Proton Fall or Bicarbonate Rise

Glycolysis is the primary step for major energy production in the cell. There is strong evidence suggesting that glucose consumption and rate of glycolysis are highly modulated by cytosolic pH/[H+], but those can also be stimulated by an increase in the intracellular [HCO3−]. Because proton and bicarbonate shift concomitantly, it remained unclear whether enhanced glucose consumption and glycolytic rate were mediated by the changes in intracellular [H+] or [HCO3−]. We have asked whether glucose metabolism is enhanced by either a fall in intracellular [H+] or a rise in intracellular [HCO3−], or by both, in mammalian astrocytes. We have recorded intracellular glucose in mouse astrocytes using a FRET-based nanosensor, while imposing different intracellular [H+] and [CO2]/[HCO3−]. Glucose consumption and glycolytic rate were augmented by a fall in intracellular [H+], irrespective of a concomitant rise or fall in intracellular [HCO3−]. Transport of HCO3− into and out of astrocytes by the electrogenic sodium bicarbonate cotransporter (NBCe1) played a crucial role in causing changes in intracellular pH and [HCO3−], but was not obligatory for the pH-dependent changes in glucose metabolism. Our results clearly show that it is the cytosolic pH that modulates glucose metabolism in cortical astrocytes, and possibly also in other cell types.

The high energy requirement of the mammalian brain is primarily fueled by the degradation of blood-derived glucose. Astrocytes, the major glial cells in the brain, take up a large portion of blood glucose, and by aerobic glycolysis, they preferably produce more lactate (1,2). Accumulating evidence suggests that astrocytes metabolically support neurons by releasing lactate, which then could be taken up by neighboring neurons (3,4). Therefore, glucose metabolism in astrocytes was proposed to be modulated by various means of signals presumably originating from neurons (5,6). It is well documented that astrocytes respond to neuronal activity with an intracellular alkalinization (7)(8)(9). Recently, it was shown that the glycolytic rate in astrocytes can be significantly enhanced by an alkaline pH i shift derived from extracellular K ϩ -induced membrane depolarization enhancing bicarbonate uptake via the electrogenic sodium bicarbonate cotransporter NBCe1 (10), which is highly expressed in mouse cortical astrocytes (11). In fact, cytoplasmic pH is well known to be a potential regulator of glucose metabolism, particularly the glycolytic rate, in various cell types (12)(13)(14)(15). However, due to the dynamic chemical equilibrium of pH with [HCO 3 Ϫ ] and [CO 2 ], a rise in pH i at a constant [CO 2 ] will also result in a concomitant rise in cytosolic [HCO 3 Ϫ ] ([HCO 3 Ϫ ] i ). Interestingly, a shift in [HCO 3 Ϫ ] i can also be a potential signal to a variety of cellular processes, including glucose metabolism (16 -19). A rise in [HCO 3 Ϫ ] i was shown to stimulate glucose metabolism in astrocytes, via the activation of bicarbonate-dependent soluble adenylyl cyclase (20). The question whether it is the rise in bicarbonate ions or a fall in [H ϩ ] i that mediates the cytoplasmic pH-dependent augmentation of the glycolytic rate has not been resolved.
Here we address this question by using intracellular H ϩ (H ϩ i ) imaging and FRET-based intracellular glucose imaging in cultured cortical astrocytes and organotypic hippocampal slices. To functionally dissect the contribution of H ϩ and HCO 3 Ϫ on glycolysis, we measured the glycolytic rate under different conditions: 1) during a fall in H ϩ i (alkalinization) with either a concomitant fall or rise in intracellular bicarbonate, 2) during a rise in intracellular bicarbonate with either a concomitant fall or rise in [H ϩ ] i , and 3) during a fall in [H ϩ ] i in the nominal absence of CO 2 /HCO 3 Ϫ . Our results suggest that the glycolytic rate in astrocytes is primarily augmented by the fall in [H ϩ ] i and not by a rise in [HCO 3 Ϫ ] i .

