INTRODUCTION

Stress-activated protein kinase/c-Jun NH₂-terminal kinase (SAPK/JNK)¹ and p38 mitogen-activated protein kinase (p38 MAPK) are members of a family of enzymes that are activated by a variety of agents and conditions that generate cellular stress. These kinases are members of two parallel, yet independent cascades, with distinct upstream activators and downstream targets. SAPK is activated by SEK (MKK4), which is in turn stimulated by MEKK or MLK3 (1-3). Similarly, p38 MAPK is activated by MKK3 and MKK6 (4, 5). The effector pathways triggered by the stress kinases also differ: the preferred substrate of SAPK is c-Jun, a component of the transcription regulator AP-1 (6, 7), whereas MAPKAPK-2, a serine/threonine kinase, and the transcription factor ATF-2 are main targets of p38 MAPK (8, 9).

The agonists and conditions that activate SAPK and/or p38 are remarkably varied. They include hyperosmolarity, inflammatory cytokines, heat shock, ultraviolet light, and protein synthesis inhibitors such as cycloheximide and anisomycin (for review, see Refs. 10 and 11). Several of these stimuli also activate the Na⁺/H⁺ exchanger (NHE), a ubiquitous family of transmembrane proteins involved in the regulation of cytosolic pH, cellular volume, and transepithelial ion transport (for review, see Refs. 12 and 13). For example, hyperosmotic exposure, in most cells, leads to the rapid activation of the NHE (14). Heat shock, another activator of SAPK, also increases NHE activity in Vero cells (15). Moreover, cytokines, including interleukin-1 and TNF, activate NHE in myocytes (16) and fibroblasts (17), respectively. Ischemia/reperfusion, which activates SAPK (18) and enhances binding of ATF-2 and c-Jun to DNA (19), has likewise been associated with increased transport by the NHE (20). Finally, cycloheximide, a protein synthesis inhibitor that stimulates SAPK, has been shown to increase NHE activity as well (21).

It is unclear whether the activation of the stress kinases and the stimulation of the ion exchanger are related. The kinases may mediate the stimulation of the antiporter, although direct phosphorylation has been ruled out as the mechanism of NHE activation (22). Conversely, the ionic changes initiated by the exchanger may trigger kinase activation. Enhanced NHE activity is expected to increase the intracellular Na⁺ concentration and elevate the cytosolic pH (pH_i). These parameters could in turn activate SAPK and/or p38 MAPK. In this work, we examined the relationship between the cytosolic pH and the activation of the stress kinases. In particular, we analyzed the effects of alkalinization on SAPK and p38 MAPK and considered the possibility that shrinkage-induced activation of NHE is required for activation of the kinases.

EXPERIMENTAL PROCEDURES

Radiolabeled ATP was purchased from Mandel/Dupont (Guelph, Ontario, Canada). Fetal bovine serum was purchased from Life Technologies, Inc., and the cell culture medium was prepared by the Media Department at Princess Margaret Hospital (Toronto, Ontario, Canada). BCECF, indo-1, nigericin, and ionomycin were purchased from Molecular Probes (Eugene, OR). All other materials were purchased from Sigma.

Antibodies to SAPK and p38 MAPK were raised in rabbit against a pGEX vector containing full-length p54 SAPK or full-length p38, respectively.

U937 cells (American Type Culture Collection, Bethesda, MD) were grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum in a humidified environment under 5% CO₂. For kinase assays, U937 cells were preincubated for 24 h in medium supplemented with only 0.5% fetal calf serum.

To measure pH_i, U937 cells were sedimented and resuspended in a Na-HEPES-buffered solution (NHB) containing (in mM): 117 NaCl, 25 Na-HEPES, 5.36 KCl, 1.66 MgSO₄, 1.36 CaCl₂, and 25 glucose, pH 7.4, at 37 °C, at a density of 2 × 10⁵ cells/ml. This suspension was then incubated with 1 µM acetoxymethyl ester form of BCECF for 15 min at room temperature. Cells were sedimented, resuspended in fresh NHB, placed into a polystyrene cuvette, and inserted into the thermally regulated (37 °C) holder of a Perkin-Elmer 650-40 fluorescence spectrophotometer. BCECF was continually excited at 495 nm, and emission was recorded at 525 nm. For each experiment, fluorescence emission was calibrated internally versus pH_i by using the high KCl/nigericin technique (23).

