Rewiring endogenous genes in CAR T cells for tumour-restricted payload delivery

Human samples

Buffy coats from healthy donors less than 35 years old were obtained from the Red Cross with informed consent, as approved by the Red Cross and the Peter MacCallum Cancer Centre (PMCC) Human Research and Ethics Committee under HREC#01/14. Frozen apheresis samples were collected from one multiple myeloma patient and six DLBCL patients enrolled in CAR T cell clinical trials. All patients gave informed consent, in accordance with the PMCC Human Research and Ethics Committee under HREC/74245/PMCC.

Animal models

OT-I and C57BL/6 human-Her2 (hHer2) transgenic mice^50,51 were bred at PMCC. C57BL/6 wild-type mice were purchased from the Walter and Eliza Hall Institute or Australian Bioresources. The Ly5.1 congenic mice and NOD.Cg-Prkdc scid IL2rg (NSG) mice were either bred at PMCC or purchased from Australian Bioresources. The OT-3 mice were bred at the Peter Doherty Institute. All murine experiments were done with mice 6–18 weeks of age and housed in a PC2 specific pathogen-free animal facility, in accordance with the PMCC Animal Experimentation Ethics Committee under projects #E582, #E664, #E671 and #E693, and a minimum of three mice per group were used in each experiment. Mice were randomized before treatment according to tumour size to ensure that all groups had equivalent tumour burdens before therapy. Experiments were not blinded because the same investigators performed and analysed experiments, so blinding was not possible. All experiments complied with the ethical endpoints stated in the approved projects, including maximum tumour size.

Cell lines

All murine tumour cell lines were from a C57BL/6 background. The murine breast carcinoma cell lines AT-3 and E0771 were obtained from T. Stewart (PMCC) and R. Anderson (Olivia Newton-John Cancer Centre), respectively. The murine MC38 colon adenocarcinoma cell line and 24JK sarcoma cell line were provided by J. Schlom (National Institutes of Health) and P. Hwu (National Institutes of Health), respectively. The parental tumour cell lines were retrovirally transduced with a murine stem cell virus vector to obtain hHer2- and ova-expressing tumour cell lines. The human ovarian cancer cell line OVCAR-3 and breast cancer cell lines MCF-7 and MDA-MB-231 were obtained from the American Type Culture Collection. The retroviral packaging lines GP+e86 and PG13, as well as HEK293T, were obtained from the American Type Culture Collection. All cell lines were confirmed to be mycoplasma-negative by polymerase chain reaction (PCR)-based testing.

The E0771, MC38, 24JK, OVCAR-3, MCF-7, GP+e86 and PG13 cell lines were cultured in Roswell Park Memorial Institute (RPMI) 1640 medium supplemented with 10% heat-inactivated fetal bovine serum (FBS), 1 mM sodium pyruvate, 2 mM glutamine, 0.1 mM non-essential amino acids, 10 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES), 100 U ml⁻¹ penicillin and 100 µg ml⁻¹ streptomycin (complete RPMI), and maintained in a humidified incubator at 37 °C with 5% CO₂. The AT-3, MDA-MB-231 and HEK293T cell lines were cultured in Dulbecco’s Modified Eagle Medium (DMEM) supplemented with 10% heat-inactivated FBS and maintained in a humidified incubator at 37 °C with 10% CO₂.

Antibodies and cytokines

Murine anti-CD3 (clone 145-2C11) and anti-CD28 (clone 37.51) antibodies were purchased from BD Biosciences. The human anti-CD3 antibody (clone OKT3) was purchased from BioLegend. The anti-Myc tag antibody (clone 9B11) was purchased from Cell Signaling Technology. The anti-LeY idiotype antibody hu3S193 was provided by A. Scott (Olivia Newton-John Cancer Centre)⁵². The recombinant human IL-2 (hIL-2) was obtained from the National Institutes of Health or purchased from PeproTech and Miltenyi Biotec. The recombinant murine IL-7 (mIL-7) and human IL-15 (hIL-15) were purchased from PeproTech.

