Effects of N-linked glycosylation on the creatine transporter

Abstract

The CRT (creatine transporter) is a member of the Na+- and Cl--dependent neurotransmitter transporter family and is responsible for the import of creatine into cells, and thus is important for cellular energy metabolism. We established for CRT an expression system in HEK-293 cells that allowed biochemical, immunological and functional analysis of CRT wild-type and glycosylation-deficient mutants. Analysis of HA (haemagglutinin)-tagged CRT-NN (wild-type rat CRT with an HA-tag at the C-terminus) revealed several monomeric immunoreactive species with apparent molecular masses of 58, 48 and 43 kDa. The 58 kDa species was shown to be plasma-membrane-resident by EndoHf (endoglycosidase Hf) and PNGase F (peptide N-glycosidase F) treatments and represents fully glycosylated CRT, whereas the 48 kDa and 43 kDa species were glycosylation intermediates and non-glycosylated CRT respectively. Glycosylation-deficient mutants (Asn192Asp, Asn197Asp and Asn192Asp/Asn197Asp) showed altered electrophoretic mobility, indicating that CRT is indeed N-glycosylated. In addition, a prominent CRT band in the range of 75-91 kDa was also detected. Pharmacological inhibition of N-linked glycosylation by tunicamycin in CRT-NN-expressing cells gave a similar reduction in molecular mass, corroborating the finding that Asn192 and Asn197 are major N-glycosylation sites in CRT. Although the apparent Km was not significantly affected in glycosylation-deficient mutants compared with CRT-NN, we measured reduced Vmax values for all mutants (21-28% residual activity), and 51% residual activity after enzymatic deglycosylation of surface proteins in intact CRT-NN cells by PNGase F. Moreover, immunocytochemical analysis of CRT-NN- and CRT-DD-expressing cells (where CRT-DD represents a non-glycosylated double mutant of CRT, i.e. Asn192Asp/Asn197Asp) showed a lower abundance of CRT-DD in the plasma membrane. Taken together, our results suggest that plasma-membrane CRT is glycosylated and has an apparent monomer molecular mass of 58 kDa. Furthermore, N-linked glycosylation is neither exclusively important for the function of CRT nor for surface trafficking, but affects both processes. These findings may have relevance for closely related neurotransmitter transporter family members.

Figure 1. CRT cDNA expressed in HEK-293 cells

In (A) rat CRT cDNA (shaded box) was amplified by PCR and subcloned with EcoRI and BamHI into the plasmids pHAC1 (CRT-C-NN) and pHAN1 (CRT-NN) containing an HA-epitope tag (hatched box) at the N -and the C-termini respectively of the MCS (multiple cloning site). Amino acids Asn¹⁹² and Asn¹⁹⁷ are canonical N-glycosylation sites within the extracellular loop II. (B) The expression pattern of CRT-NN and CRT-C-NN in HEK-293 cells was analysed in 8 μg of total protein lysate loaded on to LDS-PAGE. Immunoreactive species with apparent electrophoretic mobilities of 75–91, 58, 48 and 43 kDa were recognized by anti-HA and anti-CRT antibodies for CRT-NN and CRT-C-NN (lanes 3, 5 and 8) on a Western blot. No signal was seen in mock-transfected cells (lanes 1 and 7). Concomitant treatment of CRT-NN- and CRT-C-NN-expressing cells with tunicamycin (5 μg/ml) abolished the 58 and 48 kDa species, and led to a band shift of the 75–91 kDa region to 70–86 kDa (lanes 4, 6 and 9). The higher-molecular-mass region at 75–91 kDa is likely to represent a homodimeric form of CRT-NN and CRT-C-NN.

Figure 2. Subcellular distribution of wild-type CRT-NN

In immunocytochemical assays, HEK293 cells expressing CRT-NN were probed with anti-HA (green), anti-calnexin (panels C and D, red) or Mitotracker (panels F and G, red). Plasma membrane was visualized using FITC-labelled wheat germ agglutinin WGA (panel J and K, green). The CRT-NN signal (A, E and H) revealed a reticular distribution of CRT-NN that co-localized with the calnexin staining (A, C and D, double arrows). The CRT-NN signal was also detectable at the plasma membrane and in cellular boundaries (J and K, arrows). The Mitotracker signal (F) showed no overlap with CRT-NN immunofluorescence (G). Mock-transfected HEK293 cells showed no staining (B). Scale bar A-G: 5 μm. Scale bar H–K: 15 μm.

