Vector construction

The NF-κB-specific promoter, named DMP, which consists of five high-affinity NF-κB binding sites and one minimal promoter sequence, was amplified from the pTsgRNA-DMP-Cas9 plasmid constructed previously [22] and was then cloned and inserted into the pAAV-MCS vector (Stratagene) using EcoRI and BamHI restriction enzymes (Thermo Fisher, Waltham, USA) to generate the pAAV-DMP plasmid. The artificially synthesised U6-gRNA sequences consisting of the U6 promoter sequence and gRNA direct repeat spacer separated by BbsI restriction sites were cloned and inserted into the pAAV-DMP vector using MIuI and EcoRI restriction enzymes to generate the pAAV-U6-gRNA-DMP backbone plasmid. The Cas13a coding sequence was amplified from the pET-His6-TwinStrep-SUMO-LwCas13a vector (Addgene, MA, USA) and was then cloned and inserted into the pAAV-U6-gRNA-DMP backbone vector using BamHI and SalI restriction enzymes to generate the pAAV-U6-gRNA-DMP-Cas13a plasmid (referred to as pHOPE). The primers used for pHOPE backbone vector construction are listed in Supplementary Table S1.

The gRNAs targeting NT (no transcript), ZsGreen, mCherry, and human and murine oncogenes, including TERT (telomerase reverse transcriptase), PLK1 (polo-like kinase 1), MYC (myelocytomatosis viral oncogene homologue), and KRAS (Kirsten rat sarcoma viral oncogene homologue), were chosen with CHOPCHOP online software (http://chopchop.cbu.uib.no/). Complementary oligonucleotides containing a 28-bp gRNA target-specific region and two flanking BbsI sites were artificially synthesised, annealed into double-stranded oligonucleotides, and ligated into the pHOPE vector via the Golden Gate Assembly method. The ligation reaction contained 10 units of BbsI (NEB, Ipswich, MA, USA), 1000 units of T4 DNA ligase (NEB), 1 μL of 10 × T4 DNA ligase buffer (NEB), 1 μL of 0.1 mg/mL bovine serum albumin (BSA) (NEB), double-stranded oligonucleotides (1 nM), 50 ng of the pHOPE backbone plasmid, and nuclease-free H2O to 10 μL. Ligation was performed in a thermal cycler as follows: 10 cycles at 37 °C for 5 min, 16 °C for 10 min, then 37 °C for 30 min, and 80 °C for 5 min. The generated plasmids were named pHOPE-NT, pHOPE-ZsGreen, pHOPE-mCherry, pHOPE-hTERT, pHOPE-hPLK1, pHOPE-hMYC, pHOPE-hKRAS, pHOPE-mTERT, pHOPE-mPLK1, pHOPE-mMYC, and pHOPE-mKRAS. One plasmid coexpressing the gRNAs targeting four human oncogenes (TERT, PLK1, MYC, and KRAS) was constructed and named pHOPE-hTPMK. In addition, another plasmid coexpressing the gRNAs targeting four murine oncogenes (TERT, PLK1, MYC, and KRAS), was constructed and named pHOPE-mTPMK. The oligonucleotides used for gRNA target construction are listed in Supplementary Table S2.

The pZsGreen and pmCherry reporter plasmids, in which the expression of ZsGreen and mCherry, respectively, was under the control of the CMV promoter, were maintained in our laboratory. The functional sequences are provided in Supplementary Table S3.

Cell culture and cell transfection

The following cancer cell lines were used in this study: SiHa (human cervical carcinoma cells), BGC823 (human gastric cancer cells), A549 (human non-small cell lung cancer cells), HT-29 (human colorectal adenocarcinoma cells), and CT-26 (mouse colorectal carcinoma cells). The following normal cell lines were used in this study: HL7702 (human normal hepatocytes), GES-1 (human gastric epithelial cells), MRC-5 (human embryonic lung fibroblasts), and HEK-293T (human embryonic kidney cells). SiHa, BGC823, and HEK-293T cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM) (Gibco, California, USA). A549, HT-29, CT-26, HL7702, and GES-1 cells were cultured in Roswell Park Memorial Institute (RPMI) 1640 medium (Gibco). MRC-5 cells were cultured in minimum essential medium (MEM) (Gibco). All media were supplemented with 10% fetal bovine serum (FBS) (Sigma-Aldrich, Missouri, USA), 100 units/mL penicillin and 100 µg/mL streptomycin (Gibco). The cell lines were originally purchased from the Cell Resource Center of the Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences, and were incubated at 37 °C in a humidified incubator with 5% CO2.

