Calmodulin binds to and inhibits GTP binding of the ras-like GTPase Kir/Gem.

Recently, a new subfamily of Ras-related GTP-binding proteins consisting of Rad (as ssociated with iabetes), Gem (immediate early ene xpressed in itogen-stimulated T-cells), and Kir (tyrosine inase-nducible as-like) was discovered. The C terminus of these proteins contains an extension of approximately 30 amino acids not present in other members of the Ras family and which exhibits all the hallmarks typical for calmodulin (CaM)-binding domains. A peptide corresponding to the putative CaM-binding domain of the Kir/Gem protein was synthesized, and its affinity for CaM was determined by fluorescence spectrometry. Titration of dansyl-CaM with the Kir/Gem peptide gave an affinity constant of 1 nM. Furthermore, a single point mutation of the peptide, W269G, abolished this high affinity interaction. Gel-shift analysis showed that the complex formation between CaM and the Kir/Gem peptide is strictly calcium-dependent. We also demonstrate with a newly developed [32P]CaM overlay technique that full-length Kir/Gem and Rad proteins bind CaM in a Ca2+-dependent fashion. The binding of CaM to glutathione S-transferase-Kir and GST-Gem inhibited the binding of GTP to Kir/Gem significantly. These results suggest the existence of a direct link between Ca2+/CaM and growth factor signal transduction pathways at the level of small Ras-like GTPases.

Calmodulin (CaM) 1 acts as the intracellular calcium sensor that translates the Ca 2ϩ signal into a variety of cellular processes. Biochemical characterization of the interaction of CaM with its targets has defined its role as an activator of multiple Ca 2ϩ /CaM-dependent enzymes important in various cellular functions including growth and cell division (1). One of the most crucial roles of Ca 2ϩ /CaM appears to be the activation of immediate early genes. Ca 2ϩ /CaM is able to stimulate a number of transcription factors such as c-Fos, c-Jun, the cyclic AMP-response element-binding protein, and the serum response factor (2,3). Except for the known action of CaM and CaM kinase II on cyclic AMP-response element-binding protein activation by phosphorylation and the involvement of CaM and calcineurin in NF-AT-dependent gene activation (4), a direct link between Ca 2ϩ -CaM and growth factor-dependent signal transduction pathways at the level of upstream signaling molecules such as Ras, inositol 1,4,5-trisphosphate signaling molecules, and mitogen-activated protein kinases has not yet been established molecularly.
Recently, a new subfamily of Ras-related GTP-binding proteins named Rad (5), Gem (6), and Kir (7) was discovered. Rad has a 61% amino acid sequence identity to Gem (6) and Kir (7), respectively, whereas the latter 2 proteins are 98% identical to each other. The 29-kDa Rad protein is highly expressed in skeletal muscle, lung, and heart of humans. It binds GTP and has low intrinsic GTPase activity which is greatly enhanced by a GTPase-activating protein (GAP) activity present in various tissues and cell lines (8). Rad has recently been shown to interact with skeletal muscle ␤-tropomyosin and the cytoskeleton of muscle cells in a guanine nucleotide-dependent manner (9), suggesting that Rad may be involved in skeletal muscle motor function and cytoskeletal organization.
Gem, a 35-kDa GTP-binding protein was cloned from mitogen-induced human peripheral blood T cells (6). The protein was found to be phosphorylated on tyrosine residues and localized to the cytosolic face of the plasma membrane. Deregulated Gem expression prevented proliferation of normal and transformed 3T3 cells, suggesting that Gem is a regulatory protein, possibly participating in receptor-mediated signal transduction at the plasma membrane.
Kir has been found to be expressed in cells transformed by ABL tyrosine kinase oncogenes (7). The Kir gene encodes a protein of 33 kDa that exhibits guanine nucleotide-binding activity. Kir was cloned by differential screening of genes present in fully malignant versus growth factor-independent cell lines expressing wild-type or mutant forms of BCR/ABL. Kir expression in BCR/ABL and v-ABL transformed B-cells renders these cells highly tumorigenic and metastatic, indicating that Kir could be involved in processes of invasion or metastasis in mammalian cells (7,10).
Interestingly, unlike most members of the Ras superfamily, Kir/Gem and Rad do not possess the lipid modification motifs CAAX, CXC, or CC (11). However, the C terminus of the Kir/ Gem and Rad proteins contains an approximately 30-amino acid extension not present in other members of the Ras family. In this study, we show that a synthetic peptide corresponding to the C terminus of the Kir/Gem protein is able to bind CaM with high affinity. Replacement of a single residue, W269G, in the peptide abolished this high affinity interaction indicating that tryptophan 269 is important for complex formation. Furthermore, binding of CaM to GST-Kir/GST-Gem inhibits GTP binding significantly.
