Purification , Identification , and Characterization of Two GTP-binding Proteins with Molecular Weights of 25 , 000 and 21 , 000 in Human Platelet Cytosol

From the Departments of Biochemistry and TNeurosurgery, Gifu University School of Medicine, 40 Tsuhamachi, Gifu 500, the $Institute of Medical Science, University of Tokyo, 4-6-1 Shiroknnedai, Minato-ku, Tokyo 108, the $Department of Life Science, Tokyo Institute of Technology, 4259 Nagatsuta, Yokohama 227, and the 11 Department of Physiological Chemistry, Faculty of Pharmaceutical Science, University of Tokyo, Tokyo 113, Japan

a variety of receptor-mediated signal transduction systems (see Refs. 1-3 for review). These G proteins communicate between membrane receptors and the effectors including adenylate cyclase, phospholipase C and Az, K+ and Ca2+ channels, and cGMP-phosphodiesterase. They have a common heterotrimeric structure consisting of a, b, and y subunits. The a subunits which bind GTP or GDP are unique, while the / 3 and y subunits are similar, if not identical, to each other. Molecular cloning of the a subunit genes (4, 5) and cDNAs (1)(2)(3) has clearly indicated the existence of multiple forms of the subunit polypeptides.
In addition to these heterotrimeric G proteins, accumulating evidence indicates the existence of another group of GTPbinding proteins having M, values of 20,000-30,000. The ras genes have been known to encode proteins with M, of about 21,000 (ras p21) which have GTP-binding and GTPase activities (6-8). Three species of ras genes, H-, K-, and N-ras have been found in mammalian tissues (9-11). Recently, a ras-related gene, YP!ll, encoding a protein with a M, of 23,500, has been found in yeast (12). Other ras-related genes, rho (13), ral (14), and SEC4 (15)) have also been reported in ApZysia, simian B lymphocyte and yeast, respectively. A GTPbinding protein with a M, of about 21,000 required for the cholera toxin-dependent ADP-ribosylation of GB has been purified to near homogeneity from rabbit liver (16) and bovine brain (17). This protein has been designated as ADP-ribosylation factor. New species of GTP-binding protein with a M, of about 21,000, termed as G,, has been found and partially purified from human placenta (18). More recently, a GTPbinding protein with M, of 24,000 has been purified from bovine brain membrane (19). Moreover, another species of GTP-binding protein was purified from bovine brain membrane (20) and cytosol of bovine adrenal gland (21) as a substrate for botulinum ADP-ribosyltransferase, which was identified as rho gene product (20,22).
As for the platelet, several kinds of GTP-binding proteins with M, values ranging 20,000-30,000 were detected in both cytosol and membrane fractions (23,24). Recently, we have purified two GTP-binding proteins with M, of 22,000 from membrane fraction, which were biochemically and immunochemically distinguishable from ras p21 (25). Thereafter, Ohmori et al. (26) also purified a major low M, GTP-binding protein from human platelet membranes and identified as smg p21 (27). The physiological functions of these low M, GTP-binding proteins are unknown, but by analogy with the G proteins with heterotrimeric structure such as G,, Gi, Gt,

M, in H u m a n Platelet Cytosol
and Go, it can be speculated that these GTP-binding proteins with low M, values wou'ld also be involved in transmembrane signaling.
In the present investigation, we have purified two GTPbinding proteins with MI of 25,000 (c25KG) and 21,000 (c21KG) to near homogeneity from the cytosol fraction of human platelet, identified, and characterized them. Partial amino acid sequence analysis of c25KG compared with other GTP-binding proteins with low M , values indicates that c25KG is a novel GTP-binding protein distinct from other GTP-binding proteins described previously. On the other hand, the determined amino acid sequences of c21KG were completely agreed with those of rap 1 protein (28), smg p21 (29), and Kreu-1 protein (30).

