FAS Inhibited Proteomics and Phosphoproteomics Profiling of Colorectal Cancer Spheroids Shows Activation of Ferroptotic Death Mechanism

Abstract

Colorectal cancer (CRC) is projected to become the third most diagnosed and third most fatal cancer in the United States by 2024, with early onset CRC on the rise. Research is constantly underway to discover novel therapeutics for the treatment of various cancers to improve patient outcomes and survival. Fatty acid synthase (FAS) has become a druggable target of interest for the treatment of many different cancers. One such inhibitor, TVB-2640, has gained popularity for its high specificity for FAS and has entered a phase 1 clinical trial for the treatment of solid tumors. However, the distinct molecular differences that occur upon inhibition of FAS have yet to be understood. Here, we conduct proteomics and phosphoproteomics analyses on HCT 116 and HT-29 CRC spheroids inhibited with either a generation 1 (cerulenin) or generation 2 (TVB-2640) FAS inhibitor. Proteins involved in lipid metabolism and cellular respiration were altered in abundance. It was also observed that proteins involved in ferroptosis—an iron mediated form of cell death—were altered. These results show that HT-29 spheroids exposed to cerulenin or TVB-2640 are undergoing a ferroptotic death mechanism. The data were deposited to the ProteomeXchange Consortium via the PRIDE repository with the identifier PXD050987.

Graphical Abstract

INTRODUCTION

Colorectal cancer (CRC) is projected to be the third most diagnosed and third most fatal cancer in the United States in 2024, with an estimated total of 152,810 new diagnoses and 53,010 total deaths.¹ Early onset colorectal cancer is also on the rise, with projections showing CRC to be the leading cause of cancer mortality in young people between the ages of 20 and 49 in the US by as early as 2030.^2,3 Research into new therapeutics and druggable targets is increasing to meet the increasing number of CRC diagnoses. It has been observed that many cancers synthesize their lipids de novo at a similar rate to the liver rather than uptake them through the diet to maintain their rapid growth and proliferation.⁴ It was identified in 1984 that Fatty Acid Synthase (FAS) was elevated in breast tumors and later was shown to be very important for cancer growth and survival.^5,6 FAS is the primary enzyme for the synthesis of the 16-chain saturated fatty acid (FA) palmitate via condensations of acetyl and malonyl CoA. Once palmitate is synthesized by FAS, it then gets added to the cellular fatty acid pool where it can then be desaturated and elongated to meet the demands of the cell. Recently, These discoveries have made FAS a druggable target of interest for treating many cancers.^6,7 The first FAS inhibitor, cerulenin, showed promise for being a treatment for obesity in the late 1990s but lost its popularity once it was observed to cause drastic weight loss in mice after only 6 days of exposure.^8,9 Additional FAS inhibitors have been developed with an increasing specificity for FAS. TVB-2640 is a newly developed FAS therapeutic shown to have increased specificity for FAS. A recent phase 1 study was conducted in humans with TVB-2640 for the treatment of metastatic solid tumors.¹⁰ The researchers observed lowered levels of triglyceride lipids in patient serum as well as a correlation for a longer time to disease progression with mutant KRAS tumors compared to wild-type KRAS tumors in participants given TVB-2640 monotherapy.¹⁰

FAS inhibition has been studied extensively using cell culture models. A study from Chang et al. showed that HT-29 CRC cells inhibited with cerulenin had lowered levels of PI3K, phosphorylated Akt and mTOR.¹¹ It has also been shown by Shiragami et al. that combining FASN inhibitors with common chemotherapeutics, such as oxaliplatin, can increase the antitumor effect of these therapeutics and slow tumor progression.¹² Recently, Barpanda et al. showed that cerulenin in HCT 116 cells activates the proteasome machinery and induces apoptosis in the cells.¹³ The work mentioned above has expanded our understanding of FASN inhibition in monolayer cell culture; however, monolayer cell culture models do not recapitulate tumor microenvironment (TME) found within an in vivo human tumor. A three-dimensional cell culture model, such as a cancer spheroid, can more accurately recapitulate the TME by developing three distinct cellular regions—proliferative, quiescent, and necrotic region—from the development of chemical and nutritional gradients.^14–16

Our research group recently characterized the full drug penetration and subsequent changes to the lipidome of HCT 116—a KRAS mutant, microsatellite instable (MSI) colorectal carcinoma cell line—and HT-29—a P53 mutant, microsatellite stable (MSS) colorectal adenocarcinoma cell line—spheroids inhibited with either cerulenin or TVB-2640.¹⁷ One of the notable observations was an increase in monounsaturated fatty acids (MUFAs) and polyunsaturated fatty acids (PUFAs) among many lipid classes and subclasses within TVB-2640 treated spheroids. The increase in MUFA and PUFA lipids suggested that the different generations of FAS inhibitors cause cells to undergo a ferroptotic death mechanism rather than an apoptotic death mechanism. Ferroptosis can be activated via many different mechanisms, including, but not limited to, activation of various phosphorylation signaling networks.¹⁸ AMP-activated protein kinases (AMPK) can play many roles in activating or inhibiting ferroptosis. Phosphorylation of acetyl-CoA carboxylate via AMPK can block the formation of lipid peroxides and prevent ferroptosis, while it has also been shown that AMPK phosphorylation of Beclin 1 can cause erastin-induced ferroptosis.^19,20 It has also been shown that inhibition of the PI3K-Akt-mTOR pathway can make cells more sensitive to ferroptosis.²¹

In this work, we have complemented our previous lipidomics study of HCT 116 and HT-29 spheroids inhibited with either cerulenin or TVB-2640 using the same drug dosing protocols as those in the previous work. In this study, we then conducted time course proteomics and phosphoproteomics to discern distinct protein and phosphorylation differences among these two generations of FAS inhibitors.

