A pair of primary colorectal cancer-derived and corresponding synchronous liver metastasis-derived organoid cell lines

Establishment of CWH22/CLM22 3D organoids

The experimental design of the procedure is summarized in Figure 1. The occupied lesions in the left liver lobe and ascending colon are presented in Supplementary Figure 1A, 1B, respectively. H&E and IHC staining indicated moderately differentiated adenocarcinomas with tubular structures of colorectal epithelial origin of colorectal epithelial origin (cytokeratin 20 (CK20)- and CDX2-positive) with wild-type p53, high proliferation rates (Ki67), and microsatellite stability (MSS; normal expression of MLH1, PMS2, MSH2, and MSH6 proteins; Supplementary Figure 1C). The treatment strategies and the response records are presented in Supplementary Figure 1D, 1F.

Figure 1. Overview of the procedure. Tissues from primary colorectal cancer and liver metastases were obtained and then divided into three parts: two parts were placed in liquid nitrogen and 4% paraformaldehyde, separately, for molecular and histopathological examinations, and the other part was immediately placed on ice and transported to the laboratory for tumor cell isolation and organoid culture. Whole-exome sequencing and histopathological analysis were conducted to determine the concordance between the tumor organoids and corresponding tumors. The organoids were subjected to STR, karyotype analysis, in vitro drug sensitivity tests, and in vivo animal model construction.

During culture, both CRC- and CLM-derived organoids, namely CWH22 and CLM22, proliferated and differentiated well. The gross appearance of the CRC and CLM tissues, as well as bright-field images of the organoids, are presented in Figure 2A. H&E staining and IHC staining revealed morphological (e.g., thin- and thick-walled cystic structures) and protein expression similarities between CWH22/CLM22 organoids and the corresponding tumor tissues (Figure 2A, 2B; Supplementary Figure 1C). Both CWH22 and CLM22 organoids circumvented replicative senescence, acquired the ability to sustain unlimited proliferation in culture (passaged more than 50 times), and were cryopreserved at different early passage stages. Two movie recordings (Supplementary Videos 1 and 2) show the bright-field morphologies of CWH22 and CLM22 ten days after recovery, with continuous zoom at different focal surfaces, demonstrating that these organoids can withstand freezing, storage, and thawing.

Figure 2. CWH22 and CLM22 organoids maintain histological and proliferative characteristics of the corresponding tumors. (A) H&E staining of the tumor organoids showing the resemblance of the organoids to the lumens and tubular structures of the primary tumor. IHC staining demonstrated consistent positive expression of CK20 and CDX2 in the organoids and corresponding tumors (scale bars, 100 μm). (B) IHC staining of CWH22 and CLM22 organoids demonstrating the consistent expression of p53, Ki67, MLH1, PMS2, MSH2, MSH6, and Villin, with the expression in CRC and CLM tissues (Supplementary Figure 1C). Scale bars, 100 μm. (C) Representative time course of CWH22 and CLM22 organoid growth for 14 days (scale bars, 500 μm) and local field amplification (scale bars, 200 μm). (D) Analysis of the proliferation rates of CWH22 and CLM22 organoids over time using 3D cell viability tests. (E) Bright-field images showing the morphology of CWH22 and CLM22 organoids at the 2nd, 5th, 10th, and 20th passages (P2, P5, P10, and P20, respectively; scale bars, 500 μm), and the expression of CK20 and Ki67 in representative differentiated organoids from CWH22 and CLM22 (scale bars, 50 μm).

Proliferation, morphological stability, and identification of CWH22/CLM22 organoid cells

Proliferation and tumorigenicity are basic features of primary and organoid cell lines, making them popular models in cancer research. CWH22 and CLM22 organoids grow rather slowly during the first five days, owing to the intentionally low number of cells initially seeded into each well; this was done to provide more 3D space in the BME for organoid growth and differentiation (Figure 2C). Subsequently, the organoids grew rapidly, and the total cell viability on the 14th day was 40 times greater than that on the first day. Moreover, the proliferation rates of this paired organoids (CWH22 and CLM22) did not differ (Figure 2D).

Apart from the proliferation conditions in the 3D culture system, stable organoid morphology was also maintained over long periods. Luminal and glandular epithelioid folds were evident at different passages (P2, P5, P10, and P20) in both CWH22 and CLM22 organoids, which maintained the structural heterogeneity well (Figure 2E). Immunofluorescence staining of the differentiated structures in CWH22 and CLM22 organoids was performed in the 20th generation, revealing that both organoids were positive for CK20, which is normally expressed in luminal cells of the colonic mucosa, and Ki67, a proliferation marker (Figure 2E).

Since CWH22 and CLM22 meet the conditions of being cell lines and possess good tumorigenicity (see the following sections), the 4th generation of these organoid cell lines were deposited at the CCTCC (No. C202218 and C202219) after testing negative for bacterial and mycoplasma contamination. The STR authentications of 21 locations (including the nine ATCC-required loci) from these two organoid cell lines and the corresponding normal mucosa tissue were assessed by the CCTCC, and a complete match was observed (Table 1). Ample evidence shows that cross-contamination or phenotypic drift of cells in culture can generate irreproducible or misleading data, and many scientific journals require cell identification based on DNA analysis of samples and cell lines used [20, 21]. The STR profile of this case differed from that of all cell lines available in different cell banks (http://cellresource.cn/str/default.aspx).