DISCUSSION
Several recent discoveries indicate that the ras genes belong to a large family consisting of about 20 genes. The amino acid sequences of these proteins predicted from the cDNAs have the consensual sequence responsible for GTP-binding activity, which has been found in heterotrimeric G proteins. All low M, GTP-binding proteins (MI 20,000-30,000) so far purified reveal GTP-bindling activity. The C terminus of the encoded proteins includes a cysteine that seems to become fatty-acylated, suggesting the distribution of these proteins both in membranes and cytosol: the acylated proteins may locate in membranes whereas the proteins located in cytosol may not be acylated. In fact, several kinds of the proteins including ras, rho, and srng 25A proteins were purified from the membrane fraction and some of them such as rho (21) and smg 25A (31) proteins have been found in the cytosol fraction. In this paper, we have detected in the cytosol fraction of human platelets at least three low MI GTP-binding proteins which were separated on the MonoQ HR 5/5 column chromatography. Among thlese three major proteins, two proteins with M, values of 25,000 and 21,000, tentatively termed as c25KG and c21KG, respectively, were purified to homogeneity.
To identify the purified c25KG and c21KG, we determined the partial amino acid sequences of the nine peptides containing 76 amino acid residnes of c25KG and of the six peptides containing 58 amino acid residues of c21KG. The results indicated that c25KG is a novel low MI GTP-binding protein, although the protein is closely related to rub and smg 25 families: the sequence revealed 60% homology to rub proteins (32) or smg p25A (33). On the other hand, the partial amino acid sequences dervied jfrom c21KG were completely identical with the predicted amino acid sequence of rap1 (28), smg21 (29), or Kreu-1 (30), suggesting that this gene encodes c21KG. The amino acid sequelnce of one of the c21KG fragments, YDPTIGVDFK, is identical with those of H-ras, K-ras, and N-ras, as well as those of rupl, smg 21, and Kreu-1. This region of ras proteins appears to be involved in the interaction with effector molecule such as GTPase activating protein (34). Thus, it will be of interest to examine whether the GTPase activity of c21KG, like ras p21, is stimulated by GTPase activating protein.
It was reported that GDP is tightly associated with many GTP-binding proteins such as PT-substrates (1) and ras gene Portions of this paper (including "Experimental Procedures" and "Results," Figs. 1-12, and Table 1) are presented in miniprint at the end of this paper. Miniprint is easily read with the aid of a standard proteins (35), even in their highly purified preparations. And it is also known that the addition of 250 mM (NH&S04 to the reaction mixture caused the release of GDP probably due to the comformational change in the vicinity of guanine nucleotide-binding site (36). It was the case with c25KG and consequently the binding rate of GTPrS exogenously added apparently increased (Fig. 10B). The apparent first order rate constants were calculated to be 0.28/min and O.O16/min in the presence and absence of 250 mM (NH4)2S04, respectively, by the analysis of first order plots (data not shown). On the other hand, c25KG was found to be distinct from other low M, GTP-binding proteins in the GTPase activity. The GTPase activity of c25KG was calcualted to be 1.8 mmol of P/mol of protein/min in the absence of (NH4)804. These results suggest that the slow GTPase rate is due to the slow hydrolysis rate.
The structure of the GTP-binding domain or its vicinity in c25KG may be similar, if not identical, to that in smg p25A but evidently distinct from those in c21KG and smg p21. N -Ethylmaleimide inhibited the GTP-binding activities of c21KG and smg p21 (29) but did not affect those of c25KG and smg p25A (19). Thus, c25KG and smg p25A appear to contain cysteine residue(s) in the GTP-binding domain or its vicinity, but c21KG and smg p21 do not.
Moreover, it is noteworthy that smg p21, which is abundant in human platelet membranes (26,27), was able to be phosphorylated by CAMP-dependent protein kinase (37), whereas the same gene product c21KG in cytosol was not phosphorylated (38). Considering the fact that smg p21 and c21KG are the same gene product of Kreu-1, it seems feasible to anticipate that smg p21 in membranes may be fatty-acylated and c21KG in cytosol may not be. Accordingly, it is conceivable that the former is associated with membranes and is the preferred substrate for phosphorylation by cyclic AMP-dependent protein kinase whereas the latter is not phosphorylated.
The biological observations of Kreu-1 indicate that its product may be involved in the negative growth regulation of Kirsten sarcoma virus-transformed NIH/3T3 (30). In fact, Kreu-1 is observed to suppress the transformed phenotype of NIH/3T3 when highly expressed (30). It could be considered that c21KG and smg p21 play an important role in proliferation of some cell types. However, since the platelet represents terminal differentiation stage in hematopoietic development, the physiological significance of c21KG abundant in cytosol remains unknown in this nonproliferating cell. In mammalian cells, the effects of GTP and its analogues on phosphoinositide metabolism and exocytosis have suggested the existence of two putative G proteins, G, and G,, that are involved in phospholipase C activation and secretion processes, respectively (39). This hypothesis is supported by the findings that 1) the soluble phospholipase C from human platelets is stimulated by a low M , GTP-binding protein (29,000) (40) and 2) low M, GTP-binding proteins (18,000-24,000) may be implicated in exocytosis in the adrenal chromaffin cell (41). Thus, it is tempting to speculate that c25KG and c21KG purified from the human platelet cytosol may function in the regulation of the activity of phospholipase C and/or the process of exocytosis in the human platelet.