If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Therapeutic disruption of immune checkpoints has significantly advanced the armamentarium of approaches for treating cancer. The prominent role of the programmed death-1 (PD-1)/programmed death ligand-1 axis for downregulating T cell function offers a tractable strategy for enhancing the disease-modifying impact of CAR-T cell therapy.
Methods
To address checkpoint interference, primary human T cells were genome edited with a next-generation CRISPR-based platform (Cas9 chRDNA) by knockout of the PDCD1 gene encoding the PD-1 receptor. Site-specific insertion of a chimeric antigen receptor specific for CD19 into the T cell receptor alpha constant locus was implemented to drive cytotoxic activity.
Results
These allogeneic CAR-T cells (CB-010) promoted longer survival of mice in a well-established orthotopic tumor xenograft model of a B cell malignancy compared with identically engineered CAR-T cells without a PDCD1 knockout. The persistence kinetics of CB-010 cells in hematologic tissues versus CAR-T cells without PDCD1 disruption were similar, suggesting the robust initial debulking of established tumor xenografts was due to enhanced functional fitness. By single-cell RNA-Seq analyses, CB-010 cells, when compared with identically engineered CAR-T cells without a PDCD1 knockout, exhibited fewer Treg cells, lower exhaustion phenotypes and reduced dysfunction signatures and had higher activation, glycolytic and oxidative phosphorylation signatures. Further, an enhancement of mitochondrial metabolic fitness was observed, including increased respiratory capacity, a hallmark of less differentiated T cells.
Conclusions
Genomic PD-1 checkpoint disruption in the context of allogeneic CAR-T cell therapy may provide a compelling option for treating B lymphoid malignancies.
The advancement of immunotherapies for the treatment of various malignancies over the past decade has included adoptive cell therapies (ACTs) such as chimeric antigen receptor (CAR)-T cells that have elicited significant disease-modifying impact [
]. Targeting of B cell malignancies through the cell surface antigen CD19 has led to profound responses in patients with acute lymphoblastic leukemia (ALL) and non-Hodgkin lymphoma (NHL) [
], leading to the approval of multiple autologous CAR-T cell therapies. Despite this progress, several limitations remain that prevent the widespread access of autologous CAR-T cell therapies for many patients, including restricted eligibility, bridging therapy requirements, variable T cell integrity and potency, high production costs and manufacturing complexity [
] enables an opportunity to overcome the challenges of autologous CAR-T cell therapy, providing an alternative with more universal application among patients with significant medical need. Limitations of allogeneic CAR-T cell therapies include their persistence, since the cells are subject to rejection by the patient's immune system, and the rapid downregulation of their antitumor activity through cell exhaustion.
Despite the remarkable responses observed in patients with B-NHL and B lymphoid leukemias treated with autologous CD19 CAR-T cell products, the durability of these responses is often limited [
Integrating CAR T cell therapy and transplantation: comparisons of safety and long-term efficacy of allogeneic hematopoietic stem cell transplantation after CAR T cell or chemotherapy-based complete remission in B cell acute lymphoblastic leukemia.
]. Further developments to enhance the potency of CAR-T cell therapies are needed to promote longer-term antitumor activity. One of the options that has surfaced is the disruption of the programmed death-1 (PD-1)/programmed death ligand-1 (PD-LI) checkpoint axis between T cells and tumor cells [
]. Preclinical and clinical studies have evaluated the combination of CAR-T cells with antibodies that disrupt this checkpoint, demonstrating some improvement in preclinical xenograft models [
], whereas treatment of patients with B-NHL with autologous CD19-specific CAR-T cells led to improved outcomes in patients whose tumor cells were PD-L1 negative [
] using a high specificity next-generation platform that uses Cas9 with guides containing both RNA and DNA residues termed chimeric hybrid RNA-DNA or chRDNA (Cas9 chRDNA) [
] to generate allogeneic CAR-T cells expressing a CD19-directed CAR that was site-specifically integrated into the T cell genome at the T cell receptor alpha constant (TRAC) locus. The Cas9 chRDNA system enables superior genome integrity following editing, such that high efficiency on-target editing is achieved without the generation of detectable off-target edits [
]. In addition, we disrupted expression of the PDCD1 locus that encodes the PD-1 checkpoint receptor by Cas9 chRDNA-mediated gene knockout (KO). These combined edits generated allogeneic CAR-T cells (CB-010) that mediated enhanced survival compared with CAR-T cells without a PDCD1 KO (CAR knock-in [CAR KI]) in an established orthotopic xenograft model of a B cell malignancy. These two CAR-T cell preparations, CB-010 and CAR KI, exhibited similar activity in short-term antigen-stimulated functional assays in vitro and showed similar rates of expansion and biodistribution in mouse models, suggesting that CB-010 cells likely become exhausted more slowly than the PD-1+ CAR KI cells. Importantly, single-cell RNA-Seq analyses demonstrated that CB-010 cells exhibited fewer Treg cells, lower exhaustion phenotypes, reduced dysfunction signatures and higher activation, glycolytic and oxidative phosphorylation (OXPHOS) signatures compared with CAR KI cells. Additionally, CB-010 cells harbor functionally enhanced energy reserves, suggesting that they may favor a higher level of OXPHOS for ATP production compared with PD-1+ CAR-T cells, thereby providing a functional advantage for sustained antitumor activity.
] specific for the TRAC and PDCD1 genes were generated by combining recombinant S. pyogenes Cas9 enzyme fused to a nuclear localization sequence in 1× CCE buffer (20 mmol/L HEPES:KOH, pH 7.4; 10 mmol/L MgCl2; 150 mmol/L KCl; 5% glycerol) and annealed chRDNA-chacr (chacr = tracrRNA/DNA hybrid) complexes (IDT, Coralville, IA, USA) at a molar ratio of 1:3 Cas9:chRDNA.
Editing efficiencies for TRAC and/or PDCD1 were determined for CB-010 CAR-T cells and TRAC KO cells. Genomic DNA was lysed using QuickExtract DNA Extraction Solution (Epicentre, Madison, WI, USA) according to the manufacturer's instructions. Cas9 target sites were amplified using a two-step polymerase chain reaction (PCR). In brief, 3.75 μL (corresponding to approximately 7500 cells) of lysate was used as a template for PCR amplification with Q5 Hot-Start HighFidelity DNA Polymerase (New England Biolabs, Ipswich, MA, USA) and unique primer pairs containing an internal locus-specific region and an outer Illumina-compatible adapter sequence. A second PCR targeting the outer-adapter sequence was performed to append unique indices to each amplicon. Cas9-targeted amplicon samples were sequenced on a MiSeq platform with 1 × 151 paired-end reads (Illumina, San Diego, CA, USA). For each site, indels were tallied if they occurred within the 20-base pair region of the putative Cas9 targeted site, and editing efficiencies were calculated by subtracting the percentage of indels in unedited control cells from the percentage of indels in electroporated cells. The depth of coverage was approximately 5000–50 000 reads per amplicon, and all samples with <500 reads were discarded.
A droplet digital PCR procedure was used to quantify the site-specific genomic insertion of recombinant AAV6 (rAAV6) donor DNA into the TRAC locus at position chr14:22550584 (hg38) using genomic DNA extracted from the edited T cells on day 10. Each droplet digital PCR was multiplexed, containing both a two-copy genomic reference HEX-labeled primer/probe set and a FAM-labeled primer/probe set for the TRAC locus. rAAV6 DNA correctly inserted into the TRAC locus was detected using primer/probe sets consisting of one genome-specific and one insert-specific primer with a homology arm specific FAM-labeled probe (BioRad Automated Droplet Generator, BioRad C1000 Touch Deep Well Thermal cycler, BioRad QX200 Droplet Reader; BioRad, Hercules, CA, USA). The site-specific genomic insertion rate or the total copies of rAAV6 donor DNA were calculated as the ratio of the copies of the target to the copies of the reference.
