pH dependence of the Slc4a11-mediated H⁺ conductance is influenced by intracellular lysine residues and modified by disease-linked mutations

Abstract

SLC4A11 is the only member of the SLC4 family that transports protons rather than bicarbonate. SLC4A11 is expressed in corneal endothelial cells, and its mutation causes corneal endothelial dystrophy, although the mechanism of pathogenesis is unknown. We previously demonstrated that the magnitude of the H⁺ conductance (G_m) mediated by SLC4A11 is increased by rises in intracellular as well as extracellular pH (pH_i and pH_e). To better understand this feature and whether it is altered in disease, we studied the pH dependence of wild-type and mutant mouse Slc4a11 expressed in Xenopus oocytes. Using voltage-clamp circuitry in conjunction with a H⁺-selective microelectrode and a microinjector loaded with NaHCO₃, we caused incremental rises in oocyte pH_i and measured the effect on G_m. We find that the rise of G_m has a steeper pH_i dependence at pH_e =8.50 than at pH_e =7.50. Data gathered at pH_e =8.50 can be fit to the Hill equation enabling the calculation of a pK value that reports pH_i dependence. We find that mutation of lysine residues that are close to the first transmembrane span (TM1) causes an alkaline shift in pK. Furthermore, two corneal-dystrophy-causing mutations close to the extracellular end of TM1, E399K and T401K (E368K and T370K in mouse), cause an acidic shift in pK, while a third mutation in the fourth intracellular loop, R804H (R774H in mouse), causes an alkaline shift in pK. This is the first description of determinants of SLC4A11 pH dependence and the first indication that a shift in pH dependence could modify disease expressivity in some cases of corneal dystrophy.

INTRODUCTION

SLC4A11 is a H⁺ (or OH^–)-transporting membrane protein and is the only member of the SLC4 family of acid-base transporters that does not transport ${\text{HCO}}_3^{-}$ (20, 26). SLC4A11 is predominantly expressed in corneal endothelial cells and cochlear fibroblasts (18, 29). Mutations in SLC4A11 cause autosomal-dominant inheritance of progressive vision loss [Fuchs endothelial corneal dystrophy (FECD)], autosomal-recessive inheritance of congenital vision loss [congenital hereditary endothelial dystrophy (CHED)], and autosomal-recessive inheritance of congenital vision loss with progressive hearing loss (Harboyan Syndrome) (6, 30, 31). Slc4a11-null mice exhibit congenital corneal edema and accelerated hearing loss (8, 9, 18). The genotype-phenotype correlations are not well understood. One of the major challenges to understanding how mutations in SLC4A11 result in disease is our incomplete understanding of the physiological role of the protein. Numerous actions have been assigned to SLC4A11. Several early hypotheses such as sodium-borate cotransport (25), Na⁺-coupled pH regulation (10, 24), and ammonium transport (34), have since been revisited and refined. Thus the current hypotheses regarding the multiple actions that could alone, or in combination, explain the physiological importance of SLC4A11 now only include pH-sensitive H⁺ (or OH^–) transport with or without cotransport of ammonia (12, 13, 20, 25, 34), water transport (29), and extracellular matrix binding (19).

Corneal dystrophy involves the loss of endothelial cell function, sometimes even loss of endothelial cells themselves, and ultimately corneal swelling (edema) and loss of transparency (33). Several studies have focused on the role of SLC4A11 in maintaining the fluid-pumping function of the endothelial cells, an action that resists edema formation. The majority of the thickness of the cornea is accounted for by a proteoglycan layer known as the stroma. The endothelial cells are located on the posterior surface of the stroma facing the aqueous humor. The endothelial-cell layer is leaky, so the stromal matrix has a natural tendency to absorb water from the aqueous humor: an action that, if unopposed, would distend the matrix causing it to scatter light and become opaque. Endothelial cells pump osmolytes such as ${\text{HCO}}_3^{-}$ and lactate (and consequently fluid) from the stroma back into the aqueous humor and thereby maintain corneal clarity (10, 29). The driving force for the fluid pump is established by the coordinated action of a number of basolateral membrane transport proteins including the Na⁺-${\text{HCO}}_3^{-}$ cotransporter NBCe1-B and the H⁺-lactate cotransporter MCT1 (3). SLC4A11 is located in this same membrane (29). With observations of water flux mediated by SLC4A11, it has been proposed that the protein is directly responsible for the movement of water (17, 29). On the other hand, the pH-sensitive H⁺ transport function could, if it responded appropriately to challenges in intracellular pH (pH_i), allow the pump to function without disturbing pH_i, even in the absence of tight functional coupling between NBCe1-B and MCT1 (20, 22). That is to say that, if the activity of one outpaced the activity of the other, SLC4A11 could act to counter a disturbance of pH.

In a previous study we showed that the magnitude of SLC4A11-mediated H⁺ conductance is increased by raising extracellular pH (pH_e) at constant pH_i or by raising pH_i at constant pH_e but we did not parameterize these observations (20). [We also found that the application of NH₃/${\text{NH}}_4^{+}$ enhanced H⁺ conductance but, as the elicited conductance persists after NH₃/${\text{NH}}_4^{+}$ removal, concluded that this effect reflects an indirect effect upon SLC4A11 mediated by the influence of NH₃/${\text{NH}}_4^{+}$ on cell pH and membrane potential (20).] In the present study we aim to further our understanding of the importance of the pH dependence of SLC4A11-mediated H⁺ transport in health and disease by determining, for the first time, values for its intracellular pH dependence (pK_i) and pH sensitivity (N_app). We use these values as a benchmark for identifying determinants of pH dependence in wild-type Slc4a11 and for assessing disturbances of pH dependence in disease-linked mutants.

METHODS

Constructs.

