Evidence that tyrphostins AG10 and AG18 are mitochondrial uncouplers that alter phosphorylation-dependent cell signaling.

Receptor agonists that initiate fluid secretion in salivary gland epithelial cells also increase protein phosphorylation. To assess contributions of tyrosine phosphorylation to secretion, changes in muscarinic receptor-initiated secretion (estimated from sodium pump-dependent increases in oxygen consumption) were measured in parotid acinar cells exposed to tyrosine kinase inhibitors. However, like the mitochondrial uncoupler carbonyl cyanide p-trifluoromethoxyphenyl hydrazone, tyrphostins AG10 and AG18 increased the rate of oxygen consumption and reduced cellular ATP by approximately 90% in the absence of the muscarinic agonist carbachol, indicating that these tyrphostins uncouple mitochondria. Exposure of isolated mitochondria to five structurally related tyrphostins demonstrated that their relative potencies as uncouplers differed from their in vitro kinase-inhibitory potencies due to different molecular requirements for the two effects. AG10 and AG18 blocked parotid phosphorylation events only at concentrations that reduced ATP content. The tyrosine kinase inhibitor genistein reduced ATP content by 15-20% and weakly uncoupled isolated mitochondria, but its inhibition of carbachol-mediated protein kinase Cdelta tyrosine phosphorylation and ERK1/2 activation appeared attributable to blocking tyrosine kinases directly. Carbachol itself rapidly reduced ATP content by 15-20%. Carbachol, 3'-O-(4-benzoyl)benzoyl adenosine 5'-triphosphate (P2X(7) receptor agonist), AG10, AG18, and carbonyl cyanide p-trifluoromethoxyphenyl hydrazone rapidly activated the fuel sensor AMP-activated protein kinase (AMPK); however, only AMPK activation by carbachol and BzATP was due to sodium pump stimulation. AG10 and AG18 also activated AMPK and/or uncoupled mitochondria in PC12, HeLa, and HEK293 cells. These studies demonstrate that some tyrosine kinase inhibitors produce cellular effects that are mechanistically different from their primary in vitro characterizations and, as do salivary secretory stimuli, promote rapid metabolic alterations that initiate secondary signaling events.

Signaling by G protein-coupled receptors engages a variety of downstream effectors and signaling cascades that rely on alterations in the phosphorylation of proteins on tyrosine and other residues. In salivary gland epithelial cells, secretagogues (i.e. receptor ligands that promote secretion), including the muscarinic receptor ligand carbachol, increase the tyrosine phosphorylation of various proteins (1)(2)(3)(4). Therefore, a likely hypothesis is that secretagogue-dependent changes in tyrosine phosphorylation contribute to fluid secretion in the parotid gland, one of the three main salivary glands. To examine this possibility, experiments were designed to block tyrosine phosphorylation using chemical inhibitors of tyrosine kinases, including AG18, a member of the tyrphostin (tyrosine phosphorylation inhibitor) family, and genistein. To evaluate the action of kinase inhibition on fluid secretion, the effects of these inhibitors were measured on carbachol-stimulated increases in the rate of O 2 consumption (QO 2 ) 1 of parotid acinar cells.
Carbachol and other ligands such as substance P and ␣-adrenergic agonists, which activate receptors that initiate fluid secretion, promote increases in the QO 2 of parotid acinar cells and other salivary gland epithelial cells (5)(6)(7)(8)(9)(10). Receptor-initiated increases in fluid movement across epithelia are accompanied by quantitative increases in the rate of cell respiration due to the tight coupling between mitochondrial oxidative metabolism and the sodium pump (Na,K-ATPase) activity (5,7,11,12), which is activated when net secretion and net reabsorption are stimulated. In parotid acinar cells, the initiation of fluid secretion is promoted by the elevation of [Ca 2ϩ ] i , the opening of potassium and chloride channels, and the entry of sodium into the cells via multiple transport systems (13,14). The sodium pump is activated by the rapid increase in the intracellular sodium/potassium ratio due to secretagogue-promoted potassium loss and sodium entry. Changes in the activity of the ATP-consuming sodium pump alter the cytosolic ATP/ADP ratio in secretory and absorptive epithelial cells, and the rates of oxidative phosphorylation increase or decrease in response to increases or decreases in net ion movement. In previous studies, we and others determined that agents that reduced or blocked fluid secretion in parotid acinar cells also reduced or blocked the muscarinic receptor-stimulated increases in QO 2 (5)(6)(7)(8). Therefore, it seemed that the contribution of tyrosine phosphorylation to fluid secretion could be examined in parotid acinar cells by determining whether tyrosine kinase inhibitors altered increases in carbachol-stimulated rates of respiration. Accordingly, the initial experimental focus was to determine whether AG18 and genistein affected the stimulation of respiration by carbachol; however, preliminary findings led to a revision of this approach and to an evaluation of the inhibitors themselves.
Genistein and some tyrphostins, including AG18, have been used as broad-spectrum tyrosine kinase inhibitors. Tyrphostins have been used to implicate tyrosine kinases in various cell signaling and biological events, such as sodium pump (15) and chloride channel (16) activation in epithelial cells, mitogenactivated protein kinase activation (17,18), and aldosterone secretion in rat cortex (19). In initial experiments with parotid acinar cells, the QO 2 was not increased by carbachol in cells exposed to AG18. This was surprising, since it was not expected that tyrosine kinases would play such a dominant role in mediating secretory effects downstream of receptor stimulation. However, subsequent experiments indicated that the ineffectiveness of carbachol was due to the fact that AG18 uncoupled the mitochondria and stimulated a nearly maximal increase in the QO 2 independent of receptor activation. A large number of tyrphostins have been synthesized based on a common structural core of benzylidinemalononitrile, and structurally related tyrphostins based on this core had differing potencies and degrees of effectiveness as kinase inhibitors (20,21). Therefore, to determine the relevant structural moieties of AG18 that produced its uncoupling effects and to determine whether the structural moieties relevant to its uncoupling activity were related to its reported effectiveness as a kinase inhibitor, the effectiveness of AG18 as a mitochondrial uncoupler was compared with a number of tyrphostins of related structure.
