AMPK phosphorylates PPARδ to mediate its stabilization, inhibit glucose and glutamine uptake and colon tumor growth

Abstract

Peroxisome proliferator–activated receptor δ (PPARδ) is a nuclear receptor transcription factor that plays an important role in the regulation of metabolism, inflammation, and cancer. In addition, the nutrient-sensing kinase 5′AMP-activated protein kinase (AMPK) is a critical regulator of cellular energy in coordination with PPARδ. However, the molecular mechanism of the AMPK/PPARδ pathway on cancer progression is still unclear. Here, we found that activated AMPK induced PPARδ-S50 phosphorylation in cancer cells, whereas the PPARδ/S50A (nonphosphorylation mimic) mutant reversed this event. Further analysis showed that the PPARδ/S50E (phosphorylation mimic) but not the PPARδ/S50A mutant increased PPARδ protein stability, which led to reduced p62/SQSTM1-mediated degradation of misfolded PPARδ. Furthermore, PPARδ-S50 phosphorylation decreased PPARδ transcription activity and alleviated PPARδ-mediated uptake of glucose and glutamine in cancer cells. Soft agar and xenograft tumor model analysis showed that the PPARδ/S50E mutant but not the PPARδ/S50A mutant inhibited colon cancer cell proliferation and tumor growth, which was associated with inhibition of Glut1 and SLC1A5 transporter protein expression. These findings reveal a new mechanism of AMPK-induced PPARδ-S50 phosphorylation, accumulation of misfolded PPARδ protein, and inhibition of PPARδ transcription activity contributing to the suppression of colon tumor formation.

5′AMP-activated protein kinase (AMPK) is a serine/threonine kinase, which consists of AMPKα (catalytic subunits) and AMPKβ/AMPKγ (regulatory subunits) (1). Activation of AMPK regulates cell growth, metabolism, autophagy, and cancer progression (2, 3). Under intracellular ATP depletion, AMPK is activated and accelerates catabolism, resulting in ATP production. In addition, AMPK can be activated by the upstream regulatory protein LKB1 in the liver (4). Some pharmacologic reagents such as metformin and 5-aminoimidazole-4-carboxamide-1-β-d-ribofuranoside (AICAR) can activate AMPK (2, 5, 6). Activation of AMPK triggers the downstream signaling pathways by inducing the targeted proteins FOXO3a or G-alpha-interacting vesicle-associated protein (GIV) phosphorylation leading to autophagy induction (7) and tumor suppression (8). Furthermore, the interaction of AMPK with peroxisome proliferator–activated receptor δ (PPARδ) regulates glucose metabolism, vascular–endothelial dysfunction, insulin resistance, and inflammation (9, 10, 11, 12). AMPK interacts with PPARδ leading to increased endurance or training adaptation in mice (9). Omentin-1 inhibits vascular–endothelial dysfunction in response to high glucose, which is associated with inhibition of the AMPK/PPARδ signaling pathway (10). METRNL alleviates insulin resistance and inflammation in response to lipid or LPS (11, 12), which is involved in activation of AMPK and PPARδ pathways. In addition, PPARδ agonist GW501516-induced glucose uptake of human skeletal muscle cells is AMPK dependent rather than PPARδ (13). As one of the members of PPARs family, PPARδ was the first identified in humans (14), which is highly expressed in colonic epithelial cells and regulates colonic cancer progression (15, 16, 17, 18, 19, 20). Ligand-binding and activated PPARδ is involved in the chronic inflammation (ulcerative colitis and Crohn's disease) and colitis-associated colorectal cancer (CRC) (15, 16, 17, 18, 19), suggesting that PPARδ activation enhances CRC. Metformin inhibits PPARδ agonist–mediated colorectal tumorigenesis in azoxymethane/dextran sulfate sodium-induced tumor model (21), whereas the molecular mechanism of this pathway on tumor progression is unclear. Here, we identified that AMPK induced PPARδ-S50 phosphorylation leading to inhibition of its transcription activity and PPARδ-mediated tumor growth.

