The transport system b0,+ mediates reabsorption of dibasic amino acids and cystine in the kidney. It is made up of two disulfide-linked membrane subunits: the carrier, b0,+AT and the helper, rBAT (related to b0,+ amino acid transporter). rBAT mutations that impair biogenesis of the transporter cause type I cystinuria. It has been shown that upon assembly, b0,+AT prevents degradation and promotes folding of rBAT; then, rBAT traffics b0,+AT from the endoplasmic reticulum (ER) to the plasma membrane. The role of the N-glycans of rBAT and of its C-terminal loop, which has no homology to any other sequence, in biogenesis of system b0,+ is unknown. In the present study, we studied these points. We first identified the five N-glycans of rBAT. Elimination of the N-glycan Asn575, but not of the others, delayed transporter maturation, as measured by pulse chase experiments and endoglycosidase H assays. Moreover, a transporter with only the N-glycan Asn575 displayed similar maturation compared with wild-type, suggesting that this N-glycan was necessary and sufficient to achieve the maximum rate of transporter maturation. Deletion of the rBAT C-terminal disulfide loop (residues 673–685) prevented maturation and prompted degradation of the transporter. Alanine-scanning mutagenesis uncovered loop residues important for stability and/or maturation of system b0,+. Further, double-mutant cycle analysis showed partial additivity of the effects of the Asn679 loop residue and the N-glycan Asn575 on transporter maturation, indicating that they may interact during system b0,+ biogenesis. These data highlight the important role of the N-glycan Asn575 and the C-terminal disulfide loop of rBAT in biogenesis of the rBAT-b0,+AT heterodimer.
System b0,+ mediates re-absorption of cystine and dibasic amino acids by exchange for neutral amino acids in the proximal tubule [1,2]. It belongs to the family of heteromeric amino acid transporters (HATs), formed by several carrier light subunits (unglycosylated proteins with 12-transmembrane domains) and two helper heavy subunits [rBAT (related to b0,+ amino acid transporter) and 4F2hc; type II membrane glycoproteins]. It is made up by rBAT and the carrier b0,+AT, linked by a disulfide bond conserved in all HATs [3,4].
The structure of the ectodomain of the helper subunit 4F2hc reveals a (β/α)8 TIM-barrel (domain A) followed by an antiparallel eight-stranded β-sheet (domain C) and is similar to α-glucosidases (but without glucosidase activity) . The structure of the carrier subunits is unknown. However, template modelling based on the related prokaryotic transporter AdiC [6,7], together with the structure of the 4F2hc ectodomain, allowed construction of a low-resolution model of the human HAT 4F2hc–LAT2 . Structural alignment strongly supports a model for the rBAT ectodomain similar to 4F2hc , with three striking differences: (i) a domain B between β3 and α3 of domain A; (ii) a ∼31-residue long C-terminal tail without homology to any other protein sequence; and (iii) 3 intramolecular disulfide bonds: one within domain B, one connecting domain C with the tail and a last one forming a C-terminal loop (residues Cys673–Cys685; Cys685 is the C-terminal residue) within the tail .
Cystinuria (OMIM 220100) is due to mutations either in rBAT or b0,+AT which cause a decrease in functional system b0,+. This leads to the formation of kidney cystine stones, the clinical sign of the disease . Type I cystinuria is caused mainly by mutations in rBAT. Most of them are found in the ectodomain and cause folding/trafficking defects . There is therefore great interest in understanding the biogenesis of this transporter. The two subunits are interdependent for traffic from the endoplasmic reticulum (ER) to the plasma membrane [12,13]. Assembly with b0,+AT prevents ERAD-mediated degradation of unassembled rBAT and is required for the oxidative folding of the rBAT ectodomain, which proceeds via formation of three consecutive disulfide bonds. The domain B disulfide (Cys242–Cys273) may form first and, like disulfide Cys571–Cys666, is essential for biogenesis. The role of the last disulfide (Cys673–Cys685) is not yet known .
