Lack of Charge Interaction in the Ion Binding Site Determines Anion Selectivity in the Sodium Bicarbonate Cotransporter NBCe1

Abstract

The Na/HCO₃ cotransporter NBCe1 is a member of SLC4A transporters that move HCO₃⁻ across cell membranes and regulate intracellular pH or transepithelial HCO₃ transport. NBCe1 is highly selective to HCO₃⁻ and does not transport other anions; the molecular mechanism of anion selectivity is presently unclear. We previously reported that replacing Asp⁵⁵⁵ with a Glu (D555E) in NBCe1 induces increased selectivity to other anions, including Cl⁻. This finding is unexpected because all SLC4A transporters contain either Asp or Glu at the corresponding position and maintain a high selectivity to HCO₃⁻. In this study, we tested whether the Cl⁻ transport in D555E is mediated by an interaction between residues in the ion binding site. Human NBCe1 and mutant transporters were expressed in Xenopus oocytes, and their ability to transport Cl⁻ was assessed by two-electrode voltage clamp. The results show that the Cl⁻ transport is induced by a charge interaction between Glu⁵⁵⁵ and Lys⁵⁵⁸. The bond length between the two residues is within the distance for a salt bridge, and the ionic strength experiments confirm an interaction. This finding indicates that the HCO₃⁻ selectivity in NBCe1 is established by avoiding a specific charge interaction in the ion binding site, rather than maintaining such an interaction.

1. Introduction

NBCe1 is a membrane protein that mediates electrogenic Na⁺-HCO₃⁻ and/or CO₃²⁻ transport across cell membrane and regulates intracellular and extracellular pH, as well as transepithelial HCO₃⁻ transport in many cells [1,2,3,4]. NBCe1 was first physiologically identified in the kidney proximal tubules [5], where it is responsible for reabsorbing two thirds of filtered HCO₃⁻. NBCe1 is highly selective to HCO₃⁻ and does not transport other anions, including Cl⁻ [5]. The gene encoding NBCe1 was expression-cloned in the late 1990s by Romero et al. [1]; since then, it has provided valuable information on the molecular and cellular physiology of Na/HCO₃ transport mediated by NBCe1 and its family proteins, collectively the Na⁺-coupled bicarbonate transporters, in humans. NBCe1 exists as multiple variants, due to different N- and C-terminal sequences, and each variant differs in tissue expression, intrinsic functional properties, and regulation [6]. The transporter has a Na⁺:HCO₃⁻ stoichiometry of 1:3 in renal proximal tubules and 1:2 in other cells, as well as in heterologous expression systems. Overall, NBCe1 plays an important role in the physiology and pathophysiology of many different organs, such as kidneys, heart, brain, eyes, enamel organs, and intestines [7,8,9,10].

The cryoEM structure of NBCe1 [11] has provided details on the protein structure and amino acid residues responsible for ion transport. NBCe1 is a homodimer with each monomer, consisting of 14 transmembrane segments (TMs), extracellular loops, and cytoplasmic regions. TMs 5–7 and 12–14 form the gate domain and TMs 1–4 and 8–11 form the core domain, while the cavity between these two domains houses an ion access pathway, through which Na⁺ and CO₃²⁻ (HCO₃⁻) move. The ion accessibility pathway is formed by TMs 1, 3, 5, 8, 10, 12, and a short loop connecting TM13 and TM14, and the diameter along the pathway varies, ranging from >12 Å in the entrance region to ~2 Å diameter in the middle of the protein. Many amino acid residues lining the pathway were previously recognized as residues critical for transport function by mutagenesis studies [12,13,14,15,16,17]. Together with the structures of the Na⁺-driven Cl/HCO₃⁻ exchanger NDCBE [18] and Cl/HCO₃ changer AE1 [19], NBCe1 structure has greatly advanced our understanding of the molecular mechanisms underlying Na⁺, Cl⁻, HCO₃⁻, and CO₃²⁻ transport via SLC4A bicarbonate transporters.