For cell transfection, 293T, SiHa, BGC823, A549, HL7702, GES-1, MRC-5, CT-26, and HT-29 cells were seeded into 24-well plates at a density of 1 × 105 cells/well and cultured overnight. Then, the cells were transfected with 500 ng of various plasmids using Lipofectamine® 2000 (Thermo Fisher) according to the manufacturer’s instructions. After being cultured for 4 h, the medium in each well was replaced with 500 μL of fresh DMEM, RPMI 1640 medium or MEM containing 10% FBS. After transfection, the cells were washed with PBS and then stained with acridine orange/ethidium bromide (AO/EB) (Sangon Biotech, Shanghai, China) according to the manufacturer’s instructions; the live cells appeared uniformly green. The cells were observed and imaged with a fluorescence microscope (IX51, Olympus, Tokyo, Japan) and counted with ImageJ2 software.

Cell viability assay

SiHa, BGC823, A549, HL7702, GES-1, MRC-5, CT-26, and HT-29 cells were seeded into 96-well plates at a density of 5 × 103 cells/well and cultured overnight. Then, the cells were transfected with 100 ng of various plasmids using Lipofectamine® 2000. After transfection, 10 μL of Cell Counting Kit-8 (CCK-8) reagent (Yeasen, Shanghai, China) was added to the cells in each well, and the cells were cultured in a 37 °C, 5% CO2 humidified incubator for 1 h. Then, the absorbance of each well was measured at 450 nm with a microplate plate reader (Tecan, Männedorf, Switzerland). Wells without transfected cells were regarded as blank wells, and the percentage of cell viability (%) was calculated as follows: cell viability (%) = [A450 (treated wells) − A450 (blank wells)]/[A450 (control wells) − A450 (blank wells)] × 100%.

Reverse transcription quantitative PCR

293T, CT-26 and HT-29 cells were seeded into 6-well plates at a density of 1.5 × 106 cells/well and cultured overnight. Then, the cells were transfected with 2500 ng of various plasmids using Lipofectamine® 2000. Twenty-four hours posttransfection, total RNA was isolated using TRIzol® reagent (Invitrogen, MA, USA) according to the manufacturer’s instructions. The concentration of RNA was quantified using a NanoDrop 2000 spectrophotometer (Thermo Fisher), and 500 ng of total RNA was reverse transcribed into complementary DNA (cDNA) using Hifair® V One-step RT-gDNA Digestion SuperMix for qPCR (Yeasen) according to the manufacturer’s instructions in a total volume of 20 µL. The mRNA expression levels of ZsGreen, mCherry, TERT, PLK1, MYC, KRAS, and RelA were quantified via quantitative PCR (qPCR) using an Analytic Jena qTOWER3 G thermal cycler. The qPCR mixture contained 10 µL of Hieff UNICON® Universal Blue qPCR SYBR Green Master Mix (2×) (Yeasen), the forward and reverse primers (0.2 µM each), 2 µL of cDNA, and nuclease-free H2O to a final volume of 20 μL. The thermal cycling programme used for qPCR was as follows: 95 °C for 2 min followed by 45 cycles at 95 °C for 10 s, and 60 °C for 30 s. Melting curve analysis revealed a single PCR product. The Ct values of the target genes were normalised by subtracting the Ct values for GAPDH and calculated by the instrument’s software. mRNA expression levels of target genes were calculated as relative quantity (RQ) values: RQ = 2−ΔΔCt, where ΔCt = Cttarget − CtGADPH, and ΔΔCt = ΔCttreatment − ΔCtcontrol. mRNA expression levels of RelA were calculated as 2−ΔCt. Each qPCR analysis was performed with at least three technical replicates. The primers used for qPCR are listed in Supplementary Table S4.

