Glutamine Regulates the Human Epithelial Intestinal HCT-8 Cell Proteome under Apoptotic Conditions

Glutamine plays a key role in the metabolism of rapidly dividing cells, including enterocytes and lymphocytes, which may contribute to its beneficial clinical effects. Gut mucosal homeostasis is achieved through a balance between cell proliferation and apoptosis. In T cells, glutamine up-regulates antiapoptotic proteins and down-reg-ulates proapoptotic proteins. In gut mucosa, glutamine prevents apoptosis in rat epithelial cell lines, whereas glutamine starvation induces apoptosis through caspase activation. Finally glutamine specifically prevents tumor necrosis factor- (cid:1) -related apoptosis in the human intestinal cell line HT-29. Comparative functional proteomics enables the characterization of each differentially expressed protein in intestinal cells in response to modifications of nutritional environment. The influence of glutamine on intestinal proteome expression in apoptotic conditions has not been studied and evaluated. This comparative proteomics study was performed in the human epithelial intestinal cell line HCT-8 under experimental apoptotic conditions to investigate the influence of glutamine on protein expression during apoptosis. The phar-maconutritional effects of glutamine were determined under 2 m M (physiological concentration) and 10 m M (pharmaconutritional concentration) conditions. About 1,800 protein spots were revealed in both conditions. Comparative assessments indicated that 28 proteins were differentially expressed significantly ( i.e. at least 2-fold modulated and Student’s t test with p < 0.05) in response to an increase of glutamine

Glutamine plays a key role in the metabolism of rapidly dividing cells, including enterocytes and lymphocytes, which may contribute to its beneficial clinical effects. Gut mucosal homeostasis is achieved through a balance between cell proliferation and apoptosis. In T cells, glutamine up-regulates antiapoptotic proteins and down-regulates proapoptotic proteins. In gut mucosa, glutamine prevents apoptosis in rat epithelial cell lines, whereas glutamine starvation induces apoptosis through caspase activation. Finally glutamine specifically prevents tumor necrosis factor-␣-related apoptosis in the human intestinal cell line HT-29. Comparative functional proteomics enables the characterization of each differentially expressed protein in intestinal cells in response to modifications of nutritional environment. The influence of glutamine on intestinal proteome expression in apoptotic conditions has not been studied and evaluated. This comparative proteomics study was performed in the human epithelial intestinal cell line HCT-8 under experimental apoptotic conditions to investigate the influence of glutamine on protein expression during apoptosis. The pharmaconutritional effects of glutamine were determined under 2 mM (physiological concentration) and 10 mM (pharmaconutritional concentration) conditions. About 1,800 protein spots were revealed in both conditions. Comparative assessments indicated that 28 proteins were differentially expressed significantly (i.e. at least 2-fold modulated and Student's t test with p < 0.05) in response to an increase of glutamine concentration in the culture medium. Twenty-four proteins were identified by mass spectrometry and associated databases. From these proteins, 34% are involved in cell cycle and apoptosis mechanisms, 17% are involved in signal transduction, and 13% are involved in cytoskeleton organization. These data were integrated in a proposed schema of the interactome under apoptotic conditions. In conclusion, this study provides the first holistic picture of proteome mod- Glutamine (Gln) is the most abundant free amino acid in the body and is essential for cell growth (1). As a non-essential amino acid, Gln is produced in sufficient quantities by most organs in physiological conditions but becomes a limiting, conditionally essential amino acid in severe catabolic states, such as major surgery, trauma, or sepsis (2). In such cases, endogenous Gln stores and synthesis decrease rapidly, resulting in Gln depletion, which is correlated with a poor prognosis (3). Recent studies have demonstrated the efficacy of Gln parenteral or enteral supplementation in reducing complication rates in critically ill and postoperative patients (4). Gln decreases bacteremia and enhances gut barrier function by stimulating enterocyte proliferation and intestinal protein synthesis (5). Gln is an energy substrate for most cells and conducts mitochondrial adenosine triphosphate formation. Gln also constitutes a precursor for nucleotides, glutamate, and especially glutathione synthesis (6). Gln is thus considered as the main respiratory fuel and metabolic precursor for rapidly dividing cells such as enterocytes and immune cells (6).
