Effect of hypoxia on steady-state pH_i

Hippocampal astrocytes bathed in a nominally CO₂/${\mathrm{HCO}}_3^{-}$-free, HEPES-buffered solution had a mean steady-state pH_i of 7.05 ± 0.05 (n=7). In the same seven experiments, subsequently exposing the astrocytes to the HEPES-buffered solution bubbled with N₂ (which reduced O₂ to ∼3% in the recording chamber) caused pH_i to decrease slightly to a lower steady-state pH_i of 7.00 ± 0.04 (P = 0.008, paired t-test) in 4.8 ± 0.7 min. In additional studies conducted in 5% CO₂/22 mM ${\mathrm{HCO}}_3^{-}$, we found that steady-state pH_i was indistinguishable under normoxic (7.23, n=3) vs. ∼3% O₂ conditions (7.25, n=3). In summary, acute hypoxia (∼3% O₂) has little effect on the resting pH_i of astrocytes either in the nominal absence or presence of CO₂/${\mathrm{HCO}}_3^{-}$.

Effect of hypoxia on acid-extrusion rates in the presence and absence of CO₂/${\mathrm{HCO}}_3^{-}$

To assess the effect of hypoxia on acid extrusion, we compared the pH_i recoveries from an intracellular acid load under normoxic and hypoxic conditions. As shown in Fig. 1A, astrocytes bathed in the HEPES-buffered solution had a steady-state pH_i of ∼6.9 (prior to point a). We then acid loaded the cells using the ${\mathrm{NH}}_4^{+}$-prepulse technique (Boron and De Weer, 1976). Applying 20 mM NH₃/ ${\mathrm{NH}}_4^{+}$ elicited an abrupt increase in pH_i (segment ab in Fig. 1A) due to NH₃ influx and protonation to form ${\mathrm{NH}}_4^{+}$. In the continued presence of NH₃/${\mathrm{NH}}_4^{+}$, pH_i slowly decreased (bc) due to ${\mathrm{NH}}_4^{+}$ influx and/or stimulation of acid-loading mechanisms. Removing NH₃/${\mathrm{NH}}_4^{+}$ caused pH_i to decrease sharply (cd) as the ${\mathrm{NH}}_4^{+}$ that accumulated during segment bc was converted to NH₃ (which diffused out of the cells) as well as H⁺. Subsequently, pH_i recovered (de) to a value similar to that prevailing before the ${\mathrm{NH}}_4^{+}$ prepulse. The segment-de pH_i recovery is primarily due to an amiloride-sensitive Na-H exchanger (Bevensee et al., 1997a), although a bafilomycin A1-sensitive H⁺ pump may also make a small contribution (Pappas and Ransom, 1993).

When the cells were exposed to a HEPES-buffered solution equilibrated with ∼3% O₂, pH_i decreased slightly (ef). After we acid loaded the now-hypoxic astrocytes a second time by applying and removing NH₃/${\mathrm{NH}}_4^{+}$ (fghi), the ensuing pH_i recovery exhibited a waveform (ijk) that was different from than seen under normoxic conditions (de). With hypoxia, the minimum pH_i attained after NH₃/${\mathrm{NH}}_4^{+}$ washout (point i) was consistently higher than with normoxia (d), consistent with stimulated acid extrusion. Moreover, with hypoxia, the rate of pH_i recovery was faster at the beginning (point i) and for most of the pH_i recovery (i.e., at relatively low pH_i values), but slower towards the end of the recovery (i.e., at relatively high pH_i values near point j).

In control normoxic experiments, pH_i-recovery rates from two sequential acid loads were very similar for astrocytes in the nominal absence or presence of CO₂/${\mathrm{HCO}}_3^{-}$ (data not shown).

