Novel cystine transporter in renal proximal tubule identified as a missing partner of cystinuria-related plasma membrane protein rBAT/SLC3A1

Significance

Although molecular identification of transporters in mammals seems almost settled, some long-proposed transporters still remain to be revealed. The second cystine transporter in renal cystine reabsorption is one of such transporters. Its genetic defect has been proposed to be responsible for a type of cystinuria distinct from that caused by the mutations of the already known cystine transporter. In this study, we have found a membrane protein SLC7A13 as the second cystine transporter with proposed characteristics, and provided a possible clue to the genetics of previously unidentified cystinuria. Intricate functional coupling of SLC7A13 with the nearby glutamate transporter is also proposed. We have solved long-lasting problems in renal cystine transport physiology and paradoxes regarding the unmatched distribution of cystine transporter components.

Abstract

Heterodimeric amino acid transporters play crucial roles in epithelial transport, as well as in cellular nutrition. Among them, the heterodimer of a membrane protein b^0,+AT/SLC7A9 and its auxiliary subunit rBAT/SLC3A1 is responsible for cystine reabsorption in renal proximal tubules. The mutations in either subunit cause cystinuria, an inherited amino aciduria with impaired renal reabsorption of cystine and dibasic amino acids. However, an unsolved paradox is that rBAT is highly expressed in the S3 segment, the late proximal tubules, whereas b^0,+AT expression is highest in the S1 segment, the early proximal tubules, so that the presence of an unknown partner of rBAT in the S3 segment has been proposed. In this study, by means of coimmunoprecipitation followed by mass spectrometry, we have found that a membrane protein AGT1/SLC7A13 is the second partner of rBAT. AGT1 is localized in the apical membrane of the S3 segment, where it forms a heterodimer with rBAT. Depletion of rBAT in mice eliminates the expression of AGT1 in the renal apical membrane. We have reconstituted the purified AGT1-rBAT heterodimer into proteoliposomes and showed that AGT1 transports cystine, aspartate, and glutamate. In the apical membrane of the S3 segment, AGT1 is suggested to locate itself in close proximity to sodium-dependent acidic amino acid transporter EAAC1 for efficient functional coupling. EAAC1 is proposed to take up aspartate and glutamate released into luminal fluid by AGT1 due to its countertransport so that preventing the urinary loss of aspartate and glutamate. Taken all together, AGT1 is the long-postulated second cystine transporter in the S3 segment of proximal tubules and a possible candidate to be involved in isolated cystinuria.

The heteromeric amino acid transporter (HAT) family is one of the major amino acid transporter families responsible for cellular uptake and epithelial transport (1–3). HATs form heterodimers composed of a 12 membrane spanning light chain (SLC7) that catalyzes transport functions and a single membrane spanning heavy chain (SLC3) essential for plasma membrane localization and stabilization of the light chains. Two heavy chains, SLC3A1/rBAT and SLC3A2/4F2hc/CD98hc, covalently bound to light chains via a disulfide bridge have been identified so far (4–6). 4F2hc interacts with most of the light chains in HATs whereas rBAT has been known to form a heterodimer only with b^0,+AT/SLC7A9. Because the rBAT-b^0,+AT complex is presented on the apical membrane of proximal tubules in the kidney and involved in the reabsorption of cystine and dibasic amino acids, the mutations of either rBAT or b^0,+AT cause cystinuria, a disorder of renal reabsorption of cystine and dibasic amino acids leading to serious renal lithiasis due to low solubility of cystine (7).

An unsolved paradox on rBAT and b^0,+AT has been the discrepancy between the distribution of rBAT and that of b^0,+AT (5, 8–10). rBAT is the most abundant in the S3 segment of proximal tubules, and its expression declines toward the S1 segment (11, 12). In contrast, the expression of b^0,+AT is highest in the S1 segment and decreases toward the S3 segment (5, 8). Furthermore, even in b^0,+AT-deficient mice, heterodimers containing rBAT still have been observed (13). Therefore, it has been proposed that unknown partners of rBAT exist in the S3 segment (5, 9, 14, 15).

The HAT family includes two members, AGT1/SLC7A13 and Asc2, whose heavy chains have not been identified (16, 17). Among them, aspartate/glutamate transporter 1 (AGT1) has been identified as an Na⁺-independent acidic amino acid transporter expressed specifically in the kidney (17). In this study, we have generated new anti-AGT1 antibodies to search for the unknown heavy chain(s), by means of coimmunoprecipitation followed by mass spectrometry, and have revealed that rBAT is a heavy chain of AGT1. AGT1 was detected at the apical membrane of the S3 segment in renal proximal tubules. A transport assay of the AGT1-rBAT heterodimer reconstituted into proteoliposomes revealed that it transports cystine as well as aspartate and glutamate. We conclude that AGT1 is a strong candidate for the “missing partner” of rBAT and a second cystine transporter in the kidney.

Results

Identification of a Heavy Chain for AGT1.

Anti-AGT1 antibodies were newly generated against C-terminal 24 amino acid peptides. The specificities of the antibodies, anti-AGT1(M) and anti-AGT1(G) (SI Materials and Methods), were confirmed using 3xFLAG-tagged AGT1 as shown in Fig. S1. In Western blot, the antibodies specifically recognized a band for AGT1 in the male kidney but not in the female (Fig. 1A). mRNA for AGT1 was also detected only in the male kidney by semiquantitative RT-PCR (Fig. S2A). There are no perfect palindrome estrogen (ERE) and androgen (ARE) response elements in the 10 kbp upstream of the AGT1 gene, as well as in the AGT1 gene on mouse chromosome 4. Instead, three imperfect palindrome AREs and a lot of half sites of ARE and ERE were found in the regions (Fig. S2B).

Fig. S1.

Characterization of newly generated anti-AGT1 antibodies, anti-AGT1(G) and anti-AGT1(M). (A) 3xFLAG-AGT1 was cotransfected with rBAT and expressed in HEK293 cells. The crude membrane fractions were prepared and subjected to SDS/PAGE and Western blot. Anti-AGT1(G) and anti-AGT1(M) antibodies recognized a band that was also detected by anti-FLAG antibody, indicating that the bands shown by the arrows are 3xFLAG-AGT1. (B) 3xFLAG-AGT1 was expressed in HEK293 cells and subjected to immunofluorescence microscopy by using the newly generated antibodies. The signals detected with anti-AGT1(G) and anti-AGT1(M) antibodies were well-matched with those with anti-FLAG antibody. (Scale bar: 10 µm.) (C) Anti-AGT1(G) and anti-AGT1(M) antibodies recognized endogenous AGT1 in the mouse kidney crude membranes (arrows) but not in the whole brain crude membranes. Anti-AGT1(G) antibody was more specific to AGT1 in kidney crude membranes than the anti-AGT1(M) antibody. (D) To confirm that the size difference between 3xFLAG-AGT1 in A and endogenous AGT1 in C is due to the 3xFLAG tag, the purified complex of rBAT and 3xFLAG-AGT1 was incubated with (+) or without (−) enterokinase for 16 h at 25 °C to cleave out the 3xFLAG-tag. Samples were analyzed by SDS/PAGE and Western blot using the anti-AGT1(M) antibody. The upper and lower bands represent 3xFLAG-AGT1 and nontagged AGT1, respectively, confirming that the size difference between 3xFLAG-AGT1 in A and endogenous AGT1 in C is due to the 3xFLAG tag. BBMV prepared from mouse kidney was examined as a control of endogenous AGT1 (BBMV).

Fig. 1.

AGT1-heavy chain heterodimer in mouse kidney. (A) Expression of AGT1 in kidney. Western blot was performed on crude membrane fractions from two female and two male mice using the anti-AGT1(M) antibody. Western blots (Left and Right) were performed in the presence (+DTT) or absence (− DTT) of 100 mM DTT, respectively. Filled arrowheads indicate AGT1. Open arrowhead points to heterodimers of AGT1 and its heavy chain (AGT1-hc heterodimer) whereas the open arrow indicates the oligomeric complex. (B) Immunoprecipitation on renal brush-border membrane vesicles (BBMVs) with the anti-AGT1(M) antibody or anti-rBAT antibody. Western blot was performed with the anti-rBAT antibody or anti-AGT1(M) antibody in the presence of 100 mM DTT. The small arrow (Left) and the arrowhead (Right) indicate the bands for rBAT and AGT1, respectively. The large arrow (Right) indicates the AGT1 homodimer. Normal rabbit IgG was used as a control for immunoprecipitation. Asterisks are the bands derived from IgG. (C) Expression of AGT1 and rBAT in the mutant mouse kidney Western blot was performed on renal BBMVs from different genotypes [WT (Slc7a9^+/+ and Slc3a1^+/+) and D140G (Slc3a1 missense mutation)] of male (M) and female (F) mice in the nonreducing condition. The anti-AGT1(G) and anti-rBAT antibodies detected AGT1-rBAT heterodimer (arrowhead) and its oligomers, including dimers of heterodimeric complexes (arrowhead 2×) and higher oligomeric complexes (arrow).