Results

AMPK induces PPARδ phosphorylation

AMPK can be activated by the agonist metformin (22). SW480 cells treated with metformin induced AMPKα phosphorylation and enhanced the interaction of PPARδ with AMPKα by immunoprecipitation analysis (Fig. 1A). Confocal analysis showed that AMPKα colocalized with PPARδ (Fig. 1B). Further analysis showed that SW480 cells treated with metformin increased PPARδ serine phosphorylation (Fig. 1C). AMPK can induce the substrate phosphorylation with the typical LxRxxSxxxL motif (23); alignment analysis by using https://www.ebi.ac.uk/Tools/msa/clustalo/ shows that PPARδ contains this typical motif (Fig. 1D). Glucose starvation can activate AMPK (7). As shown in Figure 1E, the 816.388 m/z at b8 ion fragment was from the parent ion neutral loss of one phosphoric acid (97.972 m/z), suggesting that PPARδ-S50 was phosphorylated in response to glucose starvation. To further analyze AMPK-induced PPARδ-S50 phosphorylation, SW480 cells were transfected with PPARδ or PPARδ/S50A mutant and subjected to glucose starvation. As expected, AMPK did not induce PPARδ/S50A mutant phosphorylation (Fig. 1F). AMPK-induced PPARδ-S50 phosphorylation was further demonstrated by in vitro analysis (Fig. 1G). Moreover, AMPKα shRNA-silenced cells inhibited PPARδ-S50 phosphorylation in response to metformin (Fig. 1H). These findings suggest that AMPK induced PPARδ-S50 phosphorylation. Interestingly, activated AMPK induced PPARδ-S50 phosphorylation and increased PPARδ protein levels in response to metformin (Fig. 2A), AICAR, and glucose starvation (Fig. 2, B and C). These findings suggest that AMPK induced PPARδ phosphorylation resulting in accumulation of its protein levels.

Figure 1.

AMPK induces PPARδ phosphorylation.A, SW480 cells were treated with metformin (10 mM) as indicated time. Cell lysates were subjected to immunoprecipitation and Western blot. Input: cell lysates from no metformin treatment. B, SW480 cells were treated with metformin (10 mM) for 3 h. Cells were subjected to confocal analysis. The scar bar represents 20 μm. C, SW480 cells were treated with metformin (10 mM) as indicated time. Cell lysates were subjected to immunoprecipitation and Western blot. D, alignment of the consensus of the PPARδ phosphorylation site with AMPK substrates. E, SW480 cells were treated with or without glucose starvation for 3 h. Cell lysates were subjected to immunoprecipitation. The purified PPARδ protein was identified by LC/MS/MS. F, SW480 cells were transfected with Flag-PPARδ or mutant plasmids for 36 h. After that, cells underwent glucose starvation for 3 h. Cell lysates were subjected to immunoprecipitation and Western blot. G, in vitro kinase assay was performed as described in Experimental procedures. H, SW480 cells were transfected with control shRNA or AMPKα shRNA for 36 h. Cells were treated without or with metformin (10 mM) for 3 h. Cell lysates were subjected to Western blot. AMPK, 5′AMP-activated protein kinase; PPARδ, peroxisome proliferator–activated receptor δ.

Figure 2.

AMPK induces PPARδ-S50 phosphorylation, resulting in accumulation of PPARδ protein levels.A, cells were treated with metformin (10 mM) at the indicated time. Cell lysates were subjected to Western blot. The relative PPARδ protein levels were quantified. Results are expressed as the means ± SEM (n = 3). ∗p < 0.05. B, SW480 cells were treated with 5 mM AICAR at the indicated time. Cell lysates were subjected to Western blot. C, Western blot analysis of SW480 cell lysates in response to glucose starvation. AICAR, 5-aminoimidazole-4-carboxamide-1-β-d-ribofuranoside; AMPK, 5′AMP-activated protein kinase; PPARδ, peroxisome proliferator–activated receptor δ.

