Introduction

Glutamine (Gln) is one of the key nutrients for the intestine and also has growth factor-like signaling functions, activating a number of genes involved in cell-cycle progression in the intestinal mucosa [1]. In intestinal cell models, Gln prevents increased permeability after fasting [2] or treatment with acetaldehyde [3] and stimulates amino acid transport after ischemia-reperfusion injury [4].

Gln transport from the lumen into the enterocyte is a predominantly Na+-dependent process regulated by the availability of several brush-border membrane transporters. These transporters may be differentially expressed and regulated under pathological conditions and in response to growth factors. Their relative contributions might influence the clinical outcome. System B0, the major electrogenic Gln transporter, is a broad-spectrum neutral amino acid transporter. B0AT1 was recently cloned, and mutations in its gene cause Hartnup disease [5]. Another B0 transporter, B0AT2, is a poor Gln transporter and is not found in the intestine [6]. ATB0/ASCT2 (which will be referred to as ASCT2), belonging to the ASC family, has transport characteristics, tissue, and intracellular distribution similar to system B0, but is an obligatory exchanger and is less efficient in transporting phenylalanine and tryptophan than system B0 [7–10]. ASCT2 expression is elevated in a wide spectrum of primary human cancers [11]. The relative contributions of these two transporters to total Gln transport have not been established. We will, therefore, address this activity as B0/ASCT2. ASCT2, in addition to its transport function, is a receptor to retroviruses and to placenta syncytin and its accessibility to luminal retroviruses is an important factor in designing future gene therapies [12]. System B0, (B0AT1) may be absent from Caco-2 cells, a colon carcinoma cell line that is widely used as a model for the small intestine [13]. System B0,+ (protein ATB0,+), as opposed to system B0, transports both neutral and basic amino acids, their d isomers, and arginine analogs that inhibit NO synthesis. ATB0,+ is highly concentrative, driven by 2Na+ and 1Cl− [14, 15]. ATB0,+ is upregulated in the colon of short bowel syndrome (SBS) and cancer patients [16, 17], suggesting that the colon can become an organ of net nutrient uptake during stress. System A is a highly regulated, ubiquitous system for small neutral amino acids. The predominant variant of the enterocyte is ATA2, located primarily on the basolateral membrane [18]. Recently, it was reported in abstract form that rabbit villi and crypts express Gln transport systems SN1 and SN2 and that these transporters are differentially regulated by inflammation [19]. In the current study, we will refer to the measured transport activity as the activity of a system, and when the transporter protein is studied, the name of the protein will be used.

Parenteral and enteral epidermal growth factor (EGF) enhanced enterocyte Gln transport in some animal models [20, 21]. We have characterized the transport of Gln and other amino acids in brush-border membrane vesicles (BBMV) after 70% jejunoilieal resection in the rabbit. Despite compensatory villous hypertrophy, total Gln transport is downregulated between one and four weeks after resection. This is accompanied by a concomitant downregulation of B0/ASCT2 activity and ASCT2 protein and message. The provision of two-weeks parenteral EGF plus growth hormone after resection increased Gln transport, mainly due to an increase in system B0,+ [9, 22, 23]. The increase in the broader-spectrum highly concentrative B0,+ is advantageous in the acute stage of adaptation.

To define shorter-term effects of growth factors on Gln transport, we used a monolayer of C2BBe1 (C2) cells, an enterocytes-like brush-border-expressing subclone of Caco-2. These cells were exposed for 1 h to EGF and/or growth hormone. With combination treatment, activities of systems A and B0,+ were increased 3.5-fold, while that of system B0/ASCT2 was halved [24]. Short-term upregulation of transporters can occur as a result of triggers such as feeding and growth factors, and is most probably due to either activation or stabilization of transporters already residing in the apical membrane or due to trafficking of transporters from cytoplasmic pools to the surface membrane. Transporter trafficking is most frequently phosphatidyl inositol-3 kinase (PI3K)-dependent [25–28]. Rho, a GTPase, which is activated upstream of MAPKs in EGFR signal-transduction pathways, is implicated in stabilizing transporters residing in the surface membrane and in modulating trafficking of proteins in response to EGF [29, 30].

In this study, we examined the mechanisms involved in short-term EGF upregulation of Gln transport in C2 cells. In particular, we investigated the involvement of EGFR tyrosine kinase, Erk (a MAPK) and PI3K by using pharmacological inhibitors and that of Rho by using molecular inhibition. Our hypothesis was that EGF-induced Gln transport is due to Rho and PI3K-dependent increase in surface expression of Gln transporters. We have found that EGF increased Gln transport by increasing B0,+ and B0/ASCT2 activities, partially by increasing ASCT2 surface expression. However, the signal-transduction pathways leading to increased B0,+ activity varied from those involved in increased B0/ASCT2 activity and ASCT2 surface expression.

