Animals and diets.

Mice (C57Bl/6J) and rats (Wistar) were housed in standard conditions and fed a standard diet. Animals were then either kept on a standard diet (20%), or switched to a protein-deficient diet (0% casein added) or high-protein diet (40% casein). The modified standard diet AIN93G was maintained isocaloric by adjusting the starch content (Kliba-Nafg, Kaiseraugust, Switzerland). Mice or rats housed in groups (3–4 per cage) were fed experimental diet for 8–12 days (8–10 days for mice and 9–12 days for rats). Animals were weighed daily, and experiments were always run at the same time to minimize circadian variations. All procedures for animal handling were according to the Swiss Animal Welfare laws and approved by the Kantonales Veterinäramt Zürich.

Amino acid quantification by ultra-performance liquid chromatography.

The analysis was performed at the Functional Genomics Center Zurich. Samples were deproteinized with sulfosalicylic acid (5% final concentration), and amino acid concentrations were determined using the MassTrak (pancreatic juice, plasma) or AccQTag (cell culture) Amino Acid Analysis Solution method (Waters, Milford, MA) according to the manufacturer’s instructions.

Protein and DNA measurements.

Protein content was measured by a modified Lowry method (Bio-Rad, Hercules, CA) and DNA using fluorescence quantitation kit (Sigma-Aldrich, St. Louis, MO) according to the manufacturer’s instructions.

Pancreatic juice collection.

Pancreatic juice collection was performed as described before (29). Briefly, rats were starved for 12 h and anesthetized with isoflurane, the pancreatic duct was localized, and the upper right loop of the duodenum was fixed. The biliary duct was cannulated and deviated to the lumen of the small intestine with a small incision on the wall. The pancreatic duct proximal to the sphincter of Oddi was cannulated with polythene tube (0.4-mm inner diameter and 0.8-mm outer diameter). Pancreatic juice was collected on ice. At the end of the collection, heart blood was collected and a higher dose of isoflurane was applied to euthanize animals. Every day pancreatic juice of three rats, one each diet, was collected. For experiments where the effect of cholecystokinin peptide CCK-8 and secretin was analyzed, a jugular catheter (blunt PE50) was used. Prewarmed saline infusion at 1 ml/h was maintained constant. Bolus (0.5–1 ml) of secretin (Bachem, Bubendorf, Switzerland) at a dose of 3 pmol/100 g and the cholecystokinin peptide CCK-8 (Bachem) at 125 pmol/100 g was used. In this experiment, pancreatic juice was collected every 15 min and the volume and protein content were measured. For the ultra-performance liquid chromatography measurement, samples were pooled. For the basal phase, all four samples were pooled. After CCK-8 treatment, the two samples with the highest protein content were pooled, whereas after secretin injection, the two samples with highest volume were pooled.

RNA extraction and real-time quantitative PCR.

RNA was extracted with RNeasy kit (Qiagen, Venlo, The Netherlands). Approximately 2 mm³ of frozen pancreas were homogenized with MagNa Lyser Green Beads (Roche, Basel, Switzerland) for 1 min (2 × 30 s, 6,000 rpm) or for cells ~1 × 10⁵ AR42J cells were homogenized by passing through a 20-gauge needle. RNA quantity was analyzed by ND-1000 NanoDrop UV-spectrophotometer (NanoDrop Technologies Wilmington, DE). RNA quality was analyzed by microchip analysis Agilent 2100 Bioanalyzer (NanoChip; Agilent Technologies, Santa Clara, CA). Total RNA (20 ng/µl reaction) was used as template for reverse transcription with the TaqMan Reverse Transcription Kit (Applied Biosystems, Foster City, CA) according to the distributer. Quantitative PCR (qPCR) was performed using 10 ng of pancreatic or small intestine cDNA as a template. The reaction was carried out with the Taq-Man Universal PCR master mix (Applied Biosystems) using a Prism 7700 cycler (Applied Biosystems). The abundance of the target messenger RNAs was calculated relative to a housekeeping gene [hypoxanthine phosphoribosyltransferase (HPRT)] or 18S ribosomal RNA (18S) and expressed as relative expression (R = 2^{[Ct(housekeeping gene) − Ct(test)]}), where Ct is the cycle number observed at the threshold. Primers that have not yet been published (11, 38, 42, 55) are listed in Table 1.

Acinar cell isolation.