To further explore the biogenesis of system b0,+ we focused on the C-terminal loop (residues Cys673–Cys685) of the rBAT tail and, given their recognized relevance in membrane protein biogenesis and function , on the N-glycans of rBAT. In the present work, after identification of the five N-glycosylation sites of rBAT used in vivo, we determine that the N-glycan Asn575 is of paramount importance for early traffic. Amino acid residues within the C-terminal loop are relevant for stability and/or early traffic of the transporter. Finally, a possible interaction between the Asn575 N-glycan and the C-terminal loop, relevant for biogenesis of the transporter, is suggested.
Reagents and antibodies
Reagents were purchased from Sigma unless otherwise indicated. Pro-mix in vitro cell labelling mix (L-[35S]methionine and L-[35S]cysteine) was purchased from PerkinElmer Life Sciences. Dulbecco's modified Eagle's medium (DMEM) medium without L-methionine and L-cystine and dialysed FBS were from Invitrogen. Antibodies against the N-termini of human b0,+AT and rBAT are described elsewhere [2,15].
The vectors for mammalian cell expression of human rBAT and b0,+AT were as described elsewhere . The human rBAT mutants were obtained by site-directed mutagenesis (QuikChange, Stratagene) of pCDNA3-rBAT using the mutagenic oligonucleotides shown in Table 1 (only sense oligonucleotides are shown).
|Mutagenic oligonucleotides (sense)|
|Mutagenic oligonucleotides (sense)|
Cell culture and transfection
HeLa cells were grown in DMEM (Invitrogen) supplemented with 10% fetal calf serum (heat-inactivated), 100 units/ml penicillin (Invitrogen) and 0.1 mg/ml streptomycin (Invitrogen) at 37°C in a humidified atmosphere containing 5% CO2. Calcium phosphate transient transfection of HeLa cells was performed as described  and its efficiency was between 60% and 90% in all experiments. Ten centimetre diameter plates were incubated with a mixture of DNA containing 2 μg of pEGFP (Clontech) to measure transfection efficiency, 6 μg of pCDNA3-rBAT (wild-type or mutants) and 12 μg of pCDNA3-b0,+AT, as described [2,13,16]. These conditions allow most, if not all, of the expressed rBAT to assemble with b0,+ AT . When rBAT or b0,+ AT was transfected alone, 12 or 6 μg of pCDNA3 was added respectively.
Endoglycosidase H assay
The enzyme was obtained from New England Biolabs and used following the manufacturer's protocol.
Pulse chase and immunoprecipitation protocols
Cells were transfected at 40%–50% confluence and seeded the next day in 3.5-cm diameter plates at 60%–70% confluence. 36 h after transfection, the cells were incubated for 30 min in pre-warmed L-methionine/L-cystine-free media containing 10% dialysed FBS. Next, cells were labelled with a mixture of [35S]methionine/cysteine (200 μCi/ml) for 30 min. After removal of the labelling media, the cells were incubated with pre-warmed media supplemented with 5 mM unlabelled L-methionine/L-cysteine. At the indicated chase times (or just after the pulse), cells were washed twice in cold PBS and once for 5 min in cold PBS containing 20 mM N-ethylmaleimide. Cells were collected and lysed on a rotating wheel in 0.2 ml of NET buffer [150 mM NaCl, 5 mM EDTA, 50 mM Tris (pH 7.4), 0.5% IGEPAL CA-630] with the protease inhibitors aprotinin, leupeptin, PMSF and pepstatin and with 20 mM NEM. After 30 min at 4°C, a post-nuclear supernatant was obtained by 10-min centrifugation at 10000 g at 4°C.
Immunoprecipitations were performed from equivalent amounts of radioactivity incorporated into proteins by adding an equal volume of immuno-absorbent buffer [200 mM H3BO3, 50 mM Na2B4O7, 150 mM NaCl, 1% IGEPAL CA-630 and 0.1% ovalbumin (pH 8.3)] with the same protease inhibitors as the lysis buffer and polyclonal antibodies to rBAT or b0,+AT, in combination with protein A-Sepharose. Precipitates were washed four times in borate-NaCl buffer (0.5% IGEPAL CA-630, 0.3M NaCl, 25 mM Na2B4O7 and 0.1 M H3BO4, pH 8.3) and twice in 40 mM HEPES (pH 8). Samples were run on SDS/PAGE under reducing (100 mM DTT) conditions. Gels were stained with Coomassie Brilliant Blue to control for precipitating antibodies, dried and placed on an intensifying screen for quantification with a Phosphoimager Typhoon 8600 (Molecular Dynamics).