U937 cells at a density of 2 × 10⁵ cells/ml were sedimented, resuspended in NHB, and incubated with 2 µM acetoxymethyl ester precursor of indo-1 for 25 min at room temperature. Cells were once again sedimented, resuspended in NHB, and incubated for an additional 15 min in NHB at 37 °C. Indo-1 fluorescence was monitored as described above, with the excitation at 331 nm and the emission recorded at 410 nm. Free cytosolic calcium ([Ca²⁺]_i) was calculated as described previously (24). Briefly, F_max and F_auto were obtained by adding 5 µM ionomycin and 1 mM MnCl₂, respectively, and a dissociation constant of 250 nM for the indo-1-Ca²⁺ complex (25) was used to calculate [Ca²⁺]_i.

Following incubation under the conditions specified in the text, aliquots of 5 × 10⁶ U937 cells were lysed in hypotonic lysis buffer containing 10 mM NaCl, 20 mM PIPES, pH 7.0, 5 mM EDTA, 0.5% Nonidet P-40, 0.05% -mercaptoethanol, 0.1 mM phenylmethylsulfonyl fluoride, 100 µM Na₃VO₄, 20 µg/ml leupeptin, 50 mM NaF₂, and 1 mM benzamidine. After 20 min, the lysate was sheared through a 23-gauge needle, and insoluble material was removed by centrifugation at 21,000 × g for 10 min. Samples were normalized for the amount of protein, and SAPK and p38 were immunoprecipitated from the lysates by incubating for 1 h at 4 °C with anti-SAPK (1:500 dilution) or anti-p38 (1:250) antibodies, respectively. Immune complexes were collected by adding 20 µl of protein A-Sepharose beads to the lysate and incubating for an additional 30 min at 4 °C. The beads were then washed four times with an ice-cold solution which contained 150 mM NaCl, 16 mM Na₂HPO₄, 4 mM NaH₂PO₄, and 0.1% Triton X-100.

After immunoprecipitation, the beads were sedimented and resuspended in 20 µl of kinase buffer containing 50 mM Tris-Cl, 1 mM EGTA, 10 mM MgCl₂, and 100 µM (800 nCi) [-³²P]ATP, pH 7.5. For SAPK kinase assays we added 5 µg of a fusion protein of glutathione S-transferase with residues 5-89 of c-Jun and incubated the samples at 30 °C for 30 min. The kinase reaction was terminated by adding 40 µl of 2 × Laemmli's sample buffer. Samples were resolved by SDS-polyacrylamide gel electrophoresis on 10% acrylamide gels, stained with Coomassie Blue, destained, and dried. Autoradiography was performed using Mandel/Dupont Reflection film, and radioactivity was quantified by PhosphorImager analysis using ImageQuant (Molecular Dynamics). Kinase assays for p38 were performed similarly, except that a glutathione S-transferase fusion protein comprising the 178 carboxyl-terminal amino acids of the NHE-1 isoform of the Na⁺/H⁺ exchanger was used as a substrate. The construct encoding this fusion protein was the kind gift of Dr. L. Fliegel (University of Alberta, Edmonton, Alberta, Canada). Equal loading of the immunoprecipitated kinases was confirmed in some of the experiments by immunoblotting with either anti-SAPK or anti-p38 antibodies. To resolve SAPK, which co-migrates with the heavy immunoglobulin chain, the samples were not reduced to avoid dissociation of the IgG complex.

All values are reported as the mean ± S.E. of the number of experiments specified. Statistical differences between the control and individual experimental conditions were evaluated using Student's t test. Levels of significance are indicated in the figures (*, p < 0.05; **, p < 0.01; and ***, p < 0.001).

RESULTS

Many of the substances and conditions that increase SAPK activity, can also lead to an increase in pH_i by activating NHE (see the introduction). It was therefore of interest to establish whether these events are causally related. To this end, we studied the effects of cytosolic alkalinization on SAPK activity. pH_i was manipulated by means of weak electrolytes, and the imposed changes were monitored fluorimetrically using BCECF. As illustrated in Fig. 1A, the pH_i of suspended U937 cells increased rapidly upon addition of 30 mM NH₄Cl. In six similar experiments, pH_i rose from a resting value of 7.36 ± 0.04 to 7.76 ± 0.03 within 10 s. The alkalinization was transient with pH_i returning to near-basal levels within 10 min 7.42 ± 0.03 (Fig. 1, A and B). This recovery likely reflects gradual entry of NH₄⁺ via K⁺ channels and/or the Na⁺/K⁺ pump. We next determined the effect of this alkalinization on SAPK activity. A comparable exposure of the cells to 30 mM NH₄Cl resulted in a reproducible 4-fold increase in SAPK activity (Fig. 1C), which was noticeable as early as 5 min and was sustained for up to 30 min (Fig. 1, C and D). Treatment with the weak base did not alter the efficiency of SAPK immunoprecipitation (Fig. 1C, lower panel).

