Molecular determinants of HCO₃^- and cation transport in the human cation-dependent Cl^-/HCO₃^- exchanger AE4

Abstract

The HCO₃^- transporter AE4 (SLC4A9) plays a role in NaCl reabsorption and pH sensing in the kidney, and Cl^--dependent fluid secretion in salivary glands. Sharing functional features with other Cl^-/HCO₃^- exchangers and Na⁺-HCO₃^- co-transporters, it has been proposed that AE4 mediates Cl^-/cation- HCO₃^- exchange. Our sequence alignments and molecular dynamics (MD) analysis showed that three residues, reported as critical for transport activity in other SLC4 transporters, are conserved in AE4, suggesting similarities in their ion transport mechanism. Site-directed mutagenesis and further functional experiments showed that two out of the three conserved residues (D709 and T448) are functionally relevant but in contrast to other SLC4 transporters, where transport was almost completely abolished, AE4 mutants conserved about 50% of transport activity. Additionally, alanine scanning showed that S446A and T756A decreased transport by nearly 30%. Consistent with an additive effect of mutations at positions T756 and T448, the double mutant T756A-T448I completely abolished transport in the presence of extracellular Na⁺, but interestingly, exhibited anion transporter activity in the presence of K⁺ as the main extracellular cation. MD simulations revealed that the HCO₃^- and cation coordination site is at the interface between the transmembrane segments TM3-TM10. The interaction network was importantly disrupted in the double mutant in the presence of Na⁺, but it is partially conserved in the presence of K⁺, suggesting differences in the cation coordination. In summary, we identified the putative cation coordination site of AE4 and the critical functional role of residues T756 and T448 in its transport cycle.

Graphical Abstract

NEW & NOTEWORTHY

AE4 is a HCO₃^- transporter that is important for electrolyte transport and pH regulation in epithelia, which has been proposed to mediate Cl^-/cation-HCO₃^- exchange. However, critical residues sustaining transport activity are not known. In this study, we show that AE4 can coordinate HCO₃^- and Na⁺ or K⁺ at the TM3-TM10 interface, and that residues T448 and T756 are crucial for cation binding and transport cycle.

INTRODUCTION

The SLC4 (Solute Carrier 4) family includes the product of ten genes encoding HCO₃^- transport proteins which play important roles in physiology, such as acid-base homeostasis, intracellular pH regulation and HCO₃^-, Cl^- and CO₂ transport in epithelia (1–3). Dysfunction of SLC4 transporters has been linked to various human diseases, such as renal tubular acidosis, hypertension, hemolytic anemia, and cancer (4–10). SLC4 transporters are divided into two main groups: Na⁺-independent Cl^-/HCO₃^- exchangers, including AE1, AE2, and AE3 (SLC4A1-3), and Na⁺-dependent HCO₃^- (or CO₃^2-) transporters, such as NBCe1 (SLC4A4), NBCe2 (SLC4A5), NBCn1 (SLC4A7), NBCn2 (SLC4A10), and the Na⁺-coupled Cl^-/CO₃^2- exchanger NDCBE (SLC4A8). However, the functional role of AE4 (SLC4A9) remains controversial, with conflicting evidence suggesting it may act as either a Cl^-/HCO₃^- exchanger or a Na⁺-HCO₃^- co-transporter (11–15).

AE4 has been implicated in NaCl reabsorption and acid-base sensing in the kidney, as well as saliva secretion in salivary glands, based on studies using knockout mouse models (11, 16, 17). Additionally, AE4 has been proposed to play a role in pH regulation in cancer cells (18). Our previous work demonstrated that AE4 functions as an electroneutral Cl^-/HCO₃^- exchanger that transports monovalent cations, including Na⁺ and K⁺ (19). Accordingly, we propose that AE4 is an electroneutral Cl^-/cation-HCO₃^- exchanger that shares functional features with the AE1 (SLC4A1) Cl^-/HCO₃^- exchanger and the NBCe1 (SLC4A4) Na⁺-CO₃^2- cotransporter.

Structurally, SLC4 proteins form homodimers, with each monomer consisting of 14 transmembrane segments (TMs). TM3 and TM10, which form short α-helices partially crossing the membrane, create a narrow cavity proposed to contain an ion-binding site (20–24). Despite advances in understanding the structure and function of other SLC4 transporters, the specific residues critical for AE4's transport function remain poorly characterized. In this study, we identify key amino acids within the TM3-TM10 region that are essential for ion coordination and transport activity in AE4. Our findings provide new insights into the molecular mechanisms underlying AE4's function and its unique cation-dependent transport properties.

MATERIAL AND METHODS

Plasmids.

Homo sapiens AE4 transcript variant 2 (GenBank Accession no. NM_031467) plasmids were obtained from two different sources: Origene Technologies Inc. (pCMV6-Entry, clone ID: RC212283) and Genscript USA Inc. (pcDNA3.1+(C-(K)-DYK, clone ID: OHu19533). All AE4 versions were C-terminally DDK-tagged.

Site directed mutagenesis.

Mutants S446A, S447A, T448G, D709A, S754A, T756A, I758G and double mutants were obtained from Genscript USA Inc. Mutants T448I and D709N were done on the wild-type version (Clone ID: RC212283, Origene) using the QuickChange Site Directed Mutagenesis Kit (Agilent Technologies), according to the manufacturer's instructions. All the plasmids encoding for all the mutants used in this study were confirmed by Sanger sequencing.

Cell culture and transfections.

HEK-293 cells were grown in Petri dishes containing DMEM-F12 medium (Invitrogen) supplemented with 10% (v/v) fetal bovine serum (Invitrogen) and 1% penicillin-streptomycin. Cells were maintained at 37^oC in a humidified 5% CO₂/95% air atmosphere. For transfections, cells were grown at 90% confluence in 24 well plastic plates (Corning). Transient transfections were performed using 1 µL of Lipofectamine 2000 (Invitrogen) and 1 µg of plasmid containing the wild-type or mutant versions of AE4 cDNA per well. Experiments were done 18–24 h post-transfection.

pHi measurements.

The fluorescent dye BCECF-AM (2′,7′-bis-(2-carboxyethyl)-5-(and-6)-carboxyfluorescein, acetoxymethyl ester (Invitrogen) was used to measure pHi. Cells were loaded with 2 µM BCECF-AM for 10 min at 37^oC. Imaging was performed using an inverted microscope (Nikon Eclipse TE300) equipped with a Lambda 10–2 Filter Wheel Imaging System (Sutter Instruments) coupled to a digital camera (Retiga R1, QImaging). A Plan Fluor 40X, NA 0.75, objective was used (Nikon). Images were acquired by alternate excitation at 490 and 440 nm. Emission was captured at 510 nm using Micro-Manager 1.4 microscopy software (25). The intracellular pH changes are reported as the normalized fluorescence ratios (R/R₀). Bath solutions used in this study are listed in table 1. HCO₃^--containing solutions were gassed with 95% O₂, 5% CO₂ for at least 20 min. All experiments were performed at room temperature. Data are presented as means ± SEM. Figures are representative of at least three cells per experiment, derived from a minimum of three independent transfections for AE4-expressing cells. Data from non-transfected cells are representative of at least four independent experiments. Regions of interest (ROIs) were positioned on cells that exhibited a response in the case of AE4-transfected cells, and on any cell in the case of non-transfected cells.