WES of CWH22/CLM22 organoids and corresponding tissues

Organoids derived from cancer patients recapitulate the genomic profiles of the corresponding tumors, including DNA copy number variations and mutations [12, 22, 23]. Thus, genomic DNA was isolated, and WES was performed on the two organoid cell lines, the corresponding normal mucosa, CRC, and CLM tissues (Figure 3A–3C). The organoids retained the somatic mutational spectrum observed in the patient’s tumors, and these were among the top mutated genes of CRC with liver metastases [24] (https://www.cbioportal.org/study/summary?id=crc_msk_2017; Figure 3A). Like those in the original tumors, CWH22 and CLM22 organoids harbored mutations in pathways important for CRC initiation and progression, including the WNT (APC mutation), RAS-MAPK (KRAS mutation), and TGF-β (SMAD4 mutation) pathways, as well as KMT2C and CDKN1B mutations (Figure 3A). Thus, the CWH22 and CLM22 organoid cell lines are representative of APC, KRAS, SMAD4, CDKN1B, and KMT2C quintuple-mutated CRC cell lines with wild-type TP53 and PIK3CA (Table 2). The organoids retained most somatic variants present in the corresponding original tissue (CWH22 vs. CRC tissue: 93.56%; CLM22 vs. CLM tissue: 92.74%; Figure 3B). Furthermore, the original copy number alterations at the genome-wide level were recapitulated in CWH22 and CLM22 organoids (Figure 3C). These data demonstrated that the CWH22 and CLM22 organoid lines are highly consistent with the corresponding tumor tissues at the genomic level.

Figure 3. Whole-exome sequencing (WES) and chromosome karyotype analysis of CWH22 and CLM22 organoids. (A) Overview of somatic mutations found in CRC/CLM tissues and CWH22/CLM22 organoids. (B) Histogram showing the concordance (percentage) of SNVs between CWH22/CLM22 organoids and corresponding CRC/CLM tissues. (C) The ordinate represents the multiple copy variations, the black line represents the multiple total copy number variations, and the red line represents the multiple minor copy number variations. When the value of the black line is >2, the copy number increases; when it is <2, the copy number decreases. (D, E) Chromosome karyotype pairing from one representative karyotype analysis and the chromosome numbers of 20 different metaphase cells of (D) CWH22 and (E) CLM22.

Karyotyping of CWH22/CLM22 organoids

According to WES results and analysis system of BGI, the tumor mutation burden (TMB, Mut/Mb)/microsatellite instability (MSI) sensor results for CWH22, CRC, CLM22, and CLM samples were 119.85/0.57, 106.85/0.1, 146.53/0.52, and 166.78/0.16, respectively, indicating the absence of MSI in all samples (MSI score threshold, 3.5). Combined with the normal expression levels of MLH1, PMS2, MSH2, and MSH6 proteins in tumor tissues and organoids (Supplementary Figure 1C and Figure 2B), MSS was confirmed in CWH22 and CLM22 organoids. Since MSI and chromosomal instability (CIN) are frequent events in cancer that result in the accumulation of oncogenes and inactivation of tumor suppressor genes [25–28], facilitating tumorigenesis and progression, organoid cells from the fifth passage were separately harvested for G-banding karyotype analysis. The chromosomal numbers ranged from 47 to 54 in CWH22 organoid cells and from 48 to 52 in CLM22 organoid cells while counting 20 karyotypes (Figure 3D, 3E). Chromosome karyotype pairing revealed that extra copies of different chromosomes were rather common, especially chromosomes 12 and 13 (Figure 3D, 3E), which are also stably presented in the copy number variation (CNV) at the genome-wide level (Figure 3C). Thus, from these data, we conclude that CIN was the main cause of genomic instability in these two organoids.

KRAS mutation, a common phenomenon in cancer, is found in ~40% of CRCs [29]. As this gene is located on the short arm of chromosome 12 (12p12.1, Table 2), both CNV (Figure 3C), and karyotype analyses of CWH22 and CLM22 (Figure 3D, 3E) suggested that three copies of chromosome 12 are common in these two organoids. Therefore, each organoid most likely has three copies of the KRAS gene. Moreover, the percentages of these two organoids with KRAS G12C mutations, as indicated by WES, were 60.7% and 66.2% respectively (Table 2); we speculate that KRAS G12C mutation is heterozygous and most likely to be mutation-positive on two chromosomes and wild-type on the third chromosome.

Drug response of CWH22 and CLM22 organoids

Since metastasis is a key contributor to CRC-associated deaths and effective therapies are lacking, the exploration of treatment strategies is of great importance to improve patient outcomes [30]. The use of genomics to guide cancer therapies was promoted after the widespread use of genetic testing. However, this approach has been less successful than anticipated, and less than 7% of patients in 2018 were estimated to benefit from such precision oncology [31]. Organoids have been demonstrated to be effective tools for predicting CRC responses to different treatments in clinical practice [12, 15, 17, 18, 32]; thus, we selected several compounds to determine the sensitivity of CWH22 and CLM22 organoids to these drugs.

