Human colon cancer cells (Hct116 and LoVo) were purchased from the American Type Culture Collection. The medium used was RPMI-1640 (Gibco; Thermo Fisher Scientific, Inc.) containing 10% fetal bovine serum, streptomycin and penicillin (Gibco; Thermo Fisher Scientific, Inc.) (complete medium). Cells were cultured in a constant-temperature incubator containing 5% carbon dioxide at 37°C.

When the cell confluency reached ~80%, 450 µM of CoCl₂ was added to the flask. After treatment for 48 h (Hct116) and 36 h (LoVo), most regular-sized diploid cancer cells died and only a few cells with large nuclei (PGCCs) survived. The surviving cells exhibited multinucleated and mononuclear giant cell morphology and were highly resistant to hypoxia. Following treatment, the remaining cells were cultured in a complete medium without CoCl₂. After ~15 days of recovery, the PGCCs produced daughter cells via asymmetric division. Following three repeated treatments, 30% of PGCCs and 70% of daughter cells appeared in the flask and the cells were collected for subsequent experiments.

Proteins were extracted by RIPA Buffer according to the manufacturer's instructions (Thermo Fisher Scientific, Inc.). Protein concentration was determined using Nanodrop (Thermo Fisher Scientific, Inc.). Protein samples from control cells and PDCs were separated by 10% SDS-PAGE gel for electrophoresis at a constant voltage of 80 V. After separating the protein bands, the voltage was adjusted to 120 V. After the membrane transfer, 5% skimmed milk was added to the membrane for 1 h at room temperature. According to the molecular weight of proteins, PVDF membrane was cut prior to hybridization with different primary antibodies (detailed information regarding the antibodies is provided in Table I) at 4°C overnight. For the target proteins with similar molecular weight, a membrane regeneration solution was used to elute. After washing, the corresponding secondary antibodies were added and the mixture was shaken at room temperature for 1 h. After the ECL developer (Shanghai Yeasen Biotechnology Co., Ltd.) was added, a ChemiDoc imaging system (Bio-Rad Laboratories, Inc.) was used for development and observation. ImageJ software (National Institutes of Health; 1.54D) was used to analyze and calculate the gray value of the corresponding strip and the expression index of the target protein. The experiment was independently repeated thrice.

The primers were designed by Primer 5 (http://www.premierbiosoft.com/primerdesign/). A total of 2.5×10⁶ cells were collected and total RNA was extracted using an RNA extraction kit (cat. no. 9190; Takara Biotechnology) according to the manufacturer's protocols. Transcription levels of MITF and HIF1α were detected by qPCR. The PCR conditions were set according to the instructions provided in the SYBR Green Kit (Shanghai Yeasen Biotechnology Co., Ltd.). The amplification was performed for 40 cycles (95°C 2 min, 95°C 10 sec and 60°C 30 sec). The relative amount of each mRNA level was normalized to that of the β-actin level and the difference in mRNA level was calculated using the 2^−ΔΔCq method (18). Detailed information on the primers used is provided in Table II. All experiments were repeated at least three times.

After an appropriate amount of cell precipitation from Hct116 and LoVo control cells and PDCs, 200 µl cytoplasmic protein extraction reagent A was added to the nuclear protein and cytoplasmic protein extraction kit (cat. no. P0027, Beyotime Institute of Biotechnology) and was placed on ice for lysis for 15 min. Then, 10 µl cytoplasmic protein extraction reagent B was added, placed on ice for cracking for 1 min, vortex-oscillated for 5 sec and then centrifuged at 4°C using centrifuge at 16.2 × g for 15 min. The supernatant comprised the cytoplasmic protein solution, which was then transferred to a pre-cooled Eppendorf tube. The remaining precipitate was rinsed with 500 µl phosphate-buffered saline (PBS) thrice and centrifuged at 0.6 × g at 4°C for 5 min. Subsequently, 50 µl of nuclear protein extraction reagent was added and the precipitation was dissolved on ice for 30 min. After vortex-oscillation every 1 min for 30 sec and centrifugation at 16.2 × g at 4°C for 15 min, the resultant supernatant comprised the nuclear protein solution. The two parts of the supernatant were mixed with 1/4 volume of 5× protein loading buffer and then the protein was denatured at 100°C for 10 min and stored in the refrigerator at −20°C for subsequent experiments.

