The Lymphoma Transmembrane Glycoprotein GP85 (CD44) Is a Novel Guanine Nucleotide-binding Protein Which Regulates GP85 (CD44)-Ankyrin Interaction*

In this study, we have used photoaffinity labeling by [32P]azido-GTP as well as [32P]ADP-ribosylation by pertussis toxin (PT) and cholera toxin (CT) to identify GTP-binding proteins associated with mouse T-lym-phoma plasma membranes. Our results indicate that GP85 (CD44) can be photoaffinity labeled by [32P] azido-GTP and [32P]ADP-ribosylated by both PT and CT. Using purified GP85 (CD44) obtained by Triton X- 100 extraction, wheat germ agglutinin-Sepharose, and anti-GP85 (CD44) antibody affinity chromatographies, we have further characterized GP85 (CD44) as a GTP-binding protein. GPS5 (CD44) is found to bind guanosine 5’-3-0-(thio)triphosphate (GTP-yS) in a time- and dose-dependent manner with a dissociation constant of 0.83 nM. Importantly, GP85 (CD44) appears to display a GTPase activity which hydrolyzes [-y-32P]GTP at a rate of 0.011 mol of Pi released/mol of GP85 (CD44)/min. This GTPase activity can be readily inhibited by PT- or CT-mediated ribosylation of GP85 (CD44). Most interestingly, GTP binding significantly enhances the interaction of purified

At the present time, three main classes of G-proteins have been well characterized. The first class includes G-proteins that direct ribosomal protein synthesis, exemplified by elongation factor T u (5)(6)(7). The second class includes the receptor coupled G-proteins which transmit signals from ligand-activated receptors to the effectors downstream (8)(9)(10). These Gproteins are heterotrimers consisting of a, (3, and y subunits. The a subunit binds GTP and exhibits GTPase activity. It is attached to the plasma membrane via (3 and y subunits (11). The binding of GTP to the a subunit dissociates the apy complex from the receptor and a-GTP from (37. a-GTP then activates effector enzymes such as adenyl cyclase or polyphosphoinositide phospholipase C and ion channels that gate Na+, K', or regulate MgZ+ transport (12)(13)(14)(15)(16). The regulation of the diverse functions of these effector molecules is primarily governed by structural and functional differences among a subunits (9,10). Identification of various a subunits frequently involves the use of bacterial toxins that ADP-ribosylate the a subunits, which inhibits their GTPase activity resulting in either the activation or inhibition G-protein functions (10,17). Based on the sensitivity of a subunits to bacterial toxins, the receptor-coupled G-proteins are classified as G,, Gi, Go etc. (2,3). The third class of G-proteins includes the low molecular mass  proteins that are involved in a number of important regulatory functions such as cell proliferation, differentiation, and vesicular transport (18). These proteins include ras, ral, rac, rho, rap, and rub gene families in mammalian cells and SEC 4, YPT1, ARF, and SARl gene products in yeast (19)(20)(21)(22).
One important feature shared by all three classes of Gproteins is that they bind GTP and have an intrinsic GTPase activity. The second important feature that these G-proteins share is that they function in various cellular pathways by associating with plasma membrane or the organelle membranes. For example, the a subunit attaches to the plasma membrane through the 07 complex or fatty acylation of the N terminus (11,23). The small G-proteins attach to the plasma membrane or organelle membranes through polyisoprenylation, carboxymethylation, and in some cases palmitoylation (24,25). It has been demonstrated that the membrane attachment of G-proteins is crucial for their function as signal-transducing molecules (26)(27)(28). However, most G-proteins with the exception of the signal recognition particle a contain a membrane spanning domain, and none of the known transmembrane receptor proteins binds GTP directly (2,4,29,30). The 85-kDa mouse lymphocyte transmembrane glycoprotein GP85 (Pgp-1) is a homolog of human lymphocyte homing receptor, CD44 (GP9OHermeS antigen, H-CAM (homing-cellular adhesion molecule)) (31)(32)(33). GP85 has also been shown to be an important accessory molecule in mouse lym-22073 by guest on March 20, 2020 http://www.jbc.org/ Downloaded from Regulation of GP85 (CD44)-Ankyrin Interaction by GTP phocyte homing along with the MEL-14 antigen (34,35). GP85 (CD44) is a well known T-cell differentiation antigen that has also been detected on many cell types including Bcells, macrophages, granulocytes, erythrocytes, fibroblasts, epithelial cells, and most recently on the endothelial cells (36)(37)(38)(39)(40)(41). In addition to lymphocyte homing, GP85 (CD44) is also involved in Tand B-cell adhesion, cell aggregation, and Tcell proliferation (42)(43)(44)(45). Recent work has demonstrated that GP85 (CD44) binds hyaluronic acid and collagen type I (46)(47)(48)(49). Therefore, GP85 (CD44) is most likely identical to the previously identified hyaluronic acid receptor and ECMR 111, the collagen types I and VI receptor (48,49). In addition, our laboratory has shown that the cytoplasmic domain of GP85 (CD44) binds directly to ankyrin, a cytoskeletal protein that links plasma membrane to the underlying cytoskeleton (50,51). Furthermore, GP85 (CD44) can be phosphorylated by protein kinase C and fatty acylated by palmitic acid (52,53). Both of these post-translational modifications appear to enhance the binding of GP85 (CD44) to ankyrin significantly (52,53).
In this manuscript, we report that GP85 (CD44) is a GTPbinding protein (i.e. displaying GTP binding and GTPase activity). Furthermore, it can be ADP-ribosylated by both pertussis toxin (PT) and cholera toxin (CT). Most interestingly, GTP binding to GP85 (CD44) enhances its interaction with ankyrin, and ADP-ribosylation by PT or CT abolishes the GTP-induced increase in CD44-ankyrin binding.

