Bicarbonate-Independent Sodium Conductance of Na/HCO₃ Cotransporter NBCn1 Decreases NMDA Receptor Function

Abstract

The sodium bicarbonate cotransporter NBCn1 is an electroneutral transporter with a channel activity that conducts Na⁺ in a HCO₃^–-independent manner. This channel activity was suggested to functionally affect other membrane proteins which permeate Na⁺ influx. We previously reported that NBCn1 is associated with the NMDA receptors (NMDARs) at the molecular and physiological levels. In this study, we examined whether NBCn1 channel activity affects NMDAR currents and whether this effect involves the interaction between the two proteins. NBCn1 and the NMDAR subunits GluN1A/GluN2A were expressed in Xenopus oocytes, and glutamate currents produced by the receptors were measured using two-electrode voltage clamp. In the absence of CO₂/HCO₃^–, NBCn1 channel activity decreased glutamate currents mediated by GluN1A/GluN2A. NBCn1 also decreased the slope of the current–voltage relationships for the glutamate current. Similar effects on the glutamate current were observed with and without PSD95, which can cluster NBCn1 and NMDARs. The channel activity was also observed in the presence of CO₂/HCO₃^–. We conclude that NBCn1 channel activity decreases NMDAR function. Given that NBCn1 knockout mice develop a downregulation of NMDARs, our results are unexpected and suggest that NBCn1 has dual effects on NMDARs. It stabilizes NMDAR expression but decreases receptor function by its Na⁺ channel activity. The dual effects may play an important role in fine-tuning the regulation of NMDARs in the brain.

1. Introduction

NBCn1 (SLC4A7) moves Na⁺ and HCO₃⁻ across cell membranes and regulates cellular pH and transepithelial HCO₃⁻ transport in many cells [1,2,3]. The movement occurs with the stoichiometry of 1 Na⁺ to 1 HCO₃⁻, such that both ions enter cells without altering cell membrane potential [4]. In addition to the cotransport, NBCn1 also has a channel activity that produces a HCO₃⁻-independent Na⁺ conductance [4,5]. This conductance increases intracellular Na⁺ levels ([Na⁺]_i) and causes the resting membrane potential to shift positively. The channel activity is not uniquely reported in NBCn1, as channel activities with distinct ion selectivity are also observed in other members of SLC4A [6,7,8] and SLC26A bicarbonate transporters [9,10]. Due to its Na⁺ conductance, NBCn1 channel activity is expected to alter a Na⁺ influx and [Na⁺]_i, which is critical for neuronal function in the nervous system. Nonetheless, the exact functional role of this activity is presently unclear.

In most neurons, NBCn1 is mainly localized to postsynaptic membranes, where it interacts with a variety of proteins to form a macromolecular complex [11,12,13]. Among the proteins capable of interacting with NBCn1 is the postsynaptic scaffolding protein PSD95 that recruits many postsynaptic proteins including N-methyl-D-aspartic acid receptors (NMDARs) [13]. Animal studies show that NBCn1 and NMDARs appear to be tightly associated with each other, such that a change in NBCn1 levels leads to a change in NMDAR levels [14]. NBCn1 knockout mice develop a downregulation of the NMDAR subunit GluN1 and PSD95 and are resistant to Mg²⁺/NMDA-mediated seizure and excitotoxicity [14]. The downregulation of NMDARs in these mice may also be responsible for animals’ increased alcohol consumption [15], given the importance of glutamatergic transmission for the reward pathway of the brain [16,17]. The downregulation of NMDARs in NBCn1 knockout mice is also consistent with the report [18] that NBCn1 upregulation, induced by acidification, correlates with NMDA-mediated neuronal damage in rat hippocampal primary cultures. NBCn1 is an acid-base regulator, and NMDARs are sensitive to pH; thus, from the perspective of pH physiology, the positive correlation between the two proteins is an advantage for proper regulation of the receptors. On the other hand, it is unclear whether NBCn1 channel activity is also involved.

In this study, we examined the effects of NBCn1 channel activity on NMDAR currents and the contribution of PSD95 to those effects. We expressed GluN1A/GluN2A (hereafter, GluN1·N2) and NBCn1, as well as PSD95, in Xenopus oocytes and tested whether currents evoked by glutamate (I_Glu) were affected by NBCn1 and PSD95. Oocytes were chosen for the expression system as multiple proteins were simultaneously expressed in single cells, which allowed our comparison analysis to be more reliable than other expression systems. The results show that NBCn1 channel activity decreased I_Glu mediated by GluN1·N2, and this decrease can serve to compensate an I_Glu increase by PSD95. The decreasing effects were observed at comparable levels with and without PSD95. Given the manifestation of NMDAR excitotoxicity and abnormal activity in neurodegenerative diseases, such as Alzheimer’s disease [19], we propose that stimulating NBCn1 channel activity in neurons might be a valuable approach to reducing excessive glutamate transmission under pathological conditions.