The construction of mouse Slc4a11 cDNA in the Xenopus expression vector pGH19 and the mutant “R99H” (equivalent to human CHED mutant “R125H”) has been previously reported (20). The other mutants used in this study were generated using either the QuikChange II mutagenesis kit (Agilent Technologies, Santa Clara, CA) in the case of the K343Q mutant or the Phusion site-directed mutagenesis kit (Thermo Fisher Scientific, Waltham, MA) for all other mutants, using the primers shown in Table 1 according to manufacturers’ recommendations.

Table 1.

Primers and templates used in the generation of the mouse Slc4a11 mutants used in this study

	Sequence
K343Q (equivalent to human K374Q): template: Slc4a11.pGH19
FWD	5′-GCAAGAGCGTGGGCCAATATGTCACCACC-3′
RVS	5′-GGTGGTGACATATTGGCCCACGCTCTTGC-3′

SSQ (K337S, K339S, K343Q: equivalent to human K368S, K370S, K374Q): template: Slc4a11/K343Q.pGH19
FWD	5′-CATCATTGGGAGTAGCAGTAGCGTGGGCCA-3′
RVS	5′-CCATCAGTGAAGTCCATTGGGTACACAGGA-3′

Pre-TM1-del (Δ337–342, K343Q: equivalent to human Δ368–373, K374Q): template: Slc4a11/K343Q.pGH19
FWD	5′-GGCCAATATGTCACCACCACAC-3′
RVS	5′-AATGATGCCATCAGTGAAGTCC-3′

E368K (equivalent to human FECD mutant E399K): template: Slc4a11.pGH19
FWD	5′-CCTCAACGATAAGAACACAAACGGGGC-3′
RVS	5′-GATCCAAAAGCAATGGTCGGTAGGAG-3′

T370K (equivalent to human CHED mutant T401K): template: Slc4a11.pGH19
FWD	5′-CAACGATGAGAACAAAAACGGGGCC-3′
RVS	5′-AGGGATCCAAAAGCAATGGTCGGTAG-3′

Mutant codons introduced by the primers are underlined. FWD, forward; RVS, reverse.

Oocyte preparation.

Ovaries from Xenopus laevis (Xenopus Express Inc., Brooksville, FL) were extracted in accordance with Institutional Animal Care and Use Committee protocols approved at the University at Buffalo. Frogs were anesthetized with 0.2% tricaine solution until unresponsive to toe pinch. The ovary was harvested and the frog was euthanized by exsanguination. Ovaries were cut into 1-cm fragments and washed with filter-sterilized calcium-free normal Ringer solution (Ca-free NRS: 82 mM NaCl, 20 mM MgCl₂, 5 mM HEPES, and 2 mM KCl) for 3 × 5 min. To extract the oocytes from the ovarian connective tissue, the fragments were agitated in Ca-free NRS solution containing 2 mg/ml type-1A collagenase (Sigma-Aldrich, St. Louis, MO) until the follicles began to dissociate. Oocytes were washed with Ca-free NRS for 3 × 10 min, ND96 for 1 × 10 min, and finally in OR3 medium for 1 × 10 min (OR3: 14 g Leibovitz’s L-15 medium powder; Thermo Fisher Scientific), 5 mM HEPES, 20 mL 100× penicillin-streptomycin solution (Corning Inc., Corning, NY), pH 7.50, and 200 mosmol/kgH₂O. Cells were stored in fresh OR3 at 18°C.

cRNA preparation and injection.

Plasmids were linearized with NotI enzyme and purified with the MinElute PCR Purification kit (Qiagen). The linearized DNA was converted into capped cRNA using the T7 mMessage mMachine Transcription kit (Invitrogen, Carlsbad, CA) and purified using with the RNeasy MinElute Cleanup kit (Qiagen). RNA was quantified using a Nanodrop 2000 (Thermo Fisher Scientific). cRNA was injected into oocytes using a Nanoject III programmable injector (Drummond Scientific Co., Broomhall, PA).

Electrophysiology.

Xenopus oocytes were placed into a small chamber (no. RC-3Z, Warner Instruments, Hamden, CT) on an antivibration worktable (Vision IsoStation, Newport Corp., Irvine, CA). Oocytes were superfused with solution fed from syringe pumps (Harvard Apparatus, Holliston, MA) at a rate of 2 mL/min. Glass microelectrodes were pulled from borosilicate glass (no. BF200-156-10, Sutter Instrument, Novato, CA) using a model P-1000 micropipette puller (Sutter Instrument) to achieve a tip resistance, when filled with saturated KCl solution, of 0.1–2 MΩ. The oocyte was impaled with two such microelectrodes (current passing and voltage sensing) connected to an OC275 oocyte clamp (Warner Instruments, Hamden, CT) while an oocyte bath clamp (no. 725I, Warner Instruments) was used to clamp the bath potential to 0 mV. Current-voltage (I-V) relationships were obtained by clamping the cell membrane potential (V_m) at its spontaneous value (resting potential) and then from –160 mV to +20 mV in 20-mV steps for 100 ms, returning to resting potential for 100 ms in between each step. A third, H⁺-selective microelectrode, was introduced into the opposite side of the oocyte to the I-V microelectrodes and connected to a HiZ-223 dual channel electrometer (Warner Instruments) to measure intracellular pH (pH_i) as previously described (15, 20). I-V data were digitized and acquired using a Digidata 1550 with Clampex 10.4 software (Molecular Devices LLC, San Jose, CA). Continuous recordings of pH_i and V_m were acquired using custom software (written by Dale Huffman for Walter Boron’s laboratory at Case Western Reserve University, Cleveland, OH). To raise oocyte pH_i in H₂O-injected cells, or in Slc4a11-expressing cells that are not expressing robust Slc4a11 activity (i.e., Any Slc4a11 clone when pH_e =7.50, plus R99H when pH_e =8.50), the needle of a Nanoject II microinjector (Drummond Scientific Co.), primed with 1.2 M solution of sodium bicarbonate, was introduced into the oocyte before the start of the experiment. Oocytes were subject to repeated injections of 2.3 nL to effect a gradual rise in pH_i until the injections had no further effect on pH_i. During this time and between injections, when pH_i was not rapidly changing, I-V measurements were taken. Thus, for most experiments, the oocyte was impaled with a total of four glass capillaries. However, in cells expressing any Slc4a11 construct (except R99H) when pH_e =8.50, there is sufficient driving force and H⁺ permeability (due to Slc4a11 activity) to permit spontaneous Slc4a11-mediated alkalinization of oocyte pH_i. Thus NaHCO₃ injection was not required to raise pH_i in Slc4a11-expressing cells under these conditions; we merely expedited the process by voltage-clamping V_m at a value (–30 mV) more positive than the reversal potential for H⁺ (E_H) to bolster the driving force for H⁺ extrusion (according to the Nernst equation, E_H is always more negative than –58 mV considering the >1 pH unit gradient across the plasma membrane).