Fig. S2.

AGT1 mRNA expression and the estrogen and androgen response element search. (A) Semiquantitative RT-PCR of AGT1. mRNA for AGT1 was detected only in the male mouse kidney, but not in the female mouse kidney. Total RNA was obtained from the kidneys of C57B6/J male and female mice using ISOGEN II (Nippon Gene) and was reverse-transcribed with PrimeScript RT master mix (Takara) according to the manufacturer’s protocol. Synthesized cDNA was subjected to PCR amplification using the specific primers for AGT1 mRNA (FW, gcaatggattcaaagaaggaaat; RV, ttcagcactgagctcactgtctga). The primer pair for GAPDH (glyceraldehyde-3-phosphate dehydrogenase) (FW, ccagggctgccatttgcagtggcaa; RV, acccttttggctccacccttca) was used as a control. The amplified DNA fragments were separated on an agarose gel electrophoretically and visualized by staining with SYBR Safe (Thermo Fisher Scientific). (B) Palindrome estrogen (ERE) and androgen (ARE) response elements were searched in the 10 kbp upstream of the AGT1 gene, as well as in the AGT1 gene, with the GENETYX-MAC program (GENETYX Inc.). The consensus sequence for ARE is AGAACANNNTGTTCT. For ERE, two consensus sequences were tested (ERE1, GGTCANNNTGACC; ERE2, GGTCACNNNGTGACC). No perfect palindrome was detected whereas three imperfect palindrome AREs (similarity >86%; shown with blue bars) and a lot of half sites of ARE (blue dotted bars), ERE1 (orange dotted bars), and ERE2 (magenta dotted bars) were found. Target sequence, Mus musculus strain C57BL/6J chromosome 4, GRCm38.p3 C57BL/6J National Center for Biotechnology Information (NCBI) reference sequence: NC_000070.6 from 19808727 to 19842213.

AGT1 migrated at ∼40 kDa in the reducing condition whereas the bands around 140 kDa and over 240 kDa were detected in the nonreducing condition (Fig. 1A). The band shift from 140 kDa to 40 kDa was presumably due to the breakage of a disulfide bridge between AGT1 and its heavy chain in the reducing condition. The band over 240 kDa seems to be oligomers of heteromeric complexes similar to those observed for HATs (5, 18–20).

To identify the unknown partner of AGT1, AGT1 was immunoprecipitated using anti-AGT1 antibodies (Fig. S3A). The elution fraction containing AGT1 in the immunoprecipitation was subjected to shotgun mass spectrometry. In all of the independent experiments [three times with anti-AGT1(M) and once with anti-AGT1(G)], rBAT was detected with high protein scores (21) (Fig. S3B). At the best, 19 peptides were identified on the rBAT amino acid sequence, which covered 31.7% of the protein. Peptides corresponding to AGT1 were also detected by mass spectrometry. None of the peptides corresponding to 4F2hc or b^0,+AT were detected. In addition, none of the rBAT peptides were identified from samples precipitated with normal rabbit IgG.

Fig. S3.

Identification of proteins coimmunoprecipitated with AGT1 and Western blot of mouse kidney crude membrane fractions. (A) AGT1 was successfully precipitated from kidney crude membrane fraction (Kidney MF) by anti-AGT1(M) antibody (lane 3) and detected by anti-AGT1(G) antibody (arrow, homodimer of AGT1; arrowhead, AGT1 monomer). A membrane fraction prepared from mouse brain was used as a negative control because AGT1 was not detected in brain in Northern blot (17) and Western blot (Fig. S1C). AGT1 was not precipitated from brain crude membrane fraction (Brain MF) with anti-AGT1 antibody (lane 2) and from kidney membranes when the beads without IgG were used (No IgG, lane 5). The second elution fraction (Elution 2) did not contain AGT1, but a band corresponding to IgG around 100 kDa (*, lane 4). (B) Fourteen peptides identified by LC-MS/MS are shown on the mouse rBAT amino acid sequence (bold). The transmembrane domain is indicated with an underline. (C) Western blot of mouse kidney crude membrane fractions using anti-b^0,+AT antibody (Upper) or anti-rBAT antibody (Lower) in the presence (+ DTT) or absence (− DTT) of 100 mM DTT. The amount of b^0,+AT was similar between female and male whereas rBAT was much more abundant in male compared with female. The AGT1-rBAT heterodimer (AGT1-rBAT) and b^0,+AT-rBAT heterodimer (b^0,+AT-rBAT) were distinguished by Western blot using anti-rBAT antibody because the AGT1-rBAT heterodimer did not exist in female kidneys. 2X, dimers of heterodimeric complexes.

To confirm rBAT as a partner of AGT1, immunoprecipitated samples were analyzed by Western blot. As shown in Fig. 1B, rBAT was detected after immunoprecipitation using anti-AGT1 antibody, but not with normal IgG. Moreover, AGT1 was detected by immunoprecipitation with anti-rBAT antibody (Fig. 1B). After overnight incubation for immunoprecipitation in the presence of detergent, AGT1 appeared more abundantly as homodimers (around 70 kDa in Fig. 1B, Right and Fig. S3A) than monomers (35–40 kDa in Fig. 1B, Right and Fig. S3A) whereas monomeric AGT1 was mainly detected in freshly prepared proteins (Fig. 1A and Fig. S3A).

To further confirm rBAT as an AGT1 partner, the expression of AGT1 was examined in mice carrying a missense mutation in rBAT causing the loss of rBAT protein in brush-border membrane vesicles (BBMVs) (Fig. 1C, Right) (10, 22). In the mutant mice, the band corresponding to the complex of AGT1 and its heavy chain detected by anti-AGT1 antibody was diminished in BBMVs, supporting the idea that rBAT is the heavy chain of AGT1 (Fig. 1C, Left). The Western blots of BBMVs (Fig. 1C) and crude membrane fractions from WT mouse kidney (Fig. 1 and Fig. S3C) indicate that rBAT protein, as well as AGT1, is more abundant in male than female whereas the amount of b^0,+AT has no biological sex difference (Fig. S3C). The AGT1-rBAT heterodimer and b^0,+AT-rBAT heterodimer were distinguishable by Western blot using anti-rBAT antibody because the AGT1-rBAT heterodimer did not exist in the female kidney (Fig. S3C).

One of the known functions of heavy chains of HATs is to localize the light chains to the plasma membrane (19, 23–26). To examine the role of rBAT in the localization of AGT1, AGT1 was expressed with rBAT in HEK293 cells. As shown in Fig. S4 A–C, expression of 3xFLAG-AGT1, rBAT, or 4F2hc alone resulted in the intracellular localization. In contrast, when 3xFLAG-AGT1 and rBAT were coexpressed, they both were localized on the plasma membrane (Fig. S4 D–F). 4F2hc, the other heavy chain of HATs, did not support membrane localization of 3xFLAG-AGT1 (Fig. S4 G–I).

Fig. S4.

Association of AGT1 and rBAT in cultured cells. Immunofluorescence microscopic analysis was performed on HEK293 cells expressing 3xFLAG-AGT1 (A), rBAT (B), 4F2hc (C), 3xFLAG-AGT1 and rBAT (D–F), or 3xFLAG-AGT1 and 4F2hc (G–I). 3xFLAG-AGT1 was detected by anti-FLAG antibody and is shown in cyan whereas rBAT and 4F2hc are shown in green. Blue color indicates DAPI staining. 3xFLAG-AGT1 and rBAT were colocalized on the plasma membrane when they were coexpressed (arrow heads, D–F). (Scale bars: 10 μm.)

Localization of Expression of AGT1 in Mouse Kidney.