The data described above cannot discern whether NH₄Cl exerts its stimulatory effect on SAPK by alkalinizing the cytosol or by other means. To ascertain the mechanism of activation, we compared the effects of two other weak bases, namely trimethylamine (TMA) and triethylamine (TrEA). As shown in Fig. 1, A and B, these bases also induced a brisk increase in pH_i from 7.43 ± 0.01 to 7.85 ± 0.06 and from 7.41 ± 0.01 to 7.86 ± 0.04 (n = 3), respectively. Unlike the effects of NH₄Cl, however, the alkalinization induced by TMA and TrEA persisted after 10 min (Fig. 1, A and B). The much slower decay of the alkalinization reflects the slower permeation of the protonated forms of the organic amines. Both TMA and TrEA also activated SAPK (Fig. 1C). In fact, the stimulation of the kinase was greater and progressed over the course of the experiment. After a 5-min incubation in either TMA or TrEA, SAPK activity increased 6-fold (Fig. 1D). By 10 min, the stimulation was 8-fold, and it was nearly 10-fold by 30 min. The more pronounced activation of the kinase parallels the sustained alkalinization of the cytosol. Jointly, these results suggest that SAPK is responsive to changes in the cytosolic pH, regardless of the agent used to impose them.

The effect of the weak bases was not osmotic in nature, since care was taken to maintain the total osmolarity constant by reducing the content of NaCl when the bases were added. In fact, the cell volume (measured electronically using the Coulter Channelyzer) was not detectably reduced even when the bases were added on top of the normal osmotic complement of the medium, likely because the permeation of the weak base increased the osmotic content of the cells, which compensated at least in part for the increased extracellular osmolarity.

Changes in intracellular pH are often accompanied by an increase in cytosolic calcium concentration ([Ca²⁺]_i). Because it has been previously shown in T lymphocytes that elevated [Ca²⁺]_i in conjunction with TPA stimulation activates SAPK (26), we evaluated the role of this divalent cation in SAPK activation by alkalosis. For this purpose, we compared the effects of NH₄Cl to those of calcium ionophores and of thapsigargin, an inhibitor of endomembrane Ca²⁺ATPases. By inhibiting pumping into the endoplasmic reticulum, thapsigargin unmasks an endogenous calcium "leak" which results in a transient elevation of [Ca²⁺]_i. The capacitative coupling between depleted stores and the plasmalemmal calcium channels facilitates calcium influx, inducing a sustained elevation of [Ca²⁺]_i when cells are suspended in calcium-containing media (27). These phenomena were readily reproduced in U937, as shown in Fig. 2A. When exposed to 30 nM thapsigargin, U937 cells responded with a large and sustained rise in [Ca²⁺]_i from a steady-state level of 375 ± 30 (n = 10) to 1040 ± 104 nM (n = 5) (Fig. 2, A and B). Even higher levels of [Ca²⁺]_i were attained by exposure to 1 µM ionomycin, a non-fluorescent calcium ionophore. The precise [Ca²⁺]_i levels attained with ionomycin could not be defined, because they exceeded the dynamic range of the probe used (indo-1, K_d 250 nM). By contrast, [Ca²⁺]_i did not significantly increase at any time after addition of 30 mM NH₄Cl (after 5 min [Ca²⁺]_i was 238 ± 8 nM (n = 5) (Fig. 2, A and B)). In fact, exposure to NH₄Cl after [Ca²⁺]_i had been elevated by thapsigargin resulted in a sizable decrease in the cytosolic concentration of the cation. After 5 min of incubation with TMA or TrEA [Ca²⁺]_i averaged 260 ± 1 and 343 ± 4 nM, respectively. It therefore appears unlikely that alkalinization-induced activation of SAPK is mediated by an increase in [Ca²⁺]_i. This notion was confirmed by comparing the effects of the weak base on SAPK with those elicited by thapsigargin or the calcium ionophores. Thapsigargin, ionomycin, and A23187 produced only a modest stimulation of SAPK up to 10 min after addition, much smaller than that induced by anisomycin (Fig. 1, C and D). The effects of these calcium mobilizing agents are considerably smaller than those of the weak bases (cf. Fig. 1), despite the much greater effects of the former on [Ca²⁺]_i. Thus, an increase in [Ca²⁺]_i cannot explain the stimulatory effects of weak bases on SAPK.

