Estrogen-related receptor alpha directly binds to p53 and cooperatively controls colon cancer growth through the regulation of mitochondrial biogenesis and function

The ERRα/PGC1α complex drives cancer cell survival through mitochondrial metabolic programs favoring tumor development [28, 49, 50]. Moreover, the p53/PGC1α complex drives tumor metabolism and metastasis through mitochondrial energy programs enabling mitochondria to cope with metabolic stress [45, 51]. However, the ability of both ERRα and p53 to influence cell fate decisions, such as survival and death, has not been fully investigated. To determine whether ERRα might interact with p53, we first used 293F cells to perform a flag-ERRα protein purification (Fig. 1a) followed by both liquid chromatography-tandem mass spectrometry (LC-MS/MS; Fig. 1b) and size exclusion chromatography (SEC; additional file Fig. S1A) to determine the potential protein-protein interaction between ERRα and p53 and the stability of this complex, respectively. With a considerable percentage of confidence, we found that ERRα interacts with p53 to form a stable protein complex (Fig. 1a, b and additional file Fig. S1B, C). This finding prompted us to determine whether this interaction occurs only in an artificial system through mammalian cell transfection or as an endogenous protein-protein interaction. First, using 293T cells followed by overexpression of the flag-ERRα protein and immunoprecipitation (IP)/immunoblot (IB) analysis, we confirmed the protein-protein interaction between ERRα and p53 in a transfection system (Fig. 1c). Importantly, we then performed an endogenous co-IP of ERRα and p53 using 293T cells and showed that ERRα indeed interacts with p53 in cells (Fig. 1d).

ERRα regulates an extremely broad repertoire of mitochondrial and metabolic genes, making it a very attractive target for cancer treatment [17]. To determine whether its binding to p53 could occur in cancer cells, we used isogenic human colon cancer cell lines HCT-116^p53+/+ (wild-type p53) and HCT-116^p53−/− (p53-null background). We performed an endogenous co-IP/IB analysis and observed that the wild-type p53 protein interacted with ERRα, whereas the p53-deficient cells showed no interaction with ERRα (Fig. 1e). Next, we performed a proximity ligation assay (PLA) to confirm that ERRα binds to p53. In the presence of wild-type p53, a clear immunofluorescence (IF) staining was observed, whereas in p53-deficient cells the IF staining was completely absent (Fig. 1f). To expand our findings, we used p53 mutant (missense or GOF) colon cancer cell lines, DLD-1 and HCT-15, to perform an endogenous co-IP/IB analysis. Interestingly, we also showed a marked increased binding of ERRα to the mutant form of the p53 protein (Fig. 1g).

To further determine whether ERRα directly interacts with p53, we used computational techniques to identify the binding interface and build a model of the proposed physical interaction of both proteins. Results indicate that the ERRα LBD interacts with three potentially different sites of the p53 CTD (Fig. 1h and additional file Fig. S2A-H). To directly confirm the in silico prediction, we constructed two different truncated proteins, an ERRα LBDΔ190-423 with a V5 Epitope/His tag and a p53 CTDΔ300-393 with a Flag tag. Next, we performed a Flag purification of p53 CTDΔ300-393 followed by an affinity pull-down assay and IB analysis to show that p53 CTDΔ300-393 physically interacts with ERRα LBDΔ190-423 (Fig. 1i). These results provided unequivocal evidence showing that the LBD of ERRα directly binds to the CTD of p53. Notably, we have performed endogenous IP against ERRα in 293T cells (p53 wild-type) and a colon cancer cell line DLD-1 (p53 mutant), and we showed that PGC-1α does not act as a surrogate protein ligand of the ERRα/p53 complex (additional file Fig. S1D). Overall, the interaction of the two proteins seems to occur in a colon cancer scenario regardless of p53 protein status, which could be exploited as a therapeutic opportunity for targeting ERRα in different cancer-related mutant forms of p53.

