| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
MUCOSAL BIOLOGY
1Department of Chemistry, University of Minnesota, Duluth, Minnesota; 2Department of Molecular Biosciences, University of California, Davis, California; and 3Department of General Surgery and 4Department of Internal Medicine, University of Florence, Florence, Italy
Submitted 14 December 2005 ; accepted in final form 23 January 2006
 |
ABSTRACT |
|---|
 TOP ABSTRACT MATERIALS AND METHODSRESULTS AND DISCUSSIONGRANTSREFERENCES |
|---|
-oxidation, as the main source of energy, to anaerobic glycolysis was observed owed to the alteration of nuclear- versus mitochondrial-encoded proteins and other proteins related to fatty acid and carbohydrate metabolism; 2) lower capacity for Na+ and K+ cycling; and 3) operativity of the apoptosis pathway, especially the mitochondrial one. This study of the human colon cancer proteome represents a step toward a better understanding of the metabolomics of colon cancer at early stages confined to the intestinal wall.
protein markers; apoptosis; bioenergetics
At present, most studies have focused on understanding the genetic changes that occur in tumorigenesis. The global gene expression profiles of normal and cancerous human colonic epithelium had been studied by a relatively new technique known as serial analysis of gene expression, which identifies and quantifies cellular mRNA transcript levels (83). The drawback of concentrating at the transcript level has been demonstrated by independent studies (2, 24, 38) that have shown that correlations between mRNA and protein expression cannot predict protein expression levels from quantitative mRNA data; thus comparative analyses of normal and diseased states at the protein level is most informative.
The first data describing the alteration of protein expression in transformed colon mucosa were published in the beginning of the 1980s (41, 74). In 1983, Anderson et al. (1) described differential distribution of acidic proteins in normal colon mucosa, primary colon adenocarcinoma, and colon metastases to the liver. With the use of the two-dimensional (2-D) gel electrophoresis (GE) approach, several differences in protein expression in the cytosol of colon cancer samples, at various differentiation levels, were observed (50). Unfortunately, none of these proteins were identified. Keesee et al. (44) detected the tumor-specific expression of six nuclear matrix proteins that were also present in a panel of colon tumor cell lines. This study reinforced the concept of differential protein expression by cancerous tissue, and it raised the problem of accuracy at extrapolating results from cancer cell lines to in vivo scenarios (83, 56). Specific expression of several nuclear proteins was observed by Szymczyk et al. (70) in human colorectal cancer tissue compared with normal matched mucosa. One of these proteins was a 36-kDa polypeptide that was absent in the normal colonic epithelium but present in 83% of the studied carcinomas. This protein might have served as a potential marker for colon malignancy, but, unfortunately, it was not identified, thus limiting its use and understanding of its function.
Recently, a novel approach for purification of epithelial cells from normal and tumoral colorectal mucosa has been proposed, and it represents an important step to avoid a cellular contribution other than that of epithelial cells (57) but also may restrict or underestimate the importance of the interactions between epithelial and other tissue cells to the development of a tumor.
The data on specific cancer-related proteins or cancer-related protein expression downregulation are still controversial. Recently, a 2-D electrophoresis database for human colon crypts, colorectal cancer cell lines (i.e., LIM 1215 and HT29), and colonic polyps of multiple intestinal neoplasia and p53-null mice has been set up by the Ludwig Institute for Cancer Research (43, 67, 69). Stulik et al. (69) identified proteins that exhibit differential expression by comparing normal and malignant colon tissues. Most of cancer samples were in stages II and III, spreading from stage I to IV (69). In that study, it was concluded that there were no spots unique for either normal or pathological tissue and, furthermore, that there was no clear difference in the expression of the vast majority of spots between normal and neoplastic tissue. Only a few spots for each group seemed to exhibit striking alterations in protein abundance: a fatty acid-binding protein (FABP) and SM22-
(an actin-binding protein) were significantly downregulated in tumor tissue, whereas four proteins, all members of S100 family for calcium-binding proteins, were more abundant in transformed colon mucosa.
The focus of this study was to understand the biochemical differences of human colon cancer and normal colon tissue1 from the same individual by expression proteomics, when the cancer is at relatively "early" stages, providing a more homogenous sampling than in previous studies. In addition, a higher loading of the protein sample followed by staining with a fluorescent dye facilitated the detection of low-abundant proteins, focusing on the following pathways: energy metabolism, Na+/K+ cycling, and apoptosis. This study may help our understanding of functional proteomics in colon cancer. This approach may help to identify proteins that could be useful as new tool in the diagnosis or prognosis of this type of patients and perhaps to develop new therapeutic strategies to better aim therapy on the neoplastic target while minimizing the impact on normal tissues.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTMATERIALS AND METHODSRESULTS AND DISCUSSIONGRANTSREFERENCES |
|---|
Tissue samples were obtained from patients (4 men and 4 women; median age: 64 yr; age range: 46–81 yr) who had undergone surgical resection for primary sporadic colorectal adenocarcinomas at the Department of General Surgery, University of Florence, Florence, Italy. None of the patients had taken nonsteroidal anti-inflammatory drugs for at least 3 mo before surgery. All patients were thoroughly informed about the study and gave written consent for the investigation in accordance with the ethical guidelines of Local Ethical Committee.
