Collection of fecal samples
Fecal samples from Korean CRC patients and healthy controls were collected and analyzed. Fecal samples from CRC patients were collected at National Cancer Center, Goyang between June 2002 and Apr 2004. The stools from healthy controls were collected in health check-up center at Seoul National University Hospital, Seoul and were selected through age and sex matched selection. The stool samples were collected by subjects in a tube and submitted to the institution and were frozen and stored at − 20°C.
16S bacterial rRNA analysis
Whole DNA was extracted from fecal samples using the Mag-Bind Universal Pathogen 96 Kit (Omega Bio-Tek, Norcross, GA, USA) with a Hamilton Microlab STAR liquid handler (Hamilton Laboratory Solutions, Manitowoc, WI, USA) after bead-beating the samples with TissueLyser (Qiagen, Hilden, Germany), followed by amplicon PCR targeting the V3 to V4 regions of the 16S bacterial rRNA gene using 341F and 805R primers (341F-CCTACGGGNGGCWGCAG, 805R-GACTACHVGGGTATCTAATCC). After DNA library preparation, indexing and quality checks were performed using the Nextera XT index kit (Illumina, San Diego, CA, USA) and Qubit4.0 (Thermofisher, Wilmington, DE, USA), and 300 × 2 paired-end sequencing was performed using the MiSeq system (Illumina, San Diego, CA, USA). Illumina adapter sequences of the paired- end reads were removed using Cutadapt 25. Then, the trimmed sequences were processed using QIIME2. Briefly, the reads were assigned to each sample according to a unique index; pairs of reads from the original DNA fragments were merged using an import tool in QIIME2 26. To remove low-quality bases at the end of the reads, the DADA2 software package27 wrapped in QIIME2 was applied. Alpha and beta diversity were analyzed using core-metrics-phylogenetic in the QIIME2 diversity plugin. Alpha and beta diversities were calculated using alpha- and beta-group significance in the QIIME2 diversity plugin, respectively. Alpha diversity was calculated by observed features, evenness, and Shannon Index. Beta diversity was compared by principal coordinate analysis using Bray– Curtis distances, unweighted Unifrac, and weighted Unifrac. The significance of similarity among the groups was evaluated using permutational multi- variate analysis of variance (PERMANOVA) with 999 permutations. Taxonomic annotation was performed by mapping the training reference set with primers (forward, 5′-CCTACGGGNGGCWGCAG-3′; reverse, 5′-GACTACHVGGGTATCTAATCC-3′) and extracting the V3–V4 region using GreenGenes version 13_8 28. Linear discriminant effect size analysis (LEfSe) was per- formed to identify differential features at the species level between groups based on linear discriminant analysis (LDA) scores using Galaxy implementation 29 Linear discriminant effect size analysis (LEfSe) 29 compares abundances of bacterial species levels between normal and cancer samples with a linear discriminant analysis (LDA) score.
Sample preparations of colorectal cancer and colon epithelium for organoid culture
Tissue samples from CRC patients were obtained from surgical specimens and immediately transported to the laboratory in a fresh state. Tissue biopsies were mechanically dissociated using surgical scissors and digested using an enzyme solution containing collagenase II (1.5 mg/ml, Gibco, 17101-015-1G), hyaluronidase (20 µg/ml, Sigma Aldrich, H3506-100MG), and Y-27632 (10 µM, Sigma Aldrich, Y0503-5MG) in DME/F12 medium (supplemented with penicillin and streptomycin). The tissue was incubated at 37°C with gentle rotation for 1 hour, after which the pellets were collected by filtering through a 100 µm-pore cell strainer. The pellets were then resuspended in RGF-BME Matrigel (Trevigen, 3533-001-02) and seeded onto 24-well plates. For the initial 3 days of culture, the isolated cancer epithelium (tumor) and crypts (normal) were fed with Human Intestinal Stem Cell HISC)( medium supplemented with 10 µM Y-27632 to prevent anoikis.