The effects of tyrphostins and other agents were examined on several phosphorylation events evoked by muscarinic receptor activation in parotid cells: the Src-dependent tyrosine phosphorylation of PKC␦ and the activation of the mitogen-activated protein kinases ERK1 and ERK2. The effects of carbachol and the tyrosine kinase inhibitors also were examined on AMPactivated protein kinase (AMPK) in parotid acinar cells and other cells. AMPK is a fuel-sensing enzyme that is an ␣, ␤, ␥ heterotrimeric serine/threonine kinase, and it may be activated as a protective mechanism when the cellular ATP/AMP ratio is decreased (22,23). AMPK blocks pathways that consume ATP, such as cholesterol and glycogen synthesis, and promotes mechanisms that can increase cellular ATP content, such as glucose uptake and fatty acid oxidation (24). Since the parotid ATP content was altered by kinase inhibitors as well as by receptor stimuli, their effects on AMPK were compared.
The results of these studies highlight differences between the effects of kinase inhibitors in vitro and on a variety of cell types and demonstrate that these inhibitors can produce multiple actions and consequences that affect the interpretation of their biological actions. These studies also evaluate metabolic changes produced by secretory stimuli acting on freshly dissociated salivary gland epithelial cells and enhance our understanding of cellular responses to ligands that initiate fluid secretion.
Cell Preparations and Solutions-Freshly dispersed rat parotid acinar cells were prepared as described previously (25). The cells were suspended at 1-1.5 mg/ml in a medium (solution A) composed of the following: 116.4 mM NaCl, 5.4 mM KCl, 1 mM NaH 2 PO 4 , 25 mM Na-HEPES, 1.8 mM CaCl 2 , 0.8 mM MgCl 2 , 5 mM sodium butyrate, 5.6 mM glucose, pH 7.4. In experiments using BzATP, solution A contained 1 mM CaCl 2 and no added MgCl 2 . Cells were maintained on ice prior to use. Samples of the cell suspension were stirred and equilibrated at 37°C for at least 15 min prior to use. In some experiments, cells were pretreated with inhibitors (dissolved in dimethyl sulfoxide) or vehicle (0.1% dimethyl sulfoxide) for 20 min prior to use, as noted. PC12 cells were grown in Dulbecco's modified Eagle's medium containing 10% heat-inactivated horse serum, 5% heat-inactivated calf serum, and 1% penicillin/streptomycin. HeLa cells and HEK293 cells were grown in Dulbecco's modified Eagle's medium containing 10% fetal calf serum. All three cell lines were grown to ϳ80% confluence in 100-mm dishes.
Oxygen Consumption-For measurements using parotid cells, conditions were similar to those reported previously (26). In brief, parotid acinar cells suspended in solution A were prewarmed for 15 min at 37°C prior to placing them in a stirred chamber maintained at 37°C. A Clark-type O 2 electrode (model 125/05; Instech Laboratories) was used to record the disappearance of O 2 from a closed 400-l chamber. The output from the O 2 electrode amplifier (Instech) was recorded using a Kipp & Zonen (model BD112) flatbed chart recorder. At the end of each measurement, an aliquot of the cell suspension was collected for protein analysis (27) so that the rate of O 2 consumption (QO 2 ) could be normalized to the protein content of the cells (ϳ1 mg/ml). The QO 2 value was calculated from the linear QO 2 that was achieved upon the addition of an agent to the chamber. The ⌬QO 2 values for parotid and other cells are the differences between the O 2 consumption rates under initial conditions and the new rates achieved upon the addition of various agents.
For measurements using cultured cell lines, cells were exposed briefly to trypsin/EDTA solution (BioWhitaker), transferred to a 15-ml conical tube, subjected to centrifugation, resuspended in solution A containing 1 mg/ml trypsin inhibitor (Sigma) for 1-2 min, spun down again, and resuspended in fresh solution A. QO 2 measurements were carried out at 37°C in a similar manner to the experiments using parotid cells.
For QO 2 measurements using isolated mitochondria, aliquots of mitochondria suspended in IM were diluted ϳ1:9 with solution B (120 mM KCl, 10 mM KH 2 PO 4 , 5 mM HEPES, 1 mM EGTA, pH 7.2). Pyruvate (10 mM) and malate (0.5 mM) were added as substrates during the incubation of mitochondria at 37°C for ϳ5 min prior to sealing a 1.85-ml chamber to commence the measurement of the QO 2 . A Clark-type O 2 electrode (YSI Inc.) was used to record the disappearance of O 2 . At the end of each QO 2 measurement, a sample of the mitochondrial suspension was collected for protein analysis. The mitochondria were used at ϳ3-4 mg protein/ml. ATP Assay-In experiments to examine the effects of various agents on the ATP content (Fig. 4), aliquots of parotid acinar cells were incubated at 37°C for ϳ3 min, after which they were exposed to inhibitors or vehicle (0.1% dimethyl sulfoxide) for an additional 20 min at 37°C. Cells were rapidly sedimented, and ATP was extracted from the cell pellets by lysing them in 6% perchloric acid. After 15 min, the mixture was subjected to centrifugation (15 min, 15,000 ϫ g), the supernatants were transferred to fresh tubes and neutralized using K 2 CO 3 , and the perchloric acid-precipitated proteins were saved for quantification via protein assay (27). The ATP content in the neutralized supernatant was measured spectrophotometrically (26) after storing the sample at Ϫ80°C. In experiments in which the time-dependent effects of carbachol on ATP content were examined in control and genistein-treated cells (Fig. 6), 1.9-ml cell aliquots were treated with genistein (100 M) or vehicle (0.1% dimethyl sulfoxide) at 37°C for 20 min, and 400-l samples (with carbachol) were removed sequentially and added to microcentrifuge tubes containing ice-cold perchloric acid (6% final concentration). The samples were subjected to centrifugation and treated as described above. For each condition in each experiment, samples from 2-4 cell aliquots were collected, and the mean value of each condition was calculated as n ϭ 1 experiment. The results from multiple experiments using separate acinar cell preparations were averaged, and the S.E. values were calculated.
