INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

MUCH OF THE AMMONIA produced in the brain, or reaching the brain by traversing the blood-brain barrier, is trapped in the form of glutamine after entry into astrocytes (14), where glutamine synthase activity is primarily located (32). Uncharged NH₃ is relatively lipid soluble and, in general, diffuses freely across cell membranes (14). However, <2% of total ammonia is uncharged at pH 7.4, assuming a pK_a of 9.2. The much more abundant charged species NH⁺₄ has a hydrated ionic radius similar to K⁺ and is known to share K⁺ transport pathways (17, 20). Thus transmembrane fluxes of NH⁺₄ also are appreciable. The charged and uncharged species have distinctly different effects on pH in compartments between which they migrate. When NH₃ is the predominant form entering a cell, the association of NH₃ with H⁺ to form NH⁺₄ tends to alkalinize the cytoplasm, whereas predominant entry of NH⁺₄ can acidify by reversing this reaction. At physiological ammonia concentrations of 0.1-0.2 mM, such transmembrane fluxes may present a load to pH regulatory mechanisms without appreciably altering intracellular pH (pH_i) (9). However, significant pH changes are detected at pathophysiological blood ammonia levels of 0.5 mM (11, 22, 40). In this report, the term "ammonia" is used to indicate the total of the charged and uncharged species, whereas the separate species are identified by their chemical formulas.

The transport of ammonia in the individual cell types of the brain has received little direct attention. However, use of the ammonia-prepulse technique (3) to create an intracellular acid load is a common maneuver in studies of pH regulation in brain cells. In such studies the responses observed in neurons and glia have generally conformed to the pattern exhibited by many other cell types (7, 37). A high extracellular concentration of NH₄Cl (usually 20 mM) elevates pH_i as NH₃ rapidly enters the cell and is protonated. After dissipation of the inward concentration gradient of NH₃, the new plateau level of pH_i begins to decline slowly in the continued presence of NH₄Cl. A slow influx of NH⁺₄ is believed to contribute to this decline (3). Washout of NH₄Cl from the bathing solution then induces a large intracellular acid load as NH₃ diffuses outward and intracellular NH⁺₄ dissociates to form protons. This commonly observed pattern is consistent with the view that diffusion of NH₃ is much faster than transport of NH⁺₄. As a consequence, it is believed that brain cells are unable to maintain a concentration gradient of NH₃. According to this view, the distribution of total ammonia across cell membranes is determined solely by its pK_a and the values of pH in the extracellular and intracellular compartments (3, 14, 36, 42).

The present study arose from the observation of an unconventional response of pH_i to 20 mM NH₄Cl in primary cultures of mouse cerebral astrocytes. The response began with a typical alkalinization, but this reversed abruptly within a few seconds to be replaced by an intense sustained acidification in the continued presence of the NH₄Cl. There is a recent preliminary report of ammonia-induced acidification also in glia acutely isolated from the honey bee drone retina (13). An acidifying response to ammonia has been described and explored extensively in the thick ascending limb of Henle's loop, where it is attributable to unusually rapid transport of NH⁺₄ via K⁺ transport pathways, especially across the apical membrane of the epithelial cells (16, 17, 24). Here we have characterized the effects of NH₄Cl on pH_i in mouse astrocytes, and we have attempted to interpret these effects in terms of the fluxes of NH₃ and NH⁺₄. We used selective inhibitors of K⁺ transport to identify entry routes for NH⁺₄. Comparative measurements were made in rat astrocyte cultures, which are known to exhibit a membrane permeability for K⁺ that is much lower than in mouse astrocyte cultures (45, 46). The membrane transport of NH⁺₄ is likely to impact the regulation of pH and the distribution of K⁺ in astrocytes, particularly in metabolic disorders associated with hyperammonemia.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Astrocyte culture. Cell cultures of astrocytes were prepared from the cerebral hemispheres of 1-day-old outbred mice (CD-1, Charles River) or the hippocampi of 1-day-old outbred rats (CD, Charles River), as described previously in detail (10). The cultures were grown to confluence in 35-mm dishes each containing a 9 × 31-mm glass coverslip. They were used when 3-6 wk old.