5-Fu, oxaliplatin, SN-38, regorafenib, etoposide (VP-16), and sotorasib are common drugs used for CRC treatment and preclinical study. The dose-response curves and IC₅₀ values of CWH22 and CLM22 organoids exposed to the selected drugs, as shown in Figure 4A, were consistent with previous reports indicating no significant difference in the drug sensitivity of CRC and paired LM organoids from the same patient for certain drugs [15, 18]. The IC₅₀ values of 5-Fu, oxaliplatin and SN-38 for CWH22/CLM22 organoids were 3.3/3.78 μM, 15.47/18.38 μM and 7.54/10.67 nM respectively, indicating that CWH22/CLM22 organoids were more sensitive to these drugs than LoVo, SW620, T84, and SW480 cells, according to the data from the Genomics of Drug Sensitivity in Cancer database (GDSC; https://www.cancerrxgene.org/; 10–415 μM for 5-Fu; 28–35 μM for oxaliplatin; 0.0139–39 μM for SN-38). Next, we conducted combined treatment with 5-Fu + oxaliplatin (FO, 1:1) according to the patient’s treatment regimen. Pictures of CWH22, CLM22, and another resistant organoid CRC16-Org (pre-experimental results indicating original resistance to 5-Fu and oxaliplatin) exposed to three concentrations of FO (2 μM, 5 μM, and 10 μM) are shown in Figure 4B. Further analysis revealed higher sensitivity of CWH22 and CLM22 organoids to FO treatment compared to CRC16-Org (Figure 4C; the organoid cell viability decreased by 60.96 ± 3.26%, 60.75 ± 4.74%, and 13.58 ± 5.99%, respectively, upon treatment with 5 μM FO). Clinically, the shrinkage of liver metastatic lesions demonstrated apparent sensitivity to the first eight cycles of the mFOLFOX + bevacizumab treatment strategy; the coronal image records of the liver during the treatment are presented in Figure 4D.

Figure 4. Drug treatment sensitivities of CWH22 and CLM22 organoids to traditional chemotherapies. (A) Dose-response curves of CWH22 and CLM22 organoids exposed to increasing concentrations of 5-fluorouracil, oxaliplatin, SN-38, regorafenib, VP-16, and sotorasib, and the half maximal inhibitory concentrations (IC₅₀) of each drug. (B) Representative images of CWH22 and CLM22 organoids exposed to three concentrations of 5-Fu and oxaliplatin (2 μM, 5 μM, and 10 μM; scale bars, 200 μm). (C) Quantitative analysis of CWH22, CLM22, and CRC16-Org cell viability following exposure to different concentrations of FO combined treatment (2 μM, 5 μM, and 10 μM; n = 3; data represented as the mean ± SD, ^**p < 0.01). (D) Preoperative and postoperative MRI coronal images of liver lesions during the first eight cycles treatment with mFOLFOX + bevacizumab strategies (scale bars, 5 cm).

Sotorasib was the first FDA-approved drug targeting KRAS mutations (G12C) in cancer. We determined the IC₅₀ values following sotorasib treatment for CWH22, CLM22, two organoids harboring KRAS G12D mutations (CRC1-Org and CRC16-Org), and one organoid with wild-type RAS (CRC27-Org). The IC₅₀ values were 12.07 μM, 10.95 μM, 28.31 μM, 31.37 μM, and 21.49 μM for CWH22, CLM22, CRC1-Org, CRC16-Org, and CRC27-Org, respectively (Figures 4A and 5A). Representative bright-field images of KRAS G12D- and G12C-mutated organoids exposed to 10 μM and 20 μM sotorasib are presented in Figure 5B. Based on the in vitro cell viability data, the organoids harboring KRAS G12C mutations (CWH22 and CLM22) were more sensitive to sotorasib than were those harboring KRAS G12D mutations (CRC1-Org, CRC16-Org) or the wild-type RAS (CRC27-Org; Figure 5C). Moreover, a KRAS G12D inhibitor (MRTX1133) and a pan-KRAS inhibitor (BI-2865) were used to treat all these five organoids, and the dose-response curves and IC₅₀ values are presented in Figure 5D, 5E. As expected, KRAS G12C-mutated (CWH22, CLM22) and unmutated (CWH27-Org) organoids were less sensitive to MRTX1133 than were KRAS G12D-mutated organoids (CRC1-Org and CRC16-Org); however, their sensitivity seemed to be unrelated to the KRAS mutation status when treated with BI-2865. Basic information of CRC1-Org, CRC16-Org, and CRC27-Org is presented in Supplementary Table 1.

Figure 5. Drug treatment sensitivities of different organoids to KRAS G12C, KRAS G12D and pan-KRAS inhibitors. (A) Dose-response curves of KRAS G12D-mutated organoids (CRC1-Org and CRC16-Org) and RAS-unmutated organoid (CRC27-Org) exposed to increasing concentrations of sotorasib (a KRAS G12C inhibitor), and the corresponding IC₅₀ values. (B) Representative images of CWH22, CLM22, CRC1-Org, and CRC16-Org organoids exposed to sotorasib at concentrations of 10 μM and 20 μM (scale bars, 200 μm). (C) Quantitative analysis of the viability of CWH22, CLM22, CRC1-Org, CRC16-Org and CRC27-Org (with different KRAS mutation backgrounds) following exposure to 10 μM and 20 μM sotorasib (n = 3, data represented as the mean ± SD, ^**p < 0.01). (D, E) Dose-response curves and IC₅₀ values for (D) the KRAS G12D inhibitor (MRTX1133) and (E) the pan-KRAS inhibitor (BI-2865) for organoids with different KRAS mutation statuses.