Electrophysiology solutions.

The basic recipe for ND96 bath solution used in electrophysiological experiments is 93.5 mM NaCl, 2 mM KCl, 1.8 mM CaCl₂, 1 mM MgCl₂, and 5 mM of buffer. For pH 7.50 solution (pH_e =7.50), the buffer was HEPES, while for pH 8.50 solution (pH_e =8.50) the buffer was Bicine.

Data analysis.

Conductance (G_m) was calculated as the slope of I-V plots from –120 mV to +20 mV using Microsoft Excel. The lowest voltages (–160 mV and –140 mV) were omitted from slope calculation due to a noticeable rectification in this range. We fit plots of pH_i versus G_m to the Hill equation (as defined in Ref. 23) using the solver function of Microsoft Excel.

$$\text{Hill\hspace{0.17em}equation:\hspace{0.17em}}\frac{G_{\text{m}}}{G_{\text{m,max}}}=\frac{1}{1+{\left(\raisebox{1ex}{${\text{EC}}_{50}$}\!\left/ \!\raisebox{-1ex}{$[{\text{OH}}^{-}]$}\right.\right)}^{N_{\text{app}}}}$$

Hill is normally used for ligand or agonist assays in which EC₅₀ is the concentration of agonist that elicits a half-maximal activity of the protein. In our case, we consider OH^– as our agonist (as it rises with pH). Our measure of pH dependence is pK_i = pH_i at 50% G_m,max = 14 – log(EC₅₀). N_app is the apparent Hill coefficient, which in simple reactions can indicate the number of binding sites for the agonist on the protein. However, the interpretation of the Hill coefficient is most precise in systems where cooperativity of binding increases with each subsequent binding event. In more complex reactions, for example, proteins with multiple binding sites or in our case dimeric membrane proteins with many charged residues with unknown importance in pH sensing, we may only conclude that N_app represents a minimum number of titratable residues (32) and note that this value can also be influenced by the strength of their cooperativity (27). Statistical analyses were performed either by t test with Bonferroni correction for multiple comparisons, ANOVA, or general linear model as appropriate using MiniTab 19 software.

RESULTS

Investigating the pH_i dependence of Slc4a11 activity at pH_e = 7.50.

We superfused Slc4a11-expressing or H₂O-injected-control oocytes with ND96 solution (pH_e =7.50) and gathered multiple I-V relationships from these cells by voltage clamp as we caused pH_i to increase by injecting 1.2 M NaHCO₃ solution. Figure 1A shows representative pH_i and V_m traces gathered during such an experiment. The black traces indicate data from an Slc4a11-expressing oocyte, while the gray traces indicate data from a H₂O-injected oocyte. Initially, we see that the Slc4a11-expressing oocyte has a more acidic pH_i and a more depolarized V_m than the H₂O-injected oocyte, consistent with our previous observations of Slc4a11-expressing cells (20). NaHCO₃ injections cause both cells to depolarize and alkalinize. Samples of the multiple I-V relationships gathered from the cells during the experiments shown in Fig. 1A are shown in Fig. 1B (the representative mSlc4a11-expressing cell) and Fig. 1C (the representative H₂O-injected cell). At their starting (i.e., most acidic) pH_i, the I-V plots for each cell look similar. However, as pH_i rises, the slope of the I-V relationship (i.e., the conductance, G_m) of the mSlc4a11-expressing cell greatly increases (Fig. 1B), as shown by the data plotted in Fig. 1D for all of the I-V plots gathered in this experiment (black circles). On the other hand, the conductance of the H₂O-injected cell does not greatly increase despite a comparable rise in pH_i (Fig. 1, C and D, gray circles). Figure 1E shows the pH_i versus G_m relationships for nine mSlc4a11-expressing cells (black circles). The gray dashed line represents the maximum value for G_m observed in any of the six H₂O-injected cells at their most alkaline (data points are omitted here for clarity but are presented on an expanded axis in Supplemental Fig. S1A; see https://doi.org/10.6084/m9.figshare.12400946). It is not clear, from any of the pH_i/G_m relationships gathered from individual Slc4a11-expressing cells, whether a maximum conductance (G_m,max) has been reached. However, estimating a minimum pK_i value for each cell (pH_i at which G_m is half of maximum observed G_m in each experiment) reports a range of 7.44–7.82. This indicates that pK_i for Slc4a11 at pH_e =7.50 is unlikely to be lower than 7.44.

Fig. 1.

The intracellular pH (pH_i) dependence of Slc4a11 activity when extracellular pH (pH_e) = 7.50. A: representative recording of pH_i and membrane potential (V_m) versus time as an oocyte expressing Slc4a11 (black trace) or a control H₂O-injected oocyte (gray trace) is injected with NaHCO₃ to raise pH_i. The scattered dots represent the excursions at which current-voltage (I-V) plots were taken (which was faster than our sampling rate) B: a sampling of current-voltage relationships (I-V plots) gathered from the Slc4a11-expressing cell during the experiment shown in A. C: a sampling of I-V plots gathered from the H₂O-injected cell during the experiment shown in A. D: the relationship between pH_i and membrane conductance (G_m) for the Slc4a11-expressing (black) and water-control oocyte (gray) calculated from all of the I-V plots gathered during the experiment shown in A. E: data similar to that shown in D for a greater number of Slc4a11-expressing oocytes. Gray dashed line shows maximum G_m calculated from a greater number of H₂O-injected oocytes (see Supplemental Fig. 1A for data points; https://doi.org/10.6084/m9.figshare.12400946).