Table 1. Solutions used in this study.

Solutions were continuously gassed with 5% CO₂/95% O₂.

Component	Solution Name
(mM)	A	B	C	D	E	F
Na+	145	145
K+	4.3	4.3	4.3	4.3	145	145
NMDG+			120	120
Ca2+	1	1	1	1	1	1
Mg2+	1	1	1	1	1	1
Choline			25	25
Cl−	128.3	4	128.3	4	124	4
HCO3−	25	25	25	25	25	25
Gluconate		124.3				120
Glutamate				124.3
Glucose	5	5	5	5	5	5
HEPES	10	10	10	10	10	10
pH	7.4	7.4	7.4	7.4	7.4	7.4

Immunoblot.

Cells were lysed for 30 min on ice in modified RIPA lysis buffer (1% Nonidet P-40, 0.1% SDS, 0.1% sodium deoxycholate, 100 mM NaCl, 2.5 mM EDTA, 2.5 mM EGTA, 50 mM HEPES, pH 7.4), containing protease and phosphatase inhibitors. 80 μg of each sample were electrophoresed using gradient 4–20% express Plus Page Gels (Gen Script), transferred to a nitrocellulose membrane and probed with 9A3 anti FLAG antibody (Cell Signaling Technology, Beverly, MA, USA) 4°C overnight. Antibody binding was detected with corresponding HRP conjugated secondary antibody (Pierce, Rockford, IL, USA) for 2 hours, followed by chemiluminescence detection with Super Signal West Pico PLUS (Thermo Scientific) and visualized using a Syngene G: BOX digital imaging system.

Sequence alignment and logos.

Amino acid sequence alignments were done using Clustal Omega (26). The following sequences from GeneBank were used: AE1 (SLC4A1): NP_000333.1; AE2 (SLC4A2): NP_001186621.1, NP_001186623.1, NP_001186622.1; AE3 (SLC4A3): NP_005061.3; AE4 (SLC4A9): NP_113655.2; BTR1 (SLC4A11): NP_114423.1, NP_001167561.1, NP_001167560.1; NBCe1 (SLC4A4): AAC51645.1, NP_001091954.1, ABQ43327.1; NBCe2 (SLC4A5): NP_067019.3, AAK97072.1, AAK97073.1, AAL48291.1, AAL50802.1; NBCn1 (SLC4A7): NP_003606.3, ACH61962.1, ACH61960.1, ACH61961.1, NP_001245309.1, NP_001245308.1, ACH61958.1; NDCBE (SLC4A8): AAY79176.1, ABJ09587.1, ABJ91577.1, BAA34459.1; NBCn2 (SLC4A10): NP_071341.2, AAQ83632.1. Sequence logos were done using the WebLogo application (27).

Molecular modeling and dynamics simulations.

The homology model was generated using the SWISS-MODEL platform (28), with the wild-type human AE4 isoform 2 sequence (NP_113655.2) as the target and the rat Slc4a8 cryo-EM structure (PDB code: 7rtm) as the template (22). This model was used initially to identify putative important residues at the TM3-TM10 interphase. To identify the ion coordination site of AE4, additional models of wild-type and the double mutant T756A-T448I were built. Ion pairs Na⁺ and HCO₃^- and K⁺ and HCO₃^- were added to the putative S1 coordination site (based on similarity with Slc4a8 S1 site). Four models were obtained: wild-type AE4 with Na⁺/HCO₃^-, double mutant T756A-T448I with Na⁺/HCO₃^-, wild-type AE4 with K⁺/HCO₃^- and double mutant T756A-T448I with K⁺/HCO₃^-. All models were subjected to molecular dynamics (MD) simulations for 100 ns each, executed using Desmond v2019-1 software (Bowers K., Chow E., Xu H., Dror R.O., Eastwood M.P. Scalable Algorithms for Molecular Dynamics Simulations on Commodity Clusters; Proceedings of the 2006 ACM/IEEE Conference on Supercomputing; Tampa, FL, USA. 11–17 November 2006; pp. 7695–7700) and OPLS2005 force field (29–31). The Slc4a9 structures were prepared using Protein Preparation Wizard, Maestro suite (32). The systems were embedded into a pre-equilibrated POPC (1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine) bilayer membrane model and solvated using the SPC (single point charge) water model. To neutralize the systems, Na⁺/Cl^- ions were included; NaCl was added to a concentration of 0.15 M. The systems were simulated for 20 ns using NPT ensemble and applying positional restraints to ions and proteins. Positional restraints of 1.0 kcal*mol-1 *Å-2 were applied to the protein and to the ions placed in coordination sites (Na⁺, K⁺ and HCO₃^-). Temperature and pressure were kept constant at 300 K and 1.01325 bar respectively, by coupling to a Nose-Hoover Chain thermostat (33) and Martyna-Tobias-Klein barostat (34). The integration step was set to 2 fs. For MD productive runs, the positional restraints were removed. Each run was performed during 100 ns using an NPγT (semi-isotropic ensemble) with a constant surface tension of 0.0 bar Å. The root-mean-square deviation (RMSD) values of the backbone atoms were calculated for each trajectory. To analyze ion-residue interactions, the distances and the hydrogen bonds were computed. We also computed hydrogen bond occupancy, defined as the percentage of frames in which a particular hydrogen bond occurs. All calculations were performed using VMD v1.9.4a38 (35).

RESULTS

Model of the transmembrane region of AE4

Considering the high sequence identity in the transmembrane region among members of the SLC4 family, we anticipate that the overall folding of human AE4 (hAE4) is likely similar to that of other SLC4 proteins. We initially focused on transmembrane helices 3, 8, and 10, which have been reported to play key roles in ion transport for several SLC4 proteins. Functional studies have identified positions S465, E681, and R730 (AE1 numbering) in this region as critical for transport activity (36–38). Interestingly, these residues appear to be a source of functional divergence between Na⁺-dependent and Na⁺-independent SLC4 transporters. Further structural studies have demonstrated that these residues are also directly involved in ion coordination at the TM3-TM10 interface (20–24, 39, 40).