rAAV preparation

HEK-293T cells were added to 75-cm2 flasks at a density of 5 × 106 cells/flask and cultured overnight. Then, the cells were transfected with 4 μg of pAAV-RC (Stratagene), 4 μg of pAAV-Helper (Stratagene), and 4 μg of one pAAV plasmid (pHOPE-mTERT, pHOPE-mPLK1, pHOPE-mMYC, pHOPE-mKRAS, or pHOPE-mTPMK) using Lipofectamine® 2000. Seventy-two hours posttransfection, the cells were purified with the AAVpro® Purification Kit Maxi (TaKaRa) according to the manufacturer’s instructions. The final recombinant AAV (rAAV) products were referred to as rAAV-HOPE-mTERT, rAAV-HOPE-mPLK1, rAAV-HOPE-mMYC, rAAV-HOPE-mKRAS, and rAAV-HOPE-mTPMK.

Animal experiment

Four-week-old female BALB/c mice weighing 18–22 g and purchased from Changzhou Cavens Laboratory Animal Co., Ltd (China), were used in this study. All mice were maintained under specific pathogen-free conditions. Each mouse was subcutaneously injected with 1 × 106 CT-26 cells in the unilateral inner thighs to establish the CT-26 homograft model of colorectal cancer and maintained for tumour formation. On the 8th day after tumour cell implantation, the tumour-bearing mice were randomly divided into five groups—the PBS (n = 8 mice), rAAV-HOPE-mPLK1 (n = 8 mice), rAAV-HOPE-mMYC (n = 8 mice), rAAV-HOPE-mKRAS (n = 8 mice), and rAAV-HOPE-mTPMK (n = 8 mice) groups—and then injected intravenously with PBS or 1 × 1010 vg of rAAV-HOPE-mPLK1, rAAV-HOPE-mMYC, rAAV-HOPE-mKRAS, or rAAV-HOPE-mTPMK every other day for a total of three injections. Body weight and tumour size were measured every other day. The tumour volume (V) was calculated as V = (Dd2)/2, where D and d are the lengths of the major and minor tumour axes, respectively. The mice were euthanized and photographed on the 14th day after the first AAV injection. The tumour and spleen were collected from each mouse and were then weighed and photographed. Serum samples were collected for biochemical analysis, including measurement of alanine transaminase (ALT), aspartate amino transferase (AST), alkaline phosphatase (ALP), and gamma-glutamyl transferase (γ-GT) to assess liver function and urea, creatinine (CREA), and uric acid (UA) to assess kidney function.

Library preparation and scRNA-seq analysis

Fresh tumour tissues collected from mice injected with PBS, rAAV-HOPE-mPLK1, rAAV-HOPE-mMYC, rAAV-HOPE-mKRAS, or rAAV-HOPE-mTPMK were immediately placed in GEXSCOPE Tissue Preservation Solution (Singleron Biotechnology) at 2–8 °C. Then, the tissue samples were dissociated into single-cell suspensions using a Singleron PythoN™ Tissue Separator (Singleron Biotechnology) and sCelLiVE® Tissue Preservation Solution (Singleron Biotechnology) according to the manufacturer’s instructions. Finally, the tissue samples were stained with trypan blue (Sigma-Aldrich), the cell viability was evaluated under a fluorescence microscope (Olympus).

Single-cell suspensions with a concentration of 1 × 105 cells/mL were loaded into microfluidic devices provided with the Singleron Matrix Single-Cell Processing System (Singleron Biotechnology). Then, the scRNA-seq libraries were prepared using GEXSCOPE® Single-Cell RNA Library Kits (Singleron Biotechnology, #5180011) according to the manufacturer’s instructions, diluted to 4 nM and pooled for sequencing. All libraries were subjected to 150 bp paired-end sequencing on an Illumina NovaSeq 6000 instrument.