Immunoprecipitations and Western Blotting-Parotid acinar cells were exposed to various agents or vehicle and then were collected by a brief spin in a microcentrifuge (Model 5415C, Eppendorf, Wesbury, NY). The supernatant was removed, and cells were lysed in ice-cold lysis buffer as previously described (2). Lysates cleared of insoluble proteins were added to fresh tubes and diluted 1:1 with 2ϫ sample buffer. In some experiments, prior to adding sample buffer, PKC␦ was immunoprecipitated using ϳ0.5 g of monoclonal PKC␦ per 1 ml of lysate. Immunoprecipitated proteins were collected using 4 mg of protein A-Sepharose beads, washed, and diluted with 2ϫ sample buffer. The composition of the buffers and the washing protocols for the immunoprecipitations were performed as described previously (2). For Western blotting, nitrocellulose filters containing the transferred proteins were exposed to the following dilutions of antibodies: anti-PKC␦ (polyclonal), 0.2 g/ml; anti-Tyr(P), 1 g/ml; anti-phospho-ERK1/2, 1:1000; anti-ERK2, 0.1 g/ml; anti-phospho-AMP kinase, 1:1000; anti-AMP kinase, 1:1000. Proteins were visualized using chemiluminescence techniques and x-ray film. Data-The mean values ϩ S.E. of n number of independent experiments (each obtained from a separate cell preparation) are as indicated throughout. Differences between control and experimental samples for the accumulated data were evaluated using a two-tailed Student's t test. All immunoblots are representative of at least three separate experiments.

Effects of Tyrphostins on QO 2 of Parotid Cells and Isolated
Mitochondria-The five tyrphostins and other compounds used in these studies are shown in Fig. 1. These tyrphostins have closely related benzylidenemalononitrile-based structures and were developed originally as potential epidermal growth factor receptor (EGFR) kinase inhibitors (20,21). Subsequently, a number of studies have used several of these compounds, including AG18, as general tyrosine kinase inhibitors to block tyrosine phosphorylation and events downstream of tyrosine phosphorylation. The tyrphostins are structurally similar to tyrosine, and many of them are competitive with exogenous substrate (21).
In initial experiments, carbachol did not increase the QO 2 in AG18-treated cells (not shown), a finding that could be consistent with a major dependence of fluid secretion on tyrosine phosphorylation. However, the QO 2 of cells treated for 15-20 min with AG18 (100 M) consistently was much larger than that for untreated cells. In contrast, the basal QO 2 and the carbachol-stimulated QO 2 for cells treated with AG9 (100 M, 20 min) were similar rates to those found in cells treated with vehicle (0.1% dimethyl sulfoxide) alone. AG9 is a relatively ineffective tyrosine kinase inhibitor that is structurally related ( Fig. 1) to AG18 and often is used as a negative control for tyrphostins. When tyrphostins were examined for acute effects on parotid cell respiration, AG9 (100 M) did not produce an acute increase in the basal QO 2 , but AG18 (100 M) produced a rapid and large increase in the QO 2 (Fig. 2). Respiratory increases produced by the muscarinic agonist carbachol were blocked in cells treated with the sodium pump inhibitor ouabain, but ouabain did not block the respiratory increase produced by AG18 ( Fig. 2A). Exposure of parotid cells to the cationophore nystatin, which increases intracellular sodium and decreases intracellular potassium levels (7), increased the QO 2 in a receptor-independent manner, since the sodium pump was activated by the increased intracellular sodium/potassium ratio. Ouabain reduced the nystatin-stimulated QO 2 , but AG18 stimulated the QO 2 of ouabain-treated parotid cells that had been treated first with nystatin or carbachol ( Fig. 2A). This suggested that AG18 promoted the uncoupling of mitochondria and was not acting as a cationophore to stimulate the sodium pump. In fact, the increased QO 2 values of parotid cells exposed to 100 M AG18 quantitatively were similar to those of cells treated with the mitochondrial uncoupler FCCP (Fig. 2B). Like AG18, the effects of FCCP on cell respiration also were not blocked by ouabain (Fig. 2A). These results suggested that AG18, a tyrphostin used as a general inhibitor of tyrosine kinases, is a mitochondrial uncoupler and thereby unsuitable for the initial experimental design.