CAR-T cell production
CB-010 CAR-T cells were generated from healthy donor–derived peripheral blood mononuclear cells. To summarize, T cells were isolated from cryopreserved peripheral blood mononuclear cells (STEMCELL Technologies, Vancouver, Canada) using RoboSep-S (STEMCELL Technologies) and the EasySep Human T cell Isolation Kit (STEMCELL Technologies) and activated for 3 days in the presence of anti-CD3/CD28 beads (Dynabeads; Gibco, Billings, MT, USA) along with recombinant human interleukin 2 (rhIL-2; 100 units/mL). Beads were removed and cells were expanded for 24 h with rhIL-2 before nucleofection and transduction. Cas9 chRDNA complexes to edit the TRAC and PDCD1 loci were generated by combining purified Cas9 enzyme, 1× CCE buffer (20 mmol/L HEPES:KOH, pH 7.4; 10 mmol/L MgCl2; 150 mmol/L KCl; 5% glycerol) and annealed chRDNA-chacr complexes at a molar ratio of 1:3 Cas9:chRDNA. T cells were nucleofected using a 96-well Shuttle System (Lonza, Basel, Switzerland) with Cas9 chRDNA complexes and subsequently transduced with rAAV6 (Signagen, Frederick, MD, USA) engineered to deliver the non-targeting CAR transgene or the CB-010 CAR transgene. Cells were cultured for an additional 8 days in ImmunoCult-XF T Cell Expansion Medium (STEMCELL Technologies) supplemented with 5% CTS Immune Cell Serum Replacement (Gibco) and rhIL-2.
Human tumor cell lines, recombinant cell lines and patient-derived tumors
Cell lines were obtained from ATCC or DSMZ. The patient-derived DLBCL tumor xenograft (CTG-3020) study was conducted at Champions Oncology (Hackensack, NJ, USA). The NALM-6-ffLuc-eGFP/PD-L1 (referred to as NALM-6/PD-L1) clonal cell line was generated via lentiviral transduction. Viral packaging was performed by transfecting HEK293 cells with either a cis-plasmid encoding ffLuc-eGFP or a cis plasmid encoding PD-L1 using TransIT-293 transfection reagent (Mirus, Marietta, GA, USA). Viral supernatants were harvested, filtered with a 0.45-μm syringe filter, and concentrated using a Lenti-X Concentrator (Clontech, Mountain View, CA, USA). NALM-6 parental cells were transduced with virus followed by selection with blasticidin and puromycin antibiotics. Single-cell NALM-6-ffLuc-eGFP/PD-L1 clones were isolated by fluorescence-activated cell sorting (FACS; Sony Sorter SH800Z) and CD19, eGFP, and PD-L1 expression was monitored by flow cytometry (Intellicyt iQue Screener Plus).
In vitro cytotoxicity, cytokine secretion and proliferation assays
Cytotoxicity was evaluated in vitro using 9 different CD19+ cell lines. K562 cells were used as a CD19-negative control. Briefly, target cells (T; tumor cell lines) were labeled with CellTrace Violet (CTV; C34557; Thermo Fisher Scientific, Waltham, MA, USA) to distinguish them from effector T cells (E) and cells were co-cultured at E:T ratios of 0:1, 0.05:1, 0.10:1, 0.20:1, 0.33:1, 1:1, 3:1 and 10:1 (3 co-culture wells/E:T ratio). Cytotoxicity was measured by gating on the CTV cell population (target cells) and live cells as measured by propidium iodide after 48 h of co-culture. Data were analyzed by flow cytometry (Intellicyt iQue Screener Plus). Specific lysis was calculated using the following equation for each well:
Supernatants were harvested from E:T co-cultures at 48 h and monitored for the presence of cytokines with MultiCyt QBeads and analyzed using the Intellicyt iQue Screener Plus.
Antigen-dependent proliferation of CB-010 cells was evaluated in vitro using CD19+ cell lines; CD19-negative K562 cells served as the control. To summarize, these target cells were co-cultured at a ratio of 1:1 with CB-010 CAR-T cells or TRAC KI cells labeled with CTV in the absence of IL-2. Proliferation was measured by reduction in CTV fluorescence 48 or 96 h after initiation of the co-culture using flow cytometry (Intellicyt iQue Screener Plus). Gating for CTV reduction was based on CTV-negative K562 cells at the same time point.
Flow cytometry analysis
Flow cytometry was performed using antibodies specific to human TCRα/β, CD19, PD-L1 and CD45 (BioLegend, San Diego, CA, USA). CAR was detected using a two-step process with a CD19 CAR detection reagent labeled with biotin and an anti-biotin antibody conjugated with phycoerythrin (PE; Miltenyi Biotec, Bergisch Gladbach, Germany). Cells were washed with FACS buffer (phosphate-buffered saline [PBS] containing 2% fetal bovine serum and ethylenediaminetetraacetic acid), incubated with antibodies for 30 minutes at 4°C, and washed twice with FACS buffer before analysis. Blood, bone marrow, and spleen samples from mice were treated with ammonium chloride solution (STEMCELL Technologies) for red blood cell lysis, washed twice, and incubated with TruStain FcX anti-mouse CD16/32 (BioLegend) for 10 min at 4°C before staining to detect CD45 and CAR expression. Samples were analyzed using a BD LSRFortessa X-20 instrument (BD Biosciences, San Diego, CA, USA), and flow cytometry data were analyzed using FlowJo flow cytometry analysis software v.10 (BD Biosciences).
Measurement of CD19 antigen density
CD19 antigen density on the cell surface of NHL cell lines (NALM-6, DOHH-2, SC-1, MAVER-1, JeKo-1, REC-1, Z-138, Toledo, and WSU) and primary B cells was determined using BD Quantibrite Beads, PE Phycoerythrin Fluorescence Quantitation Kit (BD Biosciences) in combination with primary antibodies conjugated with PE and specific to human CD19 antigen (BioLegend). All procedures were performed in accordance with the manufacturer's instructions. Antigen density was calculated based on mean fluorescence intensity of the stained cells using the standard curve generated from the mean fluorescence intensity of four populations of beads conjugated with different levels of PE.
Transcriptional profiling of CAR-T cells from orthotopic tumor xenograft mouse models
CD19+ NALM-6/PD-L1 tumor cells were engrafted orthotopically into NOD-SCID-gamma (NSG) mice for 21 days to generate an established tumor xenograft model. A single dose of CB-010 or CAR KI cells was administered on day 0. To evaluate the transcriptional profile of CB-010 and CAR KI cells, animals were sacrificed at an early time point (day 7) and at a late time point (day 60 and day 54 for CB-010 and CAR KI, respectively). Blood, bone marrow, and spleen samples were harvested for each animal, processed and pooled for sorting of human CD45+ CB-010 or CAR KI cells using a Sony SH800 cell sorter. Sorted cells were profiled using the 10x Genomics single cell RNA sequencing pipeline described to follow.
Construction of single-cell RNA-Seq libraries was performed using the Chromium Next GEM Single Cell 3’ Reagent Kits v3.1 (Dual Index) with Feature Barcode technology for Cell Surface Protein (10x Genomics, Pleasanton, CA, USA). For cell surface protein analysis, each sample was stained with the TotalSeq-B Human Universal Cocktail according to the manufacturer's instructions (BioLegend). The 3′-gene expression (RNA) and cell surface protein (ADT) libraries were constructed and sequenced using the NextSeq 2000 platform (Illumina).
