The intestinal peptide transporter PEPT1 provides bulk quantities of amino acids to epithelial cells. PEPT1 is a high-capacity and low-affinity solute carrier of the SLC15 family found in apical membranes of enterocytes in small intestine and distal colon. Surprisingly, murine PEPT1 (mPEPT1) has an apparent molecular mass of ∼95 kDa in the small intestine but ∼105 kDa in the large intestine. Here we describe studies on mPEPT1 protein glycosylation and how glycans affect transport function. Putative N-glycosylation sites of mPEPT1 were altered by site-directed mutagenesis followed by expression in Xenopus laevis oocytes. Replacement of six asparagine residues (N) at positions N50, N406, N439, N510, N515, and N532 by glutamine (Q) resulted in a decrease of the mPEPT1 mass by around 35 kDa. Electrophysiology revealed all glycosylation-deficient transporters to be functional with comparable expression levels in oocyte membranes. Strikingly, the mutant protein with N50Q exhibited a twofold decreased affinity for Gly-Sar but a 2.5-fold rise in the maximal inward currents compared with the wild-type protein. Elevated maximal transport currents were also recorded for cefadroxil and tri-l-alanine. Tracer flux studies performed with [14C]-Gly-Sar confirmed the reduction in substrate affinity and showed twofold increased maximal transport rates for the N50Q transporter. Elimination of individual N-glycosylation sites did not alter membrane expression in oocytes or overall transport characteristics except for the mutant protein N50Q. Because transporter surface density was not altered in N50Q, removal of the glycan at this location appears to accelerate the substrate turnover rate.
- peptide transport
- N-linked glycosylation
- Western blot
the intestinal peptide transporter PEPT1 (SLC15A1) mediates proton-coupled electrogenic epithelial influx of di- and tripeptides derived from luminal protein breakdown. PEPT1 is a prototype member of the proton-coupled oligopeptide transporter family found in all prokaryotes and eukaryotes (16, 17, 32). In addition to the transporters' nutritional role in providing bulk quantities of amino acids to circulation, the proteins also allow uptake of a variety of pharmacologically active compounds, including β-lactam antibiotics (24, 54), angiotensin-converting enzyme inhibitors (60, 35), and other peptidomimetics and prodrugs (11). Thus PEPT1 is important for the oral bioavailability of drugs.
A unique feature of PEPT1 is that it, with few exceptions, can transport about 400 different dipeptides and over 8,000 tripeptides (51). In rodent intestine, membrane abundance of PEPT1 decreases from proximal to distal (32) with highest levels found in epithelial cells of villus tips. PEPT1 is also expressed in astrocytes, in bile duct epithelial and renal proximal tubular cells, and nuclei and lysosomes of the exocrine pancreas and placenta (8, 39). The murine PEPT1 isoform consists of 709 amino acid residues and shows 85% sequence identity to its human ortholog (hPEPT1). On the basis of a membrane topology model derived from epitope mapping of hPEPT1 (14), the murine transporter is predicted to have 12 transmembrane-spanning domains (TMD) as shown in Fig. 1. Amino- and carboxy-termini are both facing the cytosol. With specific reference to the transmembrane helices 9 and 10, a large extracellular loop consisting of 202 amino acids protrudes (17). Previous studies demonstrated that PEPT1 has a distinct spatial expression also in colonic tissues of mice, rats, and humans with highest levels in distal colon (59).
Surprisingly, Western blot detection in colon revealed a consistently higher apparent molecular mass of the protein accounting to around 105 kDa, whereas, in the same animals, the protein in the small intestine possessed an apparent mass of around 95 kDa. Enzymatic deglycosylation in protein extracts of jejunum and colon yielded an identical protein mass, suggesting that observed PEPT1 mass shift is caused by N-glycans. Targeted screening for glycosylation sites present in murine PEPT1 delivered several N-X-S/T sequons as putative acceptor sites for N-glycans. N-X-S/T sequons consist of asparagine (N), a variable amino acid except proline (X), and serine (S) or threonine (T) and are important determinants for protein N-glycosylation. To probe whether these sites are used for PEPT1 glycosylation, we generated several mutant proteins by exchanging asparagines in sequons by glutamine. Using Xenopus laevis oocytes as a heterologous expression system, we assessed the changes in mPEPT1 molecular mass and surface expression. By two-electrode voltage clamp and by tracer flux studies, all mutant proteins were analyzed in oocytes for changes in the transport kinetics.
