The beta-subunit of the signal recognition particle receptor is a novel GTP-binding protein without intrinsic GTPase activity.

The beta-subunit of the signal recognition particle receptor (SRbeta), a member of the Ras family of small molecular weight GTPases, is involved in the targeting of nascent polypeptide chains to the protein translocation machinery in the endoplasmic reticulum membrane. We purified SRbeta from an expressing strain of Escherichia coli and investigated the properties of the isolated GTPase. We find that, unlike other Ras family GTPases, most SRbeta purifies bound to GTP, and SRbeta-bound GTP is not easily exchanged with solution GTP. SRbeta possesses no detectable GTPase activity. Although a stable interaction between SRbeta and ribosomes is observed, SRbeta is not stimulated to hydrolyze GTP when incubated with ribosomes or ribosome-nascent chains. A GTPase mutant harboring a mutation in a region predicted to be functionally important, based on observations made in related GTPases, binds GTP with faster kinetics and appears to be a less stable protein but otherwise displays similar properties to the wild-type SRbeta GTPase. Our results demonstrate that as an isolated GTPase, SRbeta functions differently from the Arf- and Ras-type GTPases that it is most closely related to by sequence.

Protein translocation across the mammalian endoplasmic reticulum (ER) 1 membrane is a cotranslational process believed to be regulated by three GTPases. Nascent polypeptide chains being synthesized in the cytosol are sampled by the signal recognition particle (SRP) as they emerge from the ribosome (1). The GTPase in SRP, SRP54, binds to signal sequences in nascent polypeptides as they emerge from the ribosome, forming ribosome-nascent chain-SRP ternary complexes (2,3). The ternary complexes are directed to the ER because of the affinity of SRP for its cognate receptor (SRP receptor; SR) on the surface of the ER membrane. Through a series of GTPasecontrolled steps the ribosome-nascent chain is transferred to the protein-conducting channel, or translocon, which facilitates translation of the nascent chain across, or integration into, the ER membrane (reviewed in Ref. 4).
The concerted action of GTPases ensures that the targeting step and nascent chain transfer step are unidirectional processes (5). SRP54 and one of the subunits of the SRP receptor, SR␣, bind GTP in a cooperative manner (6). Binding of GTP by SRP54 and SR␣ increases the affinity of these proteins for one another, thereby maintaining a direct physical link between the ribosome-nascent chain and the ER membrane (7,8). Sequence comparisons revealed that the SR␣ and SRP54 GTPases define a specific subfamily of GTPases conserved in prokaryotes, yeast, and mammals that is now referred to as the SRP family of GTPases (9,10). The third GTPase that appears to be involved in regulating translocation is SR␤. It has been proposed that release of SRP from the nascent chain, and subsequent transfer of the nascent chain to the translocon, is controlled by SR␤ (11).
Unlike SRP54 and SR␣, SR␤ shares significant homology within the GTP binding consensus sequences, or G boxes, of Ras-type GTPases (12). Structural analysis indicates that SR␤ also bears significant structural homology to Ras-type GTPases (13). However, it differs from other Ras-type GTPases in two respects. First, although other Ras-type GTPases require prenylation at the carboxyl terminus to enable a reversible interaction with membranes (14), SR␤ is permanently integrated into the ER membrane by an amino-terminal transmembrane domain. Second, SR␤ contains a cysteine within the G1 GTPase consensus sequence where most Ras-type GTPases contain a glycine. The other exception is members of the Arf family of GTPases that all contain an aspartic acid at this position. The identity of this amino acid appears to be crucial to the activity of Arf and Ras GTPases (15)(16)(17), which raises the possibility that SR␤ differs functionally from both Arf-like and other Rastype GTPases.
The search for protein factors that influence the activity of SR␤ has led to the observation that a factor associated with ribosomes possesses measurable GAP activity and may also function as a guanine nucleotide dissociation factor for SR␤. Incubation of ribosome-nascent chain complexes with either the SR␣/SR␤ dimer or a proteolysis product of the dimer lacking the SR␣ GTPase (SR⌬␣) both increases the GTPase activity of SR␤ and decreases the affinity of SR␤ for nucleotides (18). The influence of the ribosome on SR␤ suggests a direct physical contact between the two. Supporting this prediction, crosslinking experiments have revealed an interaction between SR␤ and a protein component of the ribosomal 60 S subunit (11). Whether this ribosomal protein is responsible for the observed GAP/guanine nucleotide dissociation factor activity is still unresolved.
