The N-terminal moiety of CDC25(Mm), a GDP/GTP exchange factor of Ras proteins, controls the activity of the catalytic domain. Modulation by calmodulin and calpain.

This work describes the in vitro properties of full-length CDC25Mm (1262 amino acid residues), a GDP/GTP exchange factor (GEF) of H-ras p21. CDC25Mm, isolated as a recombinant protein in Escherichia coli and purified by various chromatographic methods, could stimulate the H-ras p21·GDP dissociation rate; however, its specific activity was 25 times lower than that of the isolated catalytic domain comprising the last C-terminal 285 residues (C-CDC25Mm285) and 5 times lower than the activity of the C-terminal half-molecule (631 residues). This reveals a negative regulation of the catalytic domain by other domains of the molecule. Accordingly, the GEF activity of CDC25Mm was increased severalfold by the Ca2+-dependent protease calpain that cleaves around a PEST-like region (residues 798-853), producing C-terminal fragments of 43-56 kDa. In agreement with the presence of an IQ motif on CDC25Mm (residues 202-229), calmodulin interacted functionally with the exchange factor. Depending on the calmodulin concentration an inhibition up to 50% of the CDC25Mm-induced nucleotide exchange activity on H-ras p21 was observed, an effect requiring Ca2+ ions. Calmodulin also inhibited C-CDC25Mm285 but with a ∼100 times higher IC50 than in the case of CDC25Mm (∼10 μM versus 0.1 μM, respectively). Together, these results emphasize the role of the other domains of CDC25Mm in controlling the activity of the catalytic domain and support the involvement of calmodulin and calpain in the in vivo regulation of the CDC25Mm activity.

in rat (p140-rasGRF) (4) and human (H-GRF) (5,6). These rasGEFs have been described to be specific for the central nervous system (4 -9). Some evidence has also been reported for the existence of full-length and truncated forms of these exchange factors in other tissues (10,11). Experiments in vivo suggest that the upstream connection of this GEF involves G-protein-coupled receptors (9,12,13) and not hormone-receptor-bound tyrosine kinases via the adaptor protein GRB2, as has been found for SOS, a ubiquitous rasGEF (14 -16). CDC25 Mm contains in the N-terminal moiety two domains of pleckstrin homology (PH1 and PH2), one of DBL homology (DH) and a coiled-coil region that follows the PH1 domain (cf. Ref. 17). PH domains are frequently present in signaling proteins and represent regions of interactions with specific ligands such as the ␤␥-subunits of heterotrimeric G-proteins (18,19) and phospholipids (20). Coiled-coils are specific tertiary structures involved in protein interactions (21) and the DH is a domain sharing similarity with a GDP/GTP exchange factor of members of the Rho family (2,4,22,23). Farnsworth et al. (24) reported that in vivo the activity of the homologous p140-rasGRF from rat brain is enhanced by raising the calcium concentration, an effect associated with the binding of calmodulin, and that p140-rasGRF and calmodulin form a stable complex. A direct action of calmodulin was supported by the presence in the N-terminal region of CDC25 Mm of an IQ domain, a sequence frequently found in proteins interacting with calmodulin (25,26). Very recent experiments in vivo have indicated that PH1, coiled-coil and IQ domains act cooperatively to facilitate the activation of p140-rasGRF by calcium (17). Moreover, the presence in p140-rasGRF of two adjacent PEST sequences has suggested potential cleavage by the Ca 2ϩ -dependent protease calpain (26). Concerning in vitro properties, whereas C-terminal catalytic fragments spanning 256 to 488 amino acid residues have been biochemically characterized (3,5,27,28), little is known about the functional properties in vitro of the full-length molecule comprising 1262 (CDC25 Mm ) or 1244 (p140-rasGRF) amino acid residues, of which the purification has yet to be reported. Therefore, with the aim at deepening our knowledge of the mechanisms controlling the activation of H-ras p21, we have produced and purified the full-length CDC25 Mm as recombinant protein in Escherichia coli and characterized its properties under well defined conditions in vitro. The purified GEF shows a specific activity much lower than its isolated catalytic domain or the C-terminal half-molecule; it is inhibited by calmodulin and is specifically cleaved by calpain.  