Figure 3. Canonical N-glycosylation sites in CRT and removal of N-glycosylation sequons in CRT-NN by site-directed mutagenesis

In (A) the human and rat CRT protein sequence was aligned with rat GAT1 and rat 5HTT. Canonical N-glycosylation sequons (shown in italics) lay within the large extracellular loop II. Amino acid Asn¹⁹⁷ is conserved throughout the family of monoamine and γ-aminobutyric acid neurotransmitter transporters, whereas Asn¹⁹² is not conserved. High consensus is denoted by dark grey shading; low consensus is shown as light grey. (B) Canonical N-glycosylation sequons at positions Asn¹⁹² and Asn¹⁹⁷ were removed in CRT-NN by site-directed mutation of A to G. Two single mutants (CRT-ND and CRT-DN) deficient in one canonical N-glycosylation sequon and one double mutant (CRT-DD) deficient in both glycosylation sequons were generated. (C) Lysates of HEK-293 cells expressing pHAN1, CRT-NN, CRT-DD, CRT-ND and CRT-DN respectively were analysed on a Western blot with anti-HA antibody. For CRT-NN, administration of tunicamycin (5 μg/ml) abolished the 58 and 48 kDa species and increased the amount of the 43 kDa species (lanes 3 and 4). For CRT-DD, only the 43 kDa was detectable in lysates with (lane 6) or without (lane 5) tunicamycin administration. For CRT-ND and CRT-DN, the 58 kDa band was shifted to 51 kDa and the 48 kDa band to 46 kDa (lanes 7 and 9 respectively); tunicamycin abolished all species except for the 43 kDa core-protein (lanes 8 and 10).

Figure 4. Altered subcellular distribution after mutation of N-glycosylation sites

Expression of glycosylation-deficient mutants CRT-ND (A), CRT-DN (B) and CRT-DD (C) revealed altered subcellular distribution of the mutants. The reticular CRT immunofluorescence (panels I and IV, green) co-localized (panel III) with the ER-resident calnexin staining (panel II, red), but not with the mitochondrial Mitotracker signal (panels V and VI, red). For CRT-DD (C) and CRT-ND, no plasma-membrane-resident staining could be detected (C, panels VII–IX, arrows). For CRT-DN, a few cells showed weak immunofluorescence at the cell boundaries (B, panel I). (A) Double arrows in panels I–III indicate reticular staining. (B) Single arrows in panels I–III indicate plasma membrane staining.

Figure 5. Plasma-membrane-resident CRT-NN monomer has an apparent molecular mass of 58 kDa

In (A) enzymatic deglycosylation of unprocessed glycans was performed with EndoH_f (50 units/10 μg od protein) in HEK-293 cell lysates in vitro. For CRT-NN (NN), the 48 kDa species was deglycosylated, leading to an electrophoretic shift of 5 kDa (lane 3). The 58 kDa species was protected against EndoH_f digestion. For CRT-DD (DD), no changes in band pattern were observed upon deglycosylation (lane 4). For CRT-ND (ND) and CRT-DN (DN), the 51 kDa band was protected against EndoH_f (lanes 5 and 6 respectively). (B) Enzymatic deglycosylation of all glycans facing the extracellular space was achieved by addition of PNGase F (10 units/μl) to CRT-NN-expressing cells in vivo. Cleavage of all surface glycans in vivo produced a non-glycosylated plasma-membrane-resident CRT-NN with an electrophoretic mobility of 43 kDa (lane 2). As a control, total lysate was incubated in vitro with PNGase F, which showed deglycosylation of the 48 kDa species when accessible to PNGase F (lane 3). Corresponding effects could be observed for the higher-molecular-mass region.

Figure 6. Time-course of [¹⁴C]creatine uptake into CRT-NN-expressing HEK293 cells

Uptake of 10 μM [¹⁴C]creatine (0.14 μCi) was measured at 15, 30, 60 and 90 min. The amount of creatine uptake was expressed as nmol/mg per min, and values were the means±S.D. for six independent experiments. (A) [¹⁴C]Creatine import in HEK-293 cells expressing CRT-NN (○), pHAN1 (□) or in wild-type HEK293 cells (*) showed a 10 min lag-phase followed by a linear uptake phase of 50 min, before reaching a plateau after 60 min. From the slopes of the linear phase, an approx. 13-fold increase in CRT density at the plasma membrane of CRT-NN-expressing cells compared with wild-type or mock-transfected cells could be calculated. (B) Na⁺-dependence of CRT-NN expressed in HEK-293 cells was shown using either NaCl- (○) or LiCl (□)-containing assay buffer. LiCl-containing buffer blunted [¹⁴C]creatine uptake and lowered uptake rates below wild-type levels. (C) [¹⁴C]Creatine uptake was competed by adding a 10-fold excess of β-GPA over creatine to the assay mix after 30 min of normal uptake (□). Intracellular [¹⁴C]creatine levels remained therefore constant until the end of the measurements.

Figure 7. Kinetics of [¹⁴C]Creatine uptake in CRT-NN expressing HEK293 cells

[¹⁴C]Creatine uptake was measured as a function of variable creatine substrate concentrations (1, 1.5, 3, 5, 10, 20, 50, 100 and 200 μM) containing 0.001 μCi [¹⁴C]creatine per μM creatine within the linear phase (40 min). The velocity (v) of creatine uptake was expressed as pmol/mg per min, and values were means±S.E.M. for four independent experiments. The velocity as a function of substrate concentration followed the Michaelis–Menten model. In the inset to the main Figure, the uptake curve was linearized with the weighted Hanes-Woolf algorithm (substrate concentration [S]/velocity (v) plotted against [S]). The apparent Michaelis–Menten constant (K_m) and maximal velocity (V_max) were 37.3±6.8 μM and 467.7±49.2 pmol/mg per min respectively.