In addition to proposed roles in nascent chain transfer and SRP release, SR␤ also anchors SR␣ to the ER membrane via a tight physical interaction between the SR␤ GTPase domain and an amino-terminal domain of SR␣ (19,20). This interaction is influenced by the nucleotide-bound status of SR␤. Generation of empty SR␤ by gel filtration of an XTP binding mutant of SR␤, which has a decreased affinity for GTP, abolishes the SR␣/SR␤ interaction. Replenishing the reaction mixture with xanthosine 5Ј-diphosphate or XTP restores dimer formation, with XTP having a greater effect. Deletion of any part of the SR␤ core GTPase also abolishes the interaction with SR␣ (21).
All data gathered to date on the function of SR␤ have been obtained in the context of a heterodimer. Attempts to isolate the SR␤ GTPase for study have involved proteolytic treatment of SR with trypsin or elastase to specifically digest SR␣. This method releases the GTP binding domain of SR␣ from SR␤ but leaves an amino-terminal domain of SR␣ bound to SR␤ (10,22). Therefore, there are no data on the properties of SR␤ as an isolated GTPase. To address this issue directly we have expressed and purified from Escherichia coli a soluble version of SR␤, termed SR␤⌬TM.
Isolated SR␤⌬TM has no detectable GTPase activity, and most does not exchange GTP in vitro. The small fraction (3-6%) of SR␤⌬TM that does bind exogenous GTP undergoes a conformational change that can be detected by fluorescence spectroscopy. A direct interaction between SR␤⌬TM and the ribosome is confirmed, and the influence of the ribosome on the SR␤ GTPase is examined. Our results suggest that SR␤ GTPase function is unlike other Ras-type GTPases.

EXPERIMENTAL PROCEDURES
Plasmids-Construction of plasmids, sequencing, and site-directed mutagenesis were performed using standard techniques. All encoded products are under the control of a T7 promoter. The plasmids pMAC191 (containing a modified full-length cDNA sequence of canine SR␣), pMAC455 (encoding SR␤ md ), pMAC1083 (encoding hemagglutinin-SR␤ XTP ⌬TM), and pMAC853 (encoding SR␤⌬TM, a fusion of the carboxyl-terminal 206 amino acids of canine SR␤ with an amino-terminal hemagglutinin epitope tag) were reported previously (20,21).
Plasmid pMAC1277 encodes SR␤⌬TM fused to an amino-terminal His tag and enterokinase (EK) cleavage site. This plasmid was assembled in two steps. First pMAC701, encoding SR␤ md fused to an aminoterminal His tag and EK cleavage site, was generated by removing the SR␤ coding sequence from pMAC455 by digestion with BglII and KpnI and inserting it into pRSETB (Invitrogen) digested with the same enzymes. The sequence encoding SR␤⌬TM was then excised from pMAC853 using NcoI and EcoRI and inserted into pMAC701 digested with NcoI and EcoRI, thereby replacing the coding region for SR␤ md with that for SR␤⌬TM.
Plasmid pMAC1623 encoding SR␤ C71G ⌬TM fused to an amino-terminal His tag and EK cleavage site was generated from pMAC1277 by the method described in Ref. 23. Briefly, the entire plasmid was amplified by PCR using oligo958 (ATGGGCCCCTCGGCAACTCTGGGAAAAC; desired mutation in bold) and oligo959 (ATGGGCCCCAACAAGAACA-GCTCT). The product was then digested with ApaI, and the 3Ј overhanging ends were blunted by incubation with the Klenow fragment of DNA polymerase, and the linear DNA was circularized by ligation with T4 DNA ligase.
Plasmid pMAC1624 encodes SR␤ C71D ⌬TM fused to an amino-terminal His tag and EK cleavage site, under the control of a T7 promoter. To generate this plasmid pMAC1277 was amplified by PCR using oligo960 (ATGGGCCCCTCGACAACTCTGGGAAAA; desired mutation in bold) and oligo959. The PCR product was digested with ApaI and end-repaired and ligated as above.