Experimental procedures

Reagents and chemicals

Protease inhibitors (aprotinin, pepstatin, and leupeptin), protease inhibitors cocktail, anti-β-actin antibodies, and all other highly graded chemicals were obtained from Sigma Chemical (St Louis, MO, USA). Radiolabeled L-{3H}-Gln was purchased from Amersham Pharmacia Biotech (Piscataway, NJ, USA). Sulfo-NHS-LS-biotin, immobilized streptavidin, and SuperSignal were purchased from Pierce Chemical (Rockford, IL, USA). EGF and antibodies against EGFR, EGFR phosphotyrosine 1173, phospho-Erk (pErk), Erk, phospho-Akt (pAkt), and anti Na+/K+ ATPase were from Upstate USA (Lake Placid, NY, USA). Antibodies against EGFR phosphotyrosine 1068 were from Biosource (Camarillo, CA, USA) and antibodies against Akt were from Cell Signaling Technology (Beverly, MA, USA). Antibodies against ASCT2 were custom made as described and characterized previously [8]. Antibodies against pan Rho proteins were from BD Transduction Laboratories (San Jose, CA, USA). Secondary antibodies were obtained from Jackson ImmunoResearch (West Grove, PA, USA). Custom-prepared double-stranded pan-Rho siRNA (5′-GACAGAAAUGCUUGACUUCTT-3′) was prepared by Integrative DNA technologies (Coralville, IA, USA). Fetal bovine serum was purchased from HyClone Laboratories (Logan, UT, USA). DMEM, transferrin, gentamycin, benchmark protein standards, lipofectamin-2000, Opti-MEM, and BLOCK-iT fluorescent Oligo were from Gibco–Invitrogen Corporation (Carlsbad, CA, USA). Tyrphostin AG1478 (AG1478), tyrphostin AG9 (AG9), PD98509 wortmanin, and LY294002 were from Calbiochem (EMD Biochemicals, San Diego, CA, USA).

Cell culture

C2BBe1 (C2) (ATCC number CRL-2102), a subclone of Caco-2 with enhanced brush-border expression (ATCC, Rockland, MD, USA), was grown as previously described [8]. Only passages 1–5 were used for the experiments in the current study [31]. Sub-confluent cells were split and sub-cultured on plastic until semi-confluent for siRNA transfection or until 1 day post-confluence for the rest of experiments. Cells were serum starved 24 h prior to experiments. Cultures were preincubated for 30 min with vehicle or the following concentration of signal transduction inhibitors: AG1478 or AG9 (for negative control) (2 μmol/l), wortmanin (100 nmol/l), LY294002 (50 μmol/l) or PD98509 (50 μmol/l). The cells were then treated for 0–60 min (as indicated) with EGF (100 ng/ml) or serum-free medium (vehicle, untreated control). Treated cells were then used for transport experiments and Western blot analysis or were biotinylated for ASCT2 surface expression analysis. Caco-2 (ATCC number HTB-37), a colon carcinoma cell line, were grown as monolayers according to ATCC recommendation until 1 day post confluent. To study polarized conditions, cells were seeded on clear polyester Transwell inserts (Corning Incorporated, Corning NY, USA) that were pre-equilibrated with medium for 1 h. Cell morphology examination after approximately 6 days confirmed formation of a monolayer with tight junctions. TEER measurements using EVOM (World Precision Instruments, Sarasota FL, USA) were >350 Ω cm2. Tight junction formation was also corroborated by exclusion of 3H inulin (data not shown).

Rho siRNA transfection of C2 cells

Semi-confluent cells were serum-starved in Opti-MEM without antibiotics for 24 h after which cells were washed. A mixture of lipofectamine-2000 with Rho siRNA or scrambled RNA (scrRNA) or fluorescent oligo was incubated according to manufacturer instructions for 20 min and then added to the washed cells to obtain a final concentration of up to 500 pmol siRNA or scrRNA. Transfected cells were grown for 48 after which EGF (100 ng/ml) or vehicle was added for 10 or 60 min. Gln transport was determined as described below. In addition, surface-expressed biotinylated ASCT2 and crude extract proteins were determined by Western blot analysis.

Biotinylation of monolayer apical surface proteins

Confluent monolayers on plastic preincubated w/wo signal transduction inhibitors were treated for 0–60 min w/wo EGF. They were then incubated on the apical side with 1 mg/ml Sulfo-NHS-LS-biotin diluted in ice-cold PBS buffer for 30 min at 4°C and washed with ice-cold 0.1 mmol/l glycine to quench the reaction and rinsed with ice-cold PBS buffer. Cells on membrane inserts were biotinylated either on the apical or basolateral side or on both, as mentioned above.

Transport experiments

Transport experiments were performed in modified cluster trays [24, 32]. Cells were washed twice with 37°C choline uptake buffer (each liter contained: 145 mmol choline chloride, 3 mmol K2HPO4, 1 mmol CaCl2, 1 mmol MgCl2, 10 mmol Hepes-Tris, 5 mmol glucose, pH 7.4). Warm (37°C) uptake buffer containing {3H}-Gln (100 μmol/l, 2 mCi/l) and 145 mmol/l of either sodium chloride or choline chloride (plus or minus excess of competitive amino acid inhibitors) was added simultaneously to each well. After 2 min, the cells were washed three times each with ice-cold choline uptake buffer and the cells were solubilized with 2 g/l SDS plus 0.2 mol/l NaOH. Half of each cell homogenate was re-suspended in Cytoscint scintillation cocktail (ICN Biomedicals, Irvine, CA, USA) and radioactivity was measured using liquid-scintillation spectrometry in LKB Wallac 1214 Rackbeta (San Francisco, CA, USA). Protein concentration determination was performed on the remaining homogenate. Gln uptake activity was expressed as nmol/min/mg protein or as a fraction of the measured control uptake activity. Na+-independent transport was defined in the presence of choline chloride. The Arg inhibitable fraction of this was calculated from the total transport in the presence of choline minus the transport in the presence of choline plus Arg. Na+-dependent uptake was defined as uptake in the presence of Na+ minus uptake in the presence of choline. The contribution of systems A and B0,+ were determined by deducting the transport in the presence of the competitive inhibitors MeAIB (5 mmol/l, totally inhibits only Na+-dependent Gln transport by system A) or Arg (5 mmol/l, totally inhibits only Na+-dependent Gln transport by system B0,+) respectively, from the total Na+-dependent Gln transport. B0/ASCT2 activity was the residual Na+-dependent Gln transport in the presence of 5 mmol/l MeAIB plus 5 mmol/l Arg. System N Gln transport was determined as the residual Na+-dependent activity in the presence of Na+ or choline w/wo cysteine (5 mmol/l, transported by all the above mentioned systems, but not by system N).