Results
We have approached the question whether glucose consumption and glycolysis are stimulated by a fall in [H ϩ i ], a rise in bicarbonate, or by both, by measuring glucose in cultured astrocytes and in organotypic slices of mouse hippocampus in culture after imposing different pH and bicarbonate changes in the cells. To change [H ϩ ] and [HCO 3 Ϫ ] in astrocytes, we equilibrated the superfusing solutions with different [CO 2 ] and [HCO 3 Ϫ ] at pH values between 7 and 7.8. To evaluate the role of HCO 3 Ϫ transport into and out of astrocytes, we employed a genetic mouse model with a deletion in the gene for the electrogenic sodium bicarbonate cotransporter NBCe1 (SLC4A4) (21), which is highly expressed in mouse cortical astrocytes (11,22).  (Fig. 1C). Cytosolic alkalinization of astrocytes induced by the exposure of a solution with reduced external [CO 2 ] can be attributed to the diffusion of intracellular CO 2 out of the cell. As CO 2 leaves the cell, intracellular HCO 3 Ϫ and H ϩ convert to CO 2 , which then continues to leave the cell until equilibrium is reached with the external [CO 2 ]. In contrast, cytosolic alkalinization induced by the exposure of a solution with elevated extracellular [HCO 3 Ϫ ] Ϫ ] i decreased during the first change, but increased during the second change, respectively (H). C-E and I and J, cytosolic glucose, monitored with the FRET-based nanosensor, dropped in WT cells in both solutions, as did glucose during extracellular removal of glucose (Glu.) and lactate (Lac.) (D and E), whereas in NBCe1-KO cells, glucose dropped moderately during the first solution change, but did not change during the second solution change (I and J). Error bars represent means Ϯ S.E. **, p Յ 0.01, ***, p Յ 0.001. at constant [CO 2 ] can be attributed to the inward transport of bicarbonate via NBCe1.
The steady-state cytosolic glucose levels decreased under these pH i changes in both solutions, reflecting increased glucose consumption (Fig. 1D). In 2% CO 2 /26 mM HCO 3 Ϫ , the glucose signal decreased by 4.9 Ϯ 0.34% with a rate of 0.7 Ϯ 0.05%/ min, whereas in 5% CO 2 /61 mM HCO 3 Ϫ , the glucose signal decreased by 8.4 Ϯ 0.39% with a rate of 2.6 Ϯ 0.18%/min (Fig.  1E). Our experiments suggest that glucose consumption is stimulated by cytosolic alkalinization under conditions when [HCO 3 Ϫ ] i increased or decreased. Removing glucose and lactate from the superfusing solution led to a fast and large drop in the glucose signal of 17.8 Ϯ 0.26% with a rate of 4.3 Ϯ 0.15%/min (Fig. 1, D and E). This rapid and reversible, in some instance biphasic, decrease in cytosolic glucose upon the complete removal of external glucose and lactate is attributed to the efflux of glucose via glucose transporters (GLUT) and the basal glucose consumption of the cell (phosphorylation by hexokinase) (23). We considered the saturation point of the glucose signal under complete removal of external glucose and lactate as the lowest level of cytosolic glucose, and used this protocol to correlate the level of stimulation-dependent glucose changes in cells.
