Activation of c-Raf Kinase by Ultraviolet Light

The present study highlights retinoids as modulators of c-Raf kinase activation by UV light. Whereas a number of retinoids, including retinol, 14-hydroxyretroretinol, anhydroretinol (AR), and retinoic acid bound the c-Raf cysteine-rich domain (CRD) with equal affinity in vitro as well as in vivo, they displayed different, even opposing, effects on UV-mediated kinase activation; retinol and 14-hydroxyretroretinol augmented responses, whereas retinoic acid and AR were inhibitory. Oxidation of thiol groups of cysteines by reactive oxygen, generated during UV irradiation, was the primary event in c-Raf activation, causing the release of zinc ions and, by inference, a change in CRD structure. Retinoids modulated these oxidation events directly: retinol enhanced, whereas AR suppressed, zinc release, precisely mirroring the retinoid effects on c-Raf kinase activation. Oxidation of c-Raf was not sufficient for kinase activation, productive interaction with Ras being mandatory. Further, canonical tyrosine phosphorylation and the action of phosphatase were essential for optimal c-Raf kinase competence. Thus, retinoids bound c-Raf with high affinity, priming the molecule for UV/reactive oxygen species-mediated changes of the CRD that set off GTP-Ras interaction and, in context with an appropriate phosphorylation pattern, lead to full phosphotransferase capacity.

ligated by their respective growth factors, dimerize, become autophosphorylated, and recruit adapter molecules (Grb2) and the nucleotide exchange factor SOS to the cell membrane. The further assembly of GTP-bound Ras enables c-Raf to translocate from the cytosol to the plasma membrane, where an as yet unidentified mechanism bestows competence on c-Raf to activate the MAPK cascade (16). Phosphorylation of tyrosine residues (e.g. Tyr-340 and Tyr-341 (17)) and dephosphorylation of serine residues (Ser-259 and Ser-621 (18)) are believed to lock c-Raf into the optimally competent form.
For docking with Ras, two important contact sites in the regulatory domain of c-Raf have been identified, one centered on the stretch of amino acids 51-131, the other contained within the CRD (19,20). What remains to be identified is the initial molecular event that triggers cytosol-to-membrane translocation. Whether this involves changes in the phosphorylation pattern and consequent changes in the conformation of the regulatory domain is still unclear. The participation of lipid mediators in the activation of c-Raf has been suspected because of structural similarities with the PKC family of serine/threonine kinases (21,22), which harbor lipid binding sites in their CRD tandem repeats (23)(24)(25). Bound phosphatidylserine enhanced PKC activity (26). Several groups have identified lipid binding sites in the regulatory and catalytic regions of c-Raf (27)(28)(29)(30). Interestingly, Romero and colleagues (30) suggested that phosphatidic acid mediated c-Raf translocation from cytoplasm to membrane independently of its association with Ras.
Besides the classical RTPK signal chain, alternative activation signals exist. Ultraviolet light and ionizing irradiation as well as oxidizing agents lead to c-Raf/MAPK activation (13)(14)(15). To understand the biological significance, it is worth remembering that macrophages naturally produce substantial concentrations of hydrogen peroxide and that reactive oxygen species (ROS) are produced in every cell type by mitochondria as well as by dedicated enzyme systems. The changing view is that ROS, like nitric oxide, serve as normal intracellular messengers (31)(32)(33)(34). Also long known, the potent activating capacity of ultraviolet irradiation rests on the intracellular production of ROS (35,36).
Whereas the chemistry of oxidative activation of serine/threonine kinases is poorly understood, it stands to reason that ROS target the most susceptible groups in c-Raf, namely the thiols of cysteines, assuming a direct chemical modification and not activation of an upstream factor. Direct attack by ROS is all the more likely, since six cysteine residues are clustered within a stretch of 50 amino acids of the regulatory domain, all susceptible to oxidation. The questions of what changes the CRD may undergo during UV activation and how retinoids regulate such chemical changes are addressed in the present report. If the biology of vitamin A were a guide (37)(38)(39)(40), the expectation would be that hydroxylated retinoids (retinol and 14HRR) enhanced, whereas anhydroretinol attenuated, the UV effects. This prediction was borne out.
As previously shown, vitamin A functions as regulatory cofactor for redox regulation of c-Raf and other serine/threonine kinases (41,42). A single site capable of binding several natural vitamin A metabolites at nanomolar affinity was mapped to the CRD. Since bound retinoids influenced redox activation (42), the importance of this domain as a primary target for oxidation was suggested. Further, two zinc coordination centers exist per CRD, each composed of three thiol groups of cysteines and one imino group of histidine. Chemistry predicts that oxidation of one or more thiols would compromise the integrity of the zinc finger. Precedents that zinc finger structures serve as redox sensors of enzymes exist in bacteria, and the related structures in PKC have recently been shown to shed their bound Zn 2ϩ ions as a result of oxidation (43). 2 We therefore asked whether Zn 2ϩ ions would be liberated from the c-Raf CRD during UV irradiation and whether pro-oxidant retinoids would impact on zinc release.