Protein concentration

Protein concentration was determined using bicinchoninic acid (BCA) protein assay reagent kit (Pierce Biochemicals, Rockford, IL, USA) according to manufacturer instructions. Absorbance at 545 nm was measured using Benchmark Plate Reader (Bio-Rad, Hercules, CA, USA). Bovine serum albumin served as standard.

Preparation of crude extracts and avidin-isolated biotinylated proteins

Cells were lyzed in modified RIPA buffer (10 mmol/l Tris, 150 mmol/l NaCl, 0.5% noneident p40, 1% Triton X 100, 1 mmol/l EDTA, pH 7.4) with protease and phosphatase inhibitors (PPI: 1 mmol/l activated sodium orthovanadate, sodium fluoride and PMSF, 10 μg/ml aprotinin, pepstatin and leupeptin and Sigma protease inhibitor cocktail diluted 1:10). For isolation of biotinylated proteins, lysates with biotinylated proteins (2 mg) were incubated overnight (∼16 h) at 4°C with 100 μl immobilized streptavidin (6% cross-linked agarose). The agarose beads were recovered by centrifugation (14,000g for 30 s). The supernatant was re-precipitated with another 100 μl immobilized streptavidin and the two precipitates were combined and washed five times with PBS. The supernatant was collected for the indicated experiments. Biotinylated proteins were eluted with 160 μl Laemmli buffer (0.06 mol/l Tris–HCl, 10% glycerol, 0.1 mol/l DTT, 1% SDS, and 0.0005% BPB). The lysates (crude extracts), isolated biotinylated proteins or proteins in supernatants after avidin precipitation and non-biotinylated cells lysates in Laemmli buffer were boiled for 5 min prior to SDS PAGE.

Western blot analysis

Crude extracts (40 μg/lane) or isolated biotinylated proteins (40 μl/lane) were analyzed by 7.5% SDS-PAGE (10% for Rho detection) according to the method of Laemmli [33]. Proteins were electro transferred to PVDF membranes, and the membranes were blocked by incubation for 1 h at room temperature with blocking solution (3% nonfat dry milk in TTBS (13 mmol/l Tris, 150 mmol/l NaCl, pH 7.5, containing 0.05% Tween-20). The membranes were incubated overnight at 4°C with the primary antibodies against EGFR phosphorylated on tyrosines 1173, total EGFR, pAkt and Akt (1:1,000); pErk, EGFR phosphorylated on tyrosine 1068 and ASCT2 (1:2,000); Rho (1/500); Erk 1/5,000 and β-actin (1/100,000) in blocking solution. Blots were washed three times (2 × 5 min; 1 × 15 min) in TTBS and incubated with the appropriate secondary antibodies usually diluted 1:5,000 except for ASCT2 (1/10,000), β-actin (1/50,000), and EGFR (1:5,000). Prior to visualizing the autoradiograms on blue XAR ALF film (Labscientific; Livingston, NJ, USA), blots were washed again as previously and incubated in SuperSignal for 1–40 min. Benchmark standards were used to determine apparent molecular weight. Because each antibody has its own sensitivity, comparisons of the intensities of the bands can be made only to each antibody’s own control and not among bands of different antibodies. β-actin, total EGFR, total Erk, or total Akt were used as loading controls on stripped blots as specified in the relevant experiments. When mentioned, the membranes were stained with ponceau S to verify equal protein loading. Non-saturated autoradiograms were scanned by Kodak Gel Logic100 Imaging System and densitometry was carried out using Kodak 1D image Analysis Software (Rochester, NY, USA). Membranes that were incubated without a primary antibody did not show any staining (data not shown).

Statistical analysis

Results are reported as means ± SEM. Data were analyzed by One-way ANOVA (with Newman Keuls post-test) on GraphPad Prism statistical software package. Differences were considered significant at P < 0.05. Each value represents the mean of at least triplicate measurements per experiment with (3–13) experiments as indicated.

Results

Transport systems involved in Gln uptake in C2 confluent monolayers

Total Gln transport was measured in the presence of Na+ and amounted to 0.55 ± 0.052 nmol/min/mg. Na+-independent transport contributed 21% to total transport. Arg-inhibited Na+-independent transport was half of the total Na+-independent transport (Fig. 1a), but was too low to determine the specific carrier systems y+, y+L, and/or b0,+. The study was not designed to differentiate Na+-dependent Gln transport systems ASCT2 from B0 (B0AT1), because the latter is not present in C2 cells [13]. These systems will be combined and will be referred to as B0/ASCT2. The major system for Na+-dependent Gln transport was B0/ASCT2 and amounted to 79% of the total Na+-dependent transport (Fig. 1b). Contribution by system B0,+ was 17%. System A contributed less than 2% (Fig. 2b). There was no detectable Na+-dependent Gln transport in the presence of cysteine indicating that the cells do not express system N (Fig. 2b).