Cell Culture
The murine T-lymphoma cell line BW5147 (an AKR/J mouse lymphoma) was grown in Dulbecco's modified Eagle's medium supplemented with 10% horse serum at 37 "C in 5% COS and 95% air.

Lymphoma Plasma Membrane Isolation
Plasma membranes were isolated from mouse T-lymphoma cells as described previously (54).
Purification of GP85 (CD44) GP85 (CD44) was purified from lymphoma plasma membranes as described previously (47) with the following modifications. Lymphoma plasma membranes (1 mg of protein) were solubilized in a buffer containing 20 mM Tris-HC1 (pH 7.4), 150 mM NaCl, 1% Triton X-100 or 0.6% cholate, 10 pg/ml leupeptin, and 10 pg/ml aprotinin at 4 "C for 20 min. The solubilized membrane extracts were centrifuged at 43,000 X g for 45 min. The supernatant was incubated with WGA-Sepharose beads at 4 "C for 5 h. Following incubation the beads were washed sequentially and extensively in 20 mM Tris-HC1 (pH 7.4), 0.6 M NaC1, 0.1% Triton X-100 or 0.1% cholate in the same buffer containing 0.15 M NaCI. The WGA-bound proteins were eluted from the column in the presence of 20 mM Tris-HC1 (pH 7.4), 0.3 M N-acetyl-D-glucosamine, 0.1% Triton X-100 or 0.1% cholate and 0.15 M NaCl. The eluted proteins were incubated with rat anti-GP85 (CD44) antibody conjugated beads at 4 "C for 15 h. Following incubation, the beads were washed extensively in 20 mM Tris-HCI (pH 7.4), 0.15 M NaC1, 0.1% Triton X-100 or 0.1% cholate. The proteins bound to the anti-GP85 (CD44) antibody affinity column were eluted in 50 mM diethylamine buffer (pH 11.5) and neutralized immediately with 2 M Tris-HCI (pH 3.0). The eluted material was subsequently concentrated and measured for protein concentration by the Bradford (62) procedure and analyzed for purity by SDS-polyacrylamide gel electrophoresis and silver staining.
Guanine Nucleotide Binding Assay GP85 (CD44) was purified from lymphoma plasma membranes by Triton X-100 solubilization and sequential WGA-Sepharose and anti-GP85 (CD44) antibody affinity chromatographies as described above. The purified GP85 (CD44) preparation was used in GTPyS binding assays as described below.
Equilibrium Binding between p5S]GTPyS and GP85 (CD44)-Aliquots (0.3 pg of protein) of purified GP85 (CD44) were hound to anti-GP85 (CD44) antibody affinity column beads as described above. GP85 (CD44)-bound beads were then incubated with increasing concentrations of [35S]GTPrS (0.2-80 nM, specific activity 50,000 cpm/ pmol) in binding buffer at 30 "C for 60 min. The nonspecific binding was determined in the presence of a 100-fold excess of unlabeled GTPyS. Following binding, the beads were washed, and the beadbound radioactivity was determined.