Tumorigenicity and liver metastatic ability of CWH22 and CLM22 in vivo

To evaluate the potential of CWH22 and CLM22 organoid cells to grow in vivo, these cells were cultured and then injected subcutaneously as organoid structures into nude mice. Images of tumor-bearing mice on day 28 postinjection are presented in Figure 6A, confirming the tumorigenicity of the two organoid cell lines. Then, the organoid cells were transduced with lentivirus supernatants to express luciferase; the organoids stably expressing luciferase were marked as CWH22-luci and CLM22-luci (Figure 6B). Further, the splenic injected liver metastasis models with CWH22-luci and CLM22-luci organoid cells confirmed sound liver metastatic ability of these two organoids, as all injected mice exhibited liver metastases (Figure 6C–6F). Interestingly, the metastatic ability of CWH22 and CLM22 did not significantly differ based on analyses of the percentage of tumor-bearing liver weight to body weight (p = 0.611; Figure 6E) and number of liver metastases (p = 0.685; Figure 6F). Furthermore, H&E and IHC staining of the xenografts demonstrated that the subcutaneous tumors and transplanted liver metastases induced by CWH22 and CLM22 organoids preserved the moderately differentiated status and the molecular expression status of the corresponding tumor tissues from the patient (Figure 6G). Taken together, these results demonstrated the tumorigenicity and metastatic ability of CWH22 and CLM22 organoids, indicating that the 3D culture system in vitro sustained the malignant manifestation and expression status of the corresponding tumors.

Figure 6. Tumorigenicity of CWH22 and CLM22 organoids in vivo. (A) Images of nude mice bearing CWH22 and CLM22 subcutaneous xenografts (scale bars, 2 cm). (B) Luciferase activity tests of CWH22-luci and CLM22-luci (CWH22 and CLM22 with no luciferase expression were tested as control). (C) Representative bioluminescence images of mice with liver metastases 42 days after splenic injection. (D) Representative pictures of livers with metastatic lesions (on day 45) harvested from mice receiving splenic injections of CWH22 and CLM22 organoid cells (scale bars, 1 cm). (E) Quantitative analysis of the percentage of tumor-bearing liver weight to body weight following splenic injection (n = 5; data represented as the mean ± SD). (F) Quantitative analysis of the metastatic nodule numbers per mouse following splenic injection (n = 5; data represented as the mean ± SD). (G) H&E morphology and IHC stains of tumor tissues from nude mice in A and B, as well as the normal mucosa and tumor tissues of the patient. Representative images of the expression of CK20, CDX2, Ki67, and p53 are shown (scale bars, 100 μm).

3D organoids maintain tumor marker expression

Although not every CRC case is associated with increased levels of serum tumor markers (CEA, CA19-9, and CA72-4), the patient from whom CWH22 and CLM22 organoid cell lines were derived had abnormally elevated preoperative CEA and CA19-9 levels (132 ng/mL and 729 U/mL, respectively), which significantly decreased postoperatively (47.05 ng/mL and 151.6 U/mL, respectively). The serum CEA levels of the patient varied with the treatment response and disease progression during the entire treatment period until November 2022 (Supplementary Figure 1D, 1F), as did the trend of CA19-9 levels, consistent with a previous study showing that CEA and CA19-9 levels are clinically relevant for predicting outcomes in patients with metastatic CRC [33].

CEA secretion is an intrinsic feature of some colorectal cancer cell lines, such as LoVo, HCT116, HT-29, T84 [34–36]. To test whether CWH22 and CLM22 organoids maintain the secretion of CEA, as well as CA19-9 and CA72-4, the culture media (incubated for 72 h) were collected on day9 (approximately 1 × 10⁵ cells per well on day 7) and on day15 (approximately 5 × 10⁵ cells per well on day 12) of the 20th passage of CWH22 and CLM22 organoids to assess the levels of these markers; the data are presented in Table 3 (cultured media were collected after incubation with organoids for 72 h), indicating that these two organoid cell lines maintained the expression and secretion characteristics of the original tumors, and that there were no significant differences in the secretion capacity of these tumor markers (CEA, CA19-9 or CA72-4) between CWH22 and CLM22 organoids.

Characteristics of adherent CWH22-2D and CLM22-2D cells

CWH22-2D and CLM22-2D adherent cultured cell lines were successfully established from the xenografts derived from CWH22 and CLM22 organoids, after 2 months culture. The bright-field images and the positive immunofluorescent staining of CK20 and CDX2 in CWH22-2D/CLM22-2D cells are presented in Figure 7A. Next, subcutaneous tumor formation assay confirmed xenografts derived from CWH22-2D and CLM22-2D cells grew slower than those derived from CWH22 and CLM22 organoids in nude mice (Figure 7B, p < 0.01), and splenic injected liver metastasis formation assays confirmed that CWH22-2D and CLM22-2D cells were not that appropriate to generate liver metastasis models when compared with organoid-derived metastasis models (Figure 6G), as only one mouse tested positive in each group (Figure 7C). H&E staining of CWH22-2D- and CLM22-2D-derived liver metastasis xenografts also displayed a moderately differentiated status (Figure 7D) compared to that of the original tumors, which suggested that differentiation status is an intrinsic property of established cell lines and had no relationship with culture status (2D or 3D) in this study during early use.