To gain deeper insights into this region, we performed multiple sequence alignments (MSA) of the human versions of AE1, NBCe1, NDCBE, and AE4. The alignment revealed that residues S465 and E681 in AE1 are substituted by the homologous T448 and D709 in AE4, while R730 in AE1 is replaced by I758 (Figure 1A). Notably, these three residues in AE4 (T448, D709, and I758) are strictly conserved in NBCe1 as T485, D754, and I803, respectively, and correspond to T538, D800, and L849 in NDCBE. We then compared these residues across all human SLC4 transporters (Figure 1B). Threonine and serine are the most common amino acids at position T448 (hAE4 numbering), suggesting that the presence of a hydroxyl group at this site is essential for the proper function of SLC4 transporters (Figure 1B, left panel). Similarly, the position corresponding to D709 in AE4 is consistently occupied by either aspartate (D) or glutamate (E), indicating that a negative charge at this position is functionally important (Figure 1B, middle panel). In contrast, the residue at the equivalent position of I758 in AE4 is highly variable, with either hydrophobic (leucine or isoleucine) or positively charged (arginine or histidine) side chains. This variability suggests that this position (I758 in AE4) may contribute to functional differences among SLC4 HCO₃^- transporters (Figure 1B, right panel). Overall, our sequence analysis indicates that key residues involved in transport in other SLC4 proteins are conserved in AE4. However, since AE4 appears to mediate both Cl^-/HCO₃^- exchange and Na⁺-HCO₃^- co-transport, we speculate that the contribution of residues T448, D709, and I758 to HCO₃^- transport may be influenced by AE4-specific residues, differences in structural organization, or variations in the relative positioning of side chains at the substrate binding site.

Figure 1: Sequence analysis of transmembrane helices 3, 8 and 10 in SLC4 transporters.

Amino acid sequence alignments were done with Clustal Omega. (A) Positions of functionally important residues are shown with green boxes. S465, E681 and R730 in AE1 are substituted by T, D and I respectively in AE4 and NBCe1. R730 in AE1 is substituted by L in NDCBE. These residues are positioned in the putative ion binding pocket at transmembrane regions (TMs) 3, 8 and 10. (B) Sequence logos show the amino acidic conservation in the mentioned regions of 30 isoforms of human SLC4 transporters. Residues T448, D709 and I758 of AE4 are indicated in yellow boxes.

To obtain structural information of the TM3-TM10 interface in AE4, and the positions of T448, D709 and I758, we built a homology model using the structure of the cation-dependent Cl^-/CO₃^2- exchanger NDCBE (Slc4a8) as a template (22) (PDB code: 7rtm). The model shows 14 transmembrane segments where the N-terminus of TM1 starts with residue Q387 and the C-terminus of TM14 ends at residue I885. Both N- and C-terminus are oriented to the intracellular side and the short half helices TM3 and TM10 face each other at the middle of the membrane plane (Figure 2 A, left). In agreement with the outward-open conformation structure of NDCBE (22), the larger water-accessible pathway in our model is facing the extracellular side. Moreover, the ion accessibility pathway is lined by TM3 and TM10 in agreement with the idea that both helices play an important role in ion binding and translocation (Figure 2 A, right). The residues T448, D709 and I758 are found at TMs 3, 8 and 10 respectively (Figure 2 B), as previously reported for homologous residues in AE1, NBCe1, NDCBE and AE2 (20–23).

Figure 2: Homology model of AE4 in an outward-open conformation.

(A) The NDCBE structure (PDB ID Code 7rtm) was used as template to build a homology model of AE4. A. The side view shows 14 transmembrane segments indicated by numbers (left panel). The N and C terminus are oriented to the intracellular side. Extracellular loops 5–6 and 7–8 were removed for clarity. The putative ion binding site is lined by transmembrane segments 1, 3, 5, 8 and 10 as shown in a view from the extracellular side (right panel). (B) Residues T448, D709 and I758 are present in the putative ion binding site where TMs 3 and 10 are half-helices and meet in the center of the structure.

Relevance of T448, D709 and I758 residues in the transport

To further investigate the functional role of residues T448, D709, and I758 in the AE4 transport cycle, we conducted site-directed mutagenesis and pH imaging in HEK-293 cells transiently expressing the human AE4 (hAE4). The expression of AE4 was assessed by immunolocalization in the cell membrane and immunoblot (Figure 3 A). We previously proposed that AE4 mediates electroneutral Cl^- uptake in exchange for cation-HCO₃^- under physiological ion concentrations, with AE4 transporting Na⁺ and K⁺ ions with similar affinities (17). Given that the combined concentrations of Na⁺ and K⁺ are comparable in the cytosol and the extracellular space, the cation-dependent driving force should be negligible ([K⁺]_o + [Na⁺]_o = [K⁺]ᵢ + [Na⁺]_i). Therefore, we first characterized the Cl^-/HCO₃^- exchange activity of wild-type AE4 by measuring intracellular pH changes in AE4-expressing cells upon imposing an outward Cl^- gradient, achieved by switching from high-Cl^- to low-Cl^- extracellular solutions. Consistently, switching extracellular Cl^- from 128.3 mM to 4 mM induced a robust alkalinization, indicating HCO₃^- uptake mediated by Cl^-/HCO₃^- exchange. This response was absent in non-transfected HEK-293 cells. The intracellular pH returned to baseline when cells were re-exposed to a high-Cl^- extracellular solution (Figure 3B, black symbols). Alkalinization was observed only in transfected cells perfused with Na⁺-containing solutions (Figure 3C, black symbols). When extracellular Na⁺ was isotonically replaced with N-methyl-D-glucamine, reducing external Cl^- did not elicit intracellular alkalinization. These findings align with the Na⁺ dependence of AE4 transport observed in heterologous expression systems (19).

Figure 3. Functional characterization of wild-type AE4 in HEK-293 cells.

(A) Immunodetection of wild-type AE4 in the plasma membrane of transiently transfected HEK-293 cells and immunoblot showing the expression only in transfected cells (panels left and right respectively). (B) Cells HEK-293 expressing the human version of AE4 were loaded with BCECF-AM and intracellular pH measurements were performed. The intracellular pH changes are reported as the normalized fluorescence 490/440 ratios (R/R₀). A rapid intracellular alkalinization was observed in response to HCO₃^--containing low Cl^- (4 mM) bathing solutions (solution B). The intracellular pH was recovered when cells were re-exposed to a high Cl^- (128.3 mM) bath solution (solution A) (black symbols, n = 17). This response was absent in non-transfected cells (open symbols, n = 16). (C) The intracellular alkalinization was absent in cells expressing AE4 in response to HCO₃^--containing low Cl^- bathing solutions where Na⁺ was replaced by NMDG⁺ (solutions C to D). A rapid alkalinization was recorded switching to a Na⁺-containing bath solution (solutions B) (black symbols, n = 6). This response was absent in non-transfected cells (open symbols, n = 5). Data are shown as means ± SEM. Figures are representative of at least 3 cells per experiment from at least 3 independent transfections in the case of AE4-expressing cells (black symbols). Data from non-transfected cells are representative of at least 4 independent experiments (open symbols).