Recently, Zhekova et al. [20] performed site identification by ligand competitive saturation (SILCS) mapping of the binding pockets in human AE1 and NBCe1, followed by molecular dynamics (MD) simulations, and proposed two putative anion binding sites in the ion accessibility pathway of the proteins: central site S1 and entrance site S2. Site S2 serves as a transient binding site, to attract anions from the surrounding solution before ion movement to site S1, where the anion binding induces a protein conformational change for ion translocation. In NBCe1, site S2 is composed of Asp⁵⁵⁵, Lys⁵⁵⁸, Lys⁵⁵⁹, and Lys⁵⁶², all of which are in TM 5, and site S1 has residues from multiple TMs and loops. We have previously reported that substituting Asp⁵⁵⁵ with a Glu (D555E) causes the transporter to be permissive to other anions, including Cl⁻, NO₃⁻, SCN⁻, I⁻, and Br⁻ [17]. D555E maintains favorable access to HCO₃⁻; thus, it produces an outward current (I_NBC) when HCO₃⁻ is available but a Cl⁻ current (I_Cl) when HCO₃⁻ is unavailable. This modified selectivity should be related to a geometrical difference in the carboxyl side chain of Asp vs. Glu, due to an additional carbon backbone. On the other hand, all members of Na⁺-coupled bicarbonate transporters contain either Asp or Glu at the corresponding site, implicating that the geometrical difference in the carboxyl side chain is not the only cause for altered anion selectivity and an additional mechanism should be involved. Elucidating that mechanism will help us understand how bicarbonate transporters selectively transport HCO₃⁻, while excluding other anions.

In this study, we investigated the effect of charge interactions between residues in site 2 on anion selectivity. Candidate residues were changed by TM5 replacement and site-directed mutagenesis, and the mutant transporters were expressed in Xenopus oocytes and subjected to recordings of I_Cl, I_NBC, and intracellular pH (pH_i). The results show that I_Cl is induced by a charge interaction between residues in site 2. The two residues involved in the interaction are not simultaneously present in any member of the SLC4A bicarbonate transporters; thus, the HCO₃⁻ selectivity is maintained by avoiding a charge interaction in the ion binding site, located at the entrance of the ion accessibility pathway. We also find that Na⁺ is required for HCO₃⁻ access to the transporter, consistent with a conventional concept of a Na⁺ prerequisite for substrate movement in secondary active transporters.

2. Results

2.1. I_Cl Produced by D555E

To record I_Cl, produced by D555E, we expressed human NBCe1 and mutant D555E in oocytes and applied them with 71 mM Cl⁻ during superfusion of Cl⁻-free solution. Figure 1A shows a representative current trace, produced by NBCe1, in voltage clamp (the holding potential of −60 mV). NBCe1 did not produce measurable I_Cl, in response to bath Cl⁻, but produced an outward I_NBC upon exposure to 5% CO₂, 25 mM HCO₃⁻, consistent with its electrogenic Na/HCO₃ cotransport activity. In contrast, D555E produced an outward I_Cl, in response to Cl⁻ (Figure 1B). I_Cl was markedly decreased in the presence of CO₂/HCO₃⁻, consistent with our previous report [17] that D555E produces I_Cl, which can be inhibited by HCO₃⁻. Mean I_Cl from other oocytes (n = 6/group) is summarized in Figure 1C. On average, 70% of I_Cl produced by D555E was reduced in the presence of CO₂/HCO₃⁻ (p < 0.01, two-way ANOVA). In other experiments, we then determined I–V relationships for I_Cl and I_NBC to compare the current responses at different voltages in NBCe1 vs. D555E. As shown in Figure 1D, D555E evoked large outward currents at positive potentials in ND96 solution containing 96 mM Cl⁻ (p < 0.01, n = 5), reflecting Cl⁻ influx. However, in CO₂/HCO₃⁻ solution (Figure 1E), NBCe1 produced larger outward I_NBC than D555E at positive potentials (p < 0.05, n = 5). The two I–V curves were parallel to each other in the outward direction, as they are I_NBC. The curves crossed at a negative potential (approximately −80 mV), probably due to Cl⁻ efflux via D555E in the inward direction.

