Corneal dystrophy mutations R125H and R804H disable SLC4A11 by altering the extracellular pH dependence of the intracellular pK that governs H⁺(OH⁻) transport

Abstract

Mutations in the H⁺(OH⁻) conductor SLC4A11 result in corneal endothelial dystrophy. In previous studies using mouse Slc4a11, we showed that the pK value that governs the intracellular pH dependence of SLC4A11 (pK_i) is influenced by extracellular pH (pH_e). We also showed that some mutations result in acidic or alkaline shifts in pK_i, indicating that the pH dependence of SLC4A11 is important for physiological function. An R125H mutant, located in the cytosolic amino terminus of SLC4A11, apparently causes a complete loss of function, yet the anion transport inhibitor 4,4′-diisothiocyano-2,2′-stilbenedisulfonic acid (DIDS) can partially rescue SLC4A11/R125H activity. In the present study we set out to determine whether the effect of R125H is explained by an extreme shift in pK_i. In Xenopus oocytes, we measured SLC4A11-mediated H⁺(OH⁻) conductance while monitoring pH_i. We find that 1) the human corneal variant SLC4A11-B has a more acidic pK_i than mouse Slc4a11, likely due to the presence of an NH₂-terminal appendage; 2) pK_i for human SLC4A11 is acid-shifted by raising pH_e to 10.00; and 3) R125H and R804H mutants mediate substantial H⁺(OH⁻) conductances at pH_e = 10.00, with pK_i shifted into the wild-type range. These data suggest that the defect in each is a shift in pK_i at physiological pH_e, brought about by a disconnection in the mechanisms by which pH_e influences pK_i. Using de novo modeling, we show that R125 is located at the cytosolic dimer interface and suggest that this interface is critical for relaying the influence of pH_e on the external face of the transmembrane domain to the intracellular, pK_i-determining regions.

INTRODUCTION

SLC4A11 is a proton-conducting transmembrane protein that is robustly expressed in corneal endothelial cells (CECs) and is a divergent member of the SLC4 family of bicarbonate/carbonate transporters (1, 2). Whereas the nine other family members participate in sodium-coupled carbonate transport or chloride/bicarbonate exchange, SLC4A11 mediates highly selective proton (or hydroxyl ion: the two are thermodynamically indistinguishable) conductance (G_m) across the cell membrane (3–5). The magnitude of SLC4A11-mediated H⁺(OH⁻) conductance is independently increased by rises in either intracellular pH (pH_i) or extracellular pH (pH_e) (5). Specifically, the pH_i dependence of H⁺ transport is described by a Hill curve with a single pK_i (pH_i at which G_m is half maximal), the value of which is pH_e sensitive (6). SLC4A11 is predicted to mediate a small inward H⁺ current under physiological conditions, but experimental limitations on the extent of achievable intracellular alkalinization preclude the definition of pK_i at pH_e = 7.50. However, at pH_e = 8.50, mouse Slc4a11 exhibits a pK_i = 7.16 ± 0.01 with an apparent Hill coefficient (N_app) = 9 ± 1 representing an acidic shift in pK_i from its presumed value at pH_e = 7.50 (6). Many other actions have been proposed for human SLC4A11 including water transport, ammonia transport,¹ and cell adhesion (7–10). An alternative, mitochondrial expression of SLC4A11 has also been proposed under some circumstances (11). Our article focuses on SLC4A11-mediated H⁺(OH⁻) transport in the plasma membrane.

Mutations in SLC4A11 are associated with vision loss [Fuchs endothelial corneal dystrophy, congenital hereditary endothelial dystrophy (CHED)] and combined vision and hearing loss (Harboyan syndrome) (12). SLC4A11 supports the pump function of the single cell layer of CECs that line the aqueous humor side of the cornea. The pump, driven by the Na⁺-K⁺-ATPase and supported by the sodium-carbonate cotransporter NBCe1 and the H⁺-lactate cotransporter MCT1, removes bicarbonate and lactate from the corneal stroma, drawing out water and preventing the cornea from becoming edematous and cloudy (13). SLC4A11 is expressed alongside these transporters in the stromal-facing membrane of CECs. We hypothesize that under normal circumstances SLC4A11 would balance pH_i if MCT1 and NBCe1 action became imbalanced. That is to say, the SLC4A11-mediated H⁺ influx would be inhibited if MCT1 dominated (in response to a local lowering of pH_i) and would be enhanced if NBCe1 dominated (in response to a local elevation of pH_i) (5, 14). The physiological importance of the pH_i dependence of H⁺ transport is supported by our previous identification of an alkaline shift in pK_i as being responsible for diminished H⁺ transport activity exhibited by mouse mutant Slc4a11/R774H, the equivalent of the human CHED-causing mutant R804H (6). Another mouse mutant, Slc4a11/R99H, the equivalent of the human CHED-causing mutant R125H, could not be coaxed into action even at pH_e = 8.50, leading us to speculate that the mutant has an extremely alkaline-shifted pK_i, rendering the protein inactive under physiological and experimental conditions (6). SLC4A11/R125H is of special interest because wild-type action has been reported to be paradoxically restored by application of the SLC4 inhibitor 4,4′-diisothiocyano-2,2′-stilbenedisulfonic acid (DIDS) (4).

In this study, we sought to determine whether human SLC4A11, like mouse Slc4a11, exhibits pH-dependent H⁺ transport. We also sought to investigate the influence of R125H and R804H mutants on the activity of human SLC4A11 and test our hypothesis that the defect in SLC4A11/R125H lies with an alkaline-shifted pK_i. Our findings lead us to revisit our hypothesis about the role of these residues in support of H⁺ transport and to demonstrate that function can be restored to both by elevating extracellular pH.

MATERIALS AND METHODS

Constructs

The cloning of human SLC4A11-B (GenBank Protein Accession: NP_114423) and construction of the BSXG4.hSLC4A11-B vector have been previously reported (15). We refer to this wild-type clone as hSLC4A11. The Δ1–35 deletion mutant was created with the Phusion site-directed mutagenesis kit (Thermo Fisher Scientific, Waltham, MA), and the point mutations discussed in this article were cloned using the QuikChange II mutagenesis kit (Agilent Technologies, Santa Clara, CA) with the primers listed in Table 1.