Next, we employed site-directed mutagenesis to assess the transport activity of different AE4 variants. Our mutagenesis strategy was guided by previously reported functional data, focusing on the TM3-TM10 region in other SLC4 proteins. Transport activity was evaluated by comparing the initial alkalinization rates, determined from linear regressions fitted to the experimental data (Figure 4 A). First, we substituted residue T448, located at TM3, with isoleucine. A similar mutation at the homologous S465 in AE1 has previously been reported to abolish Cl^-/HCO₃^- exchange (36). We found that T448I reduced Cl^-/HCO₃^- exchange activity by ~50% compared to the wild-type, indicating that this position is functionally important in AE4 as well (Figure 4 A and B, blue symbols).

Figure 4. Functional contribution of residues in the TM3-TM10 region.

Putative functionally important residues at TM3-TM10 region were mutated and functional characterization was performed as described in Figure 3. (A) Cl^-/HCO₃^- exchange activity was initiated by switching from high-Cl^- to low-Cl^- bathing solutions (solution A to B), leading to a rapid intracellular alkalinization. Transport activity decreased by ~30% in the D709A mutant (cyan circles; n = 19), while T448I (blue circles; n = 17) and T448G (gray circles; n = 10) mutants exhibited a ~50% reduction compared to wild-type AE4 (black circles; n = 17). In contrast, mutants I758G (orange circles; n = 11), I758R (light green circles; n = 10), and D709N (dark green circles; n = 12) showed no significant change in activity. Non-transfected HEK-293 cells (open circles; n = 16) exhibited negligible response. (B) Summary of the initial alkalinization rates of experiments shown in A. (C) The S446A (green circles; n = 8) and T756A (cyan circles; n = 9) mutants exhibited a ~30% reduction in transport activity, whereas S447A (orange circles; n = 9) and T754A (light green circles; n = 9) mutants showed no significant changes. Data from wild-type AE4 and non-transfected HEK-293 cells (previously shown in panel A) were included for comparison. (D) Summary of the initial alkalinization rates of experiments shown in C. Rates were calculated by linear regression analysis from regions indicated with dashed red lines in individual experiments. Data are shown as means ± SEM. Figures are representative of at least 3 cells per experiment from at least 3 independent transfections. Statistical analysis was performed using one-way ANOVA followed by Bonferronís post hoc test. Significant differences: a, compared to non-transfected HEK-293 cells; b, compared to wild-type AE4 (p < 0.001). All data are presented as means ± SEM (error bars).

Replacing T448 by isoleucine eliminates the hydroxyl group, suggesting that the oxygen or hydrogen atoms might participate in polar interactions with ions or other side chains, forming part of the molecular network required for the transport cycle. However, isoleucine has a slightly bulkier side chain than threonine, raising the possibility that steric effects may also contribute to the observed reduction in activity. To test this alternative hypothesis, we mutated T448 to glycine (T448G), which removes the hydroxyl group while minimizing steric hindrance and interestingly, this mutation also reduced the activity by ~50%, similar to the effect observed in T448I (Figure 4 A and B, gray symbols). This result suggests that the reduced activity observed in mutants can be explained by the neutralization of de polar hydroxyl group of T448 and a steric effect of the lateral chain is not involved.

Next, position 709 in AE4 is occupied by an aspartate, which is conserved among Na⁺-dependent SLC4 HCO₃^- transporters and replaced by glutamate in Na⁺-independent Cl^-/HCO₃^- exchangers (2, 39). The negative charge of aspartate or glutamate has been proposed as a key determinant for ion-transport function (21, 22). To test this, we neutralized D709 by mutating it to asparagine (D709N), a substitution that has been shown to drastically reduce activity in AE1 and NBCe1 (~80 and ~90% reduction, respectively) (41, 42).

Surprisingly, D709N mutation had no impact on HCO₃^- transport in AE4 (Figure 4 A and B, green symbols). We hypothesize that the polar nature of asparagine may be sufficient to support transport. To further investigate this, we substituted D709 with alanine (D709A), a small and non-polar residue. This mutation resulted in ~30% reduction in activity, indicating that, unlike in other SLC4 proteins, D709 is not critical for transport function in AE4 (Figure 4 A and B, cyan symbols). Finally, residue I758 in AE4 occupies a functionally important position in other SLC4 proteins and is conserved only in Na⁺-dependent SLC4 transporters. Our molecular model suggests that I758 might play a role in stabilizing the structure of the putative ion-binding site, likely through hydrophobic interactions with residues in TM1 and the TM3-TM10 interface. To test this hypothesis, we mutated I758 to glycine, which would disrupt potential hydrophobic interactions. However, this mutation had no effect on Cl^-/HCO₃^- exchange activity, ruling out a critical structural role for I758 (Figure 4 A and B, orange symbols). Furthermore, replacing I758 with the positively charged arginine also did not reduce AE4 activity (Figure 4 A and B, light green symbols). This contrasts with previous findings in the NBCe1 cotransporter, where the same mutation abolished transport (21).

Our findings indicate that residues T448 and D709 contribute to the AE4 transport cycle; however, Cl^-/HCO₃^- exchange was only partially reduced in the single mutants. To identify additional residues that might support electrostatic interactions within the TM3-TM10 region, we examined potential candidates and identified S446 and S447 in TM3, as well as S754 and T756 in TM10 (Figure S1). To assess their functional relevance, we performed an alanine scan of these positions. Notably, the S446A and T756A mutants exhibited a significant (~30%) reduction in transport activity compared to wild-type AE4 (Figure 4C and D, green (S446A) and light blue (T756A) symbols), suggesting that these residues play a role in AE4 function.

Additive effect of mutations at TM3-TM10 interface abolish AE4 activity

Individual mutations in the TM3-TM10 region did not completely abolish transport, suggesting that the transport cycle might require a combination of critical residues. This implies a mechanism in which multiple residues contribute additively to AE4 function. To test this hypothesis, we generated double mutants using the T448I mutant as the background, since this single mutation already reduced maximal activity, and the isoleucine side chain is relatively similar in size to threonine. We generated three double mutants: S446A-T448I, D709A-T448I and T756A-T448I, to assess the combined effects of these residues.

Functional analysis showed that mutants S446A-T448I and D709A-T448I exhibited a ~50% reduction in activity, similar to the single T448I mutant (Figure 5 A and B, orange and green symbols, respectively). However, the T756A-T448I mutant exhibited a transport activity comparable to non-transfected HEK-293 cells, indicating that Cl^-/HCO₃^- exchange activity was abolished (Figure 5 A and B, light blue symbols).

Figure 5. AE4 activity is abolished in T448I-T756A double mutant.

(A) Intracellular pH recordings were performed as described in Figure 4. Double mutants S446A-T448I (orange circles; n = 18) and D709A-T448I (green circles; n = 19) decreased transport by ~50% compared to wild-type AE4, a response such as that observed in the individual mutant T448I (Figure 4). The double mutant T756A-T448I (cyan circles; n = 18) decreased activity to an extent comparable to non-transfected HEK-293 cells (open symbols). Data from wild-type AE4 and HEK-293 cells experiments are taken from Figure 4 and are shown for comparison. Data are from at least 3 cells per experiment from at least 5 independent transfections. (B) Summary of the initial alkalinization rates of experiments shown in A. One-way ANOVA followed by Bonferronís post hoc test was performed. a, b, and c indicate significant difference from non-transfected HEK-293 cells, wild-type AE4 and double mutant T756A-T448I, respectively (p < 0.001). All results are shown as means ± SEM (error bars).

