Role of an extracellular loop in determining the stoichiometry of Na⁺–HCO₃⁻ cotransporters

Abstract

The Na⁺–HCO₃⁻ cotransporters (NBCs) of the solute carrier 4 family (SLC4) are critical for regulating pH in cells as well as in fluids such as blood and cerebrospinal fluid. Moreover, mutations and gene disruptions in NBC are linked to a wide range of pathologies. NBCe1 (SLC4A4) is electrogenic because it has an apparent Na⁺:HCO₃⁻ stoichiometry of 1:2 or 1:3, whereas NBCn1 (SLC4A7) is electroneutral because it has an apparent stoichiometry of 1:1. Because stoichiometry influences the effect of transport on membrane potential and vice versa, a central question is what structural features underlie electrogenicity versus electroneutrality. A previous study on rat NBCe1/n1 chimeras demonstrated that the structural elements determining the electrogenicity of NBCe1-A are located within the transmembrane domain, excluding the large third extracellular loop. In the present study we generated a series of chimeras of human NBCe1-A and human NBCn1-A. We found that replacing merely the predicted fourth extracellular loop (EL4) – containing 32 amino acid residues that include 7 prolines – of human NBCe1-A with EL4 of NBCn1-A creates an electroneutral NBC. The opposite switch converts an electroneutral construct to one with electrogenic properties. The introduction of an N-glycosylation site into EL4 confirms that at least a part of it is exposed to the extracellular fluid. We hypothesize that putative EL4 either contributes to the substrate-binding vestibule or indirectly influences substrate binding by interacting with one or more transmembrane segments, thereby controlling the nature of transport.

Non-technical summary

The Na⁺–HCO₃⁻ cotransporters NBCe1 and NBCn1 play critical roles in the regulation of intracellular pH and in acid–base movement across epithelial barriers such as the kidney. NBCe1 appears to carry 1 Na⁺ for every 2 or 3 HCO₃⁻ and thus is electrogenic (carries electrical current) and NBCn1 appears to carry 1 Na⁺ for every 1 HCO₃⁻ and thus is electroneutral (carries no electrical current). We systematically substituted homologous pieces of NBCn1 into NBCe1 and made the surprising observation that a putative extracellular loop (previously thought to be an unstructured segment of protein that dangles in the extracellular fluid) plays a key role in determining the stoichiometry of NBCe1 and NBCn1. We found that replacing just the fourth extracellular loop of NBCe1 with that of NBCn1 fully eliminates the electrogenicity, and that the opposite substitution converts an electroneutral chimera into one capable of electrogenic transport.

Introduction

The five Na⁺-coupled HCO₃⁻ transporters (NCBTs) – members of the solute carrier 4 (SLC4) family – play important roles in the regulation of intracellular pH (pH_i) as well as in transepithelial acid–base transport. Indeed, NCBT mutations and gene disruptions are linked to a wide range of pathologies (Hmani et al. 1999; Igarashi et al. 1999, 2001; Bok et al. 2003; Dinour et al. 2004; Inatomi et al. 2004; Horita et al. 2005; Gurnett et al. 2008; Jacobs et al. 2008; Suzuki et al. 2010; for reviews, see Romero, 2005; Pushkin & Kurtz, 2006). Two NCBTs transport net negative charge, the electrogenic Na⁺–HCO₃⁻ cotransporters NBCe1 (SLC4A4; see Romero et al. 1997) and NBCe2 (SLC4A5; see Sassani et al. 2002; Virkki et al. 2002). The other three NCBTs transport no net charge: the two electroneutral Na⁺–HCO₃⁻ cotransporters NBCn1 (SLC4A7; see Pushkin et al. 1999; Choi et al. 2000) and NBCn2 (SLC4A10, aka ‘NCBE’; see Wang et al. 2000; Parker et al. 2008) as well as the Na⁺-driven Cl⁻/HCO₃⁻ exchanger NDCBE (SLC4A8; see Grichtchenko et al. 2001). NBCe1 and NBCe2 can operate with an apparent Na⁺:HCO₃⁻ stoichiometry of either 1:2 or 1:3. NBCn1 and NBCn2 have apparent stoichiometries of 1:1.

The splice variant NBCe1-A is expressed predominantly in the basolateral membrane of renal proximal tubule, which reabsorbs ∼80% of the filtered HCO₃⁻. In proximal-tubule cells, NBCe1-A has a stoichiometry of 1:3 (Boron & Boulpaep, 1983; Soleimani et al. 1987; Romero et al. 1997), and therefore mediates HCO₃⁻ efflux. However, when expressed in Xenopus oocytes, NBCe1-A has a stoichiometry of 1:2 (Sciortino & Romero, 1999), and thus generally mediates HCO₃⁻ influx. Increases in [Ca⁺⁺]_i (Muller-Berger et al. 2001) and/or phosphorylation of Ser⁹⁸² (Gross et al. 2001b) may shift the stoichiometry from 1:3 to 1:2. NBCe1-B has a wide distribution that includes the basolateral membrane of pancreatic duct cells (Abuladze et al. 1998; Marino et al. 1999). Here, NBCe1-B has a stoichiometry of 1:2 and mediates HCO₃⁻ influx, thereby contributing to HCO₃⁻ secretion into the duct lumen (Gross et al. 2001a). NBCe1-C has a stoichiometry of 1:2 and mediates HCO₃⁻ influx into astrocytes (Bevensee et al. 2000; Majumdar et al. 2008), contributing to pH_i regulation (Bevensee et al. 1997). Mutations in the gene encoding human NBCe1 can cause autosomal-recessive renal proximal-tubule acidosis, mental retardation, glaucoma and cataracts (Igarashi et al. 1999, 2001; Inatomi et al. 2004; Dinour et al. 2004; Horita et al. 2005; for reviews, see Romero, 2005; Pushkin & Kurtz, 2006).