GTPase Activity Assay
The GTPase activity assay was performed as described previously with the following modifications (55,56). GP85 (CD44) was purified from Triton X-100-solubilized plasma membranes by sequential WGA-Sepharose and anti-GP85 (CD44) antibody affinity chromatographies as described above. The purified GP85 (CD44)-bound anti-GP85 (CD44) antibody heads were washed extensively in the GTPase assay buffer (20 mM Tris-HC1 (pH 7.4), 5 mM MgCl,, 0.1% cholate, Regulation of GP85 (CD44JAnkyrin Interaction by GTP 22075 and 1 mM dithiothreitol). Alternatively, GP85 (CD44) was eluted from the anti-GP85 (CD44) antibody affinity column using diethylamine buffer as described above. The eluted GP85 (CD44) was dialyzed extensively against GTPase assay buffer. Aliquots (1 pg of protein) of purified GP85 (CD44) (either bound to anti-GP85 (CD44) beads or remaining in a soluble form) were preincubated with 1 pM [y-'"PIGTP (4 X 10' cpm/pmol) and 0.1 mM ATP in a reaction volume of 50 pl a t 4 "C for 30 min. The samples were then incubated a t 35 "C for various time periods. Following incubation, 100 pl of 1% bovine serum albumin, 0.1% dextran sulfate made in 20 mM phosphate buffer (pH 8.0) was added to the reaction mixtures followed by the addition of 750 pl of activated charcoal suspension containing 20 mM phosphate buffer (pH 8.0). Following incubation a t 4 "C for 30 min, the reaction mixtures were centrifuged, and %' Pi released in the supernatant was determined by liquid scintillation counting. The results are expressed as pmol of P, released per pg of protein. In control samples, the nonspecific release of Pi caused by the background level of GTPase activity was determined using rat IgG beadbound proteins obtained as described above. The nonspecific release of Pi in control samples was less than 10% of that released by GP85 (CD44)-containing samples and has been subtracted.

Binding of GP85 (CD44) to Ankyrin
Ankyrin was purified from human erythrocyte ghosts according to the procedure described by Bennett and Stenbuck (57). Ankyrin was either conjugated to CNBr-activated Sepharose beads (1 mg of protein) or '?,''I labeled (specific activity 3 X 10' cpm/ng of protein) as described before (51).

Binding of Radioactively Labeled Proteins to Ankyrin-conjugated Sepharose Beads
[:"P]Azido-GTP labeled plasma membranes were solubilized in RIPA buffer and incubated with aliquots (100 pl) of ankyrin-conjugated beads at 4 "C for 6 h either in the presence or absence of excess soluble ankyrin. Following incubation, the radiolabeled proteins bound to the beads were solubilized in the SDS sample buffer and analyzed by 7.5% SDS-polyacrylamide gel electrophoresis and autoradiography. In some cases, radioactively labeled proteins bound to the ankyrin beads were dissociated in the presence of 1% SDS in 20 mM Tris-HCI (pH 7.4). The SDS-solubilized extracts were diluted in RIPA buffer to a final SDS concentration of 0.1% followed by anti-GP85 (CD44)-mediated immunoprecipitation as described above.