Pancreata from mice or rats were excised, washed in cold HBSS (calcium and magnesium free), freed from fat and mesenteric tissue, rapidly cut with scalpel, and added to the digestion medium. The tissue was incubated for 20 min in 10 units collagenase NB8 (Serva, Heidelberg, Germany) in HBSS supplemented with 0.1 mg/ml trypsin inhibitor, 8.7 mM CaCl₂, and 25 mM HEPES pH 7.4. The digestion was stopped by addition of 50 ml of cold HBSS supplemented with 5 mg/ml BSA and 22 mM HEPES pH 7.4. For RNA extractions, the cell suspension was subsequently filtered using 100- and 40-µm nylon cells strainers (Corning, New York, NY). The pellet was washed in cold HBSS and snap frozen. For uptake studies, acini were washed and resuspended in cold BSA-Krebs-Tris solution (1 mg/ml BSA, 118 mM NaCl, 4.7 mM KCl, 1.2 mM KH₂PO₄, 1.2 mM MgCl₂, 1.25 mM CaCl₂, and 20 mM HEPES pH 7.4). For secretion experiments, only the 100-μm strainer was used and acini were washed and resuspended in cold extracellular solution supplemented with 1 mg/ml BSA (140 mM NaCl, 5 mM KCl, 2 mM NaHCO₃, 1 mM NaH₂PO₄, 1.2 mM MgCl₂, 1.5 mM CaCl₂, 3 mM glucose, and 10 mM HEPES, pH 7.4).

Gln, glucose, leucine, and palmitoyl oxidation rates and uptake of Gln in freshly isolated acini.

Uniformly labeled [U-¹⁴C]glucose (300 mCi/mmol), [U-¹⁴C]Gln (200 mCi/mmol), [U-¹⁴C]leucine (329 mCi/mmol), and [1-¹⁴C]palmitoyl (60 mCi/mmol) (Hartmann Analytic, Braunschweig, Germany) were added to a final concentration of 5 mCi/ml in 1 mM (with the exception of palmitoyl) unlabeled substrate to freshly isolated acini in 2-ml tubes. Inside the 2-ml tube, a 0.2-ml tube without cap, containing 100 µl trap solution (0.1 M KOH), was carefully added and the 2-ml tube was sealed with Parafilm in an air-tight manner. The tube was incubated at 37°C for 90 min under agitation (110 rpm). Following incubation, tubes were kept for 10 min on ice and opened carefully and 90 μl KOH solution were added to the scintillation cocktail for the measurement of carbon dioxide formation (CO₂). Cells were washed and used for DNA quantification. The data are expressed as nanomoles per CO₂ per hour per micrograms of DNA. Freshly isolated acini were also used for measuring the uptake of Gln. Cells were incubated for 5 min at 37°C in Krebs-Tris buffer containing 1 mM and 0.1 µCi/ml tritiated substrate. After being washed, cells were used to determine the uptake by liquid scintillation counting and for the measurement of protein and DNA content. Gln transport is expressed as nanomoles substrate per 5 min per micrograms of DNA. The protein and DNA measurements were also performed to estimate protein synthesis in acinar cells of mice under different diets (mg total protein/µg DNA).

Secretion of Glu in freshly isolated acini.

Freshly isolated acini were allowed to recover for 90 min in a cell culture incubator (37°C, 5% CO₂) with extracellular medium supplemented with 1 mg/ml BSA, 0.1 mg/ml trypsin inhibitor, and 2 mM Gln in the absence or presence of 6-diazo-5-oxo-l-norleucine (DON; 250 µM). The acini were washed, divided in 2-ml tubes, and resuspended in 1 ml of a solution containing BSA (1 mg/ml) (control) and 2 mM Gln, leucine, or alanine. For the acini treated with DON during the recovery, the secretion solutions were also supplemented with 250 µM of the glutaminase inhibitor DON. After a 1-h incubation at 37°C with agitation (100 rpm), tubes were spun (100 g) and the upper 500 µl of supernatant were collected. The cells were washed in cold extracellular medium and lysed in water and protease inhibitor. Glu secreted in the medium and in the cells was measured with the Amplex Red glutamic Acid/Glutamate Oxidase Assay Kit (Molecular Probes, Eugene, OR). The results are expressed as nanomoles Glu per micrograms of DNA.

AR42J culture for amino acid analysis and uptakes.

AR42J cells (CRL-1492; ATCC) were cultured as described by Sonda et al. (60), with F-12K Nut Mix Medium (GIBCO) supplemented with 20% fetal calf serum (FCS). For uptake experiments, 0.1 × 10⁶ cells per 3.8 cm² were plated and cultured for 6 days, followed by treatment with 50 nM dexamethasone for 24 h to induce differentiation. Afterwards, cells were incubated for 24 h in F-12K Nut Mix Medium-FCS in the absence or presence of 250 µM of DON, l-cycloserine (CySer), or aminooxy-acetic acid hemihydrochloride (AOA; Sigma-Aldrich). For ultra-performance liquid chromatography measurements, cells were washed with PBS and lysed with 100 µl ice-cold water containing protein inhibitor cocktail (2 µl/ml) (Sigma-Aldrich). For uptake measurements, cells were washed with amino acid free high glucose DMEM (Invitrogen, Basel, Switzerland). DMEM containing glucose (25 mM), Gln, or Glu and 0.1 µCi/ml of tritiated compounds (Hartmann Analytic) was used for 2 min uptakes at 37°C. After cells were washed with cold PBS, 120 µl of 1% SDS in water were used to scrape and sonicate cells. The lysate was used to determine uptake by liquid scintillation counting and normalization to DNA content. Transport is expressed as nanomoles per 2 min per micrograms of DNA.