NBCn1 is widely expressed in multiple tissues, including kidney (Pushkin et al. 1999), heart (Choi et al. 2000) and brain (Bouzinova et al. 2005; Cooper et al. 2005; Chen et al. 2007). NBCn1 has an apparent stoichiometry of 1:1, and thus mediates a HCO₃⁻ influx that contributes to pH_i regulation. Knock-out of the gene encoding mouse NBCn1 causes blindness and deafness – hallmarks of Usher syndrome 2B – due to degeneration of sensory receptors in the retina and inner ear (Bok et al. 2003). Indeed, both the SLC4A7 gene in humans and one locus for Usher syndrome type 2B map to chromosome 3p24 (Hmani et al. 1999). Recently, a genome-wide association study of breast cancer implicated the loci of NBCn1 and the nearby NEK10 kinase (Ahmed et al. 2009). In the MCF-7 human breast cancer cell line, expression of ΔNErbB2 – a constitutively active ErbB2 receptor mutant, truncated in the N terminus (Nt), and common in breast cancer – substantially enhances NBCn1 expression and increases the pH_i recovery rate after acid loading (Lauritzen et al. 2010). Finally, a genetic linkage analysis revealed a significant linkage of the concentration of trace element Pb near the SLC4A7 locus (Whitfield et al. 2010).

The stoichiometry of an NCBT (e.g. NBCe1-A vs. NBCn1-A) is physiologically critical because it influences the magnitude and even the direction of transport, the energetic efficiency of transport (e.g. 1 vs. 2 HCO₃⁻/Na⁺), and the effect of transport on membrane potential (V_m), which could in turn affect voltage-sensitive processes. For example, if NBCe1-A in the renal proximal tubule were suddenly to shift from electrogenic to electroneutral, the transporter would mediate the uptake (rather than the efflux) of HCO₃⁻ across the basolateral membrane, leading to a severe proximal-type renal tubular acidosis, as observed with naturally occurring human NBCe1 mutations.

No high-resolution crystal structures are available for any SLC4-related protein, and the molecular mechanism of NCBTs still largely remains unknown. A key question is what parts of the proteins determine, for example, that NBCe1 is electrogenic whereas NBCn1 is electroneutral. SLC4 family members can be divided into five major structural domains (see Supplemental Fig. S1): (1) the large cytoplasmic Nt; (2) the front half of the transmembrane domain (TMD_F), including putative transmembrane segments (TMs) 1–5; (3) the long extracellular loop 3 (EL3) between TM5 and TM6; (4) the back half of the TMD (TMD_B), including putative TM6–14; and (5) the cytoplasmic C terminus (Ct). Based on a series of NBCe1/n1 chimeras, Choi et al. (2007) found that the electrogenicity of rat NBCe1 does not require Nt, EL3, or Ct, but does require the simultaneous presence of TMD_F and TMD_B. In the present study, our goal was to identify regions of TMD_B of NBCe1 that are necessary for electrogenicity in the context of NBCe1/n1 chimeras.

We generated a new series of chimeras of human (h) NBCe1-A and hNBCn1-A – which are ∼50% identical at the amino acid level – focusing exclusively on TMD_B (see Fig. 1). We expressed the constructs in Xenopus oocytes, and evaluated function by microelectrode measurements of intracellular pH (pH_i), membrane potential (V_m), and current-voltage (I–V) relationship. Perhaps our most striking observation is that replacing just the fourth extracellular loop (EL4) of NBCe1 with that of NBCn1 yields a functional but electroneutral chimera, whereas the inverse swap converts an electroneutral chimera to one with electrogenic properties.

Figure 1. Sequence boundaries of constructs used in the present study.

Based on work on AE1 or SLC4A1 (Fujinaga et al. 1999; Zhu et al. 2003), the SLC4 family members have a large cytosolic Nt, followed by a transmembrane domain that includes an estimated 14 transmembrane segments – 13 of which are presumably α-helixes (vertical bars) – and a short cytosolic Ct. Between TM5 and TM6 is the large 3rd extracellular loop. The 12th transmembrane segment is hypothesized to be a re-entrant loop. Components of human NBCe1-A (accession NM_003759) are indicated in violet, whereas components of human NBCn1 (accession AF047033) are indicated in orange. Red triangles indicate single loops swapped between NBCe1 and NBCn1.

Methods

Generation of chimeras

Chimeras were made from hNBCe1-A (accession NM_003759) and hNBCn1-A (accession AF047033). hNBCe1-A, in Xenopus expression vector pGH19, was tagged with enhanced green fluorescent protein (EGFP) at its N-terminus (Lu et al. 2006). The pGH19-EGFP-hNBCe1-A construct was genetically modified to contain a silent HindIII site at the 3′ end of TM5 of hNBCe1-A and an EcoRI site after the TAA stop codon. For each chimera construct, the appropriate cDNA fragments of hNBCe1-A and hNBCn1-A fragments were generated by polymerase chain reaction (PCR), the two fragments were annealed, and then the chimeric fragments (corresponding to the back half of the TMD plus the cytoplasmic Ct) were generated by PCR. The chimeric cDNA and the pGH19-EGFP-hNBCe1-A vector were each digested with HindIII and EcoRI, and then ligated. Glycosylation-site mutants of hNBCe1-A were generated using QuickChange site-directed mutagenesis kit (Agilent Technologies, Inc., Santa Clara, CA, USA). Figure 1 shows diagrams and sequences boundaries for all constructs that we used in the present study.

cRNA preparation and oocyte injection

The protocols for housing and using of Xenopus laevis were approved by the Institutional Animal Care and Use Committee at Case Western Reserve University. An ovary lobe was taken from Xenopus anaesthetized with 2 g l⁻¹ of ethyl-3-aminobenzoate methanesulfonate (Sigma-Aldrich, St Louis, MO, USA), placed in Ca²⁺-free NRS solution (see below for description of solutions), cut into small pieces, and digested in 2 mg ml⁻¹ of Type 1A collagenase (Sigma-Aldrich) in Ca²⁺-free NRS solution for 60–70 min at room temperature. The separated oocytes were then rinsed ×5 with Ca²⁺-free NRS solution, then ×5 with standard ND96. Stage V–VI oocytes were selected and kept in OR3 medium at 18°C. The animals were humanely killed following the final collection of oocytes.

Plasmid DNA containing the chimeric cDNA was linearized at the end of the 3′ untranslated region of pGH19 vector by restrictive digestion with XhoI or NotI. cRNA encoding the transporters was prepared with T7 RNA polymerase using the mMessage mMachine kit (Ambion, Austin, TX, USA) according to the manufacturer's instructions. Unless specified elsewhere, 50 nl cRNA of 0.5 μg μl⁻¹ was injected into each Xenopus oocyte. The oocytes were incubated at 18°C for 4–5 days in OR3 medium until being used for electrophysiological measurements.