MATERIALS AND METHODS
Site-directed mutagenesis of the mPEPT1 gene.
Putative N-glycosylation sites in mPEPT1 (UniProt AC: Q9JIP7) were identified with the NetNGlyc 1.0 server and mutated by site-directed mutagenesis. Asparagines in positions 50, 406, 439, 510, 515, and 532 were converted into glutamine (N50Q, N406Q, N439Q, N510Q, N515Q, and N532Q) or glycine (N50G) residues (Table 1). To this end, two mutation-specific megaprimers, carrying specific restriction sites (EcoRV and XhoI), were amplified and subsequently coupled with phusion polymerase (New England Biolaboratories, Ipswich, MA). Primer combinations used for generation of megaprimers are listed in Table 2. All coupling reactions were performed at 25°C in a thermocycler (Biometra, Göttingen, Germany). Obtained constructs were digested with XhoI and EcoRV (Thermo Scientific, Waltham, MA) and cloned into pCRII-TOPO-3′ end vector (Invitrogen, San Diego, CA). Mutations were confirmed by sequencing (GATC Biotech, Constance, Germany).
Generation of cRNA.
cRNA synthesis was performed with the mMessage machine T7 kit (Ambion, Darmstadt, Germany) according to manufacturer's instructions. The expression vector carrying individual mPEPT1 constructs was digested with NotI. Following phenol extraction and ethanol precipitation, 1 μg linearized DNA was transcribed into cRNA and quality checked on an agarose gel containing 1% formaldehyde. Finally, all cRNA solutions were adjusted to a concentration of 1 μg/μl and stored at −80°C.
Xenopus laevis oocyte preparation and cRNA injection.
Xenopus laevis maintenance and oocyte harvest procedures were approved by the local authority for animal care in research (Regierung von Oberbayern, approval no. 55.2-1-54-2532.3-64-11). Oocytes were collected from frogs anaesthetized with 0.7 g/l 3-aminobenzoic acid ethyl ester (Sigma-Aldrich, St. Louis, MO) as described previously (36, 37). Separated and sorted oocytes (stage V/IV) were stored in Barth solution [88.0 mM NaCl, 1.0 mM KCl, 0.8 mM MgSO4, 0.4 mM CaCl2, 0.3 mM Ca(NO3)2, 2.4 mM NaHCO3, 10 mM HEPES, pH 7.4], containing 5.0 mM pyruvate and 0.2 mM gentamycin at 17°C overnight. After injection of 36.8 nl cRNA consisting of wild-type (WT) murine sodium glucose transporter 1 (mSGLT1) and mPEPT1 constructs in a ratio of 1:1 or water as a control, oocytes were incubated for protein expression over a period of 3–4 days at 17°C. For assessing tunicamycin effects, mPEPT1 WT cRNA (18.4 nl) was coinjected with 50 μg/ml tunicamycin (Sigma-Aldrich) dissolved in 4% DMSO providing a final concentration of 2.5 ng/oocyte (31).
Two-electrode voltage clamp (TEVC) experiments were conducted according to Amasheh and Kottra et al. (1, 36). Oocytes were placed in an open chamber and continuously superfused with Barth solution in the absence or the presence of 0.3–10.0 mM glycyl-sarcosine (Gly-Sar) or 1–20 mM cefadroxil (Sigma-Aldrich) with a flow rate of 3 ml/min. Current and potential electrodes, backfilled with 0.5 mM KCl and an electrode resistance between 1–3 MΩ, were used to voltage clamp the oocytes at −60 mV. Current flow was calculated from the difference between baseline after rinsing oocytes with Barth solution and reaching a plateau phase in the presence of substrate. Current-voltage (I-V) relations were recorded with a Tec-03 amplifier (Npi Electronic, Tamm, Germany) for the duration of 100 ms in the potential range of +80 to −160 mV. Transport currents were normalized to recording of currents generated by 1 mM α-methyl-d-glucopyranoside (pH 6.5) as substrate for mSGLT1 expressed as a reference transporter. After normalization of transport currents, kinetic constants Km and Imax as maximal transport current were determined by submitting data to a Michaelis-Menten equation with approximation by the least-square method.
Protein preparation from Xenopus laevis oocytes.