Using segment-de and -ij pH_i-recovery rates in experiments similar to that shown in Fig. 1, we calculated the pH_i dependence of total acid extrusion (dpH_i/dt × β_Total) for 20% O₂ (open circles in Fig. 2) and ∼3% O₂ (filled circles). As previously reported for single cultured rat hippocampal astrocytes (Bevensee et al., 1997a), total acid extrusion decreased at progressively higher pH_i. However, hypoxia alkali shifted the plot of φ_E vs. pH_i by 0.15-0.30 in the pH_i range 6.2-6.6, and also caused it to become almost linear.

We also performed experiments similar to that shown in Fig. 1, but on astrocytes bathed in 5% CO₂/22 mM ${\mathrm{HCO}}_3^{-}$. The squares in Fig. 2 summarize the φ_E vs. pH_i data. As previously reported for single astrocytes under normoxic conditions (Bevensee et al., 1997a), total acid extrusion was higher at all pH_i values in the presence of CO₂/${\mathrm{HCO}}_3^{-}$ (open squares) compared to the nominal absence of the physiological buffer (open circles), predominantly reflecting the activity of an electrogenic Na/HCO₃ cotransporter. For astrocytes in the presence of CO₂/${\mathrm{HCO}}_3^{-}$, hypoxia (filled squares) alkali shifted the plot of φ_E vs. pH_i compared to normoxia (open squares). The plots are better fit (P<0.0001) with a model of 2 separate cubic polynomials than a simpler model of 1 cubic polynomial indicating that the data sets are distinct. The effects in the presence of CO₂/${\mathrm{HCO}}_3^{-}$ (squares) were not as pronounced as seen in the nominal absence of CO₂/${\mathrm{HCO}}_3^{-}$ (circles). For example, hypoxia alkali shifted the plot of φ_E vs. pH_i by only 0.05-0.15 (pH_i range: 6.8-7.0) for cells bathed in CO₂/${\mathrm{HCO}}_3^{-}$ (compared to 0.15-0.3 for cells in the HEPES-buffered solution).

Effect of hypoxia on stilbene-sensitive acid-extrusion

An electrogenic Na/HCO₃ cotransporter (NBC) inhibited by stilbene derivatives is the predominant ${\mathrm{HCO}}_3^{-}$-dependent acid extrusion mechanism in cultured hippocampal astrocytes from rat (Bevensee et al., 1997a, 1997b). To characterize the effect of hypoxia on ${\mathrm{HCO}}_3^{-}$-dependent acid extrusion further, we used the stilbene derivative SITS to inhibit NBC, and computed the pH_i dependencies of SITS-sensitive and -insensitive acid extrusion under both normoxic and hypoxic conditions. As previously reported, pretreating hippocampal astrocytes for ∼6 h in tissue-culture media containing 400 μM SITS irreversibly reduces φ_E (computed from the pH_i recovery from a CO₂/${\mathrm{HCO}}_3^{-}$-induced acid load) by ∼65% (Bevensee et al., 1997a). Because other transporters likely contributed to the pH_i recovery, the SITS inhibition was probably substantially more than 65%. Irreversibly inhibiting NBC activity with SITS prior to experiments avoids the necessity of using stilbenes during experiments, which would interfere with the BCECF fluorescence signal. In the present study, we inhibited NBC activity by pretreating with 500 μM SITS in the media for 6-8.3 h. In principle, the prolonged incubation in SITS could have induced compensatory changes in the pH physiology of the cells. On the other hand, we previously found that stilbene inhibition of the CO₂/HCO₃^--stimulated pH_i increase was similar for hippocampal astrocytes incubated in SITS for ∼6 h or exposed only briefly to DIDS (Bevensee et al., 1997a).