Gut mucosal homeostasis is achieved through a balance between cell proliferation, located in the crypt of Lieberkü hn, and cell elimination by apoptosis, predominantly localized in both the crypt and villus compartment (7). In case of inflammatory bowel disease, the balance between proliferation and apoptosis is deregulated in inflamed or ulcerated area of gut mucosa (8).
At the molecular level, apoptosis is mainly orchestrated by the activation of the intracellular cysteine aspartate protease (caspase) cascade (9). Caspases are ubiquitously present in cells as latent precursors and constitute the molecular machinery for apoptosis (10). Two main apoptotic pathways leading to the activation of caspases have been described: the intrinsic pathway involving cytochrome c released from mitochondria and the extrinsic pathway induced by several antibodies or ligands interacting with surface-exposed death receptors such as TNF 1 receptor-1. The first pathway is dependent on the process of mitochondrial outer membrane permeabilization regulated by pro-and antiapoptotic members of the Bcl-2 family (11). Concerning the extrinsic pathway, death receptor ligation leads to the formation of the death-inducing signaling complex and subsequently activation of procaspase-8 by autocatalysis (9). Both apoptosis pathways are connected notably via the proapoptotic Bid protein. In fine, caspase activation, especially caspase-3 activation, induces specific morphological and biochemical changes associated with the apoptotic cellular phenotype.
Gln could modulate apoptosis-related cellular mechanisms (12,13). Indeed Gln up-regulates Bcl-2 and CD45RO antiapoptotic protein expression in lymphocytes and down-regulates the expression of Fas and Fas ligand proapoptotic proteins in the human T lymphocyte cell line Jurkat (14). In addition, Gln decreases both caspase-3 and caspase-8 activities in activated T cells (14).
The basic mechanisms of apoptotic signal transduction consist in quantitative and qualitative modifications of protein expression (10). Proteomics study of apoptotic cells allows us to determine the major biochemical actors and biomarkers of apoptosis under various environmental and pharmaconutritional conditions (15) including Gln supplementation (10). Consequently this study was designed to provide a holistic picture of protein modulations in Gln-treated (Gln concentration at 2 and 10 mM) human intestinal epithelial HCT-8 cells under CH11-induced apoptotic conditions.

EXPERIMENTAL PROCEDURES
Cell Cultures-Triplicate cultures of the human intestinal epithelial adenocarcinoma cell line HCT-8 (number 90032006, European Collection of Animal Cell Cultures, Salisbury, UK) were used between passages 44 and 46. Cell culture reagents, Dulbecco's modified Eagle's medium, Gln, non-essential amino acids, and fetal calf serum were supplied by Eurobio (Les Ulis, France). Tissue culture plates were obtained from ATGC (Noisy Le Grand, France). Cells were routinely grown as described previously (16). Culture medium without serum was used for stimulation for 24 h. Apoptosis was mediated with the mouse agonistic anti-Fas IgM monoclonal antibody CH11 used according to the manufacturer's instructions (Immunotech, Marseille, France). Cells were treated for 24 h with 125 ng/ml CH11 in 2 mM Gln (control) or in 10 mM Gln (Gln-treated), covering the range of physiological and pharmacological concentrations in these cells (16). Cells were washed three times with ice-cold PBS and isolated by scraping and centrifugation (360 ϫ g for 5 min) at room temperature.
Protein Sample Preparation and Proteomics Assays-All reagents and instruments used have been described in detail previously (16).
Protein quantification was performed with the 2-D Quant kit (Amersham Biosciences). Total protein lysates (450 g) were loaded and resolved using commercial 18-cm IPG (3.5-10) strips (Amersham Biosciences) for the first dimension. The second dimension was performed on vertical polyacrylamide (Sigma-Aldrich) 8 -16% (w/v) gradient slab gels (18 cm ϫ 16 cm ϫ 1.5 mm) and was run with a Laemmli SDS discontinuous system. Proteins were detected after Coomassie Brilliant Blue G-250 (Sigma-Aldrich) staining. The gels were destained for 2 h in 10% (v/v) methanol (Sigma-Aldrich) and 1% (v/v) acetic acid (Sigma-Aldrich) solutions. Gels were run in triplicates. The replicated gels were scanned (600 dots/inch resolution, 16 bits), and images were obtained with a computing and calibrating imaging densitometer (Model GS-700, Bio-Rad) and digitalization software (Multi-Analyst PC Software for the Bio-Rad Image Analysis Systems version 1.1, Bio-Rad).