Electrophysiology

Membrane potential and intracellular pH measurements were performed as previously described (Fei et al. 1994; Romero et al. 1997; Lu et al. 2006; Musa-Aziz et al. 2010). Briefly, a Xenopus oocyte injected with cRNA encoding a transporter, or injected with water as a control, was placed in a perfusion chamber. We then impaled the oocyte with V_m and pH_i microelectrodes, fabricated from fibre capillaries. The V_m electrode was filled with 3 m KCl and had a resistance of 1–2 MΩ. The pH_i electrode was baked at 250°C overnight and silanized ×45 min. The tip of the pH_i electrode was filled with proton-sensitive ionophore I cocktail B (Sigma-Aldrich), followed by back-filling with a phosphate buffer solution. The pH_i electrodes, calibrated with standard buffers at pH 6.0 and 8.0 before addition of the oocyte, had Nernst slopes of 55–60 mV/pH unit. A single-point recalibration was performed for pH_i electrode in ND96 (pH 7.50). The voltage due to pH_i was obtained by subtracting the signal of V_m electrode from that of the pH_i electrode (amplified by a FD223 electrometer, World Precision Instruments, Inc., Sarasota, FL, USA). Data were sampled by computer every 500 ms, using software written in-house.

In some experiments, rather than measure pH_i, we performed two-electrode voltage clamp with the OC-725C Oocyte Clamp (Warner Instruments, LLC, Hamden, CT, USA). The oocyte was held at a voltage close to the spontaneous resting potential until being subjected to the following voltage-clamp protocol: stepping in 20 mV increments from −160 mV to +60 mV, with each clamp step lasting for 120 ms, followed by a relaxation of 120 ms.

Solutions for physiological assays

All solutions were delivered from 140 ml syringes, driven by syringe pumps (Cat. no. 55-2226. Harvard Apparatus, Holliston, MA, USA), and flowed to the chamber via Tygon tubing as described previously (Lu & Boron, 2007). We titrated all solutions to pH 7.50 at room temperature (∼22°C) and adjusted the osmolality to 195 mosmol kg⁻¹. The following shows the composition of the solutions used in the present study (all concentrations in mm).

• Ca²⁺-free NRS: 82 NaCl, 2 KCl, 20 MgCl₂ and 5 Hepes.

• OR3 medium: one packet of L-15 medium was dissolved in 1.5 litres of H₂O, followed by the addition of 100 ml of penicillin–streptomycin (10,000 Units ml⁻¹ of penicillin, 10,000 μg ml⁻¹ of streptomycin; Invitrogen, Carlsbad, CA, USA) and 5 mm Hepes.

• Standard ND96: 96 NaCl, 2 KCl, 1 MgCl₂, 1.8 CaCl₂ and 5 Hepes.

• Na-free ND96: N-methyl-d-glucamine (titrated to pH 7.50 with HCl to generate NMDG⁺) replaced NaCl in standard ND96 solutions.

• 1.5% CO₂/10 mm HCO₃⁻: 86 NaCl, 2 KCl, 1 MgCl₂, 1.8 CaCl₂, and 5 Hepes; after adjusting pH to 7.50, we added 10 NaHCO₃. The solution was bubbled with 1.5% CO₂ balanced with O₂.

• Na-free 1.5% CO₂/10 mm HCO₃⁻: In the above solution, N-methyl-d-glucamine (titrated with HCl to pH 7.50 generate NMDG⁺) replaced NaCl. After the subsequent addition of 10 mm NMDG base, bubbling with 1.5% CO₂ balanced with O₂ generated 10 mm NMDG-HCO₃ and returned pH to 7.50.

• 5% CO₂/33 mm HCO₃⁻: 63 NaCl, 2 KCl, 1 MgCl₂, 1.8 CaCl₂, and 5 Hepes, followed by titration to pH 7.50 and the addition of 33 NaHCO₃. The solution was then bubbled with 5% CO₂ balanced with O₂.

Membrane-protein preparation, deglycosylation and Western blotting

Membrane proteins were prepared from Xenopus oocytes as described previously (Chen et al. 2008). Briefly, oocytes were homogenized in a buffer containing 7.5 mm NaH₂PO₄, 250 mm sucrose, 5 mm EDTA, 5 mm EGTA, pH 7.0 and 1% protease inhibitor cocktail for mammalian tissues (Cat. no. P8340, Sigma-Aldrich). The homogenate was centrifuged at 3000 g at 4°C to remove the cellular debris. The supernatant was then ultracentrifuged at 100,000 g at 4°C. The pellet of membrane protein was dissolved in a buffer containing 5% SDS, 20 mm Tris-HCl and 5 mM EDTA at pH 8.0 and stored in aliquots at −80°C until use. Total protein concentration was determined using the BCA Protein Assay Reagent (Cat. no. 23228 and no. 23224, Pierce Biotechnology, Inc., Rockford, IL, USA) according to the manufacturer's protocol.

Membrane proteins were deglycosylated with PNGase F (500 U μl⁻¹, New England BioLabs, Ipswich, MA, USA), and then separated on 4–20% SDS gel for Western blotting as described previously (Chen et al. 2010). Monoclonal mouse anti-GFP (Invitrogen, Carlsbad, CA, USA) antibody was used to probe the expression of EGFP-NBCe1-A or its mutants. We detected horseradish peroxidase (HRP)-conjugated secondary antibody by reacting with ECL plus Western Blotting Detection Reagents (Amersham Biosciences, Piscataway, NJ, USA) ×5 min, followed by exposure to X-ray film.

Data analysis and statistics

The rate of pH_i change and changes in membrane potential in unclamped oocytes are presented as means ±s.e.m. To compare the difference between means, a one-way ANOVA followed by Dunnett's comparison was performed using KaleidaGraph (v4, Synergy Software, Reading, PA, USA). P < 0.05 was considered statistically significant.