Table 1.

Primers used for human SLC4A11 mutants in this study

Mutation	Sequence
Δ1–35	FWD: 5′ ATGTCGCAGAATGGATACTTCGAGG RVS: 5′ GGTGGAGATCTATGGCCAAAGTTG
R125H	FWD: 5′ CTTCAAGGAAGAGATCCATGCGCACCGCGACCTAG RVS: 5′ CTAGGTCGCGGTGCGCATGGATCTCTTCCTTGAAG
R804H	FWD: 5′ CTCGTCCAGCACGTGGCCCTGC RVS: 5′ GCAGGGCCACGTGCTGGACGAG

Template: BSXG4.hSLC4A11. FWD, forward; RVS, reverse.

Oocyte Preparation

In accordance with University at Buffalo Institutional Animal Care and Use committee protocols, ovaries were extracted from Xenopus laevis frogs and prepared. Under 0.2% tricaine solution, frogs were anesthetized. The ovary was surgically extracted from the frog, and the frog was subsequently euthanized via exsanguination. The ovary was dissected into 1-cm chunks and placed in a solution of calcium-free Ringer solution [Ca-free NRS (in mM): 82 NaCl, 20 MgCl₂, 5 HEPES, and 2 KCl]. The ovary was washed in this solution three times for 5 min. Collagenase solution [2 mg/mL type 1A collagenase (Sigma-Aldrich, St. Louis, MO) in Ca-free NRS] was applied to the ovary to digest the connective tissue holding the oocytes in place. Oocytes were rinsed with Ca-free NRS three times for 10 min upon noticeable dissociation of the oocytes from the ovary. The cells were then washed with ND96 for 10 min, followed by a wash with OR3 for 10 min [OR3: 14 g Leibovitz’s L-15 medium powder (Thermo Fisher Scientific), 5 mM HEPES, 20 mL 100× penicillin-streptomycin (Corning Inc., Corning, NY), pH 7.50, 200 mosmol/kgH₂O]. Cells were kept at 18°C in fresh OR3 medium.

cRNA Preparation and Injection

BSXG plasmids were linearized with enzyme HindIII and purified with the Qiagen MinElute PCR Purification kit according to manufacturer’s instructions. The purified DNA was transcribed to cRNA with the T7 mMESSAGE mMACHINE kit (Invitrogen, Carlsbad, CA) and subsequently purified with the Qiagen RNeasy MinElute Cleanup kit. Final concentration of cRNA was assessed with a NanoDrop 2000 (Thermo Fisher Scientific). A Nanoject III programmable injector (Drummond Scientific Co., Broomall, PA) was used to inject the cRNA into oocytes.

Electrophysiology

Oocytes were placed into an ND96-superfused chamber (no. RC-3Z; Warner Instruments, Hamden, CT) affixed to an antivibration worktable (Vision IsoStation; Newport Corp., Irvine, CA). ND96 was perfused from syringe pumps (Harvard Apparatus, Holliston, MA) at a rate of 2 mL/min. Borosilicate glass microelectrodes (no. BF200-156-10; Sutter Instrument, Novato, CA) were pulled with a micropipette puller (model P-1000; Sutter Instrument) set to result in pipettes with a tip resistance of 0.1–2 MΩ when filled with saturated KCl solution. Oocytes were impaled with two microelectrodes (current passing and voltage sensing) connected to an oocyte clamp (OC275; Warner Instruments, Hamden, CT). A bath clamp (no. 725I; Warner Instruments) was used to clamp the oocyte bath potential to 0 mV. Current-voltage (I-V) plots were gathered by clamping the cell membrane from −160 mV to +20 mV in 20-mV steps for 100 ms, returning to resting potential for 100 ms between steps. To monitor intracellular pH (pH_i), a H⁺-selective microelectrode was used to impale the oocyte on the surface opposite to the I-V electrodes and connected to a dual-channel electrometer (HiZ-223; Warner Instruments) as previously described (5, 6, 16). I-V data were digitized and gathered with a Digidata 1550 with Clampex 10.4 software (Molecular Devices LLC, San Jose, CA). Recordings of pH_i and membrane potential (V_m) were continuously monitored with custom software (written by Dale Huffman for Walter Boron’s laboratory at Case Western Reserve University, Cleveland, OH). In water-injected control experiments and in some SLC4A11/R125H experiments where membrane conductance (G_m) was particularly low, intracellular alkalinization was induced with a microinjector (Nanoject II; Drummond Scientific Co.) by injections of 2.3 nL of 1.2 M NaHCO₃ into the oocyte. However, in most experiments performed at pH_e = 8.50 shown in this article, SLC4A11-mediated H⁺ conductance was sufficient to alkalinize pH_i when cells were voltage clamped to a value more positive than reversal potential for H⁺ (E_H) to drive H⁺ extrusion. For experiments at pH_e = 10.00, SLC4A11-mediated alkalinization was spontaneous and no voltage clamping was required to drive H⁺ extrusion. If spontaneous alkalinization was rambunctious at pH_e = 10.00, a microinjector was used to preacidify pH_i of cells with 10 nL of 0.1 M HCl. The cells were allowed to rest for at least 5 min in OR3 before being placed in the rig in ND96 for electrophysiology.

Electrophysiology Solutions

The recipes used in this protocol are based on solutions previously described (5, 6). ND96 is the basis for each electrophysiology solution and contains (in mM) 93.5 NaCl, 2 KCl, 1.8 CaCl₂, and 1 MgCl₂ with 5 mM of buffer. For pH 7.50 solutions (pH_e = 7.50), the buffer used was HEPES. For pH 8.50 solutions (pH_e = 8.50), the buffer used was bicine. For pH 9.50 and pH 10.00 solutions (pH_e = 9.50 and pH_e = 10.00), the buffer used was N-cyclohexyl-2-aminoethanesulfonic acid (CHES).

