The study was conducted according to ethical standards, the Declaration of Helsinki, and national and international guidelines, and it was approved by the Institutional Review Board of Union Hospital, Tongji Medical College, Huazhong University of Science and Technology (China).

The graphs of Nkx2.5 expression profiles in CRC and adjacent NCM samples were generated and downloaded from Oncomine microarray database (https://www.oncomine.org).

Here, 14 pairs of primary human CRC samples and their matched adjacent NCM samples were obtained from CRC patients who were admitted to our hospital and underwent surgical resection in 2015 (see Supplementary Table 1 ). Samples were collected according to the protocol approved by the Institutional Review Board. The patients signed the written informed consent form prior to commencing the study. All the enrolled individuals were Han Chinese. The data were analyzed anonymously.

Human FHC, Hela, Caco-2, DLD-1, HCT116, HT-29, SW480, RKO, SW620, LoVo, HEK-293T cells, and H9c2 cells were cultured in a medium recommended by the American Type Culture Collection (ATCC; Manassas, VA, USA) that was supplemented with 10% fetal bovine serum (FBS; HyClone, Logan, UT, USA) and 1% penicillin-streptomycin (Gibco, Carlsbad, CA, USA) at 37°C in a 5% CO2 incubator (Thermo Fisher Scientific, Waltham, MA, USA). Data related to the original tumors or xenografts of CRC cell lines (49) are listed in Supplementary Table 2 . The identities of the cancer cell lines were confirmed by short tandem repeat (STR) profiling.

The Nkx2.5 overexpression plasmid was generated by cloning the full-length of cDNA representing the complete ORF of Nkx2.5 into the pSi-Flag vector (Invitrogen, Carlsbad, CA, USA). Nkx2.5-F (5’-ATGGATCCAATGTTCCCCAGCCCTGCTCT-3’) and Nkx2.5-R (5’-TGCTACCAGGCTCGGATACCATGCAGCGT-3’) were chosen as primers.

Lentiviral vectors were transfected into HEK-293T cells in combination with the lentiviral packaging vectors (pRSV-Rev, pMD2.G, and pCMV-VSV-G) using Lipofectamine™2000 reagent (Invitrogen, Carlsbad, CA, USA). After transfection for 48 h, supernatants were collected, filtered, and used to establish the Nkx2.5-overexpression cell lines (Lenti-Nkx2.5-HCT116 and Lenti-Nkx2.5-SW480). Empty vectors were used to generate control cell lines (Lenti-NC-HCT116 and Lenti-NC-SW480). Colonies resistant to Geneticin (G418; Invitrogen, Carlsbad, CA, USA) were picked and expanded to obtain stable clone stock cells. Western blot and quantitative reverse transcription-polymerase chain reaction (RT-qPCR) assays were employed to indicate whether stable cell lines could be successfully generated.

Total RNA was isolated using TRIzol reagent (Invitrogen, Carlsbad, CA, USA) according to the manufacturer’s instructions. 2 μg of total RNA was reversely transcribed using an RNA PCR kit (Takara Biotechnology, Inc., Otsu, Japan), and resulting cDNA was used as a template for a standard PCR. The mRNA levels were detected by RT-qPCR with an ABI PRISM 7900 Sequence Detector system (Applied Biosystem, Foster City, CA, USA) according to the manufacturer’s instructions, and normalized to glyceraldehyde-3-phosphate dehydrogenase (GAPDH) gene expression. The primers used for RT-qPCR are listed in Supplementary Table 3 (33).