One possible explanation for the decreased activity observed in some mutants is a deficient cell membrane expression. Co-localization experiments comparing the expression pattern of FLAG-Tagged AE4 and the membrane marker Wheat Germ Agglutinin (WTA) conjugated to CF^® 555 dye showed no significant differences in membrane expression among mutants (Figure S2). We found a high degree of colocalization between WTA and wild-type AE4, (Mander's coefficient >0.8), (43) and no differences when comparing mutants and wild-type AE4, indicating that the decreased activity observed in the mutants was not a consequence of miss targeting to the cell membrane (Table S1). Moreover, immunoblot experiments showed that protein expression levels of the mutants were not different from wild-type AE4 (Figure S3).

In addition to the impaired ion transport activity observed in the mutants, the reduced transport capacity could be explained by alternative mechanisms related to mutations in the TM3-TM10 region: 1) Decreased driving force for HCO₃^- uptake due to increased resting intracellular HCO₃^- concentration, potentially due to an enhanced HCO₃^- uptake capacity in the mutants or a compensatory increase in HCO₃^- uptake. 2) Altered stoichiometry affecting HCO₃^-uptake. Previously, we proposed that AE4 functions as an electroneutral Cl^-/cation-HCO₃^- exchanger with a 1:1:2 stoichiometry of Cl^-: cation: HCO₃^-. However, a 1:1:1 stoichiometry (Cl^-: cation: CO₃^2-) also fits well with experimental data (19). We speculate that mutations might shift the transport mechanism from electroneutral to electrogenic by altering the stoichiometry of the transported species. Such alteration could impact the driving force for HCO₃^- uptake, reducing alkalinization.

To test this hypothesis, we first measured the resting intracellular pH in the presence of the physiological bathing solution A (table 1) and calculated the HCO₃^- concentration using the Henderson-Hasselbalch equation (see methods). The results showed no increase in intracellular HCO₃^- levels in the mutants compared to wild-type AE4. Interestingly, HEK cells expressing any AE4 variant tested in this study showed lower intracellular HCO₃^- concentrations than non-transfected cells (figure S4, A).

To assess whether AE4 mutant transport activity is influenced by membrane depolarization due to an electrogenic mechanism, we applied the K⁺-valinomycin method. Alkalinization rates remained unchanged between AE4-expressing cells treated with valinomycin in a high-K⁺ solution and those tested without the ionophore (Figure S4, B). The effect of valinomycin on membrane potential was validated by monitoring DiO/DPA fluorescence changes. These findings indicate that the transport cycle in AE4 mutants is independent of membrane potential (Figure S4, C).

The cation dependence of AE4 transport

AE4 is capable of transporting HCO₃^- in the presence of extracellular Na⁺, K⁺, Li⁺, Rb⁺ or Cs⁺ (19). Here, we found that the HCO₃^- transport rates were comparable in the presence of extracellular K⁺ or Na⁺ in wild-type AE4 (Figure 6A). This pattern was also observed in the S446A, T448I and T448G mutants. While all mutants display a similar HCO₃^- transport rates in the presence of extracellular K⁺, the single mutants D709A and T756A increased their activity by ~30% in the presence of sodium (Figure 6A). Interestingly, the double mutant T756A-T448I, which showed no detectable transport activity in the presence of Na⁺, maintained HCO₃^- transport rates comparable to the single mutants when K⁺ was the main extracellular cation (Figure 6 A). Additionally, using the non-permeant NMDG⁺ as the primary monovalent cation in the extracellular bath abolished transport in all the mutants, as well as in wild-type AE4 (Figure 6 B), confirming the strict dependence of AE4-mediated HCO₃^- transport on monovalent cations.

Figure 6. Alkalinization rates elicited by AE4 in the presence of extracellular Na⁺ and K⁺.

(A) Summary of the initial alkalinization rates from experiments where Na⁺ was switched to K⁺ as the main extracellular cation in HCO₃^--containing bathing solutions (white bars). Data from experiments using Na⁺-containing bathing solutions are shown for comparison (black bars). Intracellular alkalinization was elicited switching solutions from E to F. The transport activity was not changed in wild-type AE4 and mutants S446A, T448I and T448G in the presence of K⁺ compared to Na⁺. Mutants D709A and T756A decreased activity by ~30% while double mutant T756A-T448I recovered activity in the presence of K⁺ compared to Na⁺. Figures are from at least 3 cells per experiment from at least 3 independent transfections. (B) The transport activity was abolished in experiments where NMDG⁺ was used as the main extracellular cation (white bars). Data from Na⁺ containing bathing solutions are shown for comparison (black bars). Figures are representative of at least 3 cells per experiment from at least 3 independent transfections. One-way ANOVA followed by Bonferronís post hoc test was performed. A: Statistical analysis for K⁺ dependence experiments is shown. a and b indicate significant difference from wild-type AE4 and non-transfected HEK-293 cells, respectively (p < 0.001). c indicates significant difference between Na⁺ and K⁺ experiments (p < 0.001). B: a indicates significant difference from Na⁺ experiments (p < 0.001). All results are shown as means ± SEM (error bars).

Recently, the cryo-EM structure of the sodium-driven chloride carbonate exchanger NDCBE (Slc4a8) from Rattus norvegicus (rSlc4a8, PDB code: 7rtm) was solved in its outward-facing state (22). The transmembrane domain (TMD) contains two binding sites for ions, denoted S1 and S2. S1 is placed at the center of the exchanger, where a Na⁺ ion is coordinated by residues D800 and T804 (TM8 helix) and T847 (TM10 helix). In addition, CO₃^2- binds to T538 and G539 at TM3 helix. These residues are strictly conserved in AE4, and correspond to T448, G449, D709, T713 and T756, respectively (Figure S5, table S2). The S2, located above site S1 in NDCBE, involves residues T538, P540 and K970, and their equivalent residues in AE4 are T448, P450 and K879 (Figure S5). From this high conservation, similar roles in ion coordination of S1 and S2 sites are expected in AE4. Furthermore, according to AlphaFold 3D structure database (44), NDCBE and AE4 exhibit a similar fold (accession code: AF-Q96Q91-F1) and share 46% of sequence identity and 91% of sequence coverage (from BLASTp analysis). Table S2 highlights some similarities and differences between NDCBE and AE4 in the TMD. Additionally, our wild-type AE4 model was compared with the corresponding AlphaFold model for hSlc4a9 (access code: AF-Q96Q91-F1). The TMD domains were well-aligned, with an RMSD of 1.32 Å for the alpha carbon atoms (data not shown). Similarly, the superposition of the TMD domains of NDCBE and AE4 also exhibited low RMSD values (2.7 Å, data not shown). To determine the position of the putative ion coordination site in AE4 and the effect of the double mutation T756A-T448I on the cation binding, we built homology models with Na⁺/HCO₃^- or K⁺/HCO₃^- (placed at S1) using SWISS-MODEL web server (28). Our models were used as input for MD simulations of 120 ns, where the first 20 ns were under positional restraints. Three MD simulations were run per system. To analyze the ion coordination sites we selected MD runs where TMD domains exhibit stable RMSD which are: (1) for wild-type with Na⁺/HCO₃^- three replicas for chain A and three replicas for chain B, (2) for double mutant T756A-T448I with Na⁺/HCO₃^- two replicas for chain A, (3) for wild-type with K⁺/HCO₃^- three replicas for chain B and (4) for double mutant T756AT-448I with K⁺/HCO₃^- two replicas for chain B. Hereafter, all the analyses were carried out on these simulations, referred to as selected replicas. RMSD values for each selected replica are summarized in Figure S6.