We have previously shown that the electrogenic sodium bicarbonate cotransporter, NBCe1, is highly expressed in cortical astrocytes and responsible for fast HCO 3 Ϫ transport into and out of these cells (11,24). We therefore asked how much of the change in pH i was attributable to NBCe1 activity, and hence how much it influenced glucose consumption. We therefore repeated the experiments in astrocytes prepared from a transgenic mouse model, in which the NBCe1 gene had been deleted (NBCe1-KO Ϫ ] also in NBCe1-KO cells (Fig. 1F), by 18 Ϯ 0.6 nM in 2% CO 2 /26 mM HCO 3 Ϫ and by 5.9 Ϯ 1.1 nM in 5% CO 2 /61 mM HCO 3 Ϫ (Fig. 1G). From these values, the changes in [HCO 3 Ϫ ] i were calculated; the [HCO 3 Ϫ ] i decreased by 9.2 Ϯ 0.7 mM in 2% CO 2 /26 mM HCO 3 Ϫ and increased by 1.8 Ϯ 0.1 mM in 5% CO 2 /61 mM HCO 3 Ϫ (Fig. 1H). There was a small decrease in the steady-state cytosolic glucose in NBCe1-KO cells of 3.4 Ϯ 0.43% with a rate of 0.26 Ϯ 0.03% in 2% CO 2 /26 mM HCO 3 Ϫ , but no change in 5% CO 2 /61 mM HCO 3 Ϫ (Fig. 1, I and J). Removing glucose and lactate resulted in a large and reversible decrease of the glucose signal by 18.9 Ϯ 0.3% with a rate of 3.7 Ϯ 0.26%/min (Fig. 1, I and J). These experiments suggest that a rise in glucose consumption is not necessarily coupled to the HCO 3 Ϫ transport via NBCe1, but are in line with a pH dependence of glucose metabolism.
We next asked the question how, more specifically, the glycolytic rate, indicated by the change in cytosolic glucose in the presence of the glucose transport inhibitor cytochalasin B (Cyt B, 3 20 M), was affected in these solutions of different pH/[CO 2 ]/[HCO 3 Ϫ ] (Fig. 2). The application of cytochalasin B induced a reversible, monophasic fall in cytosolic glucose, attributed to the basal glucose consumption of the cell (23). When a change to low CO 2 or high HCO 3 Ϫ level was done dur-ing the ongoing Cyt B-induced gradual fall in the glucose level, an increase in the rate of fall was observed, corresponding in WT cells to an increase in the glycolytic rate by 420 Ϯ 22 and 625 Ϯ 78%, respectively (Fig. 2, A and B). In the same solutions, the glycolytic rate in NBCe1-KO cells increased to 363 Ϯ 26% in 2% CO 2 /26 mM HCO 3 Ϫ , but only to 189 Ϯ 8.7% in 5% CO 2 /61 mM HCO 3 Ϫ (Fig. 2, C and D). In organotypic hippocampal slices, transduced with the FRET-based nanosensor for glucose, the glycolytic rate in 2% CO 2 /26 mM HCO 3 Ϫ was enhanced to 359 Ϯ 33% in WT astrocytes and to 370 Ϯ 92% in NBCe1-KO astrocytes (Fig. 2, E-G), confirming the conclusions drawn from the experiments on cultured astrocytes. To estimate the glycolytic rate, the slope was calculated from the initial linear change of the FRET-ratio (1-2 min), induced by the exposure of a test solution. This corresponds to the cytosolic glucose changes during the early phase of pH i shifts induced by the same test solution. Because the peak of pH i changes, induced by the test solution (Fig. 1A), often occurs 1-4 min after the solution change, there may have been an underestimation of the values of the glycolytic rate presented here.
Glycolytic would be expected to elicit an intracellular acidification and an increase in the [HCO 3 Ϫ ] i , attributable to CO 2 diffusion into the cells and subsequent partial degradation to H ϩ and HCO 3 Ϫ ; in WT cells, however, the increase in [HCO 3 Ϫ ] e also stimulates NBCe1 (11). The total [HCO 3 Ϫ ] i rise was 12.7 Ϯ 1.1 mM in WT cells and 11.1 Ϯ 0.3 mM in NBCe1-KO cells (Fig. 3C). Under these conditions, glucose fell rapidly in WT cells, but fell much less in NBCe1-KO cells (Fig. 3D), indicating a change in glycolytic rate to 497 Ϯ 50% in WT cells and to 39 Ϯ 49% in NBCe1-KO cells.