The experimental protocol was similar to that shown in Fig. 1, except the astrocytes in Fig. 3 were continually exposed to 5% CO₂/22 mM ${\mathrm{HCO}}_3^{-}$. We acid loaded the cells both before (abcd) and after (fghi) acute hypoxia, and either with or without a ∼7-h pretreatment with 500 μM SITS. For clarity, we show only the pH_i recoveries (d’e’ and i’j’) from the SITS experiment. SITS pretreatment reduced the pH_i-recovery rate for both normoxia (compare segments de and d’e’) and hypoxia (segments ij and i’j’)—data consistent with reduced NBC activity in the two conditions. Note that in SITS-pretreated cells, hypoxia increased the rate of pH_i recovery (compare segments i’j’ and d’e’)—data consistent with hypoxia-induced stimulation of ${\mathrm{HCO}}_3^{-}$-independent acid extrusion (as shown in Fig. 1).

From experiments similar to those shown in Fig. 3, we computed the pH_i dependence of φ_E under the different conditions, using the same approach as described for Fig. 2. From data comparable to the segment-de and -ij pH_i recoveries in Fig. 3, we computed total φ_E vs. pH_i for normoxia (Fig. 4A, open squares) and hypoxia (filled squares). For this batch of astrocytes exposed to CO₂/${\mathrm{HCO}}_3^{-}$, hypoxia only elicited a slight alkali shift of the total φ_E vs. pH_i plot. Nevertheless, the data sets are better fit (P<0.0001) with a model of 2 separate cubic polynomials than a simpler model of 1 cubic polynomial.

We next used the segment-d’e’ and -i’j’ pH_i recoveries to compute the SITS-insensitive φ_E vs. pH_i in normoxia (Fig. 4B, open triangles) and hypoxia (closed triangles). As expected from the Fig. 1 data, hypoxia alkali shifted the SITS-insensitive φ_E vs. pH_i plot.

We obtained the SITS-sensitive φ_E vs. pH_i for normoxia (open diamonds) and hypoxia (closed diamonds) by subtracting each SITS-insensitive φ_E vs. pH_i plot (that yielded Fig. 4B) from the corresponding mean total φ_E vs. pH_i plot (Fig. 4A). Note that at pH_i 6.4, for example, the SITS-sensitive φ_E is about twice as great as the SITS-insensitive φ_E; that is, NBC accounts for about 2/3 of the pH_i recovery (assuming that SITS produces a 100% blockade of NBC) from an acid load at pH_i 6.4. This “2/3” figure underestimates the true NBC flux to the extent that SITS fails to block NBC. Under normoxic conditions, the SITS-sensitive φ_E (Fig. 4C, open diamonds) gradually decreased at progressively higher pH_i values (from 6.3 to 6.85)—a finding similar to previous observations (Bevensee et al., 1997a). We observed the opposite relationship under hypoxic conditions. In summary, hypoxia increases the SITS-insensitive φ_E, but decreases SITS-sensitive φ_E (particularly at a low pH of 6.4) for astrocytes exposed to CO₂/${\mathrm{HCO}}_3^{-}$.

Effect of hypoxia on acid loading

To examine the effect of hypoxia on acid loading, we monitored the rate of pH_i recovery in astrocytes that we alkali loaded by applying and removing 5% CO₂/22 mM ${\mathrm{HCO}}_3^{-}$ (see Bevensee and Boron, 1995). For the experiment represented in Fig. 5A, normoxic astrocytes bathed in a HEPES-buffered solution had a steady-state pH_i of ∼7.3 (prior to point a). Applying CO₂/${\mathrm{HCO}}_3^{-}$ elicited an abrupt decrease in pH_i (ab), due to CO₂ influx and formation of H⁺ and ${\mathrm{HCO}}_3^{-}$. In the continued presence of the physiologic buffer, pH_i increased (bc) due to acid-extruding mechanisms such as electrogenic Na/HCO₃ cotransport, and to a lesser extent, Na-H exchange (Bevensee et al., 1997a, 1997b). Removing CO₂/${\mathrm{HCO}}_3^{-}$ caused pH_i first to increase rapidly (cd), greatly overshooting the initial pH_i (d vs a), due to the efflux of CO₂ and the subsequent consumption of H⁺ and ${\mathrm{HCO}}_3^{-}$ to replace the lost CO₂ (Boron and De Weer, 1976). Only nominal levels of intracellular ${\mathrm{HCO}}_3^{-}$ remain at the peak of the pH_i increase (point d). Subsequently, the cells recovered (de) to a pH_i similar to that at the beginning of the experiment. The mechanism responsible for the pH_i recovery is not known. We then exposed the astrocytes to ∼3% O₂ and imposed another alkali load (fgh). (Note that, in this experiment, the initial CO₂-induced pH_i decrease is hidden by the overwhelming ${\mathrm{HCO}}_3^{-}$-dependent acid extrusion.) The subsequent pH_i recovery (hi) from the alkali load was faster than the segment-de recovery.