We computed the distances between Na⁺ and the residues from the putative S1 and S2 sites (T448, G449, D709, T713, T756 and K879) as well as hydrogen bonds involving HCO₃^- anion coordination. In addition, we measured the ion-S1 distance. This corresponds to the distance between the center of mass of one permeant ion (Na⁺, K⁺, or HCO₃^-) and residue A755 from the S1 site, providing evidence of whether ions dissociate from their pocket. In wild-type AE4 model (Figure 7 A), Na⁺ remains at the S1 site along simulation and coordinates with HCO₃^-. Moreover, Na⁺ is coordinated by the hydroxyl oxygens of T713 and forms an ionic bond with the carboxyl group of D709. A similar ionic bond was previously described in the solved cryo-EM structure of NDCBE with Na⁺/CO₃^2- (22). In contrast, the hydroxyl oxygens of residues T448 and T756 are more distant to assume direct coordination with Na⁺ unless indirectly via water molecules. In the double mutant T756A-T448I with Na⁺/HCO₃^-, MD simulations evidenced that Na⁺ is coordinated by D709, similar to wild-type (Figure 7, B). Na⁺ is sustained at S1 site and is also capable to coordinate with HCO₃^-. However, the direct interaction of Na⁺ with T713 is notably disrupted (Figure 7, B) compared to wild-type (Figure 7 A).

Figure 7. Distances from positive ions to the residues from the putative S1 coordination site in AE4.

The average distances between Na⁺ and K⁺ ions to D709, T448, T713 and T756 were computed along 80 ns for (A) three replicas of MD simulations (chain A and chain B) for wild-type AE4 with Na⁺/HCO₃^-, (B) two replicas of MD simulations (chain A) for T756A-T448I with Na⁺/HCO₃^-, (C) three replicas of MD simulations (chain B) for wild-type AE4 with K⁺/HCO₃^- and (D) two replicas of MD simulations (chain B) for T756A-T448I with K⁺/HCO₃^-. Distances were measured from Na⁺ and K⁺ to carbonyl oxygen atoms (OD1 and OD2) of D709 side chain and, Na⁺ and K⁺ to hydroxyl oxygen atoms (OG1) of T713, T448 and T756 side chains. In addition, the distances between center of mass of Na⁺ or K⁺ to the anion HCO₃^- are also computed. The ion-S1 distance, which is the distance between the residue A755 and Na⁺, K⁺ or HCO₃^- is referred as Na-S1, K-S1 and HCO₃-S1, respectively. The residue A755 belongs to the S1 site, and the ion-S1 distance is used to determine whether ions dissociated from the S1 pocket.

For wild-type AE4, with K⁺/ HCO₃^- at the S1 site, we also observed that K⁺ (similar to Na⁺) establishes an ionic bond with D709 (Figure 7, C). K⁺ is also coordinated by residue T713. Interestingly, in the double mutant, with K⁺/HCO₃^-, potassium coordination with both, D709 and T713, is still present (Figure 7, D). In both systems, K⁺ does not dissociate from its pocket (Figures 7 C, and 7 D). The residues T448 and T756 in wild-type neither evidenced a direct interaction with K⁺. In double mutant, K⁺ and HCO₃^- are interacting (Figure 7 D) while they are more distant in the wild-type. A summary of the computed distances is shown in the table supplementary 3.

In all systems, the anion HCO₃^- is mainly bonded via hydrogen bonds with sidechain of K879 (from S2 site) and the backbone amine of G449, showing high hydrogen bond occupancy values (Table 2, Figures S7 and S8). HCO₃^- does not dissociate from its pocket, as suggested by stable HCO₃^-S1 distance values, in the 7–9 Å range, except at the end of the wild-type K⁺/HCO₃^- simulations. The interaction of HCO₃^- with residues from the S2 site agrees with a previous observation that CO₃^2- could occupy this site: (1) to draw sodium ions from the solution or (2) for temporary coordination before dissociation into the external solution (22). In wild-type AE4, T448 and T756 also form bonds with the anion, and their hydrogen bond occupancy values are higher than 50% (Table 2, figures S7 and S8, upper panels). In double mutant, these interactions are missing, but the anion is still coordinated by K879 and G449 in all systems (Table 2, figures S7 and S8, lower panels). Finally, the residue T713 apparently does not play a role in HCO₃^- coordination in any of the systems (Table 2).

Table 2. Hydrogen bond occupancy for wild-type AE4 and the double mutant T756A-T448I with Na⁺/HCO₃^- or K⁺/HCO₃^-.

For each MD trajectory, the hydrogen bonds among HCO₃^- and residues from S1 and S2 sites (D709, G449, K879, T756, T713 and T448), were computed using VMD software. For each system, at least two MD replicas were analyzed. The average number of hydrogen bonds at each frame from all replicas was computed. The hydrogen bond occupancy was calculated as the percentage of frames where hydrogen bond occurs.

Hydrogen bond occupancy (%)
Systems	HCO3− -D709	HCO3− -G449	HCO3− -K879	HCO3− -T448	HCO3− -T756	HCO3− -T713
WT, Na+/HCO3−	89.2	90.1	97.8	87.3	57.0	-
DM, Na+/HCO3−	66.5	54.8	58.6	-	-	-
WT, K+/HCO3−	-	31.7	81.5	77.3	64.6	-
DM, K+/HCO3−	42.0	20.0	86.6	-	-	-

In wild-type and double mutants with Na⁺ or K⁺ at the S1 site, we detected that either Na⁺ or K⁺ forms ionic bonds with an Aspartic residue, D709 (Figure 8). This is a prerequisite to anchoring the anions in the binding pocket S1 as previously discussed (22). The residue T713, a direct coordinator of the positive ions at the S1 site (Figure 8 A, C, D), does not interact with Na⁺ in the double mutant (Figure 8 B). However, the interaction between T713 and K⁺ is preserved despite mutations (Figure 8 D). These differences in the cation coordination could be responsible for the different transport activities shown by D709A, T756A, and the double mutant when using K⁺ or Na⁺ in the extracellular bath.