When the [CO 2 ] was raised from 2 to 5%, lowering the pH e to 7.0 at constant [HCO 3 Ϫ ] of 10 mM (hypercapnic acidosis), a cytosolic acidification with a concomitant rise in bicarbonate was induced in both WT and NBC-KO astrocytes (Fig. 4, A-C). The glycolytic rate measured under this condition in primary cultured WT and NBCe1-KO astrocytes was 141 Ϯ 11 and 129 Ϯ 15%, respectively (Fig. 4, D-F). We have also measured glycolytic rate in astrocytes from hippocampal organotypic slice cultures under similar condition (hypercapnic acidosis) and found it to be changed by Ϫ37 Ϯ 20% in WT and ϩ199 Ϯ 15% in NBCe1-KO astrocytes (Fig. 4, G and H). These results suggest that glycolytic rate can be enhanced in response to a rise in cytosolic bicarbonate, even if pH i is concomitantly falling, although to a much smaller extent than during isohydric hypercapnia (Fig. 3E), when pH i is concomitantly rising.
Previous studies showed that depolarization of astrocytes by extracellular [K ϩ ] rise causes an intracellular alkalinization only in WT cells, but not in NBCe1-KO cells, which resulted in a robust stimulation of glycolysis in WT cells, but not in NBCe1-KO cells (10,11). We repeated these experiments and found that a rise in K ϩ elicits an intracellular alkalinization and a fast fall in cytosolic glucose (rise in glucose consumption) only in WT cells, but not in NBCe1-KO cells (Fig. 5, A-E). The [H ϩ ] i decreased by 29.9 Ϯ 2.1 nM in WT and by 4.05 Ϯ 0.98 nM in NBCe1-KO cells (Fig. 5, A and B), and the [HCO 3 Ϫ ] i rose by 10.7 Ϯ 0.4 mM in WT cells and by 3.17 Ϯ 0.24 mM in NBCe1-KO cells (Fig. 5C). When external K ϩ was raised from 3 to 15 mM in the normal solution (5% CO 2 /26 mM HCO 3 Ϫ , pH 7.4), a sharp drop of the glucose signal of 13.8 Ϯ 0.47% was monitored in WT cells, but a small increase of glucose by 3.4 Ϯ 0.54% in NBCe1-KO cells was seen (Fig. 5, D and E). Next, we used this depolarization protocol with 15 mM K ϩ to modulate the hypercapnic acidosis-associated H ϩ i and HCO 3 Ϫ shifts in astrocytes. Exposing cells to an increase in CO 2 (from 2 to 5%) and K ϩ (from 3 to 15 mM) at constant extracellular HCO 3 Ϫ (10 mM) resulted in an intracellular alkalinization in WT cells and acidification in NBC-KO cells with a concomitant rise in bicarbonate in both cell types (Fig. 5, F-H). The glycolytic rate under these conditions increased to 655 Ϯ 31%/min in WT cells, but only to 181 Ϯ 25% in NBCe1-KO cells (Fig. 5, I and J).The rise in extracellular K ϩ is known to stimulate glycolysis via activation of Na ϩ /K ϩ pump in astrocytes (25). The moderate increase in glycolytic rate in NBCe1-KO cells (181 Ϯ 25%) during the elevation of extracellular [K ϩ ]and [CO 2 ] (Fig. 5, I and J) may be attributed to the stimulation of Na ϩ /K ϩ pump as reported by Ref. 25. These results strongly suggest that pH i is the main determinant of glucose consumption and glycolytic rate, whereas a rise in the [HCO 3 Ϫ ] i is largely ineffective, if pH i decreases. Our results agree with those of Ref. 10, and in addi-    Ϫ ], and pH as indicated and introduced in Fig. 1, and the corresponding relative glycolytic rate determined from these changes in the glucose signal in WT (B) and in NBCe1-KO (D) cells. gluc., glucose; lact., lactate. E-G, the first part of this experiment was done also in organotypic hippocampal brain slices, showing the glucose-sensitive FRET ratio in astrocytes from WT (E) and from NBCe1-KO (F) mice, from which the relative glycolytic rates were determined (G). Error bars represent means Ϯ S.E. *, p Յ 0.05, ***, p Յ 0.001. n.s., not significant. tion, support our main conclusion that it is cytosolic alkalinization that stimulates glucose metabolism.