We further investigated whether UV-mediated oxidation of the CRD was per se sufficient for kinase activation but found that, unlike PKC ␣ and isoforms (42,43), c-Raf required the additional productive interaction with GTP-bound Ras, as well as modifications in the tyrosine and serine/threonine phosphorylation patterns. Nevertheless, the CRD emerged as the key target structure for ROS elicited by stress-related mechanisms. Several findings in this report will highlight our conclusion: the triggering of c-Raf activity by ROS generated during UV irradiation in vitro; the binding of pro-oxidant retinoids to the CRD; the conversion of thiol groups to disulfide by UV exposure in vivo; the release of Zn 2ϩ ions after UV irradiation; and the requirement for the CRD to communicate with Ras. Therefore, despite its appearance as a rigid structural element, the CRD should be more appropriately viewed as a dynamic, redoxsensitive hinge with profound regulatory importance for initiation of the c-Raf activation cycle by UV light.

MATERIALS AND METHODS
Retinoids-All-trans isomers of AR and 14HRR were synthesized as described previously (37,44). trans-Retinol and RA were purchased from Sigma and purified by high pressure liquid chromatography. [ 3 H]Retinol was purchased from PerkinElmer Life Sciences.
Plasmids-FLAG-c-Raf was a gift from Dr. Roger J. Davis (University of Massachusetts, Worcester, MA). N17-Ras was obtained from Dr. N. M. Nathanson (University of Washington, Seattle, WA). glutathione S-transferase (GST)-and FLAG-human c-Raf CRD ( 136 LTTHNFA-RKTFLKLAFCDICQKFLLNGFRCQTCGYKFHEHCSTKVPTMC-VDWSNIRQLLL 195 ; this peptide contains the natural Trp-186 residue, responsible for fluorescence emission) were constructed as previously described (41). The FLAG-c-Raf chimera was constructed by replacing the c-Raf CRD with the PKC␣ CRD. The following primers were used to isolate the PKC␣ CRD cDNA by polymerase chain reaction (template kindly provided by Dr. T. Powell, Memorial Sloan-Kettering Cancer Center, New York): 5Ј-CAT GTT CCC CTC ACA ACA CAC AAG TTC AAA ATC CAC and 5Ј-GAT GTT ACT CCA GTC CAC GCA GAG GCT GGG GAC ATT GAT. This product was used to synthesize the N and C termini of c-Raf chimeric gene using the following primers: 5Ј-GCG GCC GCG AAT TCA ATG GAG CAC ATA CAG and 5Ј-TGA AGA CAG GTG GGA TCC TTA CTA GAA GAC AGG CAG CCT, respectively. The two products were used to generate the full-length molecule by PCR, which was then cloned into the EcoRI/BamHI sites of the Sigma vector pFLAG-CMV2. Fidelity was confirmed by sequencing.
Transfection and UV Activation-COS-7 cells were transfected ac-cording to the calcium phosphate method (45). Briefly, 5 ϫ 10 5 cells were plated in 60-mm dishes the day before transfection. 2 h prior to transfection, the medium was replaced with fresh growth medium (Dulbecco's modified Eagle's medium (high glucose) supplemented with 10% fetal calf serum). 12 g of DNA/dish were mixed with CaCl 2 and phosphate buffer to form a fine precipitate, which was then dispersed over the cells. The day after transfection, the medium was removed and the cells were washed twice with 3 ml each of phosphate-buffered saline (PBS). The cells were cultured with 2 ml of retinoid-free, phenol red-free Dulbecco's modified Eagle's medium (high glucose) for 2 1 ⁄2 days prior to activation. UV irradiation was performed for 2 min at 400 milliwatts/ cm 2 using the 312-nm wavelength. Cultures were incubated at 37°C for 10 min after irradiation and harvested. c-Raf Immunoprecipitation/Kinase Assay-Cells were lysed in 100 l lysis buffer (50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 2 mM EDTA, 1 mM EGTA, 1% Triton X-100, 25 g/ml each leupeptin and aprotinin, 30 mM ␤-glycerophosphate, 30 mM NaF, 1 mM phenylmethylsulfonyl fluoride, and 1 mM vanadate). The lysates were precleared with 30 l of a 50% (v/v) protein G-agarose slurry, and the FLAG-c-Raf protein was precipitated using 30 l of anti-FLAG M2 affinity gel (Sigma). The immunoprecipitates were washed four times with lysis buffer containing 0.5 M NaCl and twice with kinase buffer (30 mM Tris-HCl, pH 7.5, 10 mM MgCl 2 , 0.5 mM EGTA, and 1 mM vanadate). The kinase reaction was performed in 20 l of kinase buffer using 200 ng of kinase-disabled His-MEK (K97M) (provided by Dr. R. Kolesnick, Memorial Sloan-Kettering Cancer Center) as substrate, 60 mM ATP, and 10 mCi of [␥-32 P]ATP (6,000 Ci/mmol). The reaction was carried out for 20 min at 30°C and terminated by the addition of 10 l of 4ϫ Laemmli buffer (46). After separation by SDS-PAGE on 7.5% gels, the proteins were transferred to polyvinylidene difluoride membrane and subjected to autoradiography, followed by Western analysis. Both Western blot and autoradiographs were evaluated by densitometry (Bio-Rad GS 700 densitometer with Quantity One software). Kinase activity values were normalized for the amount of FLAG-c-Raf precipitated. Data were analyzed by Student's t test.