Figure 7. Tumorigenicity of adherent CWH22-2D and CLM22-2D cells. (A) Bright-field images of CWH22-2D/CLM22-2D cells (left panel); and immunofluorescent staining images of CK20 and CDX2 (middle and right panels) from CWH22-2D/CLM22-2D cells. Scale bars, 50 μm. (B) Subcutaneous tumor growth curves were plotted according to the monitored tumor size every seven days for 35 days. The tumors were derived from CWH22-2D/CLM22-2D cells and CWH22/CLM22 organoids. (C) Bioluminescence imaging of mice with liver metastases 42 days after splenic injection of CWH22-2D and CLM22-2D cells, and livers with metastatic lesions harvested from mice receiving splenic injections of CWH22-2D and CLM22-2D cells. (D) Representative image of the entire liver lobe with metastases, and focal magnification. Scale bars, 100 μm.

Comparison of CWH22-2D/CLM22-2D with traditional adherent 2D cultured cell lines

Cell lines are important tools for studying tumor pathogenesis, therapies, and drug resistance mechanisms. Their properties in relation to primary tumor characteristics and tumorigenicity in animals indicate their feasibility for research application. Since most CRCs operated on at Tongji Hospital were moderately or moderately to highly differentiated adenocarcinomas, the xenografts formed in immunodeficient mice should reproduce these features to make the models more convincing for further studies.

First, CCK8 assays were conducted to compare the proliferation rates of CWH22-2D, CLM22-2D, LoVo, and SW480 cells, indicated slower growth rates of CWH22-2D and CLM22-2D cells (about 15-fold in the 14 days of culture) compared to that of LoVo and SW480 cells (proliferated about 15-fold in 8–10 days; Figure 8A). Since all these four cell lines possessed well tumorigenicity, we next induced subcutaneous tumor formation in nude mice with LoVo and SW480 cells, and H&E staining exhibited low differentiation status without any tubular or adenoid structures of the xenografts compared with those in the CWH22-2D/CLM22-2D-derived xenografts (Figure 8B). One previous report also presented the same morphological manifestation of the xenografts derived from SW480 in an immunodeficient mouse [37]. Although LoVo cells inoculated into nude mice produced tumor masses with glandular structures 4 decades ago [38], the tumor masses in our study demonstrated the dedifferentiated status of these cells (Figure 8B). Moreover, studies involving H&E and IHC staining of tumor xenografts of other 2D cultured CRC cell lines (from ATCC, such as HCT116, HT-29, Caco-2, and RKO) were searched, and we found that these cell-derived tumors in nude mice presented no evident glandular or adenoid epithelial arrangements in morphology [39–41]. For comparison, H&E staining of subcutaneous tumors derived from two moderately differentiated organoids (CRC16-Org and CLM15-Org) and two poorly differentiated organoids (CRC1-Org and CLM23-Org) are also presented (Figure 8B). Although both CRC1-Org- and CLM23-Org-derived subcutaneous tumors were poorly differentiated, the mucinous lakes and dedifferentiated glandular cavities were preserved, whereas LoVo- and SW480-derived xenografts did not preserve any morphological features associated with the colorectal epithelium (Figure 8B).

Figure 8. Differences between CWH22-2D/CLM22-2D and traditional adherent 2D cultured cell lines (LoVo and SW480). (A) Analysis of proliferation rates with CCK-8 in CWH22-2D, CLM22-2D, LoVo, and SW480 cells. (B) The differentiation characteristics of subcutaneous neoplasia in nude mice derived from LoVo and SW480 (ATCC), and CWH22 and CLM22 from our culture. Two moderately differentiated (CRC16-Org, CLM15-Org), and two poorly differentiated (CRC1-Org, CLM23-Org) organoid-derived subcutaneous neoplastic tissues were used for comparison. (C) Bright-field images and H&E staining of the structures derived from the cells (LoVo, SW480, CRC1-Org, CLM23-Org, CWH22-2D, CLM22-2D, CRC16-Org, and CLM15-Org) cultured in the same 3D culture system. Scale bars, 100 μm.

Since 3D culture can directly reveal morphology, we conducted 3D organoid cultures with LoVo, SW480, CWH22-2D/CLM22-2D (the 10th generation) cells, and compared with other two moderately differentiated organoids (CRC16-Org, CLM15-Org) and two poorly differentiated organoids (CRC1-Org and CLM23-Org). As indicated, only loose spheroid structures were observed in the bright-field and H&E images of LoVo and SW480 cells, demonstrating a completely dedifferentiated status in these two cell lines compared to other structures (Figure 8C). Thus, both 2D and 3D primary cultures could maintain the differentiation status of the original tumors in our study.

Moreover, the characteristics of several commonly used CRC cell lines from the ATCC and the cell lines established in our laboratory are presented in Table 4, illustrating the origin of these cells, MSH/CIN, CEA secretion, and mutation status of some driver genes. This approach will facilitate the selection and in-depth understanding of these cell lines to different researchers.

Clinical history and patient characteristics

In this study, the presented paired organoids were established from the tumor tissues of a patient with CRC and synchronous liver metastasis. Both the organoids and the corresponding tumor tissues were subjected to WES and histopathological analysis to determine the concordance between tumor organoids and corresponding tumors, and organoids were used for further study. Figure 1 presents an overview of the procedures.