Glycolytic Rate during Cytosolic Alkalinization in the Nominal Absence of CO 2 /HCO 3 Ϫ -So far, all experiments were performed in CO 2 /HCO 3 Ϫ -containing solutions, which might open the possibility that, although a [HCO 3 Ϫ ] rise was not necessary for stimulating glycolysis, the mere presence of millimolar concentrations of [HCO 3 Ϫ ] might be required for an alkalinizationinduced stimulation of glycolysis. Therefore, we used trimethylamine (TMA) to alkalinize the cells in the nominal absence of CO 2 /HCO 3 Ϫ . In a HEPES-buffered solution, 5 mM TMA resulted in a mean decrease of 16.8 Ϯ 1.2 nM H ϩ in WT cells and of 15.6 Ϯ 1 nM H ϩ in NBCe1-KO cells (Fig. 6, A and B). Under these conditions, TMA stimulated the glycolytic rate to 657 Ϯ 74%/min in WT cells and to 588 Ϯ 95%/min in NBCe1-KO cells (Fig. 6, C and D). The [HCO 3 Ϫ ] e and [HCO 3 Ϫ ] i attributable to dissolved air-CO 2 would be below 300 M under these conditions, whereas the [HCO 3 Ϫ ] i would presumably change by less than 150 M in TMA. This strongly suggests that the stimulation of glycolysis by intracellular alkalinization is independent of HCO 3 Ϫ . Because in the presence of CO 2 /HCO 3 Ϫ the H ϩ buffer strength is greatly increased attributable to fast HCO 3 Ϫ transport via NBCe1 (26), 5 mM TMA slightly changed pH i in astrocytes (Ͻ5 nM; Fig. 7, A and B). Under these conditions, a change in the glycolytic rate to only 198 Ϯ 25% was determined in TMA (Fig. 7, C and D).

Discussion
The present study addresses a fundamental question in cellular glucose metabolism, i.e. whether pH-dependent stimulation of glycolysis reported in different cell types, including astrocytes, is mediated by a fall in [H ϩ ] i or a rise in [HCO 3 Ϫ ] i . To functionally dissect the impact of intracellular pH and HCO 3 Ϫ on glycolysis in astrocytes, we measured the changes in steadystate glucose and glycolytic rate while altering the intracellular [H ϩ ] and [HCO 3 Ϫ ]. We show that glycolytic rate is substantially enhanced (by 4 -7-fold) when there is a fall in intracellular proton concentration with either a concomitant decrease or increase in [HCO 3 Ϫ ] i . In contrast, there was only a mild activation of glycolysis (up to ϳ2-fold) in response to a rise in [HCO 3 Ϫ ] i (ϳ12 mM), unless pH i also increased. The glycolytic rate was enhanced up to 7-fold by intracellular alkalinization, even in the nominal absence of CO 2 /HCO 3 Ϫ . These results suggest that a fall in [H ϩ ] i is sufficient and necessary to enhance the glycolytic rate in astrocytes, whereas a change in the [HCO 3 Ϫ ] i per se has little effect on glucose consumption and glycolytic rate. The impact of our conclusion on neuron-glia interactions, using acute slices and neuronal activation, still needs to be tested.