p62/SQSTM1 induces misfolded PPARδ degradation

Our data showed that AMPK induced PPARδ-S50 phosphorylation, leading to accumulation of PPARδ protein levels, whereas the mechanism is unclear. Further analysis showed that cells treated with the autophagy inhibitor (chloroquine) led to increased PPARδ protein levels (Fig. 3A), suggesting that PPARδ undergoes autophagic degradation. Moreover, metformin-treated cells increased PPARδ half-life (Fig. 3B). Consistently, the half-life of PPARδ/S50E (mimic phosphorylation) was longer than that of PPARδ/S50A (mimic nonphosphorylation) mutant (Fig. 3, C and D), suggesting that AMPK induced PPARδ phosphorylation, resulting in inhibition of PPARδ degradation. Western blot analysis showed that AMPK induced the insoluble PPARδ phosphorylation and inhibited its degradation (Fig. 3E). p62/SQSTM1 was the first identified autophagy receptor, which can induce the misfolded protein autophagic degradation (24, 25). To address whether p62 could mediate PPARδ autophagic degradation, SW480 cells were transfected with p62 shRNA. Half-life analysis showed that silence of p62 increased PPARδ half-life (Fig. 4A). Immunoprecipitation analysis showed that p62 bound to PPARδ (Fig. 4B). However, cells treated with metformin inhibited this event (Fig. 4C). In addition, compared with PPARδ, PPARδ/S50A mutant significantly enhanced the binding of PPARδ to p62, whereas PPARδ/S50E inhibited this event (Fig. 4D). These findings suggest that AMPK induced PPARδ-S50 phosphorylation, leading to inhibition of p62-mediated misfolded PPARδ degradation.

Figure 3.

PPARδ-S50 phosphorylation increases PPARδ protein stability.A, SW480 or HCT-116 cells were treated with chloroquine (30 μM) for 4 h. Cell lysates were subjected to Western blot. The relative PPARδ protein levels were quantified. Results are expressed as the means ± SEM (n = 3). ∗p < 0.05. B, SW480 cells were treated with cycloheximide (30 μg/ml) together with or without metformin (10 mM) at the indicated time. Cell lysates were subjected to Western blot. The relative PPARδ protein remaining at each time point was calculated. Results are expressed as the means ± SEM (n = 3). ∗p < 0.01. C and D, SW480 cells were transfected with Flag-PPARδ or mutant plasmids for 36 h. Cells were treated with cycloheximide (30 μg/ml) for 0, 3, and 6 h to inhibit de novo protein synthesis. Cell lysates were subjected to Western blot. The relative PPARδ protein remaining at each time point was calculated. Results are expressed as the means ± SEM (n = 3). ∗p < 0.01. E, SW480 cells were treated with metformin (10 mM) at the indicated time. Soluble and insoluble proteins were extracted as described in Experimental procedures. Extracts were subjected to Western blot. The relative PPARδ protein level at each time point was calculated. Results are expressed as the means ± SEM (n = 3). ∗p < 0.01. PPARδ, peroxisome proliferator–activated receptor δ.

Figure 4.

PPARδ-S50 phosphorylation inhibits p62-mediated misfolded PPARδ degradation.A, the half-life of PPARδ was assayed in p62 shRNA-silenced SW480 cells. The relative PPARδ protein remaining at each time point was calculated. Results are expressed as the means ± SEM (n = 3). ∗p < 0.01. B, cell lysates of SW480 cells were subjected to immunoprecipitation and Western blot. C, SW480 cells were transfected with the vector or PEBG-PPARδ for 36 h. Cells were treated with metformin (10 mM) for 1 h as indicated. Cell lysates were subjected to immunoprecipitation and Western blot. D, SW480 cells were transfected with Flag-PPARδ or mutant plasmids as indicated for 36 h. Cell lysates were subjected to immunoprecipitation and Western blot. PPARδ, peroxisome proliferator–activated receptor δ.