METHODS

Chemicals

LC-MS grade water, acetonitrile (ACN), 0.1% formic acid in water, and 0.1% formic acid in acetonitrile were obtained from Burdick & Jackson. Dimethyl Sulfoxide (DMSO) was obtained from MP Biomedicals. Sodium dodecyl sulfate (SDS), sodium orthovanadate, sodium fluoride, sodium pyrophosphate, β-glycerophosphate, trifluoroacetic acid (TFA), glycolic acid (GA), and triethylammonium bicarbonate (TEAB, 1 M, pH = 8.5) were obtained from Sigma-Aldrich. Methanol was obtained from Fisher Scientific. Cerulenin was obtained from Cayman Chemicals. TVB-2640 was obtained from Selleckchem. Trypsin/Lys-C was obtained from Promega.

Cell Culture

Both HCT 116 and HT-29 cells were purchased from American Type Cell Culture (ATCC). Cells were grown in a monolayer until 80% confluence in McCoy’s 5A media supplemented with 10% FBS, 1% l-glutamine, and 1%-penicillin/streptomycin at 37 °C with 5% CO₂. Spheroid formation was followed using previously established protocols.²² In brief, a 1.5% agarose solution was pipetted into the wells of a 96-well plate and allowed to cool. A 200 μL volume of the cells in media were seeded in each well at 7,000 cells/well. Spheroids were allowed to grow untouched for 4 days. At day 4 and every other day after, half of the media volume was replaced with fresh media until day 12. On day 12, HCT 116 and HT-29 spheroids were dosed with either cerulenin—87.3 ± 25.2 μM and 79.9 ± 4.1 μM, respectively—or TVB-2640—53.4 ± 3.9 μM and 131.1 ± 13.1 μM, respectively—at their respective half-maximum inhibitory concentration (IC₅₀) according to previous studies.¹⁷ Four time points were harvested in biological triplicate, with each time point consisting of 12-whole spheroids: 0-h (control), 6-h, 24-h, and 48-h. Harvested spheroids were washed with 1× PBS three times and stored at −80 °C.

Proteomics Sample Preparation

Spheroids were lysed in 200 μL of mammalian cell lysis buffer (6% SDS, 100 mM TEAB, 1 mM NaF, 1 mM NaVO₄, 1 mM β-glycerophosphate, 10 mM Na₄P₂O₇, LC-MS grade H₂O, and 1 comPLete protease inhibitor tablet) via probe sonication. The cell lysate was clarified by centrifugation at 12,0000 rpm for 10 min, and the supernatant was saved for further analysis. Protein quantification was conducted via the protein bicinchonic acid assay (BCA) and 400 μg of protein was aliquoted for proteomics preparation. Proteins were prepared into peptides using the S-Trap protocol. Proteins were reduced with 20 mM DTT and denatured by heating to 65 °C for 15 min. Reduced and denatured proteins were alkylated with 40 mM IAA for 30 min in the dark. Alkylated proteins were acidified to 0.9% TFA according to a previous report.²³ Acidified proteins were then diluted with 90% methanol and 100 mM TEAB and spun onto the S-Trap mini column via centrifugation at 4,000 rpm for 1 min. Trapped proteins were washed four times with 90% methanol and 100 mM TEAB followed by a final wash with no liquid to remove any excess methanol. S-Trap columns were moved to a clean tube, and Trypsin/Lys-C (Promega) was added in a 1:25 trypsin: protein (w/w) ratio in 50 mM TEAB and incubated overnight in a water bath at 37 °C. Peptides were eluted sequentially with equal volumes of 50 mM TEAB, 0.2% formic acid in H₂O, and 0.2% formic acid in 50% ACN. Approximated 75 μg of peptides from each sample was saved for proteomics sample analysis, and the remaining peptides were saved for phosphopeptide enrichment. Both sets of samples were dried down via vacuum centrifugation and stored at −80 °C.

Phosphopeptide Enrichment

Dried peptides were resuspended in IMAC binding buffer (80% ACN, 5% TFA, 0.1 M GA).²⁴ Ti-IMAC magnetic microparticles (ReSyn Biosciences) were aliquoted into a 1:4 peptide:bead (w/w) ratio and equilibrated with IMAC binding buffer three times. Peptides were mixed with Ti-IMAC particles and allowed to incubate for 25 min with light mixing. After phosphopeptide binding, Ti-IMAC particles were separated via a magnetic separator, and the nonphosphopeptides were discarded. Bound peptides were washed once with IMAC binding buffer to remove nonspecifically bound peptides. Phosphopeptides were further washed with Wash Buffer 1 (60% ACN, 1% TFA, and 200 mM NaCl) and Wash Buffer 2 (60% ACN, 1% TFA) and the wash fractions were discarded. Phosphopeptides were eluted with 1% NH₄OH for 10 min with light mixing. Eluted phosphopeptides were acidified with 10% formic acid. The elution step was repeated once. Acidified phosphopeptides were dried via vacuum centrifugation and stored at −80 °C.

LC-MS

Both proteomics and phosphoproteomics samples were desalted using a Waters HLB μElution plate (2 mg of sorbent, 30 μm particle size, 80 Å pore size, PN 186001828BA). An equal amount of peptides was taken from each proteomics samples and mixed together to be a pooled sample for chromatogram library generation.²⁵ Proteomics and phosphoproteomics samples were separated using a Waters M-Class Acuity UHPLC using a Waters nanoEase M/Z Peptide BEH C18 column (75 μm i.d. × 200 mm length) at 50 °C with a flow rate of 0.400 μL/min W using the following gradient: 2% mobile phase B (0.1% formic acid in Acetonitrile) to 20% B in 100 min. The percent composition of B was increased to 32% in 20 min followed by a one min ramp to 95% B and held at 95% for 4 min. Next, the percent B composition was ramped down to 2% in one min and held at 2% for 30 min to allow column equilibration. Proteomics samples were collected with data-independent acquisition (DIA) on a Thermo QE-HF Orbitrap using staggered windows of 16 m/z width from a range of 400–1000 m/z. Gas phase fractions (GPF) were collected using DIA with 4 m/z wide staggered windows in 100 m/z ranges from 400 to 1000 m/z to build a chromatogram library. Phosphoproteomics samples were collected using DIA with 25 m/z wide staggered windows ranging from 400 to 1200 m/z.