1 The abbreviations used are: CDC25 Mm , full-length murine CDC25 Mm gene product of 1262 amino acid residues; C-CDC25 Mm285 and C-CDC25 Mm631 , C-terminal fragments of CDC25 Mm comprising the last 285 and 631 amino acid residues, respectively; MBP, maltosebinding protein; GEF, guanine nucleotide exchange factor; GRF, guanine nucleotide releasing factor; ME, 2-mercaptoethanol; PAGE, polyacrylamide gel electrophoresis; GST, glutathione S-transferase. been replaced by a SalI site. A BamHI-SalI fragment, containing the total CDC25 Mm gene, was cloned in BamHI-SalI of pMAL-c2 (New England Biolabs) and expressed in E. coli strain SCS1 as N-terminal fusion with the maltose-binding-protein (MBP), comprising a Xa-specific cleavage site. Cell cultures (1.5 liters) were grown at 30°C in LB medium containing 50 g/ml ampicillin. The induction was started with 0.1 mM isopropyl-␤-D-thiogalactopyranoside at a cell density of 0.2 A 600 units and continued at 24°C to a density of 1.5 A 600 units. After centrifugation (7,000 ϫ g for 10 min), the resuspended pellet was sonicated 5 times for 15 s in buffer A (25 mM Tris-HCl, pH 7.5, 50 mM NaCl, 7 mM ME) containing 10% glycerol, 1 mM EDTA, and 1 mM Pefablock-SC and centrifuged for 20 min at 25,000 ϫ g at 4°C, a temperature at which all the subsequent purification steps were carried out. The supernatant was loaded on a 6-ml ResourceQ column (fast protein liquid chromatography system, Pharmacia Biotech Inc.) equilibrated with buffer A and eluted with the same buffer at a flow rate of 3 ml/min. Unlike the bulk of E. coli proteins, the largest portion of MBP-CDC25 Mm was not retained on the resin. The nonretained active fractions were mixed with 10 ml of amylose-resin (New England Biolabs) equilibrated in buffer A and gently shaken for 30 min. After centrifugation at low speed (2,500 ϫ g for 2 min), the supernatant was discarded, and the resin mixed again with 50 ml of buffer A was centrifuged. This step was repeated four times. Finally, MBP-CDC25 Mm was removed from the amylose resin by two washes with 10 ml of buffer A containing 10 mM maltose. After centrifugation, the combined supernatants were passed on a HiTrap heparin column of 5 ml (Pharmacia) that was step-eluted with 250, 400, 600, and 1000 mM NaCl solutions in 25 mM Tris-HCl, pH 7.5, and 7 mM ME (flow rate: 5 ml/min) under control of the fast protein liquid chromatography system. MBP-CDC25 Mm emerged at 600 mM NaCl. After dialysis against buffer A plus 50% glycerol, the purest fractions on SDS-PAGE, were stored at Ϫ20°C. Their activity was stable for at least 6 months.

Production and Purification of CDC25
Assay for GEF Activity-The dissociation rates of the p21⅐[ 3 H]GDP complexes were measured kinetically at 30°C after addition of a 500fold excess of unlabeled nucleotide using the nitrocellulose binding assay. The labeled p21⅐ were filtered through nitrocellulose discs (Sartorius 11306, 0.45 m), washed twice with 3 ml of ice-cold 50 mM Tris-HCl, pH 7.5, 10 mM MgCl 2 , and 100 mM NH 4 Cl, and the retained p21-bound radioactivity was measured in a liquid scintillation counter LKB/Pharmacia, model Wallac 1410.
Other Materials and Methods-C-CDC25 Mm285 and human c-H-ras p21 were isolated and purified as described (27, 29). C-CDC25 Mm631 was produced in E. coli, as GST fusion and purified by affinity chromatography on glutathione-Sepharose, under the same conditions as reported for C-CDC25 Mm285 by Jacquet et al. (27). This was followed by chromatography on HiTrap Heparin column (see above). Stepwise elution of C-CDC25 Mm631 took place at 400 mM NaCl. Calmodulin and calpain were purchased from Sigma. SDS-PAGE was carried out using a 10%/ 0.25% acrylamide/bisacrylamide gel and stained with Coomassie Blue. The anti-C-CDC25 Mm285 antibodies were produced in rabbit, and those anti-MBP were purchased from New England Biolabs. Protein concentration was determined by the Bio-Rad assay, using bovine serum albumin as a standard. The concentration of p21 was checked by [ 3 H]GDP binding. To determine the percentage of protein components, Coomassie Blue-stained gels were analyzed with the Apple scanner system, and the results were evaluated with the Scan Analysis software.