CRISPR/Cas9 editing of primary T cells

Murine T cells were activated from naive splenocytes by culturing in complete RPMI containing murine anti-CD3 (0.5 µg ml⁻¹), murine anti-CD28 (0.5 µg ml⁻¹), hIL-2 (100 IU ml⁻¹) and mIL-7 (200 pg ml⁻¹) for 24 h. Human T cells were activated by isolating peripheral blood mononuclear cells from healthy buffy coats and culturing in complete RPMI containing human anti-CD3 (30 ng ml⁻¹) and hIL-2 (600 IU ml⁻¹) for 48 h. To perform CRISPR/Cas9 editing, 37 pmol recombinant Cas9 (IDT) and 270 pmol single guide RNA (sgRNA; Synthego) were combined and incubated at room temperature for 10 min to form Cas9/sgRNA ribonucleoprotein (RNP) complexes. Then, 20 × 10⁶ murine T cells or 1 × 10⁶ human T cells were resuspended in 20 µl P3 electroporation buffer (containing 82% P3 buffer and 18% Supplement 1; Lonza), combined with RNP and electroporated in 20-µl cuvettes using a 4D-Nucleofector X Unit (Lonza) with pulse code CM137 for murine T cells and EO115 for human T cells. For CRISPR-mediated knockout, 100 μl of prewarmed media was immediately added, and T cells were incubated for 10 min at 37 °C before being transferred to an appropriate culture plate. Knockout efficiency was determined by PCR amplification of regions more than 150 bp around the sgRNA cut site in both mock and CRISPR-edited cells, Sanger sequencing of the PCR amplicons and analysis of sequencing data using the Synthego ICE analysis tool (https://ice.editco.bio/#/). For CRISPR-mediated knock-in, T cells were immediately washed out with prewarmed media to a concentration of 50 × 10⁶ cells per millilitre and added to an appropriate culture plate containing a mixture of AAV6 at a multiplicity of infection (MOI) of 10,000–100,000 and 2 µM M3814 (MedChemExpress). T cells were incubated at 37 °C for 4 h before AAV6 and M3814 were washed off and downstream protocols were performed. The sgRNA sequences used are included in Supplementary Table 4. Homologous repair templates were manufactured and cloned by NotI digest into pAAV-MCS (Agilent Technologies) by Genscript, and the resulting plasmids were packaged into AAV6 vectors by Vigene Biosciences (now Charles River Laboratories) or PackGene Biotech. The pAAV-MCS was provided by V. Wiebking and M. Porteus (Stanford University), and the homologous repair template sequences used are included in Supplementary Table 5.

Generation of murine and human CAR T cells

Retroviral supernatants were collected from the GP+e86 or PG13 packaging line for transduction of murine T cells with an anti-hHer2 CAR or human T cells with an anti-LeY CAR as previously described^53,54,55. For the generation of GP+e86 or PG13 packaging lines encoding both an anti-hHer2 CAR or anti-LeY CAR and an NFAT promoter⁵⁶ inducing GFP or IL-12 expression, NFAT–GFP or NFAT–IL-12 sequences were cloned into the murine stem cell virus vector encoding a truncated human nerve growth factor receptor, the vector was transfected into GP+e86 or PG13 packaging lines encoding an anti-hHer2 CAR or anti-LeY CAR, and the resulting packaging lines were sorted on nerve growth factor receptor by flow cytometry. Lentiviral transduction was used for the generation of human anti-Her2 CAR T cells. In brief, lentiviral packaging plasmids (pCMV-VSV-G, pMDLg/pRRE, pRSV-Rev) and a plasmid encoding an anti-Her2 CAR (Genscript) were transfected into HEK293T cells. The resulting lentiviral supernatants were collected on three consecutive days, pooled and centrifuged with Lenti-X Concentrator (Takara Bio) to concentrate the lentivirus. Concentrated lentivirus was directly added to human T cells at a MOI of 0.5 with Lentiboost (Sirion) for transduction. Following transduction, murine T cells were maintained in media containing hIL-2 (100 IU ml⁻¹) and mIL-7 (200 pg ml⁻¹) for in vitro assays or mIL-7 (200 pg ml⁻¹) and hIL-15 (10 ng ml⁻¹) for in vivo applications, and human T cells were maintained in media containing hIL-2 (600 IU ml⁻¹).

In vitro co-culture/stimulation assay

Murine and human T cells were co-cultured with tumour cells at an effector to target (E:T) ratio of 1:1 for 24 h before collection of supernatants and flow cytometry analysis of T cells. For 72-h chronic stimulation assays, supernatant was completely removed at the 24 h and 48 h timepoints and an equivalent number of fresh tumour cells were added. For T cell stimulation with anti-CD3 (0.5 µg ml⁻¹), anti-CD28 (0.5 µg ml⁻¹), anti-Myc tag antibody (1:1,000) or anti-LeY idiotype antibody (4.5 μg ml⁻¹), a U-bottomed 96-well plate was coated with 100 µl PBS containing the appropriate dilutions of antibody at 37 °C for 2 h before wells were washed twice with 200 µl PBS and T cells were added.