Isolation of acinar cell zymogen granule vesicles.

Mouse pancreata were weighed and homogenized in 0.3 M sucrose (2 mM MOPS, pH 6.8, 1 mM EDTA, 0.5 mM PMSF, 10 μg/ml aprotinin, and 5 μg/ml pepstatin) using a Teflon pestle PYREX Potter-Elvehjem tissue grinder powered by a variable-speed rotary drill for nine strokes at 3,000 rpm. The homogenate was centrifuged 10 min at 750 g at 4°C, and the supernatant was filtered through a 20-µm nylon mesh (Millipore, Bedford, MA). The filtered extracts were centrifuged at 1,750 g for 20 min at 4°C, and pellets were resuspended in 1 ml homogenization buffer and further purified by separation with an equal volume of Percoll gradient [60% vol/vol Percoll (Amersham Biosciences, Piscataway, NJ), 0.28 M sucrose, 20 mM MOPS pH 6.8, 1 mM EDTA, 0.5 mM PMSF, 10 μg/ml aprotinin, and 5 μg/ml pepstatin]. After 30 min centrifugation at 60,000 g at 4°C, the white zymogen granule layer near the bottom of the tube was collected and washed. The quality of the freshly isolated zymogen granule vesicles (ZGVs) was verified by transmission electron microscopy (TEM). ZGVs were subsequently used for surface biotinylation and amino acid measurements. Zymogen granules and total pancreas were lysed in water containing protease inhibitor cocktail (2 µg/ml) and used for amino acid and protein measurements. The results are expressed as nanomoles of amino acid per milligrams of protein.

Preparation of ZGVs for TEM.

Isolated ZGVs were fixed in glutaraldehyde at 4°C, washed, and treated with 1% OsO₄ for 1 h, embedded with 2% agar, and dehydrated. Epon blocks were cut at a thickness of 70 nm and treated with 2% uranyl acetate and lead citrate. The slides were analyzed at the Center for Microscopy and Image Analysis of the University of Zurich using a TEM Philips CM100 (1994, 40–100 kV). Pictures were taken at ×17,500 magnification.

Surface biotinylation of ZGVs.

After purification, fresh ZGVs were incubated with 1 mM 2-aminoethyl)methanethiosulfonate hydrobromide (MTSEA-biotin; Sigma-Aldrich) for 30 min at 4°C. The biotinylated ZGVs were washed, lysed, and incubated with immobilized streptavidin agarose resin beads (Pierce, Rockford, IL) for 2 h at room temperature. The beads were washed (50 mM Tris·HCl pH 7.4, 100 mM NaCl, and 5 mM EDTA), and protein was eluted with 50 µl of 2× sample buffer with 5% β-mercaptoethanol and heated for 15 min at 65°C. Proteins were separated on polyacrylamide gel (8 or 10%) and transferred electrophoretically to PVDF membranes (Immobilon-P; Millipore). Total membrane lysate or crude synaptosomes were used as positive controls for each antibody as described in the respective figure legends. Membranes were incubated overnight at 4°C with Rab3D (Abcam, Cambridge, UK), EAAT1 (Santa Cruz Biotechnology, Dallas, TX), EAAT3 (Alpha Diagnostics, San Antonio, TX), and vGLUT3 (1:1,000; SYSY, Göttingen, Germany). Appropriate horseradish peroxidase-conjugated secondary antibodies (1:5,000; Promega, Madison, WI) and Luminata horseradish peroxidase chemiluminescent substrate (Millipore) were used for chemiluminescent detection using the luminescent image analyzer LAS-4000 (Fujifilm).

Everted intestinal ring uptake of Glu and leucine.

Uptake of radiolabeled amino acids was performed by the previously described methods that were modified to eliminate mucus (13, 35, 39). Briefly, the small intestine was removed from ~2 cm distal to the stomach and 1 cm proximal to the cecum. The proximal (jejunum) and distal (ileum) segments were washed with 50 ml of Na-free Krebs-Tris solution (with 100 mM NMDG substituted for NaCl) to remove secretions and remainder of digested food. Further washes with 25 ml DEAE (1.25g/25 ml soaked overnight), 75 units/ml heparin, and 1 mM DTT and Na-free solution at room temperature were performed. The length and weight of the segments were measured. Segments were everted and cut in ~1-cm rings. For uptakes, rings were incubated in prewarmed, oxygenated (Oxycarbon) Krebs-Tris buffer (pH 7.4) containing 0.01 or 1 mM Glu and 1 mM leucine (0.1 µCi [³H]/ml) for 1 min at 37°C. After rings were washed with ice-cold buffer, they were weighed and lysed with Solvable (Perkim Elmer, Waltham, MA) and the radioactivity was determined by liquid scintillation counting. Transport is expressed as picomoles of amino acids per minute per milligrams of wet tissue.