Three to four days after cRNA injection, 20–30 oocytes expressing the different transporter genes were transferred into 200 μl lysis buffer (20.0 mM HEPES, 10.0 mM KCl, 1.5 mM MgCl2, 1.0 mM dithiothreitol) and mechanically homogenized in the presence of 1 mM protease inhibitor phenylmethanesulfonyl-fluoride (PMSF). Following centrifugation for 1 min at 4°C, supernatants were collected and protein contents determined by Bradford assay (Bio-Rad, München, Germany). Unless otherwise specified, 15 μg total protein was loaded per lane of a 10% SDS-acrylamide gel.
Enzymatic protein deglycosylation.
Enzymatic deglycosylation experiments from mouse intestinal tissues were carried out with purified brush-border membrane protein, obtained by the divalent cation precipitation technique (DivCatPre) previously described by Wuensch et al. (59) and Schmitz et al. (49). Treatment of protein extracts with the different glycosidases (New England Biolaboratories) was done following manufacturer's instructions.
Western blot analysis.
SDS-PAGE was performed with a mini-protean 3 system from Bio-Rad. Isolated protein (15 μg) was mixed with 4× Laemmli-buffer (125.0 mM Tris, pH 6.8, 8.0% SDS, 20.0% glycerol, 0.4% bromphenol blue sodium salt, 20.0% β-mercaptoethanol) and separated on a 10% SDS-acrylamide gel for 1 h and 3 h at 120–160 V. Proteins were then transferred onto nitrocellulose membranes (Whatman, Maidstone, UK) in a wet tank blotting system at 0.36 A for 25 min. Membranes were blocked in 1% bovine albumin (AppliChem, Darmstadt, Germany) for 1 h, succeeding overnight staining at 4°C with antibodies against mPEPT1 (custom-made; Pineda, Berlin, Germany; dilution 1:5,000, polyclonal rabbit IgG against COOH terminus of rat PEPT1: NH2-CVGKENPYSSLEPVSQTNM-COOH) and β-actin as loading control (dilution 1:2,000, polyclonal goat IgG, C-11 sc-1615; Santa Cruz Biotechnology, Dallas, TX). After three washings with PBS-T (0.137 mM NaCl, 2.700 mM KCl, 10.000 mM Na2HPO4, 1.800 mM KH2PO4, pH 7.4, 0.050% Tween-20), membranes were stained with IRDye-labeled secondary antibodies (dilution 1:12,000, C50113-03, C40415-02; LI-COR Biosciences, Bad Homburg, Germany) at room temperature. Fluorescence signals were detected with an infrared fluorescence Odyssey scanner and quantified with the Image Studio Lite software (v 3.1) supplied from LI-COR.
Biotinylation of Xenopus laevis surface proteins.
For biotinylation, 20 oocytes injected with appropriate cRNA were stored for 30 min in Barth solution lacking gentamycin. Following washing with PBS (0.14 mM NaCl, 2.70 mM KCl, 10.00 mM Na2HPO4, 1.80 mM KH2PO4, pH 8.0), oocytes were incubated with 0.5 mg/ml sulfo-NHS-LC-biotin (Pierce, Rockford, IL) in PBS for 15 min. Residual biotinylation reagent was removed by three washings with PBS. After incubation with quenching buffer (100 mM glycine in PBS) for 20 min on ice, oocytes were lysed 30 min with lysis buffer (1% Triton X-100, 150 mM NaCl, 20 mM Tris·HCl, pH 7.6), containing 0.5 mM protease inhibitor PMSF. Solubilized oocytes were centrifuged for 15 min at 14,000 g and 4°C, and supernatants were collected. Total protein (200 μg) was incubated with 50 μl streptavidin-agarose (Sigma-Aldrich, Taufkirchen, Germany) overnight at 4°C with gentle agitation. The beads were washed with cold PBS, followed by the addition of Laemmli buffer and heating at 95°C for 5 min to break biotin-streptavidin bonds. Obtained biotinylated proteins were directly transferred to SDS-PAGE.
Tracer flux studies with Xenopus laevis oocytes.
Oocytes expressing PEPT1 transporters were incubated in solutions of increasing Gly-Sar concentrations (0.3–50.0 mM) containing radio-labeled [14C]-Gly-Sar (56 Ci/mol) custom synthesized by GE Healthcare (Munich, Germany) at pH 6.5 for 10 min at 23°C. Oocytes were thereafter washed with Barth solution and transferred into scintillation vials. Samples were dissolved in 20% SDS for 2–3 h at 50°C with shaking (500 revolution/min), and radioactivity was counted after addition of scintillator (Eco Plus; Rotiszint, Roth, Germany) in a liquid scintillation counter (Tri-Carb 2810 TR; PerkinElmer, Waltham, MA).