Membrane preparation, SDS/PAGE and Western blot
The relative intensities of the labelled bands were determined using phosphorimaging as follows: each band was outlined by a rectangle (as tightly fitting to the band as possible) and a rectangle of identical size was drawn in the closest area without any band in the lane. The relative positions of band and background rectangles were maintained within the experiment and in similar experiments. The value for each rectangle was calculated using the local average background correction of the ImageQuant software. The final value of a band was the difference between the value of the rectangle band and the value of the rectangle background.
Double mutant cycle analysis
To study whether pairs of residues were coupled we calculated if the perturbation of maturation caused by the individual mutations x (ΔPx) and y (ΔPy) were additive or not when combined in the double mutant (ΔPxy). Additive effects suggest a lack of functional and/or structural interaction, whereas non-additivity suggests the opposite [18–20]. To quantify the maturation defect of wild-type, single and double mutants we calculated their half time of maturation (τ; time required for maturation of 50% of the assembled rBAT molecules co-precipitated immediately after the pulse); see below. As the difference in free energy between two end states is proportional to the logarithm of the equilibrium constant, we estimated the perturbation caused by a mutation x as: ΔPx=log(τx/τWT), where WT indicates wild-type. Non-additive values (i.e. ΔPxy–(ΔPx+ΔPy) ≠ 0) of more than ± 0.5 log units have been shown to have potential biological significance with respect to interactions between different amino acid residues .
To calculate the half time of maturation (τ), the endoglycosidase H data obtained from pulse chase experiments with wild-type rBAT and b0,+AT was fitted with GraphPad Prism. The maximum maturation value was set to 100% (i.e., all newly synthesized and assembled rBAT matures). The best fit was obtained with the hyperbolic equation: MT=100T/(τ+T), where MT is the percentage of mature rBAT (experimentally measured with endoglycosidase H assays) at the chase time T (Supplementary Figure S1). We assumed that, at least until 5 h of chase, the equation also fitted the maturation of the rBAT mutants analysed. Experiments to compare maturation of wild-type and mutants were performed at 300 min (5 h) of chase. Therefore, from the above equation, τ was calculated for wild-type rBAT and for each rBAT mutant as follows: τ=(100–M300)T/M300; where T was 300 min and M300 was the percentage of endoglycosidase H-resistant rBAT at 5 h of chase.
We used transient transfections of human rBAT and b0,+AT in HeLa cells to study biogenesis and functional expression of system b0,+ [9,11,13]. Individual N-glycans play diverse roles in the biogenesis of membrane proteins [14,21–23]. All rBAT orthologues have at least two N-glycan consensus sites, but neither their total number nor their positions are conserved (www.ensembl.org). Human rBAT has six N-glycosylation consensus sites (NXS/T; X is not proline): four in domain A (Asn214, Asn332, Asn495 and Asn513) and one in domains B (Asn261) and C (Asn575; Figure 1). N-glycans Asn261 and Asn575 lie within disulfide loops (in B, Cys242–Cys273; in C, Cys571–Cys666). The Cys571–Cys666 disulfide joins domain C with the C-terminal tail, which in turn ends in a third disulfide loop, Cys673–Cys685  (Figure 1).
N-glycans and C-terminal tail of human rBAT
Identification of rBAT N-glycans
Mutations that individually eliminate the N-glycan consensus site were constructed by serine/threonine to alanine replacement, as in other studies [24–26]. Wild-type and mutant rBAT were expressed in HeLa cells in the presence of 35S-Met, immunoprecipitated and the precipitates run in SDS/PAGE under reducing conditions. Without b0,+AT, rBAT remained core-glycosylated [9,11]. Five mutants ran faster than wild-type, indicating that rBAT has five N-glycans, attached to Asn261 (domain B), Asn332, Asn495, Asn513 (domain A) and Asn575 (domain C; Figure 2). The mutation C273S, which creates an additional N-glycosylation site, was also expressed and showed the expected lower mobility (Figure 2, lane 2). T216A rBAT had the same mobility as the wild-type, indicating that this site is not N-glycosylated.