As indicated earlier, TP53 remains the most elusive target in cancer [34, 52], whereas ERRα can be potentially inhibited by drugs [53‐55]. Therefore, to unravel the function and to understand the role of the ERRα and p53 complex in colon cancer, we first used short hairpin RNA against ERRα in different colon cancer cell lines to knock down ERRα protein expression. Knocking down ERRα in HCT-116^p53+/+ (p53 wild-type), DLD-1, or HCT-15 (missense or GOF p53 mutants) colon cancer cell lines reduced the expression of the p53 protein, independent of p53 status (Fig. 2a). Because ERRα and p53 each govern a wide array of mitochondrial network genes [26, 41], we investigated whether the inhibition of ERRα might affect mtOxPhos activity. We measured the protein levels of the mtOxPhos complex subunits I-V (ATP5A, UQCR2, MTCO1, SDHB, and NDUFB8) in HCT-116^p53+/+ cells and also determined the expression of PGC-1α, a key regulator of mtOxPhos activity and mitochondrial biogenesis, in these cells. Results showed a profound decrease in both mtOxPhos and PGC1α protein expression after knocking down ERRα (Fig. 2b). Furthermore, gene silencing of ERRα in HCT-116^p53+/+ cells significantly decreased the mitochondrial DNA (mtDNA) copy number and the relative levels of intracellular ATP (Fig. 2c, d). We next used IF to determine mitochondrial biogenesis through two mitochondrial markers, the cytochrome c oxidase subunit IV (COX-4) and the voltage-dependent anion channel 1 (VDAC1). We observed a significant reduction of COX-4 and VDAC1 expression in HCT-116^p53+/+ cells after knocking down ERRα (Fig. 2e). Interestingly, gene silencing of ERRα in HCT-116^p53−/− (additional file Fig. S3A) showed a substantial change in the distribution of the co-localization of both mitochondrial markers from an elongated (enlarged shMock) to a fragmented network (enlarged shERRα#81 and shERRα#83) and a significant difference in the expression of COX-4 and VDAC1 when compared with HCT-116^p53+/+ cells (additional file Fig. S3B). Finally, we observed less cell proliferation and cell cycle arrest in response to knocking down ERRα regardless of the status of p53 (Fig. 2f and additional file Fig. S3C-E). For the first time, these results provide evidence showing that gene silencing of ERRα affected p53 expression, regardless of p53 status. Targeting ERRα affected cell proliferation through the regulation of mitochondrial function and biogenesis independently of p53 status (i.e., whether p53 wild-type or deficient).

Several antagonists have been designed that lead to the degradation of ERRα [55, 56]. However, whether ERRα antagonists can block the interaction between ERRα and p53 is unknown. Thus, to extend our analysis of the effectiveness of targeting the direct binding of ERRα and p53 in colon cancer, we selected an antagonist of ERRα known as XCT790. First, to determine whether or not XCT790 might be a specific antagonist of ERRα expression in colon cancer cell lines [57, 58], we used CCD-18co normal colon fibroblasts and a panel of nine different colon cancer cell lines, including colon cancer cells expressing wild-type p53 (HCT-116^p53+/+ and Lim1215), p53 mutant (missense or GOF; HT-29, DLD-1, HCT-15, SW480, and WiDr), and p53-deficient (HCT-116^p53−/− and Caco-2) cells. XCT790 treatment decreased proliferation in almost all colon cancer cell types compared to normal colon cells (Fig. 3a). We extended our analysis by performing an IB analysis with ERRα, ERRγ, and ERα in normal CCD-18co colon fibroblasts and in a panel of colon cancer cell lines (Lim1215, HCT-116^p53+/+, HCT-116^p53−/−, HT-29, DLD-1, and HCT-15). Results showed that treatment of cells for 48 h with XCT790 (up to a concentration of 15 μM) totally abrogated the protein expression level of ERRα and slightly affected the expression level of ERα (Fig. 3b). Interestingly, the ERRα protein seemed to be highly expressed in colon cancer cells compared to normal colon fibroblasts and was inversely associated with the expression of ERRγ, which was not expressed in colon cancer cells (Fig. 3b). Similar to knocking down ERRα expression, the treatment of cells with XCT790 decreased p53 protein expression in most colon cancer cell lines, independently of p53 status (Fig. 3b). Notably, using a transient transfection system to overexpress ERRα restored the expression of p53 to normal levels in HCT-116^p53+/+ cells (additional file Fig. S4A).

Despite the orchestration of mitochondrial bioenergetics of tumors by ERRα, no scientific evidence exists underlying the mechanism by which ERRα could be therapeutically exploited as a new metabolic vulnerability in the presence or loss of wild-type p53 function. We next sought to delineate this mechanism by using quantitative proteomics analysis by SWATH-MS in the mitochondrial subcellular membrane/organelle fractionation (Fig. 3c). We compared the mitochondrial proteomics profile of four different cellular conditions, including (i) HCT-116^p53+/+ cells with intact ERRα and p53 function; (ii) HCT-116^p53+/+ cells with intact p53 function, but with suppressed ERRα expression; (iii) HCT-116^p53−/− cells with intact ERRα function, but no p53 function; and (iv) HCT-116^p53−/− cells with suppressed ERRα expression and no p53 function (Fig. 3c). A Venn diagram was used to show a comparison of the mitochondrial proteomic profile of 3 different combinations, including no ERRα (ii/i), no p53 (iii/i), and no ERRα and no p53 (iv/i). The results showed a total of 293 identical mitochondrial proteins (76% of the proteins analyzed) within the three groups that exhibited decreased expression levels (Fig. 3d, Table S1). A comparison of cells without ERRα and p53 (iv) with cells expressing the ERRα and p53 complex (i) showed that the absence of the ERRα/p53 complex affected central metabolism by altering KEGG pathways (Kyoto Encyclopedia of Genes and Genomes). The increased activation of KEGG pathways is related to central carbon metabolism in cancer and glycolysis, and profound decreases in the activation of KEGG pathways is associated with Parkinson’s disease, OxPhos, Huntington’s disease, Alzheimer’s disease, thermogenesis, and several metabolic pathways (false discovery rate ≤ 0.01; Fig. 3e, Table S1). Intriguingly, comparing the absence of ERRα only (ii) or the absence of p53 only (iii) to the presence of the ERRα and p53 complex (i) indicated that the absence of the function of a single protein (ERRα or p53) affected central metabolism, such as glycolysis and carbon metabolism, by increasing the activation of KEGG pathways, and dramatically decreased the activation of KEGG pathways associated with mitochondrial metabolism, such as Parkinson’s disease and OxPhos (FDR ≤ 0.05; additional file Fig. S4B, C, Table S1). Overall, our results showed that ERRα is highly expressed in colon cancer and directly targeting ERRα with XCT790 affected p53 expression. Collectively, these results prompted us to conclude that ERRα and p53 cooperatively control mitochondrial function; however, we could not distinguish whether the effect of ERRα and p53 on mitochondrial function is additive or an independent molecular event.

To further determine the molecular mechanisms underlying the function of the ERRα and p53 complex in cooperatively promoting colon cancer survival through the regulation of mitochondrial biogenesis and mtOxPhos, we analyzed the effect of the presence or absence of ERRα and p53 (Fig. 4a). We found that the absence of the ERRα and p53 complex (iv), but not the absence of either ERRα (ii) or p53 (iii) only, gradually decreased the protein levels of all five mtOxPhos complex subunits (complex I-V) and VDAC1 (Fig. 4b). Indeed, in HCT-116^p53−/− cells, the expression of mtOxPhos markers and VDAC1 was strongly downregulated following treatment with XCT790 (Fig. 4b). To gain more insight into the mechanism of the ERRα and p53 complex in coordinating mitochondrial function, we measured the mtDNA copy number and the relative levels of intracellular ATP. Consistent with the results above, the absence of the ERRα and p53 complex (iv), but not the absence of either ERRα (ii) or p53 (iii) only gradually decreased the mtDNA copy number and the relative levels of intracellular ATP (Fig. 4c, d). Again, HCT-116^p53−/− cells showed the most dramatic decrease in mtDNA copy number and lower levels of intracellular ATP in a time-dependent manner following treatment with XCT790 (Fig. 4c, d). Subsequently, we measured mitochondrial mass using MitoTracker Red CMXRos. Again, treated HCT-116^p53−/− cells (iv) showed a significant decrease in the mitochondrial mass compared to untreated HCT-116^p53−/− cells (iii) or HCT-116^p53+/+ control cells (i), but not HCT-116^p53+/+ cells treated with XCT790 (ii) (Fig. 4e).

Next, we used IF to quantify mitochondrial biogenesis by using two mitochondrial markers, COX-IV and VDAC1. We observed a reduction in the co-localization of COX-IV and VDAC1 in HCT-116^p53+/+ cells treated with XCT790 (ii), untreated HCT-116^p53−/− cells (iii), and treated HCT-116^p53−/− cells (iv) compared to the presence of the complex (i) (Fig. 4f). Also, we observed changes in the distribution of the co-localization of both mitochondrial markers from an elongated (enlarged) to a fragmented network, which was observed in the absence of the ERRα and p53 complex (Fig. 4f). Thus, one might reasonably expect that affecting mitochondrial biogenesis causes mitochondrial metabolic stress leading to accumulation of mitochondrial abnormalities, such as mitochondrial dysfunction, ROS production, and MMP [59].

We stained live cells first with MitoTracker Red CMXRos and with cytochrome c (cyto c). By using IF microscopy, we observed that the absence of the ERRα and p53 complex (iv), but not the absence of either ERRα (ii) or p53 (iii) only, gradually impacted the increased level of oxidized mitotracker (orange spots) and diffused cyto c across the cytosol and nucleus (Fig. 5a). We also measured the levels of intracellular ROS and the presence of mitochondrial-apoptotic proteins in the cytosol. As expected, in response to ERRα inhibition followed by treatment with XCT790 in HCT-116^p53−/− cells (iv), we observed significantly increased levels of ROS (Fig. 5b). Moreover, in the absence of the ERRα and p53 complex (iv) and also in the absence of only ERRα (ii) or p53 (iii), we observed a leaking from mitochondria to cytosol of apoptosis-inducing factor (AIF), Smac/Diablo, and cyto c (Fig. 5c). To confirm these results, we stained live cells with MitoTracker Red CMXRos and then with an AIF antibody. By using IF microscopy, we observed that the absence of either ERRα (ii) or p53 (iii) only gradually increased the level of oxidized mitotracker (orange dots) and diffused AIF into the nucleus (Fig. 5d). Notably, XCT790 caused a marked G1 cell cycle arrest and less proliferation regardless of p53 status (additional file Fig. S5A-C). However, the absence of the ERRα and p53 complex (iv) showed increased cleaved forms of two apoptotic markers, caspase-3 and poly (ADP-ribose) polymerase (PARP), and more accumulation of Annexin V apoptotic cell staining leading to an irreversible apoptotic signaling cascade (Fig. 5e, f). Overall, these results provided evidence that loss of the ERRα and p53 complex progressively affected mitochondrial biogenesis, mtOxPhos, ROS production, and MMP, leading to more mitochondrial metabolic stress and apoptosis.

To understand the clinical relevance of targeting ERRα in wild-type p53, p53 mutant (missense or GOF), and p53-deficient colon cancer patient samples, we performed direct DNA sequencing of the complete p53 genomic locus of 37 samples that came from patient-derived colon tumors. Overall, we identified 33 mutant p53 tumors and 4 wild-type p53 tumors (additional file Fig. 6a). Comparing the mutant p53 tumors, we identified 18 cases of nonsense mutant p53 tumors (p53-deficient) and 15 cases of missense mutant p53 tumors (GOF). Intriguingly, the single nucleotide polymorphism (SNP) 215C > G that translates to a P72R codon mutation was found in 63% of the patient cohort (Fig. 6a). Based on these data, we created matched HCT-116^p53−/− (p53-null background) cells containing the tumor-derived mutant form of p53 P72R fused with GFP and luciferase, along with the vector-transfected control HCT-116^p53+/+ parental cell line (additional file Fig. S6B). Colon cancer cell lines, HCT-116^p53+/+ (wild-type p53) and HCT-116^p53_P72R, were subjected to treatment with XCT790 (15 μM) to examine proliferation. The results indicate that mutant HCT-116^p53_P72R cells showed significantly increased proliferation compared with wild-type p53-expressing cells, and treatment with XCT790 was very effective at inhibiting growth of both wild-type and mutant p53-expressing cells (Fig. 6b).

To illustrate the impact of the genetic status of p53 on the efficacy of XCT-790, we used an in vivo tumor model with patient-derived colon tumor explants grown in SCID mice. We selected and established a missense mutant p53 genotype (single mutant P72R: HJG172, additional file Fig. 6a). DNA sequencing showed that HJG172 contains a single point mutation (215C > G), the most frequent in the patients analyzed (Fig. 6a). Mice bearing xenografts were treated with vehicle or one of two different concentrations of XCT790 (2.5 mg/kg or 5 mg/kg body weight) every 3 days over a 21-day period.

Notably, a preliminary random toxicity study was performed, showing that neither XCT790 concentration significantly affected liver or spleen weight in mice (n = 4; additional file Fig. S6C). As expected, neither concentration of XCT790 affected body weight, liver, or spleen weights of HJG172 (mut ^p53_72R; Fig. 6c and additional file Fig. S6D). Interestingly, we observed a significant reduction in tumor growth and tumor weight of HJG172 (mut ^p53_72R), compared to vehicle-treated control tumors, with a delay of ~ 40% tumor growth (Fig. 6d, e). Taken together, these results provide evidence that XCT790 might be considered as a potential drug to treat colon cancer patients with defective p53 status (Fig. 6f).

The 293T cells were purchased from American Type Culture Collection (Manassas, VA, USA) and maintained in high glucose Dulbecco’s modified Eagle’s medium (DMEM, Invitrogen, Camarillo, CA, USA) supplemented with 10% fetal bovine serum (FBS, Gemini Bio-Products, Calabasas, CA, USA), 100 U/mL penicillin, and 100 mg/mL streptomycin. FreeStyle^TM 293-F cells were purchased from Thermo Fisher Scientific (Carlsbad, CA, USA) and maintained in FreeStyle 293 Expression Medium. CCD-18Co normal human colon fibroblasts and human colon cancer cell lines, HCT-116^p53+/+, DLD-1, HCT-15, HT-29, SW480, WiDr, and Caco-2 were purchased from ATCC. HCT-116^p53−/− cells were kindly provided by Dr. Bert Vogelstein [72]. Lim1215 cells were obtained from the European Collection of Authenticated Cell Cultures (ECACC). HCT-116^p53+/+, HCT-116^p53−/−, and HT-29 cells were grown in McCoy’s 5A Medium (Invitrogen, Camarillo, CA, USA), supplemented with 10% FBS, 100 U/mL penicillin, and 100 mg/mL streptomycin. CCD-18Co, WiDr, and Caco-2 were grown in Minimum Essential Media (MEM, Invitrogen, Camarillo, CA, USA), supplemented with 10% FBS, 100 U/mL penicillin, and 100 mg/mL streptomycin. DLD-1, HCT-15, and Lim1215 cells were grown in RPMI-1640 (Invitrogen, Camarillo, CA, USA) supplemented with 10% FBS, 100 U/mL penicillin, and 100 mg/mL streptomycin. For Lim1215 cells, we added 0.6 μg/ml insulin and 1 μg/ml hydrocortisone. SW480 cells were maintained in HyClone Leibovitz’s L-15 medium (GE Healthcare, Grand Island, NY, USA), and supplemented with 10% FBS, 100 U/mL penicillin, and 100 mg/mL streptomycin. All cells were cultured in a humidified incubator at 37 °C with 5% CO₂. All the cell lines were cytogenetically tested and authenticated before the cells were frozen. All the cell lines were maintained in culture no longer than 10–15 passages.

Direct DNA sequencing of the complete p53 genomic locus was performed from tumor tissues extracted from human biopsies. DNA was sequenced for the exons of p53 sequence with primers listed at https://​p53.​iarc.​fr/​Download/​TP53_​SangerSequencing​_​IARC.​pdf.

Six- to eight-week-old SCID mice (Vital River Labs, Beijing, China) were used for these experiments. This study was approved by the Ethics Committee of Zhengzhou University (Zhengzhou, Henan, China). The PDX tumor mass was subcutaneously implanted into the back of SCID mice. When tumors reached an average volume of 100 mm³, randomized cohort of mice were divided into three groups and subjected to oral gavage with (1) XCT790 at 2.5 mg/kg (n = 7); (2) XCT790 at 5.0 mg/kg (n = 7); or vehicle control (n = 7). XCT790 was administrated every 3 days for 21 days. Tumor volume was calculated from measurements of three diameters of the individual tumor base using the following formula: tumor volume (mm³) = (length × width × height × 0.52). Mice were monitored until tumors reached 1.0 cm³ total volume, at which time mice were euthanized and tumors extracted.

The plasmid pCMV flag ERRα was purchased from Addgene Inc. (Cambridge, MA, USA). For overexpression of the ERRα protein, the plasmid was transfected into 293T cells by using the iMFectin PolyDNA Transfection Reagent (GenDepot, Barker, TX, USA) following the manufacturer’s suggested protocols. The transfection medium was changed after an overnight transfection and then replaced with fresh growth medium. Cells were cultured for 48 h and then harvested in phosphate-buffered saline (PBS) for further analysis, such as immunoprecipitation (IP) and immunoblotting (IB). For large-scale transient transfection of the ERRα gene, the plasmid was transfected into FreeStyle^TM 293-F cells by using polyethylenimine (PEI, Sigma-Aldrich, St. Louis, MO, USA) following the manufacturer’s suggested protocols. Cells were cultured for 48 h and then centrifuged at 4 °C for 5 min at 1000 rpm. The cell pellet was used for further analysis, such as protein purification.

The cell pellet was rinsed in PBS and disrupted in 5 ml lysis buffer (50 mM HEPES, pH 7.8, 300 mM NaCl, 1 mM EDTA, 1 mM DTT, 0.5% (v/v) NP-40, 10% (v/v) glycerol and protease inhibitor cocktail [Sigma-Aldrich, St. Louis, MO, USA]), on ice, for 30 min. The whole cell lysate was centrifuged at 4 °C for 30 min at 15,000 rpm, and the supernatant fraction was incubated at 4 °C on a rotary mixer from 3 to 6 h with ANTI-FLAG®M2 Affinity Gel (Sigma-Aldrich). After incubation, the affinity gel was washed 3 times with wash buffer (50 mM HEPES, pH 7.8, 500 mM NaCl, 1 mM EDTA, 1 mM DTT, 0.02% (v/v) NP-40, 10% (v/v) glycerol and protease inhibitor cocktail [Sigma-Aldrich]). An incubation step was performed with 5 mM ATP and 10 mM MgCl₂ diluted in wash buffer to remove chaperone protein-binding contamination. Afterwards, the affinity gel was washed 3 times in wash buffer, as described above. The affinity-bound flag-ERRα protein was eluted by adding 5 volumes of elution buffer containing 100 μg/ml of 3X FLAG® peptide (Sigma-Aldrich), 50 mM HEPES, pH 7.8, 300 mM NaCl, and 1 mM DTT and left on ice for 1 h each cycle, repeating 4 times. For every cycle, samples were centrifuged at 700×g for 3 min, and supernatant fraction containing the eluted protein was collected. The eluted protein was separated by SDS–polyacrylamide gel electrophoresis (PAGE) and stained with Coomassie Blue staining solution (Coomassie R-250, 50% methanol, and 10% acetic acid).

To identify endogenous binding partners of ERRα, the purified-ERRα elution fractions were separated by 8% SDS-PAGE and stained with Coomassie stain solution (as above). Proteins in excised gel slices were digested with trypsin (100 ng/sample) and eluted as described [73]. Eluted proteins were digested with Glu-C (100 ng/sample, Promega) for 16 h at 37 °C. LC-MS/MS analysis of enzymatic peptides was performed using Sciex TripleTOF™ 5600 coupled with Eksigent 1D+ nano LC. Raw data were processed and searched with the ProteinPilot™ software (version 4.5) using the Paragon™ algorithm. Protein identification was obtained by searching UniProtKB human database and filtered at ≥ 95% confidence cutoff.

The flag-ERRα purified protein was further purified again using size exclusion chromatography to verify the stability of the complex. The samples were purified on a Superose 6 size exclusion column (GE healthcare, Marlborough, MA, USA) in 50 mM HEPES, pH 7.8, 150 mM NaCl, and 1 mM DTT. The fractions corresponding to ERRα were eluted, collected into a new Eppendorf tube, and loaded into an SDS–PAGE for further (IB) analysis using ERRα and p53 antibodies.

Whole cell lysates were prepared with a lysis buffer (137 mM NaCl, 20 mM Tris-HCl, pH 8, 1 mM EDTA, 10% (v/v) glycerol, 1% (v/v) NP-40, and 1 × protease inhibitor tablet). A total of 1000 to 3000 μg of cell lysates were immunoprecipitated with 2 μg of each respective antibody and 40 μL of protein A/​G Sepharose beads (Santa Cruz Bio, CA, USA). The samples were rotated at 4 °C overnight. After the beads were washed 3 times with lysis buffer, the pellets were analyzed by SDS–PAGE and transferred to polyvinylidene difluoride (PVDF) membranes. Membrane was blocked with 5% non-fat dry milk for 1 h at room temperature and incubated with the appropriate primary antibody overnight at 4 °C (see Table S2). After 3 washing with PBS containing 0.1% Tween 20, the membrane was incubated with a horseradish peroxidase-conjugated secondary antibody at a 1:5000 dilution, and the signal was detected with a chemiluminescence reagent and the Amersham Imager 600 imager (GE Healthcare, Piscataway, NJ, USA).

Cell lysates were prepared with radioimmunoprecipitation (RIPA) assay buffer (50 mM Tris-HCl pH 7.4, 1% (v/v) NP-40, 0.25% (v/v) sodium deoxycholate, 0.1% (v/v) SDS, 150 mM NaCl, 1 mM EDTA, and 1 × protease inhibitor tablet). Equal loading of protein was confirmed by the DC^TM protein assay (Bio-Rad, Hercules, CA, USA). Protein sample was separated by SDS–PAGE and transferred to PVDF membranes. Membranes were blocked with 5% non-fat dry milk for 1 h at room temperature and incubated with the appropriate primary antibody overnight at 4 °C (see Table S2). After 3 washes with PBS containing 0.1% (v/v) Tween 20, the membranes were incubated with a horseradish peroxidase-conjugated secondary antibody at a 1:5000 dilution, and the signal was detected with a chemiluminescence reagent using the Amersham Imager 600 imager (GE Healthcare, Piscataway, NJ, USA).

To detect the association of ERRα with p53 by PLA, the Duolink in situ red starter kit was used (DUO92101-1KT–Sigma-Aldrich), following the manufacturer protocol exactly as described to perform this study.

The crystal structure of the ERRα LBD in complex with the PGC-1α box3 peptide was downloaded from the Protein Data Bank (PDB; entry 3D24 for ERRα/PGC-1α) and used as a template to build this model of interaction [74, 75]. Next, we found in the scientific literature that the ERα LBD binds directly to p53 [76]. The sequence identity (32.3%) and similarity (56.5%) between the LBDs of ERRα and ERα is in agreement with the highly conserved 3D structure of this domain by superposition (PDB IDs, 3D24 chain A, and 2IOG chain A). On the other hand, the NMR structure (PDB ID: 1DT7, chain X) of the p53 CTD (Lys370-Asp393) shows 22.7% identity and 40.9% similarity with PGC1α box 3 peptide. Thus, the p53 C-terminal regulatory helical domain (PDB ID: 1DT7, chain X) was superposed on PGC1α box 3 peptide in the template (PDB ID: 3D24) followed by removal of PGC1α box 3 peptide and energy minimization. As shown in additional file 2: Figure S2A-H, we developed the model interaction between the ERRα LBD and the p53 CTD. In this model, three potential interaction sites have been identified and the electrostatic surface representation of p53 binding site on ERRα. The 3-D First Fourier Transform–based protein docking algorithm of HEX 8.00 [77] was then used for docking experiments to determine the possible binding mode between ERRα and p53.

To construct the V5-His-tagged expression vector of ERRα, the Flag-tagged expression vector of p53, and the EGFP-luciferase-tagged expression vector of mutant p53^P72R, the DNA sequences corresponding to ERRα LBD (aa 193–423), p53 CTD (aa 300–393), and p53^P72R were amplified by PCR. The PrimeSTAR® GXL DNA Polymerase high fidelity enzyme (Takara Bio, Mountain View, CA, USA) was used for PCR. We designed specific primers for V5-His-ERRα (aa 193–423) as follows: F: 5′atggatccatgccagtgaatgcactggtgtc3′; R: 5′cgtctagagtccatcatggcctcgagcatc3′; Flag-p53 (aa 300–393) F: 5′ atggatcccccccagggagcactaagcga3′; R: 5′cctctagatgcatgctcgagtcag3′; and EGFP-luciferase-p53^72R F: 5′ atctcgaggatggaggagccgcagtcag3′; R: 5′ cgaccggtttagtctgagtcaggcccttctgtc3′. The V5-His-ERRα (aa 193–423), Flag-p53 (aa 300–393), and EGFP-luciferase-p53^72R PCR products were digested with BamHI-HF/Xba I and Xho I/Age I following instructions provided by the manufacturer (New England Biolabs, Ipswich MA, USA). The constructs were then inserted into the corresponding sites of pcDNA3.1/V5-His (Invitrogen by Thermo-Fisher Scientific), and pcDNA3/Flag and pLentipuro3/TO/V5/GW/EGFP/luciferase (Addgene Inc., Cambridge, MA, USA) to generate the encoding expression plasmids. Sanger DNA sequencing and the Blast program were used to confirm that the DNA was inserted into the corresponding sites.

To generate stable knockdown ERRα cells, the lentiviral vector of ERRα (shERRα) or pLKO.1-puro shRNA Non-Target, control plasmid DNA (shMock), and packaging vectors (pMD2.0G and psPAX, Addgene Inc., Cambridge, MA, USA) were transduced into 293T cells by using iMFectin PolyDNA Transfection Reagent (GenDepot, Barker, TX, USA) following the manufacturer’s suggested protocols. To generate stable overexpression p53-mutant cells, the lentiviral vector of p53_^P72R containing EGFP and luciferase genes or pLentipuro3/TO/V5/GW/EGFP/luciferase, control plasmid DNA and packaging vectors (pMD2.0G and psPAX, Addgene Inc., Cambridge, MA, USA) were transduced into HCT-116^p53−/− and HCT-116^p53+/+ cells, respectively, by using iMFectin PolyDNA Transfection Reagent (GenDepot, Barker, TX, USA) following the manufacturer’s suggested protocols. The transfection reagents were incubated with 293T, HCT-116^p53−/−, and HCT-116^p53+/+ cells in complete growth medium overnight, and then fresh medium with antibiotics (penicillin/streptomycin) was added. Viral supernatant fractions were collected at 48 h and filtered through a 0.45-μm syringe filter followed by infection into the appropriate cells together with 10 μg/mL polybrene (Sigma-Aldrich). After infection overnight, the medium was replaced with fresh complete growth medium containing the appropriate concentration ranging from 0.2 to 1 μg/mL puromycin (Gemini Bio Product) and cells incubated for an additional 1–3 weeks. The selected stable knockdown cells were used for further experiments.

Genomic DNA (gDNA) was extracted from HCT-116^p53+/+ and HCT-116^p53−/− cells by using DNAzol® Reagent (Thermo Scientific) and quantified using the Human Mitochondrial DNA (mtDNA) Monitoring Primer Set (Takara Bio, Mountain View, CA, USA) according to manufacturer’s protocol. Primer sets for ND1 and ND5 were used for the detection of mtDNA, and the primer sets SLCO2B1 and SERPINA1were used for the detection of nuclear DNA. The quantitation of mtDNA copy number was performed using a 7500 FastDX (Applied Biosystems, MA, USA) and the power SYBR green PCR master mix (Applied Biosystem, Warrington WA1 4SR, UK).

Briefly, 10,000 cells were seeded in 96-well cell culture plates. HCT-116^p53+/+ and HCT-116^p53−/− cells were treated either with DMSO as a negative control or with 15 μM XCT790 for 24 or 48 h. In the plate, the lysed cells were used for a luminescence ATP detection assay system by using ATPlite (PerkinElmer, Boston, MA, USA) with the Synergy™ Neo2 Multi-Mode Microplate Reader (BioTek, Winooski, VE, USA), following the manufacturer’s instructions. An ATP standard curve was generated, and the concentration of ATP was determined by the normalization of cell number.

For mitochondrial staining, HCT-116^p53+/+ and HCT-116^p53−/− live cells were first incubated with 0.3 μM MitoTracker Red CMXRos (Invitrogen, Eugene, Oregon, USA) for 30 min at 37 °C in 5% CO₂. For co-localization and for mitochondrial staining, colon cancer cell lines were fixed with 3.7% formaldehyde, and then permeabilized in 300 μl of 0.5% Triton X-100. In general for immunostaining, after blocking the cover slips in 10% BSA for 1 h, the primary antibody was added overnight. The day after, Alexa Fluor 488-labeled or Alexa Fluor 568-labeled secondary antibodies (Thermo Scientific) were added (see Table S2). The DAPI-Fluoromount-G® solution (Electron Microscopy Sciences, Hatfield, PA, USA) was added for nuclei staining of the cells. Fluorescence-labeled proteins were analyzed using the Zeiss Fluorescent microscopy with apotome attachment and the Zeiss software and Axiocam cameras (Carl Zeiss Microscopy GmbH, Deutschland).