Oocytes expressing transporters were embedded in paraffin and cut into 6-μm sections. Following dewaxing, antigens were unmasked in citrate buffer pH 6.0 by heating at 95°C for 5 min, and sections were blocked in 5% skim milk for 1 h. Staining of the oocytes was performed overnight at 4°C with an mPEPT1 primary antibody (custom made, dilution 1:1,000; Pineda) and a Cy3-conjugated secondary antibody with incubation at room temperature for 2 h (AffiniPure donkey anti-rabbit IgG Cy3-conjugated, dilution 1:500; Jackson ImmunoResearch, Newmarket, UK). Protein localization was detected by fluorescence microscopy (Leica DMI4000B, 40-fold magnification).
TEVC values are expressed as median ± min/max (+ inside bars marks mean values), or means ± SE. Statistical differences were calculated either by one-way ANOVA with Dunnett's posttest or two-tailed unpaired t-test by using GraphPad Prism 4.01 with significance levels as follows: *P < 0.05, **P < 0.01, ***P < 0.001.
PEPT1 mass shift in murine intestinal segments depends on glycosylation.
Our initial observation based on Western blot analysis was that the molecular mass of PEPT1 derived from small or large intestine of C57BL/6N mice differed by around 10 kDa (59). This variation was also observed in different mouse strains (C57BL/6N, C57BL/6J, AKR/J, A/J, and SV129/S6) in both male and female animals of either 8 or 40 wk of age. Germ-free C57BL/6N mice (male, 8 wk of age on standard diet) also revealed this marked mass difference, which seems therefore independent of the genetic background, sex, age, or bacterial colonization. Figure 2A shows the mass shift exemplarily for C57BL/6N mice. For assessing the contribution of glycans to overall mass, we performed deglycosylation experiments with peptide-N-glycosidase F (PNGaseF), endoglycosidase H (EndoH), and neuraminidase with membrane protein extracts collected from murine small intestine and colon (Fig. 2, B and C). PNGaseF, releasing N-linked oligosaccharides of high mannose, hybrid, and complex type from glycoproteins, reduced the mPEPT1 mass in jejunum from ∼95 kDa and in colon from 105 kDa to ∼60 kDa. Unmodified mPEPT1 is predicted to have a protein mass of 79 kDa (UniProtKB). EndoH, cleaving the chitobiose core of high mannose and some hybrid oligosaccharides of N-linked glycoproteins, did not alter the PEPT1 protein mass. Release of N-acetyl-neuraminic acids by neuraminidase, used in combination with EndoH to improve EndoH accessibility, did not change the mPEPT1 mass in small intestine or colon. These findings suggest that the PEPT transporter carries predominantly N-linked glycans of a complex type.
Site-directed mutagenesis of putative mPEPT1 N-glycosylation sites.
N-glycosylation sites were predicted using the NetN-Glyc 1.0 platform (27). Thereby, six asparagine residues, N50, N406, N439, N510, N515, and N532, were selected according to the following attributes: 1) location within an N-X-S/T sequon, 2) N-X-S/T sequon located in an extracellular PEPT1 domain, 3) high N-glycosylation prediction score (Table 3). By use of site-directed mutagenesis, several mutant proteins lacking single or multiple sequons were generated, and proteins were heterologously expressed in Xenopus laevis oocytes. Transporter mass was thereafter visualized by Western blot analysis (Fig. 3). Imaging of the maximal fluorescence intensity of immunoblots (Fig. 4) performed with the LI-COR Bioscience Imaging software showed that the mutants N50Q, N50G, N406Q, and N439Q had lower mass than the WT PEPT1 protein (∼95 kDa), whereas N532Q had a comparable mass (Figs. 3A and 4). Interestingly, the N515Q mutant revealed a mass increase, most likely attributable to a modification of the glycosylation pattern at residual asparagine residues within sequons.
Replacement of all asparagines in sequons by glutamine in mPEPT1 (Fig. 3B) resulted in a molecular mass of ∼60 kDa comparable with the results obtained by PNGaseF treatment of intestinal membrane proteins (Fig. 2, B and C). A similar mass was also obtained when glycosylation was impaired in oocytes by injection of the N-glycosylation inhibitor tunicamycin (Fig. 8C).
N-glycosylation of PEPT1 does not affect membrane targeting.
