was not possible to determine if cells carrying the inversion expressed TCR or not, a dominant population of CD3 positive cells was not detected by flow that would correspond to the clone frequency.
Cytogenetic analysis of male donor cells detected on day 47
Left hand side: G-banded chromosome analysis shows that one copy of chromosome 14 harbors a 14q11.2q32 inversion. Image obtained at 100× magnification. Right hand side: FISH analysis demonstrating the split 5' (spectrum orange) and 3' (spectrum green) TCR A/D signals (red and green arrows), confirming that the 14q11.2q32 inversion detected by G-banding involves the TCR A/D gene, in addition to one normal chromosome 14 (yellow arrow). Image obtained at 100× magnification.
Results of a TCR A/D break-apart fluorescence in situ hybridization (FISH) assay, which does not require actively dividing cells, showed rearrangement at the TCR A/D locus in between 32% and 41% of 200 cells sampled at various time points (Table S1). At all time points, karyotype and FISH analysis findings occurred in the context of declining CAR T cells, as measured by vector copy number (VCN) analysis (Figure 1), and low peripheral blood counts, with absolute lymphocyte count (ALC) peaking at 2.07 × 103 cells/μL on day 28, declining to 1.08 × 103 cells/μL by day 44 and to 0.3 × 103 cells/μL by day 51. Although chromosome 14 (q11.2q32) inversion is a feature of T cell prolymphocytic leukemia (T-PLL), a diagnosis of T-PLL was ruled out due to absence of other diagnostic criteria, including the expression of other genetic abnormalities, the presence of a high T cell count, organ involvement, or expression of T cell leukemia/lymphoma1 protein (TCL1; Table S2).17 Additionally, the rare lymphocytes observed in the bone marrow were consistent with CAR T cells by morphology and by immunophenotype.
Clonal abundance
Clonality analysis using TCRβ sequencing showed a clone with 29% and 41% abundance at days 50 and 56, respectively (Figure 1). This clone was not detected at the previous sampling point of day 26. A total of eight clones with lesser abundances ranging from 1% to 3% were identified at day 56, and all other clones were present with an abundance of less than 1%. Given the similarity in abundance of the dominant TCRβ clone to that of the 14q11 disruption by FISH, this clone was assumed to be the one containing the inversion. This dominant TCRβ clone could not be detected in the drug product above the level of detection of the assay (Figure S1).
Sequencing of inversion sites implicated V(D)J recombination as the causative event
Short- and long-read whole-genome sequencing (WGS) of a day 61 peripheral blood sample was conducted with next-generation sequencing (NGS) to examine the precise breakpoints of the inversion and determine whether it could be a consequence of TALEN editing. Sequencing confirmed the presence of the chromosome 14 inversion and identified the breakpoint sites (Figure 3). The centromeric (14q11.2) inversion site occurred within the TCR A/D locus at the start of the T cell receptor joining gene TRAJ7 (chr14:22537626; hg38 reference genome build). The telomeric (14q32) inversion site occurred at the IGHV3-69-1 pseudogene (chr14:106728167; hg38 reference genome build) within the IGHV region.
Map of chromosome 14 inversion detected and characterized from a peripheral blood sample of the patient at day 61 by short- and long-read WGS
The two breakpoints were identified at TRAJ7 (chr14:22537626; hg38 reference genome build) and IGHV3-69-1 (chr14:106728167; hg38 reference genome build). Two additional deletions (Chr14:22281734-22537626; hg38 reference genome build and Chr14:106470263-106728167; hg38 reference genome build) were observed at the centromeric inverted junction. The relative position and approximate distances from the TRAC TALEN cut site are shown. Also marked are predicted RAG cleavage sites, as inferred from RSS sequences, at each of the two breakpoints. Other genes are added as place markers to show the impact of inversion and deletion.