Plasmid pMAC1278 encodes SR␤ XTP ⌬TM fused to an amino-terminal His tag and EK cleavage site. SR␤ XTP ⌬TM was excised from pMAC1083 with NcoI and EcoRI and inserted into pMAC701 digested with NcoI and EcoRI, replacing SR␤ md with SR␤ XTP ⌬TM.
Protein Purification-Plasmids encoding either His-SR␤⌬TM or His-SR␤ C71D ⌬TM were expressed in the salt-inducible BL21SI strain by addition of NaCl to 300 mM final concentration for 2 h. All purification steps were carried out at 4°C. Cell pellets were washed once in 50 mM Na 2 HPO 4 , pH 8.0, 1 mM phenylmethylsulfonyl fluoride and resuspended in lysis buffer (50 mM Na 2 HPO 4 , pH 8.0, 500 mM NaCl, 5 mM MgOAc 2 , 1 mM phenylmethylsulfonyl fluoride, 10% glycerol (v/v)). Cells were lysed in a pressure cell, DNA was precipitated with 0.15% polyethylenamine, and lysate was centrifuged at 18,000 ϫ g for 20 min in a Beckman JA-20 rotor. The lysate was further clarified by centrifuging at 110,000 ϫ g for 1 h in a Beckman Ti50.2 rotor prior to loading on Ni-NTA-agarose (Qiagen) equilibrated in 50 mM Na 2 HPO 4 , pH 8.0, 300 mM NaCl, 5 mM MgOAc 2 , 10% glycerol. The column was washed in 10 volumes of equilibration buffer, and His-SR␤ was eluted with equilibration buffer ϩ 50 mM imidazole. Protein containing fractions were detected by BCA assay (Pierce), pooled, and dialyzed overnight in 40 mM Tris-OAc, pH 7.8, 300 mM NaCl, 5 mM MgOAc 2 , 1 mM DTT, 25% glycerol. Dialysate was diluted with 6 volumes of 40 mM Tris-OAc, pH 7.8, 5 mM MgOAc 2 , 1 mM DTT, 25% glycerol to reduce the NaCl concentration and loaded immediately onto CM-Sepharose equilibrated in 40 mM Tris-OAc, pH 7.8, 50 mM NaCl, 5 mM MgOAc 2 , 1 mM DTT, 25% glycerol. The column was washed with 10 volumes of equilibration buffer, and His-SR␤ was eluted in a single step in SR␤ elution buffer (equilibration buffer ϩ 100 mM NaCl). Protein containing fractions were detected by Bradford assay (Bio-Rad) and pooled. Protein concentration was determined by absorbance at 280 nm as described (24). Protein was frozen in small aliquots at Ϫ80°C; material used for functional studies was thawed once and discarded.
HPLC Analysis of Bound Nucleotide-10 nmols of SR␤ were diluted to 250 l in SR␤ elution buffer. An equal volume of 8 M urea, 20 mM Tris-OAc, pH 7.8, 100 mM NaCl was added, and the sample was incubated at 37°C for 30 min. The sample was centrifuged through a 5-kDa cutoff filter (Millipore), and the filtrate was added to a Bakerbond QUAT 5-m HPLC column (J. T. Baker Inc.) in 25 mM triethylamine bicarbonate, pH 7.2. Nucleotide was eluted from the column with a 5-100% gradient of triethylamine bicarbonate. Samples were analyzed with 32Karat, version 3.0 software (Beckman), and nucleotide was quantified by calculating the area under the curve and comparing to a standard curve of GTP or GDP. The recovery of nucleotides in these experiments (90%) was determined by adding a known amount of GMP as an internal control.
Fluorescence Experiments-300 nM SR␤ was incubated with 500 nM 2Ј (or 3Ј)-O-(N-methylanthraniloyl)GTP (mant-GTP) in 50 mM Tris-Cl, pH 7.6, 150 mM NaCl, 5 mM MgOAc 2 , 2 mM DTT, and 10% glycerol in a 1-cm path length quartz cuvette. All measurements were taken with a fluorometer equipped with an 815 photomultiplier detection system (PTI, London, Ontario, Canada) with a 2-nm excitation slit width and a 2-nm emission slit width and compiled with Felix, version 1.4 software (PTI). Samples were excited at 280 or 295 nm, and emission spectra were obtained by scanning from 300 to 500 nm in 2-nm increments with an integration time of 0.2 s per data point. Emission spectra were corrected by subtracting a buffer blank, and peak values were manually selected for further calculation. All calculations and data plots were performed within MS Excel 2002.