Revisiting the pH_i dependence of Slc4a11 activity at pH_e 8.50.

In a previous study, we determined that the H⁺ transport activity of Slc4a11 was substantially more robust at pH_e =8.50 than at pH_e =7.50 (20), a property that we decided to exploit for the present study to observe Slc4a11 operating at G_m,max and calculate a definite value of pK_i for Slc4a11. Figure 2A shows data equivalent to that in Fig. 1A but gathered at pH_e =8.50. Figure 2B shows a sample of the I-V relationships collected during the experiment in Fig. 2A for the Slc4a11-expressing cell while it is clamped at –30 mV. Figure 2C shows a sample of the I-V relationships for the H₂O-injected cell while it was injected with NaHCO₃. As with the equivalent data gathered at pH_e =7.50, when pH_i is at its most acidic in each cell type, I-V plots are similar between Slc4a11-expressing and H₂O-injected cells but only in Slc4a11-expressing cells is G_m steeply dependent on pH_i (Fig. 2D). In contrast to data gathered at pH_e = 7.50, at pH_e = 8.50 Slc4a11 activity reached a plateau within our measured pH_i range (Fig. 2D, black circles), allowing us to define G_m,max. Although G_m for the H₂O-injected cell also increased gradually (Fig. 2D, gray circles), it did not rise above 9 μS. Figure 2E shows data similar to that shown in Fig. 2D but for a greater number of cells. The gray line indicates the maximum value of G_m seen in any of the six H₂O-injected oocytes tested at any pH_i value in our experiments at pH_e 8.50 (omitted data points are presented on an expanded axis in Supplemental Figure S1B; see https://doi.org/10.6084/m9.figshare.12400946). Figure 3A shows each Slc4a11 data set from Fig. 2D normalized to its own G_m,max. Figure 3B shows the lines of best-fit to the Hill equation for each data set (sum of squares = 0.07 ± 0.01). From these data, we calculate that the pH_i dependence of Slc4a11 at pH_e 8.50 is described by an average pK_i of 7.16 ± 0.01 and the pH_i sensitivity is described by an average N_app of 9 ± 1.

Fig. 2.

The intracellular pH (pH_i) dependence of Slc4a11 activity when extracellular pH (pH_e) is 8.50. A: representative recording of pH_i and membrane potential (V_m) versus time as an oocyte-expressing Slc4a11 (black trace) is clamped at −30 mV or a control H₂O-injected oocyte (gray trace) is injected with NaHCO₃ to raise pH_i. B: a sampling of current-voltage (I-V) plots gathered from the Slc4a11-expressing cell during the experiment shown in A. C: a sampling of I-V plots gathered from the H₂O-injected cell during the experiment shown in A. D: the relationship between pH_i and membrane conductance (G_m) for the Slc4a11-expressing (black) and water-control oocyte (gray) calculated from all of the I-V plots gathered during the experiment shown in A. E: data similar to that shown in D for a greater number of Slc4a11-expressing oocytes. Gray dashed line shows maximum G_m calculated from any of a greater number of water-control oocytes (see Supplemental Fig. 1B for data points; https://doi.org/10.6084/m9.figshare.12400946).

Fig. 3.

Hill equation fits of Slc4a11 activity at extracellular pH (pH_e) 8.50. A: each Slc4a11 trace from Fig. 2E is shown here normalized to its own maximum conductance (G_m). B: gray lines represent best fit to the Hill equation for each trace in A. The average values for intracellular pH dependence (pK_i) and apparent Hill coefficient (N_app) were used to generate the averaged best-fit trace shown as a black dotted line. pH_i, intracellular pH.

The pH_e dependence of Slc4a11 activity.

To facilitate direct comparison of the behavior of Slc4a11- versus H₂O-injected cells at pH_e =7.50 and 8.50, and probe the activities of Slc4a11 at fixed pH_i but varying pH_e, we binned pH_i versus G_m data for each cell from Fig. 1E and Fig. 2E into three pH_i ranges: 7.00–7.29 (low), 7.30–7.59 (mid), and 7.60–7.89 (high). These data are shown in Fig. 4A alongside data gathered from H₂O-injected cells and, for the first time in this manuscript, equivalent data gathered from the inactive Slc4a11 mutant R99H (extracted from data shown in Supplemental Fig. S2A and S2B; see https://doi.org/10.6084/m9.figshare.12400946). The results of statistical analysis of these data by general linear model are illustrated in the form of a Venn diagram in Fig. 4B in which data sets that do not share a group are significantly different (P < 0.05). From this conservative analysis, we note the following:

Fig. 4.

Comparison of intracellular pH (pH_i) dependence of Slc4a11 at extracellular pH (pH_e) =7.50 and 8.50. A: for each cell which has a data set shown in Fig. 1E or Fig. 2E, we binned conductance (G_m) values into three pH_i ranges such that each cell could be attributed a single average G_m value corresponding to a specific value of pH_e in each of the three pH_i ranges. These data are plotted alongside equivalent data gathered from the inactive Slc4a11 mutant R99H (see Supplemental Fig. 2 for data points; https://doi.org/10.6084/m9.figshare.12400946). Each data point represents an individual cell and is the average of between 1 and 25 G_m measurements that were incidentally gathered within that pH_i range. B: statistical analysis of data shown in A illustrated as a Venn diagram of the groups (I, II, III, and IV) identified by general linear ANOVA model. Data sets that do not share a common group are significantly different from one another.