Immunohistochemical analysis of the male mouse kidney using an anti-AGT1(G) antibody revealed strong AGT1 immunoreactivity on the renal tubules in the outer stripe of the outer medulla and medullary ray (Fig. 2 A and B). The AGT1 protein was mainly localized on the apical membrane of the tubules (Fig. 2 B and C). No staining was detected in the female mouse kidney by the anti-AGT1(G) antibody (Fig. S5 A–C). The localization of AGT1 in the male mouse kidney was confirmed by the anti-AGT1(M) antibody (Fig. S5 D–F). The localization of AGT1 was further confirmed by in situ hybridization showing a similar pattern of positive signals in the male mouse kidney and no signal in the female kidney (Fig. S6). Structured illumination microscopy (SIM) revealed AGT1 immunoreactivity at the surface of apical membranes, which is evident compared with the fluorescence from apical membrane staining marker lectin phytohemagglutinin-L (PHA-L) (Fig. 2 D–F).

Fig. 2.

Localization of AGT1 and rBAT in the male mouse kidney. (A–C) Localization of AGT1 in immunohistochemistry. AGT1 immunoreactivity was detected on the renal tubules located in the outer stripe of the outer medulla (red two-way arrow in A) and medullary ray (asterisk in A and arrows in B) in low and middle magnification views (A and B). The “C” in A indicates the renal cortex. AGT1 immunoreactivity was localized on the apical membrane of the tubules (small arrows in C). AGT1 immunoreactivity was less detected in distal convoluted tubules characterized by macula densa (large arrow in C). (Scale bars: A, 1 mm; B, 200 µm; C, 100 µm.) (D–F) Structured illumination microscopy (SIM) analysis of AGT1. The AGT1 signal is shown in cyan (D) whereas the signal for lectin PHA-L used as an apical membrane marker is shown in green. Their merged image is shown in F. (Scale bar: 10 µm.) (G–L) Coimmunofluorescence staining of AGT1 and rBAT on the apical membrane of renal tubules. Low magnification views indicate the colocalization of AGT1 and rBAT on the tubules in the outer stripe of the outer medulla (“OSOM” in G) and medullary ray (asterisk in G). (G–I) The “C” in G indicates the renal cortex. AGT1 and rBAT are colocalized on the apical membrane of the tubules in high magnification views (J–L). Signals for AGT1 and rBAT are shown in red (G and J) and green (H and K), respectively. Merged images are shown in I and L. (Scale bars: I, 500 µm; L, 25 µm.) The anti-AGT1(G) antibody was used for immunohistochemistry and immunofluorescence experiments.

Fig. S5.

Immunohistochemistry of the female mouse kidney using anti-AGT1(G) antibody and immunohistochemistry of the male mouse kidney using anti-AGT1(M) antibody. (A–C) No signal was detected in the female mouse kidney with anti-AGT1(G) antibody. (D and E) Similar to the results shown in Fig. 2 using anti-AGT1(G) antibody, anti-AGT1(M) antibody stained the renal tubules located in the outer stripe of the outer medulla (red two-way arrow in D) and medullary ray (asterisk in D and arrows in E) in the low and middle magnification views (D and E). The “C” in D indicates the renal cortex. In the high magnification view, staining with anti-AGT1(M) antibody was localized on the apical membrane of the tubules (small arrows in F). (Scale bars: A and D, 1 mm; B and E, 200 µm; C and F, 100 µm.)

Fig. S6.

In situ hybridization of AGT1 in the kidney of male and female mice. The anti-sense probe for AGT1 hybridized to the tubules of the outer strip of the outer medulla (“OSOM”) and medullary ray (asterisk) whereas the sense probe did not exhibit hybridization signals, confirming the expression of AGT1 in the tubules of the outer strip of the outer medulla and medullary ray at the mRNA level. The “C” indicates the renal cortex. Furthermore, the lack of AGT1 signals in the female mouse kidney (Bottom) confirmed predominant expression of AGT1 in the male mouse kidney.

Colocalization of AGT1 and rBAT in the mouse kidney was demonstrated by immunofluorescence double staining (Fig. 2 G–L). Consistent with the results from immunohistochemistry in Fig. 2A, AGT1 immunofluorescence signals dominantly appeared on the renal tubule in the outer stripe of the outer medulla and medullary ray (Fig. 2G). In contrast, rBAT immunoreactivity was extended to the superficial cortex although that in the outer stripe of the outer medulla and medullary ray was the strongest (Fig. 2H), as reported previously (11, 12). The immunoreactivity of AGT1 was completely merged with that of rBAT at the apical membrane of the proximal tubules (Fig. 2 J–L).

Functional Characterization of AGT1 in Proteoliposome.

Functional properties of the AGT1-rBAT heterodimer were examined by reconstitution into proteoliposomes. AGT1 and rBAT were stably coexpressed in HEK293 cells and purified as a complex (Fig. S7). Most of the purified AGT1 formed heterodimers with rBAT. b^0,+AT was not detected in any fractions, including the elution fraction (Fig. S7). The AGT1-rBAT proteoliposomes preloaded with or without aspartate were incubated with [¹⁴C]cystine (Fig. 3A). Although the control liposomes did not accumulate radioactivity, time-dependent increase of [¹⁴C]cystine uptake was observed in AGT1-rBAT proteoliposomes without aspartate inside (Fig. 3A), indicating that the AGT1-rBAT heterodimer mediated downhill cystine influx. Such a facilitative transport of cystine displayed a hyperbolic dependence on cystine concentration, fitted to the Michaelis–Menten equation with a K_m value of 67.6 μM. In contrast, the proteoliposomes preloaded with aspartate exhibited a higher initial rate of [¹⁴C]cystine influx. The accelerated influx was transient, and the uptake peaked at around 10 min, after which the accumulated radioactivity declined gradually (Fig. 3A). This overshoot is typical of active transport driven by cotransport or countertransport in proteoliposome (27). It suggests that AGT1-rBAT heterodimer mediates countertransport when counter substrates exist. The countertransport mediated by AGT1-rBAT was confirmed by efflux experiments in which [¹⁴C]cystine was preloaded into proteoliposomes. The efflux of [¹⁴C]cystine was estimated in the presence or absence of external aspartate by measuring the radioactivity remaining in the proteoliposomes. As shown in Fig. 3B, the efflux of preloaded cystine was accelerated by external aspartate in a concentration-dependent manner.

Fig. S7.

Purification of the AGT1-rBAT heterodimeric complex. The AGT1-rBAT heterodimer was purified from Flp-In T-REx 293 cells stably expressing 3xFLAG-AGT1 with inducible rBAT (293TR-mrBAT/3xFLAG-mAGT1 cells) using anti-FLAG M2 affinity gel as described in SI Materials and Methods. AGT1 (arrowheads) was detected by anti-FLAG antibody. Western blot of rBAT in the presence of 100 mM DTT (+DTT) revealed two bands of rBAT in the membrane fraction: a lower band (bar) and a faint upper band (arrow). Presumably, the lower band is nonglycosylated or partially glycosylated rBAT, and the upper one is fully glycosylated rBAT. Only glycosylated rBAT (arrow) was purified as the heterodimeric complex although nonglycosylated or partially glycosylated rBAT (bar) was dominantly detected in the membrane fraction and flow-through fraction, suggesting that only glycosylated rBAT forms a heterodimeric complex with AGT1. No b^0,+AT was detected in the elution fraction. Silver staining after SDS/PAGE in the absence of 100 mM DTT (−DTT) showed that AGT1 was purified as the heterodimeric complex (asterisk). Some of the AGT1-rBAT heterodimers appeared as oligomeric complexes (double asterisk) in the SDS/PAGE.

Fig. 3.