Identification of the human rBAT N-glycans
N-glycan Asn575 facilitates maturation of rBAT-b0,+AT
Next, we did pulse chase experiments in HeLa cells transiently transfected with wild-type and N-glycan rBAT mutants together with b0,+AT, followed by immunoprecipitation with anti-b0,+AT antibodies to analyse b0,+AT-assembled rBAT. Immediately after the pulse, co-precipitated rBAT appears as a core-glycosylated band (∼90 kDa). A slower and wider band arises during the chase, indicating the maturation of the rBAT N-glycans in the Golgi complex [9,11]. The amount of co-precipitated rBAT did not change during a 5-h chase both for wild-type and mutants (compare 0 and 5 h of chase in Figure 3A; and data not shown), suggesting that none of the N-glycans is required for stability of the transporter (but see below). In contrast, T216A rBAT assembled with b0,+AT was degraded (less than 30% of rBAT at 5 h compared with 0 h of chase) and did not mature (Supplementary Figure S2A). This was expected: the cystinuria mutant T216M rBAT assembles with b0,+AT and is then degraded . To assess maturation of b0,+AT-assembled rBAT we used endoglycosidase H to measure core-glycosylated (endo H-sensitive) and mature (endo H-resistant) rBAT [9,11] (Figures 3B and 3C). More than 80% of wild-type rBAT was in the endo H-resistant form at 5 h of chase and the same was observed for mutants without one of the domain A or B N-glycans. Only the S577A mutation affected maturation: more than 50% of assembled S577A rBAT remained endo H-sensitive after 5 h of chase (Figures 3B and 3C). This was also seen in Figure 3(A) (compare the 5 h chase lanes of S577A with the other mutants and wild-type). Consistent with pulse chase experiments, Western blots from total membranes treated or not with endoglycosidase H detected the endo H-sensitive band only in the S577A rBAT mutant (T216A served as a control for the identification of this band, as this mutant does not mature; see above). Anyway, most of total S577A rBAT was endo H-resistant, indicating that, despite the maturation delay, it eventually matured (Supplementary Figure S2B). Moreover, S577A rBAT mutant elicited L-arginine transport activity similar to wild-type (Figure 4E). To test if the maturation delay could be due to the serine-to-alanine change instead of the absence of the N-glycan per se, maturation of mutants N575Q and N575D was also quantified. They also caused a slower maturation, similar to the S577A mutant (Supplementary Figures S2C and S2D). Overall, these data indicated that the absence of the N-glycan Asn575 reduced the rate, but not the efficiency, of transporter maturation.
Stability and maturation of rBAT N-glycan mutants
Analysis of sng mutants of rBAT
To further study rBAT N-glycans, we created the rBAT mutants carrying single N-glycans (sng mutants) and without N-glycans (Nglyc-less) and measured stability and maturation as above. The mutants assembled with b0,+AT and, at least until 5 h of chase, had similar stability compared both with wild-type and with S577A mutant, with the exception of Nglyc-less, which showed a small but significant reduction in stability (Figures 4A and 4B). Maturation of domain A and B sng mutants was slower than wild-type rBAT, but it was similar to wild-type in the sng575 mutant (Figures 4C and 4D). This suggested that the Asn575 N-glycan alone is necessary and sufficient to achieve the maximum rate of maturation and that the other N-glycans could partially compensate for its absence. The sng mutants were functional, albeit with reduced activities compared with wild-type and S577A rBAT. The Nglyc-less mutant was still transport-competent (indicating that N-glycans are not essential for carrier function), but its transport activity was strongly diminished (Figure 4E).
Role of the C673–C685 disulfide loop in biogenesis
The C-terminal tail (human Leu655–Cys685) is present in all rBAT orthologues (www.ensembl.org). It shows no homology with any other sequence. Residues Leu655–Cys666 are part of the disulfide loop Cys571–Cys666 that includes almost whole domain C. A deletion, including Cys666, of most of the rat rBAT tail is not functional . This is not surprising, because absence of the Cys571–Cys666 disulfide completely disrupts biogenesis . Another disulfide loop is formed by residues Cys673–Cys685. A double mutation eliminating this disulfide does not disturb biogenesis .