2.2. Na⁺ Prerequisite for HCO₃⁻ to Access Its Binding Site

The above results reveal that Cl⁻ transport by D555E is less favorable than HCO₃⁻ transport, when both ions are present in the bath. To determine whether this feature depends on Na⁺, we performed two sets of experiments. In the first set of experiments, we recorded I_Cl in Na⁺-free CO₂/HCO₃⁻ solution and tested whether I_Cl could be reduced under this condition. Representative recordings of I_Cl, produced by NBCe1 and D555E, are shown in Figure 2A,B. In contrast to NBCe1, D555E produced I_Cl in the absence of CO₂/HCO₃⁻ and, more importantly, in the Na⁺-free CO₂/HCO₃⁻ solution. The current amplitudes were similar in both solutions, indicating that HCO₃⁻ has negligible effect on I_Cl under the Na⁺-free condition. Figure 2C is a comparison of the mean I_Cl between groups in these two solutions from other oocytes (n = 5 NBCe1 and 10 D555E). No significant difference was observed within groups. In the second set of experiments, we induced I_Cl in Na⁺-free solution and tested whether the induced I_Cl could remain after CO₂/HCO₃⁻ application under the continued Na⁺-free condition. As shown in Figure 2D,E, whereas NBCe1 had no I_Cl, D555E produced I_Cl under the Na⁺-free condition, regardless of bath CO₂/HCO₃⁻. A slight decrease after CO₂/HCO₃⁻ application is probably due to Cl⁻ mismatch between solutions. Consistent with this result, comparison of mean I_Cl (n = 5/group), before and after CO₂/HCO₃⁻ application, resulted in no significant difference (Figure 2F). Conclusively, the results from the two sets of experiments demonstrate D555E preference to Cl⁻ over HCO₃⁻ in the absence of Na⁺, implying that Na⁺ is required for HCO₃⁻ to access its binding site.

2.3. Lack of I_Cl in the TM5-Replaced Chimeric Transporter

D555E is charge-conserved but has different geometry of the carboxyl group in the side chain due to an additional carbon backbone. This led us to postulate that Glu⁵⁵⁵ in D555E interacts with a nearby residue which results in a gain of function to select Cl⁻. To investigate this possibility, we replaced NBCe1/TM5 with NBCn1/TM5, which contains a Glu at the corresponding site of Asp⁵⁵⁵, and measured I_Cl in the chimeric transporter. First, we determined the functionality of the chimeric transporter by simultaneous recording of pH_i and I_NBC in voltage clamp (Figure 3A,B). In oocytes expressing NBCe1, the pH_i initially decreased upon CO₂/HCO₃⁻ application, due to CO₂ influx followed by H⁺ accumulation from hydration (Figure 3A). The pH_i was then recovered from an acidification (arrow) as HCO₃⁻ is continuously transported into the oocyte by NBCe1 and associates with intracellular H⁺. Applying CO₂/HCO₃⁻ also elicited an outward I_NBC (arrowhead), consistent with an influx of a net negative charge, due to 1 Na⁺ and 2 HCO₃⁻ (or 1 CO₃²⁻). Figure 3B is a recording of pH_i and I_NBC, produced by the TM5-replaced chimeric transporter, subjected to the same experimental protocol. The chimeric transporter had a slower pH_i recovery rate (dpH/dt) and smaller I_NBC in CO₂/HCO₃⁻ solution than NBCe1. Consistent with this observation, mean dpH/dt and I_NBC from 5 oocytes per group were significantly decreased in the chimeric transporter (p < 0.01 for each; Figure 3C,D). Despite such decreases, the chimeric transporter is functional, as it recovers pH_i from an acidification and produces I_NBC. Next, we measured the I_Cl and I_NBC produced by the chimeric transporter. Interestingly, the chimeric transporter did not produce measurable I_Cl, while retaining I_NBC (Figure 3E; one of 9 oocytes expressing the chimeric transporter is shown). Consistent with this result, I–V relationships exhibited negligible change in curves before and after Cl⁻ application (p > 0.05, n = 5; Figure 3F). Figure 3G is the comparison of Cl⁻ conductance (G_Cl), calculated from the slope of the I_Cl–V relationship (i.e., difference in I–V curve before and after Cl⁻ application). G_Cl of the chimeric transporter was negligible.

2.4. I_Cl Induced by Lys⁵⁵⁸ Replacement in the TM5 Chimeric Transporter

The result of negligible I_Cl in the chimeric transporter indicates that Glu⁵⁵⁵ is not the sole residue for I_Cl and additional residues are involved. Those residues should be in TM5 because other TMs were unchanged in the chimeric transporter. Asp⁵⁵⁵ is a residue in the anion binding site S2 that includes Lys⁵⁵⁸, Lys⁵⁵⁹, and Lys⁵⁶² (Figure 4A). The chimeric transporter contains Glu⁵⁵⁵, Glu⁵⁵⁸, Lys⁵⁵⁹, and Asp⁵⁶² at the corresponding sites (Figure 4B), suggesting that residues at position 558 and 562 would be responsible for I_Cl. To test this possibility, we changed Glu⁵⁵⁸ and Asp⁵⁶², individually or together, in the chimeric transporter with a Lys and tested their ability to produce I_Cl (n = 4–5/group). Figure 4C shows the I–V relationships for the chimeric transporter without mutation. As expected, no significant difference was observed in the I–V curves before and after Cl⁻ application (red line in the figure). In contrast, replacing Glu⁵⁵⁸ with a Lys (E558K) increased the slope in the outward direction upon Cl⁻ application (Figure 4D). Replacing Asp⁵⁶² with a Lys (D562K) had no effect (Figure 4E) and displayed similar I–V curves as the chimeric transporter. Consistent with these results, replacing both Glu⁵⁵⁸ and Asp⁵⁶² with Lys (E558K/D562K) increased the slope in the outward direction upon Cl⁻ application (Figure 4F). Thus, Lys⁵⁵⁸ is responsible for producing I_Cl.