Organoid culture medium
To produce a complete organoid culture medium, conditioned medium and basal culture medium were mixed in a 1:1 ratio and supplemented with various growth factors, including B27 (1X, Gibco, 17504-044-50X, 10 ml), n-acetyl cysteine (1.25 mM, Sigma Aldrich, A7250-5G), nicotinamide (10 mM, Sigma Aldrich, 72340-100G), hEGF (0.1 mg/ml, Thermo Fisher Scientific, PHG0311L-0.1mg), hFGF10 (10 ng/ml, Thermo Fisher Scientific, PHG0204-25µg), A83-01 (500 nM, Sigma Aldrich, SML0788-5MG), SB202190 (3 µM, Sigma Aldrich, S7067-5MG), Prostaglandin E2 (10 nM, Sigma Aldrich, P5640-1MG), and contamination-blocking antibiotics (Primocin, 100 µg/ml, Thermo Fisher Scientific, ant-pm-1-10*1 ml 500 mg). Conditioned medium was prefabricated in advance using a stably transfected cell line purchased from ATCC (L-WRN-transfected), which secreted Wnt-3a, R-spondin, and noggin proteins into the culture medium. The cell line was thawed and cultured in DMEM/F-12 supplemented with 10% fetal bovine serum and 1% penicillin/streptomycin after the selection process for transfected-only cells. Hygromycin B (500 µg/ml) and G418 (2.0 mg/ml) were used as selection markers. The basal culture medium was composed of GlutaMax (10 mM, Thermo Fisher Scientific, 35050-061 100 ML) and DME/F12 supplemented with 10% FBS and 1% penicillin and streptomycin.
Organoid culture
We established and cultured three sets of organoid lines (SNU-7237-TO, SNU-7293-TO, and SNU-7390S3-TO) derived from CRC patients, as well as their corresponding normal organoid lines (SNU-7237N-NO, SNU-7239N-NO, and SNU-7390N-NO). Tumor organoid lines were passaged using TrypLE express (Gibco, 12604), which was resuspended in DME/F12 supplemented with 5% BCS to dissociate the organoids into single cells. The dissociated cell solution was centrifuged at 1,500 rpm (126×g) for 3 minutes, and the pellet was mixed with RGF-BME and seeded onto a new 6-well plate. Normal organoids were passaged using Cell Recovery Solution (Corning, 354253 100 ml). To split organoids into single crypts, harvested and collected organoid pellets were gently resuspended in Cell Recovery Solution and incubated at 4℃ for 30 minutes. Matrigel was liquefied by mechanical pipetting (30–50 times), and the subsequent culture process was identical to that of the tumor organoid lines.
STR profiling
Genomic DNA was extracted from each organoid line using the AmpF1STR PCR amplification kit (Applied Biosystems, Foster City, CA). The amplified DNA was analyzed using a genetic analyzer (Applied Biosystems 3500/3500xL Genetic Analyzer, Foster City, CA) to detect minisatellites and determine their short tandem repeat (STR) profiles. The STR marker libraries used for profiling included D8S1179, D2S441, D21S11, D7S820, CSF1PO, D3S1358, TH01, D13S317, D16S539, D2S1338, D19S433, Vwa, TPOX, Penta D, Amelogenin, D1S1656 and D18S51.
Treatment of cell free microbial supernatant to organoids
A hemocytometer was used to calculate the seeding concentration. 4,000 cells per well were seeded on a white 96-opaque well plate (SPL, 30196), and four days were given for differentiation and organoid formation. Each microbial cell-free supernatant was then diluted to 20% concentration in HISC (Organoid medium, Human Intestinal Stem Cell) medium. The cell-free supernatant was obtained from six strains: RCM (Normal culture medium, Reinforced Clostridial Medium), Blautia producta (KCTC 15607), Parabacteroides distasonis (KCTC 5751), Bacteroides ovatus (KCTC 5827), Bifidobacterium adolescentis (KCTC 3216), and Bifidobacterium longum (KCTC 3218). Additionally, the pH of all microbial cell-free supernatants was adjusted to 6.8 ≤ pH ≤ 7.2 by adding 1.0 M NaOH and measured using a pH meter. The supernatants containing the culture medium and organoids were co-cultured for three days.
RLU measurement for Cell viability assay
Viable cells and cytotoxicity were assessed using a 3D organoid culture model with CellTiter-Glo® 3D, which is used for evaluating cellular ATP levels. The reagent was thawed and equilibrated to room temperature before use, and all experimental procedures were performed at room temperature. The half volume of the culture solution (containing bacterial supernatant) was replaced with CellTiter-Glo® 3D (Promega, G9683) and thoroughly resuspended using a multi-channel pipette. Plates for screening were sealed with aluminum foil and incubated for 30 minutes. After the light-excluded incubation, the luminescence of intra-organoid ATP content was quantified by Varioskan Lux (Thermo Fisher Scientific, USA).