AMS Trapping of Free Thiols in Vivo-COS cells transfected with FLAG-c-Raf CRD were UV-irradiated, treated with 1 or 10 mM H 2 O 2 , or left untreated and incubated at 37°C for 10 min. The medium was removed, cells were washed with PBS, and 150 l of 100 mM iodoacetamide in permeabilization buffer (lysis buffer containing 20 g/ml digitonin instead of 1% Triton X-100) (47) were added. After incubation for 10 min at 37°C, the reaction was terminated by the addition of 1% Triton X-100, and cells were frozen in liquid nitrogen. The FLAG-c-Raf CRD protein was immunoprecipitated using anti-FLAG M2 affinity gel as described above. The beads were resuspended in 25 l of 10 mM dithiothreitol, 100 mM Tris, pH 8, and 0.5% SDS and incubated at 42°C for 1 h. Following reduction, 25 l of 50 mM AMS, 100 mM Tris, pH 8, and 0.5% SDS were added, and thiol trapping was allowed for 30 min at 37°C (48). The reaction was terminated by the addition of 10 l of 4ϫ nonreducing Laemmli buffer and boiling for 5 min. The samples were applied to a 16.5% Tris-Tricine SDS-PAGE gel from Bio-Rad and analyzed by Western blot using monoclonal anti-FLAG M2 antibody (Sigma).
Bacteria Growth and Protein Purification-The c-Raf-CRD was expressed as GST fusion protein in the BL21/DE3 strain of Escherichia coli (Novagen). The growth conditions were as follows. The bacteria were initially grown at 37°C to an optical density at 600 nm of 0.5 and then transferred to room temperature. When the optical density reached 0.7-0.8, protein synthesis was induced by the addition of 0.5 mM isopropyl-␤-D-thiogalactopyranoside, and the cells were harvested 2 h later. Bacteria were lysed by two passages through a French press, and the proteins were purified by affinity chromatography on the glutathione-agarose matrix (Sigma) according to a standard protocol. Purity by Coomassie Blue staining of SDS-PAGE was usually Ͼ90% by this protocol.
Retinoid Binding Assay-The fluorescence emission spectra of 250 nM GST-c-Raf CRD fusion protein, equimolar concentrations of proteinretinoid complexes, and quantitative fluorescence measurement of 250 nM protein with retinoid titration at 25 nM increments, from a 75 M stock solution in methanol, were performed in PBS, purged of oxygen by sparging with helium for 15 min in a JASCO spectrofluorimeter (model FP777), as described in detail (41). The protein solution was excited at 280 nm, and the emission spectra were monitored from 300 to 550 nm, or at 330 nm for the single wavelength measurements. All measurements were repeated 4 -6 times. Binding constants were calculated by nonlinear curve fitting according to the theorem by Norris et al. (49). Complementary binding data were obtained by the method of enhancement of retinol fluorescence (41). 2 I. Korichneva, unpublished data.
In Vivo Retinol Binding Assays-COS-7 cells were transfected with FLAG-c-Raf cDNA and cultured in retinol-free medium as indicated above. Cells were preincubated with 100 nM retinol, 14HRR, AR, RA, or phosphatidylserine for 15 min prior to the addition of 5 nM [ 3 H]retinol for 30 min and harvested after washing with PBS containing 0.5% bovine albumin. Extracts were prepared by repeated freeze thawing in 0.25 M Tris, pH 8, supplemented with 25 g/ml each leupeptin and aprotinin, 30 mM ␤-glycerophosphate, 30 mM NaF, 1 mM phenylmethylsulfonyl fluoride, and 1 mM vanadate. c-Raf immunoprecipitates were washed extensively with PBS containing 0.5 M NaCl and 0.5% bovine albumin, and incorporated counts were measured by liquid scintillation counting. Results were expressed as differentials between immunoprecipitates of transfected and nontransfected cultures.