Proton Is a Key Signal for Glycolysis-Glycolysis is the primary step for major energy production in cells. The glycolytic rate is tightly controlled in different steps of 10 major enzymatic reactions involved in this process. The activity of 6-phospho frutokinase1 (PFK1), one of the pivotal regulatory enzymes of glycolysis, is known to have a steep pH dependence, in vitro, between the dynamic range of pH 7.1-7.35 (27). Intracellular alkalinization following bicarbonate uptake via NBCe1 has recently been shown to augment the glycolytic rate in astrocytes, presumably due to the pH-dependent activation PFK1 in astrocytes (10). On the other hand, activation of bicarbonatesensitive soluble adenylyl cyclase (sAC) followed by bicarbonate entry via NBCe1 was also suggested to stimulate glucose metabolism and lactate release in mammalian astrocytes (20). Because the rise of pH i and [HCO 3 Ϫ ] i may stimulate glucose metabolism independently, it was unresolved whether augmentation of glucose consumption and glycolytic rate reported in astrocytes was attributable to a rise in pH i or a rise in [HCO 3 Ϫ ] i , or whether both pH i and [HCO 3 Ϫ ] i had to increase. Our results now clearly show that it is a rise in pH i (fall in [H ϩ ] i ) that mediates the robust stimulation of glycolysis in astrocytes, presumably by releasing a H ϩ block of PKF1 catalytic activity (27,28). The measured steady-state pH of cultured astrocytes (7.17 Ϯ 0.02 in HEPES and 7.2 Ϯ 0.07 in 5% CO 2 /26 mM HCO 3 Ϫ buffered solution) lies in the window of maximum pH dependence of PFK1 and below the saturation point for the pH-dependent activation of PFK1. This correlates with the substantial activation of glycolysis (up to 7-fold) during an alkaline shift of 0.1 to 0.15 pH unit in astrocytes. In response to cytosolic acidification, glycolytic rate was moderately enhanced (up to 2-fold) or even slightly decreased in astrocytes (Figs. 3, D and E, and 4, G and F). This indicates that there might be other pathways to which glucose is channelized in addition to glycolysis, and other factors, except pH, that affect glycolysis.

Role of Bicarbonate in Glycolysis-A rise in [HCO 3
Ϫ ] i can occur via two different pathways: 1) by CO 2 diffusion into the cells and its degradation to H ϩ and HCO 3 Ϫ , which is augmented by carbonic anhydrase (CA) activity, and 2) by the direct uptake of HCO 3 Ϫ via membrane transporters such as NBCe1. The rise of [HCO 3 Ϫ ] i in cells via CO 2 pathway results in an acidification, whereas the transporter-mediated rise in [HCO 3 Ϫ ] i causes an alkalinization, as bicarbonate consumes intracellular H ϩ to produce CO 2 and H 2 O. NBCe1 is known to be the major bicarbonate transporter in astrocytes and is functionally very active in these cells (11,22). However, depending upon the intra-and extracellular bicarbonate concentrations, the direction of bicarbonate transport mediated by NBCe1 can be inwardly or outwardly directed. Therefore, the transporter activity can induce either an intracellular alkalinization or acidification (11,24). The study that showed bicarbonate-dependent augmentation of glycolysis (20), however, was performed under the condition when NBCe1 was operating in the inwardly directed manner, which is associated with cytosolic alkalinization. Our results suggest that a rise in [HCO 3 Ϫ ] i as such has no major impact on the acute activation of glycolytic rate in astrocytes. This conclusion is based on the following evidence. 1) Although a substantial rise in intracellular bicarbonate was observed (⌬ HCO 3 Ϫ ϭ ϳ12 mM) via CO 2 pathways under hypercapnic conditions, there was only a small activation of glycolytic rate (ϳ2-fold), when cells acidified. 2) A robust activation of glycolytic rate (5-7-fold) was observed during the rise of intracellular HCO 3 Ϫ (⌬ HCO 3 Ϫ ϭ Ϫ ] of 10 mM, which induced cytosolic acidification, with a concomitant rise in bicarbonate in both cell types (A-C). D-F, the glucose-sensitive FRET ratio in the presence of Cyt B showed little change during the hypercapnic acidosis in both WT (D and F) and NBCe1-KO (E and F) cells. gluc, glucose; lact, lactate. G and H, the same experiment was also done in organotypic hippocampal brain slices (G and H), showing the glucose-sensitive FRET ratio in astrocytes from WT (G) mice. The glycolytic rate was determined from the FRET ratio in hippocampal astrocytes from WT and NBCe1-KO mice (H). Error bars represent means Ϯ S.E. *, p Յ 0.05, ***, p Յ 0.001. n.s., not significant.
ϳ10 mM) only when the cells alkalinized concomitantly, attributable to NBCe1 activation in the inward direction. 3) Glycolytic rate was strongly activated during cytosolic alkalinization, which was associated with a decreasing [HCO 3 Ϫ ] i . 4) Glycolytic rate was robustly activated (up to 6-fold) during cytosolic alkalinization even in the nominal absence of intraand extracellular bicarbonate. These results suggest that it is not the rise in bicarbonate, but the alkalinization (fall of [H ϩ ] i ) that primarily activates the glycolytic rate in astrocytes.