The TRAJ7 site of the inversion within the TCR A/D locus is approximately 10 kb from the TRAC locus where TALEN gene editing takes place. Large deletions have been reported to occur during the gene editing process18,19,20 and may affect the formation of aberrant structural variants. Short- and long-read WGS, however, only showed the expected 49-bp deletion at the TRAC TALEN on-target site and did not show evidence of large deletions at that locus. Furthermore, the presence of consensus recombination signal sequences (RSS) sites at the inversion junctions suggests that RAG-mediated recombination, and not large deletions at the TRAC TALEN site, is the cause of the inversion. The sequences within 500 bp of both breakpoints of the chromosome 14 inversion (q11.2q32) were analyzed to identify possible off-target TALEN binding sites that could have been involved. No potential TALEN binding sites were found at levels less than, or equal to seven accumulated mismatches across both 15-bp monomer DNA-binding sites, which essentially ruled out the potential for an off-target editing event in the breakpoint regions. Likewise, no lentiviral vector integration site was detected at or near the site of inversion.
RSSs were observed at both sites of inversion. RSSs are typically composed of a conserved heptamer (consensus 5′-CACAGTG-3′) and nonamer (consensus 5′-ACAAAAACC-3′) sequence separated by either 12 (RSS-12) or 23 (RSS-23) nucleotides of variable sequence. The presence of RSS supports the hypothesis that the inversion was a consequence of RAG-mediated V(D)J recombination between the two distal sites. That is, the centromeric TRAJ7 site harbors an adjacent RSS-12 sequence (chr14:22537594-22537621; hg38 reference genome build) and telomeric IGHV3-69-1 site contains an adjacent RSS-23 sequence (chr14:106728124-106728162; hg38 reference genome build), both conforming to the 12/23 rule.21 While short-read sequencing identified the breakpoints for the inversion, subsequent long-read sequencing identified additional deletions at each of the two breakpoints, and prior to ligation. This is not unexpected, since the insertion of the signal end fragment generated by the RAG proteins has been observed to nearly always produce deletions or other rearrangements of target DNA.22
Inversion frequency over time
Sequence information from the short-read and long-read NGS analyses was used to design primer pairs to detect both ends of the inversion in order to quantitatively assess the level of inversion present in the patient sample, as well as in clinical lots and their corresponding peripheral blood mononuclear cells (PBMCs) prior to TALEN editing (Figure S2). The assay had a lower limit of detection of 1 in 1,000, which is consistent with reported estimates for droplet digital polymerase chain reaction (ddPCR) sensitivity.23 The available frozen peripheral blood samples from this patient and all other patients dosed with the drug product, products from other studies using similar manufacturing and gene editing methods, and donor material were analyzed with the ddPCR assay for the telomeric end of the inversion at time points spanning a period from prior to cell dosing up to day 61 (Figure 4). The earliest time point at which the inversion was detected was 47 days after CAR T cell infusion. The frequency of inversion was measured on a relative scale as percentage compared with the reference gene, Ribonuclease P/MRP Subunit P30 (RPP30), and ranged from 3% to 19% between day 47 and day 61 in the peripheral blood (Figure 4; Table S1). The longitudinal patient samples were also assayed by ddPCR for the centromeric inversion junction with detection at the same time points (data not shown). The time course of inversion detection is concordant with detection via karyology, FISH, and emergence of the implicated clone by TCRβ sequencing. It decreased in concordance with VCN and ALC by day 60 (Figure 1; Table S1). Samples from other patients were tested at two time points spanning dosing and ranging from lymphodepletion day −7 through 9 months, and the chromosome 14 inversion was below the limit of detection in all cases. No other drug products or heathy donor starting materials for drug products tested positive across all trials using the gene editing manufacturing platform.
Longitudinal detection of the telomeric breakpoint junction by ddPCR in patient peripheral blood samples
The x axis represents patient samples (in triplicate), and the y axis denotes the amplitude of the signal. Each dot represents an independent droplet measurement event. (Upper) The detection of the telomeric breakpoint; (lower) detection of the reference signal. The lower signal amplitude on day 61 may be due to the low quantity/concentration of DNA in this sample.