Hct116 and LoVo control cells and PDCs were inoculated onto a cover slide. When the cell density reached ~70%, the cells were fixed with methanol at room temperature for 30 min. Hydrophobic circles were drawn with a neutral oil pen. A peroxidase-blocking agent was used to treat the cells in the dark for 15 min and then goat serum-containing working solution was added, followed by incubation at room temperature for 20 min. The corresponding primary antibodies (Table I) were added and incubated at 4°C overnight. The next day, one-to-two drops of biotin-labeled goat anti-rat/rabbit IgG polymer were added, followed by incubation at room temperature for 30 min. A DAB color-developing solution was prepared to observe brown particles under a microscope at room temperature for 1–2 min. The color development reaction was stopped after brown staining. Hematoxylin was used to stain at room temperature for 30 sec, followed by alcohol gradient-mediated dehydration, the addition of dimethylbenzene and final mounting with neutral gum.

The cell samples were diluted to obtain samples with 50, 100 and 150 cells/ml and cultured in 24-well plates with three repeated pores in each group for 2 weeks. Cells were fixed with anhydrous methanol for 30 min and stained with 0.1% crystal violet for 30 min at room temperature. Cell colonies were counted at ×100 magnification (a single colony was defined as that containing >50 cells). The cell colony formation efficiency was assessed using the following formula: formation efficiency=number of clones/number of inoculated cells.

A wound healing assay was used to detect cell migration. Cells (1×10⁵) in the logarithmic growth stage were cultured in a 6-well plate and three repeat pores were set. Single-layer cells were scratched uniformly using sterile pipette tips to create wounds. PBS was used to wash away the detached cells. The cells were then incubated in serum-free RPMI 1640. ImageJ software (National Institutes of Health; 1.54D) was used to outline the migration area and calculate the wound-healing index according to the following formula: [(the wound area at 0 h)-(the wound area at the indicated time)]/(the wound area at 0 h). A high score indicated stronger migration ability.

For the Transwell migration assay, 200 µl of serum-free cell suspension containing 1×10⁵ cells was added to the upper chamber of the Transwell chamber and 600 µl of medium containing 20% serum was added to the lower chamber, which was cultured in an incubator for 24 h. The cells were fixed with methanol for 30 min and stained with 0.1% crystal violet for 30 min. The cells were observed under an inverted microscope and three fields (magnification, ×100) were randomly selected, images captured and cells counted. A total of three duplicate wells were set for each group of cells. The procedure for the Transwell invasion experiment was the same as that for the Transwell migration experiment. The difference was that the Transwell invasion experiment required a 200 µl sample containing 5×10⁵ cells and the invasion chamber contained Matrigel (cat. no. 354480; Corning, Inc.). After the cells were added into the chamber with Matrigel, the plates were incubated for 12 h at 37°C.

Co-IP was used to determine the interactions of SUMO1, SUMO2 and HIF1α in Hct116 and LoVo control cells and PGCs according to the manufacturer's protocols. When the cell density of the T25 flask reached 80%, the cells were collected in EP tubes. The cells were lysed using 500 µl IP lysis buffer (Thermo Fisher Scientific, Inc.) containing a halt protease and phosphatase inhibitor cocktail (1:100 dilution) for 30 min on ice and then centrifuged at 16.2 × g at 4°C for 10 min. The supernatant was transferred to an EP tube containing A/G agarose homogenate of agar glycoprotein beads and shaken at 4°C for 30 min. After incubation, the supernatant was divided into three parts: one part was used to detect the total protein level (input). Primary antibodies corresponding to rabbit IgG and the target protein were added to the other two tubes, respectively and maintained at 4°C overnight. The next day, A/G agarose homogenates of the agar glycoprotein beads were adsorbed by a magnetic grate and washed by lysis buffer (Thermo Fisher Scientific, Inc.). The samples containing rabbit IgG and primary antibodies (Table I) of the target protein were transferred to the newly washed column and incubated at 4°C for 2 h. After incubation, the supernatant was discarded and washed five times with 500 µl IP-specific cracking buffer. Finally, western blotting was performed to analyze the samples.