The Cell Ranger Single-Cell Software Suite (version 2.1.0) was used to perform sample demultiplexing, barcode assignments, and gene alignments. Raw base BCL files were demultiplexed using the Cell Ranger mkfastq pipeline into sample-specific FASTQ files. FASTQ files were processed individually using the Cell Ranger count pipeline, which reads to the pre-built GRCh38 human reference genome provided by Cell Ranger (10x Genomics). Aligned reads were then filtered for valid cell barcodes and unique molecular identifiers (UMIs). For assignment of CAR status to individual cells, a custom genome reference was built using the GRCh38 human reference genome and included the CDS sequence of the CD19 CAR transgene. A cell was considered CAR+ if it contained at least one UMI count aligning to this transgene. The CAR status was added to the metadata information of the cell; all gene expression analysis was then performed on the pre-built GRCh38 human reference genome alignment. In vivo samples were aligned to a dual mm10/GRCh38 reference genome provided by Cell Ranger to identify contaminating mouse cells. All samples were corrected for sequencing depth by down sampling all sample reads to ∼40 000 reads per cell using the downsampleReads function from the DropletUtils package before downstream analysis [
participants in the 1st Human Cell Atlas Jamboree, Marioni JC EmptyDrops: distinguishing cells from empty droplets in droplet-based single-cell RNA sequencing data.
Following sample alignment and demultiplexing, the resulting cell–gene count matrix was imported into the Seurat (version 4.2.0) analysis package for further cell filtering and analysis [
]. Cells were filtered from the final analysis based on the following criteria: a library complexity below an outlier cutoff (defined on a per-sample basis as a value below Q1 – 1.5*IQR of all cells in the sample) or a percent mitochondrial gene expression above an outlier cutoff (defined on a per-sample basis as a value below Q3 + 1.5*IQR of all cells in the sample) or 15%, whichever was lower. This resulted in a final dataset of 33 462 cells. Genes were excluded from the analysis if they did not contain at least 1 UMI count in at least 3 cells in each sample. Each sample was processed individually following the standard Seurat R functions according to the recommended workflow tutorial for normalizing with SCTransform. Following data normalization, all samples were integrated using the Seurat IntegrateData workflow using “SCT” as the normalization method and “rpca” as the reduction method. Contaminating mouse cells were removed based on the GEM call output of the dual mm10/GRCh38 alignment from Cell Ranger. Contaminating NALM-6 cells were removed based on expression of CD19 and PD-L1.
The integrated Seurat object was used to perform principal component analysis, clustering and uniform manifold projection analyses on the “integrated” assay. The first 30 principal components were used as input to both the RunUMAP() and FindNeighbors() functions of Seurat. Following cell clustering, clusters were assigned cell identities based on expression of canonical markers from both the ADT and RNA assays.
To aid in visualization of the RNA assay in this dataset, the normalized RNA assay was passed into the MAGIC package for expression level imputation [
]. These imputed RNA values were used for visualization of various marker genes and for calculation of module scores in Seurat.
For module score calculations, individual gene sets were used as input into the AddModuleScore() function of Seurat (supplementary Table 1). These scores were then scaled by mean centering all values and dividing them by their standard deviation. Significance of the differences in these scores was determined using a nonparametric Kruskal–Wallis test.
Seahorse assay for metabolic analysis
The extracellular acidification rate (ECAR) and oxygen consumption rate (OCR) were measured using the Seahorse XF T Cell Metabolic Profiling Kit on an Agilent Seahorse XFe96 Analyzer (Agilent Technologies, Santa Clara, CA, USA) according to the manufacturer's instructions. CD19-specific CAR-T cells with a PDCD1 KO (CB-010) or without a PDCD1 KO (CAR KI) were co-cultured with NALM-6/PD-L1 cells for 7 or 18 days. NALM-6/PD-L1 cells were added to the culture every 3–5 days. At the end of the co-culture, CAR-T cells were purified and rested overnight before the Seahorse assay to profile T cell energy metabolism. Spare respiratory capacity was calculated as the difference between the maximal OCR signal and the basal OCR signal.
Orthotopic and subcutaneous tumor xenograft mouse models
All animal experiments except the PDX study were performed unblinded by Caribou personnel according to national ethical guidelines in addition to the guidance and approval by the Institutional Animal Care and Use Committee of Explora Biolabs. The PDX xenograft study was performed at Champions Oncology, and all procedures were performed according to the guidelines of the Champions Oncology Institutional Animal Care and Use Committee. For all studies, euthanasia was conducted in compliance with the current requirements of the Guide for the Care and Use of Laboratory Animals, 8th Edition, and the American Veterinary Medical Association Guidelines on Euthanasia. All mice were group housed with environmental enrichment, and irradiated feed and sterilized water were provided ad libitum. Female NSG mice were purchased from The Jackson Laboratory (Bar Harbor, ME, USA) at approximately 7 weeks of age. Mice were acclimated for 7 days upon receipt, randomized according to body weight (minimally established models) or tumor volume (well-established models) and were injected with tumor cells as described to follow at approximately 8 weeks of age. Tumor cells were washed twice in prewarmed PBS, centrifuging each time at 300g for 7 min. Viability was determined after washing, and cells were prepared at 2.5 × 106 viable cells/mL for tumor engraftment in mice at 5 × 105 total cells/mouse in a final volume of 200 μL. Tumor engraftment by either tail vein injection or subcutaneous administration was performed. CAR-T cells harvested on day 10 were washed twice in prewarmed PBS and centrifuged at 300g for 7 min each time. Viability was determined after washing, and cells were prepared at 5 × 107 viable CAR+ cells/mL for dosing animals in a final volume of 200 μL/animal. Mice were dosed based on the average weight of the group on the day of dosing. PBS was used as a negative control. Body weight was measured after CAR-T cell administration. Animals were euthanized after losing >20% of their body weight compared with day 0 or after developing paralysis. Bioluminescence imaging was performed using an IVIS Spectrum system (Perkin Elmer, Waltham, MA, USA). Ten minutes before bioluminescence imaging, tumor-bearing mice received 150 mg luciferin/kg (Perkin Elmer) intraperitoneally. Mice were anesthetized with 2% isoflurane gas and imaged without an emission filter (840 nm, total light output - open filter) for 5 min.
Biodistribution and toxicity analysis
Following nucleofection, CB-010 CAR-T cells underwent a TCR depletion step to remove any remaining TCR+ cells. CB-010 CAR-T cells were incubated with biotinylated anti-TCRα/β antibodies and anti-biotin microbeads, and TCRα/β+ cells were removed using the CliniMACS Plus magnetic cell separation system (Miltenyi Biotec) and the remaining CB-010 CAR-T cells were cryopreserved in a controlled rate freezer. CAR and TCR expression were determined by cell surface staining of a sample of CB-010 CAR-T cells and donor-matched wild-type T cells using CAR detection reagent CD19-Alexa Fluor 647, TCR α/β detection reagent anti-human TCR α/β-BV421(BioLegend 306722, clone IP26), and anti-human CD3 (Becton-Dickinson [Franklin Lakes, NJ, USA], 347347, clone SK7). Expression was determined by fluorescence using a flow cytometer (LSRFortessa; BD Biosciences). Thirty female NBSGW mice were purchased from The Jackson Laboratory and arrived at approximately 8 weeks of age and were used when the mice were approximately 12 weeks of age. CB-010 CAR-T cells and wild-type T cells (generated from the same donor) were thawed in a water bath, washed twice with pre-warmed PBS and centrifuged at 300g for 7 min each time. Viability was determined after washing, and cells were prepared at 1.5 × 108 viable cells/mL for dosing animals at 3.0 × 107 total cells/animal in a final volume of 200 μL. This dose was approximately 1.2 × 109 cells/kg based on the average weight of the groups on the day of dosing. CAR-T cell dosing was performed via tail vein injection. PBS was used as a negative control and mice were dosed with an equivalent volume of PBS as the CB-010 CAR-T cell and wild-type T cell dosed mice.