Data Analysis

Samples were converted from .raw to .mzML with MSConvert with the demultiplex and peakpicking filters.²⁶ The GPF were searched against a Prosit library and combined together to make an empirically corrected GPF library (EMP-GPF) in encyclopeDIA with the Human FASTA as the background.^25,27 Proteomics samples were searched against the EMP-GPF library. Phosphoproteomics samples were searched using Thesaurus with the Phosphopedia spectral library generated from Lawrence et al.^28,29 Both the proteomics and phosphoproteomics .elib files were input into Skyline, filtered, and annotated. An MSstats .csv was exported from Skyline and the proteomics and phosphoproteomics data was further analyzed using MSstatsPTM.³⁰ Volcano plots were made with enhancedvolcano. Gene Set Enrichment Analysis (GSEA) was performed using clusterprofileR.^31,32 Kinase enrichment analysis 2 was used to analyze phosphosites.³³ The .raw files and searched .elib files were deposited to the ProteomeXchange Consortium via the PRIDE repository.^34,35 The data were given the identifier PXD050987.

Caspase-3/7 Glo Assay

HCT 116 and HT-29 spheroids were grown for 12 days. On the 12th day, spheroids and 100 μL of media was moved to clear bottom, white walled 96-well plate. Cerulenin or TVB-2640 was added to corresponding wells at the spheroid respective IC₅₀ value. Spheroids were incubated with the therapeutic for 48 h. Caspase-Glo 3/7 (Promega, Madison, WI) was prepared by mixing provided reagents together and letting reach room temperature. After 48 h, 100 μL of cell culture media was removed from the wells and 100 μL of prepared Caspase Glo-3/7 reagent was added. The spheroids were shaken for 5 min followed by incubation at room temperature for 2 h. The total luminescence was recorded. Base R scripts were used to plot figure and statistical analysis was conducted in excel.

RESULTS

Proteomics

Both HCT 116 and HT-29 spheroids were dosed with either cerulenin or TVB-2640 at their respective IC₅₀ concentrations.¹⁷ A 0 h time point was harvested and used as the control for statistical analysis and differential expression. Spheroids were harvested at 6, 24, and 48 h. All samples were prepared for both proteomics and phosphoproteomic analyses. Proteomics samples were analyzed with DIA. A DIA GPF library was collected with a pooled proteomics sample and searched against an in silico Prosit library to generate an empirically corrected chromatogram library within EncyclopeDIA.^25,27 All proteomics samples were searched with this sample specific library in EncyclopeDIA. After filtering proteomics data in Skyline and MSstatsPTM to include proteins consisting of two or more peptides and with 50% missing values removed, 6,091 proteins were quantified. Quality control analysis showed consistent normalized MS1 TIC levels throughout the experiment (Figure S1). Hierarchical clustering analysis shows the two cell types clustering apart; however, the distinct time points do not cluster around each other (Figure S2). Differential expression analysis was conducted in MSstatsPTM. A log₂FC cut off value of greater than 1 or less than −1 was used along with a Benjamini-Hochberg corrected p-value of less than 0.05 to determine if a protein was considered differentially expressed. A list of identified proteins along with their log₂FC and B–H corrected p-value are found in Supplemental Table S1. The shared differentially expressed proteins from each time point can be found in Supplemental Table S2.

Initial inspection of the HCT 116 cerulenin proteomic samples show two proteins common among all three time points that were also the only two proteins upregulated after 6 h of cerulenin exposure: Heme Oxygenase 1 (HMOX1) and Cyclin-Dependent Kinase Inhibitor 1 (CDNA1). There were 58 common proteins between the 48- and 24-h time points (Figure S3A). Some of these shared proteins are involved in portions of lipid synthesis, such as upregulation of hydroxymethylgluatryl-CoA Synthase, Cytoplasmic (HMCS1), an enzyme involved in cholesterol synthesis, and downregulation of Acyl-CoA dehydrogenase (ACD10). There were also alterations to enzymes involved in metabolism of various amino acids, such as upregulation of asparagine synthetase (ASNS) and downregulation of 2-Oxoisovalerate dehydrogenase subunit alpha, mitochondrial (ODBA), an enzyme that breaks down alpha ketoacids from branched chain amino acids.^36,37 The 24 h time point had 17 proteins unique to this time point (Figure S3A) as well as 39 proteins were up- and 38 proteins downregulated (Figure 1B). 78 proteins were unique to the 48 h time point (Figure S3A), with 42 proteins upregulated and 96 downregulated (Figure 1C). Alterations to enzymes involving lipids were observed after 48 h, such as downregulation of the protein acetyl-CoA acetyltransferase, mitochondrial (THIL)—a mitochondrial enzyme catalyzing the last step of the β-oxidation pathway—and upregulation of glycerol-3-phosphate acyltransferase 3 (GPAT3) and Fatty Acyl-CoA Reductase 1 (FACR1), enzymes involved in the synthesis of lysophosphatidic acid (LPA) and ether lipid synthesis (e.g., PC-O and PE-O), respectively.^38–40 There were also many changes to proteins involved in iron–sulfur (Fe–S) cluster synthesis and maintenance such as frataxin (FRDA), thiosulfate sulfurtransferase (THTR), complex III assembly factor LYRM7, iron–sulfur cluster assembly enzyme (ISCU), and cysteine desulfurase (NFS1).

Figure 1.

Volcano plots of differentially expressed proteins (log₂FC ≥ 1 or ≤ −1 and p-value <0.05) found within HCT 116 spheroids exposed to cerulenin or TVB-2640. (A) 6 h cerulenin, (B) 24 h cerulenin, (C) 48 h cerulenin, (D) 6 h TVB-2640, (E) 24 h TVB-2640, and (F) 48 h TVB-2640.