Given that Leu655–Cys666 is linked to domain C and that the role of the Cys673–Cys685 disulfide loop in transporter biogenesis and function is not known, in the present work, we decided to centre the study of the rBAT tail on its C-terminal loop. Biogenesis and function of loop mutants were measured as for the N-glycan mutants in the preceding section. First, the whole loop was deleted. The mutant (Δ[673–685]) assembled with b0,+AT, but it was rapidly degraded and did not mature (Figures 5A and 5B). Amino acid uptake was also abolished (Figure 5C). Therefore, the loop was essential for folding of the transporter. We further dissected this loop by alanine-scanning mutagenesis of the residues 674–684. All mutants assembled with b0,+AT, but differentially affected biogenesis (Figure 6). Indeed, when their maturation and stability at 5 h of chase were plotted, three groups appeared (Figure 6C). Group 1 (S676A, V677A, I680A, T683A, S684A) presented no maturation defect and a stability between 60% and 90%; Group 2 (S675A, L678A, N679A) were stable like wild-type and showed a maturation defect similar to the N-glycan mutant S577A; finally, Group 3 (Y674A, L681A, Y682A) matured less than Group 2 and their stability values were close to 60%. Combination of the effects of some of the mutants, especially from Group 2 and 3, may explain the strong biogenesis defect of the deletion mutant Δ[673–685] (Figure 5). None of the mutants were like cystinuria mutants  (Figure 6C), suggesting that, although some of them (i.e. Group 3) may impair folding of rBAT, the impairment is not comparable to that observed for cystinuria mutants. We confirmed this further with L-arginine transport experiments (Figure 6D): all mutants were functional, with values above ∼35% of the wild-type. Combination of maturation and stability defects correlated with the strongest decrease in transport activity (Group 3). Maturation delay alone either did not impair (S577A and S675A) or decreased transport function (but not more than 40% for L678A and N679A). A mild stability defect alone (Group 1 mutants V677A, I680A, T683A, S684A) did not affect L-arginine uptake. In contrast, the Group 1 mutant S676A displayed a stability and L-arginine uptake decreases comparable to Group 3 mutants, but without any maturation delay.
The C-terminal disulfide loop 673–685 of rBAT is essential for biogenesis
As L678P is the only cystinuria mis-sense mutation in the 674–684 region , we also analysed its biogenesis. Pulse chase experiments showed that the L678P mutant associated with b0,+AT but, in contrast with wild-type and L678A, it was rapidly degraded, suggesting a strong folding defect, similar to the other cystinuria mutations in the ectodomain of rBAT (Supplementary Figure S3A) . The less conservative change (L/P compared with L/A) may account for the strong effect. Transport activity was also greatly reduced (Supplementary Figure S3C), correlating with the low expression of mature rBAT observed in Western blots (Supplementary Figure S3B). Decreased transport activity can explain the cystinuria phenotype of L678P .
Double-mutant cycle analysis
To date, Group 2 mutants and the N-glycan mutant S577A are the only ones that cause maturation delay (similar for the four mutants) of the transporter, without changes in stability (at least until 5 h of chase, Figures 3 and 6). Despite the long primary sequence distance between the N-glycan Asn575 and the C-terminal disulfide loop (∼100 amino acid residues), the formation of the disulfide Cys571–Cys666 in the rBAT ectodomain after assembly with b0,+AT should bring nearer these two rBAT elements in the native structure. Therefore, we used a double-mutant cycle approach to investigate whether S675A, L678A and/or N679A had additive effects on the maturation phenotype of S577A. Pulse chase experiments were performed in parallel with wild-type, single and double mutants (Figure 7). At 5 h of chase, S577A–S675A and S577A–N679A were stable (like the single mutants); S577A–L678A showed a small decrease in stability (Figures 7A and 7C). Therefore, we measured stability also at 18 h: S577A and L678A stability was diminished by 35% and 44% respectively, whereas double mutants showed decreases of 52% (S577A–S675A), 60% (S577A–L678A) and 68% (S577A–N679A); (result not shown). Comparison with the reduction in stability found, already at 5 h of chase, in Group 3 mutants (40%–50%; Figure 6) and in cystinuria mutants (60%–75%; Figure 6C) , suggested that double mutations did not greatly compromise the folding of the transporter. Anyway, in order to avoid possible confounding effects of the stability decrease, we did the double-mutant cycle analysis to 5 h of chase. Endoglycosidase H assays revealed a greater delay in maturation of the double mutants compared with their single mutant controls (Figures 7B and 7C). The double mutants displayed a tendency to a decreased L-arginine transport compared with the single mutants, although it was statistically significant only in the case of the S577A–S675A mutation (Figure 7D).