Sample preparation and Western blot analysis.

Crude pancreatic and small intestinal membranes from mice treated with diets containing 0 or 40% casein were prepared as described previously (55). The pancreatic and small intestine crude membranes (25 or 50 µg) were immunoblotted as previously described and membranes incubated with primary antibodies. Antibodies against glutaminase 2 (Prosci, Poway, CA), GPT1 (Abcam), Glud (Cell Signaling, Danvers, MA), B⁰AT1 (Pineda, Berlin, Germany), and EAAT3 (Alpha Diagnostics) were diluted to 1:1,000. Zonula occludens-1 (ZO-1) and occludin (Zymed, San Francisco, CA) were diluted 1:500. Mouse monoclonal anti-actin (Sigma) signal was used to normalize the signals from pancreatic samples, and mouse monoclonal anti-NaKATPase (Santa Cruz Biotechnology) was used for normalization of small intestine sample signals. Quantification of bands was performed using the free software ImageJ (https://rsbweb.nih.gov/ij/index.html).

Immunofluorescence and morphological analysis.

Both paraffin and cryosections were used. Mouse pancreas and rat small intestine tissue samples were fixed in 3% paraformaldehyde. After fixation small intestine tissue was prepared as paraffin blocks and the pancreatic tissue as cryoblocks, as previously described (55). Sections of paraffin and cryosections (5- and 9-µm thickness, respectively) were used for immunofluorescence and hematoxylin-eosin staining. Sections were incubated with primary antibodies (diluted in PBS, 2% BSA, and 0.04% Triton X-100) overnight at 4°C, with the exception of sections treated with the Ki67 antibody, which were incubated for 30 min at room temperature. Subsequently, sections were incubated with the appropriate secondary antibodies (Table 2) and mounted in Glycergel (DakoCytomation, Glostrup, Denmark). The hematoxylin-eosin staining was carried out using a standard protocol as described elsewhere (60). Sections were viewed on a Nikon Eclipse TE300 epifluorescence microscope (Nikon Instruments, Melville, NY) equipped with a DS-5M Standard charge-coupled device camera (Nikon Instruments) and acquired with NISElements (Nikon Instruments). Images were merged using Photoshop 9. The poststaining analyses of images for crypt and villi measurements and quantification of Ki67-positive cells were done using the Imaris software (Bitplane, South Windsor, CA).

Statistics.

Pooled data are shown as means ± SE (n), where n represents the number of independent observations. For statistical comparison, ANOVA followed by Bonferroni posttest or two-tailed t-test were performed using a statistical software package (Prism v. 4.0; GraphPad, San Diego, CA).

Acinar cells secrete Glu in pancreatic juice during constitutive and stimulated secretion.

Glu showed the highest accumulation in pancreatic juice in comparison to plasma levels. In constitutively secreted pancreatic juice of unstimulated anesthetized rats the concentration of Glu was approximately sixfold higher than in plasma, whereas the concentration of all other amino acids, but aspartate, was clearly lower than in plasma (Fig. 1A). To evaluate if hormonal stimulation of acinar and ductal cells influences the secretion of Glu in pancreatic juice, Glu concentration was measured before (basal) and after stimulation with cholecystokinin (CCK-8; acinar cells) and secretin (ductal cells). Effective stimulation of acinar cells by CCK was evaluated by measuring protein secretion. Ductal stimulation was confirmed by the increased fluid volume observed after secretin treatment (Fig. 1B). CCK stimulated not only protein release from ZGVs but also induced a significant increase in the secretion of free Glu. However, the additional stimulation of ductal cells with secretin did not further increase Glu secretion. These data demonstrate that Glu is constitutively secreted by acinar cells and acinar cell stimulation induces increased Glu secretion, while stimulation of ducts does not influence Glu secretion in pancreatic juice.

Glu accumulation in acinar cells is not the result of Glu transport.

We analyzed the expression of Glu and Gln transporters in freshly isolated acini and in the acinar cell line AR42J. The Glu secreted into pancreatic juice may result from its accumulation in acinar cells due to import from the extracellular space or metabolism. As previously described for total mouse pancreas (55), Glu transporters were found to have a low gene expression relative to housekeeping genes also in isolated rat acini and AR42J cells (Fig. 2A). Neutral amino acids can be transported into acinar cells by the sodium-dependent Gln and the neutral amino acid transporters SNAT3 (Slc38a3) and SNAT5 (Slc38a5) and the sodium-independent exchangers LAT1-4F2 (Slc7a5-Slc3a2) and LAT2-4F2 (Slc7a8-Slc3a2), which are localized at the basolateral membrane of acinar cells. In freshly isolated acini and AR42J cells, the expression of Glu transporters of the Slc1 and Slc7 families and Gln transporters of the Slc38, Slc1, and Slc7 families displayed similar expression patterns (Fig. 2A). Functionally, the accumulation of radiolabeled amino acids measured in freshly isolated acini and AR42J cells showed that Glu uptake was up to sixfold lower than that of Gln (Fig. 2, B and C), supporting the hypothesis that accumulation of Glu in pancreatic juice is not primarily due to Glu transport.