Gene Set Enrichment Analysis (GSEA) shows many suppressed and activated GSEA terms (Figure 2). HCT 116 spheroids exposed to cerulenin for 24 h had activation of gene sets related to cell death, programmed cell death, and regulation of primary metabolic processes, to name a few. There was suppression of gene sets related to hydrolase activity acting on ester bonds as well as metabolic processes for organic acids, carboxylic acids, oxoacids, and amino acids (Figure 2A). 48 h after cerulenin exposure, many additional terms for cell death were activated, such as regulation of apoptotic process, regulation of programmed cell death, and apoptotic process. There was suppression of catalytic activity gene sets, consisting of between 20 and 25 gene sets, as well as suppression of small molecule, organic acid, and carboxylic acid catabolic processes (Figure 2B).

Figure 2.

GSEA results for HCT 116 spheroids exposed to cerulenin for (A) 24 h or (B) 48 h.

HCT 116 spheroids exposed to TVB-2640 had fewer alterations in protein differential expression ((Figure 1D–F). No proteins were differentially expressed after 6 h (Figure 1D). After 24 h, only six proteins were differentially expressed, and all proteins were downregulated (Figure 1E). 48 h after TVB-2640 exposure, four proteins were upregulated and only 14 proteins were downregulated (Figure 1F). Six differentially expressed proteins were shared between 24 and 48 h, and the remaining 12 proteins were distinct to the 48 h time point (Figure S3B). ASNS was also upregulated within TVB-2640 treated HCT 116 spheroids, as well as CDNA1. Inputting the differentially expressed proteins into GSEA revealed no significantly (p < 0.05) enriched gene sets. Interestingly, endoplasmic reticulum mannosly-oligosaccharide 1,2-alpha-mannosidase (MA1B1) was upregulated after TVB-2640, suggesting alterations to the cellular glycoproteome and glycocalyx, an exciting result that could prompt further investigations.

HT-29 spheroids exposed to cerulenin have many differentially expressed proteins. After 6 h, five proteins were upregulated, and these five proteins were shared among all three time points: Ferritin Light and Heavy chain (FRIL and FRIH, respectively), 3-hydroxy-3-methylglutaryl coenzyme A reductase (HMDH), thrombospondin-1 (TSP1), and growth/differentiation factor 15 (GDF15) (Figure S3C and Figure 3A). 24 h post cerulenin exposure, these numbers increased to 43 up- and 31 downregulated proteins, such as downregulation of THTR, helping form Fe–S clusters (Figure 3B). Twelve proteins were uniquely differentially expressed after 24 h exposure (Figure S3C) Lastly, 48 h after cerulenin exposure, 96 proteins were upregulated and 85 proteins were downregulated (Figure 3C). Of these proteins, 119 total proteins were unique to the 48 h time point (Figure S3C). Caspase-3 (CASP3) was found to be downregulated at this time point, and the ferroptosis suppressor proteins 1 (FPS1) was found to be upregulated. The observation of FSP1 being upregulated at the 48 h time point suggests that the spheroids were beginning to undergo ferroptosis and FSP1 was made in excess to alleviate the oxidative stress on the spheroids. 57 proteins were shared between the 24- and 48-h time points. Tumor necrosis factor receptor superfamily 10B (TR10B) was shared among the two time points and was upregulated, which coordinates the activation of many caspases to begin apoptosis.⁴¹

Figure 3.

Volcano plots of differentially expressed proteins (log₂FC ≥ 1 or ≤ −1 and p-value <0.05) found within HT-29 spheroids exposed to cerulenin or TVB-2640. (A) 6 h cerulenin, (B) 24 h cerulenin, (C) 48 h cerulenin, (D) 6 h TVB-2640, (E) 24 h TVB-2640, and (F) 48 h TVB-2640.

GSEA shows that after 24 h of cerulenin exposure, there is activation of cytoplasm gene sets as well as endomembrane and autophagosome gene sets. There is suppression of amide metabolic process, RNA biding, organonitrogen compound biosynthetic process, as well as mitochondrial gene expression gene sets (Figure 4A). After 48 h of cerulenin exposure, there is activation of inorganic ion homeostasis as well as energy homeostasis. There is suppression of many gene sets related to tRNA activity, such as aminoacyl-tRNA ligase activity, tRNA aminoacylation for protein translation, and tRNA aminoacylation for mitochondrial protein translation (Figure 4B). There were also many gene sets suppressed for mitochondrial related processes, suggesting intensive alterations to cell metabolism, such as the carboxylic acid cycle or β-oxidation.

Figure 4.

GSEA results for HT-29 spheroids exposed to cerulenin for (A) 24 h or (B) 48 h or TVB-2640 for (C) 24 h or (D) 48 h.

HT-29 spheroids exposed to TVB-2640 also had very few proteomic alterations. No significant proteins were found after 6 h of exposure, followed by five up- and seven downregulated proteins after 24 h (Figure 3 D and E). No proteins were uniquely differentially expressed to the 24 h time point and were all shared with the 48 h time point (Figure S3D). Stearoyl-CoA desaturase (SCD) was downregulated at this time point as well as the 48 h. After 48 h of TVB-2640 exposure, there were 33 up- and 57 downregulated proteins (Figure 3F). The GSEA showed only activated gene sets after HT-29 spheroids were exposed to TVB-2640 for 24 h. Activation of cell communication, signal transduction, and signaling gene sets were observed (Figure 4C). 48 h after exposure, many gene sets were activated and suppressed. Activation of the positive regulation of protein phosphorylation was observed as well as regulation of signal transduction and signaling (Figure 4D). There was suppression of response to lipid, oxidoreductase activity, heterocycle, aromatic compound, organic cyclic compound, and nucleobase-containing compound biosynthetic process gene sets (Figure 4D).