Next, we compared I_Cl and I_NBC produced by the mutant transporters. The chimeric transporter without mutation had negligible I_Cl but produced measurable I_NBC (Figure 5A). A transitional undershoot after Cl⁻ washout is probably due to endogenous Cl⁻ efflux which often occurs in some preparations of oocytes. E558K and E558K/D562K produced I_Cl (Figure 5B,D), whereas D562K did not (Figure 5C). Both E558K and E558K/D562K showed higher I_Cl amplitudes than I_NBC amplitudes, the reason of which is unclear. Comparison of mean I_Cl from 5–6 oocytes per group is summarized in Figure 5E. A significant amount of I_Cl was produced when a Lys was present at position 558 (p < 0.01, one-way ANOVA). We also compared mean I_NBC between groups to evaluate the effect of Lys mutations on Na/HCO₃ cotransport and found a decrease in I_NBC by the mutations (p < 0.01, one-way ANOVA; Figure 5F). Thus, positively charged Lys residues in site S2 appear to have negative effects on the transporter activity.

2.5. Salt Bridge between Glu⁵⁵⁵ and Lys⁵⁵⁸

The identification of Lys⁵⁵⁸ for I_Cl leads to the possibility of a charge interaction between Glu⁵⁵⁵ and Lys⁵⁵⁸. To test whether a salt bridge stability is involved, we compared I_Cl produced by E558K in solutions containing either low or high ionic strength. The solution osmolarity was maintained using mannitol. The chimeric transporter displayed negligible response to 1–96 mM Cl⁻ in superfusing solutions, with the ionic strength of 0.005 and 0.1 mol/L (Figure 6A,C), consistent with its lack of I_Cl. In contrast, E558K produced I_Cl with progressively larger amplitudes at higher NaCl concentrations when measured in solutions with the ionic strength of 0.005 mol/L (Figure 6B) but had nearly negligible I_Cl, when measured in solutions with the ionic strength of 0.1 mol/L (Figure 6D). The result is consistent with the fact that a favorable salt bridge is diminished by a high ionic strength [21]. The decreasing effect by a high ionic strength was evident from the graph of I_Cl plotted as a function of Cl⁻ concentration (Figure 6E). The result shows effective inhibition of E558K-mediated I_Cl by a high ionic strength (F_12,80 = 7.47, p < 0.01 for transporter x Cl⁻ concentration interaction, two-way ANOVA; n = 4–6/group).

2.6. Glu⁵⁵⁵–Lys⁵⁵⁸ Charge Interaction

To further determine the above salt bridge interaction, we analyzed the bond length between the carboxyl group in the side chain of Glu⁵⁵⁵ and the amino group of Lys⁵⁵⁸ using the structure editing function with Dunbrack rotamer library [22] built in ChimeraX. In NBCe1, the distance between the carboxyl group of Asp⁵⁵⁵ and the amino group of Lys558 was 5.63 Å (Figure 7A), higher than the maximum 4.0 Å required for a hydrogen bond [23]. However, in D555E, the bond length between Glu⁵⁵⁵ and Lys⁵⁵⁸ was 3.79 Å (Figure 7B). The lengths from Lys⁵⁵⁹ and Lys⁵⁶² were higher than 4.0 Å (data not shown). Thus, the bond length was consistent with a weak electrostatic interaction between Glu⁵⁵⁵ and Lys⁵⁵⁸, but neither Lys⁵⁵⁹ nor Lys⁵⁶². The importance of the Glu⁵⁵⁵/Lys⁵⁵⁸ interaction for I_Cl was further examined by replacing Glu⁵⁵⁵ and Lys⁵⁵⁸ residues with other amino acids and comparing their ability to produce I_Cl (Figure 7C). Replacing Glu⁵⁵⁵ with a neutral Asn or Gln (N–K and Q–K pairs in the figure) or Lys⁵⁵⁸ with an Asp (E–E) near completely abolished I_Cl. In contrast, replacing Lys⁵⁵⁸ with a positively charged Arg (E–R) retained measurable I_Cl. One-way ANOVA, with Sidak post-test, revealed a significant change in I_Cl for N–K, Q–K and E–E pairs compared to E–K and E–R pairs (F_5,24 = 25.49, p < 0.01; n = 4–7/group). Water-injected control showed no current.