Measurement of pH_i. The coverslip cultures were transferred to N-2-hydroxyethylpiperazine-N'-2-ethanesulfonic acid (HEPES)-tris(hydroxymethyl)aminomethane (Tris)-buffered salt solution (HTB) containing (in mM) 150 NaCl, 3 KCl, 2 CaCl₂, 0.8 MgCl₂, 5 D-glucose, and 10 HEPES. The pH of the solution was adjusted to 7.4 or other required value with Tris base, and the osmolarity was raised to 315 mosM with mannitol. K⁺ concentration ([K⁺]) was adjusted by equimolar substitution of KCl for NaCl. The astrocytes were loaded with the fluorescent pH_i indicator 2',7'-bis(2-carboxyethyl)-5(6)-carboxyfluorescein (BCECF) by incubation with a solution of its acetoxymethyl ester (2 µM; Calbiochem, San Diego, CA). The coverslip culture was then washed, mounted in a fluorometric cuvette, and allowed to equilibrate for 10 min in HTB at 35°C. Intracellular BCECF fluorescence was monitored in ratio mode by alternating the excitation wavelength between 500 and 440 nm and measuring emitted fluorescence at 530 nm. The sampling rate of 1 Hz could not fully resolve the peak of initial alkalinization in response to NH₄Cl when it was followed by rapid acidification. Consequently, the reported amplitudes of alkalinization may variably underestimate actual peak values, depending on the subsequent rate of acidification. Calibration of the fluorometric signal and estimation of absolute pH_i are considered in detail in Calibration of BCECF.

Calibration of BCECF. Questions have been raised about the adequacy of the nigericin-K⁺ method of Thomas et al. (43) for calibrating the fluorometric signal ratio obtained with BCECF. Nett and Deitmer (30) reported that the ion-sensitive microelectrode and BCECF techniques measured similar shifts of pH_i (pH_i) in leech glia. However, the absolute value of pH_i given by nigericin-K⁺ calibration of BCECF was 0.12 pH unit higher than the microelectrode measurement. Similarly, Boyarsky et al. (6) found that nigericin-K⁺-calibrated estimates of resting pH_i in BCECF-loaded smooth muscle cells were ~0.2 pH unit higher than the value predicted by a bracketing technique using "null solutions." These null solutions are mixtures of membrane-permeant weak acid and weak base designed to elicit zero pH_i at a theoretically specified value of pH_i.

View larger version (18K): [in this window] [in a new window]   Fig. 1.   Calibration of fluorometric signal ratio obtained with 2',7'-bis(2-carboxyethyl)-5(6)-carboxyfluorescein (BCECF). A: effects on intracellular pH (pHi) of pH 7.2-null (left trace) and pH 7.0-null (right trace) solutions of trimethylamine propionate (6), each followed by addition of 5 µM nigericin + 1 µM gramicidin to collapse transmembrane H+ gradient. Upward deflection of trace indicates increasing pHi. Transient acidification elicited by nigericin + gramicidin is assumed to reflect an initial inflow of H+ before dissipation of membrane potential. Extracellular pH was constant at 7.4 throughout. B: plots of signal ratio given by BCECF-loaded astrocytes responding to imposed changes in pH of HEPES-Tris-buffered salt solution (HTB) containing 5 µM nigericin-125 mM K+, with and without 1 µM gramicidin. Validity of calibration depends on pHi and extracellular pH being equalized (see MATERIALS AND METHODS). Linear regressions are shown for data obtained from 1 culture representative of 3 experiments yielding similar results.

Statistics. Values are means ± SE for the number of coverslip cultures (n). The significance of differences was determined by analysis of variance using the Newman-Keuls test for multiple comparisons.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Figure 2 illustrates the unusual effect of NH₄Cl on pH_i in mouse astrocytes. Shown superimposed is the response to an equal concentration of methylamine hydrochloride, representing a more typical pattern of response to an acute intracellular alkaline load. In the case of methylamine, rapid intracellular alkalinization induced by influx of the membrane-permeant base was followed by a slow relaxation of pH_i toward the initial resting value. The rate of this slow decline of pH_i may reflect metabolic production of acid, regulatory outward transport of HCO₃ (12), and slow entry of the relatively impermeant charged base (4). By contrast, the alkalinization produced by NH₄Cl was transient, because it was superseded by a rapid acidification leading to a sustained reduction of pH_i below the initial resting value.