Zinc Release Assay-A solution of 250 nM of GST-c-Raf CRD fusion protein in PBS was UV-irradiated at 312 nm for 2 min in the absence or presence of an equimolar concentration of retinol, AR, or retinol plus AR, followed by the addition of 3 M TSQ. The changes in the fluorescence emission spectrum of TSQ (excitation max ϭ 335 nm) were monitored from 390 -600 nm.

RESULTS
Retinoids Bind the c-Raf CRD in Vitro-Two spectrofluorimetric methods, state of the art in the field (41,49), were employed to measure binding of retinoids to the c-Raf CRD: quenching and enhancement. Quenching is based on the decrease in the intrinsic protein fluorescence due to resonance energy transfer to a suitable bound ligand. Fluorescence emanated from the natural Trp-186 residue. Preliminary evidence 3 suggested that retinol bound nearby at the second zinc chelation center formed by Cys residues 165, 168, and 184, in cooperation with His-139. Enhancement, or increase in the intrinsic fluorescence of the ligand/retinoids, is predicated on the movement of the ligated retinoids from the aqueous phase to the hydrophobic environment of the receptor/CRD protein.
We have previously shown that retinol bound c-Raf and certain PKC CRDs with nanomolar affinity (41). To test for binding of other natural retinol metabolites, solutions of bacterially expressed GST-c-Raf-CRD fusion protein were excited at 280 nm, and the changes in protein fluorescence emission spectra were recorded, brought about by the additions of stoichiometric amounts of retinoids. As illustrated in Fig. 1A, the fluorescence intensity of c-Raf CRD decreased in the presence of 14HRR, AR, or RA, indicating binding. The signal was particularly prominent for 14HRR, followed by AR, and moderately decreased by RA. In distinction from retinol, no detectable fluorescence resonance energy transfer signal was generated by the last three retinoids, probably due to their poor fluorescence properties.
Binding of the retinoids, retinol, 14HRR, and AR, to the c-Raf CRD was confirmed by the fluorescence enhancement method. Fig. 1B shows the idiosyncratic increases in the emission spectrum intensities that each retinoid produced when ligated to an equimolar amount of c-Raf CRD. The magnitude of the enhancement differed for each retinoid, reflecting the differences in their chromophore properties. RA, a poorly fluorescent retinoid, could not be evaluated by this method.
To determine the binding constants, titrations of retinol, 14HRR, AR, and RA were performed (Fig. 2). c-Raf CRD (250 nM) was excited at 280 nm, and its fluorescence emission was monitored at 330 nm after the addition of each 25 nM increment of retinoid from a 75 M stock solution in methanol. The binding curves, corrected for inner filtering, indicated that binding was saturable. When applied to a nonlinear curve fitting theorem developed by Norris et al. (49), the apparent dissociation constants in the nanomolar range were computed (Table I). Titration using GST as specificity control yielded a flat line (data not shown), no binding having taken place. Together, the results demonstrate high affinity binding of the retinoids to the CRD in vitro. Retinoids Bind c-Raf in Vivo-Retinol has been shown to bind c-Raf in vivo (41). To demonstrate binding of other reti-3 B. Hoyos, unpublished observation. noids as well, a competition assay was developed, using COS-7 cell transfectants expressing the full-length FLAG-c-Raf protein. Cells were labeled with 5 nM [ 3 H]retinol in the presence or absence of 1 ϫ 10 Ϫ7 M retinol, 14HRR, AR, and RA as cold competitors, and bound radioactivity was measured in c-Raf immunoprecipitates using anti-FLAG antibody. All retinoids tested competed for binding of [ 3 H]retinol to c-Raf, demonstrating specific binding (Fig. 3). Phosphatidylserine, a known ligand of the c-Raf CRD, competed only moderately with retinol, indicating minor overlap of binding sites (27)(28)(29). Thus, the in vivo and in vitro binding results established that the c-Raf CRD harbors a receptor site with equal specificity for four retinoids tested, extending our previous findings (41,42).