3. Discussion

In this study, we examined the effects of Asp/Glu⁵⁵⁵ and other charged residues in the entrance anion binding site S2 on Cl⁻ selectivity and made the following observations. (i) Replacing Asp⁵⁵⁵ in NBCe1 with a charge-conserved Glu induces a permissiveness to Cl⁻ that is normally not a substrate. This replacement does not alter HCO₃⁻ selectivity as I_NBC is favorably produced when both HCO₃⁻ and Cl⁻ are present. (ii) Under the Na⁺-free condition, D555E produces I_Cl even if HCO₃⁻ is available in the bath. The reason is that the anion binding site is not occupied with HCO₃⁻ in this condition; as a result, Cl⁻ is accessible to the site. Thus, Na⁺ is required for HCO₃⁻ to access its binding site. (iii) The I_Cl induced by D555E is due to a charge interaction between Glu⁵⁵⁵ and Lys⁵⁵⁸. Other Lys residues in site S2 have negligible effects on Cl⁻ transport. Glu⁵⁵⁵ and Lys⁵⁵⁸ are not simultaneously present in any member of the SLC4A bicarbonate transporters, indicating that the high HCO₃⁻ selectivity in these transporters is maintained by avoiding a charge interaction between the two residues. This molecular feature is interesting as it is generally understood that electrostatic interactions contribute to protein structure and create a suitable environment for protein function such as enzyme catalysis, protein-ligand binding, thermal stability, and macromolecular assemblies [21,24,25]. In this sense, our study provides novel evidence that the anion selection in the bicarbonate transporters is established by avoiding a specific interaction between residues in the anion binding site, rather than maintaining such interaction.

The amino acid residues in the chimeric transporter we examined correspond to Asp⁵⁵⁵, Lys⁵⁵⁸, and Lys⁵⁶² in NBCe1, all of which constitute site S2 located near the entrance of the ion accessibility pathway. Site S2 also contains Lys⁵⁵⁹, a DIDS-interacting residue [12], but we did not examine this residue for Cl⁻ selectivity because it is conserved in all SLC4A transporters. The SLC4A transporters contain either Asp⁵⁵⁵ or Glu⁵⁵⁵ but maintain a high selectivity to HCO₃⁻; thus, a residue capable of interacting with these residues should not be conserved. The CryoEM of NBCe1 [11] shows that Asp⁵⁵⁵ and Lys⁵⁵⁸ are located to the protein center, while Lys⁵⁵⁹ and Lys⁵⁶² are positioned further away from the center. The bond length between Asp⁵⁵⁵ and Lys⁵⁵⁸ is higher than the maximum length required for a salt bridge to take place, but Glu⁵⁵⁵ substitution has decreased the length (Figure 7A,B). We have previously demonstrated that D555E produces a large conductance in response to NO₃⁻, which is structurally in a trigonal planar arrangement. The effective radius of NO₃⁻ is bigger than the molecular radius of Cl⁻ (1.89 Å for NO₃⁻ vs. 1.81 Å for Cl⁻), but D555E produces a larger NO₃⁻ current than I_Cl. It is, thus, likely that site S2 is molded to sterically distinguish HCO₃⁻ or CO₃²⁻ from other polyatomic anions in a trigonal planar arrangement. The charge interaction between Glu⁵⁵⁵ and Lys⁵⁵⁸ in D555E modifies this steric arrangement in a way that other structurally similar ionic compounds, including NO₃⁻, are allowed. The modified steric arrangement also allows Cl⁻ to access the site but, given its monatomic molecule and competition with HCO₃⁻ or CO₃²⁻, we think that a Cl⁻ leak occurs at one of the three coordinating residues for peripheral oxygen atoms of HCO₃⁻ or CO₃²⁻. This interpretation is consistent with the MD simulations that Lys⁵⁵⁸ and Lys⁵⁵⁹ are the closest coordinating residues of CO₃²⁻, determined from ion density maps and contact frequency analysis.