Total cells and tissues were lysed using radioimmunoprecipitation assay (RIPA) buffer, and the protein concentration was determined with a bicinchoninic acid (BCA) protein assay kit (Pierce Biotechnology Inc., Rockford, IL, USA). Protein extracts were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE; Invitrogen, Carlsbad, CA, USA), and were then transferred onto a nitrocellulose membrane (Millipore, Billerica, MA, USA), which was incubated with various primary antibodies overnight at 4°C. After incubation with horseradish peroxidase (HRP)-conjugated secondary antibodies (dilution, 1:5000) for 1 h at room temperature, the membranes were treated with normal enhanced chemiluminescence (ECL) substrate (170-5061; Bio-Rad Laboratories Inc., Hercules, CA, USA) prior to visualization using a ChemiDoc MP imaging analysis system (Bio-Rad Laboratories Inc., Hercules, CA, USA) according to the manufacturer’s instructions, as previously described (50). The specific protein levels were normalized to GAPDH on the same nitrocellulose membrane. The following primary antibodies were used: anti-Nkx2.5 (sc-376565X; dilution, 1:500; Santa Cruz Biotechnology, Dallas, TX, USA), anti-GAPDH (sc-32233; dilution, 1:1000; Santa Cruz Biotechnology, Dallas, TX, USA), anti-p53 (#2527; dilution, 1:1000; Cell Signaling Technology, Danvers, MA, USA), and anti-p21^WAF1/CIP1 (#2947; dilution, 1:1000; Cell Signaling Technology, Danvers, MA, USA).

For colony formation, 1000 cells/well were seeded into 6-well plates. After 14 days of culture, cells were stained with crystal violet, and the cell-covered area of visible colonies was calculated. Cell viability and cell proliferation were determined by 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) and EdU cell proliferation assays, respectively. For the MTT assay, 1000 cells in 200 µL culture medium were seeded into each well of 96-well plates. After cultivation, MTT was added into each well. Optical density (OD) was measured and cell growth curves were drawn according to the OD value. For the EdU cell proliferation assay, as described previously (51), EdU DNA Cell Proliferation kit (RiboBio Co. Ltd., Guangzhou, China) was used (52). Briefly, cells (1×10⁵) were cultured in 24-well plates, and then, exposed to 50 µM EdU for 2 h at 37°C. The cells were fixed with 4% formaldehyde and permeabilized in 0.5% Triton X-100. Cells were then incubated with 200 µM Apollo reaction cocktail, and nuclei were stained with Hoechst 33342 (200 mL/well). The stained cells were visualized under a fluorescence microscope (IX71; Olympus, Tokyo, Japan). The number of cells was counted using Image-Pro Plus 6.2 software (Meyer Instruments, Inc., Houston, TX, USA).

Migration of cells was evaluated by wound-healing assay via uncoated Transwell^® cell culture inserts according to the manufacturer’s instructions (Corning Inc., New York, NY, USA). To carry out wound-healing assay, monolayer of cells was gently scratched with sterile micropipette tips, washed with phosphate-buffered saline (PBS), and incubated in a FBS-free medium with mitomycin. The areas of the gap at 0 h and the residual gap at 24 h after wounding of 10 random locations were compared. For the uncoated Transwell^® assay, the uncoated Transwell^® filter inserts with 8-µm pores (BD Biosciences, Bedford, MA, USA) in 24-well cell culture plates were used. Then, 1 × 10⁵ cells were suspended and seeded into the uncoated and precoated upper chambers of 24-well Transwell^® plates with a FBS-free medium. The medium containing 10% FBS was added to the lower chamber to serve as a chemo-attractant. After 12 h of incubation, migrated cells were stained with 0.5% crystal violet solution. Cell migration was quantified by calculating the cell-covered area in photomicrographs using Image-Pro Plus v6.2 software.

For cell apoptosis analysis, cells were cultured in FBS-free medium for 48 h, followed by staining with the Annexin V-fluorescein isothiocyanate (FITC)/7-amino-actinomycin D (7-AAD) (BD Biosciences, San Diego, CA, USA). For each assay, 1 × 10⁵ cells were incubated with Annexin V-FITC/7-AAD. After addition of 400 μL binding buffer, the cells were analyzed by fluorescence-activated cell sorting using FACScan (BD Biosciences, San Diego, CA, USA). Cell populations were classified as viable (Annexin V negative, 7-AAD negative), apoptotic (Annexin V positive, 7-AAD negative or positive), or necrotic (Annexin V negative, 7-AAD positive).

Athymic nude mice (strain BALB/c nu/nu; Charles River Laboratories, Wilmington, MA, USA) were used for tumorigenesis studies. These mice (age, 8-12-week-old) were housed and maintained under specific pathogen-free conditions in facilities approved by the Hubei Provincial Association for Laboratory Animal Sciences (China). For the subcutaneous xenograft mouse model, HCT116 cells (Lenti-Nkx2.5-HCT116 and Lenti-NC-HCT116, 1×10⁶ cells/well) were harvested and resuspended in Hanks’ Balanced Salt Solution. These two HCT116 cells were subcutaneously injected at day 0 into the right and left dorsal flanks, respectively. Tumor diameter was measured every 2 days before harvesting over the course of 32 days. Tumor volume (mm³) was calculated by measuring the shortest (x) and longest (y) diameters of the tumor using the following formula: tumor volume (mm³) = x²y/2.