We used the segment-de and -hi pH_i recoveries in Fig. 5A to plot linear fits to the pH_i dependence of the pH_i-recovery rate (dpH_i/dt) in normoxia (Fig. 5B, open circles) and hypoxia (closed circles). We did not compute acid-loading rates (i.e., φ_L) because H⁺ buffering power measurements at high pH_i obtained during alkali loads are not available. Buffering power measurements —particularly at high pH_i— are difficult to obtain in the astrocytes because of pronounced acid loading, even in the presence of inhibitors and ion substitutions that minimize acid-base transporter activity (Bevensee et al., 1997a). Hypoxia increased the pH_i-recovery rate at all pH_i values during the recovery and steepened the plot of dpH_i/dt vs. pH_i. We obtained similar results in two additional experiments. Overall, hypoxia steepened the plot of dpH_i/dt vs. pH_i during recoveries from alkali loads by ∼50% (P = 0.04) from a mean of 44 ± 5 × 10^-4 s^-1 to 65 ± 2 × 10^-4 s^-1.

Hypoxia and steady-state pH_i

It is well established that hypoxia/ischemia leads to a decrease in both pH_i and pH_o in vivo (Tombaugh and Sapolsky, 1993). At least in some brain-cell preparations, however, the effect of hypoxia/anoxia alone on steady-state pH_i is variable. Acute chemical anoxia (10 mM azide) of mouse neocortical neurons induces an abrupt decrease in the pH_i, which is then followed by either a further decrease, no change, or a marked increase in pH_i (Jϕrgensen et al., 1999). According to another study on acutely isolated CA1 neurons from the adult rat, acute anoxia elicits a transient decrease in pH_i (-ΔpH_i: 0.08-0.17) followed by a slower pH_i increase (ΔpH_i: 0.06-0.11) in the continued absence of O₂, and then a larger pH_i increase (ΔpH_i: 0.15-0.27) upon return to normoxia (Sheldon and Church, 2002). The slow pH_i increase during anoxia is due to a Zn²⁺-sensitive acid extrusion mechanism; possibly a H⁺ conductance, whereas the faster pH_i increase upon return to normoxia is due to a Na-dependent process; presumably a Na-H exchanger. Similar results were obtained on cultured postnatal rat hippocampal neurons (Diarra et al., 1999).

CO₂/${\mathrm{HCO}}_3^{-}$ can also influence the effect of anoxia/hypoxia on steady-state pH_i. For instance, acute anoxia elicits a small pH_i decrease of ∼0.06 in mouse CA1 hippocampal neurons bathed in CO₂/${\mathrm{HCO}}_3^{-}$, but a considerably larger pH_i increase of 0.46 in the neurons bathed in the nominal absence of CO₂/${\mathrm{HCO}}_3^{-}$ (Yao et al., 2001). The pH_i increase is due to anoxia-induced stimulation of the NHE because the increase can be blocked by removing external Na⁺ or applying the inhibitor HOE694.

In our study, we found that ∼3% O₂ elicited only a small pH_i decrease of 0.05 in hippocampal astrocytes exposed to a nominally CO₂/${\mathrm{HCO}}_3^{-}$-free solution. Hypoxia also had little effect on pH_i of astrocytes in CO₂/${\mathrm{HCO}}_3^{-}$. In a similar fashion, acute hypoxia of cultured cortical and acutely isolated hippocampal rat astrocytes elicited a relatively small, gradual pH_i decrease compared to a more pronounced pH_i decrease when the hypoxia was combined with a low pH_o and/or an ion-shifted Ringers solution (Bondarenko and Chesler, 2001).