Functional analysis of the AGT1-rBAT heterodimer in proteoliposomes. (A) Time course of downhill cystine influx and cystine countertransport. The proteoliposomes reconstituted with the AGT1-rBAT heterodimer and preloaded with (filled circle) or without (open circle) 4 mM aspartate, as well as control liposomes preloaded with (filled square) or without (open square) 4 mM aspartate, were incubated in 100 µM [¹⁴C]cystine. CssC, cystine; *CssC, [¹⁴C]cystine. (B) The effect of external aspartate on cystine efflux. The efflux of [¹⁴C]cystine (100 µM) was evaluated in the absence (open diamond) or presence of external nonradioactive aspartate (gray diamond, 100 µM; filled diamond, 500 µM). The radioactivity remaining in the proteoliposomes was shown as percent of loaded radioactivity. (C) Ion dependence of the transport. The uptake of [¹⁴C]cystine and [¹⁴C]aspartate (100 µM) was measured for 10 min in sodium uptake buffer (+) or potassium uptake buffer (−). The proteoliposomes were preloaded with 4 mM aspartate. *Asp, [¹⁴C]aspartate. Statistical difference was determined using the Student’s unpaired t test. Differences were considered significant at P < 0.05. (D) Substrate selectivity of AGT1. Proteoliposomes loaded with 4 mM aspartate were incubated in sodium uptake buffer containing ¹⁴C-labeled amino acids (100 µM) for 10 min. *Amino acid, radiolabeled l-amino acid. (E) The effects of acidic amino acid analogs on the transport. The uptake of 20 µM [¹⁴C]aspartate into the proteoliposome loaded with 4 mM aspartate was measured for 5 min in the presence or absence (−) of indicated compounds. The test compounds (Left) were at 5 mM except threo-β-hydroxyaspartate (THA) (2 mM). The compounds in the Right were at 0.5 mM. DHK, dihydrokainate; PDC, l-trans-pyrrolidine-2,4-dicarboxylate; SOS, l-serine-O-sulfate. The uptake values shown in C–E are the influx into proteoliposomes minus the influx into control liposomes. The potassium uptake buffer was used for A, B, and E. All data in Fig. 3, except B, represent mean ± SE; n = 3–4.

The ion dependence of the transport was examined for the countertransport mode in the condition that aspartate and K⁺ (150 mM) were loaded into the AGT1-rBAT proteoliposomes. The proteoliposomes were incubated with Na⁺ (150 mM) or K⁺ (150 mM). As shown in Fig. 3C, the influx of both cystine and aspartate was not significantly dependent on Na⁺.

The substrate selectivity of the AGT1-rBAT heterodimer was examined for the countertransport mode on the proteoliposomes preloaded with aspartate. As shown in Fig. 3D, a high level of influx was observed for aspartate, glutamate, and cystine whereas a lower level of influx was detected for glutamine, leucine, methionine, phenylalanine, valine, tyrosine, and tryptophan. Serine, cysteine, and basic amino acids, such as lysine and arginine, were not transported. To differentiate its substrate recognition from that of other acidic amino acid transporters, the effects of acidic amino acid analogs (see structures in Fig. S8) were investigated on [¹⁴C]aspartate influx. As shown in Fig. 3E, threo-β-hydroxyaspartate and cysteine sulfinate inhibited [¹⁴C]aspartate uptake as well as aspartate and glutamate did whereas the inhibition by serine-O-sulfate, cysteate, cysteine, and homocysteine was less than that by aspartate and glutamate. In contrast, α-aminoadipate, homocysteate, trans-pyrrolidine-2,4-dicarboxylate, and dihydrokainate did not inhibit the uptake. Cystine exhibited a similar level of inhibition compared with aspartate and glutamate.

Fig. S8.

Structures of test compounds used for Fig. 3E.

Colocalization of AGT1 with Acidic Amino Acid Transporter EAAC1 (SLC1A1).

In the mouse kidney, EAAC1 was detected by immunofluorescence microscopy in the cortex and outer stripe of the outer medulla, where the outer stripe of the outer medulla and medullary ray exhibited stronger fluorescence than the superficial cortex (Fig. 4A). EAAC1 immunoreactivity was detected at the apical membranes of tubules by SIM (Fig. 4B), similar to AGT1 (Fig. 4C). AGT1 immunoreactivity was colocalized with the staining of EAAC1 (Fig. 4 D and E) whereas the tubules expressing EAAC1 but not AGT1 also exist due to the wider distribution of EAAC1 in proximal tubules (Fig. 4E). To further investigate the spatial relation of AGT1 and EAAC1, the distance between the two molecules was estimated with an in situ proximity ligation assay. In this assay, the fluorescence signal is detected only when two target proteins are in close proximity to each other: The theoretical maximum distance between two target proteins is 30–40 nm (28). Kidney sections reacted with both the anti-AGT1 antibody and anti-EAAC1 antibody showed positive red fluorescent signals, suggesting that AGT1 and EAAC1 are closely placed with each other (Fig. 5A) whereas no signal was detected in the kidney sections treated with only either the anti-AGT1 antibody or anti-EAAC1 antibody (Fig. 5 B and C). The treatment with the anti-AGT1 antibody and anti-rBAT antibody produced positive signals as well as the combination of the anti-b^0,+AT antibody and anti-rBAT antibody whereas positive signals were not detected when the anti-AGT1 antibody was combined with an antibody for CD13/aminopeptidase N, localized at the apical membrane of the proximal tubules (Fig. 5 D–F).

Fig. 4.

Localization of AGT1 with acidic amino acid transporter EAAC1 in the male mouse kidney. (A) Localization of EAAC1 in immunofluorescence microscopy. EAAC1 immunoreactivity was detected strongly in the outer stripe of the outer medulla (“OSOM”) with the medullary ray (asterisk). Weaker staining was also detected in the cortex (“C”). (Scale bar: 500 µm.) (B–E) Coimmunofluorescence staining of EAAC1 and AGT1 on the apical membrane of renal tubules in SIM analysis. EAAC1 and AGT1 signals are shown in green and cyan, respectively. A merged image of B and C is shown in D. E was merged from other independent pictures. AGT1 was colocalized with EAAC1 (arrow) whereas tubules expressing EAAC1 but not AGT1 (arrow head) also exist (E). (Scale bars: D and E, 20 µm.) The anti-AGT1(G) antibody was used for immunofluorescence experiments.

Fig. 5.

Colocalization of AGT1 and EAAC1 determined by proximity ligation assay. An in situ proximity ligation assay (PLA) was performed on male mouse kidney sections. In the confocal images of the mouse kidney sections, positive PLA fluorescence signals are shown in red (arrowheads). Nuclei were stained with DAPI (blue). The positive signals were detected on the brush border region of proximal tubules treated with both anti-AGT1(G) and anti-EAAC1 antibodies (A). No signals were seen in the sections treated with anti-AGT1 antibody alone or anti-EAAC1 antibody alone (B and C). Positive PLA fluorescence signals were also obtained between b^0,+AT and rBAT (D) and between AGT1 and rBAT (E). No positive signal was seen in the sections treated with anti-AGT1 and anti-CD13 antibodies (F). (Scale bars: 10 μm.)

SI Materials and Methods

Materials.

All chemicals used in this study were of analytical grade. General chemicals were purchased from Wako. Amino acids and their related compounds were purchased from Sigma-Aldrich. Radioactive amino acids were obtained from Moravek Biochemicals and Radiochemicals, MP Biomedicals, and PerkinElmer. l-isomers of amino acids were used throughout the study.

Animal Experiments.

All animal experiments were designed and done according to the highest scientific, humane, and ethical principles and were performed under the regulations of the Osaka University Graduate School of Medicine and the Institutional Animal Care and Use Committee at the Institut d’Investigació Biomèdica de Bellvitge (IDIBELL) in a facility accredited by AALAC:B-9900010/3866. ddY mice were used to identify the partner of AGT1. The characterization of the AGT1-rBAT complex was done by using C57BL/6J mice. The mutant mice bearing the Slc3a1 missense mutation (Asp140Gly) (22) in mixed C57BL/6J-C3H genetic background were used at 3 mo of age to examine the association of AGT1 with rBAT.

Antibodies and Antibody Beads.