AMPK/PPARδ pathway inhibits cancer cell metabolism

As one of the PPARs family members, PPARδ promotes tumor growth (17, 26, 27, 28, 29), which is associated with an increase in tumor metabolism (29). To further address whether PPARδ-S50 phosphorylation could affect its transcription activity, the dual-luciferase analysis showed that S50A mutant significantly increased PPARδ activity compared with the WT (PPARδ) (Fig. 5A), and PPARδ agonist GW501516 enhanced this event. However, the S50E mutant reduced PPARδ activity and GW501515 had no effect on PPARδ/S50E activity (Fig. 5A). Immunoprecipitation analysis showed that PPARδ/S50A increased the binding of PPARδ to RXRα (a transcriptional partner of PPARδ) (Fig. 5B). These findings suggest that PPARδ-S50 phosphorylation disrupted the complex of PPARδ/RXRα, leading to inhibition of PPARδ transcription activity. PPARδ can promote the uptake of glucose and glutamine by increasing the expressions of Glut1 and SLC1A5 (29). As expected, compared with PPARδ, PPARδ/S50E mutant significantly inhibited Glut1 and SLC1A5 gene transcription (Fig. 5, C and D), which was in agreement with reduced Glut1 and SLC1A5 protein levels (Fig. 5E). Consistently, PPARδ/S50E significantly inhibited glucose consumption, lactate release, and glutamine uptake (Fig. 5, F–H). These findings suggest that PPARδ-S50 phosphorylation inhibited its transcription activity.

Figure 5.

AMPK induces PPARδ phosphorylation leading to inhibition of PPARδ-mediated metabolic pathway.A, SW480 cells were transfected with PPRE₃-lu reporter plasmids together with PPARδ or mutant plasmids for 36 h. After that, cells were treated with control (DMSO) or GW501516 (10 μM) as indicated for another 12 h. Cell lysates were subjected to dual-luciferase assay. Results are expressed as the means ± SEM (n = 3). ∗p < 0.05. B, SW480 cells were transfected with plasmids as indicated for 48 h. Immunoprecipitation and Western blot analysis were performed by using cell lysates. C and D, SW480 cells were transfected with Glut1-lu or SLC1A5-lu reporter plasmids together with PPARδ or mutant plasmids for 36 h. Cell lysates were subjected to the dual-luciferase assay. Results are expressed as the means ± SEM (n = 3). ∗p < 0.05 and ^#p < 0.01. E, SW480 cells were transfected with PPARδ or mutant plasmids for 36 h. Cell lysates were subjected to Western blot. F–H, SW480 cells were transfected with PPARδ or mutant plasmids for 36 h. Relative glucose consumption, lactate release, and L-glutamine uptake were assayed. Results are expressed as the means ± SEM (n = 5). ∗p < 0.05 and ^#p < 0.01. PPARδ, peroxisome proliferator–activated receptor δ.

AMPK/PPARδ inhibits tumor growth

To further address whether PPARδ-S50 phosphorylation could inhibit cancer cell proliferation and tumor growth, the soft agar analysis and xenograft mouse model were performed. Compared with PPARδ, PPARδ/S50A markedly enhanced cancer cell proliferation and tumor growth, whereas PPARδ/S50E mutant inhibited this event (Fig. 6, A–C), which was associated with the inhibition of Glut1 and SLC1A5 expressions in S50E-expressed tumors (Fig. 6D). These findings suggest that AMPK induced PPARδ-S50 phosphorylation leading to inhibition of PPARδ-mediated tumor growth.

Figure 6.

AMPK/PPARδ pathway inhibits tumor growth.A, colony formation analysis was performed by using soft agar in stable PPARδ or mutant-expressing SW480 cells. PPARδ or mutant expressions were determined by Western blot. Results are expressed as the means ± SEM, n = 3. ∗p < 0.01 and ^#p < 0.05. B and C, stable PPARδ or mutant-expressing SW480 cells were injected subcutaneously in nude mice for 4 weeks, and the tumor weight and volume were measured. Results are expressed as the means ± SEM, n = 5. ∗p < 0.01 and ^#p < 0.05. D, tumor lysates as indicated were subjected to Western blot. AMPK, 5′AMP-activated protein kinase; PPARδ, peroxisome proliferator–activated receptor δ.