The tissue donor, a 65-year-old woman, was admitted to the Hepatic Surgery Centre of Tongji Hospital in January 2021 with a two-week history of weakness after the discovery of a liver-occupying lesion during a physical examination. Computed tomography (CT) and magnetic resonance imaging (MRI) scans of the liver revealed an 89 × 65 mm lesion in the left liver lobe (Supplementary Figure 1A). Routine blood examination showed moderate anemia (hemoglobin level, 74 g/L). A colonoscopy revealed a space-occupying lesion in the right colon (Supplementary Figure 1B). Laparoscopic resections of the lesions from the right colon and left liver lobe were performed. The histological diagnosis revealed moderately differentiated right colon invasive adenocarcinoma with liver metastasis at stage IV A (T3N1cM1a; Supplementary Figure 1C). Several small liver metastases occurred during the recovery from the surgery, and on day 45 after surgery, the patient started treatments with mFOLFOX + bevacizumab. The small metastatic lesions shrank and were stable during the eight treatment cycles. Thereafter, capecitabine and bevacizumab were used as maintenance treatments, and the lesion appeared stable for an additional seven months. However, the disease relapsed 14 months later, and became uncontrollable in November 2022. The entire treatment strategies until November 2022 and treatment responses are presented in Supplementary Figure 1D. Supplementary Figure 1E, 1F present the treatment response of the liver lesions (MRI images) and changes in the serum carcinoembryonic antigen (CEA) concentration during the treatment.

Tissue sampling

Samples were obtained from the Hepatic Surgery Centre, Tongji Hospital, Huazhong University of Science and Technology (Wuhan, China). The collected CRC and CLM samples were immediately divided into three parts. One part was immersed in DMEM, transported on ice to the laboratory for tumor cell isolation, and cultured within 1–3 h. The remaining parts were placed in liquid nitrogen and 4% paraformaldehyde (Sigma-Aldrich, P6148) immediately after collection for molecular and histopathological examinations.

Tumor cell isolation and 3D organoid culture

For the derivation of paired tumoroids, freshly resected CRC or CLM samples were processed as previously reported [22]. The surface mucus and necrotic tissue of the tumor sample were removed before washing with ice-cold phosphate-buffered saline (PBS), and the tissue was soaked in active iodine for 1 min. Subsequently, the samples were rinsed thrice with PBS, penicillin/streptomycin, and PBS successively for each step. Then, the tissue (~0.5 cm³) was cut into 0.5 mm pieces, digested in 50% TrypLE™ Express (Gibco, 12605028)/50% DMEM at 25°C for 15 min (the suspension was mixed thoroughly at 7 and 14 min), and neutralized with the same volume of Ca²⁺-containing Hanks’ balanced salt solution (HBSS). The cell suspension was allowed to settle for 3–5 min, after which the supernatant was collected, centrifuged (500 × g), and washed twice using centrifugation at different speeds (140 × g and 200 × g) to remove cell debris and collect tumor cells. The pellet was then washed twice in advanced DMEM/F12, and cells were counted and mixed with growth factor-reduced BME (RD, 3533). In a prewarmed 24-well plate, 8,000–10,000 cells were seeded per well with 30 μL BME. After BME solidification following incubation at 37°C for 20–30 min, 600 μL of tumoroid-specific culture medium (Table 5) was added, and the plate was incubated at 37°C with 5% CO₂. The medium was completely refreshed every 3–5 days.

During the first two passages, only half of the medium was changed each time. Organoids were passaged when they reached diameters of 100–700 μm. After washing with PBS, both the organoids and BME fractions were mechanically pipetted using 250 μL TrypLE™ Express per well. Organoids were digested into single cells within 10–14 min. Following neutralization with the same volume of Ca²⁺-containing HBSS, the cells were washed with DMEM, counted, seeded at 8000 cells/30 μL BME, overlaid with organoid-specific culture medium, and cultured as described above.

After subsequent passages, all stably passaged organoids were viably frozen as single cells in culture medium with 10% dimethyl sulfoxide and stored at −80°C, and cryovials were transferred to a nitrogen tank the next day—for long-term storage. Organoid cultures were checked monthly for mycoplasma contamination using a MycoBlue Mycoplasma Detector Kit (Vazyme, D101-01/02, China). The organoid cell lines derived from the resected CRC and CLM tissues were named CWH22 and CLM22, respectively. Both were cultured and passaged using the same protocols.

Immunohistochemistry and immunofluorescence assays

Tissues from the patient and tumor-burdened nude mice were embedded in paraffin. All hematoxylin and eosin (H&E) and immunohistochemistry (IHC) staining procedures, and subsequent histological diagnoses were performed in the Pathology Department of Tongji Hospital, as previously reported [68]. The antibodies used for IHC were as follows: CDX2 (Cat. No. ZA-0520 R), CK20 (Cat. No. ZA-0574 R), MLH1 (Cat. No. ZM-0154), PMS2 (Cat. No. ZA-0542 R), MSH2 (Cat. No. ZA-0622 R), MSH6 (Cat. No. ZA-0541 R), Her-2 (Cat. No. ZA-0641 R), Ki67 (Cat. No. ZM-0166), p53 (Cat. No. ZM-0408), and Villin (Cat. No. ZA-0575 R; all from ZSGB-BIO).

The cultured organoids (usually 4–6 wells) were harvested and fixed in 4% paraformaldehyde overnight, then, wrapped in Histogel (Epredia, USA, HG-4000-012), dehydrated, and embedded in paraffin blocks. H&E and IHC were performed on the organoids as indicated above for tumor tissues. For immunofluorescence staining, Ki67 (CST 9449) and CK20 (Abcam, ab76126) were detected at the same time for CWH22 and CLM22 organoids. Images of immunofluorescence staining were obtained using confocal laser-scanning microscopy on a Nikon Digital ECLIPSE C1 system (Nikon Corporation, Japan).