Alanine-scanning analysis of the C-terminal disulfide loop
Double-mutant cycle analysis of Group 2 residues with the N-glycan Asn575
To test for additivity (see ‘Experimental’) of the maturation defects we first obtained the half-time of maturation of the mutants from the percentage of endoglycosidase H-sensitive rBAT molecules at 5 h. We assumed that the maturation kinetic model of the wild-type transporter (Supplementary Figure S1) was also followed by the mutants, at least until 5 h of chase. Then, the estimated maturation perturbation (ΔP) caused by each mutant was calculated from the logarithm of the ratio of the half-time of mutant relative to the half-time of wild-type. Figure 7(E) shows the effects for single and double mutants and compares the latter ones with the expected values provided that the maturation problems of the single mutants were additive. For S577A–S675A and S577A–L678A the results were compatible with additi-vity and therefore with independence of effects caused by each single mutant. Instead, the maturation delay caused by the double mutant S577A–N679A was partially additive (Figure 7E, panel III). As suggested by Mildvan , we used ΔP of the double mutant (ΔP=1.13 log units) as the reference point (‘inverse thinking’) to analyse the results. Restoring Ser577 to the N679A mutant decreased ΔP to 0.78 (–0.35 log units), measuring the contribution of the N-glycan to maturation. Restoring Asn679 to the S577A mutant decreased ΔP to 0.87 (–0.26 log units), measuring the contribution of Asn679 to maturation. By definition, ΔP of the wild-type is 0. If additive, a decrease of–0.61 log units [(–0.35)+(–0.26)] is expected for the wild-type (that is, upon restoration of both Asn679 and the N-glycan Asn575). The difference between the real decrease (–1.13 log units) and the expected decrease (–0.61 log units) in the wild-type is equal to–0.52 log units [–1.13–(–0.61)] and it measures the contribution of the possible co-operativity between these two elements to maturation of the wild-type transporter . This is compatible with a potential structural and/or functional interaction between the N-glycan Asn575 and the tail residue Asn679.
We have examined the number and identity of the N-glycans of rBAT and their role in biogenesis of system b0,+, with a focus on the domain C Asn575 N-glycan. Next, we showed that the Cys673–Cys685 C-terminal disulfide loop greatly contributes to stability and maturation of system b0,+. Our analysis pointed also to an interaction between the N-glycan and the disulfide loop.
N-glycans of human rBAT
The ectodomain of rBAT has five N-glycans (Asn261 in domain B, Asn332, Asn495 and Asn513 in domain A and Asn575 in domain C). A structural homology model of rBAT located the N-glycans in an external position , without interfering with helper–carrier subunit interactions (according to the structural model of 4F2hc–LAT2 ). In contrast, Asn214 is an internal residue in the same rBAT model and in 4F2hc and related α-glucosidases structures , suggesting why it is not used in vivo. Indeed, the cystinuria mutant T216M  and the T216A mutant (Supplementary Figure S2A) assembled with b0,+AT, but were retained in the ER and degraded, suggesting that perturbation of Asn214 strongly impairs folding.
The N-glycan Asn575 facilitates maturation of system b0,+
N-glycans have been shown to intrinsically enhance solubility and folding energetics of glycoproteins [14,29]. Furthermore, they have critical roles in folding, degradation and trafficking of glycoproteins within the cell, serving as signals for the recruitment of chaperones and other quality control and trafficking factors [14,21,23,30,31]. Analysis of rBAT N-glycosylation revealed that stability decreased at an early chase time only when all N-glycans were removed. Mutants of individual N-glycans and with single N-glycans did not reveal significant effects on stability. However, the amino acid transport activity decreased in the sng mutants (both in those with and without maturation problems) and the decrease was stronger when all N-glycans were eliminated (Figure 4E). Therefore, although not essential for system b0,+, N-glycans affect amino acid transport most probably due to a combination of effects on the intrinsic carrier activity and on its biogenesis. In particular, the results indicated that the N-glycan in domain C, Asn575, is important for maturation of the transporter.