Filter Binding-100 pmols of SR␤ (1 M) were incubated at the specified temperatures with 10 M GTP including 25% [ 3 H]GTP (specific activity 31 Ci/mmol) in 50 mM Tris-OAc, pH 7.8, 200 mM NaCl, 5 mM MgCl 2 , 10% glycerol, 2 mM DTT. At the appropriate time points samples were withdrawn and diluted to 2 ml in ice-cold filter binding buffer (20 mM Tris-OAc, pH 7.8, 200 mM NaCl, 5 mM MgCl 2 , 10 mM NH 4 Cl). Samples were applied to prewashed nitrocellulose discs (Whatman), and the discs were washed with 3 ϫ 3-ml filter binding buffer in a Millipore 1225 filtration sampling manifold (Millipore). Discs were dried, and bound nucleotide was quantified in a scintillation counter.
Nucleotide Exchange-Nucleotide exchange reactions were performed as described previously (25,26). Briefly, SR␤⌬TM (1 M) was incubated with 20 M GTP including 0.2 M [␥-32 P]GTP for 10 min at 30°C in final buffer conditions containing 20 mM Tris-Cl, pH 7.6, 6 mM MgCl 2 , 10 mM EDTA, 1 mM DTT, 10% glycerol. After 10 min MgCl 2 was added to a concentration of 20 mM. The extent of nucleotide exchange was quantified by filter binding. To prepare SR␤⌬TM for GTPase assays, free nucleotide was separated from bound nucleotide by repurifying SR␤⌬TM on CM-Sepharose.
UV Cross-linking-5 M SR␤ was incubated with 0.5 M [␣-32 P]GTP and the indicated concentration of unlabeled GTP in cross-linking buffer (50 mM Tris-OAc, pH 7.8, 150 mM KOAc, 5 mM MgOAc 2 , 2 mM DTT) for 20 min on ice, followed by 5 min at 24°C. Reactions were placed into a plastic weight boat on a chilled metal block and irradiated with UV light at 5000 microwatts/cm 2 for 5 min. Samples were precipitated with trichloroacetic acid and washed in ethanol:ether (1:1) to remove free nucleotide, resolved by SDS-PAGE, and analyzed using a PhosphorImager.
GTPase Assay-40 nM nucleotide-bound SR␤ or 5.0 A 260 units/ml ribosomes was incubated with 83.5 nM [␥-32 P]GTP in GTPase buffer (50 mM Tris-OAc, pH 7.8, 150 mM KOAc, 5 mM MgOAc 2 , 2 mM DTT) at 24°C. At the indicated time points samples were removed and quenched by adjusting the EDTA concentration to 50 mM on ice. Samples were spotted onto polyethylenamine cellulose TLC plates and resolved in 0.375 M KH 2 PO 4 , pH 3.5, for 1 h. Plates were dried and exposed to a PhosphorImager screen for quantitative analysis. To generate the Lineweaver-Burke plot reactions were supplemented with cold GTP to concentrations up to 5 M, and the reaction was monitored using hydrolysis of [␥-32 P]GTP to estimate hydrolysis of all GTP.
To assess the effect of ribosomes on the SR␤ GTPase, 10 nM [␥-32 P]GTP-loaded SR␤⌬TM and 20 nM 80 S ribosomes were incubated at 24°C in GTPase buffer. Samples were removed at the indicated time points and quenched with 50 mM EDTA, and [␥-32 P]GTP was resolved from 32 P i by TLC.
Ribosome Binding Experiments-Canine pancreatic ribosomes and wheat germ RNCs were prepared as described elsewhere (11,18) and stored at a concentration of 100 A 260 units/ml in 25 mM HEPES-KOH, pH 7.6, 5 mM MgOAc2, 150 mM KOAc, 1 mM DTT (ϩ1 mM cycloheximide for RNCs). 5 M SR␤ was incubated with 20 A 260 units/ml ribosomes or RNCs for 1 h at 24°C in the above buffer and then added to the top of 30 ml of linear 0.3-1.2 M sucrose gradients. The gradients were centrifuged in a SW28 rotor for 16 h at 48,000 ϫ g. 1-ml fractions were collected by bottom puncture and protein precipitated with trichloroacetic acid, resolved by SDS-PAGE, and analyzed by Western blotting using an antibody directed against SR␤.