2. Results

2.1. NBCn1 Channel Activity Increases the Baseline Conductance and Positively Shifts the Resting Membrane Potentials of Xenopus Oocytes

Figure 1A shows current–voltage (I–V) relationships for the baseline conductance in oocytes expressing NBCn1 or water-injected controls, measured at holding potentials from –120 to +60 mV in 20 mV steps. Experiments were performed in CO₂/HCO₃⁻-free ND96 solution containing 96 mM NaCl, as the channel activity is independent of bath HCO₃⁻. Compared to the control, oocytes expressing NBCn1 produced a higher slope of the I–V plot (n = 5–7/group), a hallmark for its channel activity [4,5]. The inward currents at negative voltages are due to the Na⁺, whereas the outward currents at positive voltages are due to other ions caused by the Na⁺ influx [4,5]. Figure 1B shows the Na⁺ component in the I–V plot, obtained from the difference before and after Na⁺ removal from the bath. Inward currents progressively increased at more negative voltages and the reversal potential approached to +60 mV, close to the reversal potential for a Na⁺ channel. In addition, NBCn1 channel activity caused the resting membrane potentials to be more positive (p < 0.01; Figure 1C).

2.2. NBCn1 Channel Activity Decreases I_Glu Produced by GluN1·N2

To examine whether the above activity affects NMDAR function, we expressed GluN1·N2 with or without NBCn1 in oocytes and compared I_Glu between the two groups. Figure 2A shows a representative I_Glu evoked by 100 µM glutamate (plus 10 µM glycine, no Mg²⁺) in an oocyte expressing GluN1·N2. A rapid inward current followed by a steady-state current was recorded, characteristic for GluN1·N2. Interestingly, I_Glu was decreased in the coexpression of NBCn1 (Figure 2B–D). The decrease was small at 1 ng of NBCn1 cRNA for injection, but substantial at 5 and 10 ng for injection. The membrane potentials were progressively more positive at higher amounts of NBCn1 cRNAs, consistent with increasing channel activities (data not shown). A control had no I_Glu (Figure 2E). Comparison of the mean I_Glu obtained from multiple oocytes confirmed this inhibitory effect of NBCn1 on GluN1·N2 (p < 0.05, one-way ANOVA, n = 4–5/group; Figure 2F).

The decreasing effect of NBCn1 on I_Glu was further evaluated by I–V relationships. Figure 2G shows the I–V plots for the responses evoked by glutamate in oocytes expressing GluN1·N2 (n = 6). The response yielded a plot that is characteristic for glutamate I–V; that is, while currents had a small increase in amplitude as the membrane is hyperpolarized, there were markedly increased currents as the membrane is depolarized, producing a large increase in the slope conductance (measured at zero-current voltage). Figure 2H shows the I–V plots in oocytes expressing GluN1·N2/NBCn1 (n = 5). As expected, NBCn1 channel activity increased inward currents at negative potentials in the baseline I–V plot. Glutamate application produced I_Glu; however, the magnitude was small, compared to that for GluN1·N2 alone in Figure 2G. Thus, the difference between the two I–V plots before and after glutamate application (i.e., I_Glu–V relationships) was smaller when NBCn1 was coexpressed, as shown in Figure 2I. Comparison of the I_Glu–V plots between GluN1·N2 and GluN1·N2/NBCn1 resulted in a significant reduction in the slope by NBCn1 (F_9,90 = 26.34, p < 0.01, two-way ANOVA repeated measures). The control with NBCn1 alone showed negligible I_Glu (n = 4). Together, these results demonstrate that NBCn1 channel activity decreases I_Glu mediated by GluN1·N2.

2.3. NBCn1 Channel Activity Compensates an I_Glu Increase Induced by PSD95

NBCn1 can cluster with NMDARs via PSD95 [11,12]. As PSD95 can affect both NMDARs [20] and NBCn1 [13], its contribution to the abovementioned NBCn1-mediated effects on I_Glu needs to be addressed. For this task, we performed two sets of experiments. In the first set of experiments, we examined the effect of PSD95 on I_Glu produced by GluN1·N2. As shown in Figure 3, PSD95 increased I_Glu and induced a steeper slope of the I_Glu–V relationships. The two I_Glu–V plots in Figure 3C crossed at the reversal potential, indicating that the current increase was mediated by GluN1·N2, not by any other current source. The result is consistent with the report [21] that PSD95 enhances NMDAR function by increasing the rate of channel insertion to membrane.