Data Analysis

Conductance (G_m) was calculated with Microsoft Excel as the slope of I-V plots from –160 mV to –100 mV, a range within which the I-V plots do not substantially rectify. Analysis of the data between –120 mV and +20 mV per our previous article on the subject (6) (not shown) produces the same results, albeit with lower G_m values, indicating that the shape of the I-V plot is unaltered by changes in pH_i. Data sets were also excluded from analysis if they contained fewer than three data points on either side of the estimated pK_i for the set. Plots of pH_i versus G_m were fit to the Hill equation by using the solver add-on in Microsoft Excel to solve for EC₅₀ (the hydroxyl ion concentration at which G_m is half maximal) and N_app (the apparent Hill coefficient) as shown below:

$$\text{Hill equation}:\frac{G_{\mathrm{m}}}{G_{\mathrm{m},\max }}=\frac{1}{1+{(\mathrm{E}{\mathrm{C}}_{50}/\left[\mathrm{O}{\mathrm{H}}^{-}\right])}^{N_{\mathrm{app}}}}$$

Hydroxide (OH⁻) is considered our agonist for proton transport, and thus in our study we convert our EC₅₀ value to that of pK_i (the pH_i at which G_m = 50% G_m,max):

$$\text{pK}=14-\left[-\log \hspace{0.17em}\left(\mathrm{E}{\mathrm{C}}_{50}\right)\right]$$

The Hill coefficient, represented by N, remains a complex quantifier in our example, as cooperativity of agonist binding affects the interpretation of the value. Since we do not have a clear understanding of cooperativity of protonation events, we consider the Hill coefficient to represent the minimum number of titratable residues involved in the pH dependence of our tested conditions. Because of their indeterminate nature, we have not performed statistical analyses on Hill coefficient values. Statistical analyses were performed via t test for pairwise comparisons or via a general linear model ANOVA for multiple comparisons, as appropriate, with Minitab 21 software.

Biotinylation and Immunoblotting

Biotinylation was performed to reveal any differences in surface expression between the wild-type protein and mutant proteins in this article. Biotinylation was carried out according to manufacturer’s instructions with the Pierce Cell Surface Protein Isolation kit (Thermo Fisher Scientific). Briefly, 10 oocytes were washed with PBS200 (PBS diluted to 200 mosmol/kgH₂O for oocyte stability). Cells were then placed in biotin solution for 1 h at 4°C. After quenching, cells were rinsed in Tris-buffered saline (TBS) before being placed in 1.5-mL tubes. Cells were incubated and lysed in oocyte lysis buffer [50 mL TBS + protease inhibitor tablet (Pierce) + 500 μL Triton X). Tubes of lysate were spun down at 3,000 rpm while NeutrAvidin columns were primed with lysis buffer. After spinning, 25 μL of lysate was set aside for a “total” sample, and the remaining lysate was subject to NeutrAvidin binding for 1 h. Unbound protein was eluted, and NeutrAvidin columns were washed with oocyte lysis buffer. Finally, SDS elution buffer was applied to the NeutrAvidin columns, incubated for 1 h, and then eluted and transferred to a 1.5-mL tube.

Protein samples were loaded (∼0.1 oocyte equivalent volume for total, ∼0.5 oocyte equivalent volume for surface) and run onto a 3–8% Tris-acetate gel. Bands were transferred to a PVDF membrane. After blocking in 4% TBS-Tween (TBS-T)-powdered milk overnight, the membranes were probed with a 1:1,000 dilution of our custom anti-SLC4A11 antibody that had been raised in rabbits against an epitope, LPRIIEAKYLDVMDAEHRP, that is common to the carboxy termini of both human and mouse SLC4A11 sequences. We have previously validated this antibody for use in detecting heterologously expressed mouse Slc4a11 in Xenopus oocytes (5). Membranes were subsequently washed with TBS-T and probed with goat anti-rabbit secondary antibody diluted at 1:2,000. Membranes were exposed to Pierce ECL Plus Western blotting substrate and then imaged with ECL imager. Analysis of bands was carried out with FIJI software (ImageJ).

Modeling

Three models of the SLC4A11 soluble NH₂-terminal domain (Nt) are included in this report, which were obtained with separate structural prediction tools. One such tool, AlphaFold, utilizes deep-learning networks to deliver predictions de novo, solely based upon amino acid sequence (17). Another tool, RoseTTAFold, utilizes a user-provided homologous protein to predict structure (18). Finally, I-TASSER is a tool that predicts protein structure by comparing the user-provided sequence to others on the PDB database and uses known structures and simulations to provide an estimation for protein structure (19). The full-length sequence of human SLC4A11-B was submitted to these programs, each of which returned five structural predictions. Of the five predicted models, the model with the highest predicted confidence was chosen to serve as a representative prediction for that program.

To obtain the predicted NH₂-terminal dimers, these representative models were loaded into the program PyMOL along with the structure of the NH₂-terminal dimer of protein AE1 (SLC4A1, PDB: 4KY9). Residues 80:283 (region of high confidence) of each SLC4A11 prediction were selected as the soluble domain, and the rest of the protein structure was hidden. The soluble domain was copied, and then each was aligned with a separate monomer of the AE1 NH₂-terminal dimer. Finally, the AE1 NH₂-terminal dimer structure was hidden to leave a resulting SLC4A11 NH₂-terminal dimer model.

RESULTS

Determining pK_i for Human SLC4A11-B in Xenopus Oocytes

Figure 1A shows a representative pair of pH_i and V_m traces gathered from an oocyte expressing human SLC4A11 in our pH_e = 8.50 bath solution. Figure 1A, top, shows the pH_i increase that occurs when the cell membrane is clamped to a potential of −40 mV (Figure 1A, bottom), a slight depolarization applied to drive proton efflux. Figure 1B shows a selection of representative I-V plots obtained at various values of pH_i during the course of the experiment shown in Fig. 1A. Note that membrane conductance (G_m) rises with pH_i. The relationship between pH_i and G_m is plotted for a larger number of cells in Fig. 1C. We had previously established that intracellular alkalinization does not elicit an increase in G_m in H₂O-injected oocytes at pH_e = 8.50 (6). The G_m,max for H₂O-injected cells at pH_e = 8.50 from those previous studies is represented by the gray dashed line. Figure 1D shows each of those relationships for hSLC4A11 normalized to its respective G_m,max in order to fit to the Hill equation and extract value for pK_i (an index of pH_i dependence) and the Hill coefficient (an index of pH_i sensitivity). The average pK_i for oocytes expressing hSLC4A11 was 7.04 ± 0.01, and the average Hill coefficient was 9 ± 2. The solid line in Fig. 1D shows a Hill plot generated with these average properties. The range of pK_i values that we gathered for hSLC4A11 is significantly different from that which we had previously determined with this protocol for oocytes expressing mouse Slc4a11, for which pK_i = 7.17 ± 0.01 (P < 0.01, 2-tailed, unpaired t test: pK_i value recalculated for this study based on G_m between −160 and −100 mV). The Hill coefficient is not significantly different between these human and mouse clones (P = 0.83, 2-tailed, unpaired t test).