Fluorescence Measurements-Fluorescence measurements were performed with a SPEX Fluorolog 1680 (Metuchen, NJ) double-wavelength spectrophotometer connected to a SpectrAcq Control Module and run on a Datamax version 1.03 software (Jobin Yvon/Spex) as described by Anagli et al. (12). The concentration of the peptide was determined by amino acid analysis. The affinity constant of the Kir/Gem peptide was calculated according to the method of Stinson and Holbrook (13).
Gel Electrophoresis in the Presence of Urea-Gel-shift analysis of CaM-Kir/Gem complexes were performed according to Ref. 14.
Overlay Assay-Generation of the GST-CaM construct, labeling of the GST-CaM fusion protein, and CaM-binding analysis were carried out according to Ref. 15. Standard molecular biology techniques were according to Ref. 16.
Guanine Nucleotide Binding-GTP binding to GST-Kir/GST-Gem was determined using a nitrocellulose filtration assay (5). GST-Kir/ GST-Gem (20 pmol) was incubated in an exchange buffer containing 50 mM Tris-HCl, pH 7.4, 1 mM DTT, 1 mg/ml bovine serum albumin, 1 mM MgCl 2 , and 3 Ci of [␣-32 P]GTP (Amersham Zü rich, specific activity 3000 Ci/mmol) with end to end rotation at room temperature. At each time point, aliquots of 40 l were withdrawn in duplicate and directly filtered through BA 85 nitrocellulose filters (Schleicher & Schuell) followed by washing with 10 ml of filtration buffer (50 mM Tris-HCl, pH 7.4, 0.1 mM DTT, 1 mM MgCl 2 . The radioactivity remaining on the filters was determined by scintillation counting.

RESULTS AND DISCUSSION
To investigate the possibility of a direct link between calcium and the Ras signal transduction pathways, the amino acid sequences of several recently discovered proteins implicated in signal transduction events were compared with those of known high affinity CaM-binding proteins. Specifically we have looked for amino acid sequences, 20 -24 residues in length, which match the described structural features of known CaM-binding sequences (17,18). We discovered that the C-terminal extensions of the Kir/Gem and Rad proteins exhibit all the hallmarks typical for CaM-binding peptides. Fig. 1 shows the putative CaM-binding C-terminal sequences of Kir/Gem and Rad and for comparison the well characterized CaM-binding domains of CaM kinase II, smMLCK, skMLCK, and the Ca 2ϩ -pump are aligned.
We first asked whether the putative CaM-binding domain of Kir/Gem indeed is able to bind CaM. Therefore, the peptide KARRFWGKIVAKNNKNMAFKLKSKS corresponding to residues 264 to 288 of the reported Kir sequence or residues 265 to 289 of the reported Gem gene product was synthesized. We have introduced a single amino acid substitution, glycine for tryptophan at position 269 of the wild type Kir peptide (M1) to obtain a negative control peptide. Nondenaturing gel electrophoresis in the presence of 4 M urea revealed that the Kir/Gem peptide bound to CaM and induced a shift of the complex in the presence of 0.1 mM CaCl 2 ( Fig. 2A). Addition of 2 mM EGTA abolished completely complex formation indicating that the interaction of Kir/Gem with CaM was Ca 2ϩ -dependent (Fig.  2B). Maximum shift was observed when the CaM/peptide ratio was 1:1. A molar excess of peptide did not result in the appearance of new bands, suggesting that CaM interacts with the Kir/Gem peptide with a 1:1 stoichiometry. No significant shift was observed when the mutant peptide M1 was analyzed (data not shown).
The interaction between the Kir/Gem peptide and CaM was also studied using fluorescence spectroscopy. As shown in Fig.  3A, titration of dansylcalmodulin with the Kir/Gem peptide saturated in the nanomolar concentration range in the presence of 0.1 mM CaCl 2 . The maximum in the emission spectrum of the binding peptide underwent a shift to lower wavelength and an increase in intensity upon complex formation with dansylcalmodulin indicating protein/peptide interaction. No increase in intensity and shift of the spectrum was observed in control experiments (peptide alone plus calcium, dansylcalmodulin alone plus calcium, data not shown). The CaM-peptide interaction was Ca 2ϩ -dependent since addition of 1 mM EGTA reversed both the shift and intensity of the spectrum (Fig. 3B). Interestingly, the mutant Kir/Gem peptide (M1) is completely inactive in CaM binding under the same experimental conditions. No shift to lower wavelength and no increase in fluorescence intensity could be observed (Fig. 3A). From this result we can conclude first that the tryptophan residue at position 269 of the Kir/Gem peptide is important for the high affinity interaction of CaM with Kir/Gem and second that the tryptophan residue at position 269 directly participates in complex formation. Similar results were also obtained with other mutant peptides (data not shown).