View larger version (16K): [in this window] [in a new window]   Fig. 2.   Superimposed typical responses of pHi to methylamine hydrochloride (CH3NH3Cl) or NH4Cl (added at arrow) in mouse cerebral astrocytes. Bathing solution was HTB (pH 7.4) equilibrated with air at 35°C. Methylamine response is reproduced from Fig. 3B of Ref. 29.

Concentration dependence of the effect of NH₄Cl on pH_i in mouse astrocytes. The responses of pH_i to 0.25-20 mM NH₄Cl shown in Fig. 3 were measured in nominally HCO₃-free solution (HTB in air) to reduce the buffering power of the cytoplasm and thus amplify effects on pH_i. The initial alkalinization in mouse astrocytes peaked within the first 5 s, and its amplitude declined to the detection threshold when the concentration of NH₄Cl added was <1 mM. The mean amplitude of the transient alkalinization in response to 1 mM NH₄Cl was 0.056 ± 0.002 pH unit (n = 57, cumulative mean). As noted in MATERIALS AND METHODS, this is likely to be an underestimate because of the limitation on data sampling rate.

View larger version (24K): [in this window] [in a new window]   Fig. 3.   Concentration dependence of effect of NH4Cl on pHi in mouse and rat astrocytes. Plotted against NH4Cl concentration (logarithmic abscissa) are peak increase in pHi (alkalinization, positive pHi) relative to initial resting value (A), initial rate of acidification (negative dpHi/dt expressed as pH units/min) after peak of alkalinization (B), and peak decrease in pHi (acidification, negative pHi) relative to initial resting value (C). Values are means ± SE for mouse cerebral astrocytes (solid lines, n = 3-6) and rat hippocampal astrocytes (dashed lines, n = 4). The mean values for 1 mM NH4Cl include data published previously (Fig. 4 of Ref. 8). Sample responses to each concentration of NH4Cl (mM) are shown on right, arbitrarily displaced on vertical axis for clarity.

1

1

1

1

Comparative responses of rat astrocytes. Figure 3 includes comparative effects of 1-20 mM NH₄Cl on pH_i in primary rat hippocampal astrocytes. Whereas the amplitude of the transient alkalinization was not different, dpH_i/dt and the maximal acidification were much smaller in rat than in mouse astrocytes. For the four concentrations of NH₄Cl tested, dpH_i/dt in rat astrocytes was a consistent 10.7 ± 0.8% of the rate in mouse astrocytes. This corresponds to the smallest difference in K⁺ flux observed between mouse and rat astrocytes under various conditions of culture (i.e., 10-fold greater in mouse; see Ref. 46). If it is assumed that these different rates of acidification reflect a difference in NH⁺₄ influx, the lack of a difference in the transient alkalinization (Fig. 3A) is an indication that NH⁺₄ influx was not fast enough, even in mouse astrocytes, to materially reduce the inward gradient of NH₃ concentration before its dissipation by NH₃ diffusion. All the remaining observations were made with mouse astrocytes.

Effect of the buffer on the response to NH₄Cl. Physiological HCO₃ buffering (26 mM HCO₃ in EBS) increases buffering power by ~23 mmol · l¹ · pH unit¹ (12) to a new total of 38.8 mmol · l¹ · pH unit¹. Thus, although dpH_i/dt in response to 1 mM NH₄Cl declined to 0.203 ± 0.040 pH unit/min (n = 7) in EBS in 5% CO₂-95% O₂ (Fig. 4) and the maximal acidification declined to 0.107 ± 0.022 pH unit, there was little change in the estimate of the initial rate of H⁺ entry in the form of NH⁺₄ (31.5 nmol · min¹ · mg protein¹).