In the second set of experiments, we compared I_Glu produced by GluN1·N2/NBCn1 vs. GluN1·N2/NBCn1/PSD95. Interestingly, both groups produced similar responses to glutamate application, and their I_Glu–V plots were similar (Figure 4). The reversal potentials were slightly separated, the reason of which is unclear, although we think the separation might be related to the difference between non-clustered vs. clustered protein complexes. Regardless, the two plots were nearly superimposed, indicating no PSD95 effect on I_Glu when NBCn1 was present. Together, the results from the two sets of experiments demonstrate that NBCn1 channel activity counteracts the effect of PSD95 on NMDAR function.

2.4. I_Glu Decrease by NBCn1 Channel Activity Also Occurs in the Presence of CO₂/HCO₃⁻

The above experiments were performed in the absence of CO₂/HCO₃⁻. To examine whether the decreasing effect of the channel activity on I_Glu also occurs in the presence of CO₂/HCO₃⁻, we measured I_Glu before and 10 min after applying a solution equilibrated with 10% CO₂, 50 mM HCO₃⁻ at constant pH 7.4 and compared them. Except when glutamate was applied, Na⁺ in CO₂/HCO₃⁻ solution was replaced with N-methyl glucamine to minimize the cotransport activity, which changes pH_i and might complicate the results. Figure 5A shows representative I_Glu traces produced by GluN1·N2/PSD95. The I_Glu with similar amplitudes were observed before and after applying CO₂/HCO₃⁻. Figure 5B shows I_Glu traces from a parallel experiment with GluN1·N2/PSD95/NBCn1. The I_Glu was smaller than that for GluN1·N2/PSD95, which is expected from the inhibitory effect of NBCn1. Nonetheless, I_Glu were similar before and after applying CO₂/HCO₃⁻, comparable to those for GluN1·N2/PSD95. Consistently, the mean I_Glu from multiple oocytes resulted in no significant difference before and after CO₂/HCO₃⁻ application within groups (p > 0.05, paired two tailed Student t-test; n = 4–6/group; Figure 5C,D). NBCn1 decreased I_Glu by 43 ± 7% and 49 ± 9% in the absence and presence of CO₂/HCO₃⁻, respectively (Figure 5E). Conclusively, NBCn1 channel activity decreased I_Glu regardless of bath CO₂/HCO₃⁻.

3. Discussion

In this study, we found that NBCn1 channel activity decreases I_Glu produced by GluN1·N2 function. The channel activity induces 45–60% of the expected membrane potential change per Na⁺ decade based on the Nernst equation [4,5]. External K⁺, Mg²⁺, Ca²⁺, or internal Cl⁻ do not affect the activity [5]. This leads to the possibility that roughly half of the expected Nernst value reflects two binding sites for Na⁺, rather than one Na⁺ and an additional ion contributing the remaining half of the conductance. A hint comes from the ion transport through the similar transporter, NBCe1, that carries CO₃²⁻ [22,23]. In particular, a recent report on the molecular dynamics simulations of NBCe1 [24] proposes a potential binding site for CO₃²⁻. Given the structural similarity between NBCe1 and NBCn1 (>65%) and a charge conservation in the binding site, it is possible that NBCn1 transports 2 Na⁺ and 1 CO₃²⁻, and the channel activity is related to a leak at the Na⁺ binding site. We have a preliminary observation that a mutation in the site associated with the Na⁺ binding abolishes the channel activity (unpublished data). Furthermore, the channel activity is enhanced by the stilbene derivate DIDS [4,5], which inhibits all sodium bicarbonate transporters except NBCn1 [1,25]. As DIDS occludes the entrance of the ion passageway in NBCe1 [24], the channel activity in NBCn1 may occur through the same ion passageway without being coupled to the cotransport activity. It will be interesting to investigate the molecular nature of the channel activity for future study.