Acinar cells metabolize Gln to synthesize Glu.

Gln, which is highly accumulated in acinar cells, proved to be the favored substrate for energy production in these cells. The amount of carbon dioxide (CO₂) formed from Gln in freshly isolated acini was higher than the amount formed from glucose, palmitoyl or leucine (Leu) (Fig. 3A). Before entering the TCA cycle, Gln is metabolized to Glu, suggesting a high Gln to Glu metabolism in these cells. As shown in Fig. 3B, transcripts for both isoforms of glutaminase (GLS1 and GLS2) were present in acinar cells. The cytosolic and mitochondrial isoforms of the alanine (GPT1 and GPT2) and aspartate (GOT1 and GOT2) aminotranferases, and Glud were also expressed. The enzyme responsible for Gln synthesis from Glu, glutamine synthase (Glul) is also expressed in acinar cells (Fig. 3B). Although in isolated acini the expression of all the aforementioned enzymes was lower than in cultured cells, the relative expression patterns were similar. Additionally, AR42J differentiation is induced by dexamethasone, which is a known inducer of Glul expression (17) and may also increase the expression of other enzymes. To verify the participation of these enzymes in the synthesis of Glu, we pharmacologically inhibited GLS, GPT, and GOT in AR42J cells and measured the accumulation of Glu, Gln, Ala, and Asp. As depicted in Fig. 3C, inhibition of the glutaminases using DON significantly increased Gln and reduced Glu concentrations. The alanine aminotransferase inhibitor CySer induced only a nonstatistically significant reduction in Glu and Ala concentrations. The inhibition of aspartate aminotransferase using AOA did not change Glu or Asp concentrations (Fig. 3C). Since DON is structurally similar to Gln, we checked if it indirectly reduced Glu accumulation by competitively inhibiting transport (and therefore availability) of enzyme substrates. In the presence of DON, Gln, Glu, and glucose transport were not changed (Fig. 3D), supporting the conclusion that changes in intracellular Gln/Glu ratios were due to glutaminase inhibition. We also studied the accumulation and secretion of Glu in freshly isolated acinar cells. Gln addition to the incubation medium induced Glu secretion and increased cellular Glu accumulation. In the presence of DON, both the accumulation and secretion of Glu were reduced (Fig. 3E). We observed that cells supplemented with Ala also secreted and accumulated more Glu in the cells than the control, suggesting a role for GPT in Glu synthesis. Conversely, addition of Leu did not cause significant changes in cell content and secretion of Glu (Fig. 3E). These results support the conclusion that in acinar cells secreted Glu is predominantly synthesized from Gln.

Glu is not accumulated in the ZGV.

The secretion of free Glu occurs in unstimulated animals and was increased upon CCK stimulation as shown in Fig. 1B. To better understand the secretory mechanism of Glu, its possible accumulation in the ZGV was analyzed. Intact ZGVs (Fig. 4A) were used for Glu concentration analysis and the expression of Glu transporters in the ZGV membrane fraction. Glu, as well as Ala and Gln, was accumulated in the acinar cytosolic fraction and not in the ZGV (Fig. 4B). To enrich the ZGV membrane proteins, surface biotinylation of freshly purified ZGVs was performed before Glu transporter expression was analyzed by Western blot. Enrichment of the membrane fraction of the vesicles was indicated by the increase in the concentration of rab3D (specific ZGV membrane marker). With the use of MTSEA-biotin, the only Glu transporter detected in the ZGV membrane was the excitatory amino acid transporter EAAT1/GLAST-1 (Slc1a3) (Fig. 4C). Immunofluorescence staining suggests that EAAT1 was localized at the apical acinar cell membrane. The signal colocalized with the apical markers peanut agglutinin and actin-phalloidin. Although EAAT1 was immunoprecipitated with the ZGV membrane fraction, an intracellular EAAT1 signal was not observed in pancreatic tissue slices (Fig. 4D). These results could be explained by the presence of apical membrane material during the biotinylation or that the expression in ZGVs was too low to be detected by immunofluorescence. Taken together, these results support the conclusions that Glu was constitutively secreted and was not concentrated in ZGVs.

Glu is also accumulated in the pancreatic juice of animals fed a protein-deficient diet.