Antigens for AGT1 (Gene ID: 74087) were prepared as follows. The cDNA fragment corresponding to residues 455–478 of AGT1 was cloned between EcoRI and SalI sites of pET49b(+) (Merck KGaA) to obtain a GST-AGT1 C-terminal fusion protein and also was cloned between EcoRI and SalI sites of pMAL-c2X (New England BioLabs) to obtain a MBP-AGT1 C-terminal fusion protein. The plasmids were transformed into Escherichia coli strain BL21(DE3). The transformed E. coli were grown in LB medium at 28 °C for the GST fusion protein and 37 °C for the MBP fusion protein. When OD₆₀₀ was 0.5, protein expression was induced by 0.1 and 0.2 mM isopropyl β-d-thiogalactopyranoside for GST fusion protein and MBP fusion protein, respectively. E. coli were further incubated for 3 h. The bacteria were collected by centrifuge and washed with PBS [137 mM NaCl, 2.7 mM KCl, 10 mM Na₂HPO₄, 1.8 mM KH₂PO₄ (pH 7.4)] and then lysed by freeze-thaw and sonication in PBS with protease inhibitors (1 µg/mL aprotinin, 1 µg/mL leupeptin, 1 µg/mL pepstatin, and 1 mM phenylmethylsulfonyl fluoride). The lysate was incubated in the presence of 0.1% Triton X-100 for 30 min at 4 °C and centrifuged at 1,000 × g for 10 min to remove debris. The supernatants were further ultracentrifuged at 20,000 × g for 10 min at 4 °C. The supernatants were incubated with Glutathione Sepharose 4B beads (GE Healthcare) or Amylose Resin (New England BioLabs) overnight at 4 °C. Beads or resin was applied to a chromatography column (Poly-Prep Chromatography Columns; Bio-Rad). Glutathione Sepharose 4B beads were washed with PBS and eluted with GST elution buffer [50 mM Tris⋅HCl, 10 mM glutathione (pH 8.0)]. Amylose resin was washed with MBP washing buffer [20 mM Tris⋅HCl, 200 mM NaCl, 10 mM β-mercaptoethanol, 1 mM EDTA (pH7.4)] and eluted with MBP elution buffer [20 mM Tris⋅HCl, 200 mM NaCl, 10 mM β-mercaptoethanol, 1 mM EDTA, 10 mM Maltose (pH 7.4)]. The eluates were subjected to an Amicon Ultra-4 10 K NMWL or 30 K NMWL (Merck Millipore) for buffer exchange in PBS and concentration. The purities of final products (GST-AGT1-C and MBP-AGT1-C) were assessed with SDS/PAGE followed by Coomassie brilliant blue staining, and products were stored at −80 °C until use.

Antibodies were generated as described previously with minor modifications (42). Briefly, the recombinant fusion proteins, expressed as described above, were used as antigens. A female New Zealand White rabbit was immunized with 200 μg of each antigen for the first shot. After 4 wk, four boost shots with 100 μg of antigen each were followed at a 2-wk interval. The animals were killed for total blood collections. Purified IgG fractions were obtained from the sera by using Protein G Sepharose (Sigma-Aldrich). Antibodies against GST-AGT1-C and MBP-AGT1-C were named anti-AGT1(G) and anti-AGT1(M), respectively.

For generation of antibody beads, the purified antibodies or Normal rabbit IgG (Santa Cruz Biotechnology) were incubated with Protein G Sepharose beads (Sigma-Aldrich) in the presence of 1 mg/mL BSA overnight at 4 °C. IgG and the beads were cross-linked with 25 mM dimethyl pimelimidate dihydrochloride (Thermo Scientific Pierce Protein Research Products) and 0.1 M triethanolamine in PBS for 30 min at room temperature and washed by 0.2 M triethanolamine three times. The cross-linking reaction was quenched by 50 mM monoethanolamine in PBS, and the beads were suspended in PBS. The beads were further washed with 1 M glycine-HCl (pH 2.7) and PBS and then stored at 4 °C until use.

Western Blot Analysis.

Crude membrane fractions or renal brush-border membrane vesicles (BBMVs) were used for the characterization and identification of the AGT1-rBAT complex. To prepare the crude membrane fractions, the frozen kidney tissue was homogenized in the buffer [50 mM Tris⋅HCl (pH 7.6), 250 mM sucrose, 1 mM EDTA, 6 mM EGTA, 20 mM β-glycerophosphate, 20 mM NaF, 1 μg/mL aprotinin, 1 μg/mL leupeptin, 1 μg/mL pepstatin, and 1 mM phenylmethylsulfonyl fluoride] at the volume of 25 mL for 1 mg of tissue. Tissue debris was separated by centrifugation at 1,000 × g for 5 min at 4 °C. The supernatants were centrifuged at 10,000 × g for 5 min, and the supernatants were ultracentrifuged at 270,000 × g for 60 min at 4 °C. The sediment was suspended in 50 mM Tris⋅HCl (pH 7.6), 150 mM NaCl, 5 mM EDTA, 6 mM EGTA, 10% (vol/vol) glycerol, 20 mM β-Glycerophosphate, 20 mM NaF, and Complete EDTA-free Protease inhibitor mixture (Roche). The membrane fractions were subjected to a BCA protein assay and kept at −80 °C until use. The membrane fractions were solubilized in the presence of 1% Fos-Choline-12 (Affymetrix) and 5 mM N-ethylmaleimide on ice for 1 h and subjected to SDS/PAGE in the presence or absence of DTT. For the Western blot of rBAT, renal BBMVs were prepared from a whole kidney as described previously (43).

Western blot was performed as described in Khunweeraphong et al. (20). Briefly, the separated proteins were transferred electrophoretically to a Hybond-P PVDF transfer membrane (GE Healthcare). The membrane was preblocked in blocking solution and then incubated overnight at 4 °C with each primary antibody as indicated. Horseradish peroxidase-conjugated anti-rabbit IgG (Jackson ImmunoResearch Laboratories) was used as secondary antibody. The signal was developed by using Immobilon Western Chemiluminescent HRP Substrate (Merck Millipore) and visualized under the LAS-4000 mini luminescent image analyzer (Fujifilm). The following antibodies were used as primary antibodies: anti-rBAT antibody (12), anti-rBAT antibody T-20 (Santa Cruz Biotechnology), anti-b^0,+AT antibody (44), and anti-AGT1 antibodies generated in this study.

Immunoprecipitation and Mass Spectrometry.

The crude membrane fraction was prepared from a whole kidney or whole brain of male mice. The mice were anesthetized and perfused with cold PBS. The crude membrane fractions were prepared as described above. The membrane fractions were solubilized in 1% Fos-Choline-12 (Affymetrix) and 5 mM N-ethylmaleimide on ice for 1 h. The solubilized membranes were ultracentrifuged at 438,000 × g for 15 min at 4 °C. The supernatant was diluted 10 times with 20 mM Tris⋅HCl (pH 7.6), 150 mM NaCl, 5 mM EDTA, 6 mM EGTA, 20 mM NaF, and Complete EDTA-free Protease inhibitor mixture and then incubated overnight at 4 °C with the antibody beads for AGT1 immunoprecipitation. For the immunoprecipitation of rBAT, the incubation with anti-rBAT antibody (Santa Cruz Biotechnology) was followed by protein G beads for an additional 4 h at 4 °C. The beads were washed with 0.05% Fos-Choline-12 in PBS and eluted with 1 M glycine-HCl (pH 3.0), 0.1% Fos-Choline-12. The eluate was adjusted to pH 7–8 by 1 M Tris⋅HCl (pH 9.0) immediately and subjected to SDS/PAGE.

For mass spectrometric analysis, precipitated proteins were eluted with 8 M urea in 5% (wt/vol) sodium deoxycholate for 20 min. The eluted fraction was alkylated with 55 mM iodoacetamide and then digested with sequencing grade modified trypsin (Roche). Sodium deoxycholate was removed by means of the phase transfer surfactant (PTS) method (45). The tryptic digests were desalted with a C18 Empore disk membrane (3M), concentrated using the Speed-Vac centrifuge, and reconstituted in 0.1% formic acid and 3% (vol/vol) acetonitrile. LC–MS/MS analysis was performed by a Paradigm MS4 nanoHPLC system (Michrom BioResources, Inc.) coupled to an LTQ linear ion trap mass spectrometer (Thermo Electron Corp.) with a nanoelectrospray ionization source (AMR Inc.) as described (46). Peptides and proteins were identified by Mascot v2.2 (Matrix Science) with a maximum tolerance of 1.2 Da for MS data, 0.5 Da for MS/MS data, and strict trypsin specificity allowing for up to one missed cleavage.

Immunofluorescence Microscopy of Cultured Cells.

To express tagged AGT1 and rBAT to HEK293 cells, cDNAs for mouse AGT1 and mouse rBAT were subcloned into mammalian expression vectors. The coding sequence of mouse AGT1 was amplified from mouse AGT1 cDNA in pcDNA3.1 (pcDNA3.1-mAGT1) (17) by using primers 5′-TTTGCGGCCGCAATGGCAATGGATTCAAAGAAG-3′ and 5′-GGTGGATCCACACAGTATGGCTACAAAGGC-3′, and then the fragment was digested by NotI and BamHI and inserted into p3xFLAG-Myc-CMV24 at NotI and BamHI sites to obtain pCMV24-3xFLAG-mAGT1. Mouse rBAT cDNA was amplified by a primer pair, 5′-CACCATGGATGAGGACAAAGGCAAG-3′ and 5′-CTAACACGAGCTATAGAGGATG-3′, using mouse kidney first strand cDNA as a template, and subcloned into pENTR/d-TOPO (Invitrogen) by TOPO-TA cloning to obtain pENTR-mrBAT. pcDNA-40-mrBAT was generated from pENTR-mrBAT and pcDNA-DEST40 by the LR recombination reaction of Gateway technology (Invitrogen).