RESULTS
Purification of CDC25 Mm and C-CDC25 Mm631 -Cloning of the full-length CDC25 Mm gene in pGEX resulted in low expression and little soluble product, whereas the use of a pMAL vector allowed the production of ϳ1 mg of CDC25 Mm /1 liter of cell culture at a A 600 cell density of 1.5 units. More than 60% of the produced CDC25 Mm remained soluble after a 20-min cen-trifugation at 25,000 ϫ g. The temperature of induction was important; incubation at 24°C gave higher concentrations of soluble CDC25 Mm than at 37°C. On SDS-PAGE the MBPfused product showed the expected apparent molecular mass of 188 kDa and corresponded, after purification, to the slowest migrating band (ϳ50% of the total proteins) (Fig. 1A, lane 4). In the purified preparations, Western blot with anti-C-CDC25 Mm285 antibodies revealed only one band corresponding to the molecular mass of the MBP-fused full-length protein (188 kDa) (Fig. 1B, lane 2) while most other SDS-PAGE bands were responsive to anti-MBP antibodies (Fig. 1B, lane 1). Since no proteolysis was detectable in the cell extract, these bands consist of N-terminal incomplete translational products of MBP-CDC25 Mm . Accordingly, a second passage on amylose resin gave the same gel pattern without any further purification. Also the use of other chromatographic techniques such as ionic exchange (ResourceQ, Pharmacia, at pH from 7.0 to 9.5) and hydrophobic resins (Pharmacia kit), or of hydroxylapatite and filtration methods (Superdex, Pharmacia and Microcon-100, Amicon) did not improve purification. Together, these properties show that our CDC25 Mm preparations were not contaminated by C-terminal fragments and contained little, if any, E. coli proteins. Protein concentration by means of Aquacide II or Amicon ultrafiltration was avoided, since it caused loss of activity, likely due to aggregation. Cleavage of the fused MBP by factor Xa led to degradation of CDC25 Mm ; therefore, we have used the fused protein. C-CDC25 Mm631 was obtained Ͼ90% pure as GST-fused protein. C-CDC25 Mm285 was homogeneous on SDS-PAGE, as reported (27).
Comparison of the Activities of Full-length CDC25 Mm , C-CDC25 Mm631 , and C-CDC25 Mm285 -Experiments in vivo suggest that CDC25 Mm has a constitutive GEF activity (9). In line with this, the [ 3 H]GDP/GDP exchange rate of p21⅐[ 3 H]GDP in vitro was enhanced by the purified full-length CDC25 Mm (Fig. 2A). The stimulation of the p21⅐GDP dissociation rate constant increased linearly with increasing the CDC25 Mm concentration (Fig. 2B). Fig. 2C shows that the C-terminal halfmolecule C-CDC25 Mm631 displayed a constitutive GEF activity several times higher than the full-length molecule. Also in this case, increasing concentrations of C-CDC25 Mm631 enhanced linearly the GDP/GDP exchange activity of H-ras p21 (Fig. 2D). Similar results were obtained by determining the GDP/GTP exchange rate (not shown).
The activity of CDC25 Mm and C-CDC25 Mm631 was then compared with that of the C-terminal fragment C-CDC25 Mm285 , the best studied catalytic fragment of CDC25 Mm (3,27), slightly longer than the shortest catalytic domain isolated so far (last 256 C-terminal residues, ref 28). C-CDC25 Mm285 was by far the most active of the three rasGEF forms. At a concentration of 0.2 M, the activities of CDC25 Mm and C-CDC25 Mm631 were 4.4 and 23%, respectively, that of C-CDC25 Mm285 (Fig. 3). This reveals a negative influence on the activity of the C-terminal domain by other domains of CDC25 Mm .
CDC25 Mm is specifically cleaved by calpain around a PEST region-Cheney and Mooseker (26) identified in the homologous p140-rasGRF from rat two adjacent PEST sequences located between residues 792 and 836. PEST sequences have been proposed to confer rapid degradation to numerous proteins via the Ca 2ϩ -dependent protease calpain (30,31). To determine whether this was also the case for CDC25 Mm , the action of calpain was examined by determining its proteolytic effect and the possible consequences on the GEF activity. On Western blot (Fig. 4A), we observed with anti-C-CDC25 Mm285 antibodies that digestion of CDC25 Mm by calpain induced the appearance of two to four C-terminal fragments with a molecular mass between 43 and 56 kDa. Digestion of CDC25 Mm at 30°C proceeded rapidly: Ͼ90% in 2 min with 1 M calpain and 50% in 30 min with 0.05 M calpain. No cleavage of C-CDC25 Mm285 by calpain could be detected.