Binding of '"I-Labeled Ankyrin to GP85 (CD44)
GP85 (CD44) was purified from lymphoma plasma membranes (PT/CT-treated or without any treatment) by Triton X-100 solubilization and sequential WGA-Sepharose and anti-GP85 (CD44) antibody affinity chromatographies as described above. Aliquots (0.1 pg of protein) of GP85 (CD44) were bound to anti-GP85 (CD44) antibody affinity column beads and incubated in a reaction buffer containing 20 mM Tris-HCI (pH 7.4), 150 mM NaCI, and 0.05% Triton X-100 in the presence of various concentrations of G T P (0, 10, 100, 1,000 p~) a t 4 "C or room temperature for 2 h. 20 ng of "'I-labeled ankyrin and bovine serum albumin to a final concentration of 0.05% were added to these reaction mixtures, and the incubation was continued for 3h at 4 'C or room temperature. The nonspecific binding was determined in the presence of 2 pg of soluble ankyrin or 0.6 M NaCl as described previously. Following binding the beads were washed extensively in buffer B, and the bead-bound radioactivity was estimated.

Identification of GP85 (CD44) as a GTP-binding Protein
We have previously used two different biochemical labeling methods, [""Plazido-GTP photoaffinity labeling and ['"PI ADP-ribosylation, to identify the GTP-binding proteins in mouse T-lymphoma plasma membrane (54). Our results indicated that a 41-kDa PT-sensitive G-protein (identified as GiJ is present in these plasma membrane preparations. However, in addition to Gi,, we have detected other membraneassociated proteins (e.g. 130  were performed to establish that GP85 (CD44) is a physiologically relevant substrate for both bacterial toxins. For example, when [""PIADP-ribosylation was carried out using membranes pretreated with bacterial toxin (PT or CT) or without any treatment, no labeling of GPSs(CD44) was observed (data not shown).
To establish whether GP85 (CD44) is actually a GTPbinding protein, the lymphoma plasma membranes were treated with the photoaffinity label ['"Plazido-GTP. The results shown in Fig. 1B (lane a ) indicate that only four membrane-associated proteins (including GiJ are photoaffinity labeled with ['"Plazido-GTP. Immunoprecipitation of the labeled plasma membranes by anti-GP85 (CD44) monoclonal antibody indicates that GP85 (CD44) is labeled by [:'2P]azido-GTP (Fig. lB, lane b). However, no '"P-labeled GP85 (CD44) is immunoprecipitated if the labeling is performed in the presence of 1 PM unlabeled GTP (Fig. lB, lane c). Other unlabeled nucleotide triphosphates (including ATP, ITP, UTP) fail to block the ["'P]azido-GTP labeling of CD44 even at a 10 PM concentration (data not shown). These observations clearly indicate that GP85 (CD44) specifically binds GTP. Our preliminary data also indicate that the ADPribosylation sites and, possibly, the GTP binding site lie in a 40-kDa peptide of GP85 (CD44) that we have previously  (CD44) was purified from the lymphoma plasma membranes as described under "Materials and Methods." Aliquots of purified GP85 (CD44) (2: 0.3 pg of protein) were incubated with 1 p~ ['"SJGTPyS in the binding buffer a t 30 "C for various time periods. The nonspecific binding was determined in the presence of 100 p~ C T P y S or GTP and has been subtracted. Following incubation, GTPyS bound to GP85 (CD44) was determined as described under "Materials and Methods." Data shown are the average of triplicate determinations from a single experiment.
shown to contain the cytoplasmic domain and the ankyrin binding site' (51,52).