Discussion
The emergence of chromosomal aberrations is a theoretical concern following gene editing owing to the potential for off-target cleavage and translocations between any edited sites. The presence of a chromosome 14 inversion in a patient treated with gene-edited cells raised concerns about the potential for gene editing to be the causative factor during manufacturing. The event led to a US Food and Drug Administration (FDA)-imposed clinical hold of all studies using the same gene editing and manufacturing method. Herein we report that, while the source of the chromosome structural variant was the infused donor cells, the variant was only detected after product infusion and the sites of inversion did not involve TALEN-mediated gene editing. Similarly, lentiviral vector integration did not contribute to the inversion. The inversion was associated with V(D)J recombination sites, implicating aberrant RAG recombination as the causative event. Such V(D)J-related inversions have been reported previously.
The chromosome 14 inversion was initially described in T cell tumor lines24,25,26,27; however, it was later shown to be present in normal T cells from healthy individuals. Chromosome 14 inversions have been detected in post-thymic, matured T cells of normal individuals and is believed to be mediated by V(D)J recombination machinery.28,29 In one report, in the three healthy donors studied by karyotype analysis (>1,000 cells analyzed per donor), the inversion was present in all donors at an average of 0.15% of lymphocytes analyzed.30 In another study of 2,595 healthy individuals (20–40 cells per individual analyzed), 49 chromosomal breakpoints were detected involving either 14q11 or 14q32. Six of these infrequent rearranged lymphocytes out of a total of 53,580 cells analyzed showed a chromosome 14q11:q32 inversion.31 Studies examining the integrity of chromosomes in normal cultured lymphocytes have shown four frequent sites of chromosome breakage: 7p13 (near location of TCRγ chain), 7q35 (location of TCRβ chain), 14q11 (location of TCRα chain), and 14q32 (location of IGH chain) and suggest that these act as fragile sites specifically in normal lymphocytes as defined by the non-random frequency of chromosomal translocations or inversions associated with them.31,32 In addition to RAG-mediated gene rearrangement during T cell development, reactivation of RAG post-thymically, a process termed TCR revision, has been demonstrated in preclinical studies and now clinically.33
The clonal population containing the inversion arose in the context of an overall contraction of all donor cells in the patient, as seen in other CAR T cell trials.34,35 The fact that none of the T cells detected were from the second infusion was unexpected but could perhaps be explained by hostile marrow conditions that resulted in aplasia, with the first infusion product still declining and the second one not engrafting sufficiently. The inversion was not detected in donor or the clinical lots used to treat the patient, or from any other clinical batch or patient sample, leading to the conclusion that this event arose post infusion and is isolated to this particular patient. While the role of the inversion in clonal expansion is uncertain, there is no evidence that the expansion was the consequence of the inversion; similarly, since there was no expansion of CD3-positive grafted cells, a TCR-driven expansion is also not supported. Importantly, there was no evidence of clinically significant sequelae related to the inversion, and, specifically, there was no evidence of malignant transformation. After the CAR T cell treatment, the patient underwent allogeneic transplant due to prolonged pancytopenia on day 76 and showed immune reconstitution before experiencing disease relapse and starting new therapy on day 134. However, it was not possible to monitor for any possible long-term consequences of the inversion due to the patient’s death from disease progression.
With the advent of adoptive cell therapy, there has been a rapid expansion of industry and academic-led clinical programs that employ gene editing or engineering cell products, both of which carry a risk of genetic alternation such as off-site gene editing, chromosomal structural changes (deletions, inversions, or translocations), or insertional mutagenesis. This report illustrates that structural variants also occur naturally in human lymphocytes in vivo and can thereby be expected to be observed in the clinic going forward. In a recent publication, Machado et al. identified 1,037 structural variants across 635 normal human lymphocytes. Of these structural variants, 85% involved either the Ig or TCR regions. Furthermore, they go on to show that an estimated 12% of non-Ig-TCR and 84% of Ig-TCR structural variants were RAG mediated, due to the presence of RSS motif or heptamer being proximal to the breakpoint.29 Following the initial detection of a chromosome 14 rearrangement by karyotype analysis and a 14q11 disruption by FISH, the FDA placed the ALLO-501A trial and all related trials on clinical hold while the biological investigation was ongoing. The finding that the inversion was not caused by aberrant gene editing or engineering and that there was no clinical consequence, such as aberrant proliferation or leukemogenesis of engineered cells, resulted in release of the clinical hold, allowing the studies to be resumed. This event illustrates the need for investigators, sponsors, and regulators to work together to further understand the etiology, biology, frequency, and potential clinical consequences of genetic alterations in adoptive cellular therapy.