Fig. 1
figure 1

Total, Na+-independent, and specific systems involved in Na+-dependent glutamine (Gln) uptake in C2 cells. Gln transport was measured in confluent C2 cells in the presence of Na+ or choline in the presence or absence of 5 mmol/l Arg, MeAIB, or both, or cysteine. Values are means ± SEM, (n = 10). (a) Total Gln transport 0.55 ± 0.05 nmol/min/mg was set at 1 and ratios were calculated. Total, total Gln transport; Na ind, Na+-independent Gln transport; Arg dep, Na+ independent/Arg dependent transport is the difference between the transport with choline to the transport with choline plus Arg. *P < 0.001 compared to all others: # P < 0.05 Na ind versus Arg dep. (b) The difference between Gln transport in the presence of Na+ and choline is Na+-dep, Na+-dependent Gln transport. Arg or MeAIB inhibitable Na+-dependent Gln transport is due to systems B0,+ or A, respectively. The residual transport in the presence of Arg + MeAIB is due to B0/ASCT2 transport. The residual transport in the presence of cysteine is due to system N. Means differ significantly at P < 0.001 except for system A versus B0,+ where the difference is at P < 0.05

Full size image
Fig. 2
figure 2

EGF-induced increase in total and B0/ASCT2 Na+-dependent Gln transport is eliminated by AG1478, wortmanin, or PD98509, while the increase in B0,+ is abolished only by PD98509. Confluent C2 monolayers were preincubated w/wo protein kinase inhibitors before 60 min treatment with EGF. Na+-dependent Gln transport activity was determined for (a) total, (b) B0/ASCT2 or (c) B0,+. For each system or for total Gln transport, the activity of the control was set at 1 and ratios were calculated. Values are means ± SEM, (n = 3–13). EGF, (E); Control with vehicle, (C); AG1478, (T); wortmanin (W); PD98509 (P). * P < 0.001 compared to all others. # P < 0.05 of E versus E+PD

Full size image

EGF increases Na+-dependent Gln transport by increasing systems B0/ASCT2 and B0,+ activities

One-hour EGF treatment did not increase Na+-independent Gln transport (data not shown). After 1 h of EGF exposure, total Na+-dependent Gln uptake was increased by 70% (P < 0.001) (Fig. 2a). This increase was due to 43% increase in system B0/ASCT2 (Fig. 2b) and 129% increase in system B0,+ (Fig. 2c) activities. System A and system N were not affected by EGF treatment (data not shown).

EGFR tyrosine kinase, Erk, and PI3K are involved in EGF-induced increased B0/ASCT2 activity, while only Erk is involved in EGF-induced increase in B0,+ activity

To determine EGFR signal-transduction pathways involved in the EGF-induced effects, C2 cells were preincubated with each of the following pharmacological signal transduction inhibitors: AG1478, an EGFR tyrosine kinase inhibitor; PD98509, a MEK inhibitor which inhibits phosphorylation of Erk; and wortmanin, a PI3K inhibitor. Each of the three inhibitors abolished total EGF-induced Na+-dependent Gln transport (Fig. 2a), mainly due to elimination of EGF-induced increase in B0/ASCT2 activity (Fig. 2b). Involvement of PI3K in EGF-induced increase in Gln transport was supported by inhibition of the increase by a second PI3K inhibitor LY294002. In three independent experiments EGF increased Gln transport by 50% and 50 μmol/l LY294002 abolished the increase (P < 0.01). The increase in B0,+ activity was not significantly reduced by AG1478 or wortmanin (Fig. 2c), indicating that EGFR tyrosine kinase and PI3K are probably not involved. In contrast, PD98509 eliminated the increase in B0,+ activity suggesting involvement of Erk.

ASCT2 intracellular location in C2

In C2 and Caco-2, anti-human ASCT2 antibody reveals a band of 81 ± 1 kDa which is completely abolished by preincubation with the peptide to which the antibody was generated (Fig. 3a). This molecular weight is similar to the one obtained for rabbit ASCT2 [8] and equivalent to one of the bands described in the rat eye lens [34]. In order to determine the location of ASCT2 within C2 cells, the cells were grown on membrane inserts and biotinylated on the apical or basolateral side individually or together. The avidin-isolated biotinylated proteins were subjected to Western blot analysis. A known basolateral membrane protein, Na+/K+ ATPase was expressed almost exclusively in the basolateral membrane fraction, indicating successful separation of the two membrane domains (Fig. 3b). ASCT2 was distributed unevenly, with higher amounts on the basolateral membrane (Fig. 3b). The supernatant fraction after exhaustive avidin precipitation of simultaneously biotinylated apical and basolateral membrane proteins also contained ASCT2 (Fig. 3b, cyt). This is consistent with other studies in the rat lens, human fetal astrocytes, and several cancer cells [34–36]. As the protein concentration applied to the gels and exposure times of blots are not equivalent, the data do not allow quantification of the relative amounts of ASCT2 on the surfaces and in the cytosol.