Image Analysis-Differential analysis was performed using the Im-ageMaster TM 2D Platinum software version 5.0 (Amersham Biosciences) for spot detection, quantification, matching, and comparative analysis on a set of three simultaneously run gels per sample. The expression level was determined by the relative volume of each spot in the gel and expressed as %Volume (%Vol ϭ (spot volume/Αvolumes of all spots resolved in the gel)). This normalized spot volume takes into account variations due to protein loading and staining by considering the total volume over all the spots present in the gel. For each condition, two classes of gels were defined, namely class A for control samples (2 mM Gln) and class B for treated samples (10 mM Gln). Only groups of spots present in all gels were considered for differential analysis.
Statistical Analysis-To test the influence of 10 versus 2 mM Gln conditions, groupwise statistical comparisons were performed on groups of spots determined from triplicate gels both in 2 and 10 mM. Variations in abundance were calculated as the ratio of average values of %Vol for a group between the two classes. Only spots with a %Vol variation ratio greater than 2 (i.e. with significance set at Ϯ2-fold change) in the ImageMaster 2D Platinum group report were considered relevant. The corresponding p values were determined using the Student's t test (significance level, p Յ 0.05) after spot %Vol log transformation. This process converts the distribution of %Vol from a non-normal distribution to a normal distribution. The corresponding prerequisites for this statistical test were carefully checked (17).
Automated In-gel Digestion of Protein, MALDI-TOF Mass Spectrometric Analysis, and Database Search Based on Peptide Mass Fingerprint Spectra-The 28 protein spots were automatically excised by cutting out from the gel using the Ettan TM Spot Picker (Amersham Biosciences) and digested on the Ettan Digester (Amersham Biosciences) with proteomics grade porcine trypsin (Sigma-Aldrich). After digestion the peptide mixtures were analyzed with a Voyager-DE TM MALDI-TOF mass spectrometer (Applied Biosystems, Foster City, CA) as described previously (16). All mass experiments were acquired in triplicate with ␣-cyano-4-hydroxycinnamic acid (LaserBio Labs, Sophia Antipolis, France) as matrix. The matrix peak relative to the CHCA A 5 K 3 H ϩ form at 1060.0900 Da was excluded for peptide mass fingerprint analysis. External calibration was carried out using the peptide mixture "Proteomix-Peptide calibration Mix4" (LaserBio Labs). Moreover the tryptic autodigestion ion peaks (842.5100, 1045.5642, 2225.1202, and 2678.3822 Da) were used as internal standards to validate the external calibration procedure. This method typically resulted in mass accuracies of 80 -120 ppm. For interpretation, each acquired spectrum was processed using Data Explorer version 4.0 (Applied Biosystems) with an advanced base-line correction (peak width, flexibility, and degree were fixed respectively at 32, 0.5, and 0.1) and noise removal (standard deviation to remove noise fixed at 2). A monoisotopic peptide mass list prepared from spectra 1 The abbreviations used are: TNF, tumor necrosis factor; 2-D, two-dimensional; IB, inhibitor of nuclear factor B; LMP, lysosomal membrane permeabilization; NF-B, nuclear factor B; PCD, programmed cell death; TGF-␤, transforming growth factor-␤; COMMD, copper metabolism gene MURR1 domain; ICAT, ␤-catenin-interacting protein 1; IKK, IB kinase; MAP3K7, mitogen-activated protein kinase kinase kinase 7.
was generated manually and put in the MS-Fit ProteinProspector search engine version 4.0.5 (developed in the University of California San Francisco Mass Spectrometry Facility). When using accurate masses from MALDI-TOF spectra as input data, Swiss-Prot release 50.0 of May 30, 2006 as database, a 100-ppm mass accuracy, trypsin as enzyme, allowance of one incomplete proteolytic cleavage site, cysteine carbamidomethylation, acrylamide-modified cysteine, methionine oxidation, selection of species (Homo sapiens), and similarity of pI and relative molecular mass were specified in the search program.