View larger version (34K): [in this window] [in a new window]   Fig. 4.   Effects of bicarbonate and extracellular pH on response of pHi to 1 mM NH4Cl. A: responses of pHi to 1 mM NH4Cl in bicarbonate-buffered solution [Earle's balanced salts (EBS) in 5% CO2-95% O2], nominally bicarbonate-free solution (HTB in air), and bicarbonate-depleted solution (HTB in 100% O2) are superimposed. B: alkalinization, dpHi/dt, and acidification (as defined in Fig. 3 legend) plotted as a percentage of mean control values (n = 6-8) obtained in HTB in air at pH 7.4. Error bars, SE (n = 5-7). * P < 0.05, different from controls. Initial resting values of pHi (means ± SE, n = 5-7) are shown for each solution.

Effect of extracellular [K⁺] on the response to NH₄Cl. Evidence that NH⁺₄ enters cells via K⁺ transport pathways includes mutual inhibition of transport by these ions. Figure 5 shows the results of experiments examining the effect of extracellular [K⁺] ([K⁺]_o) on responses of pH_i to 1 mM NH₄Cl. The reduced rate and amplitude of acidification in the presence of elevated [K⁺]_o are consistent with inhibition of NH⁺₄ entry. K⁺-induced membrane depolarization will, of course, also affect the conductance of voltage-sensitive ion channels and the driving force for passive NH⁺₄ entry. The intracellular concentration of total ammonia at steady state is expected to decline, because resting pH_i increased on raising [K⁺]_o (10). This may account for the decrease in amplitude of the transient alkalinization (Fig. 5A), since less NH₃ must enter to reach a steady-state distribution of ammonia.

View larger version (23K): [in this window] [in a new window]   Fig. 5.   Response of pHi to NH4Cl depends on extracellular K+ concentration. Plotted against K+ concentration in HTB in air (logarithmic abscissa) are alkalinization (A), dpHi/dt (B), and acidification (C) elicted by 1 mM NH4Cl (as defined in Fig. 3 legend). Error bars, SE; number of coverslip cultures is shown in parentheses. D: sample response to 1 mM NH4Cl at each indicated K+ concentration. Dashed horizontal lines, initial resting pHi (mean ± SE appended). These traces are arbitrarily displaced on vertical axis for clarity.

Block of NH₄Cl-induced acidification by Ba²⁺. The sensitivity of NH₄Cl-induced acidification to inhibition by low concentrations of Ba²⁺ is shown in Fig. 6. Near-maximal inhibition of the rate and the amplitude of acidification was apparent at 10 µM Ba²⁺, similar to the Ba²⁺ sensitivity shown by K_ir currents in rat spinal cord astrocytes (34). The initial rate of NH⁺₄-induced acidification decreased by 80.6 ± 1.9% (n = 8) in the presence of 50 µM Ba²⁺. Ba²⁺ (10-50 µM) also increased the amplitude of the transient NH₄Cl-induced alkalinization by 38-56% (P < 0.05; data not shown).

View larger version (15K): [in this window] [in a new window]   Fig. 6.   Low concentrations of Ba2+ inhibit NH+4-induced acidification. A: control response of pHi to 1 mM NH4Cl (dotted trace) is superimposed on response to 1 mM NH4Cl recorded after addition of 10 µM Ba2+ (solid trace). Dashed horizontal line, initial resting pHi. Initial acidification rate (pH units/min; B) and peak acidification (C) elicted by 1 mM NH4Cl are plotted against previously added Ba2+ concentration. Error bars, where they extend beyond symbols, indicate SE (n = 7-11).

Effect of other inhibitors of K⁺ transport. Bumetanide, a selective inhibitor of the Na⁺-K⁺-2Cl cotransporter, reduced the initial rate and the amplitude of NH⁺₄-induced acidification, but less effectively than Ba²⁺. The initial acidification rate declined by 33.9 ± 3.1% (n = 6) in the presence of 100 µM bumetanide (Fig. 7C). The combination of 100 µM bumetanide and 50 µM Ba²⁺ was more effective than Ba²⁺ alone, preventing any net acidification below the resting pH_i in response to 1 mM NH₄Cl (Fig. 7A) in each of 10 trials. However, higher concentrations of NH₄Cl were able to elicit an acidification in the presence of these combined inhibitors (data not shown), raising the possibility of additional routes of NH⁺₄ entry.

