Correction: Reprogramming anchorage dependency by adherent-to-suspension transition promotes metastatic dissemination (Molecular Cancer, (2023), 22, 1, (63), 10.1186/s12943-023-01753-7)

. Following publication of the original article [1], the author reported that the below Methods section needs to be added to the body text. RNA-seq fastq files were obtained from the ENCODE database. CLC Genomics Workbench 9.5.3 software (Qiagen, Germany) was used to map the reads to the human genome (hg19, build name GRCh37) and to generate gene expression values in the form of normalized transcripts per kilobase million (nTPM). Microarray data were obtained from the Human Protein Atlas (HPA) database in the form of matrices (row – official gene symbol, column – cell line) that included transcripts per million (TPM) values. Each of these data sets were imported into R for further analysis. Then, the data from the HPA were quantile normalized. R v3.6.3 was used to perform principal components and correlation analyses. Hierarchical clustering of these data was performed with the R package Pheatmap (version1.0.12). All cell lines were grouped into adhesion cells verus suspension cells based on their known morphology and anchorage dependency. Differentially expressed genes (DEGs) were selected based on absolute fold changes > 2, absolute TPM differences > 1, and p-values < 0.05, as measured in Gaussian T-tests. Gene Ontology (GO) enrichment analysis was performed with p-values < 0.05 (Benjamini-Hochberg) considered statistically significant. A Gene Set Enrichment Analysis (GSEA) was performed with a random combination number of 1000 and a false discovery rate (FDR) of 0.05 as criteria for significant enrichment. The “.gmt” format data were obtained from the Molecular Signature Database (MSigDB) using GSEA version 4.2.3. The specific gene sets used for these analyses are specified in Table S3. List of all transcription factors (Lambert et al., 2018) were included in the screen for AST factors (Table S4). Candidate AST factors were selected based on their nTPM differences, fold changes, and p-values. Candidate human AST genes were tagged with V5 and FLAG and subcloned into the gateway entry vector pENTR4 vector (Addgene, 17,425 and 17,423). The resulting subcloned vectors were recombined with destination vector pLentiCMV vector (Addgene, 17,293) using LR recombinase (Invitrogen, 1,179,019) to generate lentiviral expression vectors. To generate doxycycline-inducible HEK293A-AST-TetR cells, TetR plasmid (Addgene, 17,492) was infected in HEK293A-AST cells. The shRNA oligonucleotides that specifically targets 4 final AST genes (i.e., IKZF1, NFE2, IRF8 and BTG2) were annealed and subcloned into the pLKO.1 vector using the following target sequences: shRNA-IKZF1, GCTATCAATCATTAAGGTCAT; shRNA-NFE2, CCCATACTCCTATGGCAACAT; shRNA-IRF8, GCCCGCATCATGATTAAAGAA; shRNA-BTG2, CTATGAGGTGTCCTACCGCAT. To generate Ikzf1^−/−-B16F10 cells, single guide RNAs (sgRNAs) targeting mouse Ikzf1 (sgRNA1, TTTACGAATGCTTGATGCTC; sgRNA2, CAAGGCAGCTCGGCTTTGTC) were cloned into BsmBI-digested Lenti-CRISPR v2 (plasmid #52,961; Addgene). All constructs were verified by sequencing. All cells were maintained in an atmosphere of 5% CO2 in a 37°C humidified incubator. HEK293A and HEK293T, and B16F10 cells were cultured in DMEM (Hyclone, SH30022.01) and MDA-MB-231, LM2 and SUIT-2 cells were cultured in RPMI (Hyclone, SH30027.01) containing 10% FBS (Hyclone, SV30207.02) and 1% penicillin/streptomycin (Invitrogen, 15,140,122). No cell lines used in this study were found in the database of commonly misidentified cell lines that is maintained by ICLAC and NCBI Biosample. Cells lines were tested and confirmed to be free of mycoplasma.