Retinoids Modulate c-Raf Activation by UV-Having demonstrated binding of retinoids to the CRD of c-Raf in vitro and in vivo, it was important to inquire into the functional significance. Because UV irradiation activated c-Raf kinase capacity by an oxidative mechanism and because retinol had emerged as pro-oxidant regulator (41,42), we tested whether retinoids served as cofactors in UV activation, using FLAG-c-Raf expressed to a high level in COS-7 cells. The results of c-Raf immunoprecipitation/kinase assays showed that retinoids indeed modulated the UV responses in a dose-dependent fashion, affecting the magnitude of kinase activation (Fig. 4) but not the kinetics (as published previously (41)). When compared with retinoid-deprived cultures, retinol and 14HRR at 1 M were found to produce an increase in phosphotransfer to disabled MEK, which peaked at 10 min (3.7-fold increase (p ϭ 0.004, n ϭ 4) and 2.3-fold increase (p ϭ 0.085, n ϭ 3), respectively). In contrast, AR and RA at 1 M produced decreases in FLAG-c-Raf kinase activity (1.8-fold decrease (p ϭ 0.019, n ϭ 3) and 13.6fold decrease (p ϭ 0.003, n ϭ 3), respectively). On their own, retinoids had no discernible effect on kinase activity. Thus, the retinol metabolites tested showed opposite effects on c-Raf kinase activation mediated by UV-irradiation; retinol and 14HRR enhanced while RA and AR antagonized the activity. Vitamin E, a known antioxidant, had no effect on UV-mediated c-Raf kinase activity (data not shown).
The Retinoids Are Functional Antagonists-Retinoids bound the c-Raf CRD equally well in vitro and competed for binding in vivo but had opposite effects on survival of cells (38 -40, 50). Because these effects were at least partially attributable to c-Raf kinase regulation in vivo (41), we tested whether retinol and AR would behave as functional antagonists in c-Raf kinase assays. As shown in Fig. 5, the inhibitory effect of AR on retinol, when used at equimolar concentrations, was evident. These results indicate that the retinoids compete for binding to c-Raf and act as functional antagonists.
UV Activation of c-Raf Is Not Reversible by Reduction-UV irradiation generates ROS (51)(52)(53). At low levels, these radicals are not only tolerated by cells but are important for normal signal transduction (33,34). Since the hallmark of physiological activating signals is their reversibility, we investigated whether UV activation of c-Raf was abrogated by reducing agents. Fig. 6     This assay is predicated on the alkylation of free thiol groups, but not cysteine residues, by iodoacetamide. Reduction of disulfide, generated during UV/ROS exposure, restores free thiol groups that are now available for reaction with AMS. Each bound AMS residue theoretically increases the molecular mass of the protein by 490 daltons. The altered size of FLAG-c-Raf CRD protein was assessed by electrophoresis in SDS-PAGE under nonreducing conditions, although conformational changes could also contribute. The product of UV-irradiated cells migrated significantly more slowly, consistent with AMS modifications of the thiol groups, compared with the reference protein from untreated cells (Fig. 7A). Oxidation of cells by two different concentrations of hydrogen peroxide caused the appearance of two differently migrating bands, the slower of which represented the higher degree of oxidation with the relatively larger number of disulfide groups that, after the in vitro reduction, were trapped by AMS.
UV Irradiation of c-Raf CRD Results in Zinc Release-The c-Raf CRD comprises two zinc-coordinated centers, one of which is formed by the three thiol groups of Cys-165, Cys-168, and Cys-184 and one imino group of His-139, whereas the other is formed by Cys-152, Cys-155, Cys-176, and His-172. Because thiol groups were modified following UV irradiation, we predicted that Zn 2ϩ would be released. Furthermore, we investigated to what extent retinoids might modulate the UV-mediated oxidation and zinc release from the c-Raf CRD. Fig. 7B illustrates that the c-Raf-CRD fusion protein indeed shed its zinc upon UV irradiation, as determined by the binding of liberated zinc to TSQ, a zinc-sensitive fluorescent probe, in agreement with recent in vivo observations. 3 In the presence of retinol, the fluorescence emission of TSQ was significantly enhanced, indicating that at similar ROS output, additional zinc nevertheless was released. By contrast, AR inhibited UVmediated release of zinc ions from the c-Raf CRD. These results are consistent with the postulated function of retinoids as redox regulators. They strengthen the idea that the CRD represents the primary target of UV irradiation and oxidation. 2 The hierarchy of in vitro effects of retinoids paralleled those seen in vivo precisely. These results for the first time also imply a central role of zinc ions in the regulation of c-Raf kinase.