View larger version (33K): [in this window] [in a new window]   Fig. 7.   Effects of inhibitors on response of pHi to 1 mM NH4Cl. A: inhibition by 50 µM Ba2+ and 0.1 mM bumetanide (bumet) is cumulative. Traces of responses to 1 mM NH4Cl are superimposed on a dashed horizontal line, indicating initial resting pHi. B: decline of acidification elicited by 1 mM NH4Cl is accelerated by 5 mM ouabain added 2 min previously. C: initial acidification rate and peak acidification (as defined in Fig. 3 legend) plotted as a percentage of mean control values. Error bars, SE [n = 6 for bumetanide and tetraethylammonium (TEA), n = 3 for ouabain]. * P < 0.05, different from controls.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

Our study of the unconventional response of pH_i to NH₄Cl in mouse astrocytes suggests that the velocity and functional impact of the influx of NH⁺₄ in mammalian astrocytes may be greater than previously supposed. None of our observations challenge the generally held view that diffusion of NH₃ is primarily responsible for the rapid distribution of ammonia across the astrocyte plasma membrane (14). However, we interpret our findings to mean that an intense influx of NH⁺₄ persists after the steady-state distribution of ammonia is attained. A driving force for NH⁺₄ entry continues to exist, because the equilibrium potential for NH⁺₄ tends toward the value of the equilibrium potential for H⁺, which is maintained less negative than the membrane potential of the astrocytes by pH-regulatory transport (36). Once a steady-state distribution of ammonia is reached, all the NH₃ formed by dissociation of entering NH⁺₄ during the steady state will diffuse out, and this cycle of NH⁺₄ influx and NH₃ efflux imposes a continuing intracellular acid load. From the initial rate of the acidification elicited by 1 mM NH₄Cl in bicarbonate-buffered solution, we estimate that NH⁺₄ entered at a velocity of at least 31.5 nmol · min¹ · mg protein¹, which is comparable to the maximal velocity of the avid Na⁺-dependent uptake of glutamate by these cells (21).

The sensitivity of NH⁺₄-induced acidification to inhibition by low concentrations of Ba²⁺ (34) strongly suggests that passive influx via K_ir channels is the primary entry route for NH⁺₄. Other lines of evidence are consistent with this view. NH⁺₄ is known to permeate K⁺ channels, and the characteristically high permeability of the astrocyte plasma membrane to K⁺ at resting potential is generally attributed to K_ir conductance (2). The high velocity of NH⁺₄ entry and the apparent lack of saturability of this influx, which are indicated by the observed rates of acidification, are compatible with passive transport. However, regulation of NH⁺₄ transport by pH_i (24) potentially could simulate this lack of saturability. Inhibition of NH⁺₄ influx by reduced [K⁺]_o as well as by elevated levels conforms to the observed [K⁺] sensitivity of K_ir conductance (20). The absence of inhibition we found with TEA, although at variance with the susceptibility of K_ir conductance seen in spinal cord astrocytes (34), is consistent with the resistance of this conductance to TEA block reported for hippocampal astrocytes (44). Finally, the marked difference in rate of NH⁺₄-induced acidification between mouse and rat astrocyte cultures mirrors the previously reported difference in Ba²⁺-sensitive K⁺ flux (45, 46).