Data Analysis-All data analysis was performed using Sigmaplot 8.02.

RESULTS
Although sequence comparisons clearly indicate that SR␤ belongs to the Ras superfamily of small molecular weight GTPases it defines its own subfamily. One method of sorting GTPase family members is to define regions of homology within the residues lining the GTP binding pocket. Ras-type GTPases contain sequences of conserved residues called G boxes that are arranged at discrete intervals throughout the primary sequence (27). Arf family GTPases are distinguished from other Ras-type GTPases by the presence of an aspartic acid instead of a glycine residue in the G1 box (Table I). The identity of this residue within SR␤ is not strictly conserved among lower eukaryotes, but higher eukaryotes contain a cysteine in this position (Table II). The G1 box of canine SR␤ differs from Arf GTPases in only one other position in which an Ala not conserved in other Ras GTPases is replaced by a Ser.
In an attempt to identify the importance of the cysteine in the function of SR␤ two point mutants at this site were generated. The first, SR␤ C71D ⌬TM, converts the cysteine to an aspartic acid, converting SR␤ into an Arf family GTPase. The second, SR␤ C71G ⌬TM, converts the cysteine to a glycine to resemble other Ras-type GTPases. Binding of SR␤ and SR␤ mutants to SR␣ was assayed by coprecipitation. SR␣, wild-type SR␤, both cysteine point mutants, and another GTPase point mutant, SR␤ XTP ⌬TM, shown previously to switch the nucleotide binding preference from GTP to XTP (11,21), were synthesized in vitro in a rabbit reticulocyte lysate system. Nucleotides were removed from some samples by gel filtration (ϪGTP) and separate reactions containing equimolar amounts of SR␣ and each of the SR␤ variants were incubated together to allow complex formation. Complexes were immunoprecipitated with an antibody against SR␣ (Fig. 1). All of the SR␤ molecules bound SR␣ in the presence of nucleotides contributed by the translation mix. As reported previously (21), binding of SR␣ to the XTP-preferring version of SR␤ was greatly reduced in the absence of nucleotide, because the reduced affinity of SR␤ XTP ⌬TM for GTP allows this SR␤ variant to be emptied by gel filtration. Binding of either cysteine point mutant to SR␣ was unaffected by nucleotide depletion demonstrating that, despite the mutation in the G1 box, these mutants retain SR␣ binding activity under conditions that serve to empty SR␤ XTP ⌬TM.
To examine the isolated SR␤ GTPase in greater detail, a recombinant protein consisting of the cytoplasmic portion of SR␤ fused to an amino-terminal hexahistidine tag (His 6 ) was expressed in E. coli. The protein (SR␤⌬TM) was purified to apparent homogeneity in two steps involving Ni-NTA-agarose and CM-Sepharose (Fig. 2a). Gel filtration analysis of the purified product confirmed that SR␤⌬TM is a monomer in solution (data not shown). One of the cysteine point mutants, SR␤ C71D ⌬TM, was expressed with a carboxyl-terminal His 6 tag and purified using conditions identical to those used to purify SR␤⌬TM (Fig. 2b).