• • All data sets gathered from H₂O-injected cells or R99H-injected cells (together: “control cells”) fall within group IV and are thus not significantly different from each other.
• • When pH_e = 7.50, G_m values gathered from Slc4a11-injected cells and from control cells are significantly different from each other but only when pH_i = “mid” or “high.”
• • When pH_e = 8.50, G_m values gathered from Slc4a11-injected cells and from control cells were significantly different from each other for any given category of pH_i.
• • G_m values gathered from Slc4a11-expressing cells at pH_e =7.50 are significantly greater when pH_i = “high” than when pH_i = “low” or “mid.”
• • G_m values gathered from Slc4a11-expressing cells at pH_e =8.50 are significantly greater when pH_i is “mid” or “high” than when pH_i is “low.”
• • When pH_i = “low” or “mid,” G_m values gathered from Slc4a11-expressing cells at pH_e = 8.50 are significantly greater than equivalent values gathered at pH_e =7.50.
• • When pH_i = “high,” G_m values gathered from Slc4a11-expressing cells at pH_e = 8.50 are not significantly different from equivalent values gathered at pH_e =7.50.

Thus, in summary, the G_m increases observed in Slc4a11-expressing cells are significantly greater than those observed in H₂O-injected cells, increase significantly with pH_i at either value of pH_e, and rise over a significantly lower pH_i range when pH_e is 8.50 compared with when is pH_e =7.50. However, whether pH_e is 7.50 or 8.50, with a high enough pH_i, G_m is ultimately the same, as if pH_e exerts its influence by altering pK_i rather than G_m,max.

To further investigate this phenomenon, we performed a series of Fig. 1-like experiments in which we probed the pH_i/G_m relationship at pH_e =7.00 (Fig. 5A) and a series of Fig. 2-like experiments in which we probed the pH_i/G_m relationship of pH_e 8.00 (Fig. 5B). As with data gathered at pH_e =7.50, when pH_e = 7.00 the relationship between pH_i and G_m did not reach a plateau and thus we can only estimate that the minimal pK_i at pH_e =7.00 is likely to be no lower than 7.50. On the other hand, when pH_e = 8.00, we could gather sufficient data for a fit to the Hill equation (Fig. 5C, sum of squares = 0.11 ± 0.03). When pH_e was 8.00, we calculated that pK_i = 7.23 ± 0.02 and N_app = 6 ± 1. Figure 5D shows the relationship between pH_e and pK_i (or our proxy: estimated minimum pK_i). pK_i at pH_e = 8.00 is significantly more alkaline than at pH_e 8.50 (P < 0.05, one-tailed unpaired t test), and N_app is reduced (P < 0.05, one-tailed unpaired t test). Overall, there is a general trend for our measures of pK_i to decrease as pH_e rises from neutral to alkaline.

Fig. 5.

Comparison of the intracellular pH (pH_i) dependence (pK_i) of Slc4a11 at extracellular pH (pH_e) 7.00 and 8.00. A and B: the relationship between pH_i and conductance (G_m) for Slc4a11-expressing oocytes at pH_e 7.00 (A) or pH_e 8.00 (B). C: gray lines represent best fit to the Hill equation for each trace in B. The average value for pK_i and apparent Hill coefficient (N_app) were used to generate the averaged best-fit trace shown as a black dotted-line. D: the range of pK_i values at each tested value of pH_e. pK_i values are calculated from Hill plots fits for pH_e 8.00 and 8.50 (solid circles), or estimated minimal pK_i values from pH_i at 50% of maximum observed G_m (open circles).

The influence of an intracellular lysine cluster on the pH dependence of Slc4a11.

Analysis of the SLC4A11 protein sequence reveals a previously unremarked cluster of lysine residues “K-x-K-x-x-x-K” at the cytosolic end of transmembrane span 1 (TM1) that is unique among SLC4 paralogs (Fig. 6A and underlined in Fig. 6B) and conserved among mammalian Slc4a11 orthologs (Fig. 6C). To assess the contribution of these residues to the pH dependence of Slc4a11, we created a series of mutant clones (Fig. 6D):

Fig. 6.

Slc4a11 residues mutated in this study. A: topological cartoon of Slc4a11 showing the 14 transmembrane spanning regions. Amino-acid residues pertinent to this study are circled and numbered according to their location in the human (black numbers) or mouse (gray numbers) polypeptide chain. The region that corresponds to the sequence alignments in B–D is emphasized. B: alignment of all human SLC4 family protein sequences showing a unique lysine-rich region in SLC4A11 (underlined) and 2 nearby residues mutated in corneal dystrophy (E399K and T401K, boxed). Gray region shows likely limits of the first transmembrane span C: alignment of a selection of vertebrate Slc4a11 sequences showing conservation of the sequence features highlighted in Fig. 5B. D: sequence of the mutants used in this study (altered regions are shown in bold).

• • “K343Q” in which we replaced only the lysine closest to TM1 with the glutamine that is found at the equivalent location in other SLC4 proteins.
• • “SSQ” in which we replaced all three lysines: the first two lysine residues (K337 and K339) with serine residues and the third with glutamine (per K343Q).
• • “pre-TM1-del” in which we deleted the cluster, replacing it with a juxtamembrane glutamine to mimic the pre-TM1 primary structure of other SLC4 proteins.

Due to the challenge associated with defining the parameters of pH_i dependence and sensitivity at pH_e =7.50, we performed our characterization of these mutants at pH_e 8.50. From data similar to those in Fig. 2A (not shown), we obtained series of I-V plots for each mutant as pH_i increased, a selection of which, from representative cells expressing either K343Q, SSQ, or pre-TM1-del, are shown in Fig. 7, A–C. The pH_i/G_m relationships for a larger number of cells from each group are shown in Fig. 7, D–F. Those data, normalized to the G_m,max for each cell, are shown in Fig. 7, G–I. When we fit these data to the Hill equation (Fig. 8: average sum of squares = 0.09 ± 0.01, 0.06 ± 0.02, and 0.09 ± 0.02 respectively), we observe that all mutants exhibit a significantly more alkaline pK_i compared with wild-type Slc4a11, but only SSQ and pre-TM1-del exhibit a significantly lower N_app (P < 0.016, i.e., P < 0.05/3 comparisons, one-tailed unpaired t test with Bonferroni correction). Supplemental Fig. S3, A–C (see https://doi.org/10.6084/m9.figshare.12400946), shows data equivalent to that shown in Fig. 7, D–F, but gathered at pH_e =7.50. These data do not allow us to determine a value for pK_i at pH_e =7.50 but, mirroring the situation at pH_e 8.50, we find that the estimated minimum pK_i values for these mutants at pH_e 7.50 tend to be more alkaline than the estimated minimum pK_i values for wild-type Slc4a11 at pH_e =7.50 (see Supplemental Fig. S3D; https://doi.org/10.6084/m9.figshare.12400946).