To elucidate whether N-glycosylation is essential for PEPT1 protein folding and plasma membrane targeting, we biotinylated Xenopus oocytes that expressed the various mPEPT1 constructs with the membrane-impermeable reagent sulfo-NHS-LC-biotin. Biotinylated surface proteins were isolated by streptavidin-agarose before Western blot quantification. None of the mutant proteins lacking individual or all N-glycosylation sites showed impairment in plasma membrane abundance (Fig. 5). This was also confirmed by immunohistochemical staining of oocyte sections (Fig. 6). Oocytes coinjected with tunicamycin showed unaltered fluorescence for the native mPEPT1 protein.
Effects of N-glycosylation on transport kinetics.
Functional analysis of the WT transporter recorded by TEVC showed growing currents when Gly-Sar concentrations were increased from 0.3 to 10.0 mM (Fig. 7A) with an Imax at 10 mM of 330 ± 107 nA at −60 mV (Fig. 7B). Kinetic analysis revealed an apparent Km of Gly-Sar of 0.66 ± 0.12 mM (Fig. 7, C and D, Table 4) comparable to 0.75 mM in mouse (21, 22) and 1.1 mM in human PEPT1 (10). Exchange of asparagine residues at positions 406, 439, and 515 did not significantly affect Km or Imax. Interestingly, amino acid exchange at position N532 showed a slight decrease in substrate affinity without changing Imax, assuming that this sequon is most likely not functional. Noticeably, the mutants N50Q and N50G both revealed about twofold reduced Gly-Sar affinities coinciding with almost identical increases in Imax compared with WT. Similar observations were made for the two PEPT1 constructs harboring multiple amino acid exchanges including N50Q. In both transporters, the Km values increased by 1.4–2.0-fold, whereas maximal inward currents were up to three times higher. The mutant transporter N406Q/N439Q/N515Q/N532Q revealed half of the substrate affinity of WT, whereas Imax remained unchanged. Additional insertion of N510 roughly halved the Km and contrarily increased Imax to the same extent. In the presence of cefadroxil (Fig. 9, A and B), the same mutants showed reduced substrate affinities and concurrently increased Imax values. At first glance, it appears that the N-linked glycans attached to the big extracellular loop between TMD 9 and 10 directly affect PEPT1 transport. It nonetheless remains conceivable that inhibition of N-linked glycosylation at multiple sequons could lead to glycan modification at residual or not favored sequons, thereby altering PEPT1 transport kinetics. Because of the opposing Imax values observed between the mutants N406Q/N439Q/N515Q/N532Q and N406Q/N439Q/N510Q/N515Q/N532Q, a modification of surface N-glycans attached to sequon N510 can be assumed. When mPEPT1 was expressed in the presence of the N-glycosylation inhibitor tunicamycin, a decreased affinity for Gly-Sar (apparent Km = 0.75 ± 0.11 mM) but elevated transport currents (Imax = 564.13 ± 94.9 nA) were found compared with WT mPEPT1 (apparent Km = 0.65 ± 0.09 mM and Imax = 437.64 ± 118.3 nA) as shown in Fig. 8, A and B.
Changes in transport kinetics are independent of substrate and are conserved between species.
To investigate whether effects of amino acid exchange at position N50 are specific for the substrate Gly-Sar, kinetic constants for the cephalosporin antibiotic cefadroxil (Fig. 9, A and B) and tri-l-alanine (Fig. 10, A and B) were determined. Similar to Gly-Sar, the N50Q mutant showed decreased substrate affinities and increased maximal currents, both for cefadroxil and tri-l-alanine. In the case of cefadroxil, substrate affinity decreased by about 20% [apparent Km(N50Q) = 1.99 ± 0.20 mM] compared with WT (Fig. 9A), whereas maximal currents increased fourfold to 2.83 ± 0.68 nA (Fig. 9B). Highest currents were obtained with cefadroxil in the sextuple mutant protein with 6.62 ± 0.67 nA, exceeding transport currents of the WT protein by more than ninefold.
For tri-l-alanine (Fig. 10A), the N50Q mutant yielded a slightly reduced affinity [apparent Km(N50Q) = 0.18 ± 0.04 mM] compared with WT [apparent Km(WT) = 0.14 ± 0.04 mM] but had also substantially increased transport currents amounting to 1.92 ± 0.41 nA compared with WT with 0.81 ± 0.15 nA (Fig. 10B). To demonstrate that the effects of N50 are not a species-specific phenomenon, the same mutation was generated in the human PEPT1 transporter (UniProt AC: P46059) (Fig. 11, A and B). TEVC recordings also revealed a significant reduction in Gly-Sar affinity in the human N50Q mutant from 0.88 ± 0.19 mM to 1.16 ± 0.18 mM (Fig. 11A), whereas maximal currents recorded at −60 mV in the mutant increased more than twice to 3.70 ± 0.65 nA compared with WT with 1.63 ± 0.23 nA as shown in Fig. 11B.