Body weight was measured during the study period up to day 71. Animals were euthanized after losing >20% of their body weight compared with day 0 (day of dosing), developing paralysis or found moribund. The animals were graded throughout the study for clinical signs indicative of development of graft-versus-host disease (GvHD; weight loss, posture, activity, fur texture and skin integrity [
]. On days 30 and 72, two mice from the PBS-treated group and three animals each from the unmodified WT T cell– and CB-010 CAR-T cell–dosed groups were euthanized and tissues were collected for biodistribution by quantitative PCR (qPCR) analysis. From the collected organs, one lung lobe, half of the spleen, one axillary lymph node, and one ovary were snap frozen in liquid nitrogen and stored at –80°C. Samples were processed into genomic DNA for qPCR of human ribonuclease P protein subunit p30 (RPP30) as a measure for detecting CB-010 CAR-T cells or wild-type T cells. Tissues for histopathology were collected on day 30 and were fixed in formalin and transferred to 70% ethanol after 24 h and transferred to IDEXX Laboratories (West Sacramento, CA, USA) for analysis. Femurs were decalcified and all tissues were trimmed and processed into slides via paraffin infiltration, embedding, and sectioning. Slides were stained with hematoxylin and eosin and anti-human CD45 antibody (Cell Signaling Technology, Danvers, MA, USA) to determine the biodistribution of CB-010 CAR-T cells or wild-type T cells.
Statistical analyses
Statistical analyses were performed using GraphPad Prism software (version 7.0d), which was used to create Kaplan–Meier survival curves and calculate the 95% confidence interval for fractional survival at any time (Mantel–Cox log-rank test). For biodistribution assessment by qPCR, a single-copy reference plasmid (pCB6244) was used to generate a standard curve (10 – 1 × 107 copies/μL). For RPP30, samples were run in triplicate, an average threshold cycle was calculated, and then standard deviation was calculated. Linear regression was used to fit the equations to determine the total number of copies of RPP30 and copies of RPP30/ng genomic DNA.
Results
Development and characterization of CB-010, allogeneic CD19-specific CAR-T cells with a knockout of PDCD1
] was implemented to generate a KO of the TRAC locus in healthy donor-derived primary human T cells. Screening of the TRAC locus for Cas9 protospacer adjacent motif sequences yielded multiple all-RNA guides (crRNA) that exhibited high editing efficiency (supplementary Figure 1A). chRDNA guides were generated and optimized from the all-RNA sequences using previously described strategies [
]. Genome editing with a TRAC-specific chRDNA guide and Cas9 by electroporation of primary human T cells exhibited high precision with no detectable off-target editing within the limits of detection of the assay (0.1%) (Figure 1A). Off-target editing sites were identified using the biochemical SITE-Seq® assay [
]. A CD19-specific CAR transgene expression cassette was designed containing the CD8 signal sequence, FMC63 scFv, AAA hinge, CD8 transmembrane domain, 4-1BB intracellular costimulatory domain, and the TCR ζ-chain endodomain (Figure 1B). The CAR transgene was site-specifically inserted into the cut site in exon 3 of the TRAC locus, and CAR expression was driven by the short-form of the human EF1α promoter [
]. To site-specifically integrate the CAR, the T cells were transduced with AAV6 encoding the CAR construct with homology arms matching the TRAC locus during genome editing. High-efficiency site-specific CAR integration led to its expression in ∼65% of the T cell population, with ∼98% TRAC gene KO efficiency (Figure 1C). Multiple batches of CB-010 were evaluated for CAR and TCRα/β expression and showed similar levels among the different lots when compared by flow cytometry (55.76 ± 9.18% CAR integration and 93.2 ± 1.94% TCRα/β KO; supplementary Figure 1E). We observed greater expression of the CAR and improved antitumor efficacy in vivo using the EF1α promoter compared with the TRAC endogenous promoter (data not shown). An additional edit was implemented in the T cells by KO of the PDCD1 gene encoding the PD-1 protein. All-RNA guides (crRNAs) were identified through PDCD1 gene tiling (supplementary Figure 1B), and a high-efficiency guide was selected for chRDNA optimization. Although off-targets were detected with the all-RNA guide to PDCD1, no off-targets were detectable with the PDCD1 chRDNA with the identical sequence (Figure 1D), and 98.2% of the cells exhibited indels at the PDCD1 locus by amplicon sequencing (Figure 1E). Cells with both the CAR KI and the PDCD1 KO are hereto referred to as CB-010 cells. DNA repair classification, which identifies the presence of insertions and deletions (indels) at KO loci by amplicon sequencing [
], was evaluated for the TRAC and PDCD1 genes and demonstrated consistency across multiple preparations of CB-010 cells (supplementary Figure 1C, 1D). Stimulation of CAR KI T cells, which have a similar frequency of CAR+TCR− cells as CB-010 cells but lack the PDCD1 KO, with two general cell activators phorbol 12-myristate 13-actetate + ionomycin, led to the expression of cell surface PD-1 in 45.4 ± 5.02% of the population (Figure 1F). In contrast, CB-010 cells did not express significant levels of cell surface PD-1 protein following phorbol 12-myristate 13-actetate + ionomycin stimulation (1.25 ± 0.007% of the population expressing PD-1).
Fig. 1Generation and characterization of genome-edited allogeneic CB-010 CAR-T cells. (A) Evaluation of chRDNA guides compared with all-RNA guides (crRNA) for TRAC genome editing efficiency in primary T cells as determined by deep sequencing at the target locus (arrow), and specificity determined by in-cell validation of off-targets identified by biochemical SITE-Seq assay. (B) CD19-specific CAR construct driven by an exogenous promoter (EF1α), CD8 signal sequence (SS), FMC63 scFv, AAA hinge, CD8 transmembrane domain (TM), the 4-1BB costimulatory domain and the CD3ζ endodomain (with the bovine growth hormone polyA sequence) for site-directed insertion into exon 3 of the TRAC locus. (C) CAR expression in a population of primary human T cells (left panel; CAR+ population in box) and knockout frequency at the TRAC locus using Cas9 chRDNA in the same population of T cells (right panel; TCR− population in box). (D) Evaluation of chRDNA guides compared with all-RNA guides (crRNA) for PDCD1 genome editing efficiency in primary T cells as determined by deep sequencing at the target locus (arrow), and specificity determined by in-cell validation of off-targets identified by biochemical SITE-Seq assay. (E) Amplicon sequencing to determine DNA repair classification and editing efficiency of the Cas9 chRDNA specific for the PDCD1 locus. (F) Knockout of the PDCD1 gene with Cas9 chRDNA lead to disruption of cell surface PD-1 expression on phorbol 12-myristate 13-actetate (PMA) + ionomycin stimulated CB-010 CAR-T cells (sky blue), assessed by flow cytometry. CAR KI cells (salmon) express the CAR but lack the PDCD1 knockout. Data represent two batches of CB-010 and CAR KI cells generated from two individual donors.