To understand more about the relationship between tyrphostin structures and their potential to act as mitochondrial uncouplers, additional tyrphostins of similar structure to AG18 and AG9 were screened by monitoring alterations in the QO 2 of parotid acinar cells. AG30 (100 M) and AG43 (100 M) did not increase the basal QO 2 , but AG10 (100 M) produced a rapid and large increase in the basal QO 2 similar to that of AG18 (Fig. 2B). Like AG18, AG10 also stimulated the QO 2 of ouabain-treated parotid cells that had been treated first with nystatin or carbachol prior to tyrphostin addition (not shown). AG10 and AG18 (10 -100 M) produced similar concentrationdependent increases in the QO 2 of parotid acinar.
To examine more fully the potential of tyrphostins to uncouple mitochondria and the structural basis of this response, the effects of these tyrphostins were measured on the respiration of mitochondria isolated from rat liver. Similar to the results obtained using dissociated parotid cells, AG10 (100 M) and AG18 (100 M) stimulated large increases in the QO 2 of isolated mitochondria (Fig. 3A). Large increases also were produced by ADP, for which O 2 consumption is coupled to ATP production (i.e. oxidative phosphorylation), and FCCP. In contrast to effects of AG10 and AG18, the mitochondrial QO 2 was not increased by AG9 (100 M), AG30 (100 M), and AG43 (100 M) (Fig. 3B). AG10 and AG18 stimulated the mitochondrial QO 2 in a concentration-dependent manner between 10 and 100 M. At 100 M, AG10 and AG18 stimulated the QO 2 to about the same level as FCCP (10 M). The concentration dependence (EC 50 ϳ20 M) of AG10 and AG18 for increasing the QO 2 of isolated mitochondria (Fig. 3B) was similar to their effect on parotid acinar cells (Fig. 2B). These findings indicate that these two tyrphostins are mitochondrial uncouplers and presumably act as weak acids to carry protons across the mitochondrial membrane. In terms of evaluating the design of molecules with biological effects, it was interesting that 1) tyrphostins of very similar structure (Fig. 1) could have very different effects on mitochondrial respiration, and 2) the relative potencies (AG30 Ͼ AG18 Ͼ AG10 Ͼ AG9 Ͼ AG43) of these compounds as inhibitors of EGFR tyrosine kinase activity (IC 50 (in M) ϭ 25, 35, 560, Ͼ1250, 6500, respectively, as reported in Ref. 21) were distinct from their potencies (AG10 ϳ AG18 Ͼ Ͼ AG9, AG30, AG43) as mitochondrial uncouplers (see "Discussion").
AG9, AG30, and AG43 also were examined for their effects on oxidative phosphorylation by measuring the ADP-stimulated QO 2 of isolated mitochondria in the acute presence (3-4min exposure) of these tyrphostins. Compared with the ADPstimulated QO 2 of mitochondria exposed only to vehicle (dimethyl sulfoxide), AG9 inhibited the ADP-stimulated QO 2 by ϳ30% (Fig. 3C). In contrast, AG30 was without effect, and AG43 inhibited by ϳ10%, which was only marginally significant (p Ͻ 0.053). This indicates that although AG9 did not uncouple the mitochondria, it still had an effect on mitochondrial oxidative metabolism and that other tyrphostins of related structure also might have similar effects.
AG18, AG10, and FCCP Reduce Intracellular ATP Content-One characteristic of mitochondrial uncouplers is their ability to reduce intracellular ATP content in cells that rely on oxidative phosphorylation for much of their energy production. Since rat parotid acinar cells strongly depend on oxidative metabolism, the effects of several tyrphostins on the parotid cell ATP content were examined. Exposure of cells to 100 M AG18 and AG10 reduced the parotid ATP content by ϳ90%, but 100 M AG9 had no effect (Fig. 4). A lower concentration (10 M) of AG18 and AG10 had no significant effect on ATP. The decreases in ATP produced by 100 M AG18 and 100 M AG10 were similar to those produced by the mitochondrial uncoupler FCCP (1 M and 10 M). Mitochondrial uncouplers were previously shown to produce rapid decreases in parotid ATP content (26). The ATP alterations are consistent with the QO 2 results from freshly dispersed parotid cells and isolated liver mitochondria and indicate that these tyrphostins uncouple mitochondria. These results also indicate that the use of these compounds in biological experiments may produce complicated results, since tyrphostins that were originally characterized by their inhibition of tyrosine kinases in vitro can lower the cellular ATP content and thereby inhibit various types of kinases and other ATP-dependent proteins in a nonselective manner.
Genistein is a Weak Mitochondrial Uncoupler-Genistein, a general tyrosine kinase inhibitor that blocks tyrosine phosphorylation in parotid (3) and other cells, also was used to investigate the relationship between tyrosine phosphorylation and fluid secretion, again using the QO 2 protocol as an indicator of secretion. Basal QO 2 values and carbachol-stimulated increases in the QO 2 (⌬QO 2 ) were measured using parotid acinar cells that were pretreated with various concentrations of genistein for 20 min. Genistein (10 -100 M) did not significantly alter the basal QO 2 , but 100 M genistein blocked the majority of the carbachol-stimulated increase in QO 2 , and 30 M was partially inhibitory (Fig.  5A). At first glance, the effects of genistein in blocking the carbachol-promoted increase in O 2 consumption suggested that carbachol-stimulated fluid secretion (i.e. sodium pump-linked QO 2 ) was highly dependent on tyrosine phosphorylation. However, since the tyrphostin experiments had indicated that tyrosine kinase inhibitors could have confounding effects on cells, several additional studies were performed.
