The colon cancer subset of The Cancer Genome Atlas (TCGA) PanCancer Atlas data on CRC was examined in patients with +1 and –1 standard deviation (SD) of mean expression of NFKB1 and TRAP1 [21]. CBioportal was used to examine the clinical profile of the patients [22, 23].

The colon cancer cell line HCT116 was a gift from the laboratory of Dr. Noah Shroyer at Baylor College of Medicine, which was originally purchased from the American Type Culture Collection (Manassas, VA, USA) under the catalog number CCL-247 without Mycoplasma contamination. The cell line was authenticated by the American Type Culture Collection (ATCC®), using seventeen short tandem repeat (STR) loci plus the gender determining locus, Amelogenin, with the commercially available PowerPlex® 18D Kit from Promega (Manassas, VA, USA). The cells were cultured in DMEM, 2 mM L-glutamine, 10% (v/v) fetal bovine serum (FBS), 100 U/mL penicillin, and 100 µg/mL streptomycin in a 37 °C incubator at 5% CO${}_2$. The cell culture reagents were purchased from Fisher Scientific (Hampton, NH, USA). The TRAP1 inhibitor gamitrinib-triphenylphosphonium (GTPP) was purchased from MedChem Express (HY-102007A) (Monmouth Junction, NJ, USA). Phospho-Histone H2AX ($\gamma$H2AX), Thioredoxin reductases 2 (TRXR2/TXNRD2) and $\beta$-actin antibodies were purchased from Cell Signaling Technology (Danvers, MA, USA). For all experiments, HCT116 cells were treated with 10 µM H${}_2$O${}_2$, 5 µM GTPP or both 10 µM H${}_2$O${}_2$ and 5 µM GTPP with control samples as untreated cells for a period of 24 hours, unless otherwise specified.

Cell death was estimated using the Trypan Blue exclusion assay. Briefly, HCT116 cells were seeded in 6-well plates and subjected to different treatment conditions for 24 hours. Subsequently, an aliquot of the cells was mixed with 0.2% trypan blue at a ratio of 1:1 by volume. The cell suspension was counted using a hemocytometer chamber under a light microscope.

The viability of HCT116 cells was examined using the Cell Proliferation assay (MTT) with reagents purchased from Roche Life Science as per the manufacturer’s guidelines (Basel, Switzerland). A total of 5000 HCT116 cells were plated in the growth media described above. Cells were then treated with a combination of 10 µM H${}_2$O${}_2$ and 0–7 µM GTPP for 24 hours. The experiment was repeated 3 times. Cell viability relative to control was plotted using GraphPad Prism from GraphPad Software (version 8, Boston, MA, USA) [24].

Cells were lysed in Mammalian Protein Extraction Reagent (MPER) buffer with 1$\times$ Halt Protease inhibitor cocktail from Fisher Scientific (Hampton, NH, USA). The protein lysate was resolved using a 4–15% SDS‒PAGE gel from Bio-Rad Laboratories (Hercules, CA, USA) and transferred onto a 0.2 µM polyvinylidene fluoride (PVDF) membrane using the iBlot2 dry blotting system from Fisher Scientific (Hampton, NH, USA) following the manufacturer’s instructions. The membranes were incubated with primary antibodies followed by secondary horseradish peroxidase (HRP)-tagged antibodies and imaged on Gel-Doc EX images from Bio-Rad Laboratories (Hercules, CA, USA).

Approximately 1 $\times$ 10${}^5$ cells (under different treatment conditions) were plated on coverslips and washed with 1$\times$ PBS three times. The cells were fixed with 4% formaldehyde for 10 minutes at room temperature and washed with 1$\times$ PBS three times. The cells were permeabilized with 0.5% Triton X-100 in PBS at room temperature for 5 minutes and washed with 1$\times$ PBS three times. The cells were then blocked with 5% normal goat serum in 1$\times$ PBS for 1 hour. After washing with PBS, the cells were incubated with the primary antibody ($\gamma$H2AX, 1:100) overnight at 4 °C. The next day, following three washes with 1$\times$ PBS, the cells were incubated with Alexa Fluor 488-conjugated goat anti-rabbit IgG secondary antibody at a concentration of 1:1000 for 1 hour at room temperature in the dark. The cells were washed with 1$\times$ PBS three times and stained with DAPI from Fisher Scientific (Hampton, NH, USA) at a concentration of 1:1000 at room temperature for 3 minutes. After the final three washes with 1$\times$ PBS, the coverslips were mounted on slides using ProLong™ Gold Antifade Mountant from Fisher Scientific (Hampton, NH, USA). The slides were analyzed using a Nikon florescent microscope, and quantification of foci was performed using ImageJ (version 2.14, https://imagej.net/software/fiji/) [25].