Several B cell tumor lines were evaluated for CD19 expression by flow cytometry to serve as targets for CB-010 and CAR KI cells (Figure 2A). Cell lines representing B cell acute lymphoblastic leukemia (B-ALL) and different subtypes of B cell non-Hodgkin lymphoma (B-NHL), including diffuse large B cell lymphoma (DLBCL), mantle cell lymphoma (MCL), follicular lymphoma (FL) and transformed follicular lymphoma (tFL), expressed cell surface CD19 with varying densities. Evaluation of CD19+ B-ALL (NALM-6), DLBCL (Toledo) and MCL (JeKo-1) cell lines in cytotoxicity assays comparing CB-010 cells with CAR KI cells demonstrated similar efficiencies of specific lysis at all E:T ratios (Figure 2B). No cytotoxicity was observed against the CD19− cell line K562 (Figure 2B). Negative control T cells with a KO at the TRAC locus only and no CAR insertion (TRAC KO) did not exhibit cytotoxic potential toward any of the cell lines. CB-010 cells were evaluated in cell lytic assays against additional B-NHL lines and exhibited robust cytotoxic potential that did not correlate with CD19 cell surface expression levels (supplementary Figure 2A). We evaluated the cytotoxic potential of CB-010 against normal B cells derived from human peripheral blood. Normal B cells from three independent healthy donors were lysed with similar dynamics in CB-010 co-cultures (supplementary Figure 3).
Fig. 2CB-010 CAR-T cells exhibit antigen-dependent in vitro cytotoxic activity, cytokine secretion, and proliferation. (A) Cell surface CD19 density across a tumor cell line panel. CD19 expression was determined by flow cytometry. (B) Cytotoxicity induced by CB-010 cells (sky blue diamond) compared with CAR KI cells (expressing the CAR but lacking the PDCD1 knockout; salmon triangle) against three CD19+ tumor cell lines (NALM6, Toledo and JeKo-1) and one CD19− tumor cell line (K562). T cells with only the TRAC gene knocked out and no CAR inserted (TRAC KO; olive triangle) were included as a negative control. (C) Secretion of interferon gamma (IFNγ) and IL-2 were evaluated after co-incubation with the same tumor cell lines and T cell preparations as indicated in panel B. (D) Antigen-dependent proliferation of CB-010 and CAR KI cells. Tumor cells expressing CD19 (NALM-6) or CD19-negative cells (K562) were co-incubated with CB-010 or CAR KI cells. Proliferation of CAR-T cells was evaluated with and without targets cells at 72 and 96 hours.
Co-incubation experiments with either CB-010 or CAR KI cells with three selected tumor cell lines led to the secretion of interferon gamma, whereas IL-2 secretion was observed in co-cultures of CB-010 or CAR KI cells with the B-NHL cell lines Toledo (DLBCL) and JeKo-1 (MCL) and at low levels with NALM-6 (B-ALL) (Figure 2C). In contrast, CB-010 cells did not secrete interferon gamma or IL-2 when co-incubated with CD19− cells (data not shown). Antigen-dependent proliferation in the absence of exogenous IL-2 addition was observed upon co-culture of CB-010 cells or CAR KI cells with CD19+ tumor cells, but not control CD19− tumor cells (Figure 2D). Evaluation of potential IL-2 dependence of CB-010 cells in vitro demonstrated that these CAR-T cells required the exogenous addition of IL-2 for growth in culture in the absence of antigenic stimulation (supplementary Figure 2B).
Antitumor activity of CB-010 in established xenograft models of B cell malignancies
The CD19+ B-ALL tumor cell line NALM-6 was engineered to express recombinant human PD-L1 (PD-1 ligand) (Figure 3A; supplementary Figure 4A). These cells were used as tumor xenografts to evaluate the efficacy of CB-010 in an orthotopically established model. Different doses of CB-010 were evaluated in comparison to negative controls in the model, and CB-010 exhibited a dose-dependent increase in survival in the mice (Figure 3B) without a concomitant change in animal body weight (Figure 3C). CB-010 was developed for initial clinical evaluation in patients with multiple different subtypes of B-NHL. To evaluate the effect of CB-010 treatment in B-NHL xenograft models of DLBCL, both the Toledo cell line and a patient-derived xenograft (PDX) model were implemented. Toledo cells were engrafted subcutaneously into NSG mice to establish tumors before the administration of a single dose of CB-010 cells. Caliper measurements were used to evaluate antitumor efficacy. Eradication and long-term suppression of tumor growth were observed after 5 × 106 CB-010 cells were administered, whereas a partial delay in tumor growth was noted with 1 × 106 CB-010 cells (supplementary Figure 5A). Animals dosed with 5 × 106 CB-010 cells/mouse exhibited significantly prolonged survival compared with control mice, up to day 75 before euthanasia (p < 0.0001, compared with non-targeting CAR KI cells [non-targeting CAR KI cells contain a site-specifically inserted CAR that lacks an scFv and the cells lack the PDCD1 KO]). Similarly, in an established orthotopic (intravenous) model of the Toledo tumor xenografts, survival was extended (supplementary Figure 5B). A PDX model of DLBCL was established subcutaneously, and a single-dose CB-010 treatment led to a statistically significant extension in long-term survival of the mice (supplementary Figure 5C). Two additional xenograft models of B-NHL, including MCL (JeKo-1) and tFL (WSU), were evaluated. The MCL line was engrafted as an established subcutaneous model (caliper measurements were used to evaluate antitumor efficacy) that was completely eradicated with the higher dose of CB-010 (supplementary Figure 5D), and CB-010 cell significantly extended survival in the orthotopically established (intravenous) version of this MCL model (supplementary Figure 5E). Finally, a single dose of CB-010 led to statistically significant survival in a highly aggressive orthotopically established (intravenous) xenograft model of tFL (supplementary Figure 5F). Although all of these tumor cell lines expressed cell surface CD19, none of them expressed native cell surface PD-L1 (supplementary Figure 5G). The PDX model was evaluated for both CD19 and PD-L1 by immunohistochemistry by the vendor. CD19 was expressed on the surface of the tumor xenograft cells, but PD-L1 appeared expressed only intracellularly (data not shown).
Fig. 3Dose–response of CB-010 antitumor activity in an established orthotopic CD19+ tumor xenograft model. (A) Flow cytometry analysis of recombinant NALM-6/PD-L1 cells, either unstained (left panel), stained for CD19 only (left center panel), stained for PD-L1 only (right center panel), or stained for both markers (right panel). (B) Nine NOD-SCID-gamma (NSG) mice per group were engrafted orthotopically with NALM-6/PD-L1 tumor cells 4 days before tail vein injection with a single dose of either PBS, TRAC KO T cells or CB-010 cells at 6.25 × 105/mouse, 1.25 × 106/mouse, 2.5 × 106/mouse or 5 × 106/mouse. Kaplan–Meier survival curves of mice to day 105 post-treatment. Comparison between TRAC KO and CB-010 cells (5 × 106), p < 0.0001. (C) Changes in mouse body weight (BW) during the study.
Development of an allogeneic CAR-T cell therapy for clinical evaluation requires mitigation of the potential for GvHD in patients. To evaluate the toxicity of CB-010 in a mouse model of GvHD, non-irradiated NOD-SCID-gamma-Kit (NBSGW) immunodeficient mice were injected intravenously with 3 × 107 CB-010 cells (maximal cell dose in 0.2-mL injection volume) compared with an equal amount of wild-type non-engineered primary human TCR+ T cells. Different parameters of GvHD were evaluated (body weight, fur texture and skin integrity) for 72 days, and the infiltration of CB-010 or control cells into mouse tissues was evaluated from animals on days 30 and 72 (Figure 4A,B). Only animals that were injected with wild-type primary human TCR+ T cells exhibited clinical signs of GvHD (loss of body weight, changes in fur texture and reduced skin integrity), compared with mice treated with either PBS or CB-010 cells. CB-010 cell infiltration was not detected in any of the examined organs in NBSGW mice at day 30 and day 72. In contrast, wild-type TCR+ T cells were detected in mouse tissues at both time points in lungs, lymph nodes, ovaries, and spleens (Figure 4C). At day 30, wild-type TCR+ T cell levels were highest in spleen and lung, followed by lymph node, and lowest in ovary. By day 72, wild-type T cell levels were highest in spleen, followed by lung, then ovary, and lowest in lymph node; however, high inter-animal variability was noted for all tissues. All mouse tissues from animals euthanized on day 30 were collected for evaluation by immunohistochemistry after staining with an anti-human CD45 antibody (n = 2 for PBS, n = 3 for wild-type TCR+ T cells or CB-010 cells per time point). Histopathological evaluation by microscopy indicated that all tissues exhibited infiltration of wild-type TCR+ T cells, whereas CB-010 did not infiltrate any tissues (supplementary Table 2).