Figure 8. Representation of the residues coordinating the ions in wild-type AE4 and double mutant T756A-T448I.

Residues from putative ion coordination site, T448, G449, P450, D709, T713, T756 and K879 are shown in the wild-type AE4 (A and C) and double mutant T756A-T448I (B and D), in complex with Na⁺ (violet sphere) and HCO₃^- (A and B) or K⁺ (blue sphere) and HCO₃^- (C and D). Amino acid side chains and HCO₃^- are shown in licorice and colored by element (oxygen in red, hydrogen in white, nitrogen in blue, and carbon in gray). The interactions are indicated by dashed lines and the residues labeled according to their sequence number in human AE4 (accession code: NP_113655). Residues labeled in red are mutations T448I and T756A and the values of average distances (in angstroms) are labeled in blue. These distances were computed along the selected MD simulations replicas for each system during the last 80 ns. Hydrogen bonds involving HCO₃^- coordination are indicated as dashed lines and the hydrogen bond occupancy percent values are shown in orange (values from Table 2). Na⁺ or K⁺ interactions are represented with solid lines and the interacting distances are shown in blue.

DISCUSSION

It has been reported that AE4 mediates Na⁺-independent Cl^-/HCO₃^- exchange as well as Na⁺-HCO₃^- co-transport (11–14). On the other hand, we previously showed that AE4 mediates monovalent cation-dependent Cl^-/HCO₃^- exchange. This transport mechanism involves the movement of the physiologically important cations Na⁺ or K⁺ in the same direction as HCO₃^-, and the Cl^- ion in the opposite direction (19). This mechanism agrees with the proposed physiological role of AE4 in saliva secretion by salivary gland acinar cells. In this context, AE4 would contribute to Cl^- uptake at the basolateral membrane by mediating the exchange of Cl^- for cation-HCO₃^-. This occurs alongside the activity of NKCC1, which is considered the primary pathway for Cl^- uptake in these cells (19, 45, 46). Moreover, a recent study showed the importance of AE4 in the pH sensing in β-Intercalated cells (β-ICs) in kidney. Authors proposed that AE4 functions as a Na⁺(or K⁺)-dependent Cl^-/HCO₃^- exchanger that mediates the HCO₃^- uptake and Cl^- exit at the basolateral side of β-ICs (16). However, the specific structural motifs supporting the transport mechanism of AE4 have not yet been studied in detail. The aim of this study was to investigate the molecular determinants of the AE4 transport using an approach that included HCO₃^- transport assays, site specific mutagenesis and molecular modeling.

The structural and functional information obtained from other SLC4 transporters reveals that the putative ion binding site would be located in the region between short half-helices TM3 and TM10. The importance of this region is supported by previous mutagenesis studies (20–23, 36–40). Furthermore, a survey of human disease-related residues in AE1 and NBCe1 transporters also pinpoints this region (37, 47). Three positions at region TM3-TM10 have received special attention in this family of transporters: 1) Residue S465 in AE1, which corresponds to T485 in NBCe1; 2) Residue E681 in AE1 which is changed to D at position 754 in NBCe1; 3) Residue R730 in AE1 which is changed to the non-polar isoleucine at position 803 in NBCe1. Mutations at these positions have been reported to drastically reduce or abolish transport in several SLC4 proteins (21, 36–38, 42, 48, 49). Our analyses unveiled sequence and three-dimensional similarities within the family and that the equivalent residues in AE4, corresponding to T448, D709, and I758. Our molecular models showed that these residues in AE4 are in a cavity lined by TMs 1, 3, 8 and 10 at the region TM3-TM10, as in other SLC4 proteins. Here, by means of transport assays, mutagenesis, and bioinformatics, we explore whether residues T448, D709 and I758 are functionally important in AE4.

It has been reported that mutating S465 in AE1 (homologous to T448 in AE4) to isoleucine abolished Cl^-/HCO₃^- exchange (36). In contrast, we found that changing T448 to isoleucine or glycine decreased activity by only ~50 % in AE4. Interestingly, the equivalent mutation in NBCe1 (T485I) decreased the transport activity in a similar fashion (50). The comparable effects observed in the T448I and T448G mutants can be mainly attributed to the loss of the hydroxyl group at position 448, as the reduction in the side chain size in T448G did not lead to a further decrease in activity. On the other hand, when we neutralized D709 by changing it to asparagine, the transport activity was not affected, in contrast to previous reports where the same maneuver drastically reduced or abolished transport in other SLC4 proteins (21, 40–42, 48, 49). Moreover, when D709 was mutated by the non-polar alanine, we found that activity decreased only by ~30%. We then, investigated the residue I758 in AE4 which corresponds to arginine in Na⁺-independent Cl^-/HCO₃^- exchangers AE1-3, while in Na⁺-dependent SLC4 proteins is replaced by leucine or isoleucine. Based on this observation, it has been speculated that the Na⁺ transport is dependent on neutral side chains in this position, which could prevent electrostatic repulsion in the region (39). Mutations at this position in AE1 abolished transport, while in NBCe1 the activity decreased by ~60% (21, 37). The orientation of the side chain of I758 in our model, as well as its chemical nature, suggests that the contribution of this residue in AE4 is not related to ion coordination. When we mutated I758 by glycine in AE4, no changes in the HCO₃^- transport were observed, suggesting that the bulky side chain of isoleucine is not required for stabilizing a particular structural conformation during the transport cycle. Moreover, activity was not affected when I758 was changed by the positively charged arginine (same residue in AEs), ruling out positive charge repulsion on Na⁺-dependent transport in AE4. Since our mutagenesis studies on the aforementioned residues only partially decreased transport, we analyzed residues S446, S447, S754, and S756, which are also present in the TM3-TM10 interface. In our hands, mutants S446A and T756A decreased function by ~30% compared to wild-type AE4. These results indicate that polar residues T448, D709, S446, and T756 partially contribute to the transport cycle in AE4, while I758 is not involved in the transport mechanism.

However, the decreased transport activity in mutants might also be explained by deficient membrane expression, or other mechanisms such as: 1) Decreased driving force for HCO₃^- uptake due to increased resting intracellular HCO₃^- concentration, as a result of increased HCO₃^- uptake capacity of mutants, or a compensatory increased HCO₃^- uptake; 2) Altered stoichiometry could result in decreased HCO₃^- uptake. To explore these possibilities, we first measured intracellular HCO₃^- levels under physiological extracellular ion concentrations, observing no significant differences between wild-type AE4 and mutants. Interestingly, cells expressing AE4 exhibited lower resting intracellular HCO₃^- concentrations compared to non-transfected HEK-293 cells. This finding aligns with the proposed transport mechanism, in which Cl^- is transported intracellularly while HCO₃^- and a cation move in the opposite direction in physiological ion concentrations (17). Then, we performed immunodetection of AE4, and HCO₃^- transport experiments modulating the membrane potential using the K⁺ ionophore valinomycin. No differences between wild-type AE4 and mutants were observed, supporting the idea that the low transport activity in the mutants is due to an alteration in the interaction network at the ion coordination site.