CoCl₂-treated cells were seeded in six-well plates until they reached 80% confluence. Approximately 20 µM of GA (15:1, MedChem Express, USA) was added to control cells and PDCs for 24 h, and the samples were evaluated using western blotting analysis and other assays described.

Methyl linoleate (ML), the main active ingredient of Sageretia thea, is a major anti-melanin-producing compound that downregulates MITF expression. To assess cell viability before and after ML treatment, Hct116 and LoVo PDCs were seeded at a density of 5,000 cells per well into 96-well plates and incubated at 37°C for 12 h. The cells were divided into five groups and each group was independently analyzed in triplicate. The cells were treated with ML at concentrations of 40, 80, 160 and 320 µM for 12, 24, 48 and 72 h. After incubation, 10 µl of CCK8 (Dojindo Laboratories, Inc.) reagent was added to each well and incubated at 37°C for 12 h. After adding the CCK8 reagent, the wells were analyzed using a Bio-Rad microplate reader at a wavelength of 450 nm (Bio-Rad Laboratories, Inc.). Optical density data are presented as the means ± standard error of the mean.

The BLOCK-iT RNAi Designer (https://rnaidesigner.thermofisher.com/rnaiexpress/) was used to design the siRNA of MITF. A total of three different siRNA sequences targeting MITF, PIAS4 and negative control siRNA oligonucleotides were obtained from Shanghai GenePharma Co., Ltd. The K391R, K477R and empty vector was purchased from Genewiz, Inc. The cells were inoculated into a 6-well plate and transfected when the cell confluence reached ~50% at 37°C. According to the manufacturer's experimental protocol, 5 µl of transfect-Mate (Shanghai GenePharma Co., Ltd.) and 5 µl of interfering sequence or plasmids were added to 100 µl of Opti-MEM (Gibco; Thermo Fisher Scientific, Inc.) to formulate the transfection complexes at room temperature. After 48 h of transfection, cell samples were collected to detect the targeted proteins using western blotting. Detailed information on the siRNA oligonucleotide sequences is provided in Tables III and IV.

All data and statistical charts were processed using GraphPad Prism 8.0.2 (Dotmatics) or SPSS Statistics 25 (IBM Corp.) software. Statistical significance was assessed by comparing mean values using the Student's t-test for independent groups. An ANOVA followed by Bonferroni correction was used among different groups. P<0.05 was considered to indicate a statistically significant difference.

Hct116 and LoVo cells were cultured in a complete medium and their morphologies are shown in Fig. 1A a and c. Control cells were epithelioid and oval in shape, with uniform cell distribution and size. When 450 µM of CoCl₂ was added to the flask, most of the regular-sized diploid cancer cells died and only a few cells with large nuclei (PGCCs) survived. After ~15 days of recovery, PGCCs produced daughter cells through asymmetric division (Fig. 1A b and d; Control represents cells without CoCl₂ treatment and treatment represents cells with CoCl₂ treatment).

The results of the plate cloning assay demonstrated that Hct116 and LoVo PDCs had greater proliferative ability than the control cells (Fig. 1B a and b) and the differences were statistically significant (Fig. 1Bc). The wound healing assay showed a significantly higher migration ability of LoVo and HCT116 PDC compared with that of control cells (Fig. 1C a and b) and the differences were statistically significant (Fig. 1Cc), indicating that the migration ability of PGCCs and their progeny cells was stronger than that of the control group. Additionally, Transwell migration and invasion experiments showed that PDC had stronger migration and invasion abilities than the control cells (Fig. 1D).