NBCn1 channel activity increases Na⁺ influx and raises [Na⁺]_i. It is thus conceivable that the channel activity reduces an electrochemical gradient of Na⁺ with the consequence of a decrease in the driving force for Na⁺ influx via NMDARs. On the other hand, our results are inconsistent with others [26,27] that report that an increase in [Na⁺]_i enhances NMDAR currents and single channel activity in cultured spinal and hippocampal neurons. These changes are mediated by increasing Ca²⁺ influx through NMDARs and overcoming the Ca²⁺-dependent inactivation of NMDAR gating. Similarly, an increase in [Na⁺]_i by a voltage-gated sodium channel modifier increases Ca²⁺ influx through NMDARs and engages activity-dependent Ca²⁺ signaling mechanisms that lead to structural plasticity in cerebrocortical neurons [28]. It is unclear why there is a disparity between our study and others. One possible reason for this disparity is protein tyrosine kinase Src, which is associated with NMDAR regulation [29]. Src is involved in the receptor sensitivity to [Na⁺]_i [26,27], and its expression levels are high in most neurons [30]. Sato et al. [31] reported that the Src expression level is very low in Xenopus oocytes, below the detection limit. If the disparity between our study and others is due to Src, then our study supports the idea that intracellular Na⁺ does not directly affect NMDARs, but instead serves as a signaling ion. Another possible reason is the magnitude of the [Na⁺]_i change by NBCn1. With an injection of >25 ng of NBCn1 cRNA per oocyte, the channel activity can raise [Na⁺]_i by ≈40 mM [4], sufficient to trigger NMDARs activation [26,27]. However, we could not use this amount of NBCn1 for coexpression with GluN1·N2 and PSD95, because Xenopus oocytes have a limited capacity to translate membrane-bound mRNAs [32].

Regardless of the reason for the disparity, our results lead to a question of whether the decreasing effect of NBCn1 channel activity on NMDARs can be recapitulated in neurons. We do not rule out the possibility that NBCn1 channel activity conversely enhances NMDAR function in neurons. Neurons express a variety of Na⁺-permeable channels, such as voltage-gated Na⁺ channels [33,34] and Na⁺ leak channels [35], and changing [Na⁺]_i beyond the resting level may occur by a wide range of physiological and pathophysiological stimuli. We envision that NBCn1 channel activity will also play a role in regulating [Na⁺]_i in the CNS.

NBCn1 channel activity decreases NMDAR function in the absence of PSD95 (Figure 2), and there is no apparent difference in I_Glu between GluN1·N2/NBCn1 vs. GluN1·N2/NBCn1/PSD95 (Figure 4). Thus, NBCn1 does not need to cluster with NMDARs to decrease receptor function. On the other hand, the clustering is critical for receptor expression and activity, because NBCn1 KO mice develop a significant downregulation of GluN1 and PSD95 in neurons [14]. Neurons from these mice display lower apoptotic cell death associated with excitotoxicity, and their membranes are less excitable. The downregulation of GluN1 and PSD95, thus, demonstrates that NBCn1 is important for NMDAR expression during mouse development. Based on these results, we propose that NBCn1 has dual effects on NMDAR. On the one hand, it stabilizes NMDAR expression by constituting a large protein complex with the receptors; on the other hand, it decreases NMDAR function by its Na⁺ channel activity. By producing such dual effects, the transporter can fine-tune the regulation of NMDARs in the brain.

The decreased effect on the receptors by NBCn1 channel activity also occurs in the presence of CO₂/HCO₃⁻ (Figure 5). The decrease was evident between the absence vs. presence of NBCn1, while there was no difference before and after CO₂/HCO₃⁻ application. CO₂/HCO₃⁻ causes intracellular acidification as CO₂ enters cells, associates with H₂O, and produces H⁺ and HCO₃⁻. The pH_i drops to 6.9–7.1 in control oocytes, while it is higher in NBCn1-expressing oocytes [12,36,37]. We inhibited NBCn1 cotransport activity by removing bath Na⁺ and adding it only when glutamate was applied. There was no difference in I_Glu before and after CO₂/HCO₃⁻ application, thus indicating that a decrease in pH_i by 0.5 unit has negligible effect on GluN1·N2. This further indicates that NMDARs are less sensitive to pH_i, in contrast to its steep dependence on extracellular pH with the half maximum inhibition of pH 7.3 [38,39]. The pH-sensitive residues are mainly located in the GluN2 N-terminal domain, determined by mutagenesis study [40] and pKa prediction of titratable residues based on the high-pH model of the GluN1-GluN2A structure [41].

What percentage of the Na⁺ current relative to cotransport would be mediated by the channel activity? This can be estimated by comparing Na⁺ ion flux (J_Na) and I_NBC under a voltage clamp condition. J_Na is calculated from (dpH/dt)Vβ_t, where V is an oocyte volume (1 μL) and β_t is total buffering power (i.e., intrinsic buffering power (β_i) + 2.3 × [HCO₃⁻]_i). We did not determine β_i, but the reported value was 18.2 mM at 7.05 < pH_i < 7.25 [42]. In our pH measurement, dpH/dt was 12.6 × 10⁻⁵ pH/sec, and [HCO₃⁻]_i was 7.05 in 5% CO₂/25 mM HCO₃⁻ solution. This gives J_Na of 426.284 × 10⁻¹⁴ mole/sec. At −60 mV, NBCn1 channel activity produces 316.3 nA, which equals to 327.83 × 10⁻¹⁴ mole/sec. Thus, the ratio of Na⁺ current/cotransport is 0.769; i.e., 76.9% when oocyte membrane is at −60 mV. The channel activity produces a Na⁺ conductance and is markedly reduced as membranes are depolarized.