The results from our study provide new insights into the mechanism underlying ion transport in NBCe1, in addition to anion selectivity. The Glu⁵⁵⁵–Lys⁵⁵⁸ pair produces I_Cl in the absence of CO₂/HCO₃⁻ and I_NBC in the presence of CO₂/HCO₃⁻; that is, the presence or absence of I_Cl reflects whether the anion binding site is occupied with HCO₃⁻ or CO₃²⁻. I_Cl is produced in Na⁺-free CO₂/HCO₃⁻; thus, the binding site is not occupied in the absence of Na⁺, indicative of Na⁺ precondition prior to anion binding. Based on this interpretation, a model of the ion binding process can be made. In NBCe1, ion transport begins with a recruitment of Na⁺ to its binding site. The Na⁺ binding then allows HCO₃⁻ or CO₃²⁻ to access its anion binding site and as a result both ions are bound to the transporter. The steric arrangement of Asp⁵⁵⁵, Lys⁵⁵⁸, and Lys⁵⁵⁹ in site S2 is critical for distinguishing HCO₃⁻ or CO₃²⁻ from other anions. The same ion recruitment process also takes place in a mutant transporter containing the Glu⁵⁵⁵–Lys⁵⁵⁸ pair, such as D555E. However, the charge interaction between the two residues modifies the steric arrangement of residues in S2, such that other anions, such as Cl⁻, are permissive; as a result, Cl⁻ is accessible to the anion binding site. Our model proposes that Na⁺ binding is a necessary first step prior to anion binding and, thus, should be independent of external HCO₃⁻ or CO₃²⁻ levels. In this sense, it is interesting to note that the apparent affinity of NBCe1 for Na⁺ is independent of external HCO₃⁻ concentrations [26]. We think that the negatively charged Asp/Glu⁵⁵⁵ facilitates Na⁺ recruitment from the extracellular fluid surrounding the transporter. One might argue that Na⁺ should overcome an electrostatic repulsion from the positively charged Lys residues before reaching its binding site. Decreased I_NBC by E5558K and E558K/D562K (Figure 5F) could be accounted for by the electrostatic repulsions from Lys residues. On the other hand, Yamazaki et al. [27] have reported that K558R, a single nucleotide polymorphism in human NBCe1, has a significantly reduced transport activity but no change in apparent Na⁺ affinity. Either way, it is premature to conclude that Lys residues influence Na⁺ recruitment to the binding site.

The MD simulation model proposes that substrate ions transiently bind to site S2 and then move to site S1, which ultimately leads to a protein conformational change for ion translocation. The TM5-replaced chimeric transporter in this study contains NBCn1-S2 but still produces I_NBC, indicative of electrogenic cotransport. Thus, I_NBC can be induced, regardless of whether site S2 is molded for HCO₃⁻ transport in NBCn1 or HCO₃⁻ or CO₃²⁻ transport in NBCe1. Then, a question arises whether the charges in site S2 are critical for HCO₃⁻ or CO₃²⁻ recruitment. The chimeric transporter contains NBCe1-S1, indicating that the production of I_NBC is determined by the anion that occupies site S1. We think that, whereas site S2 allows a transient binding of HCO₃⁻ or CO₃²⁻, S1 determines which of the two anions is translocated via the transporter. Our interpretation further suggests that NBCn1-S2 can recruit CO₃²⁻, in addition to HCO₃⁻, although HCO₃⁻ is more favorably recruited. Nevertheless, it is difficult to envision how CO₃²⁻ is selected by both NBCn1-S2, which contains negatively charged residues, and NBCe1-S2, which contains positively charged residues. Additional studies are demanded to elucidate the role of site S2 in anion recruitment.

Does I_Cl induced by the Glu⁵⁵⁵–Lys⁵⁵⁸ pair represent a channel activity or transporter activity? If I_Cl is a channel activity, we should then observe a current in response to HCO₃⁻ (I_HCO3), comparable to I_Cl in response to Cl⁻. However, we did not observe I_HCO3 under the Na⁺-free condition. The lack of I_HCO3 under the Na⁺-free condition reflects that HCO₃⁻ movement via D555E is solely mediated by electrogenic Na/HCO₃ transport that generates I_NBC. The important finding is that I_Cl is significantly inhibited by electrogenic Na/HCO₃ transport (Figure 1B,C), indicating that I_Cl competes with I_NBC. Thus, I_Cl is associated with a transporter activity. As described above, we envision that D555E modified the HCO₃⁻ binding site to produce an anion leak. On the other hand, I_Cl can be produced without Na⁺, implicating a separate channel activity. This leads us to a conclusion that I_Cl is associated with both transporter activity and channel activity, and they overlap. It is difficult to envision how the two activities overlap, and additional studies are required to address the exact nature of I_Cl.