It is important to recognize that a change, or lack thereof, in steady-state pH_i in response to hypoxia (or any other stimulus) provides little information on the activity of specific acid-base transporters. As described in Methods, pH_i is in a steady state when the rates of acid extrusion (φ_E) and acid loading (φ_L)—both of which vary with pH_i—are equal. Steady-state pH_i will increase if φ_E (e.g., due to NHE activity) rises, if φ_L (e.g., due to Cl-HCO₃ exchanger activity) falls, or both (Bevensee and Boron, 2008). This line of reasoning is important for the present study, where it is clear that hypoxia has little effect on steady-state pH_i, but certainly has differential effects on the activity of acid-base transporters (see below).

Effects of hypoxia on acid extrusion

Effects of hypoxia on acid loading

Less is known about the mechanisms by which brain cells recover from alkali loads, and even less is known about their sensitivity to hypoxia. In hippocampal neurons, removing extracellular Cl^- reveals the functional presence of the Cl-HCO₃ exchanger AE3, which most likely functions as an acid loader (Hentschke et al., 2006). As shown in Fig. 5, astrocyte pH_i recovers rapidly from an alkali load imposed by removing CO₂/${\mathrm{HCO}}_3^{-}$, and the pH_i recovery is considerably faster when the astrocytes are exposed to hypoxia. The pH_i recovery could be due to any number of ${\mathrm{HCO}}_3^{-}$-independent acid loading mechanisms, including a Ca-H pump or a Cl-OH exchanger. A Cl-HCO₃ exchanger could also be involved. According to preliminary data (Bevensee & Boron, unpublished), the recovery is inhibited by phenylpropyltetraethylammonium (PPTEA), consistent with a contribution from a K/HCO₃ cotransporter (KBC) that has previously been identified in the squid giant axon (Hogan et al., 1995a, 1995b; Zhao et al., 1995; Davis et al., 2001). If one of the aforementioned ${\mathrm{HCO}}_3^{-}$ transporters were responsible, then the transporter would have to have a very high affinity for intracellular ${\mathrm{HCO}}_3^{-}$, which presumably would have a low intracellular concentration in the nominal absence of CO₂. Nevertheless, it is possible that metabolically produced ${\mathrm{HCO}}_3^{-}$ could have provided sufficient substrate for such a ${\mathrm{HCO}}_3^{-}$-efflux mechanism. Additional studies are required to identify the acid-loading mechanisms in the astrocytes, and to determine which are stimulated by hypoxia.

Summary and significance

Although acute hypoxia has little effect on steady-state pH_i of rat hippocampal astrocytes, hypoxia stimulates ${\mathrm{HCO}}_3^{-}$-independent acid extrusion (e.g., NHE activity) and inhibits ${\mathrm{HCO}}_3^{-}$-dependent acid extrusion (e.g., NBC activity). Hypoxia also stimulates one or more acid loading mechanisms. Apparently, a low O₂ level is a signal that alters the activity of at least three pH_i-regulating mechanisms within a short timeframe of ∼10 min. It is worth noting that the effects of hypoxia on astrocyte pH_i homeostasis are reminiscent of those of growth factors on renal mesangial cells (Boyarsky et al., 1990). In quiescent mesangial cells studied in the presence of CO₂/${\mathrm{HCO}}_3^{-}$, applying any of a wide range of growth factors has little effect on steady-state pH_i but stimulates the recovery of pH_i from both acid and alkali loads—changes that would enhance the stability of pH_i in the face of acute acid and alkali loads.