High-Throughput Screening
To establish the HTS system, we used the ImageXpress Micro Confocal 4 (Molecular Devices, USA) and corresponding image analysis software (MetaXpress, Molecular Devices, USA). For screening, we selected the µ-Plate Angiogenesis 96-well (Ibidi, 89646, Germany) and calibrated it accordingly. Our method reported previously was partially applied to this study 30. We seeded 4,000 cells per well, derived from organoids, onto 10 µl of RGF-BME gel that had been previously dispensed and solidified in each well of the plate. For the side of the organoid that faced the culture medium, 4% of RGF-BME was added to the organoid medium as a feeding ingredient for the ECM. The seeded organoids were allowed to grow for 4 days and then treated with bacterial supernatant in the proportion of 25% culture in the medium. The High-throughput screening (HTS) was performed at 3, 6, and 12 days after treatment using a 4X apochromatic objective lens, z-stacking, and z-projection.
RNA sequencing
Total RNA was isolated from cell lysate using TRIzol (Qiagen, Hilden, Germany) and Qiagen RNeasy Kit (Qiagen, Hilden, Germany). Paired-end sequencing reads from cDNA libraries (101 bp) were generated with an Illumina NovaSeq6000 instrument and the sequence quality was verified with FastQC v.0.11.7 (https://www.bioinformatics.babraham.ac.uk/projects/fastqc). For data preprocessing, low-quality bases and adapter sequences in reads were trimmed using Trimmomatic v 0.38 31. The trimmed reads were aligned to the human genome (UCSC hg19) using HISAT v2.1.0, a splice-aware aligner 31. Then, transcripts including novel splice variants were assembled with StringTie v1.3.4d 32. The abundance of these transcripts in each sample was calculated as read counts or TPM (Transcript per Million mapped reads) values. To perform principal components analysis (PCA) and confirm the similarity distance among the samples, the dist and prcomp functions were used from the ggdendro (v0.1.22) and ggfortify (v0.4.11) R packages respectively. To access the internal data of gene loadings and analyze the contributing variables of PC2 which split the samples by beneficial and harmful intestinal bacteria, the loading components were calculated by advanced features of the pca function from PCAtools (v1.2.0) R package. The 20 highest loading components containing both positive and negative values in PC2 were plotted in a bar graph. Based on the raw read counts of 35,993 transcripts, single sample enrichment level analysis of the cell signaling pathway was conducted by GSEA (v4.1.0) using the hosted MSigDB gene set database of KEGG library (c2.cp.kegg.v7.4.symbols.gmt). The phenotype label was designated as either normal colorectal mucosa versus tumor tissue or corresponding tumor derivatives for the normalized enrichment score (NES) of a single sample. Independent NES of the paired sample (normal and tumor) was calculated on default fields with the setting of permutations as 1,000 and phenotype. The result was annotated by NCBI Gene ID MsigDB.v7.4.chip platform. A total of overlapped pathways (50 gene sets of KEGG pathway) which were significantly different (FDR < 0.25) compared to normal control in both multiple tumors and PDOs (patient-derived organoid) were selected to identify the recapitulations of differentially expressed pathways in derived models. Using the ComplexHeatmap (v2.2.0) R package, the heatmap of NES values was plotted and the mapped color variance was set between minimum and maximum values.
Gas chromatography
Supernatants from bacterial culture were harvested at O.D600 values of 0.35–0.40. Methanol was added to the samples in a 1:1 ratio, and the resulting mixture was vortexed and sonicated for 10 minutes. After centrifugation at 13,000 rpm (16,609×g) and 4℃ for 10 min, the supernatant was collected and dried using a speed vacuum at 10,000 ppm (1 mg/ml). The dried sample was reconstituted in 150 µl of distilled water and then transferred to a new ep-tube. The sample was dried again with a speed vacuum, and derivatization was performed. Gas chromatography was carried out using a GC-TOF-MS (LECO Corporation, US) with a Rtx-5MS column and helium gas for analysis. The split ratio was set at 20:1, and three analytical replicates were performed.