There are various pathways for the ferroptosis cascade to originate, with the most understood being inhibition of Glutathione Peroxidase 4 (GPX4) or solute carrier family 7 member 11 (SLC7A11, also referred to as xCT) to increase the amount of peroxidized lipids in the cell. In this data set, the GPX4 protein was identified and quantified; however, SLC7A11 was not identified at the protein level. GPX4 abundances remained consistent across the time points and treatment conditions across both cell lines. Another protein involved in ferroptosis, ferroptosis suppressor protein 1 (FSP1), was able to be identified and quantified in this experiment. FSP1 inhibits ferroptosis within the cell via a glutathione independent pathway, differing from the classic GPX4 ferroptotic pathway.⁴² Varying expression for FSP1 was observed for both HCT 116 and HT-29 spheroids treated with either cerulenin or TVB-2640. HCT 116 spheroids treated with either cerulenin or TVB-2640 had decreased levels of FSP1 over time. HT-29 spheroids, however, had an increasing abundance of FSP1 over time when inhibited with either therapeutic, with FSP1 being differentially expressed in HT-29 spheroids inhibited with cerulenin after 48 h.

Phosphoproteomics

Phosphoproteomic samples were analyzed using DIA and searched with Thesaurus using the Phosphopedia spectral library.^28,29 Only phosphosites with a false localization rate of 0.05 or less were kept for statistical analysis. Phosphoproteomics data was explored in Skyline and analyzed using MSstatsPTM. For differential expression analysis, MSstatsPTM conducts two different hypothesis tests—without protein abundance information and with protein abundance information—to allow for discrepancies in higher protein abundance, leading to higher phosphosite abundance to be discerned. All data reported here have been corrected for protein abundance and represent alterations in the phosphosite abundance.³⁰ A differentially expressed phosphosite was determined if the log₂FC was either ≥1 or ≤ −1 and had a Benjamini-Hochberg adjusted p-value of 0.05 or less. The normalized and log₂ transformed intensities are shown in Figure S4. The top three up- and downregulated phosphosites can be found in Table S1. A list of identified phosphosites along with their log₂FC and B–H corrected p-value are found in Supplemental Table S1. The shared differentially expressed phosphosites from each time point can be found in Supplemental Table S2.

Various phosphosites from the FAS protein were detected and quantified among all the sample conditions. Downregulation of phosphoserine 1174 on FAS was found among almost all of the samples and time points. This residue falls into a region within FAS between the dehydratase (DH) and enoyl reductase (ER) domain. HCT 116 spheroids exposed to cerulenin were observed to have increasing abundances of phosphoserine 275 and 279 over time and were most notably observed to have downregulation of phosphoserine 1174. Surprisingly HCT 116 spheroids exposed to TVB-2640 had an increase in elevation of phosphoserine 1174 from 6- to 24 h. The only differentially expressed phosphosite found in HCT 116 spheroids inhibited with TVB-2640 was phosphoserine 831, becoming upregulated from 6- to 24 h. This residue lies between the acyl and malonyl transferase domains and the DH domain. HT-29 spheroids exposed to cerulenin only had three quantifiable FAS phosphosites: phosphoserine 279, 275—found in the ketoacyl synthase (KS) domain—and 1174. No phosphosites were considered as differentially expressed after 6 h of exposure; however, phosphoserine 275 was observed to become upregulated after 24- and 48 h of exposure. HT-29 spheroids exposed to TVB-2640 were observed to have downregulation of four FAS phosphosites, with phosphoserine 286, also found in the KS domain, and 1174 were found to be statistically significant across all three time points.

HCT 116 spheroids exposed to cerulenin had many alterations at the phosphosite level (Figure 5A–C). Among all the differentially expressed phosphosites, 10 phosphosites were shared among all three-time points Figure S5A. It was observed that after 6 h of cerulenin exposure, 14 phosphosite were up- and downregulated (Figure 5A). Only 12 phosphosites were uniquely differentially expressed in the 6-h time point, one phosphosite was shared between 6 and 24 h, and five phosphosites were shared among 6- and 48 h (Figure S5A). 24 h after cerulenin exposure, the differentially expressed phosphosites rose dramatically to 115 up- and 430- down regulated sites (Figure 5B). 109 phosphosites were unique to the 24 h time point, and 425 were shared with the 48 h time point. Lastly, 48 h after cerulenin exposure was observed to have 97 up- and 1447 downregulated phosphosites (Figure 5C). The remaining differentially expressed phosphosites were unique to this time point, with 1104 phosphosites.

Figure 5.

Volcano plots of differentially expressed phosphosites (log₂FC ≥ 1 or ≤ −1 and p-value <0.05) found within HCT 116 spheroids exposed to cerulenin or TVB-2640. (A) 6 h cerulenin, (B) 24 h cerulenin, (C) 48 h cerulenin, (D) 6 h TVB-2640, (E) 24 h TVB-2640, and (F) 48 h TVB-2640.

HCT 116 spheroids exposed to TVB-2640 had slightly more alterations in their phosphoproteome compared to cerulenin treated spheroids. Six hours after HCT 116 spheroids were exposed to TVB-2640, 128 phosphosites were up-regulated and only one was downregulated (Figure 5D). Among all three time points, 24 total phosphosites were shared. 65 phosphosites were uniquely differentially expressed after 6 h, 31 were shared with 24 h, and last nine phosphosites were shared with the 48 h time point (Figure 5B). A full 24 h after TVB-2640 exposure, 101 phosphosites were up- and 14 phosphosites were downregulated (Figure 5E). 36 phosphosites were unique to the 24-h time points, with 24 phosphosites being shared with the 48-h time point. Finally, 48 h after exposure, only 111 phosphosites were up- and 20 phosphosites were downregulated (Figure 5F). 74 phosphosites were uniquely differentially expressed after 48 h of TVB-2640 exposure.