Genistein also blocked the majority of the stimulatory effect of nystatin on the QO 2 of parotid acinar cells (Fig. 5B). This indicated that genistein might be blocking cation-dependent activation of the sodium pump or blocking mitochondrial function. To test for the latter, the effects of uncouplers on the QO 2 were examined in genistein-treated parotid cells, since respiratory responses to mitochondrial uncouplers are not dependent on sodium pump activity. Genistein blocked the stimulation of the QO 2 by both FCCP and AG18 (Fig. 5B). This suggested that genistein compromises electron transport along the mitochondrial respiratory chain, particularly in response to agents that increase the level of electron transport. This suggests that the effectiveness of genistein in blocking most of the increase in the QO 2 by carbachol was not due to blocking the contributions of tyrosine phosphorylation to fluid secretion but was due to direct effects of genistein on the mitochondria of intact cells.
In contrast to its substantial inhibitory effects on cellular respiration (Fig. 5), genistein (100 M) rapidly increased the QO 2 (Fig. 3A) of isolated mitochondria. This response was much less than maximally effective concentrations of AG10, AG18, or FCCP (Fig. 3B), suggesting that acute exposure of this concentration of genistein only weakly uncoupled the isolated mitochondria (see "Discussion").
Genistein and Carbachol Reduce Intracellular ATP Content-Since genistein affected mitochondrial respiration, its effects on cellular ATP content were examined. Genistein (100 M, 20 min) reduced the ATP content in unstimulated cells by ϳ20% (Fig. 4), compared with the much larger (ϳ90%) reductions by AG18 and AG10. Since this concentration of genistein nearly completely blocked the carbachol-stimulated increase in QO 2 (Fig. 5), the ATP content also was quantified in cells exposed to carbachol to see if the ATP decrease in genisteintreated cells was much greater when oxidative phosphorylation was activated upon receptor activation. In control (vehicle-treated) cells, carbachol reduced the ATP content by 15-20%, and this reduction was maintained for cells treated with carbachol for 1-5 min (Fig. 6). This is consistent with observing a sustained stimulatory effect of carbachol on the QO 2 ( Fig. 2A) (7) via stimulation of the sodium pump, which consumes ATP and generates the ADP that provides a signal to the mitochondria to stimulate higher rates of oxidative phosphorylation. Similar (ϳ28%) reductions in the ATP content of perfused submandibular salivary glands were detected using 31 P NMR techniques when muscarinic receptor-mediated secretion was stimulated (28). In genistein-treated parotid cells, carbachol also produced a ϳ20% decline in the ATP content; however, since the basal content of ATP in genistein-treated cells was ϳ15% lower than that found in control cells, the ATP content in the combined presence of genistein and carbachol was ϳ35% lower than the control basal content (Fig. 6). These decreased amounts also were sustained for at least 5 min and were parallel to those measured in the absence of genistein.
One explanation for the nearly complete inhibition of 100 M genistein on the carbachol-stimulated increases in QO 2 (Fig. 5) could have been that genistein, or the combination of genistein plus carbachol, reduced cellular ATP to a level that blocked most of the activity of the sodium pump. However, a ϳ35% net reduction of the ATP content, which was only about twice as much as the usual amount produced by carbachol alone, would not be expected to have such a substantial effect in reducing the sodium pump-related increase in the QO 2 (29). Therefore, along with the inhibitory effect of genistein on increases in QO 2 promoted by FCCP and AG18 (Fig. 5B), these results are consistent with genistein blocking the carbachol-stimulated QO 2 by reducing mitochondrial electron transport (see "Discussion").

Effects of Tyrphostins and Other Kinase Inhibitors on Tyrosine Phosphorylation and Cell
Signaling Cascades-To determine the effectiveness of AG10, AG18, and genistein as inhibitors of tyrosine phosphorylation in parotid acinar cells, their actions were examined on the carbachol-stimulated tyrosine phosphorylation of PKC␦. In these cells, carbachol increases the tyrosine phosphorylation of PKC␦ through a Src-dependent phosphorylation, and this phosphorylation increases the activity of this protein (2). Exposure of the cells to 100 M AG10 and 100 M AG18 blocked or greatly reduced the carbachol-initiated tyrosine phosphorylation of PKC␦ (Fig. 7A). The carbachol-initiated tyrosine phosphorylation of PKC␦ also was reduced by 100 M genistein and the Src family inhibitor PP2 (10 M), as described previously (2, 3).  Lower concentrations (10 M) of AG10 and AG18 either had no effects or were much less effective than 100 M concentrations of these agents. Notably, the effects of the tyrphostins to inhibit tyrosine phosphorylation were not distinguishable from their reduction of cellular ATP content, since 1) the concentration dependences of AG10 and AG18 on blocking PKC␦ tyrosine phosphorylation and reducing ATP content were overlapping, and (2) the mitochondrial uncoupler FCCP, which also reduced ATP content to a similar degree as AG10 and AG18 (Fig. 4), also blocked PKC␦ tyrosine phosphorylation (Fig. 7B) to a similar degree as AG10 and AG18.
The activation of the mitogen-activated protein kinases ERK1 and ERK2 by carbachol also was examined in cells treated with these inhibitors (Fig. 7). ERK1 and ERK2 are activated by phosphorylation on threonine and tyrosine residues by the upstream kinase mitogen-activated protein kinase/ extracellular signal-regulated kinase kinase. AG10 and AG18 blocked ERK1/2 phosphorylation at a concentration (100 M) that reduced the ATP content, but these compounds did not block at a lower concentration (10 M) that did not reduce the ATP content. FCCP (10 M), which also reduced the ATP content, also blocked ERK1/2 activation. These data suggest that

FIG. 7. Effects of tyrphostins and other kinase inhibitors on PKC␦ tyrosine phosphorylation and ERK1/2 activation initiated by carbachol.