Amino acid exchange at N50 does not open a proton leak.
To assess whether the high maximal currents detected by TEVC are caused by a proton conductance, tracer flux studies with [14C]-Gly-Sar were performed (Fig. 12A). Oocytes expressing N50Q showed significantly higher transport activity for Gly-Sar, amounting with 50 mM Gly-Sar to 4.47 ± 0.32 pmol/min per oocyte in N50Q and 2.32 ± 0.30 pmol/min per oocyte in WT. Tunicamycin treatment slightly increased maximal [14C]-Gly-Sar uptake rates in oocytes expressing WT PEPT1 amounting to 2.59 ± 0.21 pmol/min per oocyte. Apparent Km values increased from 0.82 ± 0.19 mM in WT to 2.92 ± 0.37 mM in N50Q and to 1.26 ± 0.12 mM in tunicamycin-treated oocytes. Competition of tracer influx by the presence of a 10-fold excess of glycyl-glutamine (Gly-Gln) (Fig. 12B) revealed a reduction of tracer uptake by 82% in N50Q and inhibition of the WT mPEPT1 by 57%.
It is assumed that 50–90% of all proteins in higher eukaryotes undergo posttranslational modifications, of which the glycosylation of asparagine residues is the most frequent one (38). Previous mass-spectrometric analysis (58) identified PEPT1 as a cell surface glycoprotein. Targeted screening for asparagine residues in mPEPT1 delivered in total 35 matches, whereof 8 asparagines are positioned within an N-X-S/T consensus motif (Table 3). Acceptor sequences containing proline in center position, known to have a low N-glycosylation probability (7, 46a), as well as sequons located in an intracellular protein domain and motifs with a low N-glycosylation prediction score (<0.3) were neglected. The remaining six asparagines N50, N406, N439, N510, N515, and N532 were exchanged for glutamine. Mutant transporter proteins lacking single or multiple glycosylation sites all showed expression in the plasma membrane of oocytes with no indication of impaired targeting or membrane insertion, allowing the proteins to be studied also at the functional level. Disruption of N50, N406, and N439 was accompanied with a mass reduction of mPEPT1, whereas single mutations at N510Q and N532Q did not appear to change the protein mass. It would be erroneous to conclude that these two sequons are not N-glycosylated. Disruption of single sequons can trigger glycan modification on other sequons. This was shown by Tanaka et al. (55) for the organic anion transporter OAT1, demonstrating that simultaneous replacement of asparagines in sequons leads to oligosaccharide transfer onto previously neglected downstream acceptor sites.
Also the fact that N510 and N532 are both located in sequons where threonine is found at the third position of the N-X-S/T motif, which is known to be preferred glycosylated over serine residues (7, 25, 46), would suggest the presence of N-glycans. Simultaneous disruption of multiple sequons containing N510 and N532, however, showed a decrease in the PEPT1 protein mass of about 3–10 kDa between mutants N50Q/N406Q/N439Q/N515Q/N532Q and N50Q/N406Q/N439Q/N510Q/N515Q/N532Q as well as N406Q/N439Q/N515Q and N406Q/N439Q/N515Q/N532Q, indicating that both sequons of N510 and N515 are occupied by N-glycans. Disruption of all six N-glycosylation sites in the mouse PEPT1 protein caused a mass reduction of around 35 kDa. Moreover, the sextuple mutant revealed an altered migration pattern with increased mobility and reduced protein band size, probably attributable to altered detergent binding (57). When glycosylation in oocytes was suppressed by tunicamycin, a similar loss of mass was observed, and this matches also with the results obtained by treatment of murine intestinal brush-border membrane samples with PNGaseF.
Protein glycosylation is important for proper protein folding, sorting, membrane targeting, and protein structure and function (30). Disruption of N-glycosylation can thus cause protein misfolding and promotes proteasomal degradation (5, 57) or can change turnover rates by affecting the transporter membrane density (18). However, for the mPEPT1 and hPEPT1 mutants described here, neither immunofluorescence analysis nor oocyte surface protein biotinylation revealed any evidence for alterations in membrane protein density.