Fig. 7.

The influence of an intracellular lysine cluster on the intracellular pH (pH_i) dependence of Slc4a11 at extracellular pH (pH_e) 8.50. Three Slc4a11 mutants K343Q, SSQ, and pre-TM1-del (per Fig. 6D) were tested for their response to changes in intracellular pH (pH_i). A–C: a sampling of current-voltage (I-V) plots gathered from a representative mutant-Slc4a11-expressing cell during the experiments equivalent to that shown in Fig. 2A. D–F: the relationship between pH_i and membrane conductance (G_m) for a greater number of mutant-Slc4a11-expressing oocyte calculated from all of the I-V plots gathered for each cell. G–I: each Slc4a11 trace from Fig. 6, D–F, is shown here normalized to its own maximum conductance. V_m, membrane potential.

Fig. 8.

Hill equation fits of intracellular lysine mutants of Slc4a11 activity at extracellular pH (pH_e) 8.50. The best fit to the Hill equation for each of the traces in Fig. 7, G–I, was used to determine a range of values for intracellular pH (pH_i) dependence (pK_i) and apparent Hill coefficient (N_app) for each mutant. The average value for these constants was used to generate the averaged best-fit traces shown in the graph. *Significant difference from wild-type values (P < 0.016, one-tailed, unpaired t test with Bonferroni correction).

The influence of disease-causing mutants on the pH dependence of Slc4a11.

While the lysine-deficient mutants K343Q, SSQ, and pre-TM1-del are artificial constructs that have never been observed in a living subject, there are two disease-causing mutations on the extracellular side of TM1 that result in the introduction of an additional lysine residue: E399K (a dominant mutation linked to FECD) and T401K (a recessive mutation linked to CHED). The mutated residues (Fig. 6A and boxed in sequence alignment in Fig. 6B) are conserved among SLC4 paralogs (“D/E-x-T”) and among Slc4a11 orthologs (Fig. 6, B and C). We recreated these mutants in our mouse clone (E368K and T370K in Fig. 6D) and examined the transport properties of the mutant transporters at pH_e = 8.50. From data similar to those in Fig. 2A (not shown), we obtained series of I-V plots for each mutant as pH_i increased, a selection of which, from representative cells expressing either E368K or T370K, is shown in Fig. 9, A and B. The pH_i/G_m relationships for a larger number of cells from each group are shown in Fig. 9, D and E. Finally, those data, normalized to the G_m,max for each cell, are shown in Fig. 9, G and H. A third mutation R804H (recessive, CHED-linked: equivalent to mouse R774H) caught our attention due to its alteration of a titratable residue in an intracellular juxtamembrane loop (Fig. 6A). In Fig. 9, C, F, and I (note the change of scale in Fig. 9F), we show the relationship between pH_i and G_m for R774H at pH_e 8.50. When we fit the data for all three mutants to the Hill equation (Fig. 10: average sum of squares = 0.02 ± 0.01, 0.07 ± 0.02, and 0.06 ± 0.02, respectively), we find that E368K and T370K have a significantly more acidic pK_i compared with wild-type Slc4a11, while R774H has a significantly more alkaline pK_i compared with wild-type Slc4a11 (P < 0.016, one-tailed unpaired t test with Bonferroni correction) with no significant alteration in the N_app in any case (P > 0.016, one-tailed unpaired t test with Bonferroni correction). Although the low G_m,max of E368K and T370K precluded their study at pH_e =7.50, the G_m,max for R774H was not different from that of wild-type (P = 0.44, two-tailed unpaired t test) and thus we were able to gather data from R774H-expressing oocytes at pH_e =7.50 (Fig. 11, A and B). Mirroring the situation at pH_e 8.50, we find that the estimated minimum pK_i value for R774H tends to be more alkaline than the estimated minimum pK_i values for wild-type Slc4a11 at pH_e =7.50 (Fig. 11C).

Fig. 9.

The influence of three dystrophy-causing mutations on the intracellular pH (pH_i) dependence of Slc4a11 at extracellular pH (pH_e) 8.50. Three mutants (E369K, T370K, and R774H: equivalent to human dystrophy-causing mutants E399K, T401K, and R804H) were tested for their response to changes in pH_i. A–C: a sampling of current-voltage (I-V) plots gathered from representative mutant-Slc4a11-expressing cells during experiments equivalent to those shown in Fig. 2, A, D, E, and F: The relationship between pH_i and membrane conductance (G_m) for a greater number of mutant-Slc4a11-expressing oocytes drawn from all of the I-V plots gathered for each cell. Gray dashed line shows maximum G_m calculated from any of a greater number of water-control oocytes (from Supplemental Fig. 1B: https://doi.org/10.6084/m9.figshare.12400946). G–I: each Slc4a11 trace from D–F is shown here normalized to its own maximum conductance. V_m, membrane potential.

Fig. 10.

Hill equation fits of three dystrophy-linked mutants of Slc4a11 activity at extracellular pH (pH_e) 8.50. The best-fit to the Hill equation for each of the traces in Fig. 9, G–I was used to determine a range of values for intracellular pH (pH_i) dependence (pK_i) and apparent Hill coefficient (N_app) for each mutant. The average value for these constants was used to generate the averaged best-fit traces shown in the graph. *Significant difference from wild-type values (P < 0.016, one-tailed, unpaired t test with Bonferroni correction).