Parotid acinar cells were treated with various agents at the indicated concentrations for 20 min and then exposed to 10 Ϫ5 M carbachol for 2 min. Cells were lysed, a fraction of whole cell lysate (WCL) was reserved for immunoblotting (IB), and PKC␦ was immunoprecipitated (IP) from the remainder of the lysate using an anti-PKC␦ antibody. Immunoprecipitates and whole cell lysates were subjected to SDS-PAGE, transferred to nitrocellulose paper, and immunoblotted as indicated. Each immunoblot is representative of at least three separate experiments. AG10 and AG18, like FCCP, block multiple types of protein kinases by lowering the parotid cell ATP content rather than by directly blocking tyrosine phosphorylation.
PP2 did not block the carbachol-initiated ERK1/2 activation, suggesting that the tyrosine kinase Src was not involved. However, since ERK1/2 activation was reduced by genistein (100 M), this suggests that tyrosine phosphorylation by non-Src family kinases plays a role in the activation of ERK1/2 downstream of muscarinic receptor stimulation in parotid acinar cells (see "Discussion"). The lack of effect of PP2 on ERK1/2 activation also suggests that the tyrosine phosphorylation of PKC␦ does not play a major role in the activation of ERK1/2 by exposure of these cells to carbachol, although our previous studies indicated that a portion of carbachol-initiated ERK1/2 activation in parotid acinar cells is dependent on PKC (30).
Activation of AMPK in Parotid Cells-Since secretory stimuli, tyrphostins, and other agents lowered the ATP content of parotid acinar cells, albeit via different mechanisms, their effects on the metabolite-sensing AMPK were examined. AMPK activity was monitored using a phosphospecific antibody to Thr 172 , a site on the ␣ subunit of AMPK that, when phosphorylated, greatly increases the activity of the enzyme (31). Within 1 min of their exposure to parotid acinar cells, carbachol (10 Ϫ5 M), BzATP (10 M), and FCCP (10 M) all produced increases in AMPK phosphorylation/activation (Fig. 8A). AG10 (100 M) and AG18 (100 M) activated AMPK on the same time scale as FCCP and the other stimuli, as did the mitochondrial inhibitor rotenone (not shown). Cells exposed to genistein (100 M) for 2 min did not have a noticeable increase in AMPK activation (not shown). BzATP is a ligand for the P2X 7 receptor, the activation of which increases parotid ion fluxes, QO 2 , and other responses in a similar manner to many of the changes evoked by carbachol (8,9). Although the P2X 7 and muscarinic receptors may also activate distinct ionic currents (32,33), the activation of both kinds of receptors produces rapid increases in the sodium/potassium ratio of parotid cells.
The phosphorylation of AMPK in response to carbachol and BzATP was blocked by the sodium pump inhibitor ouabain, but ouabain was ineffective in blocking the phosphorylation produced by AG10, AG18, and FCCP (Fig. 8, B and C). These results suggest that the activation of AMPK upon stimulation of the muscarinic and P2X 7 receptors was secondary to sodium pump-mediated changes in energy metabolites and that the activation of AMPK produced by mitochondrial uncouplers was independent of the sodium pump. Notably, the increase in AMPK phosphorylation by carbachol had kinetics that were similar to its effects in decreasing the cellular ATP content (Fig. 6). The speedy effects of AG10, AG18, and FCCP on AMPK are consistent with their rapid uncoupling of the mitochondria in intact cells (Fig. 2), a process that initiates rapid decreases in ATP and other metabolic alterations. In fact, FCCP (10 M) produced a ϳ90% reduction in the parotid ATP content within 2 min (26). Uncouplers not only uncouple ATP production from O 2 consumption but also reverse the F 0 F 1 -ATPase (ATP synthase), thereby promoting the consumption of ATP rather than the production of ATP. The ineffectiveness of genistein in activating AMPK within a 2-min exposure suggests that it did not produce substantial alterations in energy metabolites within this time period.
AG10 and AG18 Uncouple Mitochondria and Activate AMPK in Multiple Cell Types-To determine whether AG10 and AG18 had similar actions on other cells, their effects were examined on PC12, HeLa, and HEK293 cells. AG10, AG18, and FCCP produced large and rapid increases in the QO 2 of each of these three cell types. In PC12 cells, the basal QO 2 (14.1 Ϯ 2.4 (n ϭ 3) nmol of O 2 /mg/min) was increased by 24.5 Ϯ 4.6 (n ϭ 3), 16.4 Ϯ 2.4 (n ϭ 3), and 24.9 Ϯ 3.1 (n ϭ 3) nmol of O 2 /mg/ml by AG10 (100 M), AG18 (100 M), and FCCP (1 M), respectively. The concentration dependences (at 10, 30, and 100 M) of the tyrphostins on increasing the QO 2 of the cell lines (not shown) appeared similar to those observed using parotid cells. AG10, AG18, and FCCP also increased AMPK activity in PC12 (Fig. 9) and HEK293 cells. These results indicate that these tyrphostins act as mitochondrial uncouplers, alter energy metabolism, and produce secondary changes in signaling proteins in a variety of cell types in addition to parotid acinar cells.