GTPases are generally purified in the GDP-bound form. This holds true for both tissue-derived proteins and recombinant proteins that lack GAP homologues in E. coli (28 -30). Therefore, we expected that SR␤ purified from E. coli would be GDP-bound. To identify and quantify the nucleotide that copurified with SR␤⌬TM, 10 nmols of SR␤⌬TM or SR␤ C71D ⌬TM were denatured in 4 M urea to release the bound nucleotide into solution. The protein was removed by filtration, and the released nucleotide was analyzed by HPLC and compared with standards of GTP and GDP examined in parallel (Fig. 3). The retention times for GDP and GTP on the HPLC column were 7.9 and 9.6 min, respectively (Fig. 3a). Surprisingly, the supernatant from denatured SR␤ contained a single major peak that eluted at 9.6 min, indicating the presence of GTP. A smaller peak was detected at 7.9 min, corresponding to a small amount of GDP (Fig. 3b). Calculating the area under the curves and comparing these values against values obtained from GTP and GDP standards and correcting for 10% loss (measured using GMP as an internal standard) revealed that 72% of SR␤⌬TM contains bound GTP whereas only 2.2% was bound to GDP. Similarly, 71% of purified SR␤ C71D ⌬TM contains GTP, and 2.8% was bound to GDP. Therefore both wild-type SR␤ and the GTPase point mutant remain bound to GTP throughout purification. The remaining 26% is not bound to nucleotide. This population of SR␤ did not bind to GTP in the timescale expected of an active empty GTPase (31, 32) (see Figs. 4 and 5). Therefore we presume that this population consists of SR␤ that has become structurally unstable in the absence of bound GTP, and the resulting loss of conformation prevented the uptake of exogenous GTP. A structurally unstable empty state is a common feature of Ras-type GTPases (31,33,34). Addition of 10 M GTP to the buffers used during purification did not decrease the percentage of empty SR␤ (data not shown), suggesting that this population does not arise from dissociation of nucleotide during purification.
Two methods were used to determine what fraction of recom- binant SR␤ is able to bind to or exchange bound GTP for exogenous GTP. The first method measured the ability of aromatic amino acids within SR␤ to transfer energy to a fluorescent GTP analogue, mant-GTP (Fig. 4). Resonance energy transfer (RET) can be measured by monitoring the decrease in fluorescence output of an excited donor molecule (aromatic amino acids) and concomitant increase in acceptor molecule (mant) fluorescence (Fig. 4a). Mant fluorescence did not change in a control cuvette lacking protein, nor was there an increase in mant fluorescence attributable to binding to SR␤ when the dye was excited directly at 350 nm (data not shown). Therefore, the increase in mant fluorescence arises solely from RET between the mant fluorophore and aromatic side chains within SR␤, including a tryptophan near the carboxyl terminus. We monitored GTP binding via the increase in mant fluorescence, because we observed a decrease in Trp fluorescence over time in the absence of mant-GTP (data not shown). We assume this is because of thermal denaturation of the purified protein in the fluorometer cuvette. Whatever the cause, by monitoring the increase in mant fluorescence we would slightly underestimate rather than overestimate the rate of GTP binding. The distance limitations of RET require that mant-GTP is bound to SR␤ for energy transfer to occur. Therefore this method provides a sensitive means to compare the rate of GTP binding to SR␤⌬TM and SR␤ C71D ⌬TM. Reactions containing 300 nM SR␤    3. Identification of nucleotide bound to SR␤. a, 1 nmol samples of GTP (black) and GDP (gray) were analyzed by HPLC on a quaternary amine column equilibrated in 25 mM triethylamine bicarbonate, pH 7.2. Nucleotide was detected by absorbance at 260 nm. The retention time for GTP was 9.6 min, and for GDP it was 7.9 min. b, 4 nmols of SR␤⌬TM (black) or SR␤ C71D ⌬TM (gray) were denatured in 4 M urea, and 25% of the nucleotide-containing supernatant was applied directly to a quaternary amine HPLC column by a series of three injections, completed within 2 min. Application of the sample was complete before the elution step began. Nucleotide was detected by absorbance at 260 nm. 65% of both SR␤⌬TM and SR␤ C71D ⌬TM were bound to GTP whereas only 2% was bound to GDP. The series of peaks eluting prior to 5 min are because of absorbance from contaminants in the urea used in the denaturing buffer. and 500 nM mant-GTP were excited at 280 nm, and energy transfer was monitored by measuring the increase in mant-GTP fluorescence emission at 340 nm (Fig. 4b). Both SR␤⌬TM and SR␤ C71D ⌬TM bound mant-GTP, with SR␤⌬TM following biphasic binding kinetics. The first mode is complete after 45 min; the second mode is slower and takes an additional hour to complete. SR␤ C71D ⌬TM shows a single mode of GTP uptake that is complete after 45 min. The initial rate of increase in mant fluorescence is greater for SR␤ C71D ⌬TM than for SR␤⌬TM, indicating that the cysteine mutation permits mant-GTP more rapid access to the GTP binding site in SR␤.