Inhibition by bumetanide of one-third of the initial rate of acidification elicited by 1 mM NH⁺₄ suggests that the Na⁺-K⁺-2Cl cotransporter also plays a significant role in mediating NH⁺₄ entry, as reported for other cell types (17, 23). However, it is not clear that the effect of a high concentration of ouabain can be taken to indicate NH⁺₄ transport via Na⁺-K⁺-ATPase. The accelerated decline of NH⁺₄-induced acidification in the presence of ouabain is compatible with progressive membrane depolarization and consequent reduction of the driving force for NH⁺₄ influx. The entry of Na⁺ together with NH⁺₄ via the Na⁺-K⁺-2Cl cotransporter would accelerate such a process. The combined action of Ba²⁺ and bumetanide eliminated net acidification by 1 mM NH⁺₄. Nevertheless, some influx of NH⁺₄ may have persisted, partially masked by pH-regulatory acid extrusion. Thus additional entry pathways for NH⁺₄ are not excluded by our data. Limits on the specificity of Ba²⁺ and bumetanide also leave open the possibility of other pathways. For example, a high concentration of Ba²⁺ was shown to inhibit NH⁺₄ flux via the K⁺/H⁺ exchanger in thick ascending limb (1). Also, 100 µM bumetanide partially inhibited K⁺-Cl cotransport in red blood cells (19).

The 10-fold lower rate of NH⁺₄-induced acidification in rat astrocytes than in mouse astrocytes more likely reflects a difference in expression of K⁺ conductance in vitro than a species difference in vivo. The expression of Ba²⁺-sensitive K⁺ flux (46), K⁺-induced alkalinization (33), and K_ir conductance in rat astrocyte cultures (15) each was found to be promoted by removal of serum or by treatment with dibutyryl-cAMP. For unknown reasons, the expression of these properties in mouse astrocytes appears less affected by constituents of the nutrient medium (21, 46). Control observations of rat astrocyte cultures in this study were necessitated by earlier reports of a conventional response of rat astrocytes to the ammonia-prepulse technique (7, 26). We did not attempt to confirm the basis for this difference in NH₄Cl response. It can be argued that other in vitro models, specifically acutely isolated astrocytes or brain slices, preserve the mature astrocytic phenotype more faithfully than cell culture. However, these models are subject to their own limitations. For example, acutely isolated astrocytes are stripped of fine terminal processes, which may be major sites of K⁺ conductance (31), during tissue dissociation. In slice preparations the ionic composition of the extracellular compartment cannot be controlled with certainty.

View larger version (22K): [in this window] [in a new window]   Fig. 8.   Theoretical relationship between transmembrane distribution ratio of total ammonia and relative membrane permeability to NH3 and NH+4 for 3 different values of pHi. Distribution ratio is given by following expression (see Eq. 3 in Ref. 4) where [H+]i and [H+]o are intracellular and extracellular H+ concentrations, Ka is association constant, PB/PBH+ is relative permeability to NH3 and NH+4, respectively, and f = FVm/RT (Vm is membrane potential and F, R and T are Faraday's constant, gas constant, and absolute temperature, respectively). Calculations used pHi values shown, extracellular pH = 7.4, pKa = 9.2, Vm = 90 mV, and temperature = 37°C. Inset: expansion of foot of curves.

The buffered solution bathing the cell cultures in this study maintained constant the ionic composition of the extracellular compartment. The extracellular space in vivo is smaller and less well buffered than the intracellular compartment, so extracellular changes can profoundly affect pH_i and transmembrane NH⁺₄ fluxes (12). Accordingly, the glial alkalinization observed in experimental models of hyperammonemia can be understood in the context of an associated rise in [K⁺]_o (39). After a 6-h infusion of ammonium acetate that elevated plasma ammonia to 0.63 mM in rats, cerebrocortical [K⁺]_o reached a mean value of 11.8 mM (39). Our data show that high [K⁺]_o and physiological bicarbonate levels markedly attenuate NH⁺₄-induced acidification, whereas K⁺-induced alkalinization (Fig. 5D) was shown previously to be amplified by physiological bicarbonate buffering (10). The above combination of increased extracellular levels of ammonia and K⁺ in bicarbonate-buffered solution would induce a net alkalinization in our in vitro astrocyte preparation, consistent with findings in vivo. The intracellular acidification reported in cortical brain slices exposed to 1 mM NH⁺₄ (11) can be attributed to the short diffusion pathway to the superfusion solution limiting accumulation of extracellular K⁺ in the slice.