Matched sets of colorectal carcinomas and normal colon mucosa used for both one-dimensional (1-D) and 2-D GE analyses were obtained within 30 min after surgical resection. The tissue samples were checked and sliced by a pathologist. The diagnoses of all samples were obtained from pathology reports.
Samples of small colorectal carcinomas were collected from the distal colon and rectum and were staged according to American Joint Committee on Cancer classification (33). Four tumors were located in the distal colon, and four tumors were located in the rectum. All eight tumors were adenocarcinomas classified as moderately differentiated. One tumor was stage I (T2, N0, M0); five tumors were stage II (T3, N0, M0; n = 5); and two tumors were stage III (T2, N1, M0). Cancer tissue (from the edge of the tumor) and adjacent normal mucosa (at least 10 cm from the tumor) were excised from each surgical specimen. Samples were washed with cold PBS and were homogenized in a 3:5 volume of lysis buffer (8 M urea, 2 M thiourea, 4% CHAPS, 40 mM Tris, 2% -mercaptoethanol or 0.1 M DTT, and 1 mM PMSF) using a motorized Teflon/glass homogeneizer (Potter-Elvejhem; total clearance: 0.05 mm). The resulting lysate was centrifuged using a Sorvall centrifuge at 12,000 g for 15 min at 4°C. Proteins were concentrated by treating the supernatants with 3 volumes of ice-cold acetone and kept at –20°C overnight. Precipitated proteins were resuspended in sample solution (8 M urea, 2 M thiourea, 0.1 M DTT, 2% carrier ampholytes, 2% Triton X-100, and 2% CHAPS), homogeneized, and subsequently stored at –80°C for further processing.
Chemicals and Biochemicals
The Multiphor II, DryStrip kit, Immobiline DryStrips with linear pH gradient 3–10, 18 cm, and Pharmalytes pH 8–10.5 and pH 3–10 were from Amersham Pharmacia Biotech (Uppsala, Sweden); the 6,000-V power supply was purchased from Hoefer. 2-D gels were cast and run in a Bio-Rad Multi Cell. Western blot analysis was done using a Bio-Rad Trans-blot apparatus (Richmond, CA). Acrylamide, urea, CHAPS, DTT, and tricine were from USB (Amersham Pharmacia Biotech); 1,4-bis-(acryloyl)piperazine (PDA), SDS, N,N,N,N-tetramethyl-ethylenediamine, ammonium persulfate, and nitrocellulose membranes were from Bio-Rad; and polyvinylidene difluoride (PVDF) and Nonidet P-40 were purchased from Boehringer (Mannheim, Germany). TRIZMA base, agarose, thiourea, tricarboxylic acid, glycine, formaldehyde, silver nitrate, methanol, acetic acid, citric acid, iodoacetamide, and sodium deoxycholate (DOC) were from Sigma (St. Louis, MO). Bicinchoninic acid (BCA) assay reagents were from Pierce (Rockford, IL). 1-D electrophoretic gels were cast and run in a Mini-Protean II System. Western blot analyses were performed using Mini Trans-Blot Transfer Cell; all remaining equipment was obtained from Bio-Rad. Deionized water (18 M
) prepared with a Barnstead system was used for all buffers.
Western Blot Analyses
Tumors and normal mucosa samples were homogenized in ice-cold buffer [0.9% NaCl, 20 mM Tris·HCl (pH 7.6), 0.1% Triton X-100, 1 mM PMSF, and 0.01% leupeptin] and centrifuged at 4°C for 10 min at 10,000 g. The protein content was determined by using a BCA protein assay kit (Pierce) with BSA as the standard. Samples were mixed 1:1 with Laemmli buffer under reducing conditions and boiled. SDS-PAGE was performed using 12% and 5% acrylamide for the separating gel and stacking gel, respectively. Proteins were transferred onto nitrocellulose membranes (Pierce). Blots were blocked with 5% BSA in PBS (Pierce) and incubated overnight at 4°C with the primary antibody [anti-B cell-associated protein 31 (BAP31) goat polyclonal antibody from Santa Cruz Biotechnology; Santa Cruz, CA]. Blots were further incubated with the appropriate secondary antibodies from Santa Cruz Biotechnology for 1 h at room temperature, incubated with SuperSignal West Pico Chemiluminescent Substrate (Pierce) for 5 min, and then exposed to CL Xposure Film (Pierce). Western blots for human caspase-9, apoptotic peptidase-activating factor (Apaf)-1, the -subunit of F0F1-ATPase, and cytochrome c oxidase subunit II were performed using anti-caspase-9 from Cell Signaling Technologies (no. 9502), anti-APAF1 from Cell Signaling Technologies (no. 4452), anti--subunit of F0F1-ATPase (A-21351 from Molecular Probes), and anti-cytochrome c oxidase subunit II (A-6404 from Molecular Probes) following the manufacturer's protocols. Blots were obtained by transferring the 2-D GE of the samples and probing sequentially (previous stripping) for the proteins. Secondary antibodies were from Santa Cruz Biotechnology. The images were taken with a Kodak 2000MM imager.