The mutation S577A eliminated the Asn575 N-glycan. Maturation of this mutant transporter was strongly delayed already at 5 h of chase (Figure 3), whereas a small reduction in stability was detected only at very long chase times. In Western blots, most of S577A rBAT was found in the mature form, suggesting that only the kinetics of maturation was impaired, not the efficiency. The maturation effect was probably mediated by the N-glycan per se, because mutations of Asn575 either to D or to Q had the same consequences. Absence of any of the other N-glycans, individually, did not disturb biogenesis. Assuming that S577A–rBAT–b0,+AT follows the same maturation kinetic model as the wild-type (see ‘Experimental’), the absence of this N-glycan delays maturation at least 5-fold (i.e. the maturation half-time of the wild-type transporter was ∼1 h compared with ≥5h of the S577A mutant). rBAT-b0,+AT heterodimers carrying a single domain A or B N-glycan matured at a rate similar to or reduced compared with S577A, indicating that they were able to substitute for the N-glycan Asn575, but only partially. Strikingly, when Asn575 was the only N-glycan present, maturation was similar to that of the wild-type transporter (Figure 4). We concluded that the Asn575 N-glycan is necessary and sufficient to support the maximum rate of maturation of system b0,+. The N-glycan site Asn575 is conserved in all known reptile and bird and in ∼2/3 of mammalian rBAT orthologues. In the structural homology model of rBAT, it is found in the loop between the antiparallel β1 and β2 strands (Cβ1 and Cβ2) of the β-sheet. This β-sheet forms the hydrophobic interface of domain C interacting with helix α6, 7 and 8 of the TIM barrel domain A . These elements are conserved in the structures of related α-glucosidases [5,32] (Figure 1). M467T, located in Aα7, is the most frequent rBAT cystinuria mutation and strongly impairs folding , most probably by disrupting interactions between domains A and C. Cys571, four residues before Asn575, is part of Cβ1 and forms a disulfide with Cys666 on the rBAT tail. This disulfide joins domain C with the C-terminal tail, which ends with the disulfide loop Cys673–Cys685  (Figure 1). With this in mind, at least two non-mutually-exclusive mechanisms may explain the central role of this N-glycan in biogenesis of human rBAT-b0,+AT. First, it may facilitate folding of the rBAT ectodomain within the heterodimer, either through direct interactions with the protein moiety (see below) or through the binding to ER lectin chaperones. Earlier work indicated that the calnexin chaperone system facilitates post-assembly folding of the rBAT ectodomain . AMYLPRED2 employs a consensus of eleven different methods to search for potential aggregation-prone regions in proteins . It found Aα7 and Cβ2 peptides among the most prominent candidates (result not shown). Calnexin binding to the N-glycan Asn575 may hinder non-native contacts of those peptides, thus enhancing folding of domain C and formation of the domain A/domain C interface. Additionally, while acting on domain C folding, it may help in the correct disulfide pairing of Cys571 with Cys666 and therefore, also of Cys673 with Cys685. Second, the N-glycan Asn575 may be part of a signal that binds a cargo receptor to accelerate the ER-exit of the transporter. ER-exit of membrane proteins often depends on cytosolic motifs recognized by the COPII sorting machinery [34–38]. In some cases, lumenal domains of membrane proteins are bound by cargo receptors which recruit the COPII sorting machinery via their cytosolic motifs [39–41]. These lumenal signals can require the presence of specific N-glycans . b0,+AT has a C-terminal cytosolic ER-exit signal, functional only when b0,+AT is assembled with rBAT , suggesting that rBAT itself provides trafficking information. The N-glycan Asn575 could be part of such information.