HEK293 cells were maintained in DMEM supplemented with heat-inactivated 10% (vol/vol) FBS, penicillin, and streptomycin. Cells were cultured at 37 °C in 5% (vol/vol) CO₂. For the immunofluorescence microscopy, cells were seeded on coverslips coated with poly-d-lysine and then subjected to transfection of mouse pcDNA3.1(+)-m4F2hc (17), pCMV24-3xFLAG-mAGT1, and pcDNA-40-mrBAT using Lipofectamine 2000 (Invitrogen) for the transient expression of 4F2hc, 3xFLAG-AGT1, and rBAT, respectively.

Two days after transfection, cells were fixed with 4% (wt/vol) paraformaldehyde in PBS for 30 min. After permeabilization in PBS containing 0.1% Triton X-100 and 1% BSA, cells were incubated with the primary antibodies for 3 h, followed by incubation with secondary antibodies for 1 h. All procedures were conducted at room temperature. Coverslips were mounted using fluorescence mounting medium (Dako) and subjected to image acquisition by an LSM510META confocal microscope (Carl Zeiss) equipped with a 63× PlanApo oil-immersion objective lens (NA1.4). DAPI (Invitrogen) was used for the staining of nuclei. Primary antibodies used were as follows: goat anti-CD98 antibody against 4F2hc (N-20; Santa Cruz Biotechnology), goat anti-rBAT antibody (T-20; Santa Cruz Biotechnology), and mouse anti-Flag antibody (M2; Sigma-Aldrich). The following secondary antibodies were used: Alexa Fluor⁴⁸⁸-conjugated donkey anti-mouse IgG (Invitrogen) and Alexa Fluor⁵⁶⁸-conjugated donkey anti-goat IgG (Invitrogen).

Immunohistochemistry and Immunofluorescence Microscopy of Mouse Kidney Sections.

Eight-week-old C57BL/6J mice were deeply anesthetized and fixed by anterograde perfusion via the aorta with 4% (wt/vol) paraformaldehyde in 0.1 M sodium phosphate buffer (NaPi) (pH 7.4). Kidneys are removed, postfixed in the same fixative for 3 h, and cryoprotected in 30% (wt/vol) sucrose. Parasagittal kidney frozen sections were cut at 10-µm thickness in a cryostat microtome (Leica) and mounted on MAS-coated glass slides (Matsunami).

The sections were washed by PBS and treated with 0.3% H₂O₂ in methanol to block endogenous peroxidase activity. Sections were placed in 0.1 M citrate buffer (pH 6.2) and autoclaved at 121 °C for 5 min for antigen retrieval and then incubated overnight at 4 °C with the anti-AGT1(G) antibody, anti-rBAT antibody (T-20; Santa Cruz Biotechnology), and/or anti-EAAC1 antibody (C-20; Santa Cruz Biotechnology) diluted in PBS containing 1% BSA. For light microscopy, sections were incubated with biotinylated goat anti-rabbit secondary antibody (for AGT1) or donkey anti-goat secondary antibody (for rBAT). Biotinylated secondary antibodies were purchased from Vector Laboratories. Staining was visualized by using a Vectastain ABC Standard kit (Vector Laboratories) and diaminobenzidine according to the manufacturer’s protocol. The sections were counterstained with Mayer’s hematoxylin and/or eosin, dehydrated in an ascending ethanol series [70, 80, 90, and 100% (vol/vol)], and xylene. Images were acquired using a BZ-9000 microscope (Keyence) and processed by a BZ analyzer (Keyence). For fluorescence microscopy, sections were incubated with secondary antibodies, and images were acquired using an AX-80 (Olympus). Lectin phytohemagglutinin-L (PHA-L) conjugated with Alexa 488 (Invitrogen) was used as an apical membrane staining marker.

Superresolution Structured Illumination Microscopy.

Superresolution structured illumination microscopy (SR-SIM) images of AGT1 and EAAC1 distribution in the mouse kidney were obtained via a 63×/1.4 oil-immersion objective using an Elyra S1 microscope (Carl Zeiss). Images were reconstructed using ZEN software (Carl Zeiss MicroImaging) based on the structured illumination algorithm developed by Heintzmann and Cremer (47).

Establishment of Flp-In T-REx 293 Cells Stably Expressing rBAT and 3xFLAG-AGT1 (293TR-mrBAT/3xFLAG-mAGT1 cells).

To stably express rBAT, the NotI and ApaI fragment of pcDNA-40-mrBAT was excised and subcloned into NotI and ApaI sites of pcDNA5/FRT/TO to obtain pcDNA5/FRT/TO-mrBAT. For AGT1, the 3xFLAG-mAGT1 cDNA was amplified from pCMV24-3xFLAG-mAGT1 by using primers 5′-GAAGTCGACATGGACTACAAAGACC-3′ and 5′-TATGATATCTC AACTTTCTTCATGTATG-3′ and was inserted into pEBMulti-Neo at SalI and EcoRV to obtain pEBMulti-Neo-3xFLAG-mAGT1.

Flp-In 293 cells (Invitrogen) were transfected with pcDNA6/TR (Invitrogen) by using Lipofectamine 2000 following the manufacturer’s protocol with slight modifications. The transfected cells were grown in DMEM containing 10% (vol/vol) FBS, penicillin, streptomycin, and 5 mg/L blasticidin. Positive colonies resistant to blasticidin were screened for the expression of Tet Repressor. A single clone stably expressing Tet Repressor was selected and named Flp-In T-REx 293 cells. Flp-In T-REx 293 cells were transfected with pOG44 (Invitrogen) and pCDNA5/FRT/TO-mrBAT by Lipofectamine 2000 and grown in DMEM containing 10% (vol/vol) FBS, penicillin, streptomycin, 5 mg/L blasticidin, and 150 mg/L hygromycin B. Positive clones resistant to blasticidin and hygromycin B were selected and named 293TR-mrBAT cells. The cells were then transfected with pEBMulti-Neo-3xFLAG-mAGT1 and grown in DMEM containing 10% (vol/vol) FBS, penicillin, streptomycin, 5 mg/L blasticidin, 150 mg/L hygromycin B, and 900 mg/L G418 sulfate. Positive clones resistant to blasticidin, hygromycin B, and G418 were selected to obtain Flp-In T-REx 293 cells stably expressing 3xFLAG-mAGT1 with inducible mrBAT named 293TR-mrBAT/3xFLAG-mAGT1 cells.

293TR-mrBAT/3xFLAG-mAGT1 cells were routinely maintained in DMEM containing 10% (vol/vol) FBS, penicillin, streptomycin, 5 mg/L blasticidin, 150 mg/L hygromycin B, and 900 mg/L G418 sulfate. The cells were cultured at 37 °C with 5% (vol/vol) CO₂ and humidity. To express rBAT, the cells were grown in 1 mg/L doxycycline hyclate-containing medium (Tet-ON) for 2 d.

Purification of the rBAT-AGT1 Complex and Its Reconstitution into Proteoliposomes.

To purify 1 μg of the complex of rBAT and 3xFLAG-AGT1, 2.5 × 10⁷ 293TR-mrBAT/3xFLAG-mAGT1 cells were used. Cell pellets were suspended in homogenization buffer containing 50 mM potassium phosphate buffer (KPi), pH 7.0, 150 mM NaCl, 10% (vol/vol) glycerol, and Complete EDTA-free Protease Inhibitor Mixture (Roche). The cells were sonicated six times for 5 s each time by a Handy Sonic UR-20P (Tomy Seiko) and centrifuged at 1,000 × g for 10 min. The supernatant was collected and centrifuged at 264,900 × g for 90 min. The crude membrane faction was pelleted and washed with 4 M urea on ice for 30 min. The urea-washed crude membrane fraction was sedimented by centrifuging at 264,900 × g for 3 h. Membrane proteins were extracted from the urea-washed crude membrane fraction with 2% (wt/vol) n-dodecyl-β-d-maltoside (DDM) (Dojindo) in homogenization buffer. The complex of rBAT and 3xFLAG-AGT1 was purified by using anti-FLAG M2 affinity gel (Sigma-Aldrich) following the manufacturer’s protocol with minor modifications. Briefly, the extracted membrane proteins were mixed with anti-FLAG M2 affinity gel overnight at 4 °C. The gel was washed with homogenization buffer containing 0.02% DDM. The purified complex of rBAT and 3xFLAG-AGT1 was eluted twice with 200 µg/mL 3xFLAG peptide in the washing buffer. The elution fractions were analyzed by Western blot and silver staining.