Authentication of CWH22/CLM22 organoids

For authentication, DNA from CWH22 and CLM22 organoid cells and normal mucosa tissue from the patient were isolated using an AxyPrep Multisource Genomic DNA Miniprep Kit (Axygen, Corning, Inc., USA), and 21 STR loci were examined and compared with the corresponding STR profile of the normal tissue at the CCTCC. After bacterial and mycoplasma contamination tests were confirmed to be negative, the CWH22 and CLM22 organoid cell lines were deposited at the CCTCC (No. C202218, C202219).

WES analysis of organoids and tumor tissues

CWH22 and CLM22 organoids were harvested for DNA extraction after being cultured for 12 days. DNA from cultured organoids and liquid nitrogen-frozen tissues (normal mucosa, CRC, and LM) of the corresponding patient was extracted using a Zymo Quick-DNA Microprep kit (Zymo Research, USA #D2030) according to the manufacturer’s protocol. Library construction and whole-exome capture of genomic DNA were performed using the SureSelectXT Human All Exon V6 (Agilent, USA) and MGIEasy Exome Universal Library Prep Set-V1.0. The captured DNA was sequenced using a MGISEQ 2000 platform (BGI-Shenzhen, China) with 150 bp paired-end sequencing. A total of >0.4 μg DNA per sample was used as input. The effective sequencing depth was above 100× (10 G) per sample. The human genome data of Hg19 from the UCSC Genome Browser (http://genome.ucsc.edu/) was used as a reference, and common mutated genes in CRC and LM associated with CRC were analyzed for each sample.

Chromosome karyotype analysis

Metaphase chromosome preparations were conducted by treating organoids in the exponential growth phase (usually at 7–10 days) with 0.6 μg/mL colchicine for 5 h. Then, enzymatic dissociation was performed with TrypLE™ Express to disperse the organoids into single cell suspensions. The suspension was centrifuged at 1,000 × g for 6 min. Preheated 0.075 mM KCl hypotonic solution was added and incubated for 15 min at 37°C, then fixed with 3:1 methanol-acetic acid (v: v; 1 min). Subsequently, the pellets were fixed twice at room temperature for 10 min, and 500 μL stationary liquid was added to resuspend and collect the cells. Afterward, cell droplets were spotted on clean, precooled microscope glass slides. The slides were immediately heated over an alcohol lamp for 1 s to allow the chromosomes to spread out. Chromosome specimens were stained with Giemsa (#BA4219, Baso Diagnostics, China) for 10 min at 25–28°C according to the manufacturer’s instructions, and the chromosomes of M phase cells were counted using an oil immersion lens (100×) under a microscope (Nikon, Tokyo, Japan). Karyotypes were determined by arranging all the photographed metaphases. The chromosomes were classified according to the International System for Human Cytogenetic Nomenclature [69].

Organoid proliferation assay

When the organoids (CWH22 and CLM22) entered the logarithmic growth phase between days 9–11 of each passage, they were mechanically and enzymatically dissociated into single cells by incubation in TrypLE™ Express for 10–14 min and resuspended in BME at a concentration of 3 × 10⁵/mL separately on ice. To assess proliferation, the suspensions were seeded onto pre-warmed 96-well plates (Greiner Bio-one, 655090) in triplicate at 3 × 10³ cells per well (10 μL/well) on days 1, 4, 7, 10, and 14 and maintained in complete culture medium (100 μL/well) for 14 days. Quantification of cell viability was performed by replacing the culture medium with 50 μL of CellTiter-Glo 3D (#G9681, Promega, USA) mixed with 50 μL of culture medium on an Infinite 200 PRO plate reader (Tecan Life Sciences) according to the manufacturer’s instructions. Organoid proliferation rates were determined, and graphs were drawn using GraphPad Prism 8 (GraphPad Software, Inc.).

2D-cultured cell proliferation tests

The establishment of CWH22-2D and CLM22-2D was done from the subcutaneous tumors derived from 3D organoids CWH22 and CLM22. The tissues were digested into single cells and clusters as presented above and then seeded in culture plates with high-glucose DMEM containing 10% FBS. When cultured for about 2 months and passaged more than 4 times, the growth rates and cellular morphology became stable, allowing further analysis.

Cells (LoVo, SW480, CWH22-2D and CLM22-2D) were seeded onto 96-well plates at 4 × 10³ cells per well in 100 μL suspensions in triplicate per day and maintained in complete culture medium. The cell proliferation rate was measured using the Cell Counting Kit-8 assay (CCK-8; Cat. No. KJ800; Dojindo Laboratories, Kumamoto, Japan) according to the manufacturer’s instructions as previously reported [68].

Tumor marker detection

Clinically, serum tumor markers (CEA, CA19-9, and CA72-4) were detected before and after surgery, and the CEA and CA19-9 levels of the patient (from whom CWH22 and CLM22 were derived) were abnormal. The culture medium of the CRC/CLM-derived organoids at passage 20 (incubated for 3 days, 500 μL) was collected for tumor marker detection. All measurements were conducted in the Laboratory Medicine of Tongji Hospital under the same conditions.