Fig. 3
figure 3

Identification of ASCT2 protein in C2 and Caco-2 cells and intracellular localization of ASCT2 protein in C2. (a) C2 and Caco-2 cells were grown as monolayers on plastic until confluent. Crude extracts were prepared and subjected to Western blot analysis (50 μg crude extract/lane) with antibody to ASCT2 in the presence or absence of 100 mol/l excess of the peptide against which the antibody was raised. (b) Post-confluent C2 cells on membrane inserts were biotinylated on the apical (Ap) or basolateral (Bl) surface or on both. Avidin isolated biotinylated proteins were subjected to Western blot analysis for ASCT2 and Na+/K+ ATPase. Ap, apical; Bl, basolateral: cyt, supernatant after precipitation of biotinylated proteins on both Bl and Ap sides (a) and (b) are representative experiments (n = 4)

Full size image

EGF increases ASCT2 surface expression in confluent C2 monolayers

We investigated whether the increase in B0/ASCT2 activity is accompanied by increased surface expression of ASCT2 protein. The confluent cells were exposed to 1 h apical EGF and then biotinylated on the exposed apical surface. The biotinylated fraction was isolated by avidin precipitation and subjected to Western blot analysis for ASCT2. The increase in ASCT2 surface expression in response to EGF was already evident after 5 min, was saturated after 10 min, and persisted for at least 1 h (Fig. 4). Thus, the 43% increase in B0/ASCT2 activity after 1 h EGF addition (Fig. 2c) was accompanied by a 153 ± 25% (n = 5, P < 0.003) increase in surface expression of ASCT2 protein (Figs. 4 and 5a). In contrast, there was no change over this time course in the total amount of ASCT2 in the crude extract (Fig. 4), indicating that de-novo synthesis of ASCT2 is not involved. Although actin is not accessible to surface biotinylation, it is coupled to accessible membranal proteins. Actin in the membrane fraction and in the crude extract served as a loading control. These results indicate that EGF-induced increase in Na+-dependent Gln transport is, at least, partly due to elevation of surface expression of ASCT2 protein.

Fig. 4
figure 4

Time course of EGF-induced increase in ASCT2 surface expression. Confluent C2 monolayers were incubated with EGF for the indicated times and apical membrane proteins were biotinylated. Crude extracts were prepared and biotinylated proteins were isolated by avidin precipitation. The proteins were subjected to Western blot analysis for ASCT2 protein or β-actin. A representative experiment (n = 4)

Full size image

EGFR tyrosine kinase, MAPK (Erk), and PI3K are involved in EGF-increased ASCT2 surface expression

To determine EGFR signal-transduction pathways involved, C2 cells were preincubated with AG1478, PD98509, or wortmanin before EGF treatment. The incubation was followed by biotinylation, avidin isolation, and Western blot analysis. The EGF-induced increase in ASCT2 surface expression was eliminated by each of the three inhibitors (Fig. 5a). Surprisingly, the inhibitors reduced ASCT2 surface expression below the control levels (Fig. 5a), while the transport activity was reduced to the control level (Fig. 2b). A possible explanation is that the remaining activity is due to system B0 or an ASCT2 variant which is not recognized by the antibody. Neither EGF nor EGF in the presence of the signal transduction inhibitors affected the total amount of ASCT2 in the crude extract (Fig. 5b). Together, the data suggest that EGFR, Erk and PI3K pathways are involved in EGF-induced increase in ASCT2 surface expression and that the increase is not due to de-novo synthesis of ASCT2 in the cells.

Fig. 5
figure 5

The Increase in ASCT2 surface expression induced by EGF is eliminated by AG1478, PD98509 or wortmanin. Confluent C2 monolayers were preincubated w/wo protein kinase inhibitors before 60 min treatment with EGF w/wo biotinylation and avidin isolation. Crude extracts (b) and avidin isolated biotinylated proteins (a) were subjected to Western blot analysis for ASCT2 and β-actin. Control, (C); EGF, (E); AG1478, (T); PD98509, (P); wortmanin, (W). (a) and (b) representative experiments, (n = 4)

Full size image

Activation status of the signaling proteins involved in EGF-induced increase of Gln transport and ASCT2 surface expression

We next determined the activation status of the involved signal-transduction proteins in response to EGF and the signal transduction inhibitors. Under any of the conditions, there was no change in total EGFR. EGF increased the phosphorylation of EGFR on tyrosines 1068 and 1173 and phosphorylation of pErk1/2. Basal amounts of pAkt (considered to be a PI3K downstream signaling protein) were detected, but there was no change in pAkt amounts in response to EGF (Fig. 6). AG1478 inhibited the phosphorylation of EGFR and pErk, but not of pAkt, while PD98509 and wortmanin did not inhibit EGFR phosphorylation, but partially reduced phosphorylation of Erk1/2 (Fig. 6). Total amounts of Erk and Akt did not change under any condition (Fig. 6). Vehicle or AG9 (a negative control for AG1478) did not affect transport or phosphorylation of signal-transduction proteins (data not shown).

Fig. 6
figure 6

Phosphorylation of signaling proteins in C2 treated w/wo EGF and w/wo inhibitors of EGFR tyrosine kinase (AG1478), MAPK/Erk (PD98509), and PI3K (Wortmanin). Confluent C2 cells were preincubated w/wo the signal-transduction inhibitors prior to incubation w/wo EGF for 10 min. Crude extracts were subjected to Western blot analysis for total EGFR (tEGFR), EGFR phosphorylated on tyrosines 1068 (pTyr1068) and 1173 (pTyr1173), pErk, total Erk, pAkt, and total Akt (a representative experiment, n = 4)

Full size image

Collectively, the data imply that EGF-induced increase in B0/ASCT2 Gln transport activity is due, at least in part, to increased ASCT2 surface expression and that both the activity and surface expression are dependent on functional EGFR tyrosine kinase, Erk, and PI3K pathways but not on Akt phosphorylation.