Discussion

PPARs contain PPARα, PPARγ, and PPARδ (30, 31, 32, 33, 34). PPARδ can be activated by the fatty acids or fatty-acid derivatives, which is involved in the mucosal inflammation and malignant transformation (26). Aberrant PPARδ expression or activation can cause metastatic progression and carcinogenesis of CRC (15, 16, 26, 29, 35). Consistently, PPARδ KO reduces colon inflammation and colitis-associated tumor growth (17). Conversely, clinical observation shows that higher expression of PPARδ increases survival in patients with CRC (36, 37). Overexpression of PPARδ or activation by agonists decreases CRC cell proliferation and colorectal tumorigenesis (37, 38, 39, 40). In contrast, silence or KO of PPARδ promotes colon cancer cell proliferation and tumor growth (41, 42, 43). These discrepancy findings could be derived from different models and ligand specificity. AMPK is the metabolic sensor that regulates cellular and organismal metabolism in eukaryotes, which is activated in response to intracellular ATP depletion (2, 22). AMPK plays a critical role in regulating cell growth and metabolism (2). However, the effect of AMPK on PPARδ-mediated cancer cell metabolism and tumor growth is unclear. AMPK induces GIV phosphorylation, leading to stabilization of epithelial tight junctions and inhibition of tumorigenesis (8), suggesting that AMPK can act as a tumor suppressor. Although the AMPK/PPARδ pathway can regulate glucose and lipid metabolism (9, 13, 44, 45, 46), the mechanism of this interaction is unclear. Here, we identified that PPARδ contains the typical motif of AMPK phosphorylation substrate (LxRxxS(50)xxxL), which was further demonstrated by LC/MS/MS and in vitro kinase assay. AMPK induced misfolded PPARδ-S50 phosphorylation, leading to inhibition of p62-mediated misfolded PPARδ autophagic degradation. Other reports suggest that KRAS could regulate autophagy (47, 48), and SW480 and HCT-116 are KRAS mutant colonic cancer cells, whereas the increased PPARδ protein levels were similar to KRAS WT (CW-2, HT-29) colonic cancer cells (Fig. 2A). Protein quality control plays an important role in refolding and degrading misfolded proteins (49, 50, 51). Enhanced degradation of misfolded proteins promotes tumorigenesis (52), and cancer cell exhibits the capacity to remove misfolded protein (53). In contrast, the accumulation of misfolded protein is associated with tumor suppression (52, 53). PPARδ-S50 phosphorylation inhibited PPARδ transcription activity, metabolic pathway, and tumor growth, suggesting that AMPK acts as tumor suppressor in cancer cells by inducing PPARδ phosphorylation. Consistently, AMPK induces the downstream GIV phosphorylation leading to tumor suppression (8). Activated PPARδ binds to the peroxisome-proliferator response element (PPRE) consensus sequence (AGGTCANAGGTCA) in the promoter of the target genes and triggers the gene expressions (30, 31, 32, 33). Glut1 and SLC1A5 are the direct targets of PPARδ (29), while AMPK induced PPARδ-S50 phosphorylation leading to inhibition of this event, suggesting that PPARδ-S50 phosphorylation suppressed PPARδ-mediated glucose and glutamine metabolic pathways. Glucose metabolism is required for biosynthesis of macromolecules in most types of cancer cells (54). In this process, Glut1 plays a critical role in glucose uptake to maintain cancer cell growth and proliferation (55, 56). In addition, SLC1A5 serves as an important regulator of essential amino acid influx (57). Depletion or silence of SLC1A5 terminates tumor progression (29, 58). Consistent with this, our results showed that AMPK-induced PPARδ-S50 phosphorylation abolished PPARδ transcription activity and reduced uptake of glucose and glutamine and, subsequently, decreased cancer cell proliferation and xenograft tumor growth. Taken together, AMPK induced PPARδ-S50 phosphorylation and reduced PPARδ transcription activity, leading to inhibition of PPARδ-mediated metabolic pathway and tumor growth (Fig. 7).

Figure 7.

Model of the AMPK/PPARδ pathway inhibits tumor growth. Activated AMPK induced PPARδ-S50 phosphorylation, leading to accumulation of misfolded PPARδ and inhibition of PPARδ transcription activity and subsequently decreased cancer cell metabolism and tumor growth. AMPK, 5′AMP-activated protein kinase; PPARδ, peroxisome proliferator–activated receptor δ.