Exposure of cells to permeating weak bases will alkalinize not only the cytoplasm but also intracellular compartments, particularly those that are maintained at an acidic pH by vacuolar H⁺-ATPases. It is therefore possible that the stimulatory effect of the weak bases on SAPK is mediated by a pH change within an endomembrane compartment. This is particularly relevant since Verheij et al. (28) showed that TNF activates SAPK via ceramide, which is generated by hydrolysis of sphingomyelin within both neutral and acidic compartments. Therefore, it was conceivable that NH₄Cl induced activation of SAPK through modulation of sphingomyelinase activity in an acidic compartment. Two approaches were used to test this possibility. First, we preincubated cells with desipramine, which has previously been shown to inhibit acidic sphingomyelinase by neutralizing the acidic compartment (29). As illustrated in Fig. 3, A and B, preincubation with 10 µM desipramine for 1 h had no effect on SAPK activity, but obliterated the ability of TNF to activate SAPK. In contrast, desipramine had no effect on the ability of NH₄Cl to activate SAPK.

The role of endomembrane acidic compartments was also evaluated using bafilomycin A₁, a potent and very selective inhibitor of vacuolar-type H⁺-ATPases. This inhibitor permeates into the cells, reaches the ATPases of intracellular compartments, and thereby dissipates their pH gradients, while affecting the cytosolic pH minimally. Unlike the weak bases, treatment with 100 nM bafilomycin for up to 1 h had only a marginal statistically insignificant (p > 0.1) effect on SAPK activity (Fig. 3, C and D). In addition, pretreatment with bafilomycin did not preclude the ability of NH₄Cl to activate SAPK. Therefore, neutralization of acidic endomembrane compartments is not likely the mechanism whereby weak bases activate SAPK.

Changes in the pH of intracellular compartments are similarly unlikely to play a role in the activation of SAPK effected osmotically or by anisomycin. This conclusion was derived from the experiments in Fig. 4. Neither desipramine nor bafilomycin, at concentrations known to inhibit acidic sphingomyelinase and the H⁺ pump, respectively, had a significant inhibitory effect on the activation of SAPK by anisomycin or by hypertonic sorbitol (Fig. 4, A-D).

p38 MAPK, a homolog of the yeast Hog1 protein, is also a member of the stress-activated protein kinase family (30). Like SAPK, p38 MAPK is activated by anisomycin, hyperosmolarity, and the cytokines interleukin-1 and TNF (10). We therefore considered the possibility that, like SAPK, p38 MAPK could also be activated by changes in cytosolic pH. Thus, U937 cells were exposed to either NH₄Cl or TMA, and the activity of immunoprecipitated p38 MAPK was assessed in vitro using as a substrate the carboxyl-terminal domain of NHE-1, which we had found earlier to be effectively phosphorylated by this kinase.² As illustrated in Fig. 5, exposure to NH₄Cl resulted in a 5-fold increase in p38 MAPK activity detectable within 5 min and maintained through 30 min. Similar results were obtained with the organic base TMA, suggesting that, like SAPK, p38 MAPK is responsive to changes in pH_i. As in the case of SAPK, the weak bases did not affect the efficiency of p38 immunoprecipitation (lower panel in Fig. 5A).

In addition to activating SAPK and p38 MAPK, hyperosmotic treatment also increases pH_i. This cytosolic alkalinization in most cells is mediated by the shrinkage-induced activation of the Na⁺/H⁺ exchanger (NHE). Since our present data show that alkalinization suffices to activate SAPK as well as p38 MAPK, we entertained the possibility that shrinkage-induced alkalinization, mediated by the NHE, is responsible for the observed activation of the kinases. When exposed to a hyperosmotic solution, U937 cells underwent a rapid intracellular alkalinization (Fig. 6A). pH increased at a rate of 0.05 ± 0.007 pH unit/min to a new steady-state pH_i of 7.58 ± 0.05 (n = 6). As reported for other cell types, this shrinkage-induced alkalinization in U937 cells was Na⁺-dependent and inhibited by amiloride (Fig. 6A). Indeed, in the absence of external Na⁺ or in the presence of the diuretic, pH_i became more acidic, at a rate of 0.02 ± 0.005 (n = 3) and 0.06 ± 0.05 pH units/min (n = 3), respectively. These findings confirm that cell shrinkage activates the NHE in U937 cells.