Gln-induced Effects on the Apoptotic HCT-8 Cell Pro-
teome-Approximately 1,800 protein spots per gel were detected within a pI range of 3-10 and a relative molecular mass range of 10 -150 kDa. 2-D gels obtained from cells exposed under apoptotic conditions in medium containing 2 or 10 mM Gln are presented in Fig. 1, A and B. Twenty-eight protein spots were differentially expressed. The proteins were identified with MALDI-TOF MS corresponding to 24 protein entries because four spots could not be identified with these spec-trometric methods. Table I summarizes the identification results of these 28 proteins. Twenty protein spots from the apoptotic HCT-8 10 mM Gln class were decreased as compared with the control 2 mM Gln apoptotic HCT-8 cells. Eight protein spots were significantly up-regulated in 10 mM Glntreated HCT-8 cells.
Among the 28 spots influenced by Gln, 24 could be identified and classified. These proteins account for three major cellular functions. Pie charts (Fig. 3) representing the distribution of the Gln-altered proteins under apoptotic conditions indicate that 34% of the 10 mM Gln-regulated proteins (Fig. 3)

FIG. 1. 2-D gel images visualized by Coomassie Brilliant Blue staining representing total proteins extracted from HCT-8 under apoptotic conditions in 2 mM Gln (A) and 10 mM Gln (B).
The numbered spots were determined by statistical analysis and correspond to the proteins analyzed by MALDI-TOF MS and associated databases. Protein identification results are depicted in Table I.

Comparison of HCT-8 cell line proteome under apoptotic conditions in response to 10 mM Gln treatment
Spot numbers with corresponding Swiss-Prot database accession number, the protein name, the main cellular function, the theoretical molecular mass and pI values, the spectrometric percentage of peptide/sequence coverage, and the modulation level are shown. App, appearance; Disp, disappearance; PRKX, protein kinase X. Supplemental Data-Supplemental material is available in the on-line version of this article, including detailed spot %Vol measurements, statistical analysis, and mass spectrometry data relative to the 28 HCT-8 proteins significantly modified by Gln supplementation under apoptotic conditions. DISCUSSION The present study has demonstrated that high Gln concentration (10 mM), under apoptotic conditions, regulates the expression of several specific proteins mainly implicated in cell cycle, protein biosynthesis, and cytoskeleton organization processes, i.e. overall in cell life span and programmed cell death (PCD). A previous study has shown that Gln supplementation prevents spontaneous apoptosis in rat small intestine-derived epithelial cells, whereas Gln starvation induces apoptosis by caspase-3 and caspase-8 sequential activation (18). Moreover Gln prevented TNF-␣-related apoptosis-inducing ligand effects in the human intestinal cell line HT-29 (19). The Gln-derived pyrimidine pathway appears to be tightly implicated in this process (20). Gln also prevents caspase-8 activation in human intestinal epithelial T84 cells cultured with the proapoptotic Clostridium difficile toxin A (21). Another potential mechanisms of Gln action is the pathway of glutathione, an antioxidant compound that detoxifies reactive oxygen species and has potent antiapoptotic effects (22,23). However, although the glutathione pathway is widely implicated in the antiapoptotic action of Gln on immune cells, Evans et al. (19) have demonstrated that this was not the case in the intestinal HT-29 cell line. Finally the protective effects of Gln on gut mucosa may be related to the induction of cytoprotective proteins, such as the heat shock protein (Hsp) family (24). For instance, Hsp-72 transcript and protein expression were up-regulated by Gln in IEC18 intestinal cryptlike epithelial cells during heat shock and apoptotic injury (25).