Fig. 4Biodistribution and toxicity profile of CB-010 CAR-T cells in a mouse model of GvHD. Non-irradiated NOD-SCID-gamma-Kit (NBSGW) mice were injected intravenously with a single dose of PBS, wild-type primary human TCR+ T cells (3 × 107), or CB-010 cells (3 × 107). (A) Clinical parameters of GvHD were evaluated on a scale of 0–2 for 72 days and scored: [a] body weight (BW) loss, [b] fur texture changes and [c] reduction in skin integrity. Mouse numbers are indicated on the left. (B) Individual animal BWs in each group were measured for 72 days. (C) Infiltration of T cells into individual tissues from three mice per group was evaluated by qPCR for copies of the human RPP30 gene at day 30 (left panel) and day 72 (right panel). LN, lymph node.
CB-010 exhibited extended antitumor activity in an established xenograft model relative to equivalent CAR-T cells without a PDCD1 KO
To understand the impact of the PDCD1 KO, CB-010 cells were evaluated in comparison to CAR KI cells in a 4-day established xenograft model of a B cell tumor expressing PD-L1 (NALM-6/PD-L1) at a dose of 5 × 106/mouse in the study, with treatment 4 days after tumor engraftment. The outcome of the comparison was that CB-010 cells provided a survival advantage relative to CAR KI cell treatment (supplementary Figure 4B). A single dose of CB-010 cell treatment in this minimally established orthotopic tumor xenograft model led to a statistically significant extension of survival of the mice compared with CAR KI cell treatment or negative controls, without a significant impact on mouse body weight (supplementary Figure 4C). We next evaluated a more established version of this orthotopic tumor xenograft model, where the NALM-6/PD-L1 tumor cells were engrafted for 23 days to increase the tumor cell bulk and metastatic burden before CAR-T cell treatments to evaluate tumor eradication potential and the durability of antitumor activity (Figure 5A). Animals across the different groups had comparable tumor burden at day 0 (start of treatment). In this study, a negative control T cell group that expressed a CAR without a targeting domain (non-targeting CAR KI cells) was included. CB-010 treatment led to enhanced survival at 160 days of measurement compared with CAR KI treatment (Figure 5B), without an impact on animal body weight for CB-010 treated mice (Figure 5E). All animals dosed with CAR KI cells were euthanized due to paralysis or because they were found dead by day 133, and all mice in the PBS- and non-targeting CAR KI cell-treated groups were euthanized due to paralysis by day 27. Tumor burden, measured by bioluminescent imaging and quantitative intensity, was both visually and quantitatively eradicated after either CB-010 or CAR KI treatments by day 7 (Figure 5C,D). Following dosing with CB-010 or CAR KI cells (1 x 107/mouse), animals showed significant reductions in tumor burden, which remained low until day 44 as assessed by imaging and bioluminescent intensity, compared with animals dosed with PBS or non-targeting CAR KI cells (1 x 107/mouse), which had all been found moribund or were euthanized by that time point. Evaluation of the immune cell compartments (blood, bone marrow, and spleen) did not demonstrate significant differences in the kinetics of CB-010 or CAR KI cells in circulation, suggesting potential differences in functional fitness between the two CAR-T cell preparations (Figure 5F).
Fig. 5PDCD1 knockout in allogenic CD19-directed CB-010 CAR-T cells promotes persistent survival in a well-established B cell tumor xenograft model. (A) Study design. Nine NOD-SCID-gamma (NSG) mice per group were engrafted orthotopically with NALM-6/PD-L1 tumor cells 23 days before tail vein injection of either PBS, non-targeting CAR KI cells (site-specifically inserted CAR lacking an scFv and lacking a PDCD1 KO), CAR KI (site-specifically inserted CD19 CAR and lacking a PDCD1 KO), or CB-010 cells. Bioluminescence imaging was collected up to day 108, and survival was followed until day 160. (B) Kaplan–Meier survival curves of the mice in the study to day 160. Comparison between CAR KI and CB-010, p < 0.0001. (C) Images generated by IVIS optical imager with color saturation bar. All images were taken with maximal saturation. (D) Bioluminescent intensity (photons/s) was measured for all groups to day 98. (E) Body weights of the mice over the course of the study. (F) Pharmacokinetics of CB-010 cells compared with CAR KI cells measured by droplet digital PCR in mouse tissues as indicated.
Single-cell transcriptional profiling reveals enhanced activation and metabolic fitness signature in CB-010 relative to equivalent CAR-T cells without a PDCD1 KO
To characterize potential differences between CB-010 cells and CAR KI cells on a molecular level, we used the 10x Genomics single-cell RNA sequencing platform to transcriptionally profile CAR-T cells isolated from of an orthotopically engrafted CD19+ NALM-6/PD-L1 tumor xenograft model at an early time point (day 7) and a late time point (day 60 and day 54 for CB-010 and CAR KI, respectively). Overall, a total of 33 462 cells were captured from both the infusion product (IP) and the CAR-T cells isolated in vivo from the multiple time points after QC filtering. In silico detection of the CAR transcript matched flow-based detection of the CAR, and CAR+ cells were detectable in all samples, with an increase in CAR+ cell proportion in both CB-010 and CAR KI in vivo (supplementary Figure 6A,B). Clustering analysis revealed that CAR+ and CAR− cells did not cluster distinctly from each other, except for a subset of CAR− CD4+ T cells at the late CAR KI sample (Figure 6A ; supplementary Figure 6A). CB-010 and CAR KI IPs also did not cluster independently from CAR-T cells from in vivo isolation at early and late time points (Figure 6A). Identification of CD4+ T cells, CD8+ T cells, and NK-like cell subtypes was determined based on surface protein and RNA expression of canonical markers (Figure 6B; supplementary Figure 6C). CD4+ and CD8+ T cells were further classified into memory, effector, and regulatory subtypes, including stem cell memory T cell (Tscm), central memory T cell (Tcm), effector memory T cell (Tem), and regulatory T cell (Treg) (supplementary Figure 6D,F). CB-010 and CAR KI IPs contained similar distribution of these cell subtypes (Figure 6A; supplementary Figure 6E). Interestingly, although CAR KI and CB-010 IPs contained roughly equivalent proportions of CAR+ Treg cells among CD4+ T cells (42.0% vs 40.3%), at the early time point CAR KI samples maintained nearly double the number of CAR+ Treg cells compared with CB-010 samples (13.2% vs 7.2%, Figure 6C). Furthermore, at the late time point, CB-010 CAR+ cells had a substantially higher proportion of overall CD4+ T cells (43.4% vs 14.7%), lower proportion of Treg cells (1.3% vs 2.5%) and higher proportion of effector memory CD4+ T cells (96.4% vs 83.3%) than CAR KI CAR+ cells (Figure 6A,C). Interestingly, by the late time point, CB-010 and CAR KI CAR+ CD8 T cells did not show substantial phenotypic divergence and were primarily Tem cells (98.6% vs 99.4%, Figure 6D). However, when comparing the CB-010 and CAR KI CAR+ CD8+ Tem cells at both the early and late time points, CB-010 cells displayed lower exhaustion and dysfunction signatures [
], both as full modules and on a single-gene basis (Figure 6E,F; supplementary Table 1). Furthermore, CB-010 CD8+ Tem cells scored higher for both glycolysis and oxidative phosphorylation signature scores [
Fig. 6CB-010 effector T cells in vivo display higher activation, reduced exhaustion and enhanced metabolic fitness signatures relative to equivalent CAR-T cells without a PDCD1 KO. (A) Uniform manifold projection (UMAP) plots of CB-010 and CAR KI cells isolated from either infusion product (IP) or various in vivo time points (early: d7, late: d54, CAR KI; d60, CB-010). Data are visualized as either all cells colored by their respective samples or as density plots showing the enrichment of CAR+ cells in the UMAP space. (B) UMAP plots of all CB-010 and CAR KI cells, colored by their assigned cell identities. (C-D) Bar plots demonstrating the indicated CD4+ T cell (C) or CD8+ T cell (D) subset proportions within CAR+ CD4+ T cells (C) or CAR+ CD8+ T cells (D). Samples are separated by CB-010, CAR KI, and time point. (E) Violin plots visualizing the scaled module scores for gene sets representing previously described T cell exhaustion and activation signatures. Significance of the differences in module scores (χ2, P) was determined with a non-parametric Kruskal–Wallis test. (F) Violin plots visualizing the expression of the indicated markers within CAR+ Tem CD8+ T cells from the indicated samples and time points. The cell surface marker (ADT) expression is displayed along the top row, and the imputed RNA (iRNA) expression is indicated along the bottom row. Feature barcoding was not included in the Late CB-010 time point due to low cell input. (G) Violin plots visualizing the scaled module scores for gene sets representing metabolic pathway signatures. Significance of the differences in module scores (χ2, P) was determined with a non-parametric Kruskal–Wallis test. OXPHOS, oxidative phosphorylation.