The c-Raf CRD Is Essential for c-Raf Function-c-Raf shares the structural CRD motif with members of the PKC family (21). Furthermore, the homologous domains in the PKC family bearing the diacylglycerol and phorbol ester binding sites have been implicated in the redox regulation of kinase function but do not promote interaction with Ras. To investigate whether these domains are functionally interchangeable, we replaced the c-Raf CRD domain with the PKC␣ C1B domain. COS-7 cells, transfected with wild type FLAG-c-Raf or the FLAG-tagged chimeric c-Raf/PKC construct, were treated with UV irradiation, serum, or PMA 2 days post-transfection. The data in Fig.   8 demonstrate that swapping the CRDs inhibited c-Raf activity in response to UV by 76% (p ϭ 0.1, n ϭ 3) and serum by 61% (p ϭ 0.1, n ϭ 3). In contrast, the PMA response was augmented 2-fold for the chimeric compared with wild type c-Raf (p ϭ 0.02, n ϭ 3). The inability of the chimera to respond to UV irradiation suggested that the c-Raf CRD was essential for c-Raf function not only in mediating the important Ras/c-Raf interaction but also as primary sensor of UV signal transmission. On the other hand, the enhanced responsiveness to PMA may be explained by Ras-independent c-Raf activation due to the fact that PMA binding to the CRD facilitated localization at the plasma membrane (55), in analogy to the well studied paradigm governing PKC cytosol-to-membrane translocation and activation (23).

UV Activation of c-Raf Is Dependent on Ras/c-Raf
Interaction-Growth factor-mediated activation of c-Raf is dependent on Ras/c-Raf interaction (16). It was therefore of interest to determine whether c-Raf activation by UV also required interaction with Ras. To address this question, the N17-Ras mutant (56) was used as a dominant negative element in FLAG-c-Raf cotransfection experiments in COS-7 cells. Fig. 9A shows the level of c-Raf kinase activity stimulated by UV in comparison with serum and PMA. The presence of the N17-Ras dominantnegative mutant caused a 66% reduction in the UV-mediated c-Raf activation (p ϭ 0.007, n ϭ 3), suggesting mandatory interaction with GTP-Ras, like receptor protein-tyrosine kinase-mediated c-Raf activation. Inhibition by 38% (p ϭ 0.09, n ϭ 3) of PMA-induced c-Raf activation in the presence of N17-Ras indicated the existence of a Ras-independent pathway, as previously documented by Marais et al. (57). UV activation, however, followed the classical, Ras-dependent pathway.
This conclusion was confirmed independently using a pharmacological inhibitor of Ras post-translation modification. Perillic acid inhibits cysteine isoprenylation, without which Ras cannot localize to the plasma membrane and remains inoperative (58). Pretreatment of COS-7 cells for 12 h with 3 mM perillic acid resulted in 80% inhibition of UV-mediated FLAGc-Raf kinase activation (Fig. 9B). These results are in agreement with those obtained by the Ras dominant-negative mutant. Taken together, they indicate that the Ras/c-Raf interaction was absolutely required not only for activation of c-Raf by growth factor but also by UV irradiation.
Tyrosine Phosphorylation and Serine or Threonine Dephosphorylation Are Essential for c-Raf Optimal UV Activation-Phosphorylation of tyrosine residues Tyr-340 and Tyr-341 is essential for c-Raf activation, one of the kinases responsible being a member of the Src family (59,60). To evaluate the role of Src kinases in the c-Raf activation by UV, herbimycin A, a potent Src inhibitor, was used (61,62). COS-7 transfectants were treated with 3 M herbimycin A for 30 min prior to stimulation. Fig. 10 shows that the activation of c-Raf mediated by UV is decreased by 47% (p ϭ 0.1, n ϭ 2) and that of serum is decreased by 71% (p ϭ 0.09, n ϭ 2) (compare lane 2 with lane 4, and compare lane 3 with lane 5) in herbimycin A-pretreated cultures, indicating that UV stimulation does not bypass the requirement for c-Raf tyrosine phosphorylation.