The purified complex of rBAT and 3xFLAG-AGT1 was reconstituted as described (41) with modification. Briefly, l-α-phosphatidylcholine type II-S from soybean (Sigma-Aldrich) and brain total lipid extract (Avanti) were purified by the method of Bligh and Dyer (48) and evaporated by a rotary evaporator. The purified lipid was resuspended in 100 mM MOPS-Tris, pH 7.0, and stored at −80 °C until use. The lipid mixture was prepared by 5:1 ratio of l-α-phosphatidylcholine type II-S to brain total lipid extract and then dissolved with 2.5% (wt/vol) n-octyl-β-d-glucoside (OG) (Nacalai Tesque). The purified complex of rBAT and 3xFLAG-AGT1 was diluted with 1.25% (wt/vol) OG and then mixed with the lipid mixture. The ratio of the protein complex to the lipid mixture was 1:500. The mixture was incubated on ice for 20 min. Proteoliposomes were reconstituted by the dilution method (41) with reconstitution buffer containing 50 mM KPi, pH 7.0, and 100 mM KCl with or without 4 mM l-aspartic acid. Mannitol (4 mM) was preloaded into proteoliposomes instead of aspartate when aspartate was not preloaded. The diluted solution was further incubated twice with Biobeads (Bio-Rad) for 2 h at 4 °C. After removing Biobeads, the proteoliposomes were sedimented by centrifugation at 264,900 × g for 3 h. The proteoliposomes were sonicated in the reconstitution buffer, frozen in liquid nitrogen, and stored at −80 °C until use.

Transport Assays on Proteoliposomes Containing rBAT and 3xFLAG-AGT1.

The proteoliposomes were thawed and sonicated and then settled at 25 °C for 30 min before the start of experiments. To initiate the transport assay, the proteoliposomes containing 5 µg of the purified proteins with reconstitution buffer were diluted 160-fold into potassium uptake buffer (50 mM KPi, pH 7.0, 100 mM KCl, and 1 mM MgCl₂) or sodium uptake buffer (50 mM NaPi, pH 7.0, 100 mM NaCl and 1 mM MgCl₂) containing l-[¹⁴C]cystine or other radioactive amino acids as indicated. The final concentration of K⁺ after dilution of reconstitution buffer containing proteoliposomes with sodium uptake buffer was 0.93 mM. For the countertransport assay (counterflow assay), the aspartate-preloaded proteoliposomes were used. For the efflux assay, proteoliposomes were preloaded with 100 µM l-[¹⁴C]cystine, and then the proteoliposomes were diluted 400-fold into potassium uptake buffer in the presence or absence of cold amino acids. The reaction was incubated at 25 °C for 10 min unless otherwise indicated and then terminated by adding 2.5 mL of ice-cold potassium uptake buffer. The mixture was filtrated immediately through a 0.22-µm cellulose filter (GSTF; Millipore), and the filter was washed once with the same cold buffer. The filter was soaked in 3 mL of Emulsifier safe mixture (PerkinElmer), and the radioactivity was counted by a β-scintillation counter (LSC-7200; Hitachi Aloka Medical).

In Situ Proximity Ligation Assay.

To detect the interaction between AGT1 and EAAC1, the Duolink in situ Proximity Ligation Assay (PLA) (Sigma-Aldrich) was used according to the manufacturer’s protocol. The kidney sections were incubated with primary antibodies overnight at 4 °C. The primary antibodies against AGT1, EAAC1, b^0,+AT, rBAT, and CD13 (C-17; Santa Cruz Biotechnology) were used to detect the proximities of the proteins. The secondary antibodies with attached PLA probes were supplied in the kit. The fluorescence signal of PLA was observed by confocal microscopy at 63× magnification.

Methods for Digoxigenin-Labeled in Situ Hybridization.

By using the primer pair for AGT1 in Fig. S2, Digoxigenin (DIG)-labeled cRNA probes for mouse AGT1 were prepared using T7 or SP6 RNA polymerase (Promega) and a DIG RNA Labeling Kit (Roche). The sections were briefly dried, fixed for 20 min in 4% (wt/vol) paraformaldehyde in PBS, washed twice with PBS for 10 min, and treated with 10 μg/mL proteinase K in PBS for 5 min. After washing in PBS and fixing with 4% (wt/vol) paraformaldehyde for 5 min, the sections were incubated with 0.25% acetic anhydride in 0.2 M triethanolamine for 10 min. The sections were incubated in prehybridization buffer [50% (vol/vol) deionized formamide, 4× SSC, 1× Denhardt’s solution] for 1 h at room temperature. For hybridization, DIG-labeled probes diluted at the appropriate concentration in hybridization buffer [50% (vol/vol) deionized formamide, 20 mM Tris⋅HCl (pH 8.0), 300 mM NaCl, 2.5 mM EDTA, 10% (wt/vol) dextran sulfate, 1× Denhardt’s solution, and 400 μg/mL yeast transfer RNA] were denatured, quenched on ice, and incubated on the sections in a humid chamber overnight at 50 °C. The hybridized sections were washed in 2× SSC and 50% (vol/vol) formamide, 2× SSC and 0.2× SSC washing buffer each time for 1 h at 50 °C, followed by DIG buffer (100 mM Tris⋅HCl and 150 mM NaCl, pH 7.5). After the sections were blocked with blocking reagent [0.5% skim milk, 0.05% Tween-20, 10% (vol/vol) goat serum], they were exposed to an anti-Digoxigenin-AP Fab fragment antibody (1:2,000; Roche Diagnostics) overnight at 4 °C, and then rinsed three times for 5 min in PBS containing 0.05% Tween 20. For the detection of hybridized DIG-labeled cRNA probes, the sections were rinsed briefly in alkaline phosphatase (AP) buffer (100 mM Tris, 100 mM NaCl, and 50 mM MgCl₂, pH 9.5) and incubated in developing buffer (AP buffer containing NBT/BCIP stock solution, 1:50; Roche Diagnostics). To stop the color development reaction, the sections were washed in PBS and mounted. All procedures were performed at room temperature unless otherwise indicated.

Discussion

On the cystine reabsorption in the renal proximal tubule, a long-standing issue that has not been solved since the discovery of rBAT and b^0,+AT is the discrepancy regarding the distribution of rBAT and b^0,+AT along the renal proximal tubules. Based on this discrepancy, it has been postulated that rBAT has an unknown partner other than b^0,+AT. It should be the main rBAT-partner in the S3 segment of the proximal tubules where the expression of b^0,+AT is substantially low whereas that of rBAT is the highest. In an attempt to reveal an unidentified heavy chain of AGT1 in the present study, we have found that AGT1 is the long-awaited rBAT-partner in the S3 segment.

By means of immunoprecipitation using anti-AGT1 antibodies followed by mass spectrometry, rBAT was identified as a binding partner of AGT1. rBAT associates with AGT1 via a disulfide bond (Fig. 1) and translocates AGT1 to the plasma membrane, consistent with the proposed role of heavy chains of HATs. In the kidney, AGT1 expression is largely restricted to the outer stripe of the outer medulla and medullary ray where it is colocalized with the strongest staining of rBAT. It indicates that AGT1 is present in the S3 segment of the proximal tubules because the strong rBAT staining is a marker for the S3 segment (11). The presence of the AGT1-rBAT heterodimer in the proximal tubules was supported by an in situ proximity ligation assay (Fig. 5D) and further was confirmed by the observation that AGT1 disappeared from the BBMV in mutant mice lacking rBAT expression (Fig. 1C). Therefore, we concluded that AGT1 is the transporter corresponding to the missing partner of rBAT in the S3 segment of the proximal tubules. Additionally, we found a remarkable biological-sex difference in the expression of AGT1 in the kidney. Although we could not find perfect palindrome estrogen (ERE) and androgen (ARE) response elements associated with the AGT1 gene (Fig. S2B), some of the imperfect palindrome AREs and/or half-sites of ARE and ERE could be involved in male-specific AGT1 expression. The functional significance of imperfect AREs and half-sites of ARE or ERE is, however, still not established (29).