CB-010 exhibited enhanced respiratory fitness relative to equivalent CAR-T cells without a PDCD1 KO
To evaluate further the potential differences between CB-010 cells and CAR KI cells with respect to differences in glycolytic/oxidative phosphorylation fitness, we stimulated the CAR-T cells with NALM-6/PD-L1 cells for 18 days in vitro. Subsequently, the cells were evaluated for ATP production, revealing an enhancement in ATP production by CB-010 cells compared with that by CAR KI cells (Figure 7A). This suggests that CB-010 cells exhibit increased respiration, eliciting higher rates of ATP production. To better understand the oxygen consumption rate of the two CAR-T cell preparations, we studied the effect of mitochondrial inhibitors as a glycolytic stress evaluation of respiration through the oxidative phosphorylation pathway (Figure 7B,C). CB-010 cells exhibited a higher respiratory capacity and extracellular acidification rate compared with CAR KI cells following inhibition of the mitochondrial proton gradient necessary for ATP production. A similar effect was observed after only 7 days of co-incubation of the CAR-T cells with NALM-6/PD-L1 cells (supplementary Figure 7). Overall, antigenic stimulation of the CAR-T cells in the context of PD-L1 led to a statistically significant overall increase in spare respiratory capacity in CB-010 cells compared with CAR KI cells (Figure 7D).
Fig. 7PDCD1 knockout modulates metabolic fitness of CB-010 CAR-T cells. (A) ATP production rate (mitochondrial ATP production rate compared with glycolytic ATP production rate) in CB-010 cells compared with CAR KI cells after antigenic stimulation with NALM-6/PD-L1 cells. (B) Extracellular acidification rate (ECAR) kinetics evaluated after treatment of cells with oligomycin A (ATP synthase inhibitor), followed by BAM15 (mitochondrial uncoupler), then Rotenone (electron transport complex I inhibitor) and Antimycin A (electron transport complex III inhibitor). (C) Oxygen consumption rate (OCR) kinetics evaluated after treatment of cells with oligomycin A (ATP synthase inhibitor), followed by BAM15 (mitochondrial uncoupler), then Rotenone (electron transport complex I inhibitor) and Antimycin A (electron transport complex III inhibitor). (D) Spare respiratory capacity (the difference between the maximal OCR signal and the basal OCR signal) of CB-010 cells compared with CAR KI cells following antigenic stimulation with NALM-6/PD-L1 cells. Data represent n = 5 replicates for CB-010 cells and n = 6 replicates for CAR KI cells, respectively. CB-010 and CAR KI cells were engineered from one individual donor. Error bars represent average ± SD. *p < 0.01 by unpaired t test between CB-010 and CAR KI cells.
]. The initial response rates have exceeded those of other targeted therapies including monoclonal antibodies, bispecific antibodies, antibody drug conjugate therapies and small molecule approaches [
]. However, there remains a significant unmet need with respect to the reliability of customized therapies that require (i) the integrity of the starting T cells derived from patients, (ii) the time to production and then infusion while a patient's disease burden worsens, (iii) the duration of response, (iv) the cell dose required for efficacy and (v) the uniformity of therapy across patients when using lentiviral transduction [
]. Such issues could be addressed by allogeneic versions of CAR-T cells that contain the features of autologous approaches but include elements that may provide more uniformity for patients, rapid availability, increased health of starting cell material and the opportunity to include armoring strategies that improve persistence, either at the level of antitumor activity or retention in blood and tissues postinfusion [
Integrating CAR T cell therapy and transplantation: comparisons of safety and long-term efficacy of allogeneic hematopoietic stem cell transplantation after CAR T cell or chemotherapy-based complete remission in B cell acute lymphoblastic leukemia.
]. In this study, we developed CB-010, an allogeneic CAR-T cell therapy that employs site-directed insertion of a CD19-directed CAR into the TRAC locus as a means of eliminating the expression of the native TCR and promoting consistent expression of the CAR, and a knockout of the PDCD1 gene that eradicates expression of PD-1 to prevent exhaustion through the PD-1/PD-L1 axis.
Our observation of improved antitumor activity of CB-010 in an established preclinical xenograft model expressing PD-L1 compared with identical cells that lack the PDCD1 edit (CAR KI) confirmed a critical role for the PD-1 KO. The results demonstrated that well-established and metastatic B cell tumors reflecting the clinical setting were eradicated with long-term efficacy, suggesting that clinical application of CB-010 may have an impact on both the therapeutic index and sustained antitumor duration at the appropriate dose. Single-cell RNA-Seq analyses demonstrated that CB-010 cells were distinguished from CAR KI (PD-1+) cells by exhibiting fewer Treg cells and lower exhaustion phenotypes [
]. Further, the CB-010 CAR-T cells exhibited higher activation, glycolytic and OXPHOS signatures, indicative of enhanced fitness. The evaluation of CB-010 by metabolic profiling with respect to energy reserves, as shown in other systems [
], indicated that CB-010 retains an advantage over CAR KI cells in terms of OXPHOS and/or fatty acid oxidation, possibly by inducing enhanced oxygen consumption through electron transport complex IV. Further, the CB-010 CAR contains the 4-1BB intracellular signaling domain, which others have shown results in mitochondrial fusion, mitochondrial biogenesis and ultimately leads to increased T cell spare respiratory capacity [
]. Previous studies have shown that CAR-T cells that secreted a PD-1 protein (sPD-1) as a PD-1/PD-L1 blockade strategy did not differ from similar cells that lacked expression of sPD-1 with respect to glycolytic energy reserves [
]. These data suggest that the intrinsic blockade through the PD-1 KO in CB-010 cells has a direct metabolic impact on cellular function that is not induced by a pre-existing interaction between the PD-1 receptor and its ligand(s). Our data suggest that antigen-driven stimulation leads to PD-1 expression and subsequent mobilization of energy stores in CAR KI cells. Early efforts showed that PD-1 signaling reduced glucose metabolism through AKT activation and blockade of PI3K signal transduction [
], suggesting that CB-010 may retain an advantage over CAR KI cells though intracellular signaling. Importantly, we did not observe differences in the biodistribution of CB-010 versus CAR KI cells in hematologic tissues from heavily tumor-burdened mice, suggesting that exhaustion dynamics [
] promoted differences in durable antitumor activity.