Hypercapnia-mediated pH i Shifts and Their Impact on Glycolysis-Isohydric hypercapnia (increase in [CO 2 ] o and [HCO 3 Ϫ ] o at constant pH o ) is expected to induce an intracellular acidification, attributable to diffusion of CO 2 into the cells, and its degradation to H ϩ and HCO 3 Ϫ (29, 30). However, one of the striking findings of the present study was that isohydric hypercapnia, i.e. switching extracellular solution from 2% CO 2 /10 mM HCO 3 Ϫ /7.4 pH to 5% CO 2 /26 mM HCO 3 Ϫ /7.4 pH, induced an intracellular alkalinization in astrocytes and activated glycolysis to more than 5-fold (Fig. 3). In NBCe1-KO astrocytes, in contrast, an acidification was evoked by the same solution change. This suggests that bicarbonate uptake via NBCe1 dominated over CO 2 diffusion in WT astrocytes during the exposure of isohydric hypercapnia, and not only suppressed the CO 2 -mediated acidification, but even caused an alkalinization (see also Ref. 26). Attributed to the acidic pH i shift induced by isohydric hypercapnia in NBCe1-KO astrocytes, and despite the substantial bicarbonate rise by ϳ12 mM, the glycolytic rate was not augmented as compared with WT, where the alkalinization was accompanied by a robust increase in glycolysis.   Ϫ ] e ), we could induce acidification in both WT and NBCe1-KO cells with a concomitant rise in bicarbonate. This acidification in both cells is presumably due to the domination of CO 2 diffusion over NBCe1-mediated bicarbonate transport into the cells. Because both 2% and 5% CO 2 -buffered solutions had the same concentration of external bicarbonate, the free energy for NBCe1-mediated bicarbonate transport would be expected to be unaltered. Although there is a net bicarbonate rise in both WT and NBC-KO cells under these conditions, the glycolytic rate was enhanced only to 1.4-fold in WT and 1.3fold in NBCe1-KO cells, suggesting that it is not the rise in bicarbonate ion stimulating glucose consumption in astrocytes. Interestingly, we could reverse the pH change induced by hypercapnic acidosis in WT cells, and thus enhance the glycolytic rate up to 6-fold, when external K ϩ was also concomitantly raised from 3 to 15 mM. Elevated extracellular K ϩ from 3 to 15 mM would depolarize astrocytes and stimulate inwardly directed NBCe1 (11,22). The bicarbonate uptake via NBCe1 masked CO 2 -induced acidification, and even caused an alkalinization, thus stimulating glycolysis.
Hypercapnia-induced alkalinization in astrocytes and stimulation of glycolysis may have potential physiological significance in astrocyte-neuron metabolic shuttling. As neurons rely more on oxidative metabolism for their energy requirements as compared with astrocytes, their activity may lead to a substantial production of intracellular CO 2 (31)(32)(33). The metabolically produced CO 2 in neurons may diffuse out, following its chemical gradient, and then may degrade into H ϩ and HCO 3 Ϫ , catalyzed by extracellular carbonic anhydrases expressed by neurons and glial cells (34). A rise in extracellular CO 2 /HCO 3 Ϫ , originating from oxidative neurons, may activate NBCe1 in astrocytes, and the subsequent intracellular alkalization would stimulate glycolysis and lactate production in astrocytes. The lactate produced in astrocytes can be shuttled back to neurons for metabolic support as proposed by the astrocyte-neuron lactate shuttle hypothesis (35)(36)(37). Indeed, it was recently shown that CO 2 exposure causes a rise in extracellular lactate in acute brain slices (38). Therefore, we hypothesize that the activityinduced production of CO 2 in neurons, in addition to elevated extracellular K ϩ , may function as a signal that stimulates glucose metabolism in neighboring astrocytes.