Quantitative and Temporal Proteome Analysis of Butyrate-treated Colorectal Cancer Cells

Colorectal cancer is one of the most common cancers in developed countries, and its incidence is negatively associated with high dietary fiber intake. Butyrate, a short-chain fatty acid fermentation by-product of fiber induces cell maturation with the promotion of growth arrest, differentiation, and/or apoptosis of cancer cells. The stimulation of cell maturation by butyrate in colonic cancer cells follows a temporal progression from the early phase of growth arrest to the activation of apoptotic cascades. Previously we performed two-dimensional DIGE to identify differentially expressed proteins induced by 24-h butyrate treatment of HCT-116 colorectal cancer cells. Herein we used quantitative proteomics approaches using iTRAQ (isobaric tags for relative and absolute quantitation), a stable isotope labeling methodology that enables multiplexing of four samples, for a temporal study of HCT-116 cells treated with butyrate. In addition, cleavable ICAT, which selectively tags cysteine-containing proteins, was also used, and the results complemented those obtained from the iTRAQ strategy. Selected protein targets were validated by real time PCR and Western blotting. A model is proposed to illustrate our findings from this temporal analysis of the butyrate-responsive proteome that uncovered several integrated cellular processes and pathways involved in growth arrest, apoptosis, and metastasis. These signature clusters of butyrate-regulated pathways are potential targets for novel chemopreventive and therapeutic drugs for treatment of colorectal cancer.