2-D GE
The first-dimensional IEF was performed on precast 18-cm IPG strips (Amersham Pharmacia Biotech) at 20°C with a maximum current setting of 50 µA/strip using an Amersham Pharmacia IPGphor IEF unit. Commercial strips with linear immobilized pH 3–10 gradient were used for isoelectric focusing. These strips were wetted with rehydration buffer containing 2 M thiourea, 7 M urea, 4% CHAPS, 0.5% Triton X-100, 100 mM DTT, and 0.4% Pharmalytes (pH 3–10) overnight in a leveled acrylic strip holder. The amount of protein loaded was 0.8–1 mg protein for preparative gels. After rehydration, the IEF run was carried out using the manufacturer's conditions for linear strips using a power supply EPS 3500 XL. Before the 2-D SDS-PAGE was carried out, strips were subjected to a two-step equilibration. The first was with an equilibration buffer consisting of 6 M urea, 30% glycerol, 2% SDS, 50 mM Tris·HCI (pH 6.8), and 1% (wt/vol) DTT. The second step was with a buffer consisting of 6 M urea, 30% glycerol, 2% SDS, 50 mM Tris·HCI (pH 8.8), and 2.5% (wt/vol) iodoacetamide. After the IPG strips were transferred onto the 2-D SDS-PAGE gel, a molecular mass marker strip was included, and the strips were sealed in place with 0.75% agarose. SDS-PAGE was performed on 1.0-mm-thick 10% polyacrylamide gels at a constant voltage at 13°C for 5 h using a Hoefer SE600.
Gel Staining
Silver staining of the gels was performed as follows: gels were fixed in 50% methanol-5% acetic acid in water for 30 min followed by washing in 50% methanol in water for 10 min. Gels were then washed again with water for 60 min and sensitized with 0.02% sodium thiosulfate for 2 min. After the gels were rinsed twice with water for 1 min each, they were incubated in chilled 0.1% silver nitrate for 40 min at 4°C. After the silver nitrate was discarded and gels were rinsed with two changes of distilled water for 1 min each, they were developed in 0.04% formalin (35% formaldehyde in water) in 2% sodium carbonate. When the desired intensity was attained, the developer was discarded, and the gel was incubated with either 1.46% EDTA disodium dihydrate or 1% acetic acid for 10 min to stop the development. The staining procedure was completed by three rinses with water for 5 min each. Stained gels were scanned using a Kodak Imager. Sypro RubyRed staining was performed by following the manufacturer's conditions, and images were taken with a Fluorochem imager.
Statistical Analysis and Image Analysis of 2-D Gels
Silver-stained or Sypro-stained gels were digitized using a Hewlett-Packard flatbed scanner with 300 dpi resolution. Scanned TIFF images were analysed using Phoretix 2D Gel Analysis Software version 5.01 (Nonlinear Dynamics; Newcastle, UK). Gel images were analyzed by using the "Spot Detection Wizard" function. Scanned gels were processed to remove vertical and horizontal streaks and enhance the spots before crosshairs were placed on the detected spots. Images were adjusted for warping by using internal markers. Spot detection parameters were set as follows: sensitivity, 9,138; operator size, 31; noise factor, 5; and background factor, 1. Spots were automatically detected and images were checked by eye for undetected or incorrectly detected spots. Spots incorrectly matched after the automatic matching were corrected by removal of the incorrect matching vector and manual insertion of a new matching vector ("User Seeds"). The total staining intensity within each detected spot, which is the amount of protein present, is defined as "spot volume." To correct for differences in staining intensity between gels and protein loading, spot volumes were divided by the volume of all spots and multiplied by the summarized total area of all spots. The resulting values represent the "normalized spot volumes." Duplicates of gels under each condition and for each patient were run simultaneously to maximize spot reproducibility. From the replicas of the control and tumor states, one "average control" and one "average tumor" were produced (Create Average Gel Option) for each patient. The average images were checked so that each spot was assigned a consistent match number on both gels (pairwise) and all gels (across patients).
Differences in intensity were calculated for each spot on every pair of gels. Statistical analysis consisted of a two-tailed Wilcoxon nonparametric paired t-test. Results were obtained as a list of spots that increased or decreased in intensity when control and cancer samples were compared at a given confidence level, e.g., P 
0.05. The Wilcoxon t-test, which requires no assumption of a normal distribution in spot intensities, takes the ranked magnitudes of differences for a spot on each pair of gels and reports the likelihood that this distribution could occur by chance; the null hypothesis is that no difference exists. To minimize false positive results, a high level of significance was required. This was achieved by two means: 1) the statistics were performed on spots that changed in all eight patients; and 2) a fourfold change in abundance relative to the control was counted. Smaller changes were arbitrarily judged to be biologically unimportant even though they could be statistically significant (5, 59). This threshold was set based on two facts: 1) >20% of all spots detected in gels run and processed in parallel (in particular, weak spots) differ in their intensities by a factor of more than two when silver staining is used (76, 80); and 2) analysis of replicate gels from a sample of rat liver mitochondria [isolated and purified through a Percoll gradient as indicated previously (31)] demonstrated that the average quantitative relative error of all spots from gel-to-gel analysis was 27%. If only the most abundant proteins were considered, then this error was 12%. However, no exclusion of spots was made based on abundance to compensate for the higher average error to avoid overlooking less-abundant proteins.