The physiological significance of our findings for ammonia distribution in the brain is likely to depend critically on the relative permeability of the astrocytic plasma membrane to the uncharged (B) and charged (BH⁺) forms of ammonia (P_B/P_BH⁺). The importance of P_B/P_BH⁺ is illustrated in Fig. 8, which shows the intracellular-to-extracellular distribution ratio of total ammonia plotted as a function of P_B/P_BH⁺ according to the expression proposed by Roos and Boron (4, 35). Fluxes of both species are assumed to be passive in this model. The steepness of the relationship derives from the characteristically large resting membrane potential of astrocytes (assigned the value 90 mV in Fig. 8). As Roos (35) noted, his expression approaches the Donnan distribution, dependent on the membrane potential but not on pH, when P_B/P_BH⁺ is very small. When P_B/P_BH⁺ is very large, as has been assumed previously for astrocytes, the distribution is determined by the transmembrane pH gradient and is independent of membrane potential. A value of P_B/P_BH⁺ much below 50-100 (Fig. 8) is unlikely in astrocytes, because the brain-blood distribution of ammonia has been found to range between 1.5 and 3 (14). However, our findings raise the possibility that P_B/P_BH⁺ could lie at the foot of the rising phase of the curve. For example, the ammonia distribution ratio is 4.5 when P_B/P_BH⁺ = 100 and pH_i = 7.2 (a typical value of pH_i in bicarbonate-buffered solution; see Fig. 4). This distribution ratio corresponds to a Nernst potential for NH⁺₄ of 40 mV, deviating from the Nernst potential for H⁺ (approximately 12 mV), but sufficiently removed from the resting membrane potential to provide a substantial driving force for passive NH⁺₄ entry. In the well-studied instance of the thick ascending limb, P_B/P_BH⁺ for ammonia is ~20 (16).

The functional implication of relative permeabilities near the center of the range plotted in Fig. 8 is that increasing levels of neuronal activity would modulate astrocytic processing of ammonia from a predominantly "buffering" mode to a predominantly "metabolizing" mode. In quiescent conditions, [K⁺]_o can fall below 3 mM, with a consequent decline of astrocytic pH_i <7.2 (10). This would reduce glutamine formation (9) and favor a severalfold accumulation of ammonia in astrocytes (Fig. 8) compared with extracellular and neuronal levels. In contrast, increasing neuronal activity would raise [K⁺]_o and, consequently, 1) increase astrocytic pH_i above 7.2 (10), 2) stimulate glutamine synthesis (9), and 3) competitively inhibit NH⁺₄ transport, effectively increasing P_B/P_BH⁺. As a result of outcomes 1 and 3, the ammonia distribution ratio would be depressed toward unity (Fig. 8). The inverse changes expected in pH_o (12), which are ignored in Fig. 8, would amplify the modulatory effect of [K⁺]_o. This scenario further implies the spatial migration of ammonia from quiescent to active brain regions, a direction opposite to that proposed for K⁺ (2).

The present findings also suggest that NH⁺₄ influx via active and passive K⁺ transport pathways could contribute initially to the elevation of [K⁺]_o and astrocyte swelling in acute hyperammonemia. In addition to competing for K⁺ transport pathways, NH⁺₄ potentially could impede K⁺ clearance by an effect of NH⁺₄-induced acidification to reduce K⁺ conductance and gap junctional permeability (38). The electrochemical gradients driving NH⁺₄ entry would also present an osmotic load to the astrocytes in the form of Na⁺ and Cl accompanying NH⁺₄ on the Na⁺-K⁺-2Cl cotransporter. There is a documented involvement of Na⁺-K⁺-2Cl cotransport in astrocyte volume regulation (25). Swelling of astrocytic end feet is observed early in hyperammonemia (39).

However, these putative early effects of NH⁺₄ on astrocytic K⁺ transport in acute hyperammonemia are inherently self-limiting, inasmuch as [K⁺]_o rises and diminishes or reverses them. Other mechanisms, such as the metabolic disturbances addressed by Sugimoto et al. (39), are likely to sustain astrocyte swelling in hyperammonemia.