Materials and methods
Patient treatment history
The patient was a 57-year-old woman with stage IV transformed follicular lymphoma (tFL) with a t(8; 14) chromosomal translocation involving cMyc rearrangement. She was refractory to all previous therapies, had undergone radiation to mesenteric lymph nodes, and was also unable to undergo autologous CAR T cell therapy due to non-expansion of autologous CAR T cells during manufacturing.
After providing written informed consent, and in accordance with protocols approved by the institutional review board (IRB) at Medical College of Wisconsin (Milwaukee, WI), the patient received cyclophosphamide 500 mg/m2/day; fludarabine 30 mg/m2/day × 3 days on days −5, −4, and −3; and ALLO-647 (anti-CD52 antibody, Allogene Therapeutics, South San Francisco, CA) 20 mg/day × 3 days on day −2, −1, and 0, and received an ALLO-501A dose of 120 × 106 CAR+ cells on day 0. A second dose of ALLO-501A, termed consolidation, was administered on day 36 from an alternate batch without preconditioning due to recent grade 3 HHV-6 reactivation and subsequent pancytopenia. Both cell infusions were from separate male donor-derived T cells. Despite being refractory to all prior therapies, she achieved a partial response on day 28, which was confirmed on day 56. The patient underwent allogeneic transplant due to prolonged pancytopenia on day 76 and showed immune reconstitution before experiencing disease relapse and starting further therapy on day 134.
Karyotyping and FISH
Standard karyotyping was performed locally by the Medical College of Wisconsin (Milwaukee, WI) on tissue from fresh peripheral blood and bone marrow samples. The 14q11.2 region was further interrogated by FISH using TRA/D Dual Color Break-Apart probes (Abbott Molecular, Abbot Park, IL), with Spectrum Orange Probe: chr14:20895431-21554322 and Green Probe: chr14:22531141-23244993 (hg38 reference genome build). The break-apart probe sets were applied to specimens, hybridized, and washed according to the standard Medical College of Wisconsin interphase FISH protocol. All cells present in the specimen were assayed, until a full 20-cell study (if able) was obtained; karyotyping (cut apart and pairing of the chromosomes) was conducted in approximately three cells per cell line. A total of 200 interphase cells were scored to determine the TCRA/D rearrangement.
HLA and STR typing
HLA and chimerism testing were performed by Versiti (Milwaukee, WI) on ALLO-501A drug products and patient genomic DNA. Eleven HLA class I and II sequences were amplified and sequenced with high-resolution NGS. Additionally, multiplex PCR amplification was conducted on eight total STR loci. Amplification products from CAR T cells and patient baseline DNA were used to uniquely identify recipient and donor alleles, and fluorescence of recipient and donor alleles was used to calculate the percentage chimerism. Using the most appropriate identified markers, the average percentages derived from the recipient and donor were reported.
TCRβ clonality data
Sequencing of the TCRβ CDR3 region in genomic DNA extracted from whole blood or ALLO-501A drug product was performed by Adaptive Biotechnologies (Seattle, WA). Clone frequencies for each sample were calculated as the number of reads representing a unique CDR3 sequence divided by the total reads for the sample. Scatterplots were used to compare clone frequencies between two samples. The sensitivity of the assay was <7 in 1 million clones, 95% confidence interval(0, 6.7 × 10−6) (Figure S1). Statistical significance of a clone’s shift in frequency in one sample relative to the other was calculated using a binomial distribution framework.