Experimental procedures

Cells, reagents, and plasmids

The human SW480, HCT-116, HT-29, and CW-2 colonic cancer cell lines were obtained from the ATCC and maintained in 10% fetal bovine serum (FBS, Gibco) Dulbecco's modified Eagle's medium (DMEM). PPRE₃-lu plasmid was described previously (31). Human PPARδ cDNA was cloned into the pcDNA3 vector, PPARδ/S50A, or PPARδ/S50E was mutated by the site-directed mutagenesis method and identified by DNA sequencing. The mutant primers: S50A forward: CCCGGAGCTCCGCGCCACCCTCACT; S50A reverse: AGTGAGGGTGGCGCGGAGCTCCGGG. S50E forward: CCCGGAGCTCCGAGCCACCCTCACT; S50E reverse: AGTGAGGGTGGCTCGGAGCTCCGGG. p62 and AMPKα shRNA plasmids were purchased from GeneChem. Plasmids were transfected into cells by using the TurboFect Transfection Reagent (Thermo Fisher Scientific). AICAR and protease inhibitor cocktail were obtained from Sigma. Metformin was purchased from Bomei biotech. Geneticin (G418 sulfate) was obtained from Thermo Fisher Scientific.

Antibodies and Western blot

PPARδ (Cat: SC-74440) was obtained from Santa Cruz. GAPDH (Cat: 60004) was obtained from Proteintech. p62 (Cat: ab56416) was obtained from Abcam. Actin (Cat: D195316), GST (Cat: D110271), and Flag (Cat: D110005) were purchased from Sangon Biotech. AMPKα (RLT026), phospho-AMPKα1/2 (RLP0575), and anti-phosphoserine (RLM3440) were purchased from Ruiying Biological. The p-PPARδ-S50 mouse polyclonal antibody was developed by using human PPARδ peptide PSSSYTDLSRSSSpPPSLLDQL (nonphosphorylation peptide as a negative control), which was purchased from Chinese Peptide Company. Western blot analysis showed that the p-PPARδ-S50 antibody did not cross-react with other PPAR family members (Fig. S1). Secondary antibodies were obtained from Jackson ImmunoResearch. Cells were lysed in the lysis buffer (50 mM Tris HCl, pH 7.4, 250 mM NaCl, 0.5% Triton X-100, PMSF, 10% glycerol, protease inhibitor cocktail). Protein concentration was determined by using the Pierce BCA Protein Assay Kit (Thermo). The samples were subjected to SDS-PAGE, transferred to a nitrocellulose membrane, and then probed by the indicated antibodies and developed by using an ECL reagent. Blots were quantified by using ImageJ.

Immunoprecipitation and GST pull-down assay

For immunoprecipitation, cell lysates were precleared with protein A/G PLUS-agarose (Cat: sc-2003, Santa Cruz). Precleared lysates with protein A/G PLUS-agarose beads were added with indicated primary antibodies for 1 h. After that, 15-μl protein A/G PLUS-agarose beads were added and rocked overnight at 4 °C. The beads were washed with ice-cold PBS and subsequently subjected to Western blot with indicated antibodies. For GST pull-down analysis, 15-μl MagBeads GST Fusion Protein Purification (C650031, Sangon Biotech Shanghai) was added to cell lysates and rocked overnight at 4 °C. After that, beads were washed with ice-cold PBS. The samples were subjected to Western blot with the indicated antibodies.

Soluble and insoluble protein fractionation analysis

Cells were incubated with the lysis buffer (50 mM Tris HCl, pH 7.6, 150 mM NaCl, 1% Triton X-100, 30 mM NaF, 1 mM EDTA, protein inhibitors) on ice for 20 min. The lysates were centrifuged at 500g for 2 min at 4 °C. The supernatants were further centrifuged at 20,000g for 20 min at 4 °C. The supernatants and pellets were collected as detergent-soluble and -insoluble fractions. Insoluble fractions were sonicated. Fractions were subjected to 7% SDS-PAGE and Western blot.

L-glutamine uptake analysis

L-glutamine uptake analysis was described previously (29). After amino acid starvation by FBS deprivation for 20 h, SW480 cells (1 × 10⁴) were incubated with D-PBS buffer (20 mM Hepes and 1 g/l D-glucose, CaCl₂, MgCl₂) for 3 h. After that, cells were cultured in DMEM containing 4 mM L-glutamine for 45 min. Cells were washed with PBS, and intracellular glutamine levels were measured using the L-Glutamine Assay kit (Goybio) and normalized to total protein levels.