Of interest, although HeLa cells exposed to AG10, AG18, and FCCP responded with an increase in QO 2 , these agents did not produce an increase in AMPK phosphorylation (not shown). Presumably this was due to the lack of LKB1 expression in FIG. 8. Activation of AMP kinase in parotid acinar cells. Parotid acinar cells were treated as indicated and lysed, and a fraction of the lysate was subjected to SDS-PAGE. Proteins were transferred to nitrocellulose and sequentially immunoblotted with anti-phospho-AMPK (P-AMPK) antibody and anti-AMPK antibody (see "Materials and Methods"). Each immunoblot is representative of at least three separate experiments. A, parotid acinar cells were equilibrated at 37°C for 20 min and then exposed to BzATP (10 M), carbachol (Carb; 10 M), and FCCP (10 M) for 1-5 min. All three agents stimulated AMPK activity within 1 min. B, parotid acinar cells were equilibrated at 37°C for 20 min and then exposed to ouabain (1 mM) for 2 min followed by BzATP (10 M), carbachol (10 M), and FCCP (10 M) for 2 min. Ouabain blocked the activation of AMPK by carbachol and BzATP but not by FCCP. C, parotid acinar cells were equilibrated at 37°C for 20 min and then exposed to ouabain (1 mM) for 2 min followed by either AG10 (100 M) or AG18 (100 M) for 2 min. Ouabain did not block the activation of AMPK by AG10 or AG18.
FIG. 9. Activation of AMP kinase in PC12 cells. PC12 cells were equilibrated at 37°C for ϳ60 min in solution A and then exposed to AG10 (100 M), AG18 (100 M), and FCCP (1 M) for 3 min. Cells were lysed, and a fraction of the lysate was subjected to SDS-PAGE. Proteins were transferred to nitrocellulose and sequentially immunoblotted (IB) with anti-phospho-AMPK (P-AMPK) antibody and anti-AMPK antibody (see "Materials and Methods"). The immunoblot is representative of at least three separate experiments.
HeLa cells. In conjunction with other proteins, LKB1 acts as an effective AMPK kinase, but it is not expressed in HeLa cells (34). Consistent with this, we found that LKB1 protein was expressed in parotid, PC12, and HEK293 cells and that this protein was lacking in our HeLa cell line (not shown). DISCUSSION These studies demonstrate that AG10 and AG18 affect cells in ways that are very different from those for which they initially were designed and characterized as inhibitors of protein-tyrosine kinases. These tyrphostins uncoupled mitochondrial ATP production from O 2 consumption and promoted the rapid depletion of parotid cell ATP in the absence of receptor agonist stimulation. Thus, these inhibitors were unsuitable for the initial purpose of their utilization in this study, which was to screen alterations in carbachol-stimulated increases in O 2 consumption as a means of determining whether tyrosine phosphorylation contributed to fluid secretion. Like FCCP, the effectiveness of AG10 and AG18 in blocking receptor-mediated phosphorylation-dependent cell signaling events appeared to be attributable to reducing the ATP content rather than to directly inhibiting tyrosine kinases. In contrast, the effectiveness of genistein in blocking carbachol-mediated increases in PKC␦ tyrosine phosphorylation and ERK1/2 activation was probably due to its direct inhibition of tyrosine kinases, since genistein produced a relatively small decrease (15-20%) in the cellular ATP content in unstimulated cells (Figs. 4 and 6) and since the carbachol-initiated ATP decreases in the presence and absence of genistein were similar. This suggests that tyrosine kinases contribute to the PLC␤-linked muscarinic receptor-initiated activation of ERK1/2. However, this conclusion assumes that the activities of the involved kinases were not substantially compromised by the initial reduction of the ATP content in cells exposed to genistein prior to exposure to carbachol.
Carbachol produced a rapid 15-20% decrease in cellular ATP content, doubtlessly due to the large increase in sodium pump activity that was promoted by a rapid increase in the cellular sodium/potassium ratio (7). In agreement with this, previously we measured a ϳ20% decrease in the cellular ATP content of freshly isolated renal cells when the sodium pump was activated (35). Despite the carbachol-initiated decrease in ATP content under normal conditions, the cellular ATP concentration may not have been substantially altered, since there is a rapid 20 -25% decrease in cell volume in parotid cells exposed to carbachol (7).
Stimulation of parotid muscarinic and P2X 7 receptors activated AMPK subsequent to energy metabolite changes. Receptor-mediated changes in sodium pump activity and corresponding changes in oxidative phosphorylation attest to the functional coupling between secretion and parotid cell metabolism; in a different way, the carbachol-initiated changes in ATP metabolism (due to sodium pump activation) and the ouabain-sensitive activation of AMPK also demonstrate the close relationship between parotid metabolic processes and ion movement. In lung epithelial cells, AMPK may coordinate ion movement and metabolic state, since it co-localizes with the cystic fibrosis transmembrane conductance regulator Cl Ϫ channel, phosphorylates cystic fibrosis transmembrane conductance regulator, and regulates its activity (36,37).
Genistein and the two tyrphostin mitochondrial uncouplers had distinct but different effects on parotid cell metabolism. Unlike AG10 and AG18, genistein did not alter the basal respiration (Fig. 5A). This suggests that genistein did not cause significant degree of mitochondrial uncoupling of intact parotid acinar cells under these conditions, although it appeared to be a weak uncoupler of isolated mitochondria (Fig. 3). Genistein may be both an inhibitor of the mitochondrial respiratory chain and an uncoupler (38), and this may explain its contrasting effects on the QO 2 of dissociated salivary cells and isolated mitochondria. Genistein also has been reported to affect several mitochondrial proteins that play a role in energy production. Genistein inhibited the F 0 F 1 -ATPase (ATP synthase) (39), the enzyme that synthesizes ATP during oxidative phosphorylation. Genistein also opened the mitochondrial permeability transition pore, which is made up of the adenine nucleotide transporter and other components, and this led to the induction of apoptosis in intact cells (40). The mitochondrial permeability transition pore plays a role in cell death and apoptosis, and its opening can uncouple the mitochondria. In fact, inhibitory effects of genistein on pancreatic cancer cell growth were postulated to occur by opening the mitochondrial permeability transition pore and promoting apoptosis (41).
The results presented here indicate that the relative abilities of tyrphostins AG9, AG10, AG18, AG30, and AG43 to act as mitochondrial uncouplers were not related to their in vitro potencies in blocking the EGFR kinase activity, which was one of the original screening assays performed to quantify their biochemical effects as kinase inhibitors. For tyrphostin compounds that all have a similar benzylidenemalononitrile structural core, kinase inhibition is much more potent for di-CN compounds that contain two -OH groups on the ring (AG18) compared with one -OH group (AG10). Thus, AG10 and AG18 had very different IC 50 values (560 and 35 M, respectively) for blocking the EGFR kinase activity (21). In contrast, the potencies of AG10 and AG18 for uncoupling isolated mitochondria were similar (EC 50 ϳ20 M) (Fig. 3B), and both tyrphostins had similar potencies in uncoupling mitochondria in intact salivary epithelial cells (EC 50 ϳ30 M) (Fig. 2B). Therefore, there was no obvious correlation between the effects of AG10 and AG18 on EGFR kinase inhibition and their uncoupling abilities. AG9, AG30, and AG43 had IC 50 values of Ͼ1250, 25, and 6500 M, respectively, in reducing EGFR kinase activity (21). None of these compounds displayed any uncoupling activity at 100 M concentrations (Fig. 3B).
The uncoupling effects of compounds such as FCCP and dinitrophenol are related to their ability to act as weak acids, which act as protonophores to shuttle protons across the mitochondrial membrane and collapse the proton gradient that drives oxidative phosphorylation. Effective uncouplers that are weak acids require an ionizable acid group and a strong electron-withdrawing group (42). Thus, based on the results of the present study, one would predict that AG10 and AG18 are weak acids that can act as protonophores and that AG9, AG30, and AG43 lack this ability. The lack of a dissociable proton might be expected for AG9, which has a methoxide group (-OCH 3 ) in place of the -OH group that is found in AG10. AG30 has a -COOH group substituted in place of a -CN group on AG18. Saturation of the double bond in AG10 converts it into AG43, and this change made AG43 a much less ineffective EGFR kinase inhibitor compared with AG10 (21). The saturation of this bond also was responsible for making AG43 ineffective as a mitochondrial uncoupler compared with AG10 (Fig.  3), probably by reducing the interaction between the electron withdrawing power of the two -CN groups and the proton at the 4-OH position of the ring.
The uncoupler effects of AG10, AG18, and FCCP on parotid cell energy metabolites led to the secondary activation of the fuel sensor AMPK. AG10 and AG18 (and FCCP) also promoted large increases in the QO 2 of PC12, HEK293, and HeLa cells, and AMPK was activated in PC12 and HEK293 cells. These findings indicate that these tyrphostins uncouple mitochondria and alter cellular energy homeostasis in a variety of cell types.
Uncouplers may not produce large changes in the ATP contents of cells that rely primarily on glycolysis, unlike the large ATP decreases that were produced in parotid cells (Fig. 4). The ATP content was only reduced ϳ15% in Chinese hamster ovary cells treated for 16 h with 200 M AG18 (18). Another tyrphostin, AG17 (3,5(di-tert-butyl-4-hydroxybenzylidene) malononitrile), which has the basic skeletal structure of AG10, but with two additional groups (C(CH 3 ) 3 ) on the ring, decreased the ATP content to ϳ10% of control values and altered the mitochondrial integrity in HL-60(TB) cells, a human leukemia cell line (43). Similar to results reported here for AG10 and AG18, AG17 was effective at blocking tyrosine phosphorylation in HL-60(TB) cells only at concentrations that altered the ATP content. However, AG17 did not appear to reduce the ATP content in a human non-Hodgkin's lymphoma cell line (44). Tyrphostins AG213 (␣-cyano-(3,4-dihydroxy) thiocynnamide), which has a structure similar to AG18 (two -OH groups but with a different group in place of one CN group), and AG126 (␣-cyano-(3-hydroxy-4-nitro)cinnamonitrile; A10) also uncouple mitochondria (38,45).
The results of the present study indicate that some of the tyrphostins that have been used as general tyrosine kinase inhibitors can block multiple types of protein kinases indirectly by altering the ATP content in cells. This may have been a component of their reported biological effects in some studies. In results analogous to these findings, rottlerin, a putative specific inhibitor of PKC␦, also uncoupled mitochondria and greatly lowered the parotid cell ATP content and thereby blocked the tyrosine phosphorylation of PKC␦ by Src (26). Since apoptosis and other biological processes involve alterations in mitochondrial proteins, it is important to recognize that tyrphostins can have effects on mitochondria, even if they are being used on cells that rely on glycolysis. In addition, as demonstrated in the present studies, tyrphostins with uncoupling effects can secondarily activate AMPK in multiple cell types, including those that have a primarily glycolytic metabolism.
These results demonstrate the importance of performing cellular metabolic analyses along with in vitro kinase assays when compounds are developed for use as biological kinase inhibitors. Rational drug design may succeed at devising molecules that are effective kinase inhibitors, but such compounds may have other effects that confound the identification of their mechanism of action on intact cells. In addition, these studies demonstrate that normal physiological events such as the stimulation of epithelial cells by ligands that modify fluid secretion can exert metabolic changes that activate kinases (e.g. AMPK) secondarily. This extends our understanding of the variety of cell signaling mechanisms that accompany receptor-mediated changes in fluid secretion.