VCN analysis
VCN analysis was performed and analyzed by Navigate Biosciences (Carlsbad, CA) utilizing a validated qPCR assay. Genomic DNA was extracted from patient peripheral blood and bone marrow aspirate samples collected in K2EDTA tubes utilizing the QIAamp Blood DNA Midi Kit (QIAGEN, Germantown, MD) and quantified using Qubit Fluorometer. Patient DNA input ranged from 20 to 200 ng per reaction and was analyzed with a validated primer probe set specific for the ALLO 501A CART drug product, 2X TaqManGene Expression Master Mix (Thermo Fisher Scientific, Waltham, MA) and molecular-grade water, ultrapure distilled water, and DNase and RNase-free water (Thermo Fisher Scientific). Analysis was conducted on the Applied Biosystems ViiA7 qPCR system in conjunction with the ViiA7 Applied Biosystems software version 1.2.4 or newer (Thermo Fisher Scientific).
Off-target TALEN assessment
Flanking sequences (~500 bp) on each side of the breakpoints were screened for potential TRAC and CD52 TALEN binding off-target sites (four sequences, two from each pair) using publicly available tools.36 Sequence analysis was performed using three approaches: (1) NCBI BLASTn suite optimized for more dissimilar sequences, (2) Fuzzy Search DNA, which accepts a DNA sequence along with a query sequence and returns sites that are identical or similar to the query, and (3) TALENoffer, which scans input sequences for off-targets of a given TALEN recognition sequence.
WGS and alignment
Peripheral blood was drawn from the patient into K2EDTA tubes on day 61 after infusion. Genomic DNA was purified from the blood samples using QIAamp DNA Blood kits (QIAGEN, Germantown, MD). Genomic DNA of ALLO-501A CAR T drug product (DP) was purified using Allprep mini kit (QIAGEN, Germantown, MD). Libraries were prepared with Illumina DNA Prep kits (Illumina, San Diego, CA). Individual DNA libraries were sequenced using an Illumina NovaSeq6000 platform (Illumina) at SP-500 cycle, 2 × 250 bp paired-end reads.37 Sequencing was performed at a depth of 100× to ensure that any variants at as low as 1% frequency would be detected. Paired-end sequencing tagged and sequenced both ends of the DNA fragments in the libraries, thus providing the ability to identify discordant alignments of paired reads to the genome to identify possible structural variants.
The alignment pipeline consists of a series of steps that started by evaluating the FastQ files for read quality with the FastQC online tool (version 0.11.8).38 Next, the reads were trimmed using Trim Galore 0.6.0 software (available at: www.bioinformatics.babraham.ac.uk/projects/trim_galore) to remove sequencing adaptors and low-quality sequences. Pair-end reads >80 bp after trimming were aligned using the BWA software package (version 0.7.17)39 and the generated alignment file was then sorted and indexed using SAMtools (version 1.9).40 Finally, alignment statistics were generated using Picard (version 2.20.3; available at https://broadinstitute.github.io/picard/), and a final quality control (QC) report was generated using the MultiQC v1.8.41
Alignment files generated during the WGS process were used as input for the identification of genomic structural variations. Duplicate reads, split reads (terminal sequence with >20 bp that do not match the reference sequence), and discordant reads that did not produce the expected alignment using SAMBLASTER (version 0.1.25)42 and Sambamba (version 0.7.1) were removed,37 and the resulting files were used by two structural variation algorithms, LUMPY (version 0.2.13)43 and Manta (version 1.6.0),44 to identify potential structural variation events with a minimum of five supporting sequencing reads. Results from both algorithms were combined and filtered to identify events affecting chromosome 14 with breakpoints in the proximity of the two FISH probes that suggested the existence of an inversion. The Integrative Genomics Viewer (IGV version 2.6.1; available at https://software.broadinstitute.org/software/igv/node/295) was used to validate possible structural variants and inspect the locations of potential inversion-related breakpoints.