Sequence motifs in G-protein superfamily
The table shows the amino acid sequences of putative four (e.g G-1, G-2, G-3, and G-4) regions plus their consensus sequences which are characteristic of G-proteins (2,4,59). The upper case numbers indicate amino acid identities used to align sequences. X represents any amino acid, whereas and # indicate hydrophobic and hydrophilic residues, respectively. I represents the exact match, and : represents a conserved change in amino acids.   WGA-Sepharose column and can be eluted by N-acetylglucosamine (Fig. 2B) (50). After passing the N-acetylglucosamine-eluted proteins through an anti-GP85 (CD44) antibody affinity column, we obtained pure GP85 (CD44), since it is the only polypeptide revealed by silver staining. The amino acid composition of this protein has been shown to be identical to that of mouse Pgp-1 (CD44) (47). T o verify further whether GP85 (CD44) is a G-protein, we have used purified GP85 (CD44) (Fig. 2C) to test its GTP binding property (Fig. 3) and GTPase activity directly (Fig. 4).
[35S]GTPyS binds GP85 (CD44) in a saturable manner with a half-time of ~1 3 min at 30 "C ( Fig.  3) and an average rate of 0.14 min". Maximal binding of [35S] GTPyS is, however, substoichiometric. Under saturation condition, the specific activity of this GP85 (CD44) preparation is 4.3 nmol of GTPyS bound per mg of protein. Since the theoretical value for an 85-kDa protein is 11.7 nmol/mg protein, a portion of the GP85 (CD44) may have lost its ability to bind G T P during rigorous purification.
The binding affinity of guanine nucleotide for GP85 (CD44) is determined by incubating aliquots of purified GP85 (CD44) with various concentrations of [35S]GTPyS at 30 "C for 1 h. Scatchard plot analysis of these data results in a linear plot reflecting a single class of binding sites with an apparent Kd of 0.83 nM, which is comparable to that reported for other Gproteins (55,56,58).
GTPase Activity of GP85 (CD44)"Since all G-proteins possess an intrinsic GTPase activity, we have tested for GTPase activity of purified GP85 (CD44). As shown in Fig. 4, the purified GP85 (CD44) molecule displays a GTPase activity which hydrolyzes [y-32P]GTP in a time-dependent manner. The maximal rate of G T P hydrolysis is 0.011 mol/mol of GP85 (CD44) protein/min. This GP85 (CD44)-associated GTPase activity appears to be comparable to other G-proteins including the ras protein (2,4). However, the GTPase activity of these G-proteins is increased by either of the two regulatory classes of proteins: the guanine nucleotide release proteins and GTPase-activating proteins.
Similarly, mRNA-programmed ribosomes strikingly increase the GTPase activity of elongation factor T u (4). It remains to be determined whether the GTPase activity of lymphoma GP85 (CD44) is also similarly regulated.
As discussed above, G-proteins are targets for ADP-ribosylation by bacterial toxins such as PT and CT. These toxins covalently modify specific a subunits by transferring the ADP-ribose moiety from NAD onto cysteine (in the case of PT) or arginine (in the case of CT) (3). ADP-ribosylation often inhibits the intrinsic GTPase activity of PT/CT-sensitive G-proteins, which leads to a modification of G-proteinrelated activities (23). Since our data indicate that GP85 (CD44) can be ADP-ribosylated by both PT and CT, we have examined the effect of PT/CT-mediated ADP-ribosylation on its GTPase activity. As shown in Fig. 4, PT/CT-mediated ADP-ribosylation of GP85 (CD44) blocks more than 80-90% of its GTPase activity. Thus, as is case for other G-proteins, the ADP-ribosylation of GP85 (CD44) also inhibits its GTPase activity. The results described above clearly indicate that GP85 (CD44) is a GTP-binding protein and contains sites for ADP-ribosylation mediated by both PT and CT.
Recently the mouse Pgp-1 (GP85/CD44) cDNA has been cloned and sequenced. The overall sequence of GP85 (CD44) does not show a significant homology with any known Gproteins. However, if the amino acid sequences of GP85 (CD44)'s cytoplasmic domain is aligned with critical regions of known G-proteins (Table I), we have found that similarities between GP85 (CD44) and several classes of G-proteins (e.g. the ras class) are present within certain consensus regions (i.e. G-1, G-2, G-3, and G-4) ( Table I). For example, the sequence between amino acid 282 and 298 of GP85 (CD44) shares some similarity with the G-1 and G-2 regions that are conserved in many G-proteins including the ras class. The G-1 region is involved in GTP binding. Two other regions of GP85 (CD44) (e.g. amino acids 295-304 and 310-316) also have a moderate amount of similarity with the G-3 and G-4 regions, respectively, that are conserved in ras and other classes of G-proteins. This sequence analysis indicates that GP85 (CD44) is a novel G-protein which displays a composite of putative four regions (e.g. G-1, G-2, G-3, and G-4). Currently, we are using deletion mutation methods to determine further the precise GTP binding domain of GP85 (CD44).