In summary, our study shows the importance of NBCn1 channel activity for NMDAR regulation. NBCn1 regulates the receptors not only by altering the Na electrochemical gradient, but also by compensating PSD95-mediated effects on receptor function. NBCn1 channel activity provides another mechanism to modulate the receptors, in addition to its cotransport activity. We envision that stimulating NBCn1 channel activity in neurons might be a valuable approach to reducing excitotoxicity and abnormal NMDAR activity in neurodegenerative diseases such as Alzheimer’s disease [19]. In this context, it is interesting to note that DIDS or SITS can provide moderate protection against NMDA toxicity and decrease NMDAR-mediated increase in intracellular Ca²⁺ in cultured rat cortical neurons [43]. Overall, our study provides new insights into the functional properties of NBCn1 and offers a basis for future studies on NBCn1-mediated regulation of glutamate transmission in the central nervous system.

4. Materials and Methods

4.1. Protein Expression in Xenopus Oocytes

Protein expression in Xenopus oocytes was performed as described previously [13]. Briefly, Xenopus laevis oocytes at stages V–VI were purchased from Ecocyte Bioscience (Austin, TX, USA). NBCn1-E, PSD95, NR1A, and NR2A were transcribed using the mMessage/mMachine transcription kit (Life Technologies, Grand Island, NY, USA). The amount of injected cRNA (in 46 nL) was 14 ng for NBCn1-E, 7 ng for PSD95, 7 ng for NR1A, and 14 ng for NR2A. Thus, the ratio of NR1A to NR2A was 1 to 2. The injection of 46 nL reduced the oocyte osmolality from 300 mOsM to 287 mOsM, which is negligible. Equal amounts of RNAs were used when multiple samples were compared. Controls were injected with sterile water. Oocytes were maintained at 18 °C for 3 days.

4.2. Two-Electrode Voltage Clamp

An oocyte was placed in the recording chamber containing ND96 solution (mM; 96 NaCl, 2 KCl, 1.8 CaCl₂, 1 MgCl₂, 10 HEPES, and pH 7.4) and 1 mM BaCl₂ and impaled with two borosilicate glass electrodes filled with 3 M KCl (tip resistance: 0.5–2 MΩ). After the resting potential was stable, the oocyte was clamped at –60 mV using the voltage-clamp amplifier OC-725C (Warner Instrument, Hamden, CT, USA). For recording I_NBCn1, current-voltage (I–V) relationships were obtained by voltage commands from −120 mV to +60 mV (100 msec, in 20 mV increments). Na⁺-free solution was made by N-methyl-D-glucamine. For recording I_Glu, the holding potential of –60 mV was used. I_Glu was monitored by applying 100 µM glutamate (with 10 µM glycine, but no Mg²⁺). I–V relationships were also obtained by voltage commands in the absence or presence of glutamate application. For recording I_Glu in the presence of CO₂/HCO₃⁻, oocytes were superfused with a solution buffered with 10% CO₂, 50 mM HCO₃⁻, and pH 7.4. Signals were collected using a Digidata 1322 interface (Molecular Devices; Sunnyvale, CA, USA) and analyzed using pClamp 10 (Molecular Devices). Signals were filtered using a Bessel lowpass filter with a cutoff frequency of 0.1 Hz. All experiments were performed at room temperature.

4.3. Statistical Analysis

Data were reported as mean ± standard error. The level of significance was determined using (i) unpaired, two-tailed Student t-test for comparison of I_Glu between GluN1·N2 vs. GluN1·N2/NBCn1 and comparison of I_NBC between control vs. NBCn1; (ii) paired, two-tailed Student t-test for comparison of I_Glu before and after CO₂/HCO₃⁻ application within the same group; (iii) one-way ANOVA for comparison of I_Glu among GluN1·N2/NBCn1 with different injection amounts of NBCn1; and (iv) two-way ANOVA for comparison of I–V relationships. The p value of less than 0.05 was considered significant. Data were analyzed using Prism 9 (GraphPad; La Jolla, CA, USA) and Microsoft Office Excel add-in Analysis ToolPak (Redmond, WA, USA).