Tyrosine kinases are an essential component of c-Raf activation. Omnipresent protein-tyrosine phosphatases counteract tyrosine phosphorylation. Hence, their inhibition may result in the constitutive activation of c-Raf. Since phosphatases are known to be deactivated by ROS (63, 65), as demonstrated especially for protein-tyrosine phosphatase 1B with its ROSsensitive cysteine in the catalytic domain (66), it was necessary to demonstrate UV activation in cells in which protein-tyrosine phosphatases were blocked by pharmacological inhibitors. FLAG-c-Raf-transfected COS-7 cells were therefore UV-irradiated in the presence of 1 mM sodium vanadate, a specific inhibitor of tyrosine phosphatases. Fig. 10 shows that UV irradiation caused a 20% increase in c-Raf activation in vanadate- To determine the influence of serine/threonine phosphatases on c-Raf activation, we treated FLAG-c-Raf-transfected COS-7 cells with 100 nM okadaic acid (67, 68) for 30 min prior to stimulation by UV. The presence of okadaic acid caused a 67% decrease in UV-mediated c-Raf activation (p ϭ 0.01, n ϭ 3) (Fig.  10, compare lane 2 with lane 6); okadaic acid alone caused a negligible increase in c-Raf basal activity (data not shown). These data indicate the importance of an okadaic acid-sensitive phosphatase, most likely protein phosphatase 2A, since this has been shown to form complexes with c-Raf in vitro and in vivo (18) during the activation of c-Raf kinase. Taken together, the data indicate that the direct chemical effect of UV irradia- FIG. 8. The c-Raf CRD is essential for kinase function. FLAGc-Raf wild type and FLAG-c-Raf chimera with the c-Raf CRD substituted for the PKC␣ C1B domain were activated by UV irradiation, serum (5% fetal calf serum), or phorbol ester (100 ng/ml PMA). The wild type responded to the stimuli, but the chimera, being unable to interact with Ras, did not respond to UV or serum. Ras-independent membrane targeting and activation may explain the effect of PMA on the chimera (55). tion on the molecule was not sufficient for c-Raf activation and did not bypass the requirements for Ras as well as tyrosine kinases and phosphatases, well known for their obligatory roles in growth factor and cytokine-mediated activation. DISCUSSION To understand the role of retinoids as regulators of c-Raf function, we pursued four seemingly separate lines of experimentation. When integrated, however, our results offer a new perspective on the function of the CRD, a subdomain of the N-terminal regulatory domain that was long suspected to play a crucial role in kinase activation. First, our previous finding that retinol bound the CRD was expanded to 14HRR, RA, and AR. These retinoids bound at the same site as retinol with nanomolar affinity, precisely replicating the findings with certain PKC CRDs (42). While they share the same ␤-ionone ring and polyene structure that presumably furnish the contact surface, they differ in the functional groups at their tail ends. The presence or absence of hydroxyls dictated biological function, as retinol and 14HRR acted as agonists, whereas AR that lacks hydroxyl groups was antagonistic (37-40, 50, 69 -71). Complementing the in vitro binding studies (Figs. 1 and 2 and Table I), we also furnish evidence for binding in intact cells and provide at least qualitative evidence that all four had affinity for full-length c-Raf in native conformation (Fig. 3). Since retinol and 14HRR are nearly ubiquitous in tissues at concentrations (2 ϫ 10 Ϫ6 and 1 ϫ 10 Ϫ7 M, respectively) exceeding the K d of the CRD-retinoid complex (2 ϫ 10 8 mol Ϫ1 ), one can safely assume that inside cells c-Raf is always loaded with retinoids.
It has long been recognized that UV irradiation and the ROS this engenders lead to activation of a number of signal pathways (32,36,72). Prominent among these is the c-Raf/MAPK axis that plays a role in physiological responses to UVB, for instance in skin cells (73)(74)(75)(76). We used UVB irradiation of cells as a quasipharmacological, convenient mode of activation, since production of the actual mediator of kinase activation, ROS, is restricted to the light period. The consequences for c-Raf were then determined free of concern that continued exposure to ROS might accumulate damage, as might be the case with most oxidizing chemicals. Brief irradiation of cells initiated events that peaked 10 min later in the expression of substantial c-Raf phosphotransferase activity, determined in immunoprecipitates by phosphorylation of the disabled substrate, MEK. The presence of retinol promoted a substantial increase in kinase activity, confirming previous findings that peroxide-mediated activation of c-Raf and PKC was facilitated by bound retinol (41,42). Extending the studies to natural retinol metabolites, it was observed that 14HRR, as predicted from its agonistic properties in cell survival assays (50,70), also enhanced UV-triggered c-Raf kinase activation, whereas AR did not (37)(38)(39)(40)69). The latter effect mirrored the physiological consequence of growth inhibition and apoptosis that cells experience when exposed to AR. We have provided evidence elsewhere that AR displaced retinol from the common receptors and thus mimicked a state of retinol deficiency, leading to growth retardation and apoptosis (37)(38)(39)(40)69). Similar physiological effects are not uncommon with cultures depleted of vitamin A (38,50,70). That retinol and AR acted as pharmacologic mutual inhibitors was also evident in the UV activation experiments (Fig. 5).