Earlier preclinical CAR-T studies evaluated disruption of the PD-1/PD-L1 axis using a variety of modalities, including the expression of dominant-negative PD-1 proteins, shRNA to knock down the PD-1 expression, PD-1 knockout with genome editing technologies, CAR-T cell secretion of soluble anti-PD-1 antibody fragments and co-administration of anti-PD-1 monoclonal antibodies [
]. The impact in vivo varied with respect to the degree of checkpoint blockade, but there has generally been a consistent antitumor promoting effect across these different studies, further validating the approach to limiting exhaustion in T cells despite the limited duration of effect. In the context of epithelial tumors, the results have been equivocal [
]. Interestingly, disrupting other checkpoints such as LAG-3 and CTLA-4 in CAR-T cells have been unsuccessful in promoting improved antitumor activity [
], further confirming the dominance of PD-1 in T cell functional downregulation. Here we demonstrated multiple potential mechanisms of the PD-1 KO relating to a reduction in exhaustion and an enhancement of metabolic function and fitness.
Several elements of our studies distinguish CB-010 relative to previous work: (i) our use of the Cas9 chRDNA technology to achieve >95% editing at the PDCD1 locus with low levels of translocation events resulted in CB-010 cells that were almost uniformly PD-1 negative and thereby exhibit profound effects. This contrasts with previous clinical evaluation of a PD-1 KO in autologous TCR-T cells, where only ∼25% of the cells harbored the checkpoint edit [
]; (ii) the safety profile of the previous autologous ACT with a PD-1 knockout was promising, given that the cells were detectable in the patients 9 months’ post-infusion without long-term adverse events [
]. Our data provide molecular phenotypic evidence that PD-1 disruption has impactful changes to the T cells that enable them to resist exhaustion through the PD-1/PD-L1 axis, likely in part due to enhanced energetic fitness. However, previous studies did not address the potential mechanism of enhanced antitumor activity [
]; (iii) the intrinsic capability of CB-010 cells to directly disrupt the PD-1/PD-L1 axis provides an advantage over the exogenous administration of anti-checkpoint monoclonal antibodies. Such antibody treatments have increased potential to elicit toxicity, necessitate repeated dosing, disrupt receptor/ligand interactions dependent on the half-life and avidity of the antibody and function through target receptor dynamics such as internalization and shedding. Clinical outcomes have not supported this combination therapy approach; (iv) we demonstrate that CB-010 is highly active in vivo against multiple B-NHL subtypes; and (v) we have shown in a preclinical model of GvHD that CB-010, when administered to immunodeficient mice at high doses, did not induce toxicity or infiltrate tissues long-term, enabling the evaluation of the product candidate in the clinic. For these reasons, a therapeutic such as CB-010, with the potential to directly and genetically resist exhaustion though a key checkpoint pathway, particularly in the B-NHL setting where the inhibiting ligand is expressed on myeloid-derived immune cells, the endothelial lining of the lymphatic architecture [
]. To our knowledge, this is the first clinical evaluation of an allogeneic CAR-T cell product candidate with a PD-1 knockout. The trial includes the evaluation of multiple histologic subtypes of B-NHL, including those where PD-L1 expression correlates with poor survival [
]. Interestingly, in a study of an autologous CD19 CAR-T cell therapy, the use of Yescarta demonstrated that patients with B-NHL with PD-L1–negative tumors had better outcomes that those with PD-L1–positive tumors [
], suggesting the importance of the PD-1/PD-L1 axis in CD19-directed CAR-T cell function. In ANTLER we observed that all six patients in the first cohort, where a dose of 40 million CB-010 CAR+ cells was administered, achieved a complete response as their best response [
Nastoupil LJ, O'Brien S, Holmes HE, et al. First-in-human trial of CB-010, a CRISPR-edited allogeneic anti-CD19 CAR-T cell therapy with a PD-1 knock out, in patients with relapsed or refractory B cell non-Hodgkin lymphoma (ANTLER study). Congress Eur Hematol Assoc 2022; poster 3103.
]. These results indicate that the PD-1 KO may play a significant role in the ability of these allogeneic CAR-T cells to impact r/r B-NHL disease. Safety outcomes from the ongoing ANTLER clinical trial indicate that the PD-1 KO did not adversely affect patients beyond the expected CAR-T treatment-induced adverse events [
Nastoupil LJ, O'Brien S, Holmes HE, et al. First-in-human trial of CB-010, a CRISPR-edited allogeneic anti-CD19 CAR-T cell therapy with a PD-1 knock out, in patients with relapsed or refractory B cell non-Hodgkin lymphoma (ANTLER study). Congress Eur Hematol Assoc 2022; poster 3103.
]. ANTLER will help validate the impact of the PD-1 KO in these allogeneic CAR-T cells and provide confirmation of the preclinical evidence that durability and therapeutic index may be improved relative to allogeneic CAR-T cells that lack this checkpoint edit.
Declaration of Competing Interest
All authors are current or former employees of Caribou Biosciences, Inc.
Funding
This work was funded by Caribou Biosciences.
Author Contributions
Conception and design of study: EL, GK, PDD, RTD, JS, EG, MvO and SBK. Acquisition of data: EL, GK, TWF, BS, PDD, RTD, SM, SCC, CW, LB, MI, LE, MS, BK, GWJL, SCS, RD and MvO. Analysis and interpretation of data: EL, GK, TWF, BS, PDD, RTD, SM, SCC, CW, LB, MI, LE, MS, BK, GWJL, SCS, RD, SG, CKF, JS, EG, MvO and SBK. Drafting or revising the manuscript: EL, GK, TWF, RTD, MB, SCC, LE, GWJL, SCS, RD, CKF, JS, EG, MvO and SBK. All authors have approved the final article.
Acknowledgments
The authors thank Arthur Owen for assay and analysis infrastructure development, as well as assistance with experimental design and custom analyses. They also thank Nelle Cronen for the animation graphics.
Integrating CAR T cell therapy and transplantation: comparisons of safety and long-term efficacy of allogeneic hematopoietic stem cell transplantation after CAR T cell or chemotherapy-based complete remission in B cell acute lymphoblastic leukemia.
Nastoupil LJ, O'Brien S, Holmes HE, et al. First-in-human trial of CB-010, a CRISPR-edited allogeneic anti-CD19 CAR-T cell therapy with a PD-1 knock out, in patients with relapsed or refractory B cell non-Hodgkin lymphoma (ANTLER study). Congress Eur Hematol Assoc 2022; poster 3103.
00091-9&id=gr1.jpg){kind=link}
00091-9&id=gr2.jpg){kind=link}
00091-9&id=gr3.jpg){kind=link}
00091-9&id=gr4.jpg){kind=link}
00091-9&id=gr5.jpg){kind=link}
00091-9&id=gr6.jpg){kind=link}
00091-9&id=gr7.jpg){kind=link}