Glucose consumption and lactate release assay

The glucose consumption and lactate release assay were described previously (29). SW480 cells (1 × 10⁴) were grown in DMEM (5.5 mmol/l glucose) with 10% FBS. After cell confluence, the medium was discarded, and then, the same FBS/DMEM was added for 12 h. The medium was collected for glucose consumption and lactate release assay. The glucose concentrations were assayed by using the assay kit (Applygen). Lactate release was assayed by colorimetric assay kit (Goybio). The cell number was determined by trypan blue exclusion assay for normalization.

In vitro kinase assay

PPARδ cDNA was subcloned into the PGEX-6P-1 vector. The GST-PPARδ or S50A mutant was expressed in Escherichia coli strain BL21. The recombinant protein was purified by glutathione-conjugated Sepharose beads. 50 ng purified AMPKα heterotrimers (Signal Chem), GST-PPARδ, or GST-PPARδ-S50A (10 ng) was added in the reaction buffer (20 mM Hepes, pH 7.4, 10 mM MgCl₂, 1 mM EGTA, 10 mM ATP, 1 mM DTT) for 40 min at 30 °C. The reactions were subjected to Western blot with indicated antibodies.

MS assay

SW480 cells were treated with or without glucose starvation for 3 h. Cell lysates were subjected to immunoprecipitation by using the PPARδ antibody. Samples with the loading buffer were boiled and then subjected to SDS-PAGE. The gel-purified phosphorylated and unphosphorylated PPARδ proteins were digested with chymotrypsin and trypsin. The digested peptides were assayed by using UPLC-Q-Exactive (Thermo Fisher) at Aixiang Biotech Company in China. Peptide coverage is 78.23% of PPARδ amino acid sequence. The data were searched against UniProt database, and the peptide false discovery rate was <1%.

Immunofluorescent analysis

SW480 cells were treated with metformin (10 mM) for 1 h. Cells were fixed (3.7% paraformaldehyde), permeabilized (0.5% Triton X-100), blocked (10% BSA), and then incubated with the indicated primary antibodies and secondary antibodies (Jackson ImmunoResearch). Immunostained cells were viewed in a confocal microscope.

Luciferase assay

For the PPARδ transcription activity assay, cells were transfected with PPRE₃-lu and Ptk-RL together with PPARδ or mutant plasmids for 36 h. Cells were treated with GW501516 as indicated for another 12 h. For the Glut1or SLC1A5 transcription activity analysis, cells were transfected with Glut1-lu or SLC1A5-lu together with Ptk-RL, PPARδ, or mutant plasmids as indicated for 36 h. Cell lysates were assayed by using a dual-luciferase reporter assay system (Promega).

Soft agar analysis

SW480 cells were transfected with PPARδ or mutant (S50A, S50E) plasmids. Cells were selected with G418 to develop stable gene-expressing cell lines. Cells (3 × 10³) were suspended in 0.35% noble agar with 10% FBS DMEM and were layered over 0.5% agar in six-well plates. A normal culture medium with 500 μg/ml G418 was added to each well. After 2 weeks, dishes were stained with 0.05% crystal violet, and colonies were counted.

Xenograft mouse model

Eight-week-old female NU/NU nude mice were purchased from the laboratory animal center of Yangzhou University. Stable PPARδ, S50A, or S50E expressing SW480 cells (1 × 10⁶) by using G418 selection were injected subcutaneously in nude mice. Each group contains five mice. After 4 weeks, the tumor volume was measured. Tumor volume = 1/2 (length × width²). All studies were carried out with the approval of the Jiangsu University Animal Care Committee.

Statistical analysis

Data are expressed as the mean ± SEM. Differences between two dependent groups were evaluated with the paired Student’s t test.

Data availability

Additional data are available as Supplementary information.

Supporting information

This article contains supporting information.

Conflict of interest

The authors declare that they have no conflicts of interest with the contents of this article.

Edited by Eric Fearon