HT-29 spheroids were also exposed to cerulenin, and then changes to their phosphoproteome were subsequently examined. From all of the differentially expressed phosphosites of HT-29 spheroids exposed to cerulenin, 151 were common among all three time points (Figure S5C). After 6 h of cerulenin exposure, 10 phosphosites were up- and 174 phosphosites were downregulated (Figure 6A). Of these differentially expressed phosphosites, 11 were unique to the 6-h time point, 12 were shared with the 24-h time point, and 10 were shared with the 48-h time point (Figure S5C). 24 h after cerulenin exposure, many more sites were altered. 114 phosphosites were up- and 1004 were downregulated (Figure 6B). A total of 206 phosphosites were uniquely differentially expressed at this time point, and 749 phosphosites were shared with the 48 h time point. After 48 h of cerulenin exposure, 142 phosphosites were up- and 1125 phosphosites were downregulated (Figure 6C). 357 phosphosites were uniquely differentially expressed at 48 h.

Figure 6.

Volcano plots of differentially expressed phosphosites (log₂FC ≥ 1 or ≤ −1 and p-value <0.05) found within HT-29 spheroids exposed to cerulenin or TVB-2640. (A) 6 h cerulenin, (B) 24 h cerulenin, (C) 48 h cerulenin, (D) 6 h TVB-2640, (E) 24 h TVB-2640, and (F) 48 h TVB-2640.

HT-29 spheroids exposed to TVB-2640 had many alterations in their phosphoproteome. Among all three time points, 1,282 phosphosites were shared (Figure S5D). A majority of the phosphosites in this experiment were downregulated (Figure 6D,E). After 6 h, only two phosphosites were up-regulated: serine 231 of HS90A and serine 631 of LARP1. There were 2,150 downregulated phosphosites. Of all the phosphosites, 413 were unique to only 6 h, 137 were shared with 24 h, and 320 were shared with 48 h (Figure S5D). 24 h after TVB-2640 exposure, 19 sites were upregulated, and 1,547 sites were downregulated (Figure 6E). 70 sites were unique to the 24-h time point and 77 were shared with the 48-h time point. 48 h after exposure, only 12 sites were upregulated, and 1,894 sites were downregulated (Figure 6F). 227 total sites were unique to this 48-h time point.

A kinase enrichment analysis of the differentially expressed phosphosites was conducted using KEA2 to understand kinase activation after FAS inhibition (Figures S6 and S7). HCT 116 spheroids treated with cerulenin had no statistically significant (B–H corrected p-value of 0.05 or less) enriched kinases after 24 h inhibition (Figure S6A). After 48 h of cerulenin exposure, three statistically significant enriched kinases were observed: MAPK14, CDK2, and GSK3B (Figure S6B). HCT 116 spheroids inhibited with TVB-2640 had no statistically significant enriched kinases among all three time points (Figure S6C–E). Similar to HCT 116 spheroids exposed to cerulenin, HT-29 spheroids had the same three kinases (MAPK14, CDK2, and GSK3B) were statistically significantly enriched after both 24- and 48 h exposure (Figure S7A–C). Many more kinases were enriched with HT-29 spheroids exposed to TVB-2640 (Figure S7D–F). After 6 h, six kinases were statistically significantly enriched: MAPK14, CDK2, GSK3B, CDK1, MAPK8, and MAPK3 (Figure S7D). All but MAPK8 were enriched and significant after 24 h exposure (Figure S7E). After 48 h of TVB-2640 exposure, seven kinases were significantly enriched: MAPK14, CDK2, GASK3B, CDK1, CSNK1E, MAPK3, and MAPK8 (Figure S7F).

As mentioned in the previous section, ferroptosis can occur via different mechanisms, and a myriad of different proteins can lead to a ferroptotic death mechanism. While SLC7A11 was not detected at the protein level, one phosphosite was detected in the phosphoproteomics data set. Within HCT 116 and HT-29 spheroids inhibited with TVB-2640, it was observed that the phosphorylated levels of serine 26 were increasing over time. Phosphorylation of SLC7A11 at serine 26 by mTORC2 has been shown to inhibit the transport activity of SLC7A11, reducing the levels of cystine available for glutathione reduction.^43,44 Another phosphosite on a different protein—Acyl-CoA synthetase long chain family 4 (ACSL4)—also involved in ferroptosis was observed to change among the treatment conditions and time points. ACSL4 catalyzes the addition of PUFA lipids to CoA for incorporation into phospholipids that subsequently become incorporated into lipid membranes. CDK1 phosphorylates ACSL4 at serine 447 which marks this protein for degradation.⁴⁵ It was observed that in HCT 116 and HT-29 spheroids inhibited with TVB-2640 had lowered levels of ACSL4 pS447 over time, while the same spheroids inhibited with cerulenin had dramatically increased levels over time. The lowered levels of ACSL4 pS447 indicate that this protein is still abundant and is capable of incorporating PUFAs into glycerophospholipids, allowing for more instances of peroxide lipid formation to occur to destabilize the membrane.

As bottom-up proteomics only infers protein abundance from peptide abundance, the activation of a protein by cleavage of a portion of a protein cannot be inferred. This is the such case for Caspase-3/7, which only becomes active once it is cleaved into a 12 and 17 Da subunit. In this data set, peptides spanning each of these subunits was detected and quantified; however, activity of the enzyme cannot be inferred from these peptide abundances. Phenotypic validation assays must be conducted in order to answer enzymatic activity questions that cannot be determined from bottom-up proteomics and phosphoproteomics data. We conducted a Caspase-3/7 luminescence glo assay to quantify total Caspase-3/7 glo activity in HCT 116 and HT-29 spheroids inhibited with either cerulenin or TVB-2640 (Figure S8). Initial inspection of the data showed that the HCT 116 control samples contained an outlier, as determined by the Grubbs test. However, the low intensity value in the HT-29 TVB-2640 treated samples did not pass the Grubbs test for being an outlier and was included in further statistical analysis. The luminescence intensity was significantly different among both the control and cerulenin treated samples among HCT 116 and HT-29 spheroids (p = 0.0397 and 0.0199, respectively, at 95% confidence). There was no statistically significant difference between HCT 116 control and TVB-2640 intensities as well as between HCT 116 cerulenin and TVB-2640 intensities (Figure S8). The difference between HT-29 cerulenin and TVB-2640 treated spheroids was observed to not be statistically significant (p = 0.119). Due to these results, it appears that perhaps a more complex death mechanism is occurring, with HT-29 TVB-2640 treated spheroid dying via an intricate combination of both apoptosis and ferroptosis. This intriguing observation will be analyzed further to determine if that is the true case.