IC₅₀ tests

In the present study, 5-fluorouracil (5-Fu; Cat. No. HY-90006), oxaliplatin (Cat. No. HY-17371), SN-38 (Cat. No. HY-13704; an active metabolite of irinotecan), VP-16 (Cat. No. HY-13629), regorafenib (Cat. No. HY-10331), and sotorasib (Cat. No. HY-114277) were selected for drug sensitivity tests (all purchased from MedChemExpress, USA). We chose 9–12 different concentrations of each drug to test, based on the results of preliminary experiments. Organoids were mechanically and enzymatically dissociated into single cells, resuspended in BME at 3 × 10⁵/mL, and seeded at 10 μL per well in prewarmed 96-well plates (Greiner Bio-one, 655090). The cells were maintained in complete culture medium (100 μL/well) as described above. Organoids were considered ready for IC₅₀ tests when they reached 100 μm in diameter, at which point the culture medium was replaced with media containing different drug concentrations. Those organoids were exposed to various drugs for six days, and the culture medium was changed on the third day. Subsequently, cell viability was assessed using the CellTiter-Glo 3D Cell viability assay (#G9683, Promega, USA ), as presented above. Dose-response curves were plotted, and the IC₅₀ values were calculated using GraphPad Prism 8.

MRTX1133 (a KRAS G12D inhibitor; Cat. No. HY-134813), BI-2865 (a pan-KRAS inhibitor; Cat. No. HY-153724), and sotorasib (a KRAS G12C inhibitor) were also introduced into the media containing the CWH22/CLM22 organoids (KRAS G12C mutated), CRC1-Org/CRC16-Org (KRAS G12D mutated), and CRC27-Org (wild-type RAS) to draw dose-response curves, calculate IC₅₀ values, and determine the sensitivity of KRAS targeted therapies in vitro.

Construction of organoids stably expressing luciferase

The luciferase gene was cloned into the pLVX-IRES-Neo plasmid using XhoI/EcoRI restriction sites to produce a luciferase-carrying lentivirus. Lentivirus supernatant was collected as previously reported [70]. To establish organoids stably expressing luciferase (CWH22-luci and CLM22-luci), single-cell suspensions of the CWH22 and CLM22 organoid cells (the second generation) were seeded on the surface of presolidified BME in 96-well plates (30 μL diluted BME, 5000 cells/well, 10 wells/organoid line). Approximately 48–72 h later, the lentivirus supernatant was used to transduce each organoid (nine wells for transduction, one well for selection control); 5 μg/mL polybrene was added to improve the transfection efficiency and was replaced with organoid culture medium after 12 h. At 48 h post-transduction, the culture medium was replaced with selection medium containing 500 μg/mL G418 for at least seven days to kill any cells that were not transfected. Once the cells in the control well died thoroughly, and the luciferase-expressing cells were grown into organoid structures 100 μm in diameter, the organoids were harvested and enzymatically dissociated into single cells using TrypLE™ Express for 10–14 min. Cells were subsequently passaged as described above. On the 10th day from the first passage post-transfection, one well of each organoid (CWH22-luci and CLM22-luci) was harvested and tested separately for luciferase activity on a Gloma 20/20 luminometer (Promega, USA) with 20 μL Vivo GloTM Luciferin reagent (10 mg/mL, Promega, P1403, USA), un-transfected organoids (CWH22-con and CLM22-con) were also harvested for control test.

Determination of tumorigenicity

A total of 20 BALB/c athymic nude mice were housed under specific pathogen-free conditions in a temperature- and humidity-controlled environment (room temperature, 20–26°C; humidity, 40–60%; 12-h light/dark cycle; free access to food and water). First, CWH22 and CLM22 organoids (the third generation) were harvested with BME and injected subcutaneously into the right hip of each mouse (1.5 × 10⁶ cells/mouse, five mice/organoid) to test tumorigenicity. On day 28, the mice were tested for the tumor–burdened status, and the subcutaneous tumors were subjected to histopathological analysis.

Next, the liver metastatic capacities of CWH22 and CLM22 organoids were tested in vivo. Ten female mice were divided into two groups, with five mice per organoid. Single-cell suspensions containing 1 × 10⁶ cells (CWH22-luci, CLM22-luci, total fifth generation) were injected into each anesthetized mouse spleen as described previously [71]. At the onset of weight loss, bioluminescence imaging indicated definite liver metastases in each mouse. On day 45, the mice were euthanized with an overdose of intraperitoneal pentobarbital sodium, and metastatic livers were harvested, measured, fixed in 4% paraformaldehyde overnight at 25–28°C, and embedded in paraffin.

The procedures for detecting the tumorigenicity and metastatic ability of CWH22-2D and CLM22-2D cells were the same as those described above for 3D organoids (1.5 × 10⁶ cells/mouse for subcutaneous injection, 1 × 10⁶ cells/mouse for spleen injection).

Statistical analysis

Statistical analyses were performed using SPSS 19 and GraphPad Prism 8. Independent sample t-test and one-way analysis of variance were used as appropriate. A p-value < 0.05 was considered to indicate statistical significance.

Availability of data and materials

The datasets used and analyzed in the current study are available from the corresponding author upon reasonable request. The WES dataset generated and analyzed in the present study is available in the NCBI repository (https://www.ncbi.nlm.nih.gov/bioproject/PRJNA902102). All the authors had access to the study data and reviewed and approved the final manuscript.