Chromium-51 release assay

Tumour cells were labelled with ⁵¹Cr by resuspending cell pellets in 50 µCi ⁵¹Cr per 1 × 10⁶ cells and incubating at 37 °C for 1 h. Next, 1 × 10^{4 51}Cr-labelled tumour cells were co-cultured with T cells at the indicated E:T ratios. As controls for background and total ⁵¹Cr levels, tumour cells were cultured either alone or with 5% Triton X-100 (Sigma-Aldrich). After a 16-h incubation, the ⁵¹Cr level in the supernatant was measured using the automatic gamma counter Wallac Wizard 1470 (PerkinElmer), and T cell killing was quantified using the following formula: [⁵¹Cr CPM (sample) − ⁵¹Cr CPM (background)]/[⁵¹Cr CPM (total) − ⁵¹Cr CPM (background)], where CPM stands for counts per minute.

Incucyte killing assay

Tumour cells expressing mCherry or GFP were co-cultured with T cells in a 384-well black, optically clear flat-bottomed plate (PerkinElmer) at the indicated E:T ratios. Plates were imaged using the Incucyte SX5 Live-Cell Analysis System every 4 h. The assay was run using the ‘adherent cell-by-cell’ scan type, using a 10× objective lens, with acquisition times of 400 ms and 300 ms for the red and green channels, respectively.

Analysis of cytokine production

Supernatants from the in vitro co-culture and stimulation assays and serum samples from mice were analysed for cytokine concentration using BD Cytometric Bead Array Flex sets for IFNγ, TNF, IL-2, IL-12/IL-23p40 (human) and IL-12p70 (murine) (BD Biosciences). Data were acquired using FACSVerse, FACSCanto II or LSR II (BD Biosciences) and analysed using the FCAP Array v.3 software (BD Biosciences).

Flow cytometry

Cells were incubated in FACS buffer (2% FBS, 2 mM EDTA in PBS) containing a 1:50 dilution of Fc receptor block (2.4G2 antibody, produced in-house) with fluorochrome-conjugated antibodies at 4 °C for 30 min in the dark. After staining, cells were washed twice with FACS buffer before analysis or intracellular staining. For cytoplasmic and intranuclear staining, cells were fixed and permeabilized using a BD Cytofix/Cytoperm Fixation Permeabilization Kit (BD Biosciences) or an eBioscience Foxp3/Transcription Factor Staining Buffer Set (Thermo Fisher Scientific) following the manufacturer’s instructions, respectively, before staining with fluorochrome-conjugated antibodies at room temperature for 30 min in the dark. Cells were then washed twice with 1× BD Perm/Wash Buffer or eBioscience Permeabilization Buffer (diluted from 10× stock) before analysis. Data were acquired on either a FACSCanto II, LSRFortessa X-20, LSR II, FACSymphony A3 or A5 (BD Biosciences), and analysed using FlowJo software (BD Biosciences). Cells were quantified using counting beads (Beckman Coulter) using the following formula: cell events of interest/bead events × number of beads per sample. FC values calculated from negative mean fluorescence intensity values were set to 0.

Adoptive-transfer experiments

For murine adoptive-transfer experiments, C57BL/6 WT, hHer2 transgenic or Ly5.1 mice were inoculated with 5 × 10⁵ AT-3-ova or 2 × 10⁵ E0771-hHer2 in the fourth mammary fat pad orthotopically or 4 × 10⁵ MC38-hHer2 subcutaneously. Tumours were allowed to establish for eight days (AT-3-ova, MC38-hHer2) or six days (E0771-hHer2) before mice were preconditioned with 0.5 Gy (for AT-3-ova and MC38-hHer2) or 4.0 Gy (E0771-hHer2) total body X-ray irradiation. For human adoptive-transfer experiments, NSG mice were inoculated with 1.25 × 10⁶ MDA-MB-231 in the fourth mammary fat pad orthotopically, or 5–6 × 10⁶ OVCAR-3 subcutaneously. Tumours were allowed to establish for 7 days (MDA-MB-231) or 10–16 days (OVCAR-3) before preconditioning with 1 Gy total body X-ray irradiation. Mice were intravenously injected with 0.1–1.5 × 10⁷ OT-I cells or 2 doses of 0.1–1.0 × 10⁷ murine or human CAR T cells on consecutive days and intraperitoneally injected with 5 doses of hIL-2 (25,000 IU per dose) on consecutive days. For experiments in NSG mice using IL-12-engineered T cells, all T cell groups were edited to disrupt TCR expression by CRISPR/Cas9-mediated TRAC knockout to minimize the risk of graft-versus-host disease.

For tumour growth experiments, tumour area was measured using callipers every 2–4 days until all mice reached an ethical end point. The ethical end point for tumour size was 150 mm².