In our model, we induced apoptosis by using the mouse agonistic anti-Fas IgM monoclonal antibody CH11, which activates the Fas extrinsic pathway leading to death-inducing signaling complex constitution and to caspase-8 and caspase-10 activations. Extrinsic and intrinsic pathways are interconnected through considerable cross-talk leading to cell death (9). In particular, the Bid protein activates the mitochondrial pathway after cleavage by caspase-8 and finally amplifies the apoptotic process (9). The intrinsic pathway involves the formation of the heptameric apoptosome complex. In fact, once cytochrome c is released from mitochondria, it can bind the Apaf-1 protein and then recruit and activate procaspase-9 (11). The cell death regulator Aven, a novel antiapoptotic member, can bind both Bcl-xL and Apaf-1 (26). In the baby hamster kidney fibrosarcoma BHK cell line and in the human breast adenocarcinoma MCF-7 cell line, Aven interferes with the ability of Apaf-1 to self-associate and inhibits PCD induced by Apaf-1 and caspase-9 (26). Aven is broadly expressed and conserved in mammalian species (26). In addition, Aven enhances the protective antiapoptotic function of Bcl-xL in chinese hamster ovary cells (27). The present study demonstrated that 10 mM Gln up-regulates Aven expression and may therefore inhibit apoptosome formation and the endstep of mitochondrial-mediated apoptosis.
Extrinsic and intrinsic pathways, via caspase-8 and apoptosome, respectively, finally lead to the cleavage of procaspase-3. Once activated, caspase-3 orchestrates apoptosis through the cleavage of key substrates resulting in demolition and packaging of dying cells (28). Actually we observed in the present study that 10 mM Gln down-regulated the caspase-3 precursor expression in the HCT-8 cell line. In agreement with this finding, a previous study has shown that Gln inhibits activation of caspase-3 in Jurkat T cells stimulated with phorbol myristate acetate plus ionomycin (14). Moreover in the HT-29 human carcinoma cell line, Gln prevented the activation of caspase-3 in response to treatment with the TNF-␣-related apoptosis-inducing ligand (19).
Mitogen-activated protein kinase kinase kinase 7 (MAP3K7), also called transforming growth factor-␤ (TGF-␤)activating kinase 1 (TAK1), is a key enzyme of the TGF-␤ pathway involved in the regulation of cell proliferation and differentiation in numerous cell types. Additionally MAP3K7 has been shown to modulate the Wnt/␤-catenin signaling pathway via the nemo-like kinase activation in Xenopus embryos and human embryonic kidney HEK293 and DLD-1 hu-man colon cancer cell lines. In these models, MAP3K7 stimulates nemo-like kinase activity and inhibits the interaction of the ␤-catenin and T cell factor-lymphoid enhancer factor nuclear factor transcriptional complex with deoxynucleotide nucleic acid (29 -31). Actually aberrant accumulation of ␤-catenin protein due to mutations of either the ␤-catenin or adenomatous polyposis coli genes plays an essential role in the development of colorectal carcinoma (32). In the present work, Gln supplementation up-regulated MAP3K7 expression, and this effect may be involved in the modulation of HCT-8 extrinsic apoptosis by Gln. Moreover we observed that Gln down-regulated ␤-catenin-interacting protein 1 (ICAT) expression. ICAT negatively regulates Wnt signaling via ␤-catenin and T cell factor interaction inhibition (33). Because ICAT is overexpressed in almost half of colorectal carcinomas (33), nutritional modulation of ICAT could be an interesting approach to prevent the activation of inflammation leading to metaplasia and finally to cancer.
Furthermore we found that Gln down-regulated the expression of the COMMD (copper metabolism gene MURR1 domain) proteins 2 and 6. These proteins belong to the recently discovered COMMD protein family. COMMD proteins are extensively conserved in multicellular eukaryotic organisms and establish multimeric complexes involved in NF-B transcriptional inhibition (34,35). In physiological conditions, NF-B is sequestered in the cytoplasm through an interaction with a family of inhibitory proteins, the inhibitor of nuclear factor B (IB). IB proteins are phosphorylated by the IB kinase (IKK) complex. After ubiquitinylation, IB proteins are rapidly degraded by the proteasome, and this finally allows nuclear translocation and activation of NF-B (36). NF-B stimulates cell proliferation and limits PCD in different cell types, including rat IE6 cells (37,38). Several studies have demonstrated that MAP3K7 activates IKK following TGF-␤1 treatment in murine hepatocytes (39) or TNF-␣ treatment in HeLa cells (40). Thus, Gln appears to positively modulate the antiapoptotic effects of NF-B both by inducing the MAP3K7/IKK pathway and by decreasing the expression of both COMMD 2 and 6 NF-B inhibitors.