Figure 1.

The intracellular pH (pH_i) dependence of human (h)SLC4A11 activity when extracellular pH (pH_e) = 8.50. A: a representative recording of pH_i (top) and membrane potential (V_m) (bottom) vs. time for a wild-type human SLC4A11-expressing oocyte. Data scatter around the trace indicates time points at which current-voltage (I-V) relationships were gathered. B: a selection of representative I-V plots obtained during the experiment shown in A. C: membrane conductance (G_m) values plotted against pH_i for n = 6 cells. D: data taken from C, but each trace is normalized to its respective G_m,max for a fit to the Hill equation. The solid black line represents the Hill equation when solved for the average pK value that governs intracellular pH dependence (pK_i) (pK_avg) and Hill coefficient of the n = 7 replicates, as shown in the inset. N_app, apparent Hill coefficient.

Figure 2A shows a cartoon representation of the topology of SLC4A11 showing the cytosolic amino terminus and the multispanning transmembrane domain (TMD). A major difference between the one Slc4a11 gene product in mice and the major corneal hSLC4A11 variant (SLC4A11-B) is the presence in the human product of a 35-amino acid (aa) amino-terminal appendage. Besides this difference, the remainder of hSLC4A11-B is 85% identical to mouse Slc4a11 (GenBank Protein Accession No. A2AJN7). To investigate which of the structural differences between mouse and human SLC4A11 is responsible for their difference in pK_i, we created a truncated form of hSLC4A11, hSLC4A11/Δ1–35, that initiates translation at Met36, and set about to determine its pK_i using the same approach as shown in Fig. 1. Figure 2B shows select I-V plots gathered from a representative hSLC4A11/Δ1–35-expressing oocyte as pH_i increased under voltage clamp. The relationship between pH_i and G_m is plotted for a larger number of cells in Fig. 2C. Figure 2D shows each of those relationships normalized to its respective G_m,max in order to fit to the Hill equation and extract value for pK_i and the Hill coefficient. The average pK_i for oocytes expressing hSLC4A11/Δ1–35 was 7.16 ± 0.05, and the average Hill coefficient was 7 ± 2. The solid black curve in Fig. 2D shows a Hill plot generated with these average properties. Figure 3 shows an overlay of the average properties of hSLC4A11, hSLC4A11/Δ1–35, and our previously determined values for mouse Slc4a11 (6). The pK_i of hSLC4A11/Δ1–35 is not significantly different from the pK_i of mSlc4a11 (P = 0.77, 2-tailed, unpaired t test).

Figure 2.

The intracellular pH (pH_i) dependence of human (h)SLC4A11/Δ1-35 at extracellular pH (pH_e) = 8.50. A: an illustration of the SLC4A11-B gene product showing the location of the 35 amino acids deleted in hSLC4A11/Δ1-35 as well as the 2 arginine mutants R125H and R804H used in later experiments. B: a selection of representative current-voltage (I-V) plots gathered from hSLC4A11/Δ1-35-expressing cells as pH_i was increased. V_m, membrane potential. C: membrane conductance (G_m) values plotted against pH_i for n = 8 cells. D: data taken from C, but each trace is normalized to its respective G_m,max for a fit to the Hill equation. The solid black line represents the Hill equation when solved for the average pK value that governs intracellular pH dependence (pK_i) (pK_avg) and Hill coefficient of the n = 8 replicates, as shown in the inset. G_m,max, maximum G_m; N_app, apparent Hill coefficient.

Figure 3.

Comparison of human (h)SLC4A11 and hSLC4A11/Δ1-35 to mouse (m)Slc4a11 at extracellular pH (pH_e) = 8.50. Traces representing the Hill equation when solved for the average pK value that governs intracellular pH dependence (pK_i) (pK_avg) and Hill coefficient observed for each construct. Only the pK_i of full-length hSLC4A11 is significantly different from that of mSlc4a11 (*P < 0.05, unpaired, 2-tailed t test). G_m, membrane conductance; G_m,max, maximum G_m.

Investigating the Effect of pH_e on pK_i

During our previous investigation of the properties of mouse (m)Slc4a11, we determined that the value of pK_i of mSlc4a11 could be significantly acid shifted by a rise in pH_e from 8.00 to 8.50 (6). To determine whether the pK_i of hSLC4A11 is also sensitive to pH_e, we sought to determine pK_i in a pH_e = 10.00 bathing solution. Figure 4A shows a selection of I-V relationships gathered from a representative H₂O-injected oocyte bathed in pH_e = 10.00 solution as pH_i was raised by injection with NaHCO₃ solution. As demonstrated for a larger number of H₂O-injected cells in Fig. 4B, raising pH_i under these conditions has little effect upon G_m, consistent with previously published control experiments performed at pH_e = 7.50 and pH_e = 8.50 (6). On the other hand, applying our pH_e = 10.00 bath solution to an hSLC4A11-expressing oocyte caused an immediate and rapid rise of G_m to G_m,max (not shown), precluding accurate determination of pK_i. Presuming that this phenomenon was consistent with a shift in pK_i to a value more acidic than resting pH_i, we preacidified subsequently tested hSLC4A11 cells with a 10-nL injection of 0.1 M HCl before applying the pH_e = 10.00 solution. Figure 4C shows voltage-clamp data gathered from a representative preacidified hSLC4A11-expressing oocyte at pH_e = 10.00 as pH_i spontaneously rose. The relationship between pH_i and G_m for a larger number of these cells is shown in Fig. 4D and is shown normalized to G_m,max in Fig. 4E. From these data, we determined a pK_i value of 6.82 ± 0.09 and a Hill coefficient of 9 ± 2. The pK_i of hSLC4A11 at pH_e = 10.00 is significantly more acidic than the pK_i of hSlc4a11 at pH_e = 8.50 (P = 0.01, 1-tailed, unpaired t test). The plot of the transmembrane pH gradient against membrane potential in Fig. 4F indicates a Nernst slope of 42 ± 5 mV/decade with respect to H⁺ permeability, which is significantly less than the ideal slope of 58 mV/decade (P = 0.02, 1-tailed t test).