As shown in Fig. 4B, calpain-treated CDC25 Mm showed a ras GEF activity ϳ3 times as strong as that of the untreated CDC25 Mm . This effect, clearly dependent on the production of C-terminal CDC25 Mm fragments with constitutive activities higher than the activity of the full-length molecule, further confirms the negative influence of the N-terminal moiety on the activity of the C-terminal catalytic domain and suggests a regulatory role by proteases on the activity of this rasGEF.
Calmodulin Action on CDC25 Mm Activity-Farnsworth et al. (24) observed that H-ras p21 activation by p140-rasGRF in vivo was enhanced by raising the concentration of Ca 2ϩ , a mechanism mediated by calmodulin, since this ubiquitous protein, known to be involved in cellular processes controlled by Ca 2ϩdependent signaling (cf. Refs. 32 and 33), was coprecipitated with p140-rasGRF. The fact that CDC25 Mm has the same recognition sequence for calmodulin (IQ motif, residues 202-229), as found in p140-rasGRF (residues 198 -225T) (26) suggests a productive interaction also between CDC25 Mm and calmodulin. As shown in Fig. 5, in our in vitro system Ca 2ϩ -calmodulin was found to inhibit the rasGEF activity of CDC25 Mm . With increasing concentrations of calmodulin, the inhibition leveled at ϳ50% the rasGEF activity, the concentration of calmodulin inducing half maximum inhibition (IC 50 ) being ϳ0.1 M (Fig. 5,  inset).
In experiments not shown, we have observed that the treatment of CDC25 Mm by calpain leading to N-terminal and Cterminal-truncated fragments abolished the inhibition of the GEF activity induced by 1 M calmodulin. Thus, one can conclude that the inhibitory effect of calmodulin on the catalytic activity of the C-terminal domain is due to an intramolecular mechanism originating from the N-terminal moiety.
As shown in Fig. 6, the omission of Ca 2ϩ and the addition of 0.1 mM EGTA led to a progressive slight decrease of the calmodulin inhibition on the CDC25 Mm activity, the obtained values displaying a marked variability. Only at 1 mM EGTA, the calmodulin effect was completely abolished.
Ca 2ϩ -calmodulin also inhibited the GEF activity of C-CDC25 Mm285 , but at a concentration about 100 times higher than in the case of the full-length molecule (IC 50 ϭ ϳ10 versus 0.1 M, respectively) (Fig. 7). Calmodulin had no effect on the intrinsic GDP exchange activity of H-ras p21.

DISCUSSION
The biochemical characterization of purified full-length CDC25 Mm shows that its noncatalytic moiety, comprising two PH, one DH, one coiled-coil and one IQ domain, has a regulatory function on the GEF activity of CDC25 Mm . In fact, the specific activity of the full-length protein is ϳ4% that of the isolated catalytic C-terminal fragment of 285 residues. The observation that the C-terminal half-molecule shows an activity, whose extent lies between the activity of the full-length CDC25 Mm and that of C-CDC25 Mm285 , further confirms a regulatory function of the noncatalytic moiety of the molecule and supports the existence in the cell of mechanisms activating this GEF. This is in agreement with experiments in vivo showing that serum stimulation of NIH3T3 transformants leads to an increase in Ras⅐GTP only in cells expressing the entire CDC25 Mm but not in cells expressing smaller C-terminal forms of CDC25 Mm (9,22). Thus, the N-terminal region is essential for the response to serum. In line with this, is also the observation that the C-terminal truncated CDC25 Mm lacking the last 600 amino acids behaves as dominant negative to the response to serum of the full-length molecule (9).
The specific cleavage by calpain around the PEST region increases the GEF activity due to the production of C-terminal fragments. The observation that the induced activity is of the same range as the activities of C-CDC25 Mm631 and the p140-rasGRF C-terminal fragment of 456 residues (4) suggests that most noncatalytic domains participate in regulating the activity of the catalytic C-terminal conserved region. This indicates that the signal activating the productive interaction between the CDC25 Mm C-terminal region and H-ras p21 evokes a global effect on the CDC25 Mm molecule.
