Surgical human colorectal adenocarcinoma samples were obtained with written informed consent and approval from the Institutional Review Board of Tongji Hospital, Tongji Medical College, Huazhong University of Science and Technology (Wuhan, China; IRB ID: 20141106); the experiments were conducted according to the principles of the Declaration of Helsinki. In total, 20 tumor specimens from CRC patients were included in the present study, and the patients were assigned case numbers CRC1-20. The patient clinical characteristics are listed in Table SI. The CRC specimens were disassociated into single primary CRC cells as described previously (15). Briefly, fresh specimens were minced into small sections with scissors. The completely minced pieces were then incubated in serum-free Dulbecco's modified Eagle's medium (DMEM)/F12 (Thermo Fisher Scientific, Inc., Waltham, MA, USA) containing 1.5 mg/ml collagenase IV (Gibco; Thermo Fisher Scientific, Inc.), 20 µg/ml hyaluronidase (Sigma-Aldrich; Merck KGaA, Darmstadt, Germany) and 1% penicillin/streptomycin (Thermo Fisher Scientific, Inc.) at 37°C for 1 to 2 h. To eliminate red blood cells, the cells were incubated in red blood cell lysis buffer (eBioscience; Thermo Fisher Scientific, Inc.) on ice for 10 min, then washed twice with PBS. The single primary CRC cells were then resuspended in PBS for use in subsequent organoid culture, sphere forming assay and animal studies.

Organoid culture was processed as previously described (15,16), with several modifications. For culture establishment, single primary CRC cells were embedded in Matrigel (growth factor reduced; phenol free; BD Biosciences, Franklin Lakes, NJ, USA) and seeded into 24-well culture plates at the indicated dosage (1,000 cells/30 µl Matrigel/well). Following Matrigel polymerization, the cells were overlaid with human colorectal cancer organoid culture medium and incubated at 37°C in humidified air containing 5% CO₂. The composition of the human colorectal cancer organoid culture medium was as follows: DMEM/F12 (Invitrogen; Thermo Fisher Scientific, Inc.) supplemented with 1X B27 (Invitrogen; Thermo Fisher Scientific, Inc.), 50 ng/ml recombinant human EGF (Sigma-Aldrich; Merck KGaA), 10 nM Gastrin (Sigma-Aldrich; Merck KGaA) and 500 nM A83-01 (Tocris Bioscience, Bristol, UK) (16). The medium was changed every 3 days. To calculate the forming efficiency, following 7 days post-embedding, organoids with diameters >100 µm were scored under an inverted microscope (Leica Microsystems, Wetzlar, Germany). The forming efficiency (%) = scored organoid number / total embedding CRC cells. For serial passages, following 7 days post-embedding, whole organoids were trypsinized using 0.025% trypsin/EDTA at 37°C for 15–20 min. Single organoid-derived cells were then resuspended with PBS and employed for a new round of organoid culture in Matrigel.

The sphere formation assay was conducted as previously described (17,18). Single primary CRC cells were resuspended in standard stem cell medium (SCM), which consisted of the following: DMEM/F12 (Invitrogen; Thermo Fisher Scientific, Inc.) supplemented with 1X B27 (Invitrogen; Thermo Fisher Scientific, Inc.), 20 ng/ml human recombinant epidermal growth factor (Sigma-Aldrich; Merck KGaA) and 20 ng/ml basic fibroblast growth factor (Sigma-Aldrich; Merck KGaA) (17). The spheres were cultured at 37°C in humidified air containing 5% CO₂. To calculate the forming efficiency, following 7 days post-plating, spheres with diameters >50 µm were scored under an inverted microscope (Leica Microsystems). The forming efficiency (%) = scored sphere number / total plating CRC cells. For serial passage experiments, at 7 days post-plating, spheres were trypsinized as aforementioned. Single sphere-derived cells were then replated in SCM for a new round of sphere forming assay.

Female, 4- to 6-week-old NOD/SCID mice (n=135) were purchased from Beijing HFK Bioscience Co., Ltd. (Beijing, China) and were maintained according to protocols approved by the Institutional Animal Care and Use Committee, Huazhong University of Science and Technology (Hubei, China; IACUC ID: 2014S652). During experimentation, all mice were kept in the animal center of the Tongji Medical college, Huazhong University of Science and Technology under specific pathogen-free (SPF) conditions (License Number: SYXK#2016-0057). For maintenance of mice, a stable breeding environment was provided: a 12-h/12-h light/dark, a temperature of 21°C and a relative humidity 50% and allowed access to water and food ad libitum. For the establishment of patient-derived xenografts (termed PDXs), n=40 mice were used, and the detailed information of PDX generation is presented in Table SI. For transplantation assays (n=5 mice were used), the single cells derived from CRC6 specimens and the corresponding organoids and spheres generating form them were resuspended in a PBS/Matrigel (BD Biosciences) mixture (1:1 volume) (15,17). The cells were then injected into the subcutaneous tissue of mouse flanks using 27-gauge needles at at dose of 125,000 cells/point. For histopathological analyses, the xenografts derived from primary cells (termed PDX, 1 mice), organoid cells (termed ODX, 2 mice) and sphere cells (termed SDX, 2 mice) were harvested on day 30 post-implantation. The mice were anesthetized and sacrificed according to the AVMA guidelines, and hematoxylin and eosin (H&E) staining was performed as previously described (15). For limited dilution assays (n=90 mice were used), 10,000, 1,000 and 100 single cells derived from organoids at 1st - 3rd generation, or 1,000, 100 and 10 single cells derived from spheres at 1st - 3rd generation were implanted per injection (5 mice for each dose at each group). For the limited dilution assays, the tumor volumes were examined every 3 days, and the mice with excessive weight loss (humane endpoint) post-injection were excluded from the study and euthanized (n=7), and these included: 1 mouse injected with organoid cells at 1st generation; 2 mice injected with organoid cells at 2nd generation; 4 mice injected with organoid cells at 3rd generation. Following sacrifice, tumors were removed from the mice and weighed to evaluate tumor development. The frequency of tumor-initiating cells and statistical significance were examined using the Extreme Limiting Dilution Analysis software (bioinf.wehi.edu.au/software/elda/index.html).

The immunofluorescence procedures for organoids and spheres were performed as previously described (19). Prior to staining, organoids or spheres were fixed in 4% PFA at 4°C for 20 min. For immunofluorescence staining, the following antibodies were used to detect antigens: Mouse-anti-human cluster of differentiation (CD)-133 (1:100; cat. no. 66666-1-lg; ProteinTech Group, Inc., Chicago, IL, USA) and rabbit-anti-human cytokeratin (CK)-20 (1:100; cat. no. 13063; Cell Signaling Technology, Inc., Danvers, MA, USA). Briefly, the sections were blocked with 5% (w/v) bovine serum albumin (Invitrogen; Thermo Fisher Scientific, Inc.) in PBS, then incubated with the primary antibodies at 4°C overnight, followed by staining with secondary antibodies conjugated to Streptavidin-Cy3 (1:100; cat. no. SA1010; Thermo Fisher Scientific, Inc.) or Alexa Flour 488 (1:100; cat. no. 705-546-147; Jackson ImmunoResearch Laboratories, Inc., West Grove, PA, USA) at room temperature for 2 h. Finally, nuclei were stained with 4′,6-diamidino-2-phenylindole (DAPI, Sigma-Aldrich Merck KGaA) at room temperature for 10 min. Images of organoids and spheres were captured under a fluorescence microscope (TRRFM; Olympus Corp., Tokyo, Japan) or a confocal microscope (FV1000; Olympus Corp.).

Total RNA was extracted using RNAiso Plus (Takara Bio, Inc., Otsu, Japan), and then cDNA was synthesized using the Mixima First Strand cDNA Synthesis kit (Thermo Fisher Scientific, Inc.). RT-qPCR analysis was conducted using an ABI PRISM 7300 Sequence Detection System instrument with Maxima SYBR-Green/ROX qPCR Master Mix (Thermo Fisher Scientific, Inc.) according to the manufacturer's instructions. The thermocycling conditions for RT-qPCR were: step1: 50°C for 2 min; step 2: 95°C for 2 min; step 3 (×40 cycles): 95°C for 15 sec and 60̊C for 1 min. The Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was used as an internal control, and the gene expression was normalized to the GAPDH to calculate relative expression level using the 2^−ΔΔCq method (20). The sequences of the primers were as follows: GAPDH forward, 5′-TCGTGGA AGGACTCATGACC-3′ and reverse, 5′-TCCACCACCCTGTT GCTGTA-3′; CD44 forward, 5′-AGCAACCAAGAGGCAAG AAA-3′ and reverse, 5′-GTGTGGTTGAAATGGTGCTG-3′; CD133 forward, 5′-TTCTTGACCGACTGAGACCCA-3′ and reverse, 5′-TCATGTTCTCCAACGCCTCTT-3′; ABCG2 forward, 5′-TCCATATCGTGGAATGCTGA-3′ and reverse, 5′-TTTCAGCCGTGGAACTCTTT-3.

The TCF/LEF reporter, which drives the expression of GFP (TOP-GFP) lentivirus, was purchased from SBO Medical Biotechnology Co. (Shanghai, China) (18). Primary cancer cells were infected as previously described (21). Primary cells were infected with TOP-GFP lentivirus at an MOI of 25 for 72 h. To separate the GFP⁺/GFP- cells, the top (GFP⁺) and bottom (GFP-) 5–10% cells were purified out by flow cytometry. To evaluate Wnt activity, the intensity and frequency of GFP⁺ cells were detected using a fluorescence microscope (TRRFM; Olympus Corp.) or flow cytometry according to the manufacturer's instructions of a FACS Aria II Cell Sorter (BD Biosciences) (18).

Fluorescence-activated cell sorting (FACS) was performed according to the manufacturer's instructions using a FACS Aria II Cell Sorter (BD Biosciences) (18). To measure CD133 expression in the organoids and spheres, the cells were stained with phycoerythrin (PE)-conjugated mouse anti-human CD133 (1:11; cat. no. 130-090-756; Miltenyi Biotec GmbH, Bergisch Gladbach, Germany), then analyzed using a BD FACS Aria II flow cytometer (18). To separate CD133⁺/CD133⁻ cells in primary CRC, tumor specimens were processed into single cells as described above (15). The cells were then stained with PE-conjugated mouse anti-human CD133 at 4°C for 15 min. For purification, only the top (CD133⁺) and bottom (CD133^−/lo) 10–20% cells were purified out (18).

The cell death of organoid- and sphere-derived cells was assessed using the Cell Counting kit-8 (CCK-8; Dojindo Molecular Technologies, Inc. Rockville, MD, USA), as well as with 5-fluorouracil (5-Fu) and oxaliplatin (OXA) treatment, which were purchased from Sigma-Aldrich (Merck KGaA). Briefly, the cells were seeded in complete medium at 30,000 cells/well in 96-well plates. Following 12 h, the cells were treated with either 5-Fu (1 µM) or OXA (1 µM). After 72 h, 10 µl CCK-8 solution was added to each well. The plates were then incubated at 37°C for 1 h, and cell viability was determined by scanning with a microplate reader (Thermo Fisher Scientific, Inc.) at 450 nm.

Statistical significance was calculated with SPSS Statistics 18.0 software (IBM Corp., Armonk, NY, USA). All measurement data are presented as the means ± standard deviation of at least 3 independent experiments. The measurement data were analyzed using the Student's t-test or one-way analysis of variance followed by a Tukey's post hoc test to determine the significant differences of means in two or multiple groups (n>2) comparisons. The enumeration data were analyzed using the Fisher's exact test to determine the significant differences of rates in two categories comparisons. P<0.05 was considered to indicate a statistically significant difference.

Sphere formation assay (22) and organoid culture (23) are 3D models that support the long-term expansion of primary tumor cells in vitro. In this study, to investigate whether these 3D models accurately simulate the tumor heterogeneity of CRC tumors, we generated spheres and organoids side by side using freshly surgical CRC specimens. H&E staining revealed that the organoids harbored parental tumor-like budding structures, while the spheres were observed as solid globes (Fig. 1A). Notably, the organoids formed more complex structures in the cells derived from high-grade tumors (such as Dukes' C and D), whereas the spheres retained similar global shapes (Fig. 1B). Since cells expressing CD133 have been reported to enrich for CSCs in CRCs (24) and cytokeratin 20 (CK20) is a differentiated cell marker (25), the present study conducted further CD133 and CK20 staining for spheres and organoids derived from primary CRCs. The results revealed that the expression of CD133 and CK20 more closely resembled that of primary tumors in the organoids than in the spheres (Fig. 1C). Quantitative analysis demonstrated that the organoids harbored similar percentages of CD133⁺ cells when compared with the corresponding primary tumors; however, sphere-forming assays significantly increased the percentage of CD133⁺ cells (Fig. 1D), implying that organoid culture may maintain the cellular heterogeneity of parental CRCs. These results demonstrate that, compared with the sphere-forming assay, organoid culture may more effectively simulated the tumor heterogeneity of primary CRCs in vitro.

In general, tumor cells are able to generate xenografts in immuno-compromised mice subcutaneously, and the PDXs can recapitulate original tumor heterogeneity (5). In order to explore the tumor heterogeneity of CRC tumors, it is also crucial to establish successful cell culture in vitro (5,26,27). However, whether cells cultured in 3D models preserve the ability to generate parental tumor-like xenografts (i.e., PDXs) remains unclear. Consistent with the findings of previous studies (5,7,25), the results of the present study demonstrated that primary CRC cells and their corresponding organoids and spheres were all capable of generating tumor xenografts in NOD/SCID mice (Fig. 2A). To determine whether ODXs and SDXs exhibit the same tumor heterogeneity of primary CRCs, the present study performed CK20 (25) staining for primary CRC tumors, PDX, ODX and SDX. As shown in Fig. 2B, the present study revealed that the expression pattern of CK20 in ODX more closely resembled primary tumors and the corresponding PDX than SDX (Fig. 2B), suggesting that organoid culture more accurately reproduced the tumor heterogeneity of primary tumors than the sphere formation assay. In order to examine the efficiency of generating organoids or spheres from primary CRC tumors, the present study performed side-by-side organoid culture and sphere-forming assays for CRC specimens (Table S1). The results revealed that organoids in 15 of the 20 CRC specimens were successfully generated (success rate, 75%), whereas spheres were only generated for 5 of the 16 CRC specimens (success rate 31%; Tables I and S1). Notably, the primary CRC cells formed more organoids than spheres when the same cell dosage was applied (Fig. 2C and D). Taken together, these results demonstrate that organoid culture possesses a higher success rate and better efficiency to simulate primary colorectal tumors than sphere-forming assay.

The cell surface marker, CD133, is widely used to enrich CSCs in CRCs (16,23). In the present experimental system, using FACS, CRC cells expressing high levels of CD133 (CD133⁺) and those expressing little or no CD133 (CD133⁻) were assessed and sorted. CD133⁺ CRC cells were found to produce more spheres and organoids than CD133⁻ cells, implying that sphere- and organoid-forming cells were enriched for CSCs (Fig. 3A). Serial sphere formation assays are generally applied to enrich and expand CSCs (27). However, the dynamics of CSCs in serial organoid cultures remain unclear. The results of the present study revealed that the levels of CD133 and CD44, the two cell surface markers widely used to enrich CSCs in CRCs (24,29), remained constant in serial organoid cultures, while they significantly and gradually increased in serial sphere formation assays, indicating that the dynamics of CSCs may differ between organoid culture and sphere-forming assays during serial passages (Fig. 3B). To visualize CSCs and differentiated cells within individual organoids and spheres during serial passages, the present study performed CD133 and CK20 immunofluorescence staining for organoids and spheres (Fig. 3C and D). Quantification analysis revealed that the proportion of CD133⁺ and CK20⁺ cells remained relatively constant in organoids during serial passages; however, there were increased levels of CD133⁺ cells and decreased levels of CK20⁺ cells in spheres during serial passages (Fig. 3E and F). Taken together, these results demonstrate that organoid culture produces steady ratios of CSC subsets in long-term cultures, whereas sphere-forming assays may enrich CSCs.

TICs, also known as CSCs, are a rare population within CRC (30). In this study, to investigate whether the frequency of TICs was altered during serial transplantations, we performed limited dilution assays to evaluate the tumorigenic capacity of organoid- and sphere-derived cells in each generation (15,30). As shown in Fig. 4A, during serial transplantations (1st to 3rd), organoid-derived cells possessed a constant number of TICs. On the other hand, serial sphere-forming assays had significantly increased numbers of TICs with serial passaging (Fig. 4B). Notably, it was observed that organoid-derived cells initiated tumor generation at almost the same time in each generation, while sphere-derived cells initiated tumor generation earlier than the former generation (Fig. 4A and B). Consistent with this finding, there were no significant differences among the tumor volumes of organoid-derived xenografts in serial transplantations (Fig. 4C), while sphere-derived cells developed increasingly larger tumors in serial transplantations (Fig. 4D). Taken together, these results indicate that organoid culture maintains the frequency of TICs (i.e., CSCs) in long-term culture, while sphere-forming assays enrich TICs.

It has been demonstrated that a high Wnt activity functionally designates the colon CSC population (18,31–33). To further evaluate the alternations in Wnt activity during serial passages, in the present study, we infected purified primary CRC cells with lentivirus, in which the TCF/LEF reporter drove the expression of GFP (TOP-GFP) and represented the cellular levels of Wnt activity (18,31,32). The transfected cells were then utilized in organoid culture and sphere-forming assay; when the organoids and spheres were growing, the expression of GFP was evaluated. The results revealed that the intensity of GFP expression was stronger in the spheres than the organoids (Fig. 5A), indicating that spheres were more highly enriched for CSCs than organoids during culture. The present results also demonstrated that during serial passage, the number of cells expressing GFP (i.e., GFP⁺ or Wnt⁺ cells) gradually increased in serial sphere-forming assays, but remained relatively constant in serial organoid cultures (Fig. 5B and C). Consistent with these findings, FACS analysis further demonstrated that the intensity of Wnt⁺ cells remained unaltered in organoids during serial passaging, while they markedly increased in spheres with serial passaging (Fig. 5D). Collectively, these data indicated that organoid cultures preserved relatively constant numbers of Wnt⁺ cells and the level of Wnt signaling in the cells during long-term culture, whereas the sphere-forming assay enriched Wnt⁺ cells and increased the intensity of Wnt signaling in the cells.

As the presence of CSCs and the activation of Wnt signaling have been demonstrated to be associated with the chemoresistance of tumor cells (31–33), we hypothesized assumed that serial sphere-forming assays, which increased the proportion of CSCs and improved the levels of Wnt signaling during serial passages, may also increase the frequency of chemoresistant cells; whereas, organoid culture may only maintain the chemoresistant capacity in the cells during serial passages. To examine this hypothesis, the present study performed serial organoid cultures or sphere-forming assays combined with conventional chemotherapeutic agents (oxaliplatin or 5-fluorouracil), and then detected the efficiency of organoid or sphere formation. As shown in Fig. 6A–D, the results indicated that, upon the administration of chemotherapeutic agents, the efficiency of organoid and sphere formation decreased; however, along with serial passages, this impairment was partially rescued in serial sphere-forming assays, but remained unchanged in organoid culture. Furthermore, in each generation, the present study detected the frequency of Wnt⁺ cells and the expression levels of adenosine triphosphate binding cassette subfamily G member 2 (ABCG2), one of the resistant markers of CRC (34), in organoids and spheres treated with chemotherapeutic agents. The results verified that, the frequency of Wnt⁺ cells, as detected by FACS, and the expression level of ABCG2, as determined by RT-qPCR, significantly increased in spheres and organoids following treatments with chemotherapeutic agents; however, during serial passaging, these markers remained unaltered in the organoid culture, while they gradually increased with serial sphere formation (Fig. 6E–H). The half-maximal inhibitory concentration (IC₅₀) analysis upon OXA and 5-Fu treatment suggested that, since the 1st generation, organoid cells exhibited a similar IC₅₀ to that of primary cells, while the sphere forming assay significantly increased the original IC₅₀ (Fig. 6I and J). These data, obtained from CRC6, indicated that organoid culture preserved the characteristics of chemoresistance in primary CRC cells. To verify the universality, a chemoresistance assay in serial passages was respectively performed in CRC1, CRC5 and CRC9. The subsequent CCK8 assay further demonstrated that the frequency of chemoresistant cells gradually increased in serial sphere-forming assays, while it remained relatively constant in serial organoid cultures (Fig. 6K and L). Overall, the results indicated that, when compared with sphere-forming assays, organoid culture can preserve the frequency of chemoresistant tumors cells in the long-term culture, and may serve as a better pre-clinical model for drug screening and discovery.