The affinity constant for the formation of the complex between CaM and the Kir/Gem peptide was determined from titration experiments with dansylcalmodulin. The Kir/Gem peptide bound CaM with high affinity (1 nM) (Fig. 4). The plot of relative fluorescence versus the molar ratio of Kir/Gem to dansylcalmodulin as shown in Fig. 5 indicates Ca 2ϩ dependence of the interaction and a stoichiometry of 1:1. This is in agreement with the results obtained using urea gel-shift analysis (Fig. 2). The formation of the CaM/peptide complex was also monitored at different free calcium concentrations (Fig. 6). Half-maximal binding was obtained with less than 10 M free calcium. In summary, using both methods, gel-shift analysis and fluorescence spectrometry, we have shown that the Kir/ Gem peptide binds CaM in a strictly Ca 2ϩ -dependent fashion and with a stoichiometry of 1:1. Furthermore, the binding of the Kir/Gem peptide to CaM is abolished when the single tryptophan residue in the Kir/Gem peptide is replaced by a glycine residue.
Additionally, we addressed the question whether the full size Kir/Gem and Rad proteins retain the ability to bind CaM. A rapid alternative method for detecting CaM-binding proteins with a 32 P-labeled CaM probe generated as a GST-fusion protein was used to detect GST-Kir/Gem and GST-Rad in an overlay assay. Fig. 7 demonstrates that full-length Kir/Gem and Rad proteins are easily detected by the 32 P-labeled GST-CaM probe. As a control, GST alone does not bind to either Kir/Gem or Rad (data not shown). Furthermore, Fig. 7 shows that a large amount of the three proteins is associated with the pellet fraction of bacterial extracts. In addition to GST-Gem (calculated molecular mass of 60 kDa), a higher molecular mass species (over 150 kDa), which also gives a strong signal, is present in the total lysate and the pellet fraction. Interestingly this band is not present in GST-Kir and GST-Rad bacterial lysates, suggesting a Gem-specific effect.
Since Kir, Gem, and Rad have been described as GTP-binding proteins (5-8), we investigated whether CaM binding affects GTP binding to Kir/Gem. As can be seen from Fig. 8, Kir/Gem binds GTP less efficiently in the presence of CaM. We can observe an inhibition of GTP binding of about 40%. This result would suggest that CaM alone is not sufficient to suppress GTP binding, but could affect the level of bound nucleotides through an unknown mechanism. No GAP or GRF for Kir/Gem and Rad have been described, but CaM could play a regulatory role in conjunction with effector molecules such as GAP's or GRF's specific for Kir/Gem.
The C terminus of the Gem gene product was shown to be important for intracellular localization (6). A C-terminal truncated form of Gem redistributes from the plasma membrane to the nucleus, whereas N-terminal truncated Gem resides mostly at the plasma membrane. CaM binding to the C terminus of Excitation was performed at 340 nm. The relative fluorescence intensities are plotted against the ratio of total concentration of the Kir/Gem peptide and the total concentration of dansylcalmodulin, as given by one representative titration experiment.
Kir/Gem and Rad proteins might modulate or regulate their intracellular localization. CaM might therefore be involved in a novel mechanism regulating the attachment to or release from the plasma membrane of these Ras-like GTPases.
Evidence for a direct link between Ca 2ϩ -CaM and the Ras signal transduction pathway was reported for the neuronal exchange factor RasGRF as a signal mediator and CaM target by Farnsworth et al. (19). Activation is mediated in a Ca 2ϩ -dependent fashion by CaM binding to RasGRF. Recently, a novel GTPase-activating protein, called IQGAP, has been reported (20,21). Interestingly, IQGAP also binds CaM (21). The existence of a CaM-binding domain in Ras-like GTPases such as Kir/Gem (this study) and in effector molecules of Ras and Ras-like G-proteins such as RasGRF and IQGAP is intriguing. One can speculate that CaM-binding motifs (similarly to the roles of SH2, SH3, PH, and others) may represent yet another important module regulating protein-protein interactions in signal transduction pathways. The physiological relevance of such interactions, however, has yet to be investigated.
That Ras-like proteins of the Kir/Gem and Rad type could play an important role in growth-related signaling has recently been demonstrated by the finding that Kir functions upstream of the ste20 kinase in activating mitogen-activated protein kinase cascades in yeast (10). Since the function of Kir to induce the pseudohyphae formation phenotype in yeast is abolished when Kir lacks the C-terminal domain Ca 2ϩ -CaM may be involved in the regulation of biological functions of small Rasrelated proteins.