Stathmin, or oncoprotein 18, is a ubiquitous cytoplasmic protein implicated in the dynamic regulation of microtubules. Stathmin promotes microtubule depolymerization during interphase and late mitosis and is regulated by changes in phosphorylation levels during cell cycle progression (41). Accordingly stathmin phosphorylation inhibits the protein microtubule destabilizing activity (41). Stathmin expression is upregulated in a wide variety of human cancers such as acute leukemia and breast and liver cancer (42). Recently stathmin has been found to be activated by the apoptosis signalregulating kinase 1-p38 cascade in rat pheochromocytoma PC12 cells (43). Here we observed that stathmin expression was down-regulated in HCT-8 cells, suggesting that Gln supplementation potentially modulates microtubule stabilization under apoptotic conditions. Autophagy consists in wrapping a part of the cytoplasm into a specific membrane and to its later digestion by lysosomal enzymes. Depending on the cellular context, autophagy can promote cell survival or induce cell death in response to endogenous and exogenous stress, such as apoptotic effectors. It is generally accepted that during these cell death conditions, lysosomal membrane permeabilization (LMP) induces cytotoxic effects and is related to activation of the mitochondrial cell death pathway (44). Actually the proapoptotic protein Bid can be cleaved by cathepsins or lysosomal proteases in HeLa cells and human breast cancer MCF-7 cells (45,46). Consequently Bid seems to be implicated as a potential mediator connecting LMP and autophagy (PCD type II) to apoptosis (PCD type I) processes. In addition, after TNF-␣ induction, LMP requires the caspase-8-mediated pathway in mice hepatocytes (47). Interestingly in our model, the expression of a protein involved in autophagy, Atg5, was markedly down-regulated by 10 mM Gln, suggesting that Gln supple-mentation can negatively modulate not only PCD type 1 (apoptosis) but also PCD type 2 (autophagy-mediated cell death). Autophagy is notably controlled by an evolutionarily conserved protein family, the autophagy proteins (Atg) (48). Sixteen Atg proteins are involved in the autophagosome biogenesis. Autophagosome represents the initial vacuole that sequesters cytoplasmic material before its delivery to the lysosome, and the (Atg12-Atg5-Atg16) 4 tetramer complex is required for autophagosome expansion and completion (48). Moreover previous studies demonstrated the involvement of the Atg5 protein in autophagic cell death (49,50). Accordingly the interaction between Atg5 and FADD (Fas-associated protein with death domain) has been shown to play a crucial role in interferon-␥-induced cell death (51). A recent study has demonstrated that Atg5 constitutes a substrate for calpains 1 and 2 in several cell types (52). Actually the truncated form of Atg5 (relative molecular mass of 24,000 Da) tightly interacts with Bcl-xL, leading to antiapoptotic protein inactivation (49, 52). Thus, we demonstrate for the first time that Atg5 protein expression and consequently autophagic cell death mechanisms can be modulated by Gln supplementation. These data provide crucial information to understand the regulatory pathways of this recently discovered proapoptotic protein.
Altogether our results provide new interesting leads to explain the apoptosis suppressive capacities of Gln. The differentially expressed proteins described above are tightly implicated in all of the apoptotic processes including the extrinsic (MAP3K7 plus downstream effectors) and intrinsic apoptosis pathway (Aven), autophagy (Atg5), and cytoskeleton regulation (stathmin). Based on these results and a bibliographic analysis, the Gln-modulated protein data described above can be integrated to provide a global schema of the intestinal interactome under apoptotic conditions (Fig. 5) depicting the relevant protein interactions and the subcellular pathways involved that ultimately contribute to the cytoprotective effects of Gln. Finally our work introduces new ways of investigation for further research involving the antiapoptotic mechanisms of Gln in intestinal cells and may lead to the discovery of new biomarkers for the understanding of inflammation as well as cancer initiation and progression.