Results

V_m and pH_i responses in oocytes expressing WT NBCe1-A and NBCn1-A

Figure 2 shows diagrams of our constructs (left side) and representative recordings of pH_i and V_m (right side) from oocytes expressing WT NBCe1-A (construct ‘A’), or NBCn1-A (construct ‘B’), or a H₂O-injected control oocyte. Before the start of the records shown in the figure, the oocytes were superfused sequentially with nominally HCO₃⁻-free ND96 solution until V_m and pH_i stabilized. The green record on the right side of Fig. 2A shows that – for the oocyte expressing WT NBCe1-A – switching the extracellular solution from one lacking CO₂/HCO₃⁻ to one containing 1.5% CO₂/10 mm HCO₃⁻ (pH 7.50) causes pH_i to fall due to CO₂ entry, and then to recover (rise) due to Na⁺-coupled HCO₃⁻ uptake (Boron & De Weer, 1976a,b;). The rate of pH_i recovery (dpH_i/dt, indicated in the figure by the dashed line) is an index of the rate of uptake of HCO₃⁻ equivalents. The red record in Fig. 2A shows that introducing CO₂/HCO₃⁻ also causes an abrupt negative shift in V_m (ΔV_m), which gradually wanes as Na⁺ and HCO₃⁻ accumulate in the cytosol. Replacing extracellular Na⁺ with NMDG⁺– in the continued presence of CO₂/HCO₃⁻– leads to a rapid and reversible fall in pH_i, as well as a near-instantaneous and reversible depolarization (Fig. 2A). The aforementioned V_m changes are indicative of electrogenic transport.

Figure 2. Representative records of pH_i (green) and V_m (red, in mV) for oocytes expressing human NBCe1-A (construct ‘A’, A) or human NBCn1-A (construct ‘B’, B), or injected with H₂O (C).

In each case, the oocyte was exposed to ND96 (CO₂/HCO₃⁻-free) solution before being acid loaded by an exposure to 1.5% CO₂/10 mm HCO₃⁻ for 15 min. In the continued presence of CO₂/HCO₃⁻, extracellular Na⁺ was temporarily replaced with NMDG⁺ for 10 min. Finally, Na⁺ was returned in the continued presence of 1.5% CO₂/10 mm HCO₃⁻ for an additional 5 min. V_m as well as pH_i were monitored with microelectrodes. The composition of Constructs ‘A’ and ‘B’ are provided in Fig. 1.

For an oocyte expressing WT NBCn1-A (construct ‘B’), the dpHi/dt is smaller than for WT NBCe1-A, but ΔV_m is nil (Fig. 2B). Also, Na⁺ removal causes a hyperpolarization rather than a depolarization (red record in Fig. 2B). This relatively large hyperpolarization – starting from a rather depolarized state – reflects an intrinsic basal Na⁺ conductance of NBCn1 (Choi et al. 2000).

The control oocyte injected with H₂O exhibits minimal pH_i recovery, none of the V_m changes characteristic of electrogenic transport, and V_m changes consistent with a minimal Na⁺ conductance (Fig. 2C).

Although not shown, we also performed two-electrode voltage-clamp experiments on oocytes expressing either human NBCe1-A or human NBCn1-A. In our standard ND96 (i.e. CO₂/HCO₃⁻-free solution), the slope conductance of oocytes expressing human NBCn1-A averaged 7.1 ± 0.3 μS (n= 6), which is similar to the value of 7.7 μS reported earlier for rat NBCn1-B (Choi et al. 2000), and much larger than our value of 0.8 ± 0.1 μS (n= 14) for H₂O-injected oocytes. On the other hand, the slope conductance of oocytes expressing human NBCe1-A was 7.7 ± 0.2 μS (n= 15) in standard ND96, and 5.6 ± 0.6 μS (n= 7) in ND96 bubbled vigorously with 100% O₂ to minimize [CO₂]_o and [HCO₃⁻]_o. Choi et al. (2007) made similar observations for rat NBCe1-A in standard ND96. Because the conductance engendered by NBCe1-A in the nominal absence of CO₂/HCO₃⁻ is nearly as large as that engendered by NBCn1-A, we concluded that it would be difficult to assess chimeras for the Na⁺-conductance property of NBCn1-A. Therefore, we focused exclusively on the electrogenic transport property of our chimeras.

The structural elements in TMD_B that are essential for the electrogenicity of NBCe1-A lie between TM6-EL4, inclusive

Choi et al. (2007) have previously demonstrated that a chimera consisting of Nt, TMD_F and EL3 of rat NBCe1-A plus TMD_B and Ct of rat NBCn1-B is electroneutral. In order to elucidate the structural elements essential for the electrogenicity of NBCe1-A, we began by making a chimera comparable to that of Choi et al. – our construct ‘C’– but from human cDNA. As shown in Fig. 3A, when an oocyte expressing construct ‘C’ is exposed to 1.5% CO₂/10 mm HCO₃⁻, V_m does not undergo an instantaneous change. As far as the pH_i record is concerned, the initial CO₂-induced acidification is followed by a modest pH_i recovery, indicating that the chimeric transporter has a modest rate of HCO₃⁻ uptake. Thus, we can conclude that construct ‘C’ is a modestly active but electroneutral chimera, confirming the previous observation that the electrogenicity of NBCe1 requires the presence of structural elements in both TMD_F and TMD_B (Choi et al. 2007).

Figure 3. Representative records of pH_i (green) and V_m (red, in mV) of oocytes expressing chimeric construct ‘C’ (A), construct ‘D’ (B) and construct ‘E’ (C).

The experimental protocols are the same as for Fig. 2. The compositions of the constructs are provided in Fig. 1.

Starting from the electroneutral construct ‘C’, we progressively restored elements of the TMD_B of NBCe1-A to determine which are necessary for the electrogenicity of NBCe1-A. The results for an oocyte expressing construct ‘D’ (Fig. 3B) – a chimera consisting NBCe1-A from Nt through the third intracellular loop (IL3) and NBCn1-A from TM7 through Ct – are similar to those for construct ‘C’, indicating that construct ‘D’ is also electroneutral. However, on the background of construct ‘C’, restoring TM6–EL4 of NBCe1-A – thereby creating construct ‘E’– restores the electrogenicity of the transporter (Fig. 3C). Taking together the results of Fig. 2A–C, we can conclude that the part of NBCe1's TMD_B necessary for electrogenicity includes some part of TM6 through EL4, inclusive.