Introduction
In developed countries, colorectal cancer is a prevalent disease with high mortality and morbidity rates (1). This disease has emerged as the top malignancy in Singapore. Environmental factors are responsible for about 80% of the cases whereas genetic predisposition accounts for the minority 20% of cases. Epidemiological evidence suggests that high intake of dietary fiber reduces the incidence and risk of this neoplasm (2,3). A wealth of studies has shown that butyrate produced from anaerobic fermentation of indigestible carbohydrate, is the molecule responsible for the chemopreventive properties of fiber-rich diet (4)(5)(6).
While butyrate serves as an energy source for normal colonocytes, in vivo and in vitro studies have shown that at physiological concentrations, this natural short-chain fatty acid mediates cell maturation with the promotion of growth arrest followed by differentiation, and/or apoptosis of cancer cells (7)(8)(9)(10)(11). These biological effects are crucial in colorectal cancer therapy as colonic transformation is characterized by multi-stage alterations of tissue homeostasis resulting in aberrant cell division and/or cell death (12,13). Butyrate has been purported as a potential anti-cancer agent. This initiated notable research in identifying proteins that contribute to its biological effects (14,15). However, most of these investigations focused on one target at any one time and were thus unable to systematically elucidate butyrate's mode of actions in an integrated manner.
Through the use of DNA microarray technology, Mariadason et al. (16) showed that butyrate induced maximal genetic reprogramming after 16h of treatment on colorectal cancer cells. In our earlier work, a functional proteomics approach using pre-fractionation strategy coupled with 2-D DIGE analysis was undertaken to identify candidate proteins regulated by 24h butyrate treatment in HCT-116 cells (17). We have also demonstrated the cell line's high by guest on May 8, 2020 https://www.mcponline.org Downloaded from sensitivity to butyrate-induced growth inhibition and apoptosis in a time-and dose-dependent manner (18). Therefore, the stimulation of cell maturation by butyrate implicated a temporal orchestration of various cellular processes.
In this article, we carried out comparative proteome analysis of HCT-116 cells treated with butyrate at three time-points with the aim to identify clusters of proteins (and pathways) that showed a consistent trend of differential expression over time. The synergistic influence of each cluster of proteins may result in the overall phenotypic response to butyrate. Herein, the chosen period of treatment (24h, 36h and 48h) spans from the induction of growth arrest and early phase of apoptosis till the late phase of cell death. In addition to providing insights into the mechanism underlying butyrate's pleiotropic effects, our study of the time dynamics of butyrate treatment could lead to the discovery of potential therapeutic targets associated with the progression of cell maturation in cancer cells. As the iTRAQ methodology permits multiplexing of 4 samples in a single experiment, it is well suited for the evaluation of the dynamic cellular response to butyrate in a time-course study (19). Here, we show the first experimental iTRAQ data for butyratetreated HCT-116 cells carried out at 24, 36 and 48 hours. triethylammonium bicarbonate/1.0% (w/v) SDS was used for extraction and denaturation of cellular proteins by boiling at 100 o C for 10min. Cellular debris was removed after centrifugation at 18,800g for 1h at 23 o C. iTRAQ labeling of each sample was performed according to the manufacturer's protocol (Applied Biosystems, Foster City, CA, USA). 100µg of protein was reduced with 5mM tris-(2-carboxyethyl)phosphine (TCEP) at 60 o C for 1h, and subsequently alkylated with 10mM methyl methane-thiosulfonate (MMTS) for 10min. After cysteine blocking, each sample was diluted to 0.05% (w/v) SDS prior to trypsinization at 37 o C for 16h. Following this, each tryptic digest was labeled for 1h with one of the four isobaric amine-reactive tags as follows: Tag 114 -24h control; Tag 115 -24h treated; Tag 116 -36h treated; and Tag 117 -48h treated samples. These four iTRAQ-derivatized samples were then pooled and passed through a strong cation exchange cartridge as recommended by the manufacturer (Applied Biosystems). This eluate (from the ion exchange step) was desalted using a Sep-Pak cartridge (Millipore), vacuum dried and reconstituted for 2-D LC.
They were then subjected to centrifugation at 18,800g for 1h at 23 o C to remove cell debris.
cICAT labeling and processing of the samples followed standard protocols (Applied Biosystems). 100µg of protein from the control and butyrate-treated cell lysate of each timepoint were each reduced with 1.25mM TCEP, and subsequently labeled with the respective isotopic light and heavy forms of the cICAT reagents, for 2h at 37 o C. Each pair of heavy and light cICAT derivatized proteins from each time-point was then pooled and trypsinized at 37 o C for 16h. Upon completion of in-situ digestion, the digested peptide mixture was cleaned up with 6 a strong-cation exchange cartridge, and then enriched with an avidin affinity cartridge. The cICAT-labeled peptides were then dried by speed vacuuming, dissolved in cleaving reagents and incubated at 37 o C for 2h. After the removal of biotin, peptides were brought to dryness again before being reconstituted for 2-D LC.
iTRAQ-labeled samples. One thousand shots were accumulated for each MS spectrum.
For MS/MS, 6,000 shots were combined for each precursor ion with signal to noise (S/N) ratio greater or equal to 100. For precursors with S/N ratio between 50 and 100, 10,000 shots were acquired. The resolution used to select the parent ion was 200. No smoothing was applied before peak detection for both MS and MS/MS, and the peaks were deisotoped. For MS/MS, only the peaks from 60 Da to 20 Da below each precursor mass, and with S/N ≥ 10 were selected. Peak density was limited to 30 peaks per 200 Da, and the maximum number of peaks was set to 125.
Cysteine methanethiolation, N-terminal iTRAQ labeling, and iTRAQ labeled-lysine were selected as fixed modifications while methionine oxidation was considered as a variable modification. One missed cleavage was allowed. Precursor error tolerance was set to 100 ppm while MS/MS fragment error tolerance was set to 0.4 Da. Maximum peptide rank was set to by guest on May 8, 2020 2. iTRAQ quantification was performed using the GPS Explorer TM software and normalized among samples. iTRAQ ratios were calculated based on the cluster areas of the iTRAQ reporter fragment peaks (114, 115, 116 and 117), and the ratios calculation included only peptides identified with C.I. % above cutoff thresholds as described below.
The average iTRAQ ratio and standard deviation (S.D.) were determined using the GPS Explorer TM software which was calculated using the following equations:  with normalized ratio (normalized against median ratio of all the cICAT pairs detected) greater than 40% were selected for fragmentation. Singletons were also selected as precursor ions. Stop conditions were implemented so that 3,000 to 6,000 shots were accumulated depending on the quality of the spectra. The resolution used for parent ion selection was 200. Peak processing and detection procedures were the same as above mentioned. Heavy and light cICAT- Estimation of false positive rate to determine cut-off score. In addition to the IPI human database, a randomized database (67922 sequences) generated using IPI human database Version 3.30 (generated using a Pearl script downloaded from http://www.matrixscience.com/help/decoy_help.html) was also used to search both the iTRAQand cICAT-labeled samples. The false positive rate was calculated by comparing the peptide hits obtained from these 2 databases at different ion score C.I. % (peptide). The minimum ion score C.I. % was set such that no more than 5% false positive rate is achieved. Based on this cut-off threshold, all the proteins identified from the random database search were single-peptide matched. Hence, proteins identified from the human database that are matched to at least 2 peptides are statistically confident. For single-peptide matched proteins, only those with ion score C.I. % greater than the highest C.I. % attained from the random database search were selected. With these cut-off thresholds, we essentially achieved 0% false-positive identification rate at protein level. In addition, those single-peptide matched proteins must be identified based on peptide which has been detected several time in one run or in replicate runs. The minimum ion score thresholds that were used for each iTRAQ-and cICAT-labeled sample were shown in the supplementary data.