Lastly, our study leads us to a discussion about a pathological implication of Cl⁻ leak mediated by mutations in NBCe1. Cl⁻ and HCO₃⁻ movements tightly coordinated in many cells, and specific transporters and channels are involved in regulating such coordination [28,29,30,31]. Obviously, Cl⁻ leak is undesirable in cells and tissues where NBCe1 is highly expressed and regulates HCO₃⁻ transport for cellular and physiological function. Myers et al. [32,33] have reported that Q913R, a mutation identified from a patient with proximal renal tubular acidosis, causes intracellular retention of NBCe1 and a gain of function activity in Cl⁻ leak. It is expected that this mutation causes a depolarization in the basolateral membrane of renal proximal tubules; as a result, the driving force for HCO₃⁻ reabsorption is decreased. A Cl⁻ leak via the mutation is also expected to alter the coupling of Cl⁻ and HCO₃⁻ movement observed in secretory epithelia, such as pancreas and salivary glands [34]. Another mutation of interest is K558R that has a reduced transport activity [27]. Our analysis of the bond length between Asp⁵⁵⁵ and Arg⁵⁵⁸ in K558R is less than 4 Å (3.82 Å with the probability of 0.1 and 3.5 Å with the probability of 0.05), implicating a salt bridge between the two residues. It will be interesting to examine whether this mutation can cause Cl⁻ leak. Additionally, depending upon NBCe1 variants, intracellular Cl⁻ can regulate the transporter activity [35]. Thus, the lack of Cl⁻ leak in NBCe1 is beneficial for cellular HCO₃⁻ homeostasis and epithelial electrolyte secretion.

In summary, by analyzing the TM5 chimeric transporter and relevant point mutants, we identified a charge interaction in site S2 as a key factor for anion selectivity and provided new insights into CO₃²⁻ or HCO₃⁻ recruitment to the binding site and ion binding sequence. Future studies will be of the molecular mechanism underlying ion selectivity and translocation in other Na/HCO₃ transporters.

4. Materials and Methods

4.1. TM5-Replaced Chimeric Transporter and Point Mutants

D555E made on human NBCe1-A (Genbank accession number: NM_003759; hereafter, NBCe1) was described previously [17]. The TM5-replaced chimeric transporter was constructed by creating restriction enzyme sites at the TM5 boundaries in NBCe1 and NBCn1. Point mutant transporters were constructed using the QuickChange Site-directed Mutagenesis kit (Agilent Technologies, Santa Clara, CA, USA). Primers were designed to replace nucleotides encoding candidate amino acids (Supplementary Table S1). PCR was carried out 95 °C for 1 min, 55 °C for 30 s, and 68 °C for 10 min for 16 cycles, and an additional 2 min per nucleotide substitution were included for extension at 68 °C. All constructs were sequenced.

4.2. Protein Expression in Xenopus Oocytes

Xenopus laevis oocytes, at stages V-VI, were purchased from Ecocyte Bioscience (Austin, Texas, USA). For cRNA synthesis, plasmids containing NBCe1 or mutant transporter DNAs were linearized and transcribed using the mMessage/mMachine transcription kit (Life Technologies, Grand Island, NY, USA). Transcribed RNAs (15–25 ng in 46 nL) were injected per oocyte. Equal amounts of RNAs were used when multiple samples were compared. Controls were water injection. Oocytes were maintained for 3–4 days at 18 °C before use.