Ex vivo analysis of immune cells

For ex vivo flow cytometry analyses of the tumour, spleen, tumour dLNs, liver, brain, lung, kidney and bone marrow, tissues were collected on days 7–9 for OT-I and murine CAR T cell experiments, and on days 7–14 for human CAR T cell experiments, unless indicated otherwise. Tumours were processed by mechanical dissociation followed by enzymatic digestion with serum-free DMEM containing 1 mg ml⁻¹ collagenase type IV (Sigma-Aldrich) and 0.02 mg ml⁻¹ DNAse I (Sigma-Aldrich) at 37 °C for 30 min with gentle shaking, then filtered through a 70-µm filter followed by a 35-µm filter before staining. Spleens were processed by macerating and filtering through a 70-µm filter, red blood cells were lysed with ACK lysis buffer, then samples were filtered through a 35-mm filter before staining. Tumour dLNs were processed by macerating and filtering through a 70-µm filter mesh before staining. Livers and brains were processed by macerating and filtering through a 70-µm filter, immune cells were isolated following density gradient centrifugation in 33% Percoll at 2,000 rpm for 12 min, red blood cells were lysed with ACK lysis buffer, then samples were filtered through a 35-µm filter before staining. Lungs and kidneys were processed by mechanical dissociation followed by enzymatic digestion with serum-free DMEM containing 1 mg ml⁻¹ collagenase type IV (Sigma-Aldrich) and 0.02 mg ml⁻¹ DNAse I (Sigma-Aldrich) at 37 °C for 30 min with gentle shaking, filtered through a 70-µm filter, red blood cells were lysed with ACK lysis buffer, then samples were filtered through a 35-µm filter before staining. Bone marrow was processed by using a needle and syringe to flush the inner cavity of the femur with FACS buffer, red blood cells were lysed with ACK lysis buffer, then samples were filtered through a 35-µm filter before staining.

For experiments requiring ex vivo stimulation, samples were stimulated with either 10 ng ml⁻¹ phorbol 12-myristate 13-acetate (PMA) and 1 µg ml⁻¹ ionomycin, 200 nM SIINFEKL peptide or a cocktail of 5 μM MC38 neoantigen peptides (Dpagt1^mut SIIVFNLL, Reps1^mut AQLANDVVL, Adpgk^mut ASMTNMELM) as well as GolgiPlug (BD Biosciences) and GolgiStop (BD Biosciences) at a 1:1,000 or a 1:1,500 dilution, respectively. Samples were incubated at 37 °C for 3 h before staining. For ex vivo co-cultures, tumour cells were seeded at 5 × 10⁴ cells per well in a flat-bottomed 96-well plate the day before tumour collection. After tumour processing, tumour samples were added to the plate and incubated overnight at 37 °C. On the following day, GolgiPlug and GolgiStop were added, and samples were incubated at 37 °C for 3 h before staining.

For flow cytometry analysis of blood, 200 µl of blood was collected by either retro-orbital or submandibular bleeds into Eppendorf tubes containing 10 µl EDTA. Red blood cells were lysed three times using 75 µl ACK lysis buffer before staining. For serum analyses, blood was centrifuged at 10,000g at 4 °C for 15–20 min and serum was collected for downstream analyses.

Immunofluorescence analysis

C57BL/6 hHer2 transgenic mice were subcutaneously engrafted with 4 × 10⁵ MC38-hHer2 for 8 days before the adoptive transfer of 5 × 10⁶ murine anti-hHer2 CAR T cells. Then, 9 days after adoptive transfer, tumours were collected, embedded in OCT compound (Scigen) in a cryomold, stored at −80 °C and sectioned at 10 μm per tissue slide.

Tissue slides were fixed with ice-cold methanol at −20 °C for 20 min, washed twice with FACS buffer at room temperature for 5 min, blocked with 0.2% bovine serum albumin at room temperature for 5 min, and stained at 4 °C overnight with fluorochrome-conjugated antibodies prepared at 1:200 dilution in FACS buffer: anti-CD4 FITC (clone RM4-5) and anti-CD8 Alexa Fluor 594 (clone 53-6.7) from BioLegend. On the following day, tissue slides were washed twice with FACS buffer at room temperature for 5 min, stained with DAPI (Thermo Scientific) at room temperature for 10 min, washed twice with FACS buffer at room temperature for 5 min and coverslipped with VECTASHIELD Antifade Mounting Medium (Vector Laboratories). Immunofluorescence images were acquired using an Olympus DP80 camera on an Olympus BX53 microscope using the cellSens Dimension program and analysed using ImageJ.