Results and Discussion
Protein identification from iTRAQ-and ICAT-labeled peptides.
783 unique proteins were identified from a total of 3,116 tryptic peptides for the iTRAQlabeled samples. On the other hand, 137 unique proteins were identified from a total of 241 peptides obtained from cICAT (see supplementary data for the lists of iTRAQ-and ICATlabeled proteins that showed temporal differential expression after butyrate treatment). Due to the difference in labeling chemistry, the result obtained from the cICAT approach complements the iTRAQ data. Recently, quantitative proteomics incorporating stable isotope tagging such as post-isolation labeling using ICAT or iTRAQ was demonstrated to be a complementary strategy to 2-DE (23,24). Most notably, a comparative study of these three proteomics methods found limited overlapping proteins between them, and iTRAQ was considered to be the more sensitive technology as compared to ICAT and 2-D DIGE (25). This underscored the importance of using various technology platforms for a more comprehensive proteomics study of complex samples.
Interestingly, a subset of proteins found in this study had also been identified in our previous work using 2-D DIGE (17), and they showed regulation in the similar manner by butyrate treatment. Such proteins include cytoskeletal 8, ornithine aminotransferase, cytochrome c oxidase polypeptide VIb and Tu elongation factor.

Temporal analysis of proteins following butyrate treatment
by guest on May 8, 2020 From the list of differentially expressed proteins obtained from this temporal study, proteins that exhibited progressive up-or down-regulation were clustered into groups on the bases of their biological functions. They could be grouped into four cellular processes, viz., A) growth arrest, B) apoptosis, C) metabolism, and D) metastasis ( Figure 1, also see supplementary data for the complete list of differentially expressed proteins). Subsequently, some of these protein candidates were validated using quantitative real-time PCR and/or western blotting.
These results are shown in Fig. 2 and 3, and they are in accord with the proteomic results.
An overview of the temporal anti-cancer effects of butyrate treatment on the various cellular processes is shown in Fig. 4. This data is obtained from the iTRAQ ratios of the proteins grouped under each cellular processes at each time point. As seen here and discussed further below, our temporal analysis showed that butyrate induced a blockage of cell cycle progression as an early event (24h) whereas the anti-metastasis effect was most apparent at the later stage (48h) of treatment.

Temporal Regulation of the Cellular Processes and Pathways induced by Butyrate.
This study has clearly identified clusters of proteins in pathways that correlate protein expression changes with the induction of anti-cancer effects. The synergistic influence of each cluster of proteins results in the overall phenotypic response to butyrate. On the bases of these observations, we proposed a model to illustrate the integrated cellular mechanism initiated by butyrate in colorectal cancer cells ( Figure 5). On the other hand, the N-terminal fragment of AKAP12 may contribute to the tumour suppressor property of this protein. These await further investigations.