The C-terminal tail disulfide loop of rBAT is essential for biogenesis of system b0,+
The C-terminal tail residues 655–666 are linked to domain C through the essential Cys571–Cys666 disulfide and the Cys673–Cys685 disulfide seems dispensable for the transporter . Moreover, it was not yet possible to functionally or structurally relate the C-terminal tail disulfide loop 673–685 to any other region of the transporter. Therefore, we decided to study this disulfide loop. Its complete deletion conferred a cystinuria-like phenotype : Δ[673–685]-rBAT assembled with b0,+AT, but did not mature and was rapidly degraded (Figure 5), indicating that the loop is essential for biogenesis of system b0,+. To identify elements responsible for the drastic phenotype, residues 674–684 where individually mutated to alanine. Group 1 mutants showed a mild decrease in stability. Four of these residues (676, 677, 683 and 684) are, within the 674–684 region, among the less conserved in rBAT orthologues. S676A displayed reduced transport activity, which correlated with the strongest decrease in stability. The functional defect of this mutant deserves further study. Unexpectedly, despite Ile680 being the most conserved residue, it showed the mildest stability defect within this group and no decrease in transport function. The other highly conserved residues are 674, 681 and 682 (Group 3). Here, conservation correlated with defective maturation and stability of the alanine mutants. Stability and functional defects were still milder compared with cystinuria mutants (Figure 6) , arguing that the individual contribution of these residues to stability is, though relevant, not a major one. Anyway, combination of the defects of Group 3 mutants may explain the Δ[673–685]-rBAT phenotype. Residues 675, 678 and 679 (Group 2) are partially conserved. Their alanine mutants had a maturation delay similar to that of the S577A N-glycan mutant (Figure 6). Interestingly, the cystinuria mutant L678P  had stronger biogenesis defects than L678A, explaining the cystinuria phenotype.
How could this disulfide loop affect biogenesis of system b0,+? Nothing is known about the structure of the C-terminal tail of rBAT, including the disulfide loop. Secondary structure prediction assigned either an α-helix or a β-strand to the disulfide loop, depending on the method (result not shown). Together with Aα7 and Cβ2, the region 678–682 is one of the best candidates for aggregation-prone regions found by AMYLPRED in the rBAT ectodomain (result not shown). This peptide contains two Group 2 and two Group 3 residues. Although just a prediction, it suggests that the acquisition of the native structure of this region is likely to be particularly important to assure the correct folding of rBAT-b0,+AT. Its location in the C-terminus of rBAT could mark its folding as an important, late quality-control step in the biogenesis of the transporter.
Double mutant analysis
We found that, despite S577A-N679A having a stronger maturation delay, the contributions of the single mutants N679A and S577A to this delay were only partially additive (Figure 7), a scenario consistent with co-operation between the Asn575 N-glycan and the residue Asn679 on maturation of the transporter. We propose that these two elements interact during biogenesis. Without structural information, it is not yet possible to directly support such an interaction. However, formation of the disulfide Cys571–Cys666 should approximate the Asn575 N-glycan to the disulfide loop 673–685. As stated above, Cβ2 (immediately following the loop where the N-glycan Asn575 is found) and Asn679 may lie in aggregation-prone regions. Non-native interactions between the Asn575 N-glycan and Asn679 may be required for folding. Alternatively, recruitment of calnexin to the N-glycan may facilitate contacts between Asn679 and regions close to the N-glycan, for instance Cβ2. Moreover, direct interactions of the N-glycan with Asn679 and perhaps other amino acid residues could energetically favour folding. In other proteins, this has been shown for some N-glycans interacting both with nearby and with distant amino acid residues [14,43,44]. Finally, the observed co-operation could arise because the Asn575 N-glycan and Asn679 may be part of a lumenal ER-exit signal that facilitates ER-exit of system b0,+. Further studies are needed to distinguish between these possibilities.
Mònica Rius performed most of the experiments and contributed to the writing of the manuscript. Laura Sala performed experiments. Josep Chillarón designed and supervised the study, performed experiments and wrote the manuscript.
We are grateful for the support of Dr A.F. Valledor, Dr T. Stratmann, Dr J. Planas, Dr J. Gutiérrez and Dr G. Viscor during the final part of the present work.
This work was supported by the Spanish Ministerio de Ciencia e Innovación [grant number BFU2009-07215/BMC] to J.C.