Figure 4.

The intracellular pH (pH_i) dependence of membrane conductance (G_m) for H₂O-injected and human (h)SLC4A11-expressing cells at extracellular pH (pH_e) = 10.00. A: a selection of representative current-voltage (I-V) plots gathered from H₂O-injected cells as pH_i was increased. V_m, membrane potential. B: G_m values plotted against pH_i for n = 6 H₂O-injected cells. C: a selection of representative I-V plots gathered from hSLC4A11-expressing cells as pH_i was increased. D: G_m values plotted against pH_i for n = 7 hSLC4A11-expressing cells. E: data taken from C, but each trace is normalized to its respective G_m,max for a fit to the Hill equation. The solid black line represents the Hill equation when solved for the average pK value that governs intracellular pH dependence (pK_i) (pK_avg) and Hill coefficient of the n = 7 replicates, as shown in the inset. F: the relationship between V_m and transmembrane pH gradient (pH_i − pH_e) for hSLC4A11-expressing cells. The Nernstian slope representing ideal H⁺ selectivity is shown in gray. G_m,max, maximum G_m; N_app, apparent Hill coefficient.

Determining pK_i for the Dystrophy-Associated Mutants SLC4A11/R125H and SLC4A11/R804H

During our previous investigation of the properties of mouse Slc4a11, we assessed the mouse equivalent of the hSLC4A11/R125H mutation, mSlc4a11/R99H, for H⁺ transport activity but were unable to discern any significant differences in G_m between mSlc4a11/R99H-expressing cells and H₂O-injected oocytes even when pH_e = 8.50 and pH_i was in the alkaline range 7.60–7.89 (6). In this study, we sought to determine whether the more acidic pK_i of human SLC4A11 would place the pK_i of hSLC4A11/R125H within determinable range.

The first indication in our study that hSLC4A11/R125H is defective with respect to H⁺ transport comes from observations made at pH_e = 7.50. A previously noted feature of oocytes expressing mouse Slc4a11 is that they are somewhat depolarized and acidified at rest in pH_e = 7.50 solution, reflecting the influx of H⁺ and the shift of V_m toward the reversal potential for H⁺ (E_H) (5). In the present study we find the values of pH_i and V_m that are presented in Table 2. ANOVA analysis indicates that, for both parameters when pH_e = 7.50, hSLC4A11/R125H-expressing cells are not different from H₂O-injected cells and that both hSLC4A11/R125H-expressing cells and H₂O-injected cells are different from hSLC4A11-expressing cells (P < 0.05).

Table 2.

Intracellular pH and spontaneous membrane potential values of oocytes at pH_e = 7.50

	pHi	V m	n
H2O injected	7.22 ± 0.01	–40 ± 4	7
hSLC4A11 cRNA injected	6.95 ± 0.02	–25 ± 2	15
R125H cRNA injected	7.22 ± 0.02	–34 ± 3	18

Values are means ± SE. pH_e, extracellular pH; pH_i, intracellular pH; V_m, membrane potential.

The second indication in our study that hSLC4A11/R125H is defective with respect to H⁺ transport comes from observations made at pH_e = 8.50. Figure 5A shows, for an oocyte expressing SLC4A11/R125H, a selection of representative I-V plots obtained at various values of pH_i when pH_e = 8.50. Figure 5B shows for a greater number of cells the relationship between pH_i and G_m. This is a pattern strikingly different from that observed for wild-type hSLC4A11 in Fig. 1C; even though similar values of G_m are exhibited by either construct at the most alkaline pH_i values, the pH_i versus G_m relationship for hSLC4A11/R125H cells did not plateau at high pH_i, precluding calculation of pK_i. However, normalizing each trace to its own maximumly achieved conductance value, we can estimate that the most acidic possible value of pK_i for hSLC4A11/R125H at pH_e = 8.50 is 7.6, far more alkaline than 7.04, the average pK_i of wild-type hSLC4A11 at pH_e = 8.50 (Fig. 1D).

Figure 5.

The intracellular pH (pH_i) dependence of membrane conductance (G_m) for human (h)SLC4A11/R125H-expressing cells at extracellular pH (pH_e) = 8.50. A: a selection of representative current-voltage (I-V) plots gathered from hSLC4A11/R125H-expressing cells as pH_i was increased. B: G_m values plotted against pH_i for n = 8 hSLC4A11/R125H-expressing cells.

Figure 6, A and B, show the results of experiments similar to those shown in Fig. 5, except that they were performed at pH_e = 10.00 upon preacidified SLC4A11/R125H-expressing oocytes. Under these conditions, the cells spontaneously hyperpolarize and alkalinize (Fig. 6C) and the relationship between pH_i and G_m did reach a plateau, allowing data to be fit to the Hill equation (Fig. 6D). We determine that the pK_i for SLC4A11/R125H is 7.14 ± 0.05 at pH_e = 10.00, with a Hill coefficient of 9 ± 3. The plot of the transmembrane pH gradient against membrane potential in Fig. 6E indicates a Nernst slope of 22 ± 2 mV/decade with respect to H⁺ permeability, which is significantly less than ideal (P < 0.01, 1-tailed, paired t test).

Figure 6.

The intracellular pH (pH_i) dependence of human(h)SLC4A11/R125H activity at extracellular pH (pH_e) = 10.00. A: a selection of representative current-voltage (I-V) plots gathered from hSLC4A11/R125H-expressing cells as pH_i was increased. B: membrane conductance (G_m) values plotted against pH_i for n = 6 hSLC4A11/R125H-expressing cells. C: a representative recording of pH_i (top) and membrane potential (V_m) (bottom) vs. time for an oocyte expressing hSLC4A11/R125H with exposure to a bath solution of pH = 10.00. Changes in pH_i and V_m are spontaneous, not requiring voltage clamp to drive proton efflux or NaHCO₃ injections for pH_i increases. D: data taken from B, but each trace is normalized to its respective G_m,max for a fit to the Hill equation. The solid black line represents the Hill equation when solved for the average pK value that governs intracellular pH dependence (pK_i) and Hill coefficient of the n = 6 replicates, as shown in the inset. E: the relationship between V_m and transmembrane pH gradient (pH_i − pH_e) for hSLC4A11/R125H-expressing cells. The Nernstian slope representing ideal H⁺ selectivity is shown in gray. G_m,max, maximum G_m; N_app, apparent Hill coefficient.