In the present study, western blotting and ICC were used to detect the expression and subcellular location of HIF1α in Hct116 and LoVo control cells and PDCs. The expression of HIF1α was higher in PDCs than in the control cells (Fig. 1E). In the control cells, HIF1α was detected only in the cytoplasm (Fig. 1F). In PDCs, HIF1α was detected in both the cytoplasm (Fig. 1F) and the nucleus (Fig. 1G). Total, cytoplasmic and nuclear HIF1α expression levels were significantly higher in PDCs than in control cells (Fig. 1H). In addition, immunocytochemical staining demonstrated the expression and subcellular localization of HIF1α in control cells and PDCs (Fig. 1I).

Co-IP was used to detect the interactions between SUMO1, SUMO2/3 and HIF1α. The total cell lysates of control and PDCs were immunoprecipitated with an anti-HIF1α antibody (Fig. 2A) and then immunoblotted with anti-SUMO1 and anti-SUMO2/3 antibodies. The results showed that HIF1α could bind to SUMO1 and SUMO2/3 in PDCs (Fig. 2B and C).

GA can inhibit the SUMOylation of important proteins that are critical during the development and progression of malignant tumors (19). GA directly binds to the SUMO E1 activating enzyme to inhibit the formation of the E1-SUMO thioester complex (20). After GA treatment, the total, cytosolic and nuclear fractions were collected to detect the expression of HIF1α. In the cytosolic fraction, the expression of HIF1α in PDCs was slightly inhibited after GA treatment compared with that in PDCs without GA treatment (Fig. 2D b and e). The total and nuclear localization of HIF1α was inhibited by GA treatment and the differences before and after GA treatment in PDCs were statistically significant (Fig. 2D a, c, d and f), indicating that SUMOylation might play an important role in the nuclear localization of HIF1α. Additionally, ICC staining was performed on the cells before and after GA treatment. Staining results showed that GA treatment inhibited the nuclear expression of HIF1α (Fig. 2E).

Functional cell experiments were performed to assess the effect of GA on PDC migration, invasion and proliferation before and after GA treatment. Cloning experiments showed that the number of PDCs decreased after GA treatment (Fig. 2F). The number of colonies of 50, 100 and 150 GA-treated PDCs was reduced compared with that of untreated Hct116 and LoVo PDCs (Fig. 2F and Ia). Wound healing experiments showed that the scratched areas of PDCs before GA treatment were significantly narrower than those after GA treatment (Fig. 2G and Ib). The results of the Transwell assay showed that the migratory and invasive abilities of PDCs treated with GA were inhibited compared with those of PDCs without GA treatment (Fig. 2H and I c and d).

The expression of MITF, PIAS4 and VHL is an important indicator influencing the expression and subcellular location of HIF1α. Total, cytosolic and nuclear fractions were collected to detect the expression of MITF, PIAS4 and VHL in control cells and PDCs. The total protein levels of MITF, PIAS4 and VHL were elevated in Hct116 and LoVo PDCs compared with those in control cells (Fig. 3A and D). In the cytosolic fraction, the expression of MITF, PIAS4 and VHL in LoVo PDCs was upregulated (Fig. 3B and D). After CoCl₂ treatment, the expression levels of MITF, PIAS4 and VHL in the nucleus were also higher than those in control cells (Fig. 3C and D). For further verification, ICC staining was performed and the results showed both nuclear and cytoplasmic expression of MITF, PIAS4 and VHL. The staining intensity of MITF, PIAS4 and VHL in PDCs was stronger than that of control cells (Fig. 3E).

MITF can bind to the HIF1α promoter and stimulate its transcriptional activity (21). To confirm whether MITF interacts with HIF1α, Co-IP was performed. When HIF1α was used as a bait protein and incubated with MITF, MITF bands appeared in the input and IP groups, indicating that HIF1α interacted with MITF (Fig. 4A). Additionally, MITF was knocked down using siRNAs. Following MITF knockdown, the mRNA expression levels of MITF and HIF1α were detected by qPCR. As shown in Fig. 4B, the mRNA expression levels of MITF and HIF1α were significantly decreased in PDCs.