Silencing of Rho by Rho siRNA decreases Rho amounts and phosphorylation of Erk1/2

In order to investigate EGFR signal-transduction pathways upstream of Erk, we have chosen to inhibit Rho, a GTPase which was shown to be involved in transporters stabilization on the membrane and in EGFR trafficking [29, 30]. Semi-confluent C2 cells were transfected w/wo increasing concentrations of a pan-Rho siRNA or with scrRNA as a negative control. After 48 h, the cells were incubated with or without EGF for 10 min and the levels of Rho protein and total and pErk1/2, a Rho downstream signaling protein, were determined. Rho protein levels decreased only in cells transfected with Rho siRNA. The decrease was concentration dependent, was evident already at 50 pmol/ml of Rho siRNA and 500 pmol obliterated Rho protein expression (Fig. 7). As expected, Rho silencing decreased pErk1/2 in a concentration-dependent manner without any change in the level of total Erk1/2 or the housekeeping protein, β-actin (Fig. 7). A fluorescent oligo at a concentration of up to 500 pmol under identical conditions was transfected into maximum 80% of the cells without causing any cell lysis even after 72 h (data not shown).

Fig. 7
figure 7

Transfection of C2 with Rho siRNA reduces Rho and EGF-induced increase in phospho-Erk (pErk). Semi-confluent C2 cells were transfected w/wo 100 pmol scrambled siRNA (scr) or 0–500 pmol Rho siRNA. After 48 h, the cells were incubated w/wo EGF for 10 min. Western blot analysis on crude extracts was performed for Rho, pErk, Erk and β-actin proteins. A representative experiment (n = 3)

Full size image

Silencing of Rho by Rho siRNA decreases Na+-dependent Gln transport

In order to investigate the effect of Rho silencing on EGF-induced Na+-dependent Gln transport, we incubated semi-confluent C2 w/wo 500 pmol Rho siRNA or scrRNA for 48 h. The cells were then exposed to 60 min incubation with EGF and Gln transport was measured. Under these conditions the cells are only 70–80% confluent. Total Na+-dependent Gln transport was 0.59 ± 0.12 nmol/min/mg. It was higher than in confluent cells (0.41 nmol/min/mg, Fig. 1b). The basolateral transporter, system A, was barely detected (0.02 ± 0.04 nmol/min/mg) indicating that as in confluent cells only apical transport is measured. The distribution of the activity between the transport systems (Figs. 8a–c) was similar under these conditions to the distribution under conditions described for Fig. 1b. In these experiments, EGF increased total Na+-dependent Gln transport by 39 ± 1.8%. The lower increase of the transport by EGF in this set of experiments versus the increase described in Fig. 2 (70%), might be due to the higher activity at the baseline for the non-confluent cells. Baseline activity and the EGF-induced increase in Gln transport were similar between non-transfected control cells and the cells transfected with scrambled RNA. EGF-induced increase in Gln transport was eliminated by Rho siRNA, but not by scrRNA transfection (Fig. 8a). EGF increased B0/ASCT2 by 32% (Fig. 8b) and that of B0,+ activity by 127% (Fig. 8c). Rho silencing eliminated B0/ASCT2 (Fig. 8b) and B0,+ (Fig. 8c) activities induced by EGF. Together, the data indicate that silencing of Rho and decreasing Erk phosphorylation inhibit EGF-induced Gln transport by abolishing EGF-induced increase in systems B0,+ and B0/ASCT2.

Fig. 8
figure 8

Rho silencing inhibits EGF-induced total and system-specific Na+-dependent Gln transport. Semi-confluent C2 cells were transfected wo/w 500 pmol Rho siRNA or scrambled RNA. After 48 h, cells were exposed to EGF for 60 min and Gln transport was measured in the presence and absence of 5 mmol/l amino acid inhibitors. Total and the system specific transport activity of each of the controls (C) was set at 1 and ratios for each condition were calculated. Values are means ± SEM, (n = 3–4). (a) Na+-dependent total Gln transport. The transport under control conditions was 0.59 ± 0.12 nmol/min/mg. *P < 0.01 compared to all others, but not between those marked *. (b) B0/ASCT2 Gln transport. The transport under control conditions was 0.41 ± 0.11 nmol/min/mg. #P < 0.05 for EGF without si Rho RNA versus EGF with si Rho RNA. (c) B0,+ Gln transport. The transport under control conditions was 0.13 ± 0.03 nmol/min/mg. * P < 0.05 differ compared to all others. Control, non-transfected cells; scr RNA, transfected with scrambled RNA; si Rho RNA, transfected with si Rho RNA

Full size image

Silencing of Rho by Rho siRNA decreases ASCT2 surface expression

In order to investigate the effect of Rho silencing on ASCT2 surface expression, semi-confluent C2 were incubated w/wo 500 pmol Rho siRNA or scrRNA for 48 h. The cells were then exposed to 60 min incubation with EGF and then biotinylated. Proteins of the crude extracts and biotinylated proteins isolated by avidin were subjected to Western analysis. EGF-induced increase in ASCT2 protein surface expression was inhibited by Rho silencing but not by scrambled RNA transfection (Fig. 9a), while total ASCT2 amount in crude extracts was not affected by EGF treatment or Rho silencing (Fig. 9b). As expected, Rho amounts decreased in cells transfected with Rho siRNA, but not in cells transfected with scrambled RNA (Fig. 9c). Together, the data, similar to the results obtained with PD98509 inhibition, support the involvement of Erk in the EGF-induced Gln transport and ASCT2 surface expression.