When we start from construct ‘E’ and restore more distal elements of the TMD_B of NBCe1, the constructs remain electrogenic but support substantial increases in both dpHi/dt and the magnitude of ΔV_m (constructs ‘F’–‘H’ in Supplemental Fig. S2A–C). Together, Fig. 3 and Supplemental Fig. S2 show that TM8–Ct, inclusive, of NBCe1-A is not necessary for electrogenicity. Nevertheless, some elements in this region greatly enhance the functional activities of the chimeras. (In our definition, functional activity includes the effects of both trafficking to the plasma membrane and the intrinsic activity of the transporter; Boron et al. 2009.)

Role of EL4 in determing electrogenicity of NBCe1-A

To refine our search for the key elements of TMD_B of NBCe1-A that are necessary for electrogenicity, we started from the electrogenic construct ‘G’ (Supplemental Fig. S2B) and generated four chimeras (constructs ‘I’, ‘J’, ‘K’, and ‘L’) by reverting – one loop or TM at a time – to the NBCn1 versions of IL3 through TM8, inclusive. Supplemental Fig. S3 shows the three such chimeras that remain electrogenic, namely the ones in which we reverted to the NBCn1 versions of IL3 (construct ‘I’, Supplemental Fig. S3A), TM7 (construct ‘J’, Supplemental Fig. S3B), and TM8 (construct ‘L’, Supplemental Fig. S3C). All three constructs also maintain substantial HCO₃⁻-transport activity. These observations confirm that TM8 of NBCe1 is not necessary for electrogenicity (see Fig. 3C), and demonstrate that IL3 and TM7 of NBCe1-A are not necessary for electrogenicity.

The key chimera in the construct-‘G’ series proved to be the one in which we reverted to the NBCn1 version of EL4 (construct ‘K’, Fig. 4A). This chimera abolishes electrogenicity while maintaining HCO₃⁻-transport activity. Thus, EL4 is likely to be the key structural element in the TM6–EL4 region that is essential for electrogenicity. Indeed, on the background of WT NBCe1-A, replacing only EL4 with the comparable portion of NBCn1 eliminates electrogenicity while maintaining HCO₃⁻ transport (construct ‘M’, in Fig. 4B). Thus, within the TMD_B of NBCe1, EL4 is the essential element for enabling electrogenicity in the context of e1/n1 chimeras.

Figure 4. Representative records of pH_i (green) and V_m (red, in mV) of oocytes expressing chimeric construct ‘K’ (A), construct ‘M’ (B) and construct ‘N’ (C).

The experimental protocols are the same as for Fig. 2. The compositions of the constructs are provided in Fig. 1.

Because substituting just the EL4 of NBCn1 into NBCe1 can so strikingly reduce the electrogenicity, it is tempting to ask whether the reverse chimera – created by substituting the EL4 of NBCe1 into NBCn1 – is electrogenic. However, this would be a fruitless experiment because we know from Choi et al. (2007) that the electrogenicity of NBCe1-A requires components of both its TMD_F and TMD_B. Therefore, starting with the electroneutral construct ‘C’– consisting of the Nt, TMD_F and EL3 from NBCe1 and the TMD_B and Ct from NBCn1 – we created construct ‘N’ by replacing just the EL4 of the TMD_B NBCn1 module with the EL4 of NBCe1. Figure 4C shows that construct ‘N’ not only mediates a modest pH_i recovery but also generates a substantial ΔV_m characteristic of an electrogenic NBC. Our results demonstrate that restoring EL4 converts an electroneutral NBC construct into one capable of electrogenic HCO₃⁻ transport. These results further demonstrate that EL4 of NBC plays critical roles in determining the electrogenicity or electroneutrality of NBC.

Note that oocytes expressing construct ‘M’ have a relatively low initial V_m (−43.3 ± 2.6 mV, n= 16) and exhibit a large hyperpolarization in response to Na⁺ removal – both characteristic of NBCn1 with its Na⁺ conductance (Choi et al. 2000). Presumably construct ‘M’, the human NBCe1-A with EL4 replaced with that of human NBCn1, also contains a Na⁺ conductance.

An interesting observation is that when we remove extracellular Na⁺ from oocytes expressing construct ‘N’, V_m sometimes becomes more positive as in Fig. 4C but sometimes becomes more negative (see Supplemental Fig. S4). We hypothesize that construct ‘N’ can mediate currents that represent both electrogenic cotransport and a Na⁺ conductance, and that these effects approximately cancel under conditions of Na⁺ removal. Taken together, the Na⁺-removal data for constructs ‘M’ (Fig. 4B) and ‘N’ (Fig. 4C and Supplemental Fig. S4) are consistent with the hypothesis that the EL4 of either NBCn1 or NBCe1 can support a Na⁺ conductance. These data are also consistent with the possibility that WT NBCe1-A mediates a Na⁺ conductance whose ΔV_m manifestation in a Na⁺-removal protocol is simply overwhelmed by electrogenic cotransport.

Figure 5 summarizes the mean initial rate of the pH_i recovery (dpHi/dt, Fig. 5B) and membrane potential changes upon application of 1.5% CO₂/10 mm HCO₃⁻ (ΔV_m, Fig. 5C) for a larger number of experiments of the kind shown in the previous figures. As we can see, the ΔV_m of chimeras ‘C’, ‘D’, ‘K’ and ‘M’ are not statistically different from that of WT NBCn1-A (construct ‘B’), whereas their dpH_i/dt values are substantially greater than that of H₂O-injected oocytes, confirming that these chimeras are electroneutral.

Figure 5. Constructs (A) and summary of pH_i recovery rates (B) and maximum V_m changes (C).

pH_i recovery rates (dpH_i/dt) of oocytes after the CO₂-induced acid load were calculated from experiments like those shown in Figs 2–4 (indicated by dashed lines in the green record of pH_i). V_m changes represent the maximum instantaneous V_m changes (ΔV_m) upon addition of 1.5% CO₂/10 mm HCO₃⁻ during the first CO₂/HCO₃⁻ impulse. Data are presented as means ±s.e.m. We used a one-way ANOVA in combination with Dunnett's multiple comparison for statistical comparisons of both the dpHi/dt and ΔV_m values with their corresponding values for H₂O-injected oocytes. All means were significantly different from those for H₂O-injected oocytes except for those indicated by NS (not significant).