SR␤ contains only one Trp located five amino acids from the carboxyl terminus. Excitation at 295 nm permits measurement of RET between this tryptophan and mant-GTP. No change in the apparent kinetics of GTP binding to SR␤ C71D ⌬TM was observed (Fig. 4c). However, SR␤⌬TM now showed a single mode of GTP binding, that resembles the second mode detected at 280-nm excitation in both slope and duration. Therefore, the first mode arises from RET between one or more of the Tyr (and Phe) residues scattered throughout SR␤⌬TM and mant-GTP, whereas the second mode arises from RET between the carboxyl-terminal Trp residue and mant-GTP. Although fluorescence spectroscopy permits determination of binding kinetics it did not permit us to determine what fraction of SR␤ can bind GTP. Therefore, a nitrocellulose filter binding assay was used to quantify the amount of GTP that could bind SR␤ (Fig. 5). 100 pmols of SR␤⌬TM (diamonds) or SR␤ C71D ⌬TM (squares) was incubated with a 10-fold molar excess of GTP, including 25% [ 3 H]GTP, at 24°C for the indicated times. To analyze GTP binding the protein was bound to nitrocellulose filters and washed extensively to remove unbound nucleotide. The nucleotide remaining on the filter, representing the amount of solution GTP retained in a complex with SR␤, was quantified by scintillation counting. At 24°C both proteins show that same t1 ⁄2 for nucleotide binding as calculated from fluorescence data. After 2 h SR␤⌬TM bound a maximum of 6.6 pmols of GTP, reflecting an occupancy of 6.6%. SR␤ C71D ⌬TM bound GTP at a faster rate than SR␤⌬TM, reaching a maximum of 3.5 pmols of GTP bound after 1 h followed by a steady decline throughout the rest of the experiment. A similar decline in binding was observed during RET experiments (Fig. 4) and may reflect structural instability of SR␤ C71D ⌬TM during extended incubation at 24°C. These data reveal that Ͻ10% of SR␤ binds exogenous GTP (de novo or by exchange). Therefore, 90% of SR␤ is already tightly bound to nucleotide or in a conformation that is unable to bind nucleotide. Both RET and filter binding experiments indicate that SR␤ binds added GTP slowly, consistent with observations made in other GTPases assayed in their nucleotide-bound states (30,35,36).
The K d of SR␤ for GTP has been reported to range from 1 M for the purified, solubilized SR dimer (12) to 20 nM for the purified SR dimer reconstituted into liposomes (18). To determine the K d of the 7% of SR␤ that can accept exogenous GTP, purified SR␤ was cross-linked to [␣-32 P]GTP in the presence of an increasing concentration of cold competitor GTP (Fig. 6, f). Binding follows a characteristic sigmoidal curve with the inflection point occurring at 2 M, demonstrating that recombinant SR␤ and solubilized SR (12) have similar affinities for GTP. An identical K d was calculated for SR␤ C71D ⌬TM (Fig. 6, OE). Therefore purified recombinant SR␤⌬TM binds GTP with a similar affinity as native SR␤ after solubilization of microsomes, and mutation of the cysteine in the G1 box does not affect the affinity of this protein for GTP. Because of the short incubation period prior to cross-linking this K d measurement reflects the loose binding conformation revealed by RET and not the majority of SR␤ that is already bound to GTP.
Because the majority of SR␤ remains bound to GTP throughout purification (Fig. 3) it is likely that the intrinsic GTPase activity of SR␤⌬TM is negligible. To experimentally verify that the SR␤ GTPase does not possess intrinsic catalytic activity, SR␤⌬TM was incubated with [␥-32 P]GTP, and hydrolysis was monitored by quantifying the liberation of the terminal phosphate by thin layer chromatography (Fig. 7). RNCs treated with N-ethylmaleimide (NEM), identical to those used in previous attempts to assay the influence of the ribosome on the SR␤ GTPase (18), demonstrated that NEM treatment is not sufficient to abolish GTPase activity associated with RNCs (Fig. 7a, NEM-RNC). Therefore, initial measurements were made in the absence of ribosomes. After 4 h of incubation at 24°C no significant GTP hydrolysis was visually apparent above a control reaction lacking SR␤ (Fig. 7a, GTP). To ensure that the assay was sensitive enough to measure a low basal rate of GTP hydrolysis, the assay was repeated with varying concentrations of GTP, and a Lineweaver-Burke plot was generated to estimate a basal GTP hydrolysis rate (Fig. 7b). From the plot an estimated K m of 4.0 M and k cat of 0.0005 min Ϫ1 are derived, reflecting a negligible rate of GTP hydrolysis for SR␤⌬TM. Consistent with the negligible rate of GTPase activity obtained using the thin layer chromatography assay, the efficiency of cross-linking [␣-32 P]GTP and [␥-32 P]GTP to SR␤ were identical (data not shown). Therefore SR␤ is unable to hydrolyze GTP, suggesting the existence of a GAP.