Verification of breakpoint sequences
High-coverage, long-read nanopore sequencing was used to verify the breakpoint sites, flanking sequence, and structure of the chromosome 14 inversion. Structural variants (SVs) relative to GRCh38 reference genome assembly were detected using NGMLR pipeline. Reads spanning chromosome 14 q11.2 and q32 region were found. Two breakpoints were identified at 22,537,626 and 106,728,167, separately, matching the two breakpoints identified in the previous WGS with Illumina short-read sequencing. In addition, while one of the inverted junctions was intact, two deletions of approximately 255 kb each were identified at the other inverted junction. While the sequence of events cannot be definitively established, large deletions have been shown to occur at double-stranded break sites induced by nucleases, including the RAG complex.45 Molecular sequence of junction regions matched with expected reference genomic regions, although there were some small deletions, insertions, and single-nucleotide polymorphisms (SNPs), which were mostly due to base-calling errors from nanopore sequencing. Finally, amplicons spanning the breakpoints were generated and sequenced, confirming the breakpoint and flanking sequences.
Lentiviral vector integration site analysis
Integration site analysis (ISA) was conducted at the Viral/Molecular High Density Sequencing Core at the University of Pennsylvania (Philadelphia, PA). This analysis was performed using the INSPIIRED pipeline.46,47 Integrated lentiviral sequences from patient and ALLO-501A DP genomic DNA were analyzed through nested PCR amplification and subsequent sequencing on Illumina MiSeq or HiSeq platforms. Novel blocking locked nucleic acid primers were utilized to reduce sequencing of undesirable internal long terminal repeat sequences, and integration site frequency was determined with the SonicAbundance method.
ddPCR method development
A linear double-stranded DNA template (gBlock) was designed as the synthetic template and positive control for method development for both the centromeric and telomeric sites of inversion based on the NGS sequence information derived from the index patient (Figure S2). As part of the initial method development, primers and probes for the ddPCR assay were designed and optimized. The synthetic template was titrated into human genomic DNA without the inversion and analyzed for specificity and sensitivity to the inversion sequence. The ddPCR assay was found to detect the signal from the synthetic template down to a sensitivity level of approximately 0.1% (Table S3).23 Increasing amounts of input DNA from DP and associated PBMCs were also tested for the inversion (Table S4). Based on these data, 50 ng of human genomic DNA corresponding to ~15,000 human genome copies was used as the input per reaction. Some patient samples yielded lower quality and/or concentration of DNA, and the input DNA amount was adjusted accordingly. Amplified copies of inversion junctions and RPP30 were measured by a ddPCR droplet reader. Each ddPCR plate consisted of a negative control (NEG CTRL: TaqMan Control Genomic DNA), positive control (POS CTRL: synthetic template was spiked into the TaqMan control genomic DNA), and a no-template control (NTC: water). The frequency of inversion was measured as percentage of the total genome copies of the reference gene (RPP30). ddPCR was performed using a Bio-Rad QX200 ddPCR system (Bio-Rad Laboratories, Hercules, CA) according to manufacturer’s instructions.
Acknowledgments
The authors appreciate and thank several specific individuals who contributed to the work presented here. Elaine Murray McCracken (Allogene Therapeutics, Inc.) managed the safety review. Julio Fernández Banet helped identify the inversion, especially the centromeric end, and David Nickle refined the inversion, especially the deletions and the telomeric end (Monoceros Biosystems, LLC). Similarly, the authors also appreciate and thank Yihe Wang, Jonathan Schultz, and Michael Sheldon (Infinity BiologiX LLC doing business as Sampled) for work on the amplification and short-read sequencing, and Eric Butz (Cascadia Drug Development Group, LLC) for flow cytometry analysis and advice. Finally, Erik MacLaren (medical leverage) assisted in the preparation of this manuscript. All funding was provided by Allogene Therapeutics.