It is known that the activity of calpain can be coordinated with that of calmodulin, as shown by the frequent presence on calmodulin-binding proteins of a PEST motif and the functional relationship between calpain action and calmodulin binding site (cf. Ref. 34). In a large number of proteins (cf. Refs. 30 and 31), PEST sequences appear to represent a signal for rapid degradation by non-ubiquitin-mediated proteolysis that can involve calpain. Interestingly, in the mouse embryo brain a 58-kDa form of CDC25 Mm was described, whereas in the more slowly metabolizing adult brain only the full-length 140-kDa protein could be detected (7). In the human brain, besides the full-length 140-kDa H-GRF, shorter forms of 43-50 kDa were reported (5) that might also be present in tissues other than brain (11). The results of our work show that C-terminal CDC25 Mm fragments of 43-56 kDa can originate from the action of calpain. Therefore, as reported in this work, activation of the GEF activity by calpain appears to be mediated by C-terminal fragments that are more active than the full-length molecule. In vivo, an eventual action of calpain could contribute to modulate the activity of CDC25 Mm . In this context, it is worth mentioning that the expression of C-CDC25 Mm285 in fibroblasts was found to enhance the GDP/GTP exchange activity on H-ras p21 and the tumor formation in the nude mice (35). Moreover, the homologous Saccharomyces cerevisiae ras-GEF Cdc25p, containing a cyclin destruction box, has been described to display a rapid metabolism (half-life time, ϳ20 min) due to proteolytic degradation (36). Whether proteases also act on the mammalian rasGEF in vivo remains to be determined, since the half-life of CDC25 Mm in neuronal cells is still unknown.
From our results, Ca 2ϩ -calmodulin interacts efficiently with CDC25 Mm , inhibiting the rasGEF activity with a IC 50 of 0.1 M, an effect dependent on Ca 2ϩ , corresponding approximately to an equimolar ratio between calmodulin and CDC25 Mm . The course of inhibition with increasing amounts of calmodulin suggests a noncompetitive effect on the GEF activity of CDC25 Mm . Hence, binding of calmodulin and p21 to CDC25 Mm should concern distinct sites. The weak interaction of calmodulin with the catalytic domain of CDC25 Mm suggests the possibility that in the three-dimensional conformation the IQ sequence at the N-terminal region (residues 202-229) and the C-terminal catalytic domain are vicinal.
Calmodulin is in general known to activate calmodulin-binding proteins. However, in some cases such as calmodulin-kinase II, calmodulin-kinase IV, myosin light chain kinase and calcineurin, incubation with Ca 2ϩ /calmodulin has been shown to induce inactivation (cf. Refs. 37 and 38). It is interesting to mention that calmodulin-kinase IV interacts with calmodulin with high affinity, comparable to that of CDC25 Mm (37), differently from calmodulin-kinase II which is inhibited only by high concentration of calmodulin (38).
Noteworthy is the relatively high concentration of EGTA required for abolishing the Ca 2ϩ -calmodulin-dependent effect. Several recent studies have indicated that the conformation of the N-terminal and C-terminal calcium binding helix-loop-helix motifs of calmodulin depends on whether calcium-is bound or not (cf. Ref. 39). Since the affinity of EGTA for calcium is several orders of magnitude higher than that of calmodulin, this finding suggests that, when occupied, the Ca 2ϩ -binding site of calmodulin is little accessible to the chelating agent.
From experiments in vivo (24), one would rather expect that Ca 2ϩ -calmodulin activates CDC25 Mm . This difference with our results in vitro suggests the existence of other components affecting the action of calmodulin on CDC25 Mm in the cell implying a more complex calmodulin-regulated mechanism. The same authors (24) report that calmodulin-bound p140-rasGRF isolated from stimulated cells and calmodulin-free p140-rasGRF isolated from unstimulated cells display the same GEF activity. The different experimental conditions may be the reason for the discrepancy between these and our results.
In conclusion, our work supports a regulatory role of the N-terminal moiety of CDC25 Mm on its catalytic domain and suggests that calmodulin and calpain act as modulators of the CDC25 Mm activity in the cell. Very recently, Mattingly and Macara (13) have reported on the basis of experiments in vivo a phosphorylation-dependent activation of CDC25 Mm exchange factor by muscarinic receptors and G-protein ␤␥-subunits. The relationship between this effect and the calmodulin-dependent regulation remains an open question.