Animals fed a protein -ree diet had reduced pancreatic mass and total protein content (Fig. 5, A and B). This adaptation was caused by the absence of protein arriving to the intestine, since sugar and fat were present and the diets were isocaloric, as described in methods. Additionally, animals showed an expected loss of body weight compared with animals maintained under control (20%) casein diet (Fig. 5, C and D). Despite decreased body weight and pancreatic volume, secreted pancreatic juice Glu concentrations were comparable to control animals fed the normal-protein diet. Pancreatic juice Glu was slightly but significantly elevated in animals fed the high-protein (40% casein) diet (Fig. 5E), but neither the protein-free nor high-protein diet condition resulted in significant changes relative to the control animals in plasma Glu concentrations (Fig. 5F). Therefore, pancreatic juice Glu levels were not regulated by the plasma concentration changes or passive transport.

We then analyzed the expression of Glu and Gln transporters and the accumulation of Gln in freshly isolated acinar cells from animals maintained in protein-free and high-protein diets. In acinar cells, isolated from animals maintained on a protein-free diet, SNAT5 (Slc38a5) expression was significantly decreased more than twofold at the RNA and protein levels (Fig. 6, A and B), whereas other transporters were not affected. However, neither the uptake of Gln (Fig. 6C) nor its oxidative metabolism was changed (Fig. 6D). Similarly, the oxidation of glucose and leucine was unchanged (Fig. 6D). We analyzed the expression of glutaminase (GLS), alanine and aspartate aminotransferases (GPT and GOT), asparagine synthetase (ASNS), and the branched chain aminotransferase (BCAT2), since we showed (Fig. 3) that the accumulation and secretion of Glu are dependent on Glu synthesis. Acinar cell GLS2 and GPT1 mRNA and protein levels were upregulated in animals maintained on a protein-free diet, while the protein expression of Glu dehydrogenase (Glud) was decreased (Fig. 6, E and F). BCAT2 and ASNS levels were unchanged and Glul expression was increased (Fig. 6E). Together, these results suggest that exocrine pancreas of rodents kept under protein restriction for short periods reduced protein synthesis, preserved Gln uptake and oxidative phosphorylation rate, and maintained free Glu secretion. We suggest that Glu accumulation was maintained by upregulation of enzymes synthesizing Glu from Gln.

Small intestine enterocyte proliferation is decreased in animals fed a protein-deficient diet.

The mass of small intestine, crypts depth, and villi length were measured to provide an indirect measurement of the number of cells and cellular proliferation. For animals fed a protein-deficient diet, the intestinal mass of jejunum and ileum was only slightly and nonsignificantly reduced (Fig. 7A). Analysis of villi length indicated a reduction in cell number in the jejunum (Fig. 7B). Villi shortening can be caused by a reduction in crypt cell proliferation, an increase in cell cycle length, the differentiation and migration of the cells toward the tip of the villi, or an increase of anoikis (56, 73). In the animals receiving the protein-deficient diet, the depth of crypts was decreased in jejunum and ileum (Fig. 7C), but the proliferation marker Ki67 was reduced only in the jejunum (Fig. 7D) coinciding with a decrease in villi length. In addition, the expression of cyclins involved in the G₁/S checkpoint (cyclin E2 and D1) was also decreased in the jejunum of animals fed a protein-deficient diet (Fig. 7E). The rate of detachment induced apoptosis (anoikis) was verified by caspase 3 expression, which was lower in animals receiving the protein depleted diets (Fig. 7F). These observations suggest that in the animals receiving a protein-deficient diet the renewal rate of jejunal villi was reduced. The loss of epithelial barrier tightness is another alteration that can follow malnourishment. The mRNA and protein levels of the tight junction proteins occludin and ZO-1 were analyzed along the length of the small intestine (Fig. 7, F and G). We observed no changes on the mRNA level. At the protein level, ZO-1 expression was increased slightly in the ileum and occludin was unchanged. The mucus layer integrity was indirectly analyzed by the expression of secreted mucin (mucin 2). The expression of mucin 2 mRNA was higher toward the ileum and did not differ in the groups receiving different diets. This data agrees with the periodic acid-Schiff staining of the rat samples showing normal number and distribution of goblet cells (data not shown). Together these data suggest that there was no loss of barrier function.

The metabolism but not the transport of Glu is altered in the small intestine during protein restriction.