We then tested whether the alkalinization generated by the antiporter is responsible for stimulation of SAPK. Cells were stimulated with hypertonic sorbitol, and the extent of SAPK activation was tested in otherwise untreated cells or under conditions shown above to preclude antiporter-mediated alkalinization of the cytosol. As shown in Fig. 6B, SAPK was comparably activated in Na⁺-containing and Na⁺-free media, in the presence and absence of amiloride. Summarized data from multiple experiments are presented in Fig. 6C. We conclude that, while alkalinization alone can activate SAPK and p38 MAPK, osmotic stimulation of the NHE is not responsible for the activation of SAPK.

DISCUSSION

Because a variety of stimuli concomitantly activate SAPK and the NHE, we considered the possibility that these events are related. Our data indicate that a cytosolic alkalinization of a magnitude comparable to that attained by stimulating the antiport suffices to activate SAPK and p38 MAPK. While these data are suggestive of a causal relationship, subsequent experiments demonstrated that activation of the kinases occurs even when NHE-induced alkalinization is precluded pharmacologically or by ionic substitution. These findings rule out that the activation of SAPK and p38 MAPK is secondary to the activation of Na⁺/H⁺ exchange. The converse relationship, namely that the NHE is stimulated by a pathway involving the stress kinases, remains a viable possibility. Alternatively, the two events may lie on parallel pathways, which could conceivably share common upstream elements. In this regard, independent studies have shown that members of the Rho family of small GTP-binding proteins can stimulate SAPK (2, 31) as well as NHE activity (32). Cdc42 has been found to activate Rac which in turn can activate Rho (33). These GTP-binding proteins regulate the formation of filopodia, lamellipodia, and stress fibers, and it is noteworthy that NHE-1, the "housekeeping" isoform of the antiporter, has been reported to accumulate at or near these structures (34). Hence, it is possible that activation of Cdc42, Rac, and/or Rho promotes the interaction between the NHE and the cytoskeleton, thereby increasing antiport activity, as well as activating SAPK pathways.

Alternatively, heterotrimeric G proteins could be the common step leading to the parallel activation of NHE and the stress kinases. Prasad et al. (35) demonstrated that constitutively active GTPase-deficient mutants of G₁₂ and G₁₃ promote the activation of SAPK. Interestingly, G₁₃ has also been shown to activate the NHE (36), seemingly via pathways involving small GTP-binding proteins of the Rho family and MEKK1 (32). Thus, G₁₃ may give rise to the coordinate, yet independent activation of SAPK and NHE by cell shrinkage or other stimulants.

Parallel yet independent activation of the stress kinases and of the NHE is also suggested by the diverging time courses of these events in cells challenged with hypertonic solutions: cation exchange is noticeable and attains maximal rate within seconds, while full osmotic activation of SAPK or p38 MAPK is delayed, reaching maximal level tens of minutes after cell shrinkage (see Ref. 37). This temporal disparity suggests that the two responses may have different functional roles in cell volume homeostasis. We speculate that the early response of the antiporter is intended to accomplish the acute phase of regulatory volume increase, an immediate defense against osmotic perturbation. In addition, however, chronic exposure of cells to hyperosmolarity is known to be counterbalanced by a slower accumulation of organic osmolytes (38). The latter process depends on an increase in biosynthetic enzymes (39) and in the abundance of organic osmolyte transporters (40), which are in turn associated with elevated mRNA levels (41). Therefore, it is conceivable that activation of the stress kinases signals an increase in transcription via activation of c-Jun or ATF-2, to prepare the cell for a prolonged period of hyperosmotic exposure.

In summary, stressful situations seemingly activate both NHE as well as SAPK and p38 MAPK. While alkalinization such as that generated by Na⁺/H⁺ exchange is capable of stimulating the stress kinases, neither chemical (anisomycin) nor physical stresses (hypertonicity) require a pH change to exert their stimulatory effect on the kinases. Nevertheless, it is conceivable that other situations leading to stimulation of Na⁺/H⁺ exchange may secondarily lead to activation of the stress kinases. Stimulation of the exchanger can be induced by integrin engagement, activation of mitogenic receptors, and by some hormones, and some of these treatments also result in activation of stress kinases.

Because upon osmotic cell shrinkage stimulation of NHE precedes activation of the kinases, we find it unlikely that SAPK and/or p38 MAPK mediates the stimulation of the antiporter. Instead, we favor the hypothesis that the two events are parallel yet independent responses, perhaps triggered by a common early event such as activation of small or heterotrimeric G proteins. The divergent activation of these pathways may provide the cell with separate complementary responses to the early and sustained phases of stressful perturbations.