4.3. Two-Electrode Voltage Clamp

An oocyte was placed in the recording chamber containing ND96 solution (in mM; 96 NaCl, 2 KCl, 1.8 CaCl₂, 1 MgCl₂, and 10 HEPES, pH 7.4) and impaled with two borosilicate glass electrodes filled with 3 M KCl. The tip resistance was 0.5–2 MΩ. After stabilization of the resting membrane potential, the oocyte was clamped at –60 or 0 mV using the voltage-clamp amplifier OC-725C (Warner Instrument, Hamden, CT, USA). For recording I_Cl, an oocyte was superfused with Cl⁻-free solution, which replaced all NaCl in ND96 solution with Na/gluconate, and then with 71 mM Cl⁻ solution, which replaced 25 mM NaCl with the same amounts of Na/gluconate. A small amount of Cl⁻ (<3 mM) was included to minimize junction potential. For recording I_NBC, 25 mM NaCl was replaced with the same amounts of NaHCO₃ equilibrated with 5% CO₂. Recording I_Cl in CO₂/HCO₃⁻ solution was achieved after I_NBC reached steady-state. Na⁺-free solutions were made by substituting Na⁺ with N-methyl-D-glucamine NMDG. Current-voltage (I–V) relationships were determined by a staircase voltage command between −120 to +60 mV, with 20 mV increments for 100 ms duration. The voltage command was applied immediately after a current reached steady state. Voltage signals were sampled by Digidata 1322A (Molecular Devices; San Jose, CA, USA) and data were acquired using pClamp 10 (Molecular Devices). Signals were filtered using an 8-pole low pass Bessel filter, with a cutoff frequency of 0.1–1 Hz. Recordings were made at room temperature.

4.4. Measurement of Intracellular pH (pH_i)

Oocyte pH_i was measured using a proton-selective glass electrode, as described before [36]. Briefly, a pH electrode was made with a borosilicate glass capillary that was silanized, filled with the proton ionophore 1 cocktail B (cat no: 95293, MilliporeSigma, St. louis, MOP, USA), and back-filled with pH 7.0 phosphate buffer. The pH electrode was connected to high impedance electrometer FD223 (World Precision Instruments; Sarasota, FL, USA) and electrode signals were amplified using a custom-made subtraction amplifier. Current and voltage electrodes were filled with 3 M KCl (resistance of 0.5–2 MΩ) and connected to an OC-725C amplifier. Signals from pH, current, and voltage electrodes were collected using Digidata 1322A. The voltage electrode signal was subtracted from the pH electrode signal using pClamp 10. The voltage/pH slope was calibrated by placing electrodes in the chamber filled with pH 6.0 and 8.0 standards. Slopes were typically at the range of 53 ± 3 mV/pH. An oocyte, in the recording chamber containing ND96 solution, was impaled with pH, voltage, and current electrodes and clamped at 0 mV. Once pH and base line current were stabilized, solutions were switched to 5% CO₂, 25 mM HCO₃⁻ (pH 7.4). The rate of pH change (dpH/dt) was determined by drawing a line during the first 4 min of recovery from CO₂-induced acidification.

4.5. Salt Bridge Experiment

For assessment of salt bridges, an oocyte expressing the mutant transporters was clamped at 0 mV and superfused with 96 mM Na/gluconate (plus 5 mM mannitol) or 197 mM mannitol, plus a small amount of chloride (<3 mM), until base line currents became stable. Then, a series of test solutions containing 1, 10, 20, 40, and 96 mM of NaCl were applied. NaCl in each test solution replaced the equivalent amount of mannitol or Na/gluconate. Each test solution was bracketed with NaCl-free solution to maintain steady-state baseline between test solutions. The ionic strength (I) was determined using the equation:

$$\mathrm{I}=\frac{1}{2}{{\displaystyle \sum }}_{i=1}^n\mathrm{C}i\mathrm{Z}{i}^2,$$

where Ci is the molar concentration of ion i (mol/L), and Zi is the charge number of that ion.

4.6. Analysis of Charge Interaction in Site S2

Analysis of the binding site S2 was performed with CryoEM structure of the human NBCe1 (accession code: 6CAA) from the RCSB Protein Data Bank using the molecular visualization program UCSF ChimeraX 1.1 [37]. A hydrogen bond between the side chain carboxy group of Asp⁵⁵⁵ and amino group of nearby Lys residues was identified when the distance between them was <4 Å. For D555E or the TM5-replaced chimeric transporter, amino acid changes were analyzed using the structure editing function with Dunbrack rotamer library [22] built in ChimeraX. A hydrogen bond was identified from the rotamer probability of higher than 0.05.

4.7. Statistical Analysis

Data were reported as mean ± standard error. The level of significance was determined using (i) unpaired, two-tailed Student t-test for comparison between NBCe1 and D555E or chimeric protein; (ii) paired, one-tailed test for comparison of single transporters in two different solutions; (iii) one-way ANOVA for comparison of I_Cl or I_NBC among multiple mutants; and (iv) two-way ANOVA for comparison between I_Cl vs. I_NBC among multiple mutants. The p value of less than 0.05 was considered significant. Data were analyzed using Prism 7 (GraphPad; La Jolla, CA, USA) and Microsoft Office Excel add-in Analysis ToolPak (Redmond, WA, USA).