DISCUSSION

In this study, we conducted a time course proteomics and phosphoproteomics study of two different colon carcinoma spheroids inhibited with either a generation 1 FAS inhibitor (cerulenin) or a generation 2 FAS inhibitor (TVB-2640). Many FAS inhibitors have been developed with mixed success rates. One of the first FAS inhibitors discovered, cerulenin, was popular until it was observed to cause drastic weight loss in mice. Many years later, TVB-2640 was developed and shown to be much more specific for FAS. TVB-2640 is currently in a phase I clinical trial for the treatment of solid tumors.¹⁰ As the popularity of FAS inhibition rises for cancer therapeutics, there is a growing need for understanding the distinct molecular differences with inhibiting FAS. This study characterized the time course alterations in the proteome and phosphoproteome of HCT 116 and HT-29 spheroids inhibited with either cerulenin or TVB-2640. These two different cell lines were chosen due to their differences in oncogenic mutations, as well as their microsatellite stability status. HCT 116 cells are a KRAS mutant and MSI cell line where as HT-29 cells are a P53 mutant and MSS cell line. The microsatellite stability status shows whether a tumor will have functioning DNA mismatch repair (MMR) machinery. If a tumor is found to have an MSI status, that tumor will have a high mutational burden due to the inoperable MMR machinery, whereas an MSS status has a functional MMR machinery. The microsatellite stability status is also important for understanding treatment options as different treatments will work for MSI versus MSS tumor. For example, is has been observed that MSI CRC tumors have success at treatment with immune check point inhibitors whereas MSS CRC tumors were not.^46–48 Between the proteome and phosphoproteome, TVB-2640 treated spheroids had much fewer altered proteins and phosphosites. This difference is possibly due to the higher specificity of TVB-2640 for FAS and subsequently fewer off-target effects.

Recently, we conducted untargeted LC-MS lipidomics on HCT 116 and HT-29 spheroids inhibited with either cerulenin or TVB-2640.¹⁷ Numerous lipidomic alterations were observed; however, a handful of vital biological questions remained unanswered. One unanswered question proposed was the exact death mechanism that occurs with FAS inhibited spheroids. Most cells will undergo apoptotic cell death via the activation of various caspases. An iron mediated form of cell death, ferroptosis, has become popular for cancer therapeutics.^18,49 In the previously collected lipidomics data, there were elevated levels of double bonds in various lipid classes in TVB-2640 spheroids. Most notable were increases in unsaturated phosphatidylcholine (PC) and phosphatidylethanolamine (PE) lipids, which make up the majority of the cell membrane. It was proposed that the elevated unsaturated PC and PE lipids found in TVB-2640 treated spheroids may be causing an enhanced ferroptotic effect.

Within the proteomic and phosphoproteomic data sets found in this study, a number of proteins and phosphosites involved in ferroptosis were identified and quantified. Most notable was the difference in the expression of FSP1 in the samples. FSP1 works by first either reducing ubiquinone to ubiquinol along with NADH/NADPH or reducing vitamin K to VKH2. Ubiquinol and VKH2 then operate as antioxidants, capturing free radicals and preventing lipid peroxidation.⁴² Within the phosphoproteomic data set, the phosphosites of SLC7A11 and ACSL4 were identified and quantified. These specific phosphosites control the activity of these two proteins by phosphorylation from mTORC2 and CDK1, respectively.^43–45 As previously mentioned, SLC7A11 is responsible for importing reduced glutathione into the mitochondria for the reduction and scavenging of lipid peroxides by GPX4. ACSL4 is responsible for the addition of PUFAs to glycerophospholipids, increasing the abundance of PUFA lipids found within the cell membrane.

HT-29 spheroids inhibited with either cerulenin or TVB-2640 were observed to have increasing levels of FSP1, elevated abundance of SLC7A11 pS26 that reduces the activity of this transporter, and TVB-2640 treated spheroids had decreased abundance of ACSL4 pS447 indicating consistent incorporation of PUFA lipids into PC and PE lipids. It was also observed that HT-29 spheroids treated with cerulenin had a downregulation of CASP3. These results show that HT-29 spheroids—an MSS, TP53 mutant colorectal cancer cell line—treated with either cerulenin or TVB-2640 are undergoing a ferroptotic death mechanism. This result agrees with the previously collected lipidomics data where it was observed that PC and PE lipids with two more degrees of unsaturation were elevated compared with controls as well as fatty acids.¹⁷ The decreased ability of these spheroids to scavenge lipid peroxides coupled with increased abundances of PUFA PC and PE lipids, as well as fatty acids, shows that ferroptosis is occurring within these spheroids.

HCT 116 spheroids treated with either cerulenin or TVB-2640 die via a different death mechanism, most likely apoptosis. First, the GSEA analysis shows that HCT 116 spheroids inhibited with cerulenin have activation of many gene sets related to programmed cell death and the apoptotic process (Figure 2). Specific proteins and phosphosites involved in apoptosis were then identified within the data sets. HCT 116 spheroids inhibited with cerulenin had significant downregulation of pyruvate dehydrogenase kinase 1 (PDK1) after 48 h. PDK1 protects cells against apoptosis, and it has been shown that knockdown of PDK-1 inhibits tumor growth and decreases invasiveness.⁵⁰ It was also observed that HCT 116 spheroids treated with cerulenin had significant upregulation of mortality factor 4-like protein 2 (MO4L2), a component of the NuA4 histone acetyltransferase complex, assisting in activating genes by acetylation of histones H4 and H2A. It has been observed that higher expression of MO4L2 confers suppression of apoptosis.^51,52 Also, HCT 116 spheroids exposed to cerulenin had downregulation of RelA-associated inhibitor (IASPP), a protein that inhibits the TP53 mediated pathway of apoptosis.^53,54 The combination of elevated MO4L2 attempting to suppress apoptosis in combination with downregulation of IASPP, activating TP53 apoptosis, suggests that HCT 116 spheroids exposed to cerulenin are dying via an apoptotic mechanism. Due to the minimal alterations in the proteome and phosphoproteome of HCT 116 spheroids inhibited with TVB-2640, the exact mechanism of death cannot be conclusively determined. However, due to the upregulation of MA1B1, glycoproteomics and glycomics may be conducted at a future time to determine how glycosylation is influencing the mechanism of death.⁵⁵