Briefly, 20,000 cells were seeded in 24-well plates. The next day, cells were treated either with DMSO as a negative control or XCT790 (15 μM). Cells were incubated from 0 to 96 h. A survival assay was performed by using the cell proliferation Resazurin sodium salt reagent (Sigma-Aldrich), and the fluorescence excitation/emission wavelengths were measured at 545/595 nm with the Synergy™ Neo2 Multi-Mode Microplate Reader (BioTek, Winooski, VE, USA). Alternatively, 20,000 cells for non-targeting and knockdown of ERRα expression and for non-targeting and overexpression of p53^P72R were seeded in 24-well plates. Cells were incubated from 0 to 108 h, and a survival assay was performed by using the IncuCyteS3 live-cell imaging system (Essen BioScience, Tokyo, Japan), and then the IncuCyteS3:2019 software was used to quantify cell proliferation.

HCT-116^p53+/+ and HCT-116^p53−/− cells were treated either with DMSO as a negative control or with 15 μM XCT790 for 48 h. Afterwards, cells were digested with trypsin and harvested in PBS. Cytosolic, membrane/organelle, and nuclear fractions were isolated using ProteoExtract® subcellular proteome Extraction kit (Millipore, Danvers, MA, USA) following the manufacturer’s suggested protocols. To screen differential protein expression across the samples in membrane/organelle fractions, MS/MS^ALL with SWATH^TM acquisition was conducted. Raw data was collected through information-dependent acquisition (IDA) and SWATH^TM acquisition Sciex TripleTOF™ 5600 with Eksigent 1D+ nano LC. The protein identification, MS peak extraction, and statistical analysis were performed with ProteinPilot^TM (version 4.5), PeakView^TM (version 2.2), and MarkerView^TM (version 1.2), respectively.

For mitochondrial staining, HCT-116^p53+/+ and HCT-116^p53−/− cells were incubated with 0.3 μM MitoTracker Red CMXRos (Invitrogen, Eugene, Oregon, USA) for 30 min at 37 °C in 5% CO₂. Afterwards, cells were stained with 3 μM 4′,6-diamidino-2-phenylindole dihydrochloride (DAPI, Sigma-Aldrich) for 15 min at 37 °C in 5% CO₂. For ROS measurement, HCT-116^p53+/+ and HCT-116^p53−/− cells were incubated with 0.1 μM 2′,7′-dichlorofluorescin diacetate (H₂DCFDA, Sigma-Aldrich) for 30 min at 37 °C in 5% CO₂. Afterwards, cells were stained with 3 μM DAPI for 15 min at 37 °C in 5% CO₂. All assays were performed using a BD FACSAria II flow cytometer (San Jose, CA, USA), and the BD FACSDiva Software version 8.0.2. Data were analyzed with the BD FlowJo version 10 software.

After XCT790 treatment, cell colony formation and cell survival were analyzed using colony-forming assay and staining with crystal violet and formaldehyde solution. Briefly, 40,000 cells were seeded in 6-well cell culture plates with 2 ml fresh growth medium. The next day, cells were treated either with DMSO as a negative control or XCT790 (15 μM). After treatment, cells were grown for another 7 days, and on day 7, cells were stained with crystal violet solution for 1 h and de-stained in plain water.

HCT-116^p53+/+ and HCT-116^p53−/− cells were treated either with DMSO as a negative control or with 15 μM XCT790 for 48 h. Cells were then harvested and incubated with 1x Annexin V binding buffer containing Annexin V-FITC (BD Biosciences, Franklin Lake, NJ, USA) and propidium iodide (PI, Thermo Scientific) for 15 min in the dark at room temperature. Fractions of apoptotic cells were measured using flow cytometry. Also, for accessing the cell cycle progression of HCT-116^p53+/+, HCT-116^p53−/−, and DLD-1 cell lines, cells were harvested and incubated with propidium iodide (PI, Thermo Scientific) and RNase for 30 min in the dark at room temperature to measure cellular DNA content. All analyses were performed on LSRFortessa X-20 flow cytometer (San Jose, CA, USA) and the BD FACSDiva Software version 8.0.2. Data were analyzed with the BD FlowJo version 10 software.

Values are presented as mean values ± S.E.M. All experiments were conducted at least 2 to 4 times with representative data shown. Comparisons among values for all groups were determined by Student’s t test using the GraphPad Prism 8.0 software (San Diego, CA, USA). Differences between samples with a p value of ≤ 0.05 were considered statistically significant. For the analyses of proteomics, comparisons between groups were made using multiple t tests with a false discovery rate of 0.05.