Figure 7 shows the results of similar experiments performed on oocytes expressing hSLC4A11/R804H. Figure 7A shows a selection of I-V plots gathered from a representative cell expressing hSLC4A11/R804H as it was alkalinized under voltage clamp in our pH_e = 8.50 solution. Figure 7B shows a selection of I-V plots gathered from a representative preacidified cell expressing SLC4A11/R804H as it spontaneously alkalinized in our pH_e = 10.00 solution. Figure 7C shows the pH_i versus G_m relationships for a larger number of cells that underwent these protocols, and Fig. 7D shows these data normalized to their respective G_m,max, superimposed upon two solid lines that represent Hill curves drawn from the average pK_i and Hill coefficients calculated from the pH_e = 8.50 (black curve) and pH_e = 10.00 (gray curve) data sets. We find that pK_i for R804H is significantly more acidic at pH_e = 10.00 than at pH_e = 8.50 (P = 0.02, 1-tailed, unpaired t test).

Figure 7.

The intracellular pH (pH_i) dependence of human (h)SLC4A11/R804H activity at extracellular pH (pH_e) = 8.50 or 10.00. A: a selection of representative current-voltage (I-V) plots gathered from hSLC4A11/R804H-expressing cells at pH_e = 8.50 as pH_i was increased. V_m, membrane potential. B: a selection of representative I-V plots gathered from hSLC4A11/R804H-expressing cells at pH_e = 10.00 as pH_i was increased. C: membrane conductance (G_m) values plotted against pH_i for hSLC4A11/R804H-expressing cells at pH_e = 8.50 (n = 6) and pH_e = 10.00 (n = 7). D: data taken from C, but each trace is normalized to its respective G_m,max for a fit to the Hill equation. The solid lines represent the Hill equation when solved for the average pK value that governs intracellular pH dependence (pK_i) and Hill coefficient of the replicates at pH_e = 8.50 (black line) and pH_e = 10.00 (gray line), as shown in the inset. *Significantly more acidic than at pH_e = 8.50 (1-tailed t test, with Bonferroni correction for multiple comparisons). G_m,max, maximum G_m; N_app, apparent Hill coefficient.

However, the pK_i for both R804H and R125H mutants at pH_e = 10.00 is statistically indistinguishable from the pK_i of wild-type hSLC4A11 at the lower pH_e = 8.50 (P > 0.05, unpaired 2-tailed t test). Thus, both mutants can attain wild-type pK_i values, but with a greater change in pH_e.

Investigating the Ability of SLC4A11 Mutants to Conduct H⁺(OH⁻) at pH_e = 10.00

Figure 8A extracts data from Fig. 4B, Fig. 4D, Fig. 6B, and Fig. 7B along with values from additional data sets that were unsuitable for pK_i determination but which reliably reported G_m,max. These data show that G_m,max at pH_e = 10.00 is unexpectedly greater for hSLC4A11/R125H than for wild-type hSLC4A11 (P < 0.05 ANOVA with Tukey’s post hoc analysis). To determine whether this result represents a per-molecule increase in H⁺(OH⁻) conductance, we performed seven sets of biotinylation experiments on cells expressing these three constructs. A representative Western blot from one data set is shown in Fig. 8B. The densitometric analysis of the blots from all experiments is shown in Fig. 8C. We find that G_m,max of hSLC4A11/R125H-expressing cells (corrected for the G_m,max of H₂O-injected cells) is approximately twofold greater than that of wild-type SLC4A11-expressing cells, but we also find that hSLC4A11/R125H is approximately threefold more abundant in the plasma membrane than wild-type hSLC4A11. Thus the difference in G_m,max is predominantly explained by differences in plasma membrane expression, but we cannot rule out a diminution of H⁺ (OH⁻) conductance. We make a similar observation for SLC4A11/R804H, for which G_m,max is unchanged while surface expression is approximately twofold greater than wild-type hSLC4A11.

Figure 8.

Comparison of maximum membrane conductance (G_m,max) among human (h)SLC4A11 and mutants hSLC4A11/R125H and hSLC4A11/R04H at extracellular pH (pH_e) = 10.00. A: comparison of G_m,max among hSLC4A11 and mutants hSLC4A11/R125H and hSLC4A11/R04H. *P < 0.05 by ANOVA. B: representative Western blot of biotinylated membrane protein extracts from oocytes expressing hSLC4A11 or 1 of the 2 mutants. C: quantification of n = 7 replicates of the biotinylation experiment from B, providing an index of relative surface expression of mutants compared to wild-type hSLC4A11. *Significantly different from wild type (P < 0.05, unpaired t test).

Hypothetical Models

Figure 9A shows an overlay of the three de novo models of the cytosolic NH₂ terminus of human SLC4A11 generated by I-TASSER (Fig. 9B), RoseTTA (Fig. 9C), and AlphaFold (Fig. 9D). Despite obvious differences in the predicted structure of the sequence away from the dimer interface, the three models show regions of similar structure. Alpha helices exist in each model, consisting of the following groups of residues: 120–136, 148–161, 172–181, 265–277, 279–287. Beta strands exist in each model, consisting of each of the following groups of residues: 138–141, 228–234, 249–254. In every model, the R125 residue is situated in alpha helices at the putative Nt dimer interface.

Figure 9.

Models of the cytosolic NH₂-terminal domain of SLC4A11. Shown are models of the cytosolic NH₂-terminal dimer interface generated using a variety of modeling programs to estimate where the R125H residue is or what it could be interacting with in the domain. A shows all 3 models overlaid for comparison. B shows the model generated from the I-TASSER program. C: RoseTTA. D: AlphaFold.