Fig. 11.

The influence of R774H on the intracellular pH (pH_i) dependence (pK_i) of Slc4a11 at extracellular pH (pH_e) =7.50. A: a sampling of current-voltage (I-V) plots gathered from representative R774H-expressing cells during experiments equivalent to that shown in Fig. 1A. B: the relationship between pH_i and membrane conductance (G_m) for a greater number of R774H-expressing oocytes drawn from all of the I-V plots gathered for each cell. Gray dashed line shows maximum G_m calculated from any of a greater number of water-control oocytes (from Supplemental Fig. 1A: https://doi.org/10.6084/m9.figshare.12400946). C: the range of pK_i values for wild-type and R774H at each tested value of pH_e. pK_i values are calculated from Hill plots fits (pH_e 8.50) or estimated minimal pK_i values from pH_i at 50% of maximum observed G_m (pH_e =7.50). V_m, membrane potential.

DISCUSSION

The pH dependence of wild-type Slc4a11.

Our data are consistent with a scheme in which the pH_i dependence of the Slc4a11 H⁺ conductance is governed by a single pK_i, the value of which is dependent on pH_e. The more alkaline pH_i becomes, the greater G_m becomes (e.g., Fig. 1E) but the more acidic pH_e becomes, the more alkaline pK_i becomes (Fig. 5D) and the lower G_m becomes. For cells expressing Slc4a11, there are obvious advantages to increasing G_m as pH_i increases; this would tend to promote H⁺ influx to counter local rises in pH_i. We have previously postulated that the role of Slc4a11 is to counter rises in pH_i caused by NBCe1 action, maintaining the transcellular passage of ${\text{HCO}}_3^{-}$ with minimal challenge to local pH_i at the basolateral membrane. However, what of the dependence on pH_e? Any acidic shift in pH_e might tend to cause a slight but counterintuitive dampening of G_m. However, this may be a minor drawback for the broader benefit of pH_e dependence as this also means that G_m is lowered in the face of pathological extracellular acidosis (while pH_i is normal) protecting pH_i from the extracellular insult. In this way, Slc4a11 could respond acutely to counteract a rise in pH_i while resisting the tendency to cause intracellular acidosis in the presence of an inwardly directed H⁺ gradient. This would be consistent with its hypothesized role in pH_i regulation and would be another point of differentiation from H⁺ channels such as H_v1, which do sense H⁺ gradients but have a pH-gradient- and voltage-dependent directionality that favors H⁺ extrusion (4, 5). The pH_e dependence of Slc4a11 might also help to keep the H⁺ conductance quiescent as the protein progresses through the secretory pathway, the lumen of which is acidic and equivalent to the extracellular space.

We encountered some unexpected technical issues during our study that precluded us from defining a value for pK_i at pH_e = 7.50. Under this condition, pK_i is so alkaline that we could not raise oocyte pH_i high enough to establish the requisite pH_i/G_m plateau that would allow us to fit data to the Hill equation. That is not to say that Slc4a11 is inactive, or operating in a different mode, under these conditions (see Supplemental Fig. S4 and S5; see https://doi.org/10.6084/m9.figshare.12400946), but it is likely to be only operating at a fraction (we estimate ~10%) of its maximum capacity at rest due to the substantially more alkaline pK_i. It is for this technical reason that we can only perform statistical comparisons among values for pK_i and N_app that we gathered at pH_e 8.50. We can, however, use data gathered at pH_e =7.50 to read off a minimal value of pK_i based on the highest observed value of G_m in each experiment. We are wary of using these values in statistical comparisons as the sets necessarily include a greater variability than the values gathered at pH_e 8.50 because they are influenced by the expression level of Slc4a11 in each cell and do not have any precise physiological meaning. However, these data are still valuable as they show that the statistically significant differences in pK_i observed at pH_e 8.50 are preserved, at least in trend (with the sole exception of “pre-TM1-del”), at the physiological value of pH_e =7.50. For these same technical reasons, we are also unable to ascertain the nature of the mathematical relationship between pH_e and pK_i, beyond the general trend (Fig. 5D).

Determinants of pH dependence.

The Hill coefficient for Slc4a11 indicates that the pH_i dependence of H⁺ conductance is governed by more than one titratable residue. Indeed, structure-function studies of a related acid-loading protein the Cl⁻/${\text{HCO}}_3^{-}$ exchanger AE2 (SLC4A2), which is also activated by rises in pH_i and pH_e (14), show that the pH_i dependence of Slc4 proteins can be complex and involve interactions between numerous pH-sensitive regions, both intracellular and extracellular (28, 35). However, Slc4a11 bears no more similarity to AE2 than it does to any other, less pH-sensitive Slc4 ortholog. Instead, sequence alignments of Slc4 sequences drew our attention to a unique cluster of lysine residues in Slc4a11 near to TM1 which is reminiscent of a critical lysine reside that is a located at the intracellular end of TM1 of the pH-dependent renal outer medullary K⁺-channel (ROMK, K_ir 1.1) (7). ROMK is activated by acidic pH with a pK of ~6.8, but when the lysine is mutated to methionine, the result is such a pronounced acidic shift in pK (pK ~5.3) that it effectively renders ROMK constitutively active. Moreover, the mutation causes the Hill coefficient to drop from 3.9 to 0.5 (7). In the case of Slc4a11, the effect of removing lysines from this region is a significant alkaline shift in pK_i, but the change is less dramatic than with ROMK, presumably because Slc4a11 includes numerous other critical pH-sensitive determinants of transport that have yet to be identified. Thus the unique juxta-TM1 lysine residues influence the pH dependence of Slc4a11 but are not required for the unique H⁺ conducting action of Slc4a11. Finally, in regard to why lysine replacement in Slc4a11 might cause an alkaline shift in pK_i rather than the acidic shift observed for ROMK, we believe that this reflects the complex cooperativity among the multiple determinants of pH dependence that are likely to be different between the two proteins. The effect upon Slc4a11 pK_i of other amino-acid substitutions at this location has not been tested but could reveal the extent of the influence of K343, and other lysine residues in this region, upon pK_i. However, it is possible that not all substitutions will be well tolerated; for example, in the related transporter AE1 (SLC4A1), introducing an alanine at the K343-equivalent position (Q404) compromises cell surface presentation (11). Finally we note that, although our study was not powered to assess changes in ion selectivity of the mutants, our data are not suggestive of such a phenomenon (see Supplemental Figures S5; see https://doi.org/10.6084/m9.figshare.12400946).