Fig. 9
figure 9

Inhibition of EGF-induced increase in ASCT2 surface expression by Rho silencing. Semi-confluent C2 cells were transfected wo/w 500 pmol Rho siRNA or scrambled RNA. After 48 h, cells were exposed to EGF for 60 min. Crude extracts (b and c) or biotinylated isolated surface proteins (a) were subjected to Western blot analysis. (a) Isolated biotinylated surface proteins probed for ASCT2 and β-actin. (b) Crude extracts (CE) probed for ASCT2 and β-actin. (c) Crude extracts probed for Rho and stained for total protein on the membrane by Ponceau S staining. The sum of all Ponceau S-stained protein bands in (c) was set at 1 and the ratio calculated. C, control; scr, transfected with scrambled RNA; si, transfected with Rho siRNA. Representative experiment (n = 3)

Full size image

We conclude that Rho silencing decreased Rho protein and EGF-induced Erk1/2 phosphorylation in parallel to decreasing EGF-induced increase in total, B0/ASCT2 and B0,+ Gln transport, and ASCT2 surface expression. Collectively, these data suggest that Rho silencing decreased Gln transport, at least in part, by reducing ASCT2 surface expression in an Erk-sensitive manner.

Discussion

Gln becomes conditionally essential for the stressed enterocytes and serves as a growth factor and immune modulator [1]. As Gln transporters are decreased in several pathological conditions such as in short-bowel syndrome [9, 22], interventions are needed to increase their long-term activity by de-novo synthesis of transporters. However, under certain conditions, such as in response to feeding after fasting, salivation (saliva contains EGF), and ischemia reperfusion injury with swelling, short-term increases in Gln transporters are crucial [2, 4, 37]. We have investigated the mechanisms involved in short-term upregulation of Gln transporters by EGF in an intestinal cell line. We have found that EGF increases Gln transport by elevating B0,+ and B0/ASCT2 activities and ASCT2 surface expression through differential signaling pathways.

A weakness of the study is the lack of differentiation between systems B0 and ASCT2 activity. These two proteins’ substrate specificity is almost identical. B0 transports phenylalanine and tryptophan more efficiently, but these amino acids partially inhibit both B0 and ASCT2 transport [5, 7–10, 13]. Thus, identification of each of these transporters’ contributions to total Gln transport using competitive inhibition with amino acids is not feasible. We cannot exclude the possibility that although B0AT1, the protein responsible for B0 activity in the intestine, is absent from Caco-2 cells [13], it might be expressed in the sub-clone, C2, and contribute to the control levels and to subsequent EGF-induced increase in Gln transport activity.

Mechanisms which may be involved in increasing Gln transport by EGF are:

  1. 1

    de-novo synthesis;

  2. 2

    increasing surface expression; and

  3. 3

    activation of existing transporters.

The involvement of these mechanisms in Gln transport induced by EGF was investigated for the portion of the activity due to ASCT2 protein. Although de-novo synthesis of some transporters, for example Glut 4, can occur in response to a short-term growth factor treatment [38], de novo synthesis is not likely to operate in the current study. This is evident from the lack of EGF-induced increase in the total amount of ASCT2 protein (the major Gln transporter in C2 cells) as measured in the crude extracts of the cells (Figs. 4 and 5b).

An alternative mechanism is a short-term increase in surface expression of the transporters. For example, ASCT2 in human fetal astrocytes is mostly cytoplasmic, but in response to supplementation of Glu or Gln, it is redistributed to the membrane [35]. ASCT2 surface expression was increased in response to EGF (Figs. 4 and 5a) implying the increase in surface expression of this transporter as a major mechanism for increasing the activity.

Activation of surface existing transporters by EGFR signaling proteins was not directly investigated in the current study. An example of such a mechanism is p38 MAPK activation of serotonin transport activity via trafficking-independent process [39]. This mechanism is probably not involved in the EGF-induced transport activity due to B0/ASCT2 activity, because elimination of the EGF-induced increase in ASCT2 surface expression abrogated the increase in the transport activity, leaving no fraction of activity unaccounted for (Figs. 2c and 5a). On the contrary, not all membranal ASCT2 present after EGF treatment is active, as EGF increased B0/ASCT2 activity by 43%, while the surface expression of ASCT2 increased by 153% (Figs. 2c, 4, and 5a). As antibodies for B0,+ protein are not available, we did not conduct an investigation of the mechanisms involved in EGF-induced Gln transport due to this transporter.

The steady-state surface expression of transporters results from an equilibrium between endocytosis and recycling of the proteins. Increased ASCT2 surface expression in response to EGF could have been due to increased rate of trafficking from internal pools and/or to a decreased rate of endocytosis due to stabilization of the resident surface ASCT2. We used Rho silencing, because Rho is responsible for retention of Na+/H+ exchanger (NH3) in the apical membrane of epithelial cells [29]. Silencing of Rho abrogated the EGF-induced increase in ASCT2-dependent Gln transport and in ASCT2 surface expression. Although EGF-induced retention of ASCT2 at the membrane through Rho activation is possible, we cannot conclude from the current data whether a Rho-sensitive decrease in endocytosis or increase in transcytosis is responsible for the elevated ASCT2 expression on the surface membrane.