Further, we analyzed three double mutants to determine whether the individual effects of single mutations might be additive. Our results show that mutants S446A-T448I and D709A-T448I decreased transport about 50%, like the single mutant T448I. Interestingly, double mutant T756A-T448I abolished transport, which is consistent with an additive effect of the individual mutations. We speculate that residues T448 and T756 might be important in different stages of the transport cycle in the presence of Na⁺, contributing partially to the overall process. Therefore, mutating both residues might result in a more pronounced disruption of the molecular interactions involved in the transport mechanism.

Previously we demonstrated that the mouse version of AE4 mediates K⁺-dependent Cl^-/HCO₃^- exchange when it is expressed in CHO-K1 cells (19). Here, we show that the human version of AE4 transports HCO₃^- in the presence of K⁺ as the main extracellular cation and transport was abolished when K⁺ is replaced by the impermeant NMDG⁺. The transport activity of wild-type AE4 and mutants S446A, T448I and T448G was comparable to that observed in the presence of Na⁺. Intriguingly, mutants D709A and T756A reduced activity by an additional ~20% in the presence of K⁺ compared to Na⁺, suggesting that these positions might play an important role in the K⁺-dependent HCO₃^- transport. Surprisingly, the double mutant T756A-T448I exhibited transport activity comparable to individual mutants T756A or T448I in the presence of K⁺, in contrast to the result with Na⁺, where the transport was abolished. This unexpected result might indicate that the protein-ion interaction network at the region TM3-TM10 could be different to support transport in the presence of Na⁺ or K⁺.

Molecular dynamics simulations revealed that, in wild-type AE4, Na⁺ is coordinated by the hydroxyl oxygens of the side chains of D709 and T713, and likely indirectly through water molecules by T756 and the HCO₃^- ion. On the other hand, HCO₃^- is primarily stabilized through hydrogen bonds with side chains of K879, T448, T756 and the backbone amine of G449. These findings align with our functional mutagenesis assays, which demonstrated reduced HCO₃^- transport in the D709, T448 and T756 mutants in the presence of Na⁺.

Remarkably, our simulations also indicated that K⁺ is coordinated by a similar interaction network involving the same residues that coordinate Na⁺. However, some interactions were absent in the presence of K⁺, specifically the T756-cation and D709-HCO₃^- interactions. This subtle difference may account for the decreased transport activity observed in the individual D709A and T756A mutants when exposed to K⁺ compared to Na⁺

Molecular dynamics simulations in the double mutant T756A-T448I showed that Na⁺ and HCO₃^- are still stabilized at the ion coordination site, although the interactions supported by hydroxyl oxygens of T448 and T756 are missed. Moreover, the interaction between Na⁺ and T713 is disrupted in the double mutant compared to wild-type AE4. This result suggests that the disruption of interactions supported by T448, T756 and T713, might be the cause of the abolition of transport in the double mutant in the presence of Na⁺. Finally, our dynamics simulations in the double mutant with K⁺ showed that interactions between T448 and T756 with ions are missed. However, the ionic bond between K⁺ and T713 is conserved, suggesting that this interaction is important to explain the remaining transport activity observed in this mutant in the presence of K⁺.

Previously, we proposed that AE4 is an electroneutral Cl^-/cation-HCO₃^- exchanger with a 1:1:2 stoichiometry of Cl^-: cation: HCO₃^- (19). Here, no changes in bicarbonate transport were observed in membrane depolarization experiments using valinomycin, supporting the hypothesis of an electroneutral transport mechanism mediated by AE4. Moreover, our molecular models showed the coordination of one HCO₃^- ion at the proposed ion binding site. This model requires a second HCO₃^- binding site according to the electroneutral transport mode of AE4. Such a mechanism has previously proposed to explain the electrogenic NBCe1 Na⁺-2HCO₃^- co-transport (51). However, the coordination of one CO₃^2- cannot be ruled out. This hypothesis agrees with an electroneutral transport mechanism with a 1:1:1 stoichiometry of Cl^-: cation: CO₃^2-, and the presence of three ion coordination sites, as previously proposed for the other SLC4 Na⁺-driven anion exchanger NDCBE (22). One plausible explanation is that the transported molecule may depend on the protonation state of the D709 side chain. The proton of HCO₃^- directly interacts with the oxygen atom of the D709 side chain (Na⁺-HCO₃^- molecular simulation, Figure 7A). However, protonation of the D709 side chain might instead facilitate interaction with the CO₃^2- ion. Therefore, we speculate that HCO₃^- or CO₃^2- might be coordinated within the TM3-TM10 interface of AE4, given the high amino acid conservation observed in this region between NDCBE and AE4.

Regarding Cl^- transport, it has been reported that this anion requires Na⁺ to stabilize at site S1 in the open outward-facing conformation of NDCBE (22). Although we did not perform molecular simulations using Cl^-, it is plausible that this mechanism is also necessary for AE4. The coordination of Cl^- in the external cavity is essential for Cl^- uptake through AE4 under physiological ion concentrations. We speculate that Cl^- may be stabilized by Na⁺ (the main extracellular cation) and that both are translocated to the intracellular side, while HCO₃^- (or CO₃^2-) is stabilized by K⁺ (the main intracellular cation) in the open inward-facing conformation, with both being translocated to the extracellular side. To maintain the electroneutrality of the transport cycle, the following stoichiometries are proposed: 1:1:1:1 (Cl^-: Na⁺: HCO₃^-: K⁺) or 1:1:1:2 (Cl^-: Na⁺: CO₃^2-: K⁺). Alternatively, Cl^- may be stabilized by the positive charge of K879 at site S1. This mechanism resembles the one proposed for AE2, where R1060 interacts with Cl^- in the external cavity, and the presence of a cation is not required (23). In this model, the proposed stoichiometries are 1:1:2 (Cl^-: cation: HCO₃^-) or 1:1:1 (Cl^-: cation: CO₃^2-). However, further experiments are required to investigate these hypotheses in greater detail.

In summary, this study provides evidence supporting a crucial functional role of the cavity formed by helices 1, 3, 8, and 10 in the AE4 transporter, consistent with findings in other SLC4 transporters. Our mutagenesis studies and molecular dynamics simulations identified residues S446, T448, D709, and T756 as key contributors to the transport cycle and ion coordination. Additionally, simulations revealed that G449, P450, T713, and K879 also participate in ion coordination, though their functional significance remains to be determined in future studies.

Notably, residues T448 and T756 play a central role in the cation dependence of AE4, as the double mutant T448I-T756A either abolished or significantly reduced HCO₃^- transport in the presence of Na⁺ and K⁺, respectively.

Supplementary Material

https://doi.org/10.6084/m9.figshare.28498067