Cluster B: Apoptosis
As demonstrated in Figure 5, the proteins in Cluster B function as tumour suppressors, heat shock proteins and chaperones, players in the oxidative phosphorylation pathway, or ubiquitination-proteasome pathway respectively. The temporal changes in expression of these proteins contribute to the initiation of apoptosis by butyrate in HCT-116 cells.
Tumour suppressors. As shown in Figure 1, tumour suppressors, such as galectin-1, metallothionein-1X, prohibitin-2, and ras-related protein Rap-1A, displayed a temporal increase in expression level upon butyrate treatment in this study. These proteins contribute to tumour growth suppression by butyrate. For example, galectins are multifunctional β-galactoside lectins with roles including cell adhesion, growth regulation, invasion, and apoptosis (29). The identification of up-regulated galectin-1 here (validated with western blot in Figure 3) corroborated with previous work that showed its association with butyrate's actions (30,31). We also found metallothionein-1X was markedly up-regulated by butyrate and this was confirmed by real-time PCR (Figure 2 Similarly, our results also showed that voltage-dependent anion-selective channel protein 1 (VDAC1) and ADP/ATP translocase 2 (ANT2) were found to be concurrently up-regulated by butyrate. Their expression levels were shown to increase, particularly after the 36h time-point (Figures 1 and 4). cancer clinical trial. HSP90 is responsible for maintaining the stability of many oncogenic proteins with biological functions in cellular proliferation and apoptosis. HSP90 is known to be dysfunctional in tumours (46,47), and was detected to be up-regulated in transformed cells.
Inhibitors of this anti-apoptotic protein triggered cancer cell death synergistically with butyrate treatment (48). The reduced expression of chaperones as shown here will deter proper protein folding leading to protein aggregation, ultimately resulting in cell death in cancer cells.
Ubiquitination-Proteasome pathway. Proteasome activator subunit 2, ubiquitinactivating enzyme E1 and F-box only protein 2 are some of the proteins in the ubiquitinationproteasome pathway that were also noted to be differentially regulated by butyrate, as shown in our results (Figure 1). Degradation of proteins via the ATP/ubiquitin-dependent pathway mediates apoptosis (49). Targets of the 26S proteasome include proteins in heat shock response and cell cycle control (50,51); both systems were found to be down-regulated in this study (Figures 1 and 4). The butyrate-induced apoptotic cascades are associated with the ubiquitindegradation system, and inhibitors of the proteasome act synergistically with butyrate in anticarcinogenic therapy. In support of this, Pei et al. (52) found that the simultaneous application of a proteasome inhibitor and butyrate could induce apoptosis. Both Yu et al. (53) and Giuliano et al. (54) showed similar synergistic effects between proteasome activity and butyrate. Hence, butyrate-regulated ubiquitination-proteasome pathway would affect the levels of survival-and apoptosis-related proteins in cancer cells.

Cluster C: Metabolism
Our data identified a repertoire of biosynthetic enzymes, including those involved in the Krebs cycle and pentose phosphate pathway, to be up-regulated by butyrate in a time-dependent by guest on May 8, 2020 manner ( Figure 1). The change in the expression levels for most of these proteins was shown to be more pronounced after 36h of treatment. Examples of these metabolic enzymes were malate dehydrogenase, oxoglutarate (alpha-ketoglutarate) dehydrogenase, transaldolase and transketolase. This suggested that butyrate altered the metabolic machinery of HCT-116 cells.
Most tumours including CRC depend on the enhanced glycolysis instead of oxidative phosphorylation for ATP production, even in the presence of oxygen, a phenomenon known as the "Warburg effect" (55). The metabolic enzymes found to be up-regulated by butyrate in this study are involved in various glucose metabolic pathways which thus promote glucose metabolism. However, unlike other metabolic enzymes, alpha-enolase was shown to be downregulated by butyrate. This may retard the rate of glycolysis since enolase catalyzes the formation of phosphoenolpyruvate, a precursor of glycolytic end-product pyruvate. Furthermore, several enzymes functioning in the oxidative phosphorylation pathway were up-regulated by butyrate (as discussed earlier).
Butyrate demonstrates phenotypical specificity whereby it causes growth arrest followed by differentiation and/or apoptosis in carcinoma cells but promotes proliferation in normal cells (56). Colonic carcinoma cells derive energy via metabolism of glucose whereas normal colonic epithelial cells oxidize butyrate as the key fuel source for cellular proliferation (57)(58)(59). Butyrate has been reported to induce apoptosis in the presence of glucose and pyruvate but promote growth in the absence of these alternative energy sources (60). Herein, butyrate altered the metabolic profile of cancer cells, resulting from an enhanced expression of several metabolic enzymes. Metabolism of other energy sources as fuel thus avails butyrate to effect its anti-cancer actions in HCT-116 cells.
by guest on May 8, 2020 In addition, proteins functioning in amino acids and lipid/cholesterol metabolic pathways, such as ornithine aminotransferase, asparagine synthetase, argininosuccinate synthase, delta 1pyrroline-5-carboxylate synthetase and enoyl-CoA hydratase, were up-regulated in this study.
Leschelle et al. (61) and Tabuchi et al. (62) have demonstrated stimulated lipogenesis by butyrate. Ruemmele et al. (63) and Della Ragione et al. (64) found that inhibiting protein synthesis by cycloheximide blocked butyrate-induced apoptosis. In this work, vesicular transport proteins, which function in protein synthesis such as vesicle trafficking protein SEC22b (verified by real-time PCR and western blot), clathrin heavy chain 1 and NAPB protein, were identified to be up-regulated. These pathways were grouped under Cluster C in the proposed model ( Figure 5).