Transport ratios of Na⁺ to HCO₃⁻ by chimeras

To quantify the currents carried by our chimeras, we performed two-electrode voltage-clamp experiments. Figure 6A shows the I–V relationship obtained on an oocyte expressing the electrogenic construct ‘H’, first in the absence (blue) and then in the presence of CO₂/HCO₃⁻ (red). The difference curve (green) shows the HCO₃⁻-dependent I–V relationship for this electrogenic construct. Figure 6B summarizes comparable difference I–V relationships – representing mean data – for constructs ‘H’, ‘M’ and ‘N’ as well as for H₂O-injected oocytes. Our goal is to compare the currents to the rates of pH_i recovery. For example, the leftmost green bar in Fig. 6C shows that construct ‘H’ had a mean dpHi/dt of ∼20 × 10⁻⁵ pH units s⁻¹, whereas the rightmost green bar shows that the H₂O-injected oocytes had a mean dpHi/dt of ∼2 × 10⁻⁵ pH units s⁻¹ (all data from Fig. 5B). Thus, corrected for the H₂O background, construct ‘H’ had a mean dpHi/dt of 18 × 10⁻⁵ pH units s⁻¹. The leftmost orange bar shows that construct ‘H’ had a HCO₃⁻-dependent current – at a V_m of −60 mV (the value during a pH_i recovery) – of ∼260 nA.

Figure 6. Current–voltage (I–V) relationships (A and B), and comparison of HCO₃⁻-dependent currents with corresponding mean pH_i-recovery rates of oocytes expressing construct ‘H’, construct ‘M’, or construct ‘N’, or injected with H₂O (C).

A, representative I–V curves of an oocyte expressing construct ‘H’, first in the ND96 (nominally CO₂/HCO₃⁻-free) solution (blue) and then in 5% CO₂/33 mm HCO₃⁻ (red). The HCO₃⁻-dependent current (green) was computed by subtracting the current in ND96 from the current in CO₂/HCO₃⁻ at the corresponding V_m. B, dependence of mean HCO₃⁻-dependent current on V_m for oocytes expressing construct ‘H’ (n= 7), ‘M’ (n= 7), or ‘N’ (n= 6), or for oocytes injected with H₂O (n= 8). C, mean pH_i-recovery rates (green bars) and HCO₃⁻-dependent currents () at −60 mV. The pH_i data are taken from Fig. 5B, and the current data, from panel B in this figure. NS, not significantly different from H₂O-injected oocytes by one-way ANOVA.

Construct ‘M’ had a mean dpHi/dt, corrected for H₂O, of 5 × 10⁻⁵ pH units s⁻¹. Thus, if construct ‘M’ were electrogenic, we might expect the current at −60 mV to be (5/18) × 260 nA ≈ 70 nA. In fact, the I–V curve for construct ‘M’ lies along the abscissa (Fig. 6C, 2nd pair of bars). Thus, we can conclude that this chimera must operate almost entirely in the electroneutral mode.

Construct ‘N’ had a mean dpHi/dt, corrected for H₂O, of 5 × 10⁻⁵ pH units s⁻¹– indistinguishable from that of construct ‘M’. The I–V curve for construct ‘N’ is slightly biphasic, indicating that an unknown, HCO₃⁻-dependent conductance must have been operative at extreme negative voltages but must have decayed at more positive voltages. At a V_m of −60 mV, where the contribution from this unknown current should have been minimal, construct ‘N’ had a mean current of ∼35 nA (Fig. 6C, 3rd pair of bars). Thus, the dpHi/dt-to-current ratio of construct ‘N’ ([5 × 10⁻⁵ pH units s⁻¹]/[35 nA]= 1.4 × 10⁻⁶ pH units s⁻¹ nA⁻¹) is twice that of construct ‘H’ ([18 × 10⁻⁵ pH units s⁻¹]/[260 nA]= 0.7 × 10⁻⁶ pH units s⁻¹ nA⁻¹). That is, for a given amount of current, construct ‘N’ carries twice as many HCO₃⁻ equivalents as construct ‘H’, suggesting that the construct ‘N’ mediates not only electrogenic transport, but also some degree of electroneutral transport.

Topological localization of EL4 of NBCe1-A

Topological models of the SLC4 family predict that ‘EL4’ is extracellular (Zhu et al. 2003; Romero et al. 2004; Stewart et al. 2007b). Moreover, in AE1 (SLC4A1) – an anion exchanger of the SLC4 family, which is ∼30% identical to NBCe1 and NBCn1, and should therefore have the same fold – EL4 is the lone site of N-glycosylation (Groves & Tanner, 1994). In order to confirm that EL4 is extracellular in the NBCs, we performed N-glycosylation studies on human NBCe1-A.

Rat NBCe1-A contains three N-glycosylation sites on the third extracellular loop (Choi et al. 2003). We eliminated the three consensus N-glycosylation sites on EL3 of human NBCe1 to generate the mutant construct N^592,597,617Q. Based upon this mutant, we introduced a consensus N-glycosylation motif ‘NSSA’ into the centre of the predicted fourth extracellular loop in a position analogous to the glycosylation site of AE1 (Groves & Tanner, 1994). We designate the resulting construct as ‘EL4:Glc’. Figure 7A shows the sequence alignment of EL4 of human NBCe1, human NBCn1, and the ‘EL4Clc’ construct (with the consensus N-glycosylated motif ‘NSSA’ underlined). Figure 7B shows that PNGase F treatment greatly reduces the molecular weight of human NBCe1-A (lanes 1–2), indicating that the wild-type human transporter is highly N-glycosylated when heterologously expressed in Xenopus oocytes. PNGase F treatment has no effect on the molecular weight of the NBCe1-A mutant N^592,597,617Q that lacks all three glycosylation sites (lanes 3–4), indicating it is not N-glycosylated when expressed in Xenopus oocytes. PNGase F treatment slightly reduced (by ∼5 kDa, as indicated by the dashed lines in Fig. 7B) the molecular weight of the NBCe1-A construct ‘EL4:Glc’ (lanes 5–6), which is thus N-glycosylated when expressed in Xenopus oocytes. This observation proves that at least the central portion of the EL4 of NBCe1 is extracellular, although we cannot say how much of the flanking regions of EL4 are accessible to the extracellular fluid.

Figure 7. Sequence alignment of EL4 of hNBCe1-A, hNBCn1-A and a hNBCe1-A mutant EL4:Glc (A), and effects of PNGase F treatment on WT hNBCe1-A or mutants expressed in Xenopus oocytes (B).