We have investigated a few of EGFR signal-transduction pathways which might be involved in EGF induced increase in Gln transport. The inhibition of EGF-induced increase in total and B0/ASCT2 transport and in ASCT2 surface expression by wortmanin or PD98509, individually, indicates that both Erk and PI3K are involved in Gln transport upregulation (Figs. 3 and 4). In the presence of a combination of the two inhibitors there was no additional increase in the inhibition of the transport and surface expression (data not shown), suggesting that Erk and PI3K use the same pathway(s). The partial inhibition of Erk1/2 phosphorylation by wortmanin may indicate that PI3K is activated by EGF upstream to Erk. This reinforces the suggestion that Erk and PI3K may use the same pathways(s). A similar relationship between Erk1/2 and PI3K was reported for the EGF-induced increased in organic anion transporter (OAT3) [40].

As expected, EGFR and Erk phosphorylation were inhibited by AG1478 and Erk phosphorylation was inhibited by PD98509 (Fig. 6). Akt (also called PKB) is often, albeit not always, activated downstream of PI3K by many triggers in different tissues and its phosphorylation serves as an indication of involvement of the PI3K pathway [41]. In C2 cell monolayers on plastic, EGF did not activate Akt (Fig. 6). In addition, we have previously found that, on membrane inserts, neither apical nor basolateral EGF activated Akt of C2 [42]. This is despite the involvement of PI3K in the EGF-induced upregulation of Gln transport and ASCT2 surface expression, as suggested by wortmanin inhibition of the EGF-induced increase in B0/ASCT2 activity and ASCT2 surface expression and by LY294002 inhibition of the increase in Gln transport. Although not expected, the lack of Akt participation in PI3K dependent pathway is not surprising. For example, the short-term EGF induced increase in glutamate transport in astrocytes is due to a PI3K-dependent increase in GLT1 (EAAT2) and in GLAST (EAAT1) activities and surface expressions. However, while the increase in GLT1 is dependent on Akt activation, that of GLAST is not [28]. It should be emphasized that dependence on a particular pathway for enhancing surface expression of proteins depends on the cell context. For example, the membrane abundance of human ASCT2 expressed in Xenopus oocytes is dependent, among other signaling proteins, on activation of Akt [43], while increasing the membrane expression of ASCT2 in C2 cells is Akt-independent (Fig. 6).

Upregulation of system B0,+ is advantageous to the cells because of its broad spectrum and its highly concentrative action [14, 15]. We were not able to detect mRNA and the protein of the cloned variant ATB0,+ [14] in C2 even after EGF treatment. These cells most probably contain an as yet unknown variant of ATB0,+. Thus, investigation of system B0,+ at the protein level was not possible. As EGFR activation is expected to underlie most effects of EGF, a surprising finding was the lack of inhibition by AG1478 of increased system B0,+ activity. This suggests that an increase in EGFR tyrosine kinase activity is not responsible for the transport augmentation. EGFR is a member of the Her/ErbB family of receptors, some of which may also be activated by EGF through forming heterodimers with EGFR [44]. Therefore, EGF activation of an ErbB receptor, other than activation of EGFR tyrosine kinase, may have started the signal-transduction pathway leading to Erk and B0,+ activation by EGF. Notably, although EGF-induced B0,+ activity was not significantly inhibited by AG1478 or wortmanin, the EGF-induced increase in total Gln transport under these conditions was abolished (compare Figs. 2a, b). This can be explained by a combination of two factors: First, the contribution of B0,+ to total Na+-dependent Gln transport, also in the presence of EGF, is low and amounts to only 20%. Second, although not statistically significant, both AG1478 or wortmanin reduced the EGF-induced increase in B0,+ activity to some extent (Fig. 2b).

Although trigger-induced short-term upregulation of transporters due to increased surface expression is often dependent on activation of PI3K [25–28], there are instances, for example adipocyte swelling-induced increase in Gln transport, where MAPK, but not PI3K are activated [37]. Similarly, the short-term increase in B0,+ activity was dependent on activation of Erk, but not PI3K, because PD98509 and Rho silencing through Erk inhibition, but not wortmanin, inhibited the increase.

Many signal-transduction pathways downstream of EGFR, such as p38 MAPK, may be involved in short-term transporters regulation [45] in addition to the PI3K and MAPK/Erk pathways which were studied here. Regulation of transporters trafficking and activity is a complex process which may involve intercalating pathways originating from activation by EGF of EGFR or other ErbB family members. In addition, long-term treatment with EGF, likely involves different signal-transduction pathways, such as MAPK or PLC/PKC with the end result affecting de novo synthesis of transporters [46, 47]. These complex processes are still under investigation.

In conclusion, short-term luminal EGF treatment of C2 cells increases total Gln transport by increasing system B0/ASCT2 and system B0,+ activities and ASCT2 surface expression. While B0/ASCT2 activity and ASCT2 surface expression are dependent on functional EGFR tyrosine kinase, Erk, PI3K and Rho, B0,+ is dependent only on functional Rho and Erk. Gln has several important functions in mammalian cells including: protein synthesis; as a source of nitrogen for nucleotides, amino sugars and glutathione synthesis; for regulation of acid–base homeostasis in the kidney; ureagenesis in liver; hepatic and renal gluconeogenesis; oxidative fuel for intestinal and immune cells; inter-organ nitrogen transport; and precursor of neurotransmitters. It is also an immune modulator and can serve also as a growth factor [1]. Elucidating the mechanisms by which growth factors such as EGF regulate Gln transport is crucial for designing interventions to upregulate its availability especially under conditions of malabsorption. Apart from the importance of ASCT2 as an amino acid transporter, delineating its mechanisms of regulation might be crucial for designing gene therapy, because it is a retrovirus receptor [12].