ML, the main active ingredient in S. thea, is a major anti melanin-producing compound that downregulates MITF expression. CCK8 was used to screen the appropriate concentration of ML and 40, 80 and 160 µM were finally selected (Fig. 4C). After ML treatment, the western blotting results showed that the total, cytosolic and nuclear expression of HIF1α, SUMO1 and SUMO2/3 was inhibited to varying degrees. Nuclear expression of HIF1α in PDCs was inhibited, suggesting that MITF may play an important role in the nuclear localization of HIF1α. Compared to the control cells, the expression levels of nuclear SUMO1 and SUMO2/3 also decreased after ML treatment, indicating that MITF can regulate SUMOylation of HIF1α and further affect the nuclear location of HIF1α (Fig. 4Da-c). Following ML treatment, the western blotting results showed that the expression of VHL was inhibited (Fig. 4Dd). Functional experiments were also performed on the PDCs before and after ML treatment. The proliferation, migration and invasion abilities of PDCs were inhibited after ML treatment, as demonstrated by plate cloning (Fig. 4E and Ga), wound healing (Fig. 4F and Gb) and Transwell experiments (Fig. 4H and I).

PIAS4 belongs to the PIAS protein family, which are protein inhibitors that activate STAT proteins. PIAS4 is often involved in PTM as it acts as a SUMO E3 ligase. VHL is an E3 ubiquitin ligase. PIAS4 mediates the SUMOylation of VHL and reduces the activity of its ubiquitin E3 ligase, contributing to the stabilization of HIF1α (22). The present study demonstrated no significant change in the protein expression levels of VHL, HIF1α and MITF following the knockdown of PIAS4 using siRNA (Fig. 5A and Da-c and Fig. S1). In addition, no significant differences were observed in the expression of cytoplasmic and nuclear VHL and HIF1α after PIAS4 knockdown (Figs. 5B, C and Dd-i and S1).

The proteins that can undergo SUMOylation have a known consensus motif: ΨKXE (Ψ: a hydrophobic amino acid, K: lysine residue, X: a variable residue amino acid, E or D: a glutamic acid) (23). Two lysine sites, K391 and K477, were revealed by the GSP-SUMO1.0 (The Cuckoo Workgroup) computer analysis software and the amino acid sequences of the two sites were 390–393 (LKKE) and 476–479 (LKLE), which met the characteristics of SUMOylation (Table V).

To determine whether HIF1α is SUMOylated at K391 and K477 sites, the plasmid with mutated lysine to arginine was transiently transfected. Compared with the empty plasmid and blank control groups, the expression level of HIF1α was higher in the HIF1α overexpression and the mutant groups. The expression level of HIF1α in the HIF1α overexpression group was lower than that in the mutant group, which indicated that HIF1α was not degraded in the mutation group because lysine was mutated to arginine and ubiquitination could not bind the lysine sites (Fig. 5Ea). The expression level of cytoplasmic HIF1α in different groups was consistent with that of total protein (Fig. 5Eb). However, the expression level of nuclear HIF1α in the K391R+K477R group decreased, indicating that HIF1α with mutated arginine sites cannot be modified by SUMOylation or enter the nucleus (Fig. 5Ec). The expression levels of HIF1α with K391R, K477R, K391R and K477R double mutants in ML-treated cells are shown in Fig. 5Ed.

To further investigate the effects of mutation of HIF1α at sites K391 and K477 on cell proliferation, migration and invasion in PDCs, cell functional experiments, including plate cloning, wound healing, Transwell migration and invasion assays, were performed. The results showed that the proliferation (Fig. 6A and Da), migration (Fig. 6B and Db) and invasion (Fig. 6C and E) abilities of Hct116 and LoVo PDCs after mutation were significantly lower than those in the non-mutation group.