DISCUSSION

NH₂-Terminal Differences between Mouse and Human SLC4A11

Unlike mouse Slc4a11, which expresses only one gene product, the human SLC4A11 gene expresses three: SLC4A11-A (GenBank Protein Accession No. NP_001167561), -B (used in this study), and -C (GenBank Accession No. NP_001167560). In each of the three human variants, the initiator methionine codon of the mouse ortholog is downstream of an alternative ATG codon, which serves as an alternative translational start site. Thus SLC4A11-A has an additional 62-amino acid (aa) Nt appendage, SLC4A11-B has an extra 35 aa, and SLC4A11-C has an extra 19 aa. Reports differ as to which variants are predominantly expressed in corneal endothelial cells, but most human corneal transcripts encode either SLC4A11-B or SLC4A11-C (20–22). Our finding that human SLC4A11-B has a more acidic pK_i than mSlc4a11, but only when the unique appendage is present, suggests that the appendage is responsible for this difference and therefore essentially acts as a conditional autostimulatory domain (e.g., in the summary of our experiments at pH_e = 8.50, shown in Fig. 3, the human protein has a greater G_m,max than the mouse protein but only over the pH_i range 6.9–7.3). The 35-aa appendage includes a number of titratable residues (Fig. 2A), but the mechanism of their individual influence upon pK_i will require further study. Others have noted that SLC4A11-C mediates a greater H⁺ flux than SLC4A11-B and thus the 19-aa appendage of SLC4A11-C may exert a similar influence on pK_i (20). The influence of an amino-terminal appendage on transport activity brings to mind the autostimulatory and autoinhibitory Nt appendages of the related SLC4 protein NBCe1 (23). However, NBCe1 is not known to be intrinsically pH dependent (24, 25), and so these domains may exert their effects upon NBCe1 activity in a different way. Another possibility is that, in vivo, the Nt appendages of SLC4A11 variants are binding sites for regulatory proteins, masking the influence of the domain upon pK_i, just as the binding of calcineurin B-homologous proteins (CHP) to the Na⁺/H⁺ exchanger NHE1 shifts the pK_i of NHE1 into physiological range (26)_.

It is interesting to consider the physiological relevance of the difference in pK_i between human and mouse SLC4A11. The corneal endothelial cell densities are similar in mice and humans and decrease with age in both species (27–29). However, the overall corneal thickness differs greatly between the two species, being around four times thicker in humans than in mice (30, 31). Thus, the same density of cells has to pump fluid from a drastically larger space. An acidic pK_i for H⁺ transport in human corneal endothelial cells could imply that the pH_i of human CECs is more acidic than that in mice or that the cytoplasmic buffering of CECs is less powerful in humans than in mice and subject to more dynamic excursions during bicarbonate and lactate pumping. However, to our knowledge, no such comparative measurements have been made.

The Effect of Mutants R125H and R804H on Slc4a11 Action

hSLC4A11/R125H has been reported by others to accumulate normally in the plasma membrane of mammalian cells (32, 33); thus the mutation is assumed to affect SLC4A11 action. The surface expression status of hSLC4A11/R804H is less clear, the result depending on the assay (33). We find that both mutants are capable of mediating substantial H⁺(OH⁻) conductance in oocytes when stimulated by a rise in pH_e. hSLC4A11/R125H mediates a robust cellular alkalinization at pH_e = 10.00 (Fig. 6C) comparable even to that mediated by hSLC4A11 at pH_e = 8.50 (Fig. 1A). The per-molecule G_m,max of both mutants appears to be at least 30% lower than wild-type hSLC4A11 at pH_e = 10.00, but this analysis could be complicated if the pH_e dependence of G_m,max is also influenced by the mutations. The relationship between pH and V_m is sub-Nernstian at pH_e = 10.00 (Fig. 4F and Fig. 6). Because of the relative scarcity of H⁺(OH⁻) compared to other ions, even a 20 mV/decade response of V_m to pH indicates a many thousandfold preference for H⁺(OH⁻) over other ions. The change in slope indicates either that hSLC4A11 and hSLC4A11/R125H become slightly less ideally selective to H⁺(OH⁻) at elevated pH_i or that these conditions activate an endogenous non-H⁺(OH⁻) permeability.

Hill fits of hSLC4A11/R125H data confirm our hypothesis that the mutation results in a disabling alkaline shift in pK_i and confirm that the R804H mutation also compromises hSLC4A11 action because of a smaller alkaline shift in pK_i. However, we have demonstrated that a rise in pH_e to 10.00 causes pK_i for both mutants to shift to values similar to those exhibited by wild-type hSLC4A11 at pH_e = 8.50 (and therefore presumably more acidic than the undefined wild-type hSLC4A11 pK_i at pH_e = 7.50); thus the mutants are capable of achieving a physiologically relevant pK_i. Therefore, we propose that the primary defect with both mutants lies not with pK_i per se but with the mechanism by which pH_e exerts an influence upon pK_i. Consequently, substantial wild-type character can be restored to the mutants by raising pH_e to a greater extent. Extracellularly applied DIDS, a compound that inhibits the related SLC4 protein AE1 by locking it in an outward-facing conformation (34), is also reported to partly restore H⁺(OH⁻) transport activity of hSLC4A11-C/R109H (the equivalent of R125H in that variant) in mammalian cells (4). Thus it is possible that DIDS mimics the effects of raising pH_e by titrating extracellular residues such as lysine and/or by inducing a similar conformational change. Despite testing a variety of protocols, we have been unable to demonstrate that DIDS shifts pK_i in our oocyte system (data not shown), perhaps because the alkaline pH_e that we require to determine pK_i interferes with the ability of DIDS to interact with SLC4A11 in the appropriate way. Nonetheless, our demonstration that there are means other than DIDS of restoring action to two hSLC4A11 mutants opens up new therapeutic possibilities.