Author contributions
B.J.S., G.J.O., S.G.G., D.C., A.M., M.L., and R.G.A. were responsible for overall experimental strategy. G.J.O., S.G., V.K., T.F., D.H., A.G., Y.H., J.C., A.N., S.P., M.A.B., T.P., and M.L. were responsible for experimental design, execution, and analysis. A.B., N.N.S., M.H., and K.M.B. were responsible for patient management and site-based assays. B.J.S., G.J.O., S.G.G., V.K., and R.G.A. were responsible for drafting the manuscript. G.J.O., T.F., S.P., H.D.E., and M.L. were responsible for sample and data logistics. B.J.S., G.J.O., D.C., A.M., and R.G.A. are responsible for the integrity of the work as a whole.
Declaration of interests
B.J.S., G.J.O., S.G., V.K., T.F., D.H., A.G., Y.H., A.N., J.C., A.B., S.P., M.A.B., T.P., H.D.E., S.G.G., D.C., A.M., M.L., and R.G.A. were employees of Allogene Therapeutics, Inc. at the time of this work. M.H. reports research support/funding from Takeda Pharmaceutical Company, ADC Therapeutics, Spectrum Pharmaceuticals, and Astellas Pharma; consultancy in the last 2 years with Incyte Corporation, MorphoSys, Kite, Genmab, SeaGen, Gamida Cell, Novartis, Legend Biotech, Kadmon, ADC Therapeutics, Omeros, and Abbvie; speaker’s bureau funding from Sanofi Genzyme, AstraZeneca, BeiGene, and ADC Therapeutics; and has participated in a data monitoring committee for Genetech and Myeloid Therapeutics, Inc. N.N.S. reports consultancy with Miltenyi Biotec; honoraria and speakers bureau funding from Incyte; serving on advisory boards for Kite, BMS, TG therapeutics, Lilly Oncology, Legend, Novartis, Umoja, and Incyte; and receiving institutional research support for clinical trials from Miltenyi Biotec. He is a member of the scientific advisory board for Tundra Therapeutics.
Footnotes
Supplemental information can be found online at https://doi.org/10.1016/j.ymthe.2022.12.004.
Supplemental information
References
Articles from Molecular Therapy are provided here courtesy of The American Society of Gene & Cell Therapy
Citations & impact
Impact metrics
Citations of article over time
Alternative metrics
Discover the attention surrounding your research
https://www.altmetric.com/details/140278385
Smart citations by scite.ai
Explore citation contexts and check if this article has been
supported or disputed.
https://scite.ai/reports/10.1016/j.ymthe.2022.12.004
Article citations
Allogeneic and other innovative chimeric antigen receptor platforms.
Clin Hematol Int, 6(3):61-72, 27 Sep 2024
Cited by: 0 articles | PMID: 39351308 | PMCID: PMC11441714
Progress and pitfalls of gene editing technology in CAR-T cell therapy: a state-of-the-art review.
Front Oncol, 14:1388475, 07 Jun 2024
Cited by: 1 article | PMID: 38912057 | PMCID: PMC11190338
Review
Free full text in Europe PMC
Non-viral expression of chimeric antigen receptors with multiplex gene editing in primary T cells.
Front Bioeng Biotechnol, 12:1379900, 31 May 2024
Cited by: 0 articles | PMID: 38882639 | PMCID: PMC11177325
Genome Editing in Engineered T Cells for Cancer Immunotherapy.
Hum Gene Ther, 34(17-18):853-869, 01 Sep 2023
Cited by: 2 articles | PMID: 37694593 | PMCID: PMC10623081
Review Free full text in Europe PMC
Unbiased assessment of genome integrity and purging of adverse outcomes at the target locus upon editing of CD4+ T-cells for the treatment of Hyper IgM1.
EMBO J, 42(23):e114188, 02 Nov 2023
Cited by: 3 articles | PMID: 37916874 | PMCID: PMC10690452
Go to all (6) article citations
Similar Articles
To arrive at the top five similar articles we use a word-weighted algorithm to compare words from the Title and Abstract of each citation.
Funding
Funders who supported this work.