Processing and scRNA-seq data analysis

The Singleron CeleScope pipeline was used to analyse the raw reads and generate a gene count matrix with default parameters [47]. Briefly, the raw data in FASTQ format were mapped to the GRCm38 (mm10) reference genome using the STAR algorithm. Then, the unique molecular identifier (UMI) and barcode counts were determined to generate gene–barcode matrices for each sample. Genes without expression values in all cells were removed. A gene–barcode matrix containing gene expression counts was generated for subsequent analysis. All additional analyses except for CNV analysis were performed using the Seurat V5.0.0 (http://satijalab.org/seurat/) R toolkit [48]. Cells with <200 or >7500 expressed genes were excluded to eliminate the influence of low-quality cells. The maximum percentage of UMIs mapped to mitochondria was set to 10%. Finally, a total of 33,612 cells were retained after quality control for further downstream analysis, specifically, 23,707 cells from rAAV-HOPE-treated samples and 9905 cells from PBS-treated samples.

Dimensionality reduction

For each dataset, the top 2000 variable features were identified using the ‘vst’ method. The datasets were anchored and integrated with the integration procedure in the Seurat package to remove batch effects between samples. A linear scaling transformation was applied to the identified variable features using the ScaleData function with default parameters. Principal component analysis (PCA) was performed on the scaled features for dimensionality reduction. The first 30 principal components were used to identify the neighbours and cluster the cells with a resolution of 0.1. The cell clusters were visualised on 2D uniform manifold approximation projection (UMAP) plots. The genes specifically expressed in each cell cluster were identified using the FindAllMarkers function. The cell types were defined based on the expression of well-established gene markers.

Identification of cancer cells

To identify malignant and non-malignant cells, two approaches were used to distinguish malignant cells from non-malignant cells in each sample. First, we identified malignant cells using the marker genes Mki67, Sox4 and Sparc. During this procedure, we found a cell cluster with high expression of malignant epithelial cell marker genes, including Epcam, Krt8, Krt18 and Krt19. We defined the cells in this cluster as cancer cells. To verify the identified cancer cells, the inferCNV R package was used to quantify copy number variants [49], with the rest of the cell types used as references. CNV analysis was performed using Seurat v4.4.0 (http://satijalab.org/seurat/).

Identification of specifically expressed genes and enriched functions in each group

To compare the features of the rAAV-HOPE-treated and PBS-treated groups, the differentially expressed genes (DEGs) in each cell type were identified using the FindMarkers function with the default settings. The lists of significant DEGs were uploaded to the Enricher (https://maayanlab.cloud/Enricher/) online tool, and the top 10 terms according to the adjusted p value were retained.

Cell‒cell communication analysis

CellPhoneDB [50] was used to analyse and compare cell‒cell communication between rAAV-HOPE-treated and PBS-treated samples. The mouse gene names were converted to the corresponding human gene names with a script developed in house.

Differential regulatory activity analysis

Regulon analysis was performed by utilising pySCENIC [51], a computational tool for identifying regulons and assessing the activity score of regulons in individual cells, with default parameters. The differentially activated regulons were defined using the Wilcoxon rank–sum test. Fold changes in activity scores and p values were also calculated.

Characterisation of the differentiation trajectories of cancer cell subclusters

The differentiation trajectories of different cancer cell subclusters were determined using Monocle2 [52]. Genes expressed in at least 10 cells and in more than 10% of cells were used to perform pseudotime analysis.

Calculation of the stemness score

The stemness score was calculated with CytoTRACE [53], an established computational framework for stemness evaluation based on transcriptional diversity. The parameters used in this analysis were all set to the default values.

Statistical analysis

The data are expressed as the means ± standard deviations (SD). The statistical significance of differences between two groups was analysed by two-tailed unpaired Student’s t test. Comparisons among three or more groups were performed by one-way or two-way analysis of variance (ANOVA) with the Tukey test for multiple comparisons. Differences for which p < 0.05 were considered statistically significant.