Effect of GTP on GP85 (CD44)-Ankyrin Interaction
The most extensively characterized plasma membrane-cytoskeleton organization is that which exists in erythrocytes and lymphocytes (50,51,60,61). In lymphocytes, we have identified several transmembrane receptors that are directly linked to membrane-associated cytoskeletal proteins (51,61). For example, GP85 (CD44) is directly linked to an ankyrinlike protein (50-53). Ankyrin is a cytoskeletal protein which in erythrocytes links anion transporter band 3 protein to spectrin (63). Most interestingly, the GP85 (CD44)-ankyrin interaction appears to be regulated by certain post-translational modifications of GP85 (CD44). For example, phosphorylation of GP85 (CD44) by protein kinase C or fatty acylation by palmitic acid significantly enhances the GP85 (CD44)-ankyrin binding interaction (52,53).
In the following set of experiments, we have examined the effect of GTP binding to GP85 (CD44) on the ability of GP85 (CD44) to interact with ankyrin. First, we determined whether ['"P]azido-GTP-labeled GP85 (CD44) binds ankyrin-conjugated Sepharose beads. As shown in Fig. 5, lune A , the 85-kDa protein is the only protein among the several ["'Plazido-GTP-labeled lymphoma plasma membrane proteins (Fig. 1B) that binds to the ankyrin beads. This 85-kDa protein is identified as GP85 (CD44) because it is specifically immunoprecipitated by anti-GP85 (CD44) monoclonal antibody (Fig.  5 , lune B ) . The binding of GP85 (CD44) to the ankyrin beads is specific because GP85 (CD44) does not bind ankyrin beads in the presence of excess soluble ankyrin (Fig. 5, lane C).
We next examined the effect of GTP on the binding of '*'Ilabeled ankyrin to GP85 (CD44). Our results clearly indicate that GTP increases the binding of '251-labeled ankyrin to GP85 (CD44) in a dose-dependent manner (Table  11). A maximum increase of 2.2-2.5-fold over the control is observed at a 100 p~ G T P concentration (Table 11; Fig. 6, A and B ) .
Since our data indicate that GP85 (CD44) can be ADPribosylated by both PT and CT, we have examined the effect of GP85 (CD44) ADP-ribosylation on GTP-induced increase in GP85 (CD44)-ankyrin interaction. As shown in Fig. 6, C and D, PT/CT-mediated ADP-ribosylation of GP85 (CD44) almost completely blocked the GTP-induced increase in the binding of ankyrin to GP85 (CD44). The inhibition of the increase in the ankyrin binding by PT/CT-mediated ADPribosylation is also observed at all concentrations of GTP (data not shown). However, in control experiments, both PTand CT-mediated ADP-ribosylation had no effect on the basal level of ankyrin binding to GP85 (CD44) (data not shown). This observation suggests that PT/CT-mediated ADP-ribosylation of GP85 (CD44) interferes functionally with the GTP binding domain.
In conclusion, the results presented in this study indicate that the mouse lymphocyte cell adhesion molecule GP85 (CD44) is a novel class of G-proteins (i.e. displaying GTP binding and GTPase activity) that are transmembrane receptor glycoproteins. Most importantly, the binding of G T P increases the binding of GP85 (CD44) (a receptor-G-protein) to ankyrin, the membrane-associated cytoskeletal protein required for signal transduction and receptor patching/capping. Although at present it is not clear whether GP85 (CD44) exists in vivo in a GTP-bound state, it is possible that the binding of putative ligands (e.g hyaluronic acid) to GP85 (CD44) on the cell surface induces G T P binding which, in turn, enhances GP85 (CD44)-ankyrin interaction during lymphocyte activation.