Since UV irradiation led to production of ROS, held responsible for the initiation of the c-Raf activation cycle, it was desirable to show direct chemical modifications of c-Raf. This was all the more necessary, since ROS reportedly activates protein-tyrosine kinase receptors (77-81), tyrosine kinases (14,82,83), and Ras (84, 85), but inactivates phosphatases (66,86). These signal transduction molecules act to different degrees as upstream modulators of c-Raf, confounding the issue. However, the following observations indicate that c-Raf was indeed a direct target of ROS. First, ROS scavengers prevented c-Raf activation (Fig. 6). Second, analysis of the thiol content indicated distinct changes after UV irradiation of cells (Fig. 7A). The lowest number of thiol groups was found in peroxide treated cells, under conditions that promoted kinase activation. UV-induced changes could not be further assigned to specific cysteines due to the complexity of the molecule. Third, oxidation of cysteines in the CRD appeared the most likely scenario, in analogy to other CRD-containing enzymes, such as PKC ␣, ␤2, ⑀, and (43), and the bacterial chaperone Hsp33 (48,87). Oxidizing even one cysteine residue ought to compromise the chelation of Zn 2ϩ . Indeed, both Hsp33 and PKC ␣ set free their Zn 2ϩ ions under oxidizing conditions, and so did c-Raf (see below).
In the case of the Hsp33 chaperone, increased enzymatic capacity was directly attributed to a controlled unfolding of the molecules, and since this process was reversible by reduction, it was proposed that zinc fingers served as redox-sensitive hinges (48). Activated c-Raf did not revert to the inactive form by reduction with N-acetylcysteine, in contrast to PKC, where repeated oxidation/reduction cycles turned kinase activity on or off (42). The c-Raf CRD might therefore not obey the same paradigm of a reversible redox switch. On the other hand, early events in the activation cycle might depend on oxidative opening of the molecule, whereas the secondary modifications that c-Raf experiences by interaction with GTP/Ras might lock the kinase into an active conformation inured to reducing conditions.
Using the reductionist approach, we determined that ROS, generated by UV irradiation in vitro under the same conditions as in vivo, was capable of causing the rapid release of Zn 2ϩ ions from the GST-c-Raf CRD fusion protein. We quote this observation as the fourth argument for direct effects on c-Raf. Although GST fusion proteins were not ideal tools to study protein function because of an inherent uncertainty of how well bacteria mastered folding mammalian peptides into their proper configuration, our results represent a credible in vitro correlate to the functional activation studies in vivo. This correlation was strengthened by the finding that GST-Raf fusion protein permitted substantially higher Zn 2ϩ ion release when loaded with equimolar retinol. Conversely, AR suppressed Zn 2ϩ release, whereas the equimolar mixture of retinol and AR behaved like retinoid-free protein (Fig. 7A). Thus, the biology of retinoids was mirrored by their biophysical effects on the isolated CRD.
The importance of the CRD for regulation of c-Raf kinase has been widely documented. Major contact sites enabling communication with GTP/Ras are embedded in this domain. The classical pathway, initiated by receptor phosphotyrosine kinases, is dependent on complex formation with Ras at the membrane. The alternate, redox-dependent pathway follows a similar route. First, redox activation of c-Raf required the presence of a CRD competent in Ras recognition. Substituting the PKC ␣ CIB CRD for the c-Raf CRD abolished its activation by UV irradiation but shunted this chimeric kinase to the PKC pathway, similar to results described by Avruch and colleagues (55), since it became highly responsive to phorbol ester (Fig. 8). Second, competitive inhibition of the Raf/Ras interaction by overexpression of mutant N17-Ras was another indication of dependence on Ras. After modification by ROS, generated during UV irradiation, presumably involving oxidation of cysteine residues of the CRD, relocation of Zn 2ϩ ions, and consequent conformational change, the interaction with GTP/Ras was still required. Furthermore, preventing prenylation and insertion of Ras into the membrane curtailed UV activation of c-Raf.
Our studies re-emphasize the importance of agonistic retinoids for controlled activation of the c-Raf/MAPK by oxidizing agents. They also illuminate the role of zinc as an essential dynamic component of c-Raf. Both, control by retinol and zinc, converge on the same domain in the kinase. It is interesting to note that nutritional vitamin A and zinc deficiency produce many of the same symptoms: night blindness, sterility, defective wound healing, and abnormal skin regeneration. More research is needed to understand whether the underlying common parameter involves the CRD of c-Raf and related serine/ threonine kinases. In distinction from its role as catalytic center in numerous enzymes, zinc occurs as a structural component in dozens of cytoplasmic and nuclear proteins including, notably, signal transduction molecules and transcription factors. Although "zinc fingers" are commonly thought of as rigid structures that enable in the case of transcription factors the intercalation into the DNA double helix, their role in the newly emerging redox regulation is apt to change our perception. The zinc fingers should be viewed as redox-regulated, dynamic, and reversible hinges, as proposed for the bacterial chaperone, Hsp33. Another view is that they act as redox sensors with provisions in mammalian cells for fine-tuning by different retinoids. Such a control element would allow c-Raf and its homologues to make constant adjustments in enzymatic output, as dictated by the changing redox status of cells.