Enzymatic Digestion of Protein Spots
Silver-stained or Sypro -stained protein spots were excised manually with a sterile scalpel blade and transferred to a sterile microcentrifuge tube. Each excised spot was destained, washed, reduced with DTT, alkylated with iodocetamide, and digested in gel with trypsin. The peptide extraction was performed with ammonium bicarbonate and acetonitrile.
Matrix-Assisted Laser Desorption Ionization-Time-of-Flight Mass Spectroscopy Analysis of Tryptic Peptides
Samples were speed vacuumed to dryness after proteolytic digestion and then reconstituted to 15 µl with 95:5 water-acetonitrile and 0.1% trifluoroacetic acid (TFA), vortexed gently, and centrifuged at 2,000 g for 30 s. The desalting/concentration procedure was performed by using C18 ZipTip according to Millipore (Billerica, MA) instructions. A C18 resin was hydrated with 2 x 10-µl aliquots of 50:50 water-acetonitrile and 0.1% TFA and then equilibrated with 2 x 10-µl aliquots of aqueous 0.1% TFA. Peptides were bound to C18 by pipetting up and down in the sample five times with the pipettor set to 10 µl. Desalting was performed by pipetting 5 x 10-µl aliquots of aqueous 0.1% TFA through the resin. Elution was performed by pipetting 1.5 µl of 60:40 water-acetonitrile and 0.1% TFA over the resin three to four times. One microliter of the extracted sample from each tube was dispensed onto a matrix-assisted laser desorption ionization (MALDI) sample plate along with 1 µl of matrix solution consisting of 10-mg/ml -cyano-4-hydroxycinnamic acid (Sigma) in 0.1% TFA and 50% acetonitrile and spotted on a MALDI target. Mass analyses were performed according to a previously published method (Ref. 21 and the references therein) using a PerSeptive Biosystems Voyager-DE STR MALDI-time of flight (TOF) mass spectroscopy (MS) system (Framingham, MA) located at the Mass Spectrometry Facility (University of Minnesota). Samples were allowed to dry under ambient conditions. For each sample, the average of 256 spectra was acquired in the delayed extraction and reflector mode. The average of four scans (each containing 64 spectra) that passed the accepted criterion of peak intensity was automatically selected and saved. Spectra were automatically calibrated upon acquisition using a two-point calibration with residual porcine trypsin autolytic fragments [842.51 and 2,210.10 (M + W) ions]. Assignment of peaks was done manually; measured peptide masses were excluded if their masses corresponded to trypsin autodigestion products or if they were from identified proteins adjacent to the spot being analyzed.
2-D Map of Human Colon Cancer, Database Search, and Identification of Proteins
The proteins were identified by searching in SWISSPROT and National Center for Biotechnology Information nonredundant databases using MS-Fit (Protein Prospector, University of California, San Francisco, CA) and peptide mapping (Profound, version 4.10.5, Rockefeller University). All mass searches were performed using a mass window between 1,000 and 100,000 Da and included human, rat, and mouse sequences. The search parameters allowed for oxidation of methionine, NH2-terminal acetylation, carboxyamidomethylation of cysteine, and phosphorylation of serine, threonine, and tyrosine. The criteria for positive identification of proteins were set as follows: 1) at least four matching peptide masses; 2) 50 ppm or better mass accuracy; and 3) molecular mass and isoelectric point (pI) of identified mature proteins should match estimated values (within a 20% error) obtained from image analysis.
|
RESULTS AND DISCUSSION |
|---|
|
TOPABSTRACTMATERIALS AND METHODSRESULTS AND DISCUSSIONGRANTSREFERENCES |
|---|
The purpose of this study was to understand the biochemical differences of certain metabolic pathways in human colon cancer and normal colon mucosa by identifying proteins differentially expressed that may be useful as a diagnostic or prognostic marker or as a drug target. To this end, colon samples from normal tissue and tumors from each patient were analyzed by 2-D GE. Gel images of control and tumor samples were superimposed for each patient to detect differentially expressed proteins (as indicated in MATERIALS AND METHODS; Fig. 1). These spots were excised and trypsinized in situ, and peptides were analyzed by MS. Peptide masses along with the pI, molecular mass, and species were then processed to identify the protein by using ProteinProspector or Profound software, and the results were refined by using MASCOT software. With the use of this approach, 11 and 15 proteins from healthy and cancerous tissues, respectively, were identified and classified by the pathway in which they were involved (Table 1). One of these proteins (BAP31) was confirmed by using Western blot analysis, whereas others (Apaf-1, caspase-9, cytochrome c oxidase subunit II, and the -subunit of F0F1-ATPase) were identified solely by this latter technique, thus taking the numbers of differentially expressed proteins to totals of 14 and 18 for normal and cancerous tissues, respectively (Fig. 2). The choice of proteins selected for Western blot analyses was based at defining the role of mitochondria in various pathways during colon cancer: mitochondria as a source of mitochondial DNA (mtDNA) for mtDNA-encoded proteins such as cytochrome c oxidase subunit II; as a key component of apoptosis via the formation of the apoptosome (constituted by Apaf-1, cytochrome c, and ATP) and the ensuing activation of the initiator caspase-9; and, finally, as the main source of ATP in aerobic tissues by analyzing the levels of the catalytic subunit (-subunit) of F0F1-ATPase. Using the data shown in Table 1, we focused on the following major processes: energy metabolism, Na+ and K+ cycling, and apoptosis.