Cluster D: Metastasis
Cytoskeleton-associated proteins. In correlation to previous reports on cytoskeletal organization of cancer cells (65,66), the data here showed increased expression of various cytoskeleton-related proteins by butyrate (Figure 1). The overall increase in the expression level of these proteins was higher after the 36h time-point. The concerted temporal up-regulation of these proteins such as cytoskeletal 18, cytoskeletal 19, epiplakin and filamins may lead to a strengthened cytoskeletal scaffold and reduced metastasis potential of carcinoma cells (Cluster D in Figure 5). Several of these identified proteins function as crosslinkers in the intermediate filament network, modulating cell adhesion, motility and invasiveness. Real-time PCR was conducted for cytoskeletal 19 (Figure 2). LIM domain and actin-binding protein, also known as the elevated expression of epithelial protein lost in neoplasm (EPLIN), identified by cICAT, diminish the invasiveness of cancer cells. EPLIN is a cytoskeleton-associated protein whose down-regulation in cancer cells may facilitate motility of these cancer cells (67). Our results by guest on May 8, 2020 showed that the anti-metastasis effect was induced as a later event, after growth inhibition and apoptosis (Figures 1 and 4). The anti-metastasis effect shown here corresponds to the in vivo study done by Velazquez et al. (68) that demonstrated inhibition of seeding and growth of colorectal metastases to the liver by intravenous infusion of butyrate in mice. Butyrate is currently evaluated in clinical trials and has shown optimistic results (69). These signature clusters of butyrate-regulated pathways could serve as potential therapeutic targets or proteomics markers to assess drug candidates' efficacy or toxicity. Our data clearly showed that in addition to targeting proteins involved in cell cycle blockage, apoptotic and anti-metastatic pathways, butyrate also alters the metabolic profile of the cancer cells to induce its anti-cancer effects. A better understanding of the mechanism whereby butyrate mediates its therapeutic actions would certainly aid in the design of better therapeutic intervention. Thus, multi-drugs regimen(s) that have synergistic effects on these clusters of pathways may be a promising pharmacological strategy for chemoprevention of colorectal

Supplemental Data
The supplementary data include the minimum ion score thresholds used, determination of cutoff threshold of fold changes for proteins with single peptide match, and the list of iTRAQand cICAT-labeled proteins that showed temporal differential expression after butyrate treatment.
The ratio and sequence of peptides that are matched to the proteins (for both ≥2 peptides and single-peptide based identifications), and the MS/MS spectra for single-peptide based protein identifications are also included.     Metabolic machinery of the cells was altered with an increased expression of several metabolic enzymes. (D) Expression of cytoskeleton-associated proteins was increased to strengthen cytoskeletal scaffold and lower the metastasis potential of HCT-116 cells.     . Western blots of proteins identified to have differential expression from iTRAQ data. 3a. Western blot confirmed differential expression of these proteins. GAPDH was used as the loading control. For AKAP12, decreased expression of full length protein (~200kDa) was detected but an increased presence of a protein fragment ~40kDa was seen over time. 3b. 2-DE (pH 3-10) western blot for AKAP12 was performed to confirm the increased expression of the fragment protein at ~40kDa (Circled in Figure 3b).