In this study, the localization of AGT1 was examined by two newly generated antibodies raised against the C-terminal intracellular domain of AGT1 produced by expressing it in Escherichia coli. In the immunohistochemistry of mouse kidney cryosections, basically identical results were obtained by using these two antibodies on the localization of AGT1 (Fig. 2 and Fig. S5), which was further confirmed by in situ hybridization (Fig. S6). Using these antibodies, AGT1 was localized to the apical membrane of the S3 segment (Fig. 2). In a previous study, an antiserum raised against a synthetic oligo peptide corresponding to the predicted amino acid sequence of AGT1 stained the basolateral membrane of proximal straight tubules and distal convoluted tubules in mouse kidney paraffin sections (17). Although the staining was eliminated in the absorption experiments using the antigen peptide, it is possible that the antibody recognized epitopes in tissue sections other than AGT1.

For functional analyses using Xenopus oocytes in the previous study, we generated fusion proteins in which AGT1 was connected with 4F2hc or rBAT to ensure plasma membrane sorting (17). Both fusion proteins expressed Na⁺-independent transport for aspartate and glutamate although their uptake levels were low. Because such fusion of proteins may affect the function, we purified an AGT1-rBAT heterodimer (Fig. S7) and reconstituted it into proteoliposomes. This proteoliposome reconstitution allowed us to examine the function of the heterodimer, avoiding distortion by protein fusions and excluding possible effects from other proteins, amino acids, and inorganic ions in the cells.

By proteoliposome reconstitution, we have revealed that the AGT1-rBAT heterodimer transports cystine as well as aspartate and glutamate. In the previous study, cystine transport activity was not detected for the fusion proteins in Xenopus oocytes although cystine weakly inhibited aspartate uptake (17). In addition, the AGT1-rBAT heterodimer also mediated a lower level of transport of some neutral amino acids, which was not detected in the fusion proteins. The profile of inhibition of [¹⁴C]aspartate uptake by acidic amino acid analogs for the AGT1-rBAT heterodimer in proteoliposome shown in Fig. 3E was similar to that for the AGT1-4F2hc fusion protein (17). AGT1 would recognize cystine as an anionic amino acid similar to other HAT member xCT/SLC7A11 that transports glutamate and cystine by recognizing cystine as an anionic amino acid. AGT1 is, however, still distinct from xCT in that AGT1 transports aspartate with a short side chain and escapes α-aminoadipate with a longer side chain (Fig. 3E), suggesting that AGT1 prefers anionic amino acids with shorter side chains.

In the apical membrane of the proximal tubule S3 segment, we have found that AGT1 is colocalized with Na⁺-dependent acidic amino acid transporter EAAC1 (30, 31) (Fig. 4). An in situ proximity ligation assay strongly suggests that they are present in the immediate vicinity of each other (Fig. 5). Such a colocalization in close proximity would enable efficient functional coupling between two transporters with overlapping substrates. As depicted in Fig. S9, aspartate and glutamate would be released into the luminal fluid via AGT1 when AGT1 reabsorbs substrates, including cystine, from the luminal fluid. The released aspartate and glutamate could be taken up and salvaged by closely located EAAC1 to prevent their urinary loss. The cystine taken up by AGT1 is converted to cysteine in the cytoplasm of epithelial cells, which ensures directional flow of cystine from tubular fluid to the cytoplasm of tubular epithelial cells. Therefore, it is proposed that AGT1 and EAAC1 in concert function as a cystine reabsorption system as a whole using aspartate and glutamate as internal-coupling substrates (Fig. S9). Through such cooperation, EAAC1 may furthermore drive and boost AGT1 by providing aspartate and glutamate to the intracellular substrate-binding site of AGT1, particularly when enough concentration of counter substrates is not available at the intracellular surface of the plasma membrane.

Fig. S9.

Proposed functional coupling of AGT1 with EAAC1. Cystine (CssC), as well as aspartate (Asp⁻) and glutamate (Glu⁻), is reabsorbed by AGT1 in exchange for Asp⁻ and Glu⁻ in the cytoplasm of tubular epithelial cells. Asp⁻ and Glu⁻ released into tubular lumen via AGT1 are proposed to be efficiently taken up by EAAC1 located in close proximity to AGT1, which prevents leakage of Asp⁻ and Glu⁻ into urine. CssC taken up by AGT1 is converted to Cys in the epithelial cells, which ensures directional flow of CssC from tubular fluid to the cytoplasm of tubular epithelial cells.

Two distinct cystine transport systems were once reported in the rodent kidney (32). One of them had no interaction with dibasic amino acids, which corresponds well with the AGT1-rBAT heterodimer characterized in the present study. An additional feature of such a cystine transport system described in BBMV was that its transport activity is increased in the presence of Na⁺. In our present study, Na⁺ dependence was not significant in the proteoliposome (Fig. 3C). It might be possible that some Na⁺-dependent transporters such as Na⁺-dependent acidic amino transporter EAAC1 as described above could cooperate and provide AGT1 with counter substrates to drive it, which makes the cystine transport somewhat Na⁺-dependent. It is reminiscent of the renal urate reabsorption system in which Na⁺-dependent monocarboxylate transporters provide counter substrates and drive Na⁺-independent urate transporter URAT1 to confer Na⁺ dependence on the urate reabsorption system (33).

In the physiological condition, more than 90% of cystine is reabsorbed in early proximal tubules, where b^0,+AT is the partner of rBAT, leaving a small portion of cystine to be reabsorbed by AGT1-rBAT. The small contribution to cystine reabsorption proposed for AGT1-rBAT may explain why cystine is not increased in the urine in dicarboxylic amino aciduria that is caused by the dysfunction of EAAC1 (34) although EAAC1 is supposed to drive AGT1 in some conditions as discussed above. However, when b^0,+AT is defective in mice, the fractional excretion of cystine is only 11%, and the remaining cystine reabsorption has been attributed to uncharacterized cystine transport (35). The small fractional excretion of cystine in b^0,+AT-defective mice suggests that cystine transporters other than b^0,+AT possess high reserve capacity to compensate largely the loss of b^0,+AT function, similar to the renal glucose reabsorption system in which SGLT1 in late proximal tubules makes up the glucose reabsorption by means of its high reserve capacity when SGLT2 in early proximal tubules is inhibited (36, 37). In fact, in the digenic inheritance of cystinuria mice, it was reported that the prevalence and severity of lithiasis are increased in the double homozygote (Slc7a9^−/-Slc3a1^−/-) compared with the single one (Slc7a9^−/-Slc3a1^+/+) (10), which supports the significant contribution of the heterodimer of rBAT and a light chain other than b^0,+AT to tubular cystine reabsorption. Notably, cystinuria patients who have no urinary hyperexcretion of dibasic amino acids were found in a German family and a British family (38, 39), as well as in dogs (40). Although the mutation T123M in b^0,+AT was reported for patients from the German family with such isolated cystinuria (38), AGT1 mutations might also explain additional cases of this type of cystinuria.

Taken all together, the evidence strongly suggests that the AGT1-rBAT heterodimer corresponds to a transport system involved in cystine reabsorption in renal proximal tubules. The identification of this second cystine transporter in proximal tubules has answered long-lived questions regarding cystine reabsorption. Simultaneously, it raises a series of new questions, such as the actual physiological contribution of AGT1 in renal cystine reabsorption, the biological sex difference in AGT1 expression, and the relevance of AGT1 to cystinuria, that require further studies to address these subjects.

Materials and Methods

SI Materials and Methods includes additional information on experimental methods.

Animal Studies.

All studies were approved by the Osaka University Animal Care and Use Committee and the Committee for the Use and Care of Animals in Institut d'Investigació Biomèdica de Bellvitge.

Antibodies, Immunoprecipitation, and Mass Spectrometry.

Anti-AGT1 antibodies were newly generated in this study as described in SI Materials and Methods. Immunoprecipitation of AGT1 or rBAT was performed, and the immunoprecipitated proteins were subjected to mass spectrometry to identify a heavy chain for AGT1 (SI Materials and Methods).

Immunohistochemistry and Immunofluorescence Microscopy.

Immunohistochemistry and immunofluorescence microscopy were performed as described in SI Materials and Methods. Superresolution SIM images were obtained by using an Elyra S1 microscope (Carl Zeiss). A duolink in situ proximity ligation assay (PLA) (Sigma-Aldrich) was used according to the manufacturer’s protocol.

Transport Assays on Proteoliposomes.

rBAT and 3xFLAG-AGT1 were purified as a heterodimeric complex and reconstituted into proteoliposomes by the dilution-freeze/thaw method (41). The proteoliposomes were subjected to transport assays as described in SI Materials and Methods.