|
|
|
ATP is mainly provided by -oxidation of short-chain fatty acids, for they are the preferred metabolic fuel in normal colonocytes (4, 30, 61). These fatty acids (e.g., acetate, propionate, and butyrate) are generated at high levels in the colon by bacterial fermentation of dietary fiber and unabsorbed carbohydrates (4, 30, 61). If ATP is mainly provided by -oxidation followed by oxidative phosphorylation in normal colonocytes, then it is expected that proteins associated with this pathway would be differentially expressed. Confirming this view, NADH-ubiquinone oxidoreductase B18 (a component of complex I of the mitochondrial electron transport chain) and the -subunit of F0F1-ATPase (complex V) were found to be overexpressed when evaluated by proteomics and Western blot analysis, respectively. These results suggested the preferential utilization of aerobic metabolism by normal tissue.
Data from colon cancer indicated the overexpression of mitochondrial transcription factor 1 [also known as mitochondrial transcription factor A (TFAM)]. TFAM is encoded by nuclear DNA and controls the synthesis of mitochondrial respiratory chain components (cytochrome c oxidase I-III, NADH dehydrogenase subunits 1–6, cytochrome b, and ATPase 6 and 8) by initiating the transcription of mtDNA (25, 29, 52, 71). An overexpression of TFAM supports the notion that in cancer cells mtDNA has been preserved, for TFAM stabilization requires the presence of intact mtDNA. Furthermore, cells depleted of mtDNA are characterized by the absence of detectable TFAM protein levels but with normal levels of TFAM transcript (28, 45). When one of the mtDNA-encoded proteins, cytochrome c oxidase subunit II (a component of complex IV), was evaluated by Western blot analysis, its expression was found to be two times higher in cancer than in the normal colon (Fig. 2). These results confirmed the presence of intact mtDNA in cancer cells and supported the operativity of TFAM in cancer cells as an initiator of the transcription at the mtDNA level. These results (relatively lower expression of NADH dehydrogenase B18 and -subunit of ATPase and higher expression of cytochrome c oxidase subunit II) might force tumor cells to find a pathway other than -oxidation/oxidative phosphorylation to provide ATP.
An indication that this shift had occurred was provided by the low expression of FABP [liver-type (L-)FABP] and the high expression of lactic dehydrogenase (LDH).
Although the precise physiological functions of FABPs remain uncertain, several roles have been proposed. Among them are 1) enhancing the uptake of fatty acids into cells and facilitating their transport to intracellular organelles, 2) protecting enzymes and membranes from the effects of free fatty acids and their acyl-CoA derivatives, 3) maintaining a large intracellular pool of fatty acids for rapid mobilization, and 4) targeting fatty acids to specific metabolic pathways (for reviews, see Refs. 9 and 35). The low expression of FABP in colon cancer is in agreement with published data describing a significant decrease in FABP mRNA abundance in experimentally induced colon cancers of rats (17) and in FABP protein expression in transformed human colon mucosa (69). A FABP expression change has been reported to occur in other situations. Animals fed a poor-fermentable fiber diet (fibers protective against colon cancer) significantly increased the FABP mRNA level in exfoliated colonocytes during tumor development (19), whereas a loss of adipocyte-type FABP during bladder carcinoma progression has been also described (18). Enterocytes contain both intestinal (I-) and L-FABPs (36) aside from another, immunologically distinguishable FABP that is associated with the microvillus membrane (68). Although the mRNAs specifying I- and L-FABP are among the most abundant sequences that accumulate in the small intestinal epithelium (8, 37), it is unclear why the intestine requires the expression of two cytosolic FABPs. Gang and Ockner (34) have shown that in the intestine, luminally derived fatty acids are largely reesterified in the enterocyte to form triacylglycerol, whereas fatty acids derived from the intestinal blood supply are utilized for energy production and phospholipid biosynthesis (75). One attractive but unproven hypothesis is that the two FABPs contribute to this metabolic compartmentalization. If this hypothesis is correct, then the low expression of L-FABP in tumor colonocytes could reinforce the concept that these cells do not perform -oxidation owing to a lack of normal supply of fatty acids via the low expression of L-FABP.