3′ bulk RNA-seq analysis

RNA-seq libraries were prepared from RNA samples using the Quant-seq 3′ mRNA-seq Library Prep Kit for Illumina (Lexogen) following the manufacturer’s instructions. Single-end, 75–100 bp RNA-seq short reads were generated by NextSeq sequencing (Illumina) and CASAVA 1.8.2 was subsequently used for base calling. RNA-SeQC v.1.1.7 was used to assess the quality of output⁵⁷, and Cutadapt v.2.1 was used to remove random primer bias and poly-A-tail-derived reads. Sequence alignment against the mouse reference genome mm10 or the human reference genome hg19 was done using HISAT2. Finally, the Rsubread software package 2.10.5 was used to quantify the raw reads of genes defined from Ensembl release 96 (ref. ⁵⁸). Gene counts were normalized using the trimmed means of M-values method and converted into log₂ counts per million using the EdgeR v.3.8.5 package^59,60. Differential gene expression between groups was derived using the quasi-likelihood F-test statistical test method based on the generalized linear model (glm) framework from EdgeR. Principal component analysis was done on normalized counts based on the most-variable genes. Adjusted P-values were computed using the Benjamini–Hochberg method. All differentially expressed genes were classified as significant based on a false discovery rate cut-off of less than 0.05. MA plots were used to represent differential gene expression between groups. Unbiased GSEA was used on a preranked list of differentially expressed genes identified by RNA-seq analysis. GSEA was done against Hallmark and C2 (canonical pathways) curated gene sets from the Molecular Signatures Database. Annotated in Fig. 5f and Extended Data Fig. 9a are genes associated with the C2 IL-12 signalling pathway (Gene set IL12_STAT4_PATHWAY).

scRNA-seq analysis

We analysed an scRNA-seq atlas of publicly available and previously published datasets of tumour-infiltrating, healthy tissue and peripheral blood T cells from 21 different cancer types, following previously described methods³³. Data integration was done by per-cell size-factor normalization and per-gene z-score scaling across cells for each dataset. Each dataset was then partitioned into mini-clusters to reduce noise before batch-correction with the Harmony package⁶¹. Seurat was then used to further cluster integrated datasets into meta-clusters. To determine binarized expression of genes of interest across different cancer types above mean expression, the scale and average.cell functions from the sscVis package (https://github.com/Japrin/sscVis/) was used, and total frequency of cells from each scRNA-seq dataset was plotted as a box-plot. Numbers of patient samples analysed were as follows: n = 1 (blood: BCL; tumour: FTC, OV); n = 2 (blood: BRCA, multiple myeloma, CHOL; tumour: AML, BCL); n = 3 (tumour: multiple myeloma); n = 4 (tumour: SCC, CHOL); n = 7 (tumour: STAD, ESCA); n = 8 (tumour: HNSCC); n = 9 (tumour: UCEC); n = 10 (tumour: RC, THCA, NPC); n = 11 (blood: HCC; tumour: BCC); n = 14 (tumour: BRCA); n = 16 (blood: LC, CRC); n = 17 (tumour: HCC); n = 18 (tumour: CRC); n = 22 (tumour: PACA); n = 26 (tumour: LC); n = 44 (tumour: MELA).

TCR sequencing analysis

RNA was extracted from processed tumour dLN samples and used for TCR sequencing using the RNeasy kit (Qiagen) and the QIAseq Immune Repertoire RNA Library kit (Qiagen) following the manufacturer’s instructions. In brief, RNA was reverse transcribed into cDNA using TCR-specific primers, and unique molecular identifiers were ligated to each double-stranded cDNA molecule. The TCR region was enriched using a set of primers specific to the TCR constant regions and a universal primer complementary to the adaptor. Then, the product was amplified using universal primers that incorporate Illumina sequences and indices. The resulting samples were pooled and sequenced in a MiSeq (Illumina, v.3 chemistry) with paired-end 300-bp reads and a custom sequencing primer (Qiagen). Read processing and analysis were performed using Qiagen’s web resources (GeneGlobe Data Analysis Centre, software version 1.0), and clonotype calls and quantity estimates were generated using the IMSEQ software (http://www.imtools.org).

Statistical analysis

All statistical analyses were performed using Graphpad Prism. The unpaired and paired student’s t-tests were used to determine statistical significance between pairs of data. To determine significance between multiple groups with one independent variable, one-way ANOVA and Tukey’s multiple-comparisons test were used. To determine significance between multiple groups of data with two independent variables, a repeated measures two-way ANOVA with the Geisser–Greenhouse correction and Tukey’s multiple-comparisons test were used. To determine significance between multiple survival curves, the log-rank Mantel–Cox test was used.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.