In astrocytes, the differential effects of hypoxia on various acid-base transporters may help optimize the overall pH-regulatory ability of the astrocytes exposed to different forms of hypoxia/ischemia (see Lascola and Kraig, 1998). The stimulation of NHE activity by hypoxia would not be surprising based on a similar finding in neurons (see Introduction). Such stimulation would certainly help astrocytes regulate their pH_i when faced with intracellular acidosis caused by low-grade, or normoglycemic ischemia. Because this form of ischemia can also lead to a depolarization-induced astrocytic alkalosis, the stimulation of acid loading mechanisms would also be beneficial. However, during hyperglycemic and complete ischemia, astrocytes undergo compartmentalized acidosis and membranes become less permeable to ${\mathrm{HCO}}_3^{-}$. Hypoxia-induced inhibition of NBC activity would contribute to the reduced ${\mathrm{HCO}}_3^{-}$ permeability of the astrocytes. Further studies are required to examine the effects of hypoxia on pH_i/pH_o regulation in the brain in-vivo, where normal oxygen tensions are likely to be less than the 20% used in the present in-vitro study (Ndubuizu and LaManna, 2007). Such future studies will be complicated however because in-vivo oxygen tensions are dynamic and display spatial and temporal heterogeneity.

In summary, the O₂ level can be included in the list of signals such as changes in pH_i, pH_o, and ion concentrations that can regulate acid-base transporter activity of astrocytes during ischemia/hypoxia. The extent to which each of these signals contributes to altered transporter activity will depend on the type and severity of the ischemia/hypoxia.

Culturing astrocytes

Astrocytes from the hippocampi of Sprague-Dawley rats were harvested and cultured as previously described (Bevensee et al., 1997a). Briefly, hippocampi from 0-3 day-old rats were harvested and placed in ice-cold PBS supplemented with 33 mM glucose. The tissues were subjected to mechanical digestion with pipettes of decreasing pore diameter, and then enzymatic digestion with ∼0.1% trypsin (GIBCO BRL, Life Technologies Inc.) for 1 min. The cell suspension was subjected to several rounds of centrifugation (400 × g) followed by cell resuspension; twice with PBS + glucose, and twice with the tissue culture media consisting of minimum essential medium (MEM) supplemented with 28 mM glucose, 2 mM l-glutamine, 100 units ml^-1 penicillin/streptomycin (GIBCO BRL, Life Technologies Inc.), and 10% fetal calf serum (Gemini Bioproducts, Inc. or GIBCO BRL, Life Technologies Inc.). After the final centrifugation, the cell suspension was divided into two 25-cm² tissue-culture flasks. On day 4 after plating, the flasks were agitated at ∼100 RPM on an American Rotator V (Model R4140) shaker for approximately two hours to dislodge contaminating oligodendrocytes. 10-12 days after the initial plating, the flasks were again agitated for at least 12 hours before the cells were trypsinized and replated. From two preparations, astrocytes passaged three or four times were plated onto 1-cm² glass coverslips for pH_i measurements. Prior to cell plating, coverslips were sequentially washed with concentrated RadiacWash® (Biodex Medical Systems, Shirley, NY), 70-100% ethanol, and deionized water. Experiments were performed on the astrocytes 1-4 weeks after being plated onto coverslips.

Measuring pH_i

The pH_i of astrocyte populations was measured using the pH-sensitive dye BCECF and a SPEX Fluorolog-2 dual-beam spectrofluorometer (model CM1T10E; SPEX Industries, Inc., Edison, NJ) as described by Bevensee et al. (1999). Prior to an experiment, astrocytes on a coverslip were loaded with BCECF by incubating the coverslip in the standard HEPES-buffered solution containing 10 μM of the cell permeant acetoxymethylester form of the dye, BCECF-AM. The coverslip was then secured in a custom-designed flow-through cuvette, which fit into a temperature-controlled cuvette holder that was mounted in the excitation light path of the spectrofluorometer. The dye was alternately excited with 502-nm and 440-nm light every 3 s, and the emitted light at 530 nm from each excitation was captured and integrated for 1 s. The emitted light from 502-nm excitation (I₅₀₂) is pH sensitive, whereas that from 440-nm excitation (I₄₄₀) is relatively pH insensitive. Thus, the fluorescence-excitation ratio (I₅₀₂/I₄₄₀) is predominately a function of pH. Normalized ratios were converted to pH_i values using the high-K⁺/nigericin technique (Thomas et al., 1979), as modified for a single-point calibration (Boyarsky et al., 1988). Background I₅₀₂ and I₄₄₀ signals from cells without dye were subtracted from total I₅₀₂ and I₄₄₀ signals. All experiments were performed at ∼37°C.