The overexpression of LDH found in cancerous tissue might reflect the stimulation of ATP production by glycolysis to overcome the decreased ATP formation by oxidative phosphorylation (Table 1). Of interest, based on the proteomic data, a shift in the LDH isoenzyme pattern was also observed. LDH is a tetramer constituted by the combination of the M and H isoforms. The LDH expressed in cancerous tissue was identified as type 5, constituted only by M subunits. This isoform has a low affinity for its substrates and operates essentially toward the formation of lactate from pyruvate, working optimally under anaerobic conditions. These findings are in agreement with other studies (16, 79) performed with intestinal tumors and adenocarcinomas.
Combining these results (relatively lower expression of NADH-dehydrogenase B18 and the -subunit of F0F1-ATPase and higher expression of cytochrome c oxidase IV subunit II), it could be hypothesized that tumor cells shifted from -oxidation as their main energy source to glycolysis because of an imbalanced stoichiometry between mtDNA- and nuclear DNA-encoded proteins, resulting in a dysfunctional oxidative phosphorylation. This hypothesis may provide the molecular basis for the "respiratory impairment" of cancer cells postulated by Warburg (79) more than 80 years ago to explain the invariably high ratio of fermentation to respiration in cancer metabolism. However, our results extend this observation to compartments other than mitochondria, i.e., the bloodstream side of colonocytes (vide infra under lower capacity to control cation cycling).
An interesting consequence of the shift from -oxidation to glycolysis by tumor colonocytes is that an accumulation of one of the main fuels, butyrate, may ensue. Considering that butyrate plays a role in histone acetylation (11), it is possible that higher concentrations of butyrate may be required to induce the transcriptional activation of the Bax gene, as reported by Mandal et al. (47), thereby inhibiting proliferation and inducing apoptosis. In this regard, short-chain fatty acids in the colon are effective at reducing cancer incidence (12, 21, 77). This effect has been explained through several potential mechanisms: inhibition of histone deacetylases (39), induction of apoptosis (7, 40), modulation of gene expression (10), and induction of cell arrest (78).
Buttgereit and Brand (15) documented that in mammalian cells, macromolecular biosyntheses are the most sensitive processes to energy supply, followed by calcium and sodium cycling. The rate of protein and polynucleotide synthesis decreases to 40% when respiration is inhibited by 30% (15). If indeed in tumor tissue there is an overall lower capacity for ATP production, then several processes should be affected: 1) a decline in macromolecule biosynthesis should be observed (accompanied or not by an increased proteolysis); 2) a lower capacity to regulate sodium/potassium cycling; and 3) a minor impact on other processes relatively less dependent on ATP, such as apoptosis.
Decline in macromolecule biosynthesis. Regarding the first scenario, an average of 167 ± 10 protein spots was identified in gels from normal colon tissue, whereas a consistently lower number (105 ± 12, P < 0.05) was obtained from colon cancer samples.2 This difference in protein expression may represent halted protein synthesis, as indicated above, without excluding a simultaneous increased proteolytic activity (48, 64, 72).
Lower capacity to control cation cycling.
The differing Na+ and K+ concentrations inside and outside of eukaryotic cells are maintained largely by an antiport protein known as Na+-K+-ATPase. Na+-K+-ATPase catalyzes ATP-dependent transport of three sodium ions out and two potassium ions into the cell per pump cycle, thereby generating a transmembrane sodium gradient. The activity of Na+-K+-ATPase is involved in the control of cellular pH, osmotic balance, and Na+-coupled transport of nutrients such as amino acids and vitamins into cells. Another transporter, the sodium-glucose cotransporter (SLC5A1), takes advantage of this ion gradient by efficiently cotransporting glucose (or galactose) and sodium into cells. As a consequence of cotransporter activity, glucose becomes more concentrated inside the cell, which it then exits, down its concentration gradient, via a passive uniport glucose transporter (GLUT5 or GLUT2). This mechanism allows the intestine to collect glucose/galactose from digested food and then release it into the bloodstream. The overexpression of SLC5A1 found in normal colonocytes (Table 1) is indicative of appropriate expression of Na+-K+-ATPase and by an adequate ATP supply. In colon tissue, oxygen consumption decreases by about 30% in the presence of the Na+-K+-ATPase inhibitor ouabain (6), indicating that a significant fraction of ATP serves to power Na+-K+-ATPase. The lower ATP capacity by colon cancer tissue may have a significant impact on this process. The lower expression of SLC5A1 (present study) would lead to a lower entry of glucose and sodium. Intracellular sodium (among other factors) modulates the transcription of both components of Na+-K+-ATPase (i.e., the - and -subunits), increasing the levels of these mRNAs and proteins (46, 49, 81). At early stages of cancer, less expression (number of pumps) of Na+-K+-ATPase would be required to match the lower SLC5A1 expression to spare ATP while preserving sodium homeostasis. From our proteomic data, we found no detectable expression of the Na+-K+-ATPase 1-subunit in cancer tissue, although a considerable amount was observed in matched control tissue. No data on the Na+-K+-ATPase -subunit was obtained from either cancer or normal tissue owed to the underrepresentation of high-molecular-mass proteins and/or proteins with extreme pI in 2-D GE (82). Similarly, in other types of cancer, lower protein expression of the -subunit of Na+-K+-ATPase (with a constant or lower expression of the -subunit) has been reported: renal clear cell carcinoma (53), oncogenic transformation of Madin-Darby canine kidney (MDCK) cells with Moloney sarcoma virus (MSV-MDCK; Ref. 55), or early stages of bladder cancer (32).