Kinetic characteristics of rheogenic transporters can most easily be characterized by electrophysiology (3, 12, 37). With the TEVC technique, we demonstrate kinetic changes in peptide transporters lacking N-glycosylation. The most impressive changes were obtained for position N50, where the exchange N50Q and N50G caused a reduction of substrate affinity accompanied by a gain of velocity. The kinetic changes were found with three different substrates. An increase in transport currents can be caused by 1) increased substrate transport, 2) increased ion conductance induced by the mutation, or 3) increased transporter density. However, quantification of protein density in oocyte plasma membranes via biotinylation proved that unaltered protein levels and tracer flux studies confirmed also increased uptake of labeled substrate. This means that N50 mutant proteins have indeed a higher transport activity.
That glycosylation affects transporter function has been shown for the glucose transporter GLUT1 (2) and the ammonium transporter MEP2 (41), in which tunicamycin causes a loss in substrate affinity. Treatment of mPEPT1-expressing oocytes with tunicamycin led to a protein mass comparable to that of the sextuple mutant. However, there were no significant effects on cell-surface expression in oocytes but robust changes in affinity and transport rate assessed by both current measurements and tracer flux analyses. Substrate affinities declined by 50–72% with Gly-Sar as substrate, whereas transport rates increased between 52 and 57%.
In understanding how glycans attached to N50 affect the transport characteristics with a gain of maximal transport rate when removed, protein structure and membrane topology need to be inspected. However, structural information on mammalian peptide transporters is very limited and relies on homologies with crystal structures of other members of the major facilitator superfamily, like the prokaryotic homologue of PEPT1 from Shewanella oneidensis (43). On the basis of modeling data from prokaryotic SLC15 members (19, 44), N50 is placed between TMD 1 and 2, just above the membrane projecting into the extracellular space. Glycans attached to N50 would thus be in a loop comprising about 32 amino acids, close to the membrane surface, and this may constrain necessary movements of amino-terminal TMDs in the transport cycle (Fig. 13). Removal of the glycans could thus increase mobility of the protein. Membrane domains 1 and 2 are known to contribute to the substrate binding and translocation pore, and changes in substrate affinity, as shown for all N50 variants, are therefore not unexpected. Assessing the precise glycan mass and architecture requires more sophisticated approaches, but immunoblots reveal a slight decrease of around 1–5 kDa when glycosylation of N50 is prevented.
Taking a total of around 35-kDa mass shift for an expected glycan-free protein, the majority of glycans are therefore attached to asparagine residues in the large extracellular loop between TMD 9 and 10, predominantly N406 and N439. Recent crystallization of this loop domain from the mouse and rat PEPT1 transporter disclosed two immunoglobulin-like structures, transiently interacting with trypsin for luminal concentration of peptides rich in arginine and lysine (4). Peptide transporter proteins from prokaryotes are lacking this big extracellular loop but have very similar functional characteristics (17, 29, 56), arguing against a role of the loop in the transport process.
On closer examination of N50, N406, N439, N510, N515, and N532, it appears that there exists a strong conservation within species (Fig. 14). Protein sequence alignment of 50 different vertebrate species showed strongest conservation of N50 and N439 within N-X≠P-S/T sequons, amounting to 58–60%. Whereas N50 is mainly found in mammals, birds, and reptiles, N439 is primarily preserved along mammalian species. With 44–46%, N510 and N515 both are highly conserved in mammals, exhibiting 30% cooccurrence. N406 and N532 have a minor prevalence of 10–20% in mammals. In consideration of these findings, it appears that some sequons are more conserved between species than others, implying a more important function. Although N50 and N439 are both located in the two most conserved sequons, our study demonstrates that only disruption of N-glycosylation of N50 alters PEPT1 transport kinetics, whereas N439 has no obvious effects. Therefore, in silico analysis of sequon conservation can be a good tool to specify and predict the most important N-glycan acceptor sites in proteins although they do not provide information on sequon occupancy. Moreover, protein sequence alignment alone is not potent enough to completely replace functional protein studies.