The Glu arriving into the lumen of the small intestine is transported into the enterocytes where it is metabolized for energy production or exported to the body. The Glu transporter EAAT3/EAAC1, a member of the excitatory amino acid family (Slc1a1), is the only transporter identified in enterocytes for dicarboxylic amino acids (36). This high affinity Na⁺-dependent transporter is localized at the apical membrane of the enterocyte, toward the crypts as depicted in Fig. 8A. EAAT3 does not colocalize with B⁰AT1, a neutral amino acids transporter localized at the tip of the villi. Uptake experiments with two concentrations of Glu [10 µM (Fig. 8B) and 1 mM (Fig. 8C)] in the presence and absence of Na⁺ were performed. Only slight nonsignificant diet-induced changes were observed. High-affinity Na⁺-dependent transport of Glu, likely reflecting EAAT3 activity, tended to be reduced in the jejunum under a protein-free diet whereas the unknown low-affinity Na⁺-dependent transport of Glu was slightly increased in the ileum. As a control, the Na⁺-dependent transport of Leu, attributed to B⁰AT1, was analyzed and also found to be unchanged (Fig. 8D). Interestingly, the RNA and protein expression of B⁰AT1 were reduced in the jejunum of animals fed the protein-free diet, while that of EAAT3 was increased (Fig. 8E). Even though we did not observe changes in transport capacity of Glu, the expression of enzymes involved in Glu metabolism was changed in enterocytes. As depicted in Fig. 9A, Glud protein expression was increased in enterocytes of animals fed a protein-deficient diet. The aminotransferases (GOT and GPT) expression at mRNA level had a tendency to decrease and Gln synthase increased in these animals (Fig. 9B). The changes in Glud expression suggest that although cells did not accumulate increased amounts of Glu, they more efficiently metabolized Glu to produce energy. Together these data suggest that during protein diet restriction, metabolism of Glu secreted from the acinar cells to the small intestine lumen was altered to maintain enterocyte energy production.

Secretion of free amino acids by the exocrine pancreas.

The presence of free amino acids in the secretion of exocrine glands, such as the salivary, lacrimal, and mammary glands, has been previously reported, but their role is not always clear (53, 66). In maternal milk, Glu concentration is higher than that of any other amino acid and almost tenfold higher than in plasma. Additionally, the amino acids Gln, alanine, and aspartate are also highly concentrated in milk (58). In pancreatic juice, Glu and aspartate, but not neutral amino acids, were more concentrated. Pancreatic Glu secretion occurred during basal and increased under cholecystokinin induction (CCK-8 peptide), suggesting ZGVs could be involved in the accumulation and release of Glu, as proposed for β-cells (28, 43, 49). In β-cells, the vesicular transporter (vGLUT3/Slc17a8) and EAAT2/Slc1a2 were suggested to accumulate Glu and regulate pH in the vesicles (26). However, EAAT2 expression is very low and its ablation did not change Glu accumulation and release (74). As for islets, we also detected a member of the sodium-dependent excitatory amino acid family (EAAT1/GLAST1/Slc1a3) in the ZGV membrane. However, by immunofluorescence we could not confirm its localization to the zymogen granules; rather it was localized in the apical membrane. Since the transport stoichiometry and mechanism do not support a role in Glu efflux, the physiological role of EAAT1 in acinar cells remains unclear. As previously shown for other solutes (30), Glu secretion may take place via active transport on the apical plasma membrane. Therefore, a still unknown carrier, active during basal and stimulated secretion, may be responsible for the secretion of Glu by pancreatic acinar cells.

Origin of Glu secreted by the exocrine pancreas.

We found that the exocrine pancreatic secretion is rich in Glu consistently with previous reports by Fukushima et al. (25) and that it originates from the metabolism of neutral amino acids. Similarly to exocrine pancreas, it has been show by others that the lactating mammary cells secreted Glu is synthesize from neutral amino acids. Mammary cells indeed avidly take up Leu that is metabolized to Glu by the branched chain amino acid aminotransferase (BCAT) (49). In contrast, we show here that in pancreatic acinar cells, although BCAT is expressed, Leu did not stimulate Glu secretion. Schachter and Buteau (57) also demonstrated that in pancreatic slices the Glu produced from Leu was accumulated intracellularly and that the secreted Glu originates mostly from Gln. We additionally show here that Leu was not an efficient source of energy for acinar cells that preferred to utilize Gln in comparison to Leu, glucose, or palmitoyl. Consistent with this observation, intravenously administered Gln was shown to accumulate rapidly in pancreas (14, 51). Furthermore, we demonstrated here that acinar cells express both enzymes (Gls and Glud) required for Glu synthesis from Gln and for further channeling into the TCA cycle. In addition, the inhibition of glutaminase efficiently decreases Glu secretion. Thus Glu accumulation in the lactating mammary gland and in acinar cells results from intracellular synthesis and not from active transport of the excitatory amino acid. That acinar cells use a more abundant substrate for the synthesis of Glu is compatible with the composition and period of the secretion. In contrast to milk, pancreatic juice is produced constitutively throughout life and does not accumulate neutral amino acids or Gln. The fact that both milk and pancreatic juice ultimately reach the lumen of the small intestine highlights the potentially important role of Glu for enterocytes and leads us to propose that Glu is utilized in energy transfer.

Maintaining Glu secretion during protein restriction.