However, we cannot exclude that in addition to a low expression of the Na+-K+-ATPase -subunit, changes in -isoform expression and/or isoform distribution might have occurred. Recently, it has been observed that there is a switch from the 1-subunit (normal mucosa) to the 3-subunit (well to moderately differentiated human colon cancer) with essentially preservation of the activity [ratio of activity of 3/ to 1/ = 0.81 (63)]. Although it is known that several / combinations are functional, 3/ complexes have a significant lower turnover number (about one-fifth to one-fourth) and lower affinity for sodium [8 and 25 mM for 1/ and 3/, respectively (26)]. Thus, by switching to the 3/ combination, Na+-K+-ATPase can be transformed to a system of decreased energy supply (13, 54), a system typical of tumor cells, with an activity that matches that of SLC5A1 expression.
The lower expression of the Na+-K+-ATPase -subunit might lead to reduced Na+-K+-ATPase activity, which, if sustained, might result in events that favor the progression of cancer. This is based on the crucial role that intracellular sodium homeostasis, regulated by Na+-K+-ATPase, plays in the development of tight junctions and induction of polarity in epithelial cells (54). Espineda et al. (32) showed that in E-cadherin-expressing MSV-MDCK cells, expression of the -subunit induced the formation of junctional complexes (e.g., tight junction and desmosomes), significantly reducing the motility and invasiveness of these cells (55). These studies revealed that the -subunit of Na+-K+-ATPase might be involved in the mechanisms leading to the suppression of invasiveness of carcinoma cells. It was suggested that increased -subunit expression might reduce the invasiveness of tumor cells, thus reducing the chance of tumor spread and, consequently, its recurrence.
In this regard, to evaluate the probability of metastasis, we compared our proteomic profiles with a gene set considered as a predictor of both metastasis and poor clinical outcome (56). This set was derived from the gene expression profiles of adenocarcinoma metastases of a variety of cancer versus unmatched primary adenocarcinomas. In our study, only one protein (of the 17 gene signatures) reported in Ref. 56 was found to be upregulated (elongation factor 4E-like 3), and, more importantly, this was found in normal instead of cancer tissue. This indicated that the colon cancer cases that were studied in the present work (despite the fact of having a low expression of the -subunit of Na+-K+-ATPase) had a low probability of metastasis, at the moment of sample collection, without excluding the involvement of other factors in the overexpression of elongation factor 4E-like 3.
Apoptosis. Proteins differentially expressed in colon cancer were BAP31, caspase-9, Apaf-1, and p53-dependent stress-inducible protein 27 (SIP27). BAP31 is ubiquitously expressed, involved in mechanisms of the regulation of apoptosis (51). BAP31 is highly enriched at endoplasmic reticulum membranes (14), where it forms both a homo- and hetero-oligomer with closely related BAP29 (14). BAP31 is a substrate of caspase-8 and can associate with procaspase-8 isoforms in response to apoptotic signaling/stimuli by the model oncogene E1A (14). The expression of BAP31 protein was confirmed by Western blot analysis performed on six of the eight cancer samples (Fig. 2). In agreement with our proteomic results, BAP31 protein levels were significantly higher in neoplastic tissue specimens than in normal tissue.
SIP27 is involved in the regulation of p53-related apoptosis, and its expression is dependent on p53 (73). SIP is a proapoptotic protein dependent on p53 activation, and it has been reported to be overexpressed in a variety of tumors. On the basis of this evidence, it was postulated that it acts as stress-induced gene with antitumor properties (73). The SIP overexpression in our samples of cancer suggests that p53 is functionally active.
Overexpression of proapoptotic BAP31 and p53-induced SIP27, in addition to caspase-9 and Apaf-1, might indicate that the mechanisms of apoptosis at these stages of cancer are still present and, probably, responsive to apoptotic stimuli.
This assumption is supported by the coexpression of neurofibromatosis type-2 protein (NF-2), which efficiently works as a tumor suppressor protein (3, 66) (Table 1). Taken together, these data suggest that colon cancer localized or confined to the intestinal wall may have an active defense mechanism against tumor progression. This underlines the importance of detecting cancer at these stages to favor apoptosis through therapeutic approaches. Results of clinical trials published in last decade on the efficacy of adjuvant chemotherapy in colon cancer are in line with these results.