As shown by the previously published lipidomics data, specific proteins involved in various steps of lipid synthesis were observed to be altered. HCT 116 spheroids inhibited with cerulenin had an upregulation of enzymes involved in LPA, PC-O, and PE-O synthesis. This data corroborate the lipidomics data as these same spheroids were observed to have elevated levels of PC-O lipids with one or more double bonds compared with HCT 116 spheroids exposed to TVB-2640. HT-29 spheroids exposed to TVB-2640 were observed to have downregulation of palmitoyl-protein thioesterase 1 (PPT1) after 48 h. PPT1 removes fatty acyl lipids (e.g., palmitate) from cysteine residues.⁵⁶ These results agree well with the lipidomics data, with TVB-2640 treated HT-29 spheroids having decreased levels of free fatty acids.¹⁷ This would create further downstream issues for lipid homeostasis, as there would be less available fatty acids to synthesize damaged lipids brought on by ferroptosis. HT-29 spheroids exposed to cerulenin were observed to have upregulation of Patatin-like phospholipase-domain containing protein 2 (PLPL2), an enzyme that hydrolyzes TGs from lipid droplets.^57,58 These same spheroids also had upregulation of hyrdroxymethylglutaryl-CoA synthase (HMCS1), an enzyme involved in synthesizing mevalonate, a precursor for cholesterol synthesis.⁵⁹ HT-29 spheroids were also observed to have dark Oil Red-O staining of lipid droplets in the previous lipidomics study.¹⁷ From these proteomics data, the dark staining can be attributed to increased cholesterol synthesis and the increased hydrolysis of TG lipids from lipid droplets denotes their increased production. Perilipin-3 and −4 were detected in this data set; however, they were not differentially expressed. It is possible that these LDs are swollen to potentially sequester the hydrophobic cerulenin to halt the effects of the therapeutic.^60,61

It was also observed that alterations to proteasome proteins were among the varying FAS treatments. The E2 ligase Ubiquitin Conjugating Enzyme E2 S—UBE2S—was observed to be increasing in the cerulenin treatment in HCT 116 spheroids and decreasing in TVB-2640 treated spheroids (Figure 1B). The role of UBE2S is to elongate ubiquitin chains on proteins, increasing the degradation of these proteins.⁶² It has been observed that alterations to the proteasome can have influences on ferroptosis, as well as certain proteasome members specifically ubiquitylating and deubiquitylating specific ferroptosis proteins.^63–65 The proteasome will be examined in the future by conducting a ubiquitinomics experiment to discern differences in the proteasome pathways.

CONCLUSION

Here, we profiled the proteome and phosphoproteome of HCT 116 and HT-29 spheroids inhibited with two different generations of FAS inhibitors. Alterations to many different proteins and phosphosites were observed. Alterations to enzymes involved in lipid metabolism and FeS cluster synthesis and maintenance was observed, showing that, in combination with previously collected lipidomics data from our research group, these therapeutics are causing distinct alterations to cellular metabolism and respiration. Most notably observed in these data were changes in proteins involved in ferroptosis and apoptosis. Through the increase in abundance of FSP1, reduced activity of the SLC7A11 transporter, and activation of ACSL4, HT-29 spheroids exposed to either cerulenin or TVB-2640 undergo a ferroptotic death mechanism. In order to understand the interplay of apoptosis and ferroptosis occurring in the complex spheroid model, more experiments will be conducted to determine whether these differences are unique to different regions of the spheroid.

Supplementary Material

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.jproteome.4c00252.

Figure S1: Normalized and log₂ transformed TIC for all label-free quant proteomic samples; Figure S2: Hierarchical clustering of proteomics samples showing separation based on cell types but not between conditions or time points; Figure S3: Venn diagram of differentially expressed proteins (log₂FC ≥ 1 or ≤ −1 and p-value <0.05) among the three experimental time points; Figure S4: Normalized and log₂ transformed TIC for all label-free quant phosphoproteomic samples; Figure S5: Venn diagram of differentially expressed phosphosites (log₂FC ≥ 1 or ≤ −1 and p-value <0.05) among the three experimental time points; Figure S6: KEA plots showing enriched kinases found within HCT 116 spheroids treated with cerulenin or TVB-2640; Figure S7: KEA plots showing enriched kinases found within HT-29 spheroids treated with cerulenin or TVB-2640; Table S1: Top three up- and downregulated phosphosites found within HCT 116 and HT-29 spheroids inhibited with either cerulenin or TVB-2640 for 6-, 24-, or 48 h (PDF)

Supplementary Table S1: Log₂FC and B–H Corrected p-values of all quantified proteins found in HCT 116 and HT-29 spheroids exposed to either cerulenin or TVB-2640 for either 6-,24-, or 48 h (XLSX)

Supplementary Table S2: Common differentially expressed proteins found from Venn Diagram results (XLSX)

Supplementary Table S3: Log₂FC and B–H Corrected p-values of all quantified phosphosites found in HCT 116 and HT-29 spheroids exposed to either cerulenin or TVB-2640 for either 6-,24-, or 48 h (XLSX)

Supplementary Table S4: Common differentially expressed phosphosites found from Venn Diagram results (XLSX)

Data Availability Statement

Raw LC-MS proteomics files and searched .elib files have been uploaded to the ProteomeXChange consortium via the PRIDE repository under the identifier PXD050987.