In-situ cell death was determined by TUNEL staining of tumor tissue sections (In Situ Cell Death Detection Kit, cat. no. 11684817910; Roche Applied Science, Penzberg, Germany). Staining was performed in strict accordance with the manufacturer’s instructions. Briefly, frozen slides were dried and fixed with 4% paraformaldehyde. The Proteinase K solution and 1% Triton X-100 were added. TdT (terminal deoxynucleotidyl transferase), fluorescein-isothiocyanate (FITC)-labeled d-UTP mixed solution, and DAPI (4’,6-diamidino-2-phenylindole) were added for labeling and staining. After washing with PBS, anti-fluorescence quenching mounting medium was added for sealing. The slides were observed and photographed under a fluorescence microscope (IX71; Olympus, Tokyo, Japan). The cells with green fluorescence were apoptosis-positive cells. TUNEL-positive cells were counted using Image-Pro Plus v6.2 software in 10 randomly selected fields from each slide.

The fraction of proliferative cells in the xenograft tumor section was assessed with Ki-67 staining. Paraffin-embedded tissue sections were deparaffinized in xylene and rehydrated through graded ethanol solutions. Sections were boiled in antigen retrieval solution (EnVision FLEX Target Retrieval Solution; Agilent Technologies, Inc., Santa Clara, CA, USA) for antigen retrieval. Afterwards, endogenous peroxidases were inactivated by immersing in 3% hydrogen peroxide. The slides were blocked with 10% bovine serum albumin (BSA), incubated with the anti-Ki-67 antibody (GA62661-2; Agilent Technologies, Inc., Santa Clara, CA, USA) overnight, and then, with the corresponding secondary antibody for 1 h at room temperature. Finally, the sections were developed with DAB color solution. Sections were counterstained with hematoxylin and preserved with resinene. The cells with brown-stain were proliferation-positive cells. Ki-67-positive cells were counted using Image-Pro Plus v6.2 software in 10 randomly selected fields from each slide.

Total genomic DNA was extracted from each cell line (FHC, DLD-1, HCT116, HT29, SW480). Eight pairs of primers (see Supplementary Table 4 ) were used to amplify the exonic regions of TP53. PCR amplification was performed using MyGene™ Series Peltier Thermal Cycler (A300; Hangzhou LongGene^® Scientific Instruments Co., Ltd., Zhejiang, China) and PrimeSTAR GXL DNA Polymerase (cat. no. R050Q; Takara Biotechnology, Inc., Otsu, Japan). DNA sequencing was carried out by using an 3730xl DNA Analyzer (Applied Biosystem, Foster City, CA, USA). All fragments were sequenced from both strands. DNA sequencing analysis was undertaken by using Chromas software (Technelysium Pty Ltd., Queensland, Australia).

Co-IP was conducted via a Co-IP kit (Pierce^™ Co-Immunoprecipitation-Kit, No. 26149; Thermo Fisher Scientific, Waltham, MA, USA), according to the manufacturer’s instructions as previously reported (53, 54). Briefly, anti-Nkx2.5 antibody (sc-376565X; Santa Cruz Biotechnology, Dallas, TX, USA) was immobilized by incubating with AminoLink plus coupling resin in the Pierce spin columns to prepare anti-Nkx2.5-coated resin, and control mouse IgG antibody (sc-2025; Santa Cruz Biotechnology, Dallas, TX, USA) was used to prepare IgG-coated resin with the same procedure. Total cell protein extracts of HCT116 cells (Lenti-NC-HCT116 and Lenti-Nkx2.5-HCT116) were prepared using ice-cold IP lysis buffer. Extracts were added to the Pierce spin columns with antibody-coated resins, resulting in making four combinations (Lenti-NC-HCT116 + IgG, Lenti-NC-HCT116 + Anti-Nkx2.5, Lenti-Nkx2.5-HCT116 + IgG, and Lenti-Nkx2.5-HCT116 + Anti-Nkx2.5). After incubation and elution, the eluates containing the protein complexes were collected. Finally, aliquots were denatured by heating at 100°C and analyzed by SDS-PAGE and Western blotting.