The cells were lysed in MPER buffer (+1X Halt Protease inhibitor cocktail), and the cell debris was pelleted by centrifugation at 14,000 $\times$g for 10 minutes at 4 °C. For each sample, the supernatant was collected, and 100 µg of protein was processed for proteomic analysis as previously described [26]. Briefly, the protein lysates were reduced with 10 mM Dithiothreitol (DTT) at 50 °C for 1 hour, followed by alkylation using 25 mM iodoacetamide at room temperature for 30 minutes in the dark. The proteins were precipitated using a 1/4th volume of 100% (w/v) trichloroacetic acid (TCA) for 30 minutes on ice. The samples were centrifuged at 14,000 $\times$g for 10 minutes at 4 °C. The pellets were rinsed twice with ice-cold acetone followed by centrifugation at 14,000 $\times$g for 10 minutes. The protein pellets were air dried and resuspended in 50 mM ammonium bicarbonate. Trypsin was added at a 1:30 ratio in two steps with a 2-hour incubation in between with vortexing every 30 minutes. The samples were finally incubated overnight at 37 °C (approximately 18 hours). The samples were centrifuged briefly, and the peptide concentration was measured using the BCA Gold assay. The samples were dried completely in Savant™ SpeedVac™ (ThermoFisher Scientific, Waltham, MA, USA) and resuspended at 1 mg/mL in 0.1% formic acid for MS analysis.

One microgram of each digested sample was analyzed with a Q Exactive™ HF-X Orbitrap ™ mass spectrometer (ThermoFisher Scientific, Waltham, MA, USA) interfaced with an UltiMate 3000 HPLC (ThermoFisher Scientific) as described previously [27]. The sample was first loaded into a 5-mm trap column packed with 5 µM/100 Å C18 material (Thermo Fisher Scientific) using 98% buffer A (0.1% formic acid in water) and 2% buffer B (0.1% formic acid in acetonitrile) at a flow rate of 3 µL/min. The peptides were separated in a 25 cm analytical column packed with 5 µM/18 Å C18 material using a 90-minute linear gradient ramping from 2% to 35% buffer B versus buffer A at a flow rate of 0.35 µL/min. Mass spectrometric analysis was performed using data-dependent acquisition (DDA) mode. The survey scan was performed with 60K resolution from 400 to 1600 m/z with an automatic gain control (AGC) target of 3e6 and a max injection time of 50 msec. Monoisotopic masses were then selected for further fragmentation for the 25 most abundant precursor ions with 2 to 4 plus charges. Precursor ions were isolated using the quadrupole with an isolation window of 1.6 m/z. Higher energy collisional dissociation (HCD) was applied with a normalized collision energy of 28%. Tandem mass spectrometry (MS/MS) scans were carried out with a resolution of 7500. The AGC target for MS/MS was set to 5e5, and the maximum injection time was limited to 22 msec.

The MS data were searched against the UniProt human protein database for peptide/protein identification using the Comet algorithm embedded in the Trans-Proteomic Pipeline [28, 29]. Carbamidomethylation of cysteine was set as a fixed modification, and oxidation of methionine and deamidation of asparagine were set as variable modifications. The peptide assignment was validated with PeptideProphet, and a probability score $\ge$0.9 in correspondence with an error rate of 0.01 was applied to filter the peptides. Skyline software (version 21.1.0.146, https://skyline.ms/) was used for quantitative analysis of the DDA data [30, 31]. The spectral library-based platform described in previous studies was used for quantitative analysis [26, 32, 33]. The composite spectral library was built using all of the DDA data collected from the samples analyzed. Quantification was performed at the MS1 level using the sum of the first 3 monoisotopic peaks. The abundance of each peptide was normalized to the total ion current (TIC) and presented as ion per million (IPM) using the following formula: Normalized Intensity (IPM) = Peptide Intensity/TIC $\times$ 1000000. Protein quantification was achieved by summation of the normalized intensities of the corresponding peptides.

The log-rank test was used to compare the survival of colon cancer patients in the TCGA dataset. A p value of 0.05 was considered significant. To compare the occurrence of new neoplasms post initial therapy, a chi-squared test was performed, and a p value 0.05 was considered significant. Differential expression analysis of TCGA data was performed on cBioportal, keeping a threshold of 10% false discovery rate (FDR) [23]. Gene set enrichment analysis (GSEA) was performed using the WebGestalt online tool [34, 35]. An FDR cutoff of 5% was applied to GSEA.

The protein ratios between the treatment groups and the control group were analyzed by calculating log2FoldChange. Differential protein expression was calculated using the limma package in R (version 3.17, https://www.bioconductor.org/) [36]. The differentially expressed genes were filtered by an FDR of 0.15. Hierarchal clustering, sample correlation and principal component analysis were performed using the online bioinformatics platform idep.96 [37]. Cluster enrichment was performed using Metascape [38].

To examine the impact of TRAP1 expression on colon cancer, we analyzed the mRNA expression of TRAP1 and its association with progression-free survival and disease-specific survival using TCGA data. First, we used the mRNA expression of NFKB1 to represent the oxidative stress level of a patient. Patients with NFKB1 expression 1 SD higher or lower than the mean value were designated as the patients with high oxidative stress or low oxidative stress, respectively. Accordingly, patients with a TRAP1 expression more than 1 SD higher than the average were classified as high TRAP1, and those below 1 SD were classified as low TRAP1. Under low oxidative stress conditions, patients with either low or high TRAP1 expression did not show a significant difference in either progression-free or disease-specific survival (Fig. 1A,B). Notably, under high oxidative stress conditions, however, the patients with high TRAP1 expression had a significantly shorter survival time for both progression-free and disease-specific survival in comparison to those with low TRAP1 levels (p 0.05, Fig. 1A,B).

Fig. 1.

Tumor necrosis factor receptor-associated protein 1 (TRAP1) expression under oxidative stress is associated with disease outcomes in The Cancer Genome Atlas (TCGA) colorectal cancer data. (A) Progression-free survival. (B) Disease-specific survival in patients with high and low expression of TRAP1 under high and low oxidative stress. (C) Percentage of patients who develop a new neoplasm postinitial therapy. (D) Volcano plot, (E) Gene set enrichment analysis of differentially expressed genes in patients with high and low TRAP1 expression under high oxidative stress (+1 SD NFKB1 expression).

Additionally, for patients with high oxidative stress, the group with high TRAP1 expression had a higher risk for developing new neoplasms postinitial therapy than the group with low TRAP1 expression (p 0.0001) (Fig. 1C). In contrast, patients with low oxidative stress showed no significant difference between low and high TRAP1 expression in the risk of developing a new neoplasm post initial therapy (Fig. 1C).

Our analysis of differential mRNA expression between TRAP1 high- and TRAP1 low-expressing tumors under high oxidative stress showed 974 differentially expressed genes, as visualized by the volcano plot in Fig. 1D. GSEA-based query using the Reactome pathway database showed an enrichment of pathways associated with RNA splicing, cellular metabolism, G protein coupled receptor signaling and cell division [35, 39] (Fig. 1E). Supplementary Tables 1,2 summarize the list of differentially expressed genes and enrichment results, respectively. Altogether, the gene expression analysis of the TCGA CRC dataset suggested that TRAP1 expression was associated with patient survival when the tumors were under high oxidative stress. This prompted us to further explore the implication of TRAP1 for CRC therapeutic benefits in the context of high oxidative stress at the functional level using a proteomic approach.

To examine the proteome alterations induced by oxidative stress and/or modulated by TRAP1, colon cancer HCT116 cells were treated with H${}_2$O${}_2$ to mimic oxidative stress in colon cancer, with or without TRAP1 inhibition using GTPP. The H${}_2$O${}_2$ concentration reflecting a state of oxidative stress was used [40]. A cell viability assay with 0–7 µM GTPP in combination with 10 µM H${}_2$O${}_2$ was performed to determine the appropriate dose for treatment. The concentration of H${}_2$O${}_2$ was selected based on the H${}_2$O${}_2$ dose response [13]. Supplementary Fig. 1 shows the dose response curve after 24 hours of treatment. A dose of 5 µM GTPP in combination with 10 µM H${}_2$O${}_2$ showed an average of 57.66% cell viability. This dose was selected to treat the cells for proteomic analysis.

Using the spectral library-based platform, a total of 3240 proteins were identified (FDR 0.01) and quantified in these samples. The adjusted p value and fold change for each protein in the three treatment groups, including H${}_2$O${}_2$ only, GTPP only and H${}_2$O${}_2$+GTPP, compared to the control group are presented in Supplementary Table 3. The top 1000 most variable proteins were visualized as a heatmap of z score across the four groups (Fig. 2A). Samples were clustered together with respect to their treatment groups. Additionally, samples were further clustered with respect to GTPP or no-GTPP treatments. Principal component analysis (PCA) confirmed the results from hierarchal clustering (Fig. 2B). PCA also demonstrated that GTPP treatment drove 18% of the variance (PC1) in proteomic profiles, followed by H${}_2$O${}_2$ treatment, which drove 14% of the variance (PC2) (Fig. 2B). Gene ontology enrichment of the pathways in each principal component showed enriched pathways related to negative regulation of metabolic process and negative regulation of biosynthetic process in PC1 (Fig. 2C). PC2 showed an enrichment of DNA-templated RNA transcription and regulation of biosynthetic processes (Fig. 2C).

Fig. 2.

Proteomic profiling of TRAP1 inhibition in colon cancer cells under oxidative stress. (A) Heatmap showing the Z score of top 1000 most variable proteins. (B) Principal component analysis of proteomic expression in control or after treatment with combination of GTPP+H${}_2$O${}_2$ or H${}_2$O${}_2$ or GTPP alone. (C) Gene ontology of PCs driving the response.

The protein ratios between different treatment groups were calculated using the limma package and are presented in Fig. 3A–C as volcano plots. The differential proteins in each treatment group compared to the control were visualized as a bar graph separating up- and down-regulated proteins (Fig. 3D). The inhibition of TRAP1 with GTPP appeared to substantially disrupt the proteome of HCT116 cancer cells, resulting in the highest number of differentially expressed proteins. These data demonstrate that GTPP is the major driver of protein expression changes in cells after treatment.

Fig. 3.

Differential protein expression after TRAP1 inhibition under oxidative stress in colon cancer cells. (A–C) Volcano plots of protein ratios in comparison to the control group for GTPP+H${}_2$O${}_2$ (A), GTPP (B), and H${}_2$O${}_2$ (C). (D) Differentially upregulated and downregulated proteins with different treatments.

To gain insight into the role of TRAP1 in colon cancer under oxidative stress, we performed enrichment analysis of the differentially expressed proteins using Metascape [38]. The top 20 most enriched clusters for the three treatment groups are presented in Fig. 4A–C. Exposure to H${}_2$O${}_2$ (oxidative stress condition) induced clusters related to response to oxidative stress, transcription, nucleo-cytoplasmic transport, and metabolism of small molecules (Fig. 4A). Two of these enrichment clusters, neutrophil degranulation and cellular response to stress, were shared by all three treatments, including H${}_2$O${}_2$ alone, GTPP alone or GTPP+H${}_2$O${}_2$ (Fig. 4D). Cells exposed to GTPP alone displayed an enrichment of pathways related to RNA metabolism, endocytosis, proteolysis, and metabolic reprogramming of colon cancer (Fig. 4B). Metabolic reprogramming of colon cancer was also an enriched cluster in cells treated with the combination of GTPP and H${}_2$O${}_2$ (Fig. 4C). In addition, cells treated with GTPP+H${}_2$O${}_2$ also displayed enrichment of the heat shock factor 1 (HSF1) pathway (Fig. 4D). The HSF1 gene plays an important role in several cellular processes important for cancer development. It has been reported that HSF1 expression is associated with sporadic colon cancer [41]. The enrichment analysis of the differential proteins suggests that TRAP1 inhibition under oxidative stress alters the protein expression in cellular metabolism and heat-shock response pathways, prompting us to further investigate these two pathways.

Fig. 4.

Functional enrichment of differentially expressed proteins. Enriched clusters were examined in differentially expressed proteins. Comparisons are as follows H${}_2$O${}_2$ vs. Control (A), GTPP vs. Control (B), GTPP+H${}_2$O${}_2$ vs. Control (C). Venn diagram of primary overlapping and exclusive clusters in the three comparisons (D).

Previous reports examining the role of TRAP1 in cancer have indicated a role of TRAP1 in the induction of aerobic glycolysis [42, 43]. Our proteomic data also showed an enrichment of metabolic reprogramming in the colon cancer pathway. To enable further inquiry into the direction of the shift in metabolism, we examined the protein expression of genes regulating cellular energetics identified by the enrichment analysis. The shift from oxidative phosphorylation to aerobic glycolysis was evident based on the induction of glycolytic genes in the cells undergoing TRAP1 inhibition (Fig. 5A). Fructose-bisphosphate aldolase B (ALDOB) catalyzes the hydrolysis of fructose 1,6 biphosphate (FBP) to glyceraldehyde 3 phosphate and dihydroxyacetone in glycolysis. Our study showed a significant increase in the expression of ALDOB in the cells treated with GTPP+H${}_2$O${}_2$ compared to the control (p 0.001) and cells treated with H${}_2$O${}_2$ alone (p 0.01) (Fig. 5A). Glutaminase 1 (GLS1) is a glutamine metabolism enzyme that catalyzes the conversion of glutamine to glutamate [44]. TRAP1 inhibition under oxidative stress conditions resulted in repression of GLS1 expression (Fig. 5A). HCT116 cells treated with GTPP+H${}_2$O${}_2$ showed a significantly stronger repression of GLS1 expression (p 0.001) relative to the control and the cells treated with H${}_2$O${}_2$ alone (p 0.05). Lactate dehydrogenase (LDHA) is a critical glycolytic enzyme that catalyzes the conversion of pyruvate to lactate, while reduced nicotinamide adenine dinucleotide (NADH) is oxidized to nicotinamide adenine dinucleotide (NAD). We observed significantly higher expression of LDHA in cells with TRAP1 inhibition under oxidative stress (Fig. 5A). LDHA expression was significantly induced in cells treated with GTPP+H${}_2$O${}_2$ compared to the cells treated with H${}_2$O${}_2$ (p 0.01) or control (p 0.01) (Fig. 5A), indicating a shift to glycolytic conditions in TRAP1-inhibited cells under oxidative stress. Pyrroline-5- carboxylate reductase 1 (PYCR1) catalyzes the conversion of pyyroline-5-carboxylate to proline while oxidizing NADH to NAD [45]. It enhances the TCA cycle under hypoxic conditions [45]. PYCR1 expression upon TRAP1 inhibition under oxidative stress is also indicative of an enhanced shift to glycolysis. Our data show that TRAP1 inhibition with (p = 0.06) and without (p 0.01) oxidative stress induces PYCR1 expression when compared to untreated cells (Fig. 5A). The SUCLG2 gene encodes succinate dehydrogenase (SDH). This enzyme catalyzes the conversion of succinate to fumarate. Under inflammatory stress, the SDH enzyme has been shown to drive the metabolic shift from oxidative phosphorylation to aerobic glycolysis [46]. Our data showed a significant induction of SUCLG2 expression under TRAP1-inhibited conditions relative to the untreated cells both with (p 0.05) and without oxidative stress (p 0.01) (Fig. 5A). Transaldolase 1 (TALDO1) is an enzyme in the pentose phosphate pathway that catalyzes the conversion of sedheptulose-7-phosphate (S7P) and GAP into erythrose-4-phosphate and fructose 6 phosphate [47, 48]. The repression of TALDO1 under TRAP1 inhibition was observed both with (p 0.01) and without oxidative stress (p 0.01). Transketolase (TKT) is an enzyme in the pentose phosphate pathway (PPP) [49]. Its activity enables the generation of sugar phosphates for glycolysis [49]. We observed an increase in TKT expression under oxidative stress in TRAP1-inhibited cells (p 0.01). Together, these results suggest that inhibition of TRAP1 under oxidative stress conditions induces a shift to aerobic glycolysis in colon cancer cells.

Fig. 5.

TRAP1 inhibition and oxidative stress induced metabolic reprogramming in colon cancer cells and activation of the heat shock response. TRAP1 inhibition under oxidative stress altered the expression of proteins regulating cellular metabolism (A) and altered the expression of proteins regulating heat shock response pathway (B).

HSF1 is a transcription factor conventionally known to respond to cellular stress [50] and is central to NAD+ metabolism [51]. Our data showed an enrichment of proteins in the HSF1 transactivation pathway (Fig. 4D) with TRAP1 inhibition and oxidative stress. Calcium/calmodulin-dependent protein kinase 2 (CAMK2) is a serine threonine kinase reported to increase the phosphorylation and transactivation of HSF1 [52]. The CAMK2 proteins CAMK2A, CAMK2B and CAMK2G showed an increasing trend after treatment with GTPP with and without oxidative stress (Fig. 5B). HSPA1B is a heat shock response protein belonging to the HSP70 family of chaperone proteins associated with poor prognosis in colon cancer [53]. TRAP1 inhibition under oxidative stress showed a significantly higher expression of HSPA1B compared to the cells under oxidative stress alone or TRAP1 inhibition alone (p 0.01) (Fig. 5B). HSPA1L is also a HSP70 family of heat shock proteins associated with poor prognosis in colon cancer [53]. TRAP1 inhibition under oxidative stress showed a trend of induction of HSPA1L protein when compared to untreated cells (p = 0.05). Prostaglandin E synthase 3 (PTGES3) is an oncogene overexpressed in colon cancer [54]. TRAP1 inhibition under oxidative stress significantly repressed PTGES3 expression in cells under oxidative stress (p 0.01). Together, our results show alterations in several stress proteins downstream of the HSF1 transcription factor under treatment with GTPP+H${}_2$O${}_2$.

Induction of apoptosis after GTPP treatment due to an increase in ROS and the release of cytochrome C in colon cancer cells has been reported in several studies. Proteomic data from our study showed a significant increase in the expression of DNA damage and the apoptotic gene SMC2 (Fig. 6A). HCT116 cells treated with GTPP+H${}_2$O${}_2$ showed higher SMC2 expression than control cells. SMC2 expression was also significantly higher in cells treated with GTPP+H${}_2$O${}_2$ than in cells treated with H${}_2$O${}_2$ or GTPP alone. To obtain a quantitative estimate of the impact of TRAP1 inhibition under oxidative stress, cell death and DNA damage were evaluated after treatment. Cell death was increased almost 4-fold in cells treated with GTPP+H${}_2$O${}_2$ compared with cells treated with H${}_2$O${}_2$ (p 0.001) or GTPP (p 0.001) alone (Fig. 6B). Induction of DNA damage as indicated by double strand DNA breaks was estimated by $\gamma$H2AX staining (Fig. 6C–D) and western blotting (Fig. 6E). Our data show a significant increase in double-strand DNA breaks after treatment with H${}_2$O${}_2$ or GTPP compared to control cells. Combination treatment with H${}_2$O${}_2$ and GTPP resulted in the highest degree of DNA damage when compared to no treatment or single treatment of H${}_2$O${}_2$ or GTPP alone.

Fig. 6.

TRAP1 inhibition induces cell death and DNA damage under oxidative stress. (A) SMC2 expression was significantly increased in HCT116 cells treated with GTPP+H${}_2$O${}_2$ compared to control, or cells treated with 10 µM H${}_2$O${}_2$ alone (p 0.05). (B) TRAP1 inhibition under oxidative stress in colon cancer cells increases the cell death by more than 4-fold. Treatment with the combination of GTPP+H${}_2$O${}_2$ induced significantly more cell death compared to treatment with H${}_2$O${}_2$ (p 0.001) or GTPP (p 0.001) alone. (C) $\gamma$H2AX expression in HCT116 cells was examined after 24 hours of treatment. Representative images are of nuclear stain in blue (DAPI) and $\gamma$H2AX in green are shown. (D) Percent positive $\gamma$H2AX were compared across using one way ANOVA followed by Tukey post hoc analysis. (E) Western-blot confirmed the expression of $\gamma$H2AX following the treatment with TRAP1 inhibitor under oxidative conditions.