The numbers in panel A indicate the numbering of amino acid residues of hNBCe1-A. The residues highlighted in black are identical between hNBCe1 and hNBCn1, whereas those highlighted in grey represent conservative substitutions. The bolded characters are proline residues. The hNBCe1-A triple mutant N^592,597,617Q was generated by mutating the Asn residues to Gln in the three consensus N-glycosylation motifs in EL3 of hNBCe1-A. Mutant EL4:Glc was created by introducing an N-glycosylation motif ‘NSSA’ into EL4 of the triple mutant N^592,597,617Q at the position indicated in panel A. Regarding panel B, membrane proteins were extracted from oocytes expressing EGFP-tagged WT hNBCe1-A, the triple mutant N^592,597,617Q, or EL4:Glc, and were treated in the absence or presence of PNGase F, separated by SDS-polyacrylamide gel electrophoresis, blotted onto a PVDF membrane, and then probed with an anti-GFP antibody. Shown here is a blot representative of four experiments.

Electrophysiology assays demonstrate that both the triple mutant (N^592,597,617Q, Fig. 8A) and the construct with the N-glycosylation motif introduced in EL4 (EL4:Glc, Fig. 8B) remain functional as electrogenic NBCs. Figure 8C summarizes pH_i recovery rates of the triple mutant and EL4:Glc in comparison with WT human NBCe1-A as well as H₂O-injected oocytes. Not surprisingly, the mutation on the glycosylation sites on EL3 of NBCe1-A modestly reduces the functional expression of the transporter.

Figure 8. Representative records of pH_i (green) and V_m (red, in mV) of oocytes expressing the triple mutant N^592,597,617Q (A) or mutant EL4:Glc (B), and summary of pH_i recovery rates of oocytes expressing transporters or injected with H₂O (C).

Records of pH_i and V_m (in mV) were obtained, using a protocol identical to that in Figs 2–4. In panel A, the oocyte was expressing N^592,597,617Q, the hNBCe1-A mutant incapable of being N-glycosylated. In panel B, the oocyte was expressing EL4:Glc, the hNBCe1-A mutant with the N-glycosylation motif ‘NSSA’ introduced – on the background of the triple mutant N^592,597,617Q – into the fourth extracellular loop. Panel C summarizes the pH_i recovery rates of oocytes expressing WT hNBCe1-A, the triple mutant N^592,597,617Q, or EL4:Glc, or of control oocytes injected with H₂O. In each case, values represent pH_i recovery rates (dpH_i/dt) in the presence of 1.5% CO₂/10 mm HCO₃⁻. Data are presented as means ±s.e.m. We used a one-way ANOVA in combination with Dunnett's multiple comparison for statistical comparisons of the dpHi/dt values with that of H₂O-injected oocytes. All means were significantly different from that for H₂O-injected oocytes (P < 0.0001) and all means were significantly different from that for hNBCe1-A expressing oocytes (P= 0.026 for N^592,597,617Q vs. hNBCe1-A, and P= 0.0002 for EL4:Glc vs. hNBCe1-A).

We also made a second N-glycosylation mutant EL4:2Glc in which we introduced two N-glycosylation motifs into EL4. PNGase treatment showed that EL4:2Glc is also slightly N-glycosylated (see Supplemental Fig. S4), confirming that EL4 is localized on the extracellular side. However, functional analysis showed that EL4:2Glc is nearly inactive (see Supplemental Fig. S5).

Discussion

Ionic stoichiometry of construct ‘N’

Figure 6C indicates that the apparent HCO₃⁻ flux (as evidenced by the dpHi/dt values) mediated by construct ‘N’ is about twice as large as expected based on our data for the electrogenic construct ‘H’. If construct ‘H’– like WT NBCe1-A when expressed in Xenopus oocytes – operates with an apparent Na⁺:HCO₃⁻ stoichiometry of 1:2 (i.e. 2 HCO₃⁻/net negative charge; see Sciortino & Romero, 1999), then construct ‘N’ would appear to operate with a stoichiometry of about 3:4 (i.e. 4 HCO₃⁻/net negative charge). Note that, although we would like to compare the dpHi/dt-to-current ratios of constructs ‘H’ and ‘N’ with that of WT NBCe1-A, the pH_i recovery for WT NBCe1-A (i.e. construct ‘A’, Fig. 2A) occurs over a much higher range of pH_i values than for either construct ‘H’ (Supplemental Fig. S2C) or construct ‘N’ (Fig. 4C). Thus, the dpHi/dt for WT NBCe1-A would be artificially depressed, rendering the comparison meaningless.

Assuming that construct ‘H’ has a 1:2 stoichiometry, the simplest explanation is that an individual construct ‘N’ slips between stoichiometries of 1:2 and 1:1, mediating approximately one electrogenic transport cycle for every two electroneutral cycles. (Of course, we cannot rule out the possibility that one-third of the constructs are consistently electrogenic, and two-thirds consistently electroneutral. If the slippage hypothesis is true, then it is intriguing to speculate that even WT NBCe1-A may occasionally mediate an electroneutral cycle.) Thus, although we can fully convert NBCe1 into an electroneutral transporter by replacing its EL4 with that from NBCn1 (Fig. 4B), the opposite replacement converts the electroneutral construct ‘C’ into the partially electrogenic construct ‘N’. Thus, other portions of the TMD_B of NBCe1 must be required for full electrogenicity (i.e. to eliminate slippage into a partially electroneutral mode) – at least on the background of construct C.

Role of EL4 in determining stoichiometry of NBC

The most striking feature of the predicted extracellular loop ‘EL4’ is that, for both NBCe1 and NBCn1, 7 of the predicted 32 amino acids are prolines (see bold letters in Fig. 7A). The large number of proline residues makes it unlikely that this segment is a transmembrane helix. Moreover, our N-glycosylation study demonstrates that presumed EL4 is indeed exposed to the extracellular side of the membrane. Because no high-resolution structure of an SLC4 family member is available, we do not accurately know the borders of EL4 with putative TM7 and TM8. Note that among the presumed 32 amino acids of EL4, hNBCn1 has a Pro at position 705 (i.e. the putative second residue of EL4), and both hNBCe1 and hNBCn1 have a Pro at position 733 (i.e. two residues from the putative end of EL4). Moreover, the final two EL4 residues of both NBCe1 and NBCn1 are Trp. Thus, it is likely that very little of the putative EL4 is part of TM7 or TM8, and that nearly all of it can be exposed to the extracellular fluid.