For luciferase reporter vector construction, the pGL3.0 p21^WAF1/CIP1 luciferase reporter vector (p21^WAF1/CIP1-Luc) was provided by Promega (Madison, WI, USA). Luciferase activity assay was undertaken as previously described (55, 56). In brief, the reporter vectors were transfected into HCT116 cells (Lenti-NC-HCT116 and Lenti-Nkx2.5-HCT116) using Lipofectamine 2000 reagent (Invitrogen, Carlsbad, CA, USA). Then, pRL-SV40 vector expressing Renilla luciferase was co-transfected as an internal control to normalize transfection efficiency. Luciferase activities were measured 24 h after transfection. The relative luciferase activity was calculated as luciferase activity of reporter vectors in HCT116 cells (Lenti-NC-HCT116 and Lenti-Nkx2.5-HCT116).

The statistical analysis was performed by using GraphPad Prism 8.4 software (GraphPad Software Inc., La Jolla, CA, USA). All data were expressed as mean ± standard deviation (SD). The significant differences were analyzed by the log-rank test for KM survival analysis, Wilcoxon matched-pairs signed-rank test and paired t-test for comparing paired data between two groups, Mann-Whitney U test and unpaired t-test with Welch’s correction for comparing unpaired data between two groups, and Mann-Whitney U test or Brown-Forsythe and Welch’s analysis of variance (ANOVA) with Dunnett’s post-hoc test for making multiple comparisons. P < 0.05 was considered statistically significant.

To indicate whether Nkx2.5 expression differs between CRC and NCM samples, microarray data that were extracted from ONCOMINE database were analyzed. A significantly higher expression of Nkx2.5 was detected in colorectal adenoma (Gaspar Colon dataset) and CRC tissue (TCGA Colorectal dataset) than that in NCM tissue ( Figure 1A ). To verify the results from database, Nkx2.5 profiles from 14 paired CRC and NCM samples ( Supplementary Table 1 ) were examined by Western blotting and RT-qPCR. Nkx2.5 protein level (P = 0.0052 for Wilcoxon matched-pairs signed-rank test, P = 0.0139 for paired t-test) and mRNA expression (P = 0.0085 for Wilcoxon matched-pairs signed-rank test, P = 0.0264 for paired t-test) were observed to be significantly higher in CRC samples than those in NCM samples ( Figures 1B–D ).

Protein and mRNA levels of Nkx2.5 in a panel of eight CRC cell lines (Caco-2, DLD-1, HCT116, HT-29, SW480, RKO, SW620, and LoVo), one normal colon epithelial cell (FHC), and one rat cardiomyocyte (H9c2, Nkx2.5-positive cell) were analyzed. Protein and mRNA levels of Nkx2.5 were high in H9c2, low in FHC, and diversely expressed among CRC cell lines ( Figures 2A, B ). We analyzed the correlation between histological grades (49) of the original tumors ( Figure 2C ) (57–59) and Nkx2.5 levels in the CRC cell lines ( Figure 2B ). It was observed that higher Nkx2.5 protein (P = 0.0286 for Mann-Whitney U test, P = 0.0051 for unpaired t-test with Welch’s correction; Figure 2D ) and mRNA (P = 0.0286 for Mann-Whitney U test, P = 0.0443 for unpaired t-test with Welch’s correction; Figure 2D ) levels significantly correlated with higher histological grades.

To explore the function of Nkx2.5, Nkx2.5-overexpressed cells (Lenti-Nkx2.5-HCT116) were established using a lentiviral vector in HCT116 cells. Being derived from a moderately poor differentiated CRC cancer, HCT116 cell line showed to have a moderate proliferation rate and apoptosis rate among CRC cell lines (57–59). Thus, HCT116 could be an appropriate cellular model to show how the proliferation and/or apoptosis was affected (60). Moreover, Nkx2.5 was moderately expressed in HCT116 cells ( Figures 2A, B ), thus overexpression of Nkx2.5 in HCT116 cells may result in a positive phenotype. HCT116 with Nkx2.5 overexpression showed significantly higher Nkx2.5 protein and mRNA levels ( Figure 3A ; Mann-Whitney U test: P < 0.01; unpaired t-test with Welch’s correction: P < 0.0001) than those in the control cells with an empty vector (Lenti-NC-HCT116).

In colony formation assay, Nkx2.5 overexpression dramatically reduced the colony formation efficiency in HCT116 cells ( Figure 3B , Mann-Whitney U test: P < 0.01 and unpaired t-test with Welch’s correction: P < 0.0001). In MTT and EdU proliferation assay, Nkx2.5 overexpression significantly inhibited cell growth and decreased proliferation rate in HCT116 cells ( Figures 3C, D , Mann-Whitney U test: P < 0.0001 and unpaired t-test with Welch’s correction: P < 0.0001). These results suggest that Nkx2.5 expression contributed to the reduced growth or proliferation in HCT116 cells. Additionally, flow cytometry revealed a remarkably increased apoptotic cell fraction in Nkx2.5 overexpressed cells (26.57 ± 3.91% versus 13.84 ± 1.80%; Figure 3E , Mann-Whitney U test: P < 0.01 and unpaired t-test with Welch’s correction: P < 0.001). In wound-healing and Transwell assays, Nkx2.5 overexpression did not affect HCT116 cell migration. ( Figure 3F ).

Taken together, the above-mentioned findings demonstrate that Nkx2.5 plays a significant role in suppressing proliferation and promoting apoptosis of HCT116 cells in vitro. It may also be noted that migration of HCT116 cells was not influenced by Nkx2.5 overexpression.

For in vivo assay, Lenti-Nkx2.5-HCT116 and Lenti-NC-HCT116 cells were subcutaneously injected into nude mice. Nkx2.5 overexpression remarkably inhibited formation and growth of tumor cells in vivo (P < 0.01 for Mann-Whitney U test; Figure 4A ), according to the measurement of the volume and weight of xenograft tumors (P < 0.01 for Mann-Whitney U test; Figure 4B ). Moreover, Ki-67 staining and TUNEL staining showed a lower proliferation rate (P < 0.01 for Mann-Whitney U test; Figure 4C ) with a higher apoptosis rate (P < 0.01 for Mann-Whitney U test; Figure 4D ) in Nkx2.5 overexpressed xenograft tumors. These findings suggest that Nkx2.5 functions as a tumor suppressor in HCT116 cells in vivo.

In order to confirm the role of Nkx2.5 in CRC cells, studies were also carried out in a second cell line SW480 which is derived from a poorly differentiated CRC having high proliferation rate and a low apoptosis rate among CRC cell lines (57–59). It is therefore highly appropriate for testing of the tumor suppressive effect of a gene or a drug. Moreover, in contrast to the low expression of Nkx2.5 in HCT116 cells, Nkx2.5 was highly expressed in SW480 cells. Thus, we attempted to assess whether a higher Nkx2.5 expression could suppress SW480 cells. Then, Nkx2.5-overexpressed and the control SW480 cells (Lenti-NC-SW480 and Lenti-Nkx2.5-SW480) were established. The transfection efficiency was confirmed by Western blotting and RT-qPCR ( Figure 5A , P < 0.01).

Overexpression of Nkx2.5 in SW480 cells did not influence neither the efficiency of colony formation ( Figure 5B ) nor cell growth evaluated by MTT assay ( Figure 5C ). EdU proliferation assay showed no significant difference in proliferation rate between Lenti-NC-SW480 and Lenti-Nkx2.5-SW480 cells ( Figure 5D ). In addition, flow cytometry showed no significant difference in apoptosis rate between Lenti-NC-SW480 and Lenti-Nkx2.5-SW480 cells (4.99 ± 1.13% versus 4.80 ± 2.52%; Figure 5E ). These outcomes suggest that overexpression of Nkx2.5 did not affect proliferation or apoptosis of SW480 cells. In wound-healing and Transwell assays, the migration of SW480 cells also could not be affected by Nkx2.5 overexpression ( Figure 5F ).

In contrast to the results in HCT116 cells, these results suggest that Nkx2.5 overexpression could not affect proliferation, apoptosis, or migration of SW480 cells.

In order to determine why different CRC cell lines exhibit different patterns when Nkx2.5 is overexpressed, we measured the levels of antitumor protein p53 (38, 40, 61–63) and its downstream effector p21^WAF1/CIP1 (64, 65) since Nkx2.5 behaves like p53 and it may interact with p53-related pathway as another NK2 family member Nkx2.1 (17–19, 25). Western blotting showed that p53 and p21^WAF1/CIP1 were diversely expressed in CRC cell lines (P < 0.001 for Brown-Forsythe and Welch’s ANOVA with Dunnett’s post-hoc test), while Hela cell (66) served as a negative control ( Figure 6A ). Strikingly, HCT116 cells showed a moderately lower level of p53, whereas a relatively higher level of p21^WAF1/CIP1 was observed in HCT116 cells compared with that in SW480 cells (P < 0.01 for Mann-Whitney U test; Figure 6A ).

Lenti-Nkx2.5-HCT116 cells possessed the same level of p53, while a higher level of p21^WAF1/CIP1 than that in Lenti-NC-HCT116 cells (P < 0.01 for Mann-Whitney U test; Figure 6B ). Lenti-NC-SW480 and Lenti-Nkx2.5-SW480 cells had the same levels of p53 and p21^WAF1/CIP1 ( Figure 6B ). This result suggests that Nkx2.5 increased p21^WAF1/CIP1 expression in HCT116 but not in SW480. The p53 level in Lenti-NC-HCT116 and Lenti-Nkx2.5-HCT116 cells was remarkably lower than that in Lenti-NC-SW480 and Lenti-Nkx2.5-SW480 cells, while p21^WAF1/CIP1 level in Lenti-NC-HCT116 and Lenti-Nkx2.5-HCT116 cells was significantly higher than that in Lenti-NC-SW480 and Lenti-Nkx2.5-SW480 cells (P < 0.01 for Mann-Whitney U test; Figure 6B ). This result suggests that p53 loses its transcriptional activity for p21^WAF1/CIP1 in SW480 but not in HCT116, even though SW480 cells possess high level of p53.

Put together, these results indicate that Nkx2.5 overexpression can increase p21^WAF1/CIP1 expression without influencing p53 level in HCT116 cells, while p53 seems to lose its function in SW480 cells.

Because of the abnormal protein level of p53/p21^WAF1/CIP1 in SW480 and SW620 cells, the mutational status of TP53 in CRC cell lines was assessed as in previous studies (57, 67) and the CCLE database (https://xenabrowser.net/heatmap/) ( Figure 7A ). DNA sequencing analysis was also conducted to confirm the mutational status of TP53 in CRC cell lines ( Figure 8 ).

To investigate the relationship between the levels of Nkx2.5, p53, and p21^WAF1/CIP1, eight CRC cell lines were assigned to two groups (wild-type p53 versus mutated p53; Figure 7A ). No significant differences in the levels of Nkx2.5, p53, and p21^WAF1/CIP1 were observed between these two groups. Since either mutational (38) or transcriptional (40) inactivation of TP53 could lead to functional disruption of p53, the CRC cell lines were allocated to two groups in another way (high expression of wild-type p53 versus mutated or low expression of wild-type p53, median level was chosen as cutoff value; Figure 7B ). A significantly higher expression of p21^WAF1/CIP1 was observed in the former group (high expression of wild-type p53: HCT116 and RKO cells). Since mutated p53 was highly expressed in high malignancies and reported as an oncogene gaining new activities (68, 69), we assigned the CRC cell lines to two groups (wild-type or low expression of mutated p53 versus high expression of mutated p53, median level was chosen as cutoff value; Figure 7C ). A markedly higher expression of Nkx2.5 was noted in the latter group (high expression of mutated p53: DLD-1, SW480, and SW620 cells).

Co-IP assay showed that Nkx2.5 could interact and co-precipitate with p53 in HCT116 cells ( Figure 9A ). The luciferase activity assay showed an additive effect on the activity of the p21^WAF1/CIP1 promoter by Nkx2.5 plasmid ( Figure 9B ). The results highlight that Nkx2.5 can increase p21^WAF1/CIP1 expression by interacting with p53 as a coactivator.