Western blot analysis disclosed differences in the PEPT1 mass between small and large intestine of mouse attributable to altered protein N-glycosylation. It is not unusual that a glycoprotein appears in different glycoforms under physiological conditions (45). Variations in PEPT1 glycosylation patterns could arise from changes in the N-glycosylation site occupancy or by modification of surface glycan structures catalyzed by cellular glycosidases and glycosyltransferases (33). Intestinal glycosyltransferases have been shown to be developmentally regulated, region specific, and susceptible for changes by the endogenous microflora (15). Because of the observation that the molecular mass of PEPT1 resembles gut segments of germ-free and conventionally raised mice, the microbiota might play a minor role in modulating glycosyltransferase activities or in the degradation of glycoconjugates (23). To verify these issues, further glycosyltransferase expression and activity studies, combined with a structural glycan analysis, need to be performed. Besides, in vitro experiments have shown that environmental factors can significantly change the glycan profile of a protein. Low glucose levels, for instance, have been found to reduce sequon occupancy in murine myeloma cells (52). At low cellular oxygen and pH concentrations, glycosyltransferase activities are known to alter significantly (40, 47). As the gastrointestinal tract is a highly complex organ system, varying PEPT1 glycoforms might result from the interplay of various environmental factors.
Besides, there is evidence for the existence of organ-related protein glycoforms (6). To verify this, the glycosylation status of PEPT1 was examined in different mouse organs. Beyond the gut, highest PEPT1 expression levels were reported for bile duct epithelial cells and the proximal tubule of kidney. However, the majority of available data are solely based on mRNA expression studies. Although immunostaining of the extrahepatic biliary tract localized PEPT1 in the apical membrane of cholangiocytes in mice (34), there is currently no information on the transporter protein mass. Because of its low abundance, all our efforts to determine the PEPT1 mass in mouse liver were unsuccessful so far.
In contrast, expression studies in dogs showed a PEPT1 mass of about 78 kDa in liver (28). However, it is currently unclear whether species-specific differences exist in PEPT1 glycosylation. For murine kidney, a PEPT1 mass of 75 kDa was reported (48). Compared with the gut, exhibiting a PEPT1 mass of ∼95 kDa in small intestine and 105 kDa in colon, the PEPT1 mass in liver and kidney rather corresponds to the predicted protein core mass of 79 kDa (UniProtKB). PEPT1 expressed in human pancreatic cancer cells was detected in multiple glycoforms varying from 90–120 kDa (26). However, in a diseased state such as cancer, N-glycosylation of proteins is known to be dysregulated (13). Under this aspect, tumor cells are of limited suitability to predict the glycosylation status of a healthy cell. Nonetheless, immunoblots performed with the human colon carcinoma cell line Caco-2, exhibiting postdifferentiation characteristics of small intestinal cells (20), resulted in a similar PEPT1 mass as observed in the small intestine of mouse (data not shown). As PEPT1 transport significantly differs between kidney and gut (9), the individual PEPT1 glycoforms might also have altered functional and kinetic properties. To confirm this, additional transport studies need to be performed.
In summary, our study demonstrates that glycosylation of intestinal PEPT1 amounts to ∼1/3 of its overall mass and that it differs between small intestine and distal colon. This phenomenon of varying protein glycosylation of an intestinal membrane transporter within the same organ is hitherto a unique observation and raises new questions on the biological significance of glycosylation. We identified six asparagine residues used for N-glycosylation with five residues located in the large extracellular loop between TMD 9 and 10. Removal of N-glycans by site-directed mutagenesis or by preventing glycosylation with tunicamycin did not alter membrane protein abundance but changed the transport kinetics. The phenomenon of reduced substrate affinity but an increased maximal transport rate was found upon an exchange of N50 to glutamine or glycine. This gain of function in velocity may be attributed to sterical interference of surface glycans with the substrate binding site of PEPT1.
This work was funded by the Deutsche Forschungsgemeinschaft (DFG) as part of the Research Training Group GRK1482.
No conflicts of interest, financial or otherwise, are declared by the authors.
Author contributions: T.S., K.E.G., and H.D. conception and design of research; T.S., T.B., and G.K. performed experiments; T.S. and T.B. analyzed data; T.S. and K.E.G. interpreted results of experiments; T.S. prepared figures; T.S. drafted manuscript; T.S., K.E.G., G.K., and H.D. edited and revised manuscript; T.S., T.B., K.E.G., G.K., and H.D. approved final version of manuscript.
We thank Veronika Mussack for assisting in TEVC experiments and Helene Prunkl for taking care of the animals and support in oocyte handling and preparation. We also acknowledge Dr. J. Stolz for project suggestions and assistance in the preparation of the manuscript.
- Copyright © 2016 the American Physiological Society