As described by Bevensee and Boron (1995), the rate of acid extrusion (φ_E) or acid loading (φ_L) is the product of the rate of change in pH_i (dpH_i/dt) and total intracellular H⁺ buffering power (β_Total)¹. φ_E and φ_L are “pseudofluxes” with the units of moles per unit cell volume per unit time (i.e., μM s^-1). β_Total is the sum of the theoretically computed ${\mathrm{HCO}}_3^{-}$ buffering power (β_HCO3 = 2.3 × [${\mathrm{HCO}}_3^{-}$]_i) and the intrinsic buffering power (β_i). The pH_i dependence of β_i for cultured rat hippocampal astrocytes has previously been described by the linear relationship β_i = -9.98 × pH_i + 80.5 (Bevensee et al., 1997a). The pH_i dependence of β_i for astrocytes cultured from mouse cortex is unaffected by 2 h of oxygen and glucose deprivation, followed by 1 h of reoxygenation (Kintner et al., 2005).

Solutions

The standard HEPES-buffered solution contained (in mM): 125 NaCl, 3 KCl, 1 CaCl₂, 1.2 MgSO₄, 2 NaH₂PO₄, 32 HEPES, 10.5 glucose, and NaOH to pH 7.4 at 37°C. NH₃/${\mathrm{NH}}_4^{+}$-containing solutions were made by replacing 20 mM NaCl with an equimolar amount of NH₄C1. For the 10-μM nigericin solution, Na⁺ was replaced with 105 mM K⁺ and NMDG⁺. For the 5% CO₂/22 mM ${\mathrm{HCO}}_3^{-}$ solution (pH 7.4), 32 mM HEPES was replaced with 22 mM NaHCO₃ plus 3.4 mM NaCl to maintain a constant ionic strength, and the solution was equilibrated with either 5% CO₂/balance air or 5% CO₂/balance N₂ in hypoxia experiments. For hypoxia experiments performed in the nominal absence of CO₂/${\mathrm{HCO}}_3^{-}$, solutions were equilibrated with 100% N₂. P_O2 of the superfusate in the cuvette was ∼23 torr (∼3%) as measured with a platinum wire electrode (Croning and Haddad, 1998; Yao et al., 2001).

For stock solutions, BCECF-AM was prepared in dimethylsulfoxide (DMSO) and nigericin in 100% ethanol. BCECF-AM was obtained from Molecular Probes, Inc. SITS, nigericin, and other chemicals were obtained from Sigma Chemical Co. (St. Louis, MO).

Statistics

Data are reported as mean ± standard error of the mean (SEM). Levels of significance were assessed using either the paired or unpaired Student’s t-test, and P < 0.05 was considered significant. dpH_i/dt vs. pH_i during recoveries were fit with either third-order polynomials (following acid loads) or lines (following alkali loads) using a least-squares method. F tests were performed on cubic polynomial fits to plots of acid extrusion vs. pH_i using GraphPad Prism 5 (GraphPad Software, Inc.) to determine if data sets were distinct. Each pair of data sets shown in Figs. 2 and 4 was fit significantly better (P<0.05) with a model of 2 separate cubic polynomials than a simpler model of 1 cubic polynomial indicating that the data sets were distinct. Reported n values represent number of experiments, even from a single batch of cells.