Regarding the homology between the putative EL4s of NBCe1 and NBCn1, note that Val⁷⁰⁴, Asn⁷³², Pro⁷³³, Trp⁷³⁴ and Trp⁷³⁵ are identical in NBCe1 and NBCn1. Thus, the key amino acids that determine electrogenicity/electroneutrality must lie between hNBCe1 residues Asp⁷⁰⁵ and Glu⁷³¹, inclusive. Of these 27 residues, 14 are identical between hNBCe1 and hNBCn1, including 5 of the 6 remaining Pro residues.

What is the mechanism by which EL4 determines the electrogenicity/electroneutrality of NBCe1/n1? One possibility is that EL4 contributes to the substrate-binding vestibule. Alternatively, EL4 may not be directly involved in substrate binding, but rather may influence – for example, by exerting force on one or more TMs – the conformation of the vestibule.

In the present study, we demonstrated that replacing only the EL4 of NBCe1 with that of NBCn1 eliminates electrogenicity of NBCe1. It is impossible to determine from the present data whether the effect is due to the absence of the EL4 of NBCe1 or to the presence of the EL4 of NBCn1. One way of addressing this issue would be to replace the EL4 of NBCe1 with the EL4 of NBCe2, or NBCn2, NDCBE, or other members of the SLC4 family.

Conversely, in the present study, we showed that replacing only the EL4 of NBCn1 with that of NBCe1 – on the background of construct ‘C’– converts a fully electroneutral chimera into one with at least a partial electrogenic character. The reason that we started with construct ‘C’ is that Choi et al. (2007) showed – in rat NBCe1/n1 chimeras – that neither TMD_F nor TMD_B, alone, of NBCe1 can engender electrogenicity. Again, we do not know whether the effect is due to the absence of the EL4 of NBCn1 or the presence of EL4 of NBCe1. And again, this issue could be addressed by extending the replacements to other members of the SLC4 family.

Significance

Understanding the molecular basis of ion translocation by transporters is of fundamental importance in physiology. In the present study, we provide evidence that very little of the putative fourth extracellular loop is part of a putative adjacent transmembrane helix, that at least a portion of putative EL4 is exposed to the extracellular fluid, and that EL4 plays a key role in determining the stoichiometries of NBCe1 and an NBCe1/n1 chimera.

At least three other papers have reported that modifications of putative extracellular loops can alter substrate affinity or the pH sensitivity of transporters. First, Tamura et al. (1995) were able to change the γ-aminobutyric acid (GABA) affinity of the GABA transporter GAT1 by making mutations in either of two extracellular loops. Second, Stewart et al. (2007a) showed that mutations in the putative EL4 can alter the sensitivity of anion exchanger AE2 (SLC4A2) to extracellular pH. Third, Puntheeranurak et al. (2007) demonstrated that three putative extracellular loops contribute to the extracellular vestibule of the Na⁺–glucose cotransporter SGLT1. Moreover, mutational studies indicate that the four cysteines on EL4 of SGLT1 play important roles in determining the affinity for d-glucose.

Certain manoeuvres involving extracellular loops can have dramatic but potentially less specific effects on the function of transporters. For example, Compton et al. (2010) reported that the extracellular loop between transmembrane segments 3 and 4 is essential for amino acid transport by the archaeal glutamate transporter. They found that cleavage of the loop abolished transport activity while leaving the structure of the transporter intact as demonstrated by circular dichroism measurements and size-exclusion chromatography. Working with the electroneutral Na⁺–phosphate cotransporter NaPi-IIc, Ghezzi et al. (2009) found that adding a fluorophore to an artificially introduced Cys on EL3 inhibits cotransport.

Not surprisingly, changes to residues in transmembrane domains can have effects that are both more specific and more dramatic than those summarized above for manipulations of extracellular loops. For example, Bacconi et al. (2005) converted NaPi-IIc from its natural electroneutral mode to an electrogenic mode by substituting three homologous residues from TM3 of the electrogenic NaPi-IIa. To our knowledge, the present experiments with NBCe1/n1 chimeras are the first to reveal that changes to a putative loop of a cotransporter or exchanger can produce effects that are dramatic (i.e. alteration of electrogenicity or stoichiometry) and yet specific (e.g. not global inhibition or blockade). Of course, our observations and those of others on extracellular loops may be part of a continuum of effects.

Concluding thoughts

In the present study, we demonstrate that the putative fourth extracellular loops of NBCe1 and NBCn1 play critical roles in determining stoichiometries of these NCBTs. We hypothesize that EL4 – presumably with unidentified components of TMD_F and perhaps with other components of TMD_B– forms the kernel of the ion-translocating machinery of NBCe1 and NBCn1, and maybe other members of the SLC4 family. It is intriguing to speculate that structure(s) on the intracellular side of NBCe1 and NBCn1 play a role analogous to that of EL4 on the extracellular side. In a broader physiological context, our EL4 observations extend the aforementioned work on the extracellular loops of GAT1, AE2 and SGLT1, and support the evolving notion that such loops can have important roles in the action of ion transporters.

Glossary

Abbreviations

Ct

carboxyl terminus

EGFP

enhanced green fluorescent protein

EL

extracellular loop

IL

intracellular loop

NBC

Na⁺–HCO₃⁻ cotransporter

NCBT

Na⁺-coupled HCO₃⁻ transporter

NDCBE

Na⁺-driven Cl⁻/HCO₃⁻ exchanger

NMDG

N-methyl-d-glucamine

Nt

N terminus

PCR

polymerase chain reaction

SLC

solute carrier

TM

transmembrane segment

TMD

transmembrane domain

TMD_B

back half of TMD (putative TM6–TM14 onward)

TMD_F

front half of TMD (TM1–TM5)

WT

wild type

Author contributions

Conception and design of the experiments: L.-M.C. and W.F.B; plasmid construction: L.-M.C and Y.L.; electrophysiological data collection: L.-M.C; Western blotting: Y.L.; data analysis and interpretation: L.-M.C. and W.F.B.; drafting: L.-M.C and W.F.B.; final approval: L.-M.C., Y.L. and W.F.B.

Supplementary material

Figure S1

Figure S2

Figure S3

Figure S4

Figure S5

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer-reviewed and may be re-organized for online delivery, but are not copy-edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors