Oncogene (1999) 18, 7794 ± 7802 ã 1999 Stockton Press All rights reserved 0950 ± 9232/99 $15.00 http://www.stockton-press.co.uk/onc pathways regulated by arsenate and arsenite

Amy C Porter1, Gary R Fanger2,3 and Richard R Vaillancourt*,1

1Department of Pharmacology and Toxicology, College of Pharmacy, The University of Arizona, Tucson, Arizona, AZ 85721-0207, USA; 2Program in Molecular Signal Transduction, Division of Basic Sciences, National Jewish Medical and Research Center, Denver, Colorado, CO 80206, USA

Arsenate and arsenite activate c-Jun N-terminal kinase c-Jun, which forms a heterodimer with c-Fos to form (JNK), however, the mechanism by which this occurs is the AP-1 transacting factor. Activation of JNK is not known. By expressing inhibitory mutant small GTP- commonly associated with inhibition of cell growth binding proteins, p21-activated kinase (PAK) and and/or (Kyriakis and Avruch, 1996). mitogen-activated protein kinase/extracellular signal- However, under certain circumstances, sustained JNK regulated kinase kinase kinases (MEKKs), we have activation, as opposed to transient activation, is identi®ed speci®c proteins that are involved in arsenate- required for an apoptotic response (Guo et al., 1998). and arsenite-mediated activation of JNK. We observe a Toxic doses of inorganic arsenicals, such as arsenite distinct di€erence between arsenate and arsenite signal- and arsenate, produce pleiotrophic e€ects. Using ing, which demonstrates that arsenate and arsenite are cultured macrophages as an in vitro system to study capable of activating unique proteins. Both arsenate and arsenic toxicity, it was observed that 80% of the dead arsenite activation of JNK requires Rac and Rho. cells were necrotic while 20% of the cells were

Neither arsenate nor arsenite signaling was inhibited by apoptotic at the LD50 dose of 5 mM arsenite and a dominant-negative mutant of Cdc42 or Ras. Arsenite 500 mM arsenate (Sakurai et al., 1998). In addition to stimulation of JNK requires PAK, whereas arsenate- cell death, chronic arsenic exposure has been associated mediated activation of JNK was una€ected by inhibitory with malignant transformation and DNA hypomethy- mutant PAK. Of the four MEKKs tested, only MEKK3 lation of epithelial cells (Zhao et al., 1997). Further- and MEKK4 are involved in arsenate-mediated activa- more, epidemiological studies demonstrate that arsenic tion of JNK. In contrast, arsenite-mediated JNK is a human carcinogen (Smith et al., 1998). However, activation requires MEKK2, MEKK3 and MEKK4. the poor mutagenicity of inorganic arsenicals (Jacob- These results better de®ne the mechanisms by which son-Kram and Montalbano, 1985; Lee et al., 1985; arsenate and arsenite activate JNK and demonstrate Rossman et al., 1980) suggests that the production of di€erences in the regulation of signal transduction diverse physiological e€ects caused by arsenic are likely pathways by these inorganic arsenic species. due to the activation of various signaling pathways and not a genotoxic e€ect. Consistent with this line of Keywords: arsenic; Rac; Rho; PAK; MEKK3; MEKK4 reasoning is the observation that arsenic activates multiple MAPK pathways (Liu et al., 1996). Conse- quently, the spectrum of physiological responses caused by arsenic range from carcinogenesis to cell death. Introduction Both arsenate and arsenite activate JNK. However, the mechanism by which these arsenic species activate c-Jun N-terminal kinase (JNK) is a member of the this pathway has not been characterized. One report stress-activated protein kinase family and is activated suggests that arsenite activates JNK by inactivation of by cellular stress, such as osmotic shock and a JNK phosphatase (Cavigelli et al., 1996). Since irradiation [reviewed in (Treisman, 1996)]. JNK arsenite has been shown to bind and inactivate proteins activation requires the small GTP-binding proteins containing thiol functional groups, arsenite may bind Ras, Rac, Cdc42 and Rho (Coso et al., 1995; Derijard to thiol groups in the active site of a JNK phosphatase. et al., 1994; Minden et al., 1995; Teramoto et al., Because phosphorylation of JNK is critical for its 1996). These small GTP-binding proteins associate with activation, this mechanism does not explain how JNK and activate a variety of serine/threonine kinases that is phosphorylated and thus activated in the ®rst place. are important in JNK activation including p21- In addition, since arsenate does not bind to thiol activated kinase (PAK), as well as mitogen-activated groups, this mechanism does not explain how arsenate protein kinase/extracellular-signal-regulated kinase ki- activates JNK. Thus, it is possible that arsenate and nase kinases [MEKKs; (Fanger et al., 1997a)]. The arsenite are capable of regulating speci®c signal MEKKs phosphorylate and activate JNK kinase transduction pathways associated with JNK activation. (JNKK), which phosphorylates and activates JNK. It has been well established that arsenite is an Upon activation, JNK alters speci®c gene transcription activator of the stress-activated protein kinase path- events via phosphorylation of the transcription factor, ways (Cavigelli et al., 1996; Liu et al., 1996). We have determined that arsenate [As(V)], which is the oxidized precursor of arsenite [As(III)], can also activate JNK. Based on this observation, we wanted to determine *Correspondence: RR Vaillancourt whether arsenite and arsenate utilize the same signaling 3 Current address: Corixa Corporation, 1124 Columbia Street, Seattle, Washington, WA 98104, USA proteins to activate JNK. We set out to map these Received 3 December 1998; revised 2 September 1999; accepted signaling pathways by using a series of inhibitory 7 September 1999 mutant proteins that are important for JNK activation Arsenic regulates Rac, Rho, PAK, MEKK3 and MEKK4 AC Porter et al 7795 by other stimuli. We have transfected inhibitory for the anion transport protein. To overcome this mutant small GTP-binding proteins, PAK, and competition, HEK 293 cells were incubated in MEKK1-4 into human embryonic kidney (HEK) 293 phosphate-free DMEM, which contained serum, cells to determine which proteins were capable of immediately prior to treatment with increasing con- blocking activation of JNK by either arsenate or centrations of arsenate. arsenite. The data from our experiments show that A time course of JNK activity demonstrated that arsenate and arsenite activate JNK through di€erent arsenate and arsenite activate JNK at di€erent rates signal transduction pathways, which require speci®c (Figure 1). The cells were treated with 300 mM arsenate small GTP-binding proteins and serine/threonine for various times and JNK activity was assayed by kinases. precipitation with GST-c-Jun (1 ± 79). Cell lysates were The goal of this study was to determine whether prepared and incubated with GST-c-Jun, which is arsenate and arsenite activate JNK through the same bound to Sepharose beads. GST-c-Jun associates with upstream proteins. One feature that di€erentiates endogenous JNK and the protein complex is pre- arsenate from arsenite is the relative permeability of cipitated by centrifugation, washed with bu€er, and each inorganic arsenic species across the plasma then incubated with [g-32P]ATP. JNK phosphorylates membrane. Arsenate must be transported into the cell its substrate, c-Jun, if it is activated by upstream by the organic anion transport protein, whereas kinases. Phosphorylated c-Jun was resolved by arsenite freely crosses the plasma membrane (Kenney electrophoresis and detected by autoradiography. and Kaplan, 1988). In the presence of phosphate, Maximal activation of JNK by arsenate was 40-fold arsenate is poorly transported into the cell since over basal. Maximal levels of JNK activity were phosphate competes with arsenate for binding to the reached between 1 and 2 h. Arsenite activation of anion transport protein. Thus, the mechanism by JNK followed a consistently di€erent time course. which arsenic signal transduction is initiated by each Arsenite treatment resulted in nearly a 30-fold arsenic species is potentially di€erent. stimulation of JNK. Maximal stimulation with arsenite was measured between 30 and 60 min in contrast to the longer time required for maximal stimulation with arsenate. At 3 and 4 h, both arsenate Results and arsenite stimulated JNK activity was lower than the maximal stimulation, suggesting that activation of Arsenate is an activator of JNK JNK is a reversible event. For both arsenic species, We used HEK 293 cells and transient transfection of there were no detectable cytotoxic e€ects during the cDNAs that encode candidate proteins as an approach respective time courses (data not shown). to identify proteins that function upstream of JNK. We HEK 293 cells were treated with various concentra- measured JNK activity in HEK 293 cells to establish tions of arsenate and arsenite. Both inorganic arsenic that the signaling proteins were expressed in these cells. species activated JNK in a dose-dependent manner Since arsenate is transported into the cell by the anion (Figure 2). At doses below 30 mM of arsenate, there transport protein (Kenney and Kaplan, 1988), a was no stimulation of JNK activity. However, a competition exists between arsenate and phosphate 100 mM dose of arsenate produced a greater than

Figure 1 Time course of JNK activation by arsenate and arsenite. HEK 293 cells were treated with 300 mM arsenate (a) or arsenite (b) for the times indicated. GST-c-Jun that was bound to Sepharose conjugated with glutathione was incubated with 40 mg of cell lysate. Endogenous JNK was precipitated by centrifugation and the protein complex was incubated with [g-32P]ATP. Phosphorylated GST-c-Jun was resolved by electrophoresis and the autoradiogram is shown (in lower panels). Activity is expressed graphically as fold stimulation over basal, quanti®ed using a Packard Instant Imager Electronic Autoradiography SystemTM (Packard Instrument Co, Meriden, CT, USA). This is a representative experiment of three Arsenic regulates Rac, Rho, PAK, MEKK3 and MEKK4 AC Porter et al 7796 sixfold stimulation of JNK activity. When the dose of binding proteins so that we could measure JNK-1 arsenate was 300 mM, JNK activity increased to almost activity from only the transfected cells. Following 20-fold over basal levels of JNK activity. In parallel stimulation with a dose of arsenite or arsenate that experiments, arsenite increased JNK activity to a provides maximal JNK activity, HA-JNK-1 was similar level at a 100 mM dose of arsenite as that immunoprecipitated from cells with the 12CA5 observed for arsenate. At a 300 mM dose, arsenite monoclonal antibody and an immune complex assay typically produced a greater activation of JNK than was performed where immunoprecipitated HA-JNK-1 arsenate. Arsenite activated JNK 36-fold over basal was incubated with bacterially expressed GST-c-Jun levels while arsenate activated JNK 20-fold in this and [g-32P]ATP. Phosphorylated GST-c-Jun was experiment. The level of JNK activity that we observed resolved by SDS ± PAGE and detected by autoradio- was maximal for both compounds at 300 mM. We did graphy. not observe any increase in JNK activity at concentra- Expression of RhoN19 inhibited greater than 90% of tions as high as 3 mM (data not shown). Thus, the arsenite-stimulated JNK activity (Figure 3a,b). In arsenate, like arsenite, activates JNK in HEK 293 contrast, expression of RacN17 provided a modest cells, although arsenite exposure induced a consistently statistically signi®cant inhibition of arsenite-stimulated higher maximal response. JNK activity, whereas expression of RasN17 and Cdc42N17 had no e€ect. The data that are presented in Figure 3a represent the average of three separate Dominant-negative Rac and Rho inhibit JNK activation experiments and are calculated relative to the arsenite- by arsenite or arsenate treated cells transfected with only the vector, pCMV5. We transfected the cDNA encoding small GTP-binding proteins that are known to function upstream of JNK to better de®ne the mechanisms by which arsenite and arsenate activate JNK. We used the inhibitory forms of Cdc42 (Cdc42N17), Rac (RacN17), Ras (RasN17), and Rho (RhoN19) to determine which small GTP-binding proteins inhibit the activation of JNK by arsenite or arsenate. Since the transfection eciency was less than 100%, we co-transfected the cDNA encoding HA- JNK-1 with the cDNAs encoding the small GTP-

Figure 3 Inhibitory mutant small GTP-binding proteins inhibit arsenite-stimulated JNK activity. HEK 293 cells were transfected with HA-tagged JNK-1 and mutant small GTP-binding proteins as described in the Materials and methods. Two days after transfection, cells were treated (+) with 300 mM arsenite or with vehicle (7) for 30 min. After treatment the cells were lysed and JNK-1 was immunoprecipitated with the 12CA5 monoclonal antibody. Immunoprecipitated JNK activity was assayed by using Figure 2 (a) Dose response curve of JNK activation by arsenate GST-c-Jun as a substrate as described above. The graph shows and arsenite. HEK 293 cells were treated with arsenate for 2 h or arsenite-stimulated JNK1 activity from three separate experiments arsenite for 30 min. Endogenous JNK was precipitated with GST- (a). Basal activity was de®ned as JNK1 activity from untreated c-Jun that was bound to Sepharose and a solid-phase kinase assay cells transfected with pCMV5. Activity from treated cells was performed in the presence of [g-32P]ATP. Phosphorylated transfected with pCMV5 was arbitrarily set at 100%. The data GST-c-Jun was separated by SDS ± PAGE using a 10% gel (b). represent the mean+s.d. from three separate experiments. Activity is expressed graphically as fold stimulation over basal. *P50.05 versus control. In b and c, a representative experiment Phosphorylated GST-c-Jun was identi®ed by autoradiography (b) of at least three separate experiments shows phosphorylated GST- and quantitated as described above. This is a representative c-Jun (b) and a Western blot of immunoprecipitated JNK1 experiment of at least three separate experiments protein (c) Arsenic regulates Rac, Rho, PAK, MEKK3 and MEKK4 AC Porter et al 7797 In this ®gure, a representative experiment is shown in b function in arsenate signaling to JNK (Figure 4b,c). and c. Similar levels of HA-JNK-1 were immunopre- Moreover, these data demonstrate that arsenate and cipitated as determined by immunoblotting HA-JNK-1 arsenite utilize similar GTP binding proteins to activate with a JNK speci®c antibody (Figure 3c). These data JNK. demonstrate that endogenous Rho and Rac function upstream of JNK in the pathway for arsenite- Dominant-negative PAK inhibits arsenite-dependent JNK stimulated JNK activity, while Ras and Cdc42 do not activity appear to function in arsenite-dependent signal transduction. Heterologous expression of PAK results in the We next examined whether arsenate activates JNK activation of JNK, suggesting that PAK functions via the same small GTP-binding proteins as were upstream of JNK (Bagrodia et al., 1995). We expressed identi®ed for arsenite. The cells were treated for 2 h kinase-inactive inhibitory bPAK (KM-bPAK) to since this time point yielded maximal activation of determine if the activity of this kinase was required JNK (see Figure 1a). HA-JNK-1 was immunoprecipi- for arsenite- and arsenate-mediated JNK activation. tated from cells that expressed inhibitory mutant forms HEK 293 cells were treated with arsenite or arsenate, of Cdc42, Rac, Ras, and Rho. We observed in multiple as described above, and HA-JNK-1 was immunopre- experiments that JNK activity was inhibited in cells cipitated. KM-bPAK inhibited *60% of the arsenite- expressing RacN17 and RhoN19, but not RasN17 and stimulated JNK activity (Figure 5, compare lanes c and Cdc42N17 (Figure 4a). The data from a representative f). However, even though KM-bPAK was e€ective at experiment demonstrate that endogenous Rac and Rho inhibiting arsenite-stimulated JNK activation there was

Figure 4 Inhibitory mutant small GTP-binding proteins inhibit arsenate-stimulated JNK1 activity. HEK 293 cells were trans- fected as described above and 2 days after transfection, cells were Figure 5 Dominant-negative PAK inhibits arsenite-stimulated treated (+) with 300 mM arsenate or with vehicle (7) for 2 h. JNK1 activity. HEK 293 cells were transfected with HA-tagged After treatment the cells were lysed and JNK1 was immunopre- JNK1 and dominant-negative PAK. Two days after transfection, cipitated with the 12CA5 monoclonal antibody. Immunoprecipi- cells were treated with 300 mM arsenate or 300 mM arsenite. After tated JNK1 activity was assayed by using GST-c-Jun as a treatment the cells were lysed and HA-JNK-1 was immunopre- substrate as described above. The graph shows arsenate- cipitated with the 12CA5 monoclonal antibody. GST-c-Jun was stimulated JNK1 activity from three separate experiments (a). used as a JNK substrate. Basal (hatched bars), arsenite-stimulated Basal activity was de®ned as JNK1 activity from untreated cells (black bars), and arsenate-stimulated (open bars) JNK activity transfected with pCMV5. Activity from treated cells transfected was determined either in the absence of KM-PAK (lanes a ± c) or with pCMV5 was arbitrarily set at 100%. The data represent the in the presence of KM-PAK (lanes d ± f). Activity is expressed as mean+s.d. from three separate experiments. *P50.05 versus relative JNK1 activity. The autoradiogram is shown below the control. In b and c, a representative experiment of at least three graph. The Western blot, showing immunoprecipitated JNK1, is separate experiments shows phosphorylated GST-c-Jun (b) and a shown at the bottom of the ®gure. This is a representative Western blot of immunoprecipitated JNK1 protein (c) experiment of at least three separate experiments Arsenic regulates Rac, Rho, PAK, MEKK3 and MEKK4 AC Porter et al 7798 no e€ect on arsenate-mediated JNK activity (Figure 5, people who drink and bathe in arsenic contaminated compare lanes b and e). From these data we conclude (Smith et al., 1998; Yeh et al., 1968), Yet, the that arsenite, but not arsenate, requires PAK to molecular mechanism by which inorganic arsenic activate JNK, which provides evidence that arsenate species cause cancer is unknown. and arsenite activate JNK through di€erent signaling It is clear that arsenate must enter the cell to have its mechanisms. e€ect on JNK, likely via an anion transport protein (Kenney and Kaplan, 1988). In fact, our data indicate that the e€ects of arsenate are limited when the cells Dominant-negative MEKK2, MEKK3 and MEKK4 are dosed in phosphate containing media (data not inhibit arsenite-dependent JNK activity: Dominant- shown). This suggests that there is a competition for negative, MEKK3 and MEKK4 inhibit the transporter between phosphate and arsenate. Upon arsenate-dependent JNK activity entry of arsenate and arsenite into the cellular milieu, Since the MEK kinase (MEKK) family members the proteins that transduce the signal to the MAP regulate JNK activity, we examined the role of all kinases as well as transcription factors have not been four MEKKs in arsenite and arsenate signaling to described. Therefore, we determined if arsenate and JNK. Given the di€erences that we found between arsenite were capable of regulating speci®c signal these two arsenic species at the level of PAK, we transduction pathways that are typically associated predicted that the MEKK proteins would di€erentially with JNK activation. a€ect arsenate and arsenite signaling to JNK. HEK Using inhibitory small GTP-binding protein con- 293 cells were transfected with kinase-inactive inhibi- structs including Cdc42, Rac, Ras and Rho, we show tory mutants of the di€erent MEKK family members (KM-MEKK1-4) and HA-JNK-1. Following treatment with arsenite, HA-JNK-1 was immunoprecipitated and phosphorylation of GST-c-Jun was determined. We consistently found that expression of KM-MEKK2, KM-MEKK3 and KM-MEKK4 signi®cantly inhibited arsenite-stimulated Jun kinase activity by greater than 60% (Figure 6b). As in Figure 3, the data were calculated relative to the arsenite-stimulated Jun kinase activity in cells transfected with pCMV5. In similar experiments using arsenate, we found that expression of KM-MEKK3 and KM-MEKK4 signi®cantly inhibited arsenate-stimulated JNK activity by greater than 80% (Figure 7b). In contrast to our ®ndings with arsenite, KM-MEKK2 did not inhibit arsenate- stimulated JNK activity, which suggests that this kinase is not involved in arsenate-mediates JNK activation. Neither arsenite nor arsenate-stimulated JNK activity was signi®cantly inhibited by a dominant negative MEKK1. Thus, arsenite appears to require the activity of MEKK2, MEKK3 and MEKK4 to activate JNK while arsenate appears to require the activity of only MEKK3 and MEKK4.

Discussion

Arsenite has been shown to activate the ERK, JNK, and p38 MAP kinases (Adler et al., 1995; Cavigelli et al., 1996; Liu et al., 1996; Meier et al., 1996; Rouse et al., 1994; Trigon and Morange, 1995). In addition, arsenite induces the expression of heat shock proteins including heme oxygenase (Elbirt et al., 1998). For historical reasons, arsenite has been the inorganic Figure 6 Inhibition of arsenite-stimulated JNK1 activity by arsenic species that has received the most study dominant-negative MEKKs. HEK 293 cells were transfected with HA-tagged JNK1 and dominant negative MEKK1-4. Two days because it is considered one of the most toxic forms after transfection, cells were treated (+) with 300 mM arsenite or of arsenic. However, human exposure to arsenate with vehicle (7) for 30 min. After treatment the cells were lysed frequently occurs due to contaminated water supplies and JNK1 was immunoprecipitated with the 12CA5 monoclonal (Stoner et al., 1977). We show for the ®rst time that antibody. Immunoprecipitated JNK activity was assayed by using GST-c-Jun as a substrate. The graph shows arsenite-stimulated arsenate, like arsenite is capable of activating JNK. JNK1 activity from three separate experiments (a). Basal activity Consequently, arsenate is another inorganic chemical was de®ned as JNK1 activity from untreated cells transfected with species of arsenic that deserves attention with regard to pCMV5. Activity from treated cells transfected with pCMV5 was the signal transduction pathways that are regulated by arbitrarily set at 100%. The data represent the mean+s.d. from arsenic. This is particularly important since there is three separate experiments. *P50.05 versus control. In b and c,a representative experiment of at least three separate experiments epidemiological data to link inorganic arsenic, such as shows phosphorylated GST-c-Jun (b) and a Western blot of arsenite and arsenate, to multiple forms of cancer in immunoprecipitated JNK1 protein (c) Arsenic regulates Rac, Rho, PAK, MEKK3 and MEKK4 AC Porter et al 7799 expression of PIP5K mRNA in Arabidopsis thaliana (Mikami et al., 1998). It is tempting to speculate that oxidative stress caused by inorganic arsenic species may also induce mammalian PIP5K through a Rho or Rac-dependent pathway. The exact role for Rac in arsenite and arsenate signal transduction to JNK is unclear. It has been shown that Rac associates with cytosolic proteins to form the multicomponent respiratory burst of NADPH oxidase in neutrophils (Freeman et al., 1996). We have found that HEK 293 cells chemically reduce arsenate to arsenite (data not shown), which may result in oxidative stress in the cell as glutathione stores are depleted. It is not clear if these cells are capable of producing the methylated products of arsenite. Thus, the involvement of small GTP-binding proteins such as Rac in arsenite- and arsenate-mediated signal transduc- tion may be due to oxidative stress mechanisms (Figure 4). Consistent with this hypothesis is the observation that the activation of all three MAP kinases was prevented by the free radical scavenger, N-acetyl-L- cysteine (Liu et al., 1996). It is possible that the Rac and Rho small GTP-binding proteins that are required for arsenite or arsenate signaling to JNK also play a role in the metabolism of arsenate to arsenite. Rac may also play a role in arsenic-induced carcinogenesis. Activated Rac (RacV12) and its e€ector, Tiam1, will activate JNK (Michiels et al., 1997) and both proteins are essential for malignant transformation Figure 7 Inhibition of arsenate-stimulated JNK1 activity by of NIH3T3 and lymphoma cells (Habets et al., 1994; dominant-negative MEKKs. HEK 293 cells were transfected with Michiels et al., 1995; Qiu et al., 1995; van Leeuwen et HA-tagged JNK1 and dominant negative MEKK1-4. Two days after transfection, cells were treated (+) with 300 mM arsenate or al., 1995). In addition, the e€ector site of Rac has been with vehicle (7) for 2 h. After treatment the cells were lysed and implicated in the control of mitogenesis through JNK1 was immunoprecipitated with the 12CA5 monoclonal superoxide production (Joneson and Bar-Sagi, 1998). antibody. Immunoprecipitated JNK activity was assayed by These observations have signi®cant public health using GST-c-Jun as a substrate. The graph shows arsenate- stimulated JNK1 activity from three separate experiments (a). implications as arsenate is the most prevalent arsenic Basal activity was de®ned as JNK1 activity from untreated cells species in surface or oxygenated water (Braman and transfected with pCMV5. Activity from treated cells transfected Foreback, 1973), which means that arsenate, [As(V)], in with pCMV5 was arbitrarily set at 100%. The data represent the addition to arsenite [As(III)] are carcinogens that mean+s.d. from three separate experiments. *P50.05 versus control. In b and c, a representative experiment of at least three humans are likely to encounter in their lifetime. Thus, separate experiments shows phosphorylated GST-c-Jun (b) and a a better understanding of the Rac pathway may Western blot of immunoprecipitated JNK1 protein (c) characterize a mechanism to describe how arsenic causes cancer, which may ultimately provide us with that speci®c GTP-binding proteins are involved in therapeutic targets to treat arsenate-induced cancer. arsenate- and arsenite-mediated JNK activation. The mechanism by which arsenic causes cancer is Interestingly, the data indicate that there is no unknown. Mutations in the ras gene have been involvement of Cdc42 in either arsenate or arsenite associated with many human cancers, including skin activation of JNK, although Cdc42 has been shown in melanoma [reviewed in (Bos, 1989)]. Based on those some cases to be involved in JNK activation (Coso et observations, Ras and Ras-dependent pathways have al., 1995; Minden et al., 1995). Rac and Rho both play been implicated as important mediators for human a role in arsenite- and arsenate-mediated JNK carcinogens. UV light is an example of a potent skin activation, although Rho was the most e€ective carcinogen that activates JNK through Ras (Derijard inhibitory mutant small GTP-binding protein that et al., 1994). It is interesting to note, however, that Ras inhibited JNK activation by arsenite and arsenate. plays little role in the activation of JNK by either The phospholipid, phosphatidylinositol 4,5- arsenate or arsenite. It is likely that Ras plays a role in bisphosphate (PIP2), is a precursor for the production the regulation of the ERKs, as arsenite induces of second messengers, such as inositol trisphosphate, anchorage-independent growth of mouse epidermal diacylglycerol, and 3,4,5-PIP3, by phospholipase C cells in soft agar (Huang et al., 1999). Our result is (PLC) and phosphatidyinositol 3-kinase (PI3-K). It is consistent with previous studies indicating that Ras interesting to note that Rho and Rac associate with does not play a role in the activation of JNK or p38 phosphatidyinositol 4-phosphate 5-kinase [(PIP5K); MAPK (Liu et al., 1996). From our results we can reviewed in (Ren and Schwartz, 1998)] and regulate conclude that there are mechanistic di€erences between the production of PIP2 (Ren et al., 1996). Thus, Rho UV light and arsenic-mediated activation of JNK. and Rac may provide a link between oxidative stress The involvement of PAK in arsenite- and not signaling and inositol phosphate production. It has arsenate-mediated JNK activation is an indication recently been shown that osmotic stress induces the that there are di€erences in signaling pathways Arsenic regulates Rac, Rho, PAK, MEKK3 and MEKK4 AC Porter et al 7800 between arsenate and arsenite in the cell. The precise di€erent pathways. Thus, it is not surprising to observe role that PAK plays in mediating the biological e€ects that arsenate and arsenite utilize two dissimilar of arsenite is not yet entirely clear. PAK1 has been MEKKs, MEKK3 and MEKK4, to activate JNK. shown to activate p38, another member of the stress In summary, we have shown that arsenate and activated protein kinase family (Bagrodia et al., 1995). arsenite activate JNK by utilizing di€erent, as well as In addition, a link between PAK and the ERK overlapping, signal transduction pathways (Figure 8). pathway has been established as PAK3 phosphory- Each of the proteins that we have identi®ed as lates Raf-1 in vitro and in vivo (King et al., 1998). PAK participating in the activation of JNK may also is also important in mediating the morphological regulate other signal transduction pathways and may changes of the cytoskeleton that occur during account for the many e€ects of arsenic. Activation of apoptosis (Zhang et al., 1995). With regard to JNK these pathways likely contributes to some of the regulation, it has been shown that Rho regulates JNK biological e€ects observed from chronic low-dose in HEK 293 cells independent of PAK (Teramoto et arsenic exposures frequently associated with arsenic al., 1996). Therefore, another e€ector kinase, other toxicity. Our results provide a framework for future than PAK, is required for JNK activation by arsenite studies to characterize the early events in arsenic (Figure 5). Additional e€ector kinases that are known signaling. activators of JNK and may be activated by arsenite include GCK (germinal center kinase), Tpl-2 (tumor progression locus 2), MLK3 (mixed lineage kinase), Materials and methods DLK (dual leucine zipper bearing kinase), and TAK1 [(TGF-b-activated protein kinase); reviewed in (Fanger Materials et al., 1997a)]. Alternatively, arsenite may by-pass the MAP4K level of the signaling cascade. Consistent with Sodium arsenate, ACS certi®ed, was purchased from JT this notion is the work of Fanger et al. (1997b) who Baker Chemical Co. (Phillipsburg, NJ, USA). Sodium arsenite, ACS certi®ed, was purchased from Fisher Scientific showed that Rac and Rho physically associate with (Tustin, CA, USA) and chemicals for protein electrophoresis MEKK4. Collectively, these data demonstrate a direct were purchased from Bio-Rad Laboratories (Hercules, CA, link between small GTP binding proteins and the USA). Anti-HA mouse monoclonal antibody was purchased MEKK family of proteins in the regulation of JNK by arsenate and arsenite. There are di€erences in the involvement of the MEKKs in arsenate and arsenite signal transduction as well. Dominant-negative MEKK2, MEKK3 and MEKK4 block arsenite activation of JNK. Only MEKK3 and MEKK4 block arsenate activation of JNK. The physiological role that the four MEKKs perform is just beginning to emerge. For example, MEKK4 interacts with the stress-inducible GADD4S proteins (Takekawa and Saito, 1998), which positions MEKK4 to regulate cellular responses to environ- mental stress such as arsenite and arsenate. An analysis of the amino acid sequence of these proteins provides us with some insight as to the function of these proteins. A similarity in amino acid sequence suggests a similarity in function, as all four MEKK proteins activate JNK (Blank et al., 1996; Gerwins et al., 1997; Lange-Carter et al., 1993; Minden et al., 1994; Xu et al., 1996; Yan et al., 1994). However, although the MEKKs have similar protein kinase domains at the carboxyl-terminus (Hanks et al., 1988), there are di€erences in the substrate speci®city for each of these kinases. For example, MEKK4 Figure 8 Summary of signal transduction pathways activated by activates the JNK (Gerwins et al., 1997) and p38 arsenate and arsenite. Arsenate and arsenite activate di€erent (Takekawa et al., 1997) pathways but not the ERKs, proteins to regulate JNK, which functions in the stress-activated di€erentiating it from MEKK1, 2 and 3, which are protein kinase pathway (SAPK). A SAPK pathway is a sequential capable of also activating the ERK pathway. MEKK2 protein kinase cascade where a protein, generically referred to as a MAP Kinase Kinase Kinase Kinase (MAP4K), phosphorylates and MEKK3 share a 94% amino acid sequence and activates a MAP Kinase Kinase Kinsase (MAP3K), which homology in their kinase domain (Blank et al., 1996), repeats the cycle by phosphorylating and activating the next while there is much less homology with MEKK1 and kinase in the cascade. The small GTP binding proteins are MEKK4 (Gerwins et al., 1997). There is even less localized upstream of the sequential protein kinase cascade. The anion transport protein regulates entry of arsenate into the cell, homology between MEKKs when MEKK2 and 3 are while arsenite, which is an uncharged arsenic species, enters the compared to MEKK1 and 4. A comparison of amino cell by di€usion. The small GTP binding proteins that are acid sequence in the putative regulatory domain, in regulated by arsenate and arsenite include Rac and Rho. Cdc42 particular, the sequence outside the kinase domain, and Ras do not appear to play a signi®cant role in arsenite and arsenate signaling to JNK. PAK plays a role in arsenite- shows a high level of dissimilarity between the dependent JNK activity. MEKK3 and MEKK4 are involved in MEKKs, which suggests that these kinases are both arsenate and arsenite activation of JNK, while MEKK2 may regulated by di€erent proteins and that they regulate be involved in the activation of JNK by arsenite Arsenic regulates Rac, Rho, PAK, MEKK3 and MEKK4 AC Porter et al 7801 from Boehringer Mannheim (Indianapolis, IN, USA). 100 mM HEPES, pH 7.2. HEK cells were treated with a ®nal Recombinant GST-c-Jun (1 ± 79) was expressed in the concentration of 300 mM arsenate for 2 h in phosphate free JM109 strain of E. coli and puri®ed using glutathione media with serum or 300 mM arsenite in media containing Sepharose (Amersham Pharmacia Biotech Inc) as described serum for 30 min. The cells were then washed twice with ice- previously (Hibi et al., 1993). cold PBS and lysed in 20 mM Tris, pH 7.6, 0.5% NP-40, 0.25 M NaCl, 3 mM EDTA, 3 mM EGTA, 1 mM PMSF,

2mM Na3VO4,5mg/ml leupeptin and 1 mM DTT. Soluble Cell culture and transfection proteins were collected after centrifugation at 15 000 r.p.m. Human embryonic kidney (HEK) 293 cells were maintained in in an Eppendorf centrifuge. The cell lysate (1 ± 3 mg) was Dulbecco's modi®ed Eagle's medium with 5% newborn calf incubated with a monoclonal antibody (0.4 mg/ml), clone serum, 5% calf serum, 100 units/ml penicillin and 100 mg/ml 12CA5, that recognizes the HA epitope (Boehringer streptomycin at 378C under 5% CO2. Transient transfection of Mannheim, Indianapolis, IN, USA) at a dilution of 1 : 250. HEK 293 cells was performed using phosphate, After 90 min, 10 ml of Protein A Sepharose (1 : 1 slurry, followed by DMSO shock (Cullen, 1987). cDNA constructs Sigma Chemical Co., St. Louis, MO, USA) was added and encoding pGEX-c-Jun (1 ± 79), pCMV5-RasN17, pcDNA3- the samples were maintained at 48C with rotation for at least RhoN19, pSRa3-HA-JNK1, pcDNA3-MEKK1 (KM), 1.5 h. The immune complexes were collected by centrifuga- pCMV5-MEKK2 (KM), pCMV5-HA-MEKK3 (KM), tion at 2000 r.p.m. for 1 min in an Eppendorf centrifuge. pCMV5-MEKK4, and pcDNA3-HA-bPAK (KM) were gifts Then, the immune complexes were washed twice with 1 ml of of GL Johnson. Constructs encoding pCMV5-RacN17 and lysis bu€er and once with kinase bu€er (20 mM HEPES, pCMV5-Cdc42N17 were gifts from N Dhanasekaran. pH 7.2, 20 mM b-glycerophosphate, 10 mM pNpp, 10 mM

To characterize the role of small GTP binding proteins in MgCl2,1mM DTT and 50 mM Na3VO4). The activity of the activation of JNK, cells were transfected with 10 mgof precipitated JNK was assayed with 1 mlof[g-32P]ATP epitope-tagged HA-JNK-1 and 20 mg dominant-negative (10 mCi/ml) and 1 mg of GST-jun in 40 ml of kinase bu€er small GTP-binding proteins [RacN17, Cdc42N17, RasN17,or at 308C for 20 min. The reaction was terminated by the RhoN19]. To characterize the role of the MEKKs in the addition of 40 mlof26Laemmli sample bu€er (Fling and activation of JNK, cells were transfected with 10 mgof Gregerson, 1986). Phosphorylated GST-c-Jun was separated epitope-tagged HA-JNK-1 and 20 mg of dominant-negative from unincorporated radioactivity by SDS ± PAGE using a MEKKs [(KM)-MEKK1, (KM)-MEKK2, (KM)-HA- 10% gel and identi®ed by autoradiography, then quanti®ed MEKK3, or (KM)-MEKK4]. Finally, to characterize the as described above. role of PAK in the activation of JNK, cells were transfected with 10 mg of epitope-tagged HA-JNK-1 and 20 mgof Statistical analysis epitope-tagged, dominant-negative PAK [(KM)-HA-bPAK]. Cells were plated on 100-mm dishes 1 ± 2 days prior to Statistical analysis was performed using pair-wise one-way transfection. Two days later, the cells were treated with analysis of variance (ANOVA). The data were analysed using 300 mM arsenate or arsenite and harvested for JNK activity. the SigmaStat (SPSS, Inc., Chicago, IL, USA) program and results were considered signi®cant at P50.05. The vertical bars that are starred in each graph represent the standard Solid phase JNK assay deviations of three separate experiments. Triplicate determi- HEK cells were treated with arsenate or arsenite in DMEM nations were signi®cantly di€erent from the pCMV5 control containing serum. Stock solutions of 300 mM sodium arsenate by one way analysis of variance with a secondary or sodium arsenite were prepared on the day of the experiment in Bonferroni's test. 100 mM HEPES, pH 7.2. The cells were then washed with ice- cold PBS and scraped into 600 ml of lysis bu€er (20 m Tris, M Immunoblotting pH 7.6, 0.5% NP-40, 0.25 M NaCl, 3 mM EDTA, 3 mM EGTA,

1mM PMSF, 2 mM Na3VO4,5mg/ml leupeptin and 1 mM In order to determine whether equal amounts of HA-JNK-1 DTT). Cell debris was removed by centrifugation at were immunoprecipitated for each condition in a particular 14 000 r.p.m. for 10 min at 48C. The protein concentration experiment, Western blots were performed as follows. The was determined as described previously, using BSA as a Protein A Sepharose beads were washed once as described standard (Bradford, 1976). A solid-phase kinase assay was above, and then 1 ml of lysis bu€er was added to the beads. A performed in which 40 mg of protein from cell lysates was volume of bu€er, which was equivalent to 100 mg of protein, incubated at 48C with 5 ml of a 1 : 1 slurry of fusion protein, based on the protein concentration that was used at the GST-c-Jun, which was bound to Sepharose beads (Hibi et al., beginning of the experiment, was removed from the second 1993). Following an incubation of 1.5 h with rotation, the beads 1 ml wash. The sample was resolved by SDS ± PAGE and the were collected at 2000 r.p.m. for 1 min. The precipitated proteins were transferred to nitrocellulose. The blot was proteins were washed twice with 1 ml of lysis bu€er and once incubated with JNK antibody (catalog number sc-571, Santa with kinase bu€er (20 mM HEPES, pH 7.2, 20 mM b- Cruz Biotechnology, Santa Cruz, CA, USA), at a dilution of glycerophosphate, 10 mM p-nitrophenylphosphate (pNpp), 1 : 1000 for 1 h at room temperature. After washing three times

10 mM MgCl2,1mM DTT and 50 mM Na3VO4). The activity with T-TBS, the blot was incubated with horseradish of precipitated JNK was assayed with 1 mlof[g-32P]ATP peroxidase-conjugated goat anti-rabbit IgG, at a dilution of (10 mCi/ml) in 40 ml of kinase bu€er at 308C for 20 min. The 1 : 1000 for 45 min at room temperature. The blot was washed reaction was stopped by the addition of 40 mlof26SDS ± three times with T-TBS and the presence of HA-JNK-1 was PAGE Laemmli sample bu€er. Phosphorylated GST-c-Jun was visualized by enhanced chemiluminescence followed by auto- separated from unincorporated radioactivity by SDS ± PAGE radiography. using a 10% gel (Fling and Gregerson, 1986), then identi®ed by autoradiography and quanti®ed by using a Packard Instant Imager Electronic Autoradiography SystemTM (Packard Instru- ment Co, Meriden, CT, USA). Acknowledgments JNK assay This work was supported, in part, by grants from the American Cancer Society (IRG 110T), the Southwest Stock solutions of 300 mM sodium arsenate or sodium Environmental Health Sciences Center (ES 06694), and arsenite were prepared on the day of the experiment in the Arizona Disease Control Research Commission. Arsenic regulates Rac, Rho, PAK, MEKK3 and MEKK4 AC Porter et al 7802 References

Adler V, Scha€er A, Kim J, Dolan L and Ronai Z. (1995). J. Meier R, Rouse J, Cuenda A, Nebreda AR and Cohen P. Biol. Chem., 270, 26071 ± 26077. (1996). Eur. J. Biochem., 236, 796 ± 805. Bagrodia S, Derijard B, Davis RJ and Cerione RA. (1995). J. Michiels F, Habets GG, Stam JC, van der Kammen RA and Biol. Chem., 270, 27995 ± 27998. Collard JG. (1995). Nature, 375, 338 ± 340. Blank JL, Gerwins P, Elliott EM, Sather S and Johnson GL. Michiels F, Stam JC, Hordijk PL, van der Kammen RA, (1996). J. Biol. Chem., 271, 5361 ± 5368. Ruuls-Van Stalle L, Feltkamp CA and Collard JG. (1997). Bos JL. (1989). Cancer Res., 49, 4682 ± 4689. J. Cell. Biol., 137, 387 ± 398. Bradford MM. (1976). Anal. Biochem., 72, 248 ± 254. Mikami K, Katagiri T, Iuchi S, Yamaguchi-Shinozaki K and Braman RS and Foreback CC. (1973). Science, 182, 1247 ± Shinozaki K. (1998). Plant J., 15, 563 ± 568. 1249. Minden A, Lin A, Claret FX, Abo A and Karin M. (1995). Cavigelli M, Li WW, Lin A, Su B, Yoshioka K and Karin M. Cell, 81, 1147 ± 1157. (1996). EMBO J., 15, 6269 ± 6279. Minden A, Lin A, McMahon M, Lange-Carter C, Derijard Coso OA, Chiariello M, Yu JC, Teramoto H, Crespo P, Xu B, Davis RJ, Johnson GL and Karin M. (1994). Science, N, Miki T and Gutkind JS. (1995). Cell, 81, 1137 ± 1146. 266, 1719 ± 1723. Cullen BR. (1987). Meth. Enzymol., 152, 684 ± 704. Qiu RG, Chen J, Kirn D, McCormick F and Symons M. Derijard B, Hibi M, Wu IH, Barrett T, Su B, Deng T, Karin (1995). Nature, 374, 457 ± 459. M and Davis RJ. (1994). Cell, 76, 1025 ± 1037. Ren XD, Bokoch GM, Traynor-Kaplan A, Jenkins GH, Elbirt KK, Whitmarsh AJ, Davis RJ and Bonkovsky HL. Anderson RA and Schwartz MA. (1996). Mol. Biol. Cell., (1998). J. Biol. Chem., 273, 8922 ± 8931. 7, 435 ± 442. Fanger GR, Gerwins P, Widmann C, Jarpe MB and Johnson Ren XD and Schwartz MA. (1998). Curr.Opin.Genet.Dev., GL. (1997a). Curr.Opin.Genet.Dev.,7, 67 ± 74. 8, 63 ± 67. Fanger GR, Johnson NL and Johnson GL. (1997b). EMBO Rossman TG, Stone D, Molina M and Troll W. (1980). J., 16, 4961 ± 4972. Environ. Mutagen., 2, 371 ± 379. Fling SP and Gregerson DS. (1986). Anal. Biochem., 155, Rouse J, Cohen P, Trigon S, Morange M, Alonso- 83 ± 88. Llamazares A, Zamanillo D, Hunt T and Nebreda AR. Freeman JL, Abo A and Lambeth JD. (1996). J. Biol. Chem., (1994). Cell, 78, 1027 ± 1037. 271, 19794 ± 19801. Sakurai T, Kaise T and Matsubara C. (1998). Chem. Res. Gerwins P, Blank JL and Johnson GL. (1997). J. Biol. Toxicol., 11, 273 ± 283. Chem., 272, 8288 ± 8295. Smith AH, Goycolea M, Haque R and Biggs ML. (1998). Guo Y-L, Baysal K, Yang L-J and Williamson JR. (1998). J. Am. J. Epidemiol., 147, 660 ± 669. Biol. Chem., 273, 4027 ± 4034. Stoner JC, Whanger PD and Weswig PH. (1977). Environ. Habets GG, Scholtes EH, Zuydgeest D, van der Kammen Health Pers., 19, 139 ± 143. RA, Stam JC, Berns A and Collard JG. (1994). Cell, 77, Takekawa M, Posas F and Saito H. (1997). EMBO J., 16, 537 ± 549. 4973 ± 4982. Hanks SK, Quinn AM and Hunter T. (1988). Science, 241, Takekawa M and Saito H. (1998). Cell, 95, 521 ± 530. 42 ± 52. Teramoto H, Crespo P, Coso OA, Igishi T, Xu N and Hibi M, Lin A, Smeal T, Minden A and Karin M. (1993). Gutkind JS. (1996). J. Biol. Chem., 271, 25731 ± 25734. Genes Dev., 7, 2135 ± 2148. Treisman R. (1996). Curr. Opin. Cell. Biol., 8, 205 ± 215. Huang C, Ma W-Y, Li J, Goranson A and Dong Z. (1999). J. Trigon S and Morange M. (1995). J. Biol. Chem., 270, Biol. Chem., 274, 14595 ± 14601. 13091 ± 13098. Jacobson-Kram D and Montalbano D. (1985). Environ. van Leeuwen FN, van der Kammen RA, Habets GG and Mutagen., 7, 787 ± 804. Collard JG. (1995). Oncogene, 11, 2215 ± 2221. Joneson T and Bar-Sagi D. (1998). J. Biol. Chem., 273, Xu S, Robbins DJ, Christerson LB, English JM, Vanderbilt 17991 ± 17994. CA and Cobb MH. (1996). Proc. Natl. Acad. Sci. USA, 93, Kenney LJ and Kaplan JH. (1988). J. Biol. Chem., 263, 5291 ± 5295. 7954 ± 7960. Yan M, Dai T, Deak JC, Kyriakis JM, Zon LI, Woodgett JR King AJ, Sun H, Diaz B, Barnard D, Miao W, Bagrodia S and Templeton DJ. (1994). Nature, 372, 798 ± 800. and Marshall MS. (1998). Nature, 396, 180 ± 183. Yeh S, How SW and Lin CS. (1968). Cancer, 21, 312 ± 339. Kyriakis JM and Avruch J. (1996). J. Biol. Chem., 271, Zhang S, Han J, Sells MA, Cherno€ J, Knaus UG, Ulevitch 24313 ± 24316. RJ and Bokoch GM. (1995). J. Biol. Chem., 270, 23934 ± Lange-CarterCA,PleimanCM,GardnerAM,BlumerKJ 23936. and Johnson GL. (1993). Science, 260, 315 ± 319. Zhao CQ, Young MR, Diwan BA, Coogan TP and Waalkes Lee TC, Oshimura M and Barrett JC. (1985). Carcinogenesis, MP. (1997). Proc. Natl. Acad. Sci. USA, 94, 10907 ± 10912. 6, 1421 ± 1426. LiuY,GuytonKZ,GorospeM,XuQ,LeeJCandHolbrook NJ. (1996). Free Rad. Biol. Med., 21, 771 ± 781.