Although these results on apoptosis might seem promising, they should be interpreted with caution, for not all caspases have been evaluated (i.e., initiators 4 and 5 and effectors 3, 6, and 8), and none of their activities were tested. This is important because as the correlation between transcript and protein levels might be misleading, the association between protein levels and activities might not be direct for critical events (such as posttranslational modifications and substrate and effector concentrations, among others) need to be considered for a comprehensive understanding of the operativity of a process.
Cell Cycle and Signaling
In this study, ubiquitin-conjugating E2 enzyme variant (UEV; Table 1) was found overexpressed in colon cancer tissues. This finding could be related to the role that this protein plays in in vitro differentiation and cell cycle (20). UEV constitutes a new subclass of proteins (originally named CROC-1) similar in sequence and structure to the ubiquitin-conjugating enzymes known as E2 enzymes (42, 60). UEV protein is downregulated upon differentiation of HT29-M6 cells (a tumor cell line derived from HT29); the constitutive expression of exogenous human UEV in these cells inhibits their capacity to differentiate upon confluence and causes both the entry of a larger proportion of cells in the division cycle and an accumulation in G2-M, accompanied by a profound inhibition of the mitotic kinase cdk1. Thus the fact that UEV has been found in tumor tissue suggests that at these stages of colon cancer (without evidence of extraorgan involvement), proliferation and cell differentiation might have been halted. More experimental evidence needs to be sought to confirm this possibility.
Proteins Related to Normal Functions of Colon Tissue
Several proteins related to colon functions were found differentially expressed in normal tissue. Of interest, insulin-like peptide 5 (INSL5), a member of the insulin/relaxin family, was also found overexpressed in the normal colon. Although the role of this protein is unknown, it could be speculated that it plays a role in gut motility based on the sequence homology with relaxin-3. INSL5 has been recently reported to be the ligand for G protein-coupled receptor 142 (GPCR142) based on the partial overlapping tissue expression profiles of INSL5 and GPCR142 (especially in the colon) combined with their high-affinity interaction. This recent finding will help to understand the role of INSL5 and those of other relaxin-related peptides.
Other proteins related to transcription (Zn finger protein and inducible cAMP repressor protein), translation (protein translation factor SUI1), and posttranslational (prefoldin) events and detoxification of xenobiotics/endogenous metabolites (glutathione transferase) were also found to be differentially expressed in normal colon tissue (Table 1).
Bearing in mind that these proteins were found in "normal colon tissue" (see definition under footnote 1) but still originated from cancer patients, it could be speculated that this tissue may already harbor molecular differences compared with normal mucosa from subjects with no evidence of colonic disease. If so, then some of these proteins could be used as biomarkers for colon cancer risk.
Conclusions
Our methodology permitted us to make unique observations. First, there is a marked difference in the expression of important, nonabundant proteins between normal and cancer colon mucosa. This difference could be exploited to develop new diagnostic strategies, as recently pointed out (83). One of the most relevant issues in proteomic analysis is that without cell fractionation or protein enrichment, abundant proteins (in general, housekeeping and structural proteins) are the most likely ones to be resolved and identified (58, 59). Accordingly, two important enzymes implicated in tumor growth by promoting angiogenesis, inducible cyclooxygenase-2 and nitric oxide synthase (inducible nitric oxide synthase), found to be upregulated in human colorectal cancer by a different experimental approach (22, 23) were not observed in our study. However, the experimental modifications introduced in this study allowed us to detect a significant fraction of differentially expressed proteins, especially some low-abundance ones (e.g, G proteins) and some membrane-bound ones (e.g., NADH-ubiquinone oxidoreductase B18 subunit), with profound implications for our understanding of cancer metabolism.
Second, these samples (from colon cancer at early or intermediate stages, all confined to the colon wall without metastasis) show the expression of proteins that are involved in mechanisms of defence against cancer, suggesting that a therapeutic strategy should be based on this differential biochemistry.
Finally, this study supports the evidence of a preference by cancer tissues to utilize glycolysis, as originally hypothesized by Warburg (79). This observation has been explained through a limited availability of substrates to mitochondria (62) or by a selective repression of the expression of the -subunit of ATPase (27), although our results indicate that more than one mitochondrial component may be altered during the progression toward tumorigenesis.
|
GRANTS |
|---|
|
TOPABSTRACTMATERIALS AND METHODSRESULTS AND DISCUSSIONGRANTSREFERENCES |
|---|
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
1 Normal colon tissue or normal mucosa was defined as macro- and microscopically normal-appearing tissue of nonneoplastic mucosa of colon cancer patients. However, adjacent "normal" mucosa (even at 10 cm of more of the tumor) may already harbor some aberrancies in protein expression compared with that from normal subjects without evidence of colonic disease. 
2 The majority of the proteins were originated from colonocytes. For this comparison, multiple spots from the same protein (probably with various posttranslational modifications) were counted once. The actual numbers of individual spots were 345 and 283 for normal and cancer tissues, respectively.
|
REFERENCES |
|---|
|
TOPABSTRACTMATERIALS AND METHODSRESULTS AND DISCUSSIONGRANTSREFERENCES |
|---|