During total parenteral nutrition and protein-deficient diet, the exocrine pancreatic volume and the synthesis of digestive enzymes are diminished (3, 18). However, despite the reduction of pancreas volume and protein synthesis, the secretion of free Glu was unchanged in our animals receiving a protein-deficient diet. The fact that Glu secreted by acinar cells mainly results from imported Gln suggested that maintaining the expression of neutral amino acid transporters may be important to sustaining Glu secretion during malnutrition. We report here that an acute depletion of dietary proteins did not influence neutral amino acid transporters expression in acinar cells, with the exception of a strong decrease of SNAT5/Slc38a5. SNAT5 reduced expression did not impact on Gln uptake. This lack of effect may be explained by its transport mode and the presence of other transporters. It has been demonstrated in astrocytes that SNAT5 favors Gln efflux rather than influx (75). Additionally, Gln can be accumulated in acinar cells by the other transporters displaying various kinetic modes [symporters and antiporters (exchangers), sodium dependent and independent] and affinities (µm to mM range). The mechanism by which SNAT5 expression is regulated during protein restriction and its effect on acinar cells are unknown. In highly proliferating cells, SNAT5 was shown to be directly targeted by cMyc, increasing its expression (67). Other members of the SNATs (Slc38) family were suggested to play a role in signaling under starvation and cell volume changes via the m mammalian target of rapamycin pathway (15, 47, 54). It was also recently shown that depletion of two substrates of SNAT5, Gln and serine, activates the tumor suppressor p53 (10, 40). p53 Induced by amino acid depletion increases the expression of prosurvival genes, for example, Gls2 (9, 53), which was also increased in pancreas of animals maintained on a protein-deficient diet. The increase in Gls2 expression and activity is suggested to support cell survival by increasing NADH production, ameliorating the redox balance of the cells and providing protection from reactive oxygen species. Additionally, Glu participates in glutathione synthesis and feeds the TCA cycle (34, 68, 69). Gls2 increased expression could also explain the maintenance of Glu synthesis and secretion under protein restriction. Whether there is a connection between SNAT5 downregulation and the prosurvival mechanism activation of Gls2 expression in acinar cells during protein restriction is not known. Based on our results, we suggest that acinar cells adapt to protein depletion and amino acid availability via metabolic regulation rather than via increasing amino acid transporter expression.

Fate of Glu in the small intestine.

We found that Glu transport was maintained and the expression of the enzyme Glud increased in enterocytes of animals fed protein-deficient diet. Luminal Glu is the most efficiently metabolized substrate in enterocytes, although arterial glucose and Gln also play roles in providing energy to intestinal epithelial cells (12, 46, 52). Boutry et al. (6) suggested that the high enterocyte metabolism of luminal Glu, rather than luminal or arterial Gln, spare Gln as a metabolic fuel. In this context, the release of Glu from pancreatic juice may be important for maintaining intestinal homeostasis under protein starvation. The site of the highest Glu metabolism may be the villi, where enzymes involved in the metabolism of Glu are localized (72). The only transporter for Glu identified in small intestine enterocytes is the high-affinity sodium-dependent transporter EAAT3 (Slc1a1), which is however localized toward the crypts (22, 36). Therefore, Glu absorption may involve other unknown carrier(s) as was recently proposed by Fanjul et al. (23). That the absorption of Glu and Leu were unchanged in our animals challenged with protein-deficient diet may be due to the high transport capacity of small intestine enterocytes. The absorptive capacity of Glu in the intestine was estimated to be fourfold higher than the normal amount ingested daily (37). For that reason, minor changes in transporters may not cause significant changes in total or in sodium-dependent transport. Although absorptive capacity was not changed, a reduction in proliferation and the continuous renewal of the epithelia were observed, suggesting that cells changed from a highly proliferating to a more quiescent state under protein restriction. The physiological transition from proliferating to quiescent cells involves metabolic adaptations, and Glu is central for these adaptations. In quiescent epithelial cells, Glud expression is increased and Glu metabolism is driven toward the production of energy. In the proliferative phase, aminotransferases metabolism of Glu is favored additionally generating nonessential amino acids to support protein synthesis (16). In our animals receiving a protein-deficient diet, the increase in Glud expression and reduced cell proliferation suggests that enterocytes adapted their metabolism to the temporary nutrient deprivation. As we also observed in acinar cells, enterocyte adaptations under protein restriction altered metabolism rather than amino acids transport into cells. The possible advantage of this strategy may be to quickly accumulate substrates as soon as available without the energy costs of transcribing and translating new transporters.

In summary and as depicted in Fig. 10, we suggest an interorgan relationship between exocrine pancreas and small intestine for Gln-Glu utilization. This involves the uptake and metabolism of Gln in acinar cells, the secretion of Glu into the lumen of the small intestine and the uptake and use of Glu by enterocytes. As recently showed by Colloff et al. (16), epithelial cells may modulate their Gln-Glu metabolism to cope with physiological needs. Likewise, our results suggested that under protein restriction the exocrine pancreas Glu secretion is maintained by upregulation of Glu-synthesizing enzymes and in the enterocytes Glu utilization by upregulation of Glu-metabolizing enzyme without changes in transport capacity of amino acids. Additional studies are necessary to identify the secretory mechanism of Glu and further characterize the up- and downstream mechanisms controlling the metabolic adaptation in acinar cells.