The contribution of altered pH dependence to disease.

The potential for loss of function under physiological conditions that could be caused by an alkaline shift in pK_i led us to examine a few select disease-causing mutants for alterations in pK_i. E399K (E368K in our mouse clone) and T401K (T370K in our mouse clone) were of interest because they represent the extracellular, juxtamembrane introduction of lysine residues, providing a counterpoint to our intracellular lysine replacement experiments at the intracellular end of TM1. Both E386K and T370K exhibit a significant acidic shift in pK_i, demonstrating the principal that disease-causing mutations can influence pH dependence. Although a pK_i shift in either direction could disrupt a finely balanced regulatory mechanism, it is unlikely that the pK is a major etiological factor in these examples; a greater issue is that both mutants are significantly retained in the endoplasmic reticulum (ER) when expressed in mammalian cells (1). However, this shift in pK does raise a couple of interesting speculative considerations. 1) If some Slc4a11 mutants are effectively hyperactive at physiological pH, small molecule chaperones that increase plasma membrane expression (2) may restore adequate H⁺ conductance with less rescued protein. (It would also be interesting to learn whether such chaperones, in aiding protein folding, normalize pK_i.) 2) An ER-retained mutant with sufficiently acid-shifted pK_i might allow H⁺ leak from the ER as the alkaline shift imposed up pK_i by the relative acidity of the lumen (~7.1, equivalent to pH_e) may no longer be sufficient to move pK_i out of physiological pH_i range and hold Slc4a11 inactive; a gain of function that could contribute to the dominance of disease in some cases. However, to be clear, we have no data that speak to the activity of either mutant at pH_e <8.50.

R804H (R774H in our mouse clone) caught our attention as a candidate mutation that might raise pK_i as, like our lysine-substituted constructs, it represents the replacement of a basic amino acid residue in an intracellular, juxtamembrane region. We note however, that three-dimensional structural models of SLC4 proteins (e.g., PDB ID: 6CAA, 4YZF) indicate that this arginine is not located in the immediate vicinity of the pre-TM1 lysines. In the case of R774H, we observed an alkaline shift in pK_i that would likely cause a greater than 50% decrease in G_m under physiological conditions, an alteration that could contribute to disease etiology. G_m,max for R774H is the same as for wild-type Slc4a11 in our experiments, but whether this altered pK is the only functional defect with R774H is unknown. We are unaware of any assessment of H₂O transport by this mutant, and it is unclear whether the mutant protein is substantially withheld from the plasma membrane in mammalian cells. This latter issue has been preliminarily assessed in mammalian cells by two methods as part of a wider screen of 54 SLC4A11 mutants; the data regarding R804H were not altogether consistent and will require further verification (1). One method (a bioluminescence resonance energy transfer assay that reports colocalization with plasma membrane expressed K-Ras) suggested normal plasma membrane abundance, while a second method (quantification of the fraction of SLC4A11 with mature glycosylation, considered by those authors to be the more reliable of the two assays) suggested a 50% decrease in plasma membrane abundance. Reduced plasma-membrane expression or not, a substantial reduction in G_m would be anticipated to impact the expressivity of disease and also compromise the usefulness of therapeutic strategies to enhance plasma membrane expression for this mutant. R804H is one of several arginine to histidine mutations that have been reported in Slc4a11. Many of these mutants are intracellularly retained, but there is one particularly notable exception, R125H (R99H in mice, see Fig. 6A) that has been used as a control in many experiments, including in the present study, due to its lack of H₂O or H⁺ transport activity despite a normal plasma membrane abundance. The reason behind the functional importance of R125 is unknown, but our study raises the possibility that R125/R99 is a major determinant of pK_i and that the pK_i of R125H/R99H is so alkaline shifted as to render the mutant protein inactive under experimental conditions. Three pieces of evidence speak to this possibility: 1) we have demonstrated that a similar intracellular arginine to histidine substitution R774H causes an alkaline shift in pK_i; 2) in some experiments performed at pH_e 8.50 (when pK_i would be at its most acidic), the trend of the pH_i versus G_m relationship for R99H suggests that its activity could exceed that of H₂O-injected cells if pH_i was sufficiently alkaline (see Supplemental Figure S2B: see https://doi.org/10.6084/m9.figshare.12400946); and 3) H⁺ transport by R125H (R109H in that study) is partially rescued by preincubation with DIDS (12), a compound that derivatizes extracellular lysine residues. This phenomenon that could be explained if the reaction disrupts the relationship between pH_e and pK_i, causing a normalizing of pK_i.

Ultimately, the pathogenesis of endothelial dystrophy is complex and involves interplay among multiple stresses (e.g., oxidative stress, unfolded protein response, and mitophagy) (21). There may even be numerous mechanisms by which Slc4a11 mutation can lead to disease such as loss of function (16, 20, 29), the accumulation of misfolded SLC4A11 in the endoplasmic reticulum leading to unfolded protein response (30), or loss of extracellular matrix-interacting determinants (19). Our study has suggested a mechanism by which some mutations can result in loss of function, raises some considerations for personalization of therapeutic intervention, and increases our understanding of the structure-function relationships in this unusual Slc4 protein.

GRANTS

This work was supported by National Eye Institute R01 award 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. and M.D.P. 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. and M.D.P. 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.