Interactions between the Nt Appendage and R804H

In previous work we determined the pK_i of mSlc4a11/R774H, which, recalculated to be consistent with this study, is 7.28 ± 0.05 at pH_e = 8.50 (6). This value is +0.11 more alkaline than the equivalent value for wild-type mSlc4a11, whereas the R804H mutation in hSLC4A11 results in an average +0.27 alkaline shift in the same parameter. If we consider mSlc4a11/R774H to be the murine equivalent of hSLC4A11/Δ1–35/R804H, we would conclude that the effect of R774H/R804H is less potent in the absence of 1–35. Furthermore, because wild-type mSlc4a11 has a more alkaline pK_i at pH_e = 8.50 than wild-type hSLC4A11, the result is that mSlc4a11/R774H effectively has the same pK_i at pH_e = 8.50 as hSLC4A11/R804H (P = 0.06, unpaired 2-tailed t test). Again, if we consider mSlc4a11/R774H to be the murine equivalent of hSLC4A11/Δ1–35/R804H, we would conclude that the effect of Δ1–35 is negated in the presence of the R774H/R804H mutation. That is to say, these data indicate that R774/R804 and 1–35 act with synergy to influence pK_i at pH_e = 8.50. One possible explanation for this could be that both R804 and 1–35 contribute to the Nt-TMD interface such that, for example, the engagement of R804 with the Nt draws 1–35 into closer proximity to the TMD, where it can influence pK_i. Because of the likelihood that the pH_e dependence of pK_i is determined by the conformation of numerous intracellular elements in relation to the TMD, it seems likely that any element that disturbs the pH_e dependence of pK_i would have consequences for the ability of another element to exert its usual effect.

The Effect of pH_e upon pK_i

The paradoxical influence of intracellular NH₂-terminal domain residues on the pH_e dependence of transport recalls studies of the related Cl^–/HCO₃^– exchanger protein SLC4A2 (AE2), but those studies report a direct effect of intracellular mutations on the pK_e value that describes pH_e dependence of Cl^– transport (35, 36). It is not clear whether, if we had approached our experiments in a different way, we could define a pK_e for hSLC4A11-mediated H⁺(OH⁻) transport (enumerated as G_m), but our findings indicate that there may be a pK_e that describes the pH_e dependence of pK_i. Unfortunately, our approach provides too low of a resolution of pH_e versus pK_i data to determine the value of this pK_e or to determine statistically significant differences in this value between constructs. However, our data are most consistent with the notion that the predominant effect of the mutations is an alkaline shift in pK_e rather than a loss of pH_e sensitivity (Hill coefficient). If this is the case, then pK_e for hSLC4A11 pK_i would most likely fall within the range 7.50–8.50, whereas pK_e for hSLC4A11/R125H pK_i would most likely fall within the more alkaline range 8.50–10.00. Presumably the link between extracellular conditions and intracellular pK_i is communicated via conformational changes in the TMD. SLC4 proteins form dimers with dimer-dimer interfaces between TMDs and also between cytosolic Nt domains (34, 37, 38). For some proteins like NBCe1, the Nt is a regulatory binding partner of the TMD and is necessary for function (39, 40). For others, like AE2, the Nt is regulatory but dispensable for function (41). For SLC4A11, the Nt is a binding partner of the TMD that is necessary for (at least) H₂O transport action (42) and H⁺(OH⁻) conductance (B. N. Quade, E. J. Myers, and M. D. Parker, data not shown). In the absence of an SLC4A11 structure, or indeed of a full-length high-resolution structure of any SLC4 protein, we can only speculate as to how R804 and R125 might facilitate the communication of pH_e change to the intracellular environment. R804 is located in an intracellular loop and may directly communicate structural changes in the TMD to the intracellular portions of SLC4A11. On the other hand, R125 is located in the Nt, in a region that is not represented in crystal structures of isolated SLC4 Nt domains. Our three independent, de novo models of the hSLC4A11 Nt are somewhat divergent, but all place R125 in an alpha helix at the cytosolic Nt dimer interface, indicating that the contact between Nts expedites the communication of TMD structural change across the intracellular portions of the protein. If R125 and R804 affect the pH_e dependence of pK_i rather than pK_i per se, they are not themselves titratable residues that directly contribute to the value of pK_i but would be conceptually similar to teeth on a gear wheel that convey structural changes between the TMD and Nt domains. In such a model, a loss of pH_e sensitivity would be represented as a loss of teeth, which would lower the mechanical advantage of the engagement between the extracellular and intracellular portions of SLC4A11, requiring a greater change in pH_e to effect a similar change in pK_i. On the other hand, an alkaline shift in pH_e dependence that we hypothesize to be a better match for our data would be represented as an initial misalignment and disengagement of the gears between the Nt and TMD, as depicted in the cartoon in Fig. 10. Only when pH_e has caused a substantial TMD structural change can the Nt gear engage (perhaps at a different location on the TMD gear than normal) to produce a functional transport protein and convey structural change to the Nt, resulting in a pH_e-dependent change in pK_i.

Figure 10.

Speculative cartoon showing in concept how extracellular pH (pH_e) might influence pK value that governs intracellular pH dependence (pK_i). A: SLC4A11 is represented as a gear rack [transmembrane domain (TMD)] enmeshed with a pinion gear [NH₂-terminal domain (Nt)]. The second monomer of the dimer is not shown, but we imagine it directly behind the units shown, mirroring the actions shown, with contact points between TMDs and between Nts (e.g., the red pivot point shown in the Nt) that facilitate the coordination of conformational changes between monomers. The pink box reports the conformation of TMD at pH_e = 8.50 as well the pK_i = 7.1 that results from the conformation of the Nt engagement with the TMD. B: as we have no information about the nature of the conformational changes, for the sake of simplicity we represent pH_e-dependent conformational changes in the TMD as a leftward or rightward shift. A change in pH_e causes the conformation of the TMD to change, shifting the relationship between the Nt and TMD and consequently pK_i. C: in mutant SLC4A11 the Nt and TMD may be misaligned and/or poorly engaged. However, pK_i may be restored by an extreme pH_e shift. R804H, being in an intracellular TMD loop, may influence Nt-TMD interactions. R125H, being close to the Nt dimer interface, may influence the alignment of the Nt dimer and its ability to engage the TMD. Note that this scheme does not represent SLC4A11 activity.

GRANTS

This work was funded by NIH Grant NEI-EY028580 to M.D.P.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

B.N.Q. and M.D.P. conceived and designed research; B.N.Q. and A.M. performed experiments; B.N.Q. analyzed data; B.N.Q. and M.D.P. interpreted results of experiments; B.N.Q. and M.D.P. prepared figures; B.N.Q. drafted manuscript; B.N.Q. and M.D.P. edited and revised manuscript; B.N.Q., A.M., and M.D.P. approved final version of manuscript.