A candidate GAP for SR␤ has been proposed recently (18) to reside within the ribosome. Because ribosomes are a significant source of GTPase activity, this activity must be abolished to unambiguously assign GTPase activity arising from SR␤ in reactions containing both SR␤ and ribosomes. The use of alky-lating reagents to modify ribosomal proteins has been shown to reduce, but not abolish, the activity of certain ribosome-associated GTPases (37) (see also Fig. 7a). Therefore in addition to 80 S ribosomes, isolated ribosomal subunits and RNCs were treated with NEM in an attempt to decrease background ribosome-associated GTPase activity enough to detect GTP hydrolysis arising from SR␤. Treatment with NEM had no effect on the GTPase activity of 80 S ribosomes or RNCs (see Fig. 7a and Supplemental Fig. 1), but the GTPase activity of isolated 60 S subunits, already significantly decreased compared with intact ribosomes, was abolished following treatment with NEM (see Supplemental Fig. 1). Incubation of SR␤ with NEM-treated 60 S subunits did not result in any additional GTP hydrolysis above background levels (data not shown).
Although isolated 60 S ribosomal subunits did not stimulate the SR␤ GTPase, it is possible that ribosome-associated GAP activity requires an intact 80 S ribosome. The GTPase activity of 80 S ribosomes precluded analysis with exogenous nucleotide; therefore we chose to analyze hydrolysis of GTP bound by SR␤⌬TM upon incubation with RNCs. If a protein within the ribosome acts as an SR␤ GAP then incubation of SR␤ with ribosomes in the absence of added GTP should result in the hydrolysis of SR␤-bound GTP to GDP. If the ribosome functions as a guanine nucleotide releasing factor then incubation of SR␤⌬TM with ribosomes should lead to the release of SR␤⌬TM bound GTP or GDP. We assessed the effect of adding ribosomes to SR␤⌬TM bound to GTP by removing the ribosomes by centrifugation at the end of the incubation and then assayed for FIG. 6. Cross-linking of GTP to purified SR␤⌬TM and SR␤ C71D ⌬TM. 50 nM SR␤⌬TM or SR␤ C71D ⌬TM were incubated with [␣-32 P]GTP in the presence of increasing concentrations of unlabeled competitor GTP for 20 min on ice, followed by 5 min at 24°C. Samples were transferred to a plastic weigh boat on a chilled metal block, and GTP was cross-linked to SR␤ by UV irradiation. The protein was isolated by TCA precipitation and separated by SDS-PAGE, proteins were fixed with acid/methanol, and the gel was washed to remove unbound nucleotide. Radioactivity was quantified from dried gels using a PhosphorImager and plotted against the concentration of added GTP. The apparent K d for both SR␤⌬TM (black diamonds) and SR␤C71D⌬TM (gray squares) was ϳ2 M. Although there is no visually apparent difference between samples containing SR␤ and GTP or GTP alone there is a small change that can be detected using a PhosphorImager. b, the assay was repeated using GTP concentrations ranging from 0.2 to 5 M and analyzed as in a. Radioactive regions were quantified using a PhosphorImager, and the data were displayed as a Lineweaver-Burke plot. Analysis of the plot estimates a GTP hydrolysis rate of Յ0.0005/min. nucleotide in the supernatant and bound to SR␤⌬TM using the HPLC method described above. Using this approach we were unable to detect any increase in hydrolysis of GTP because of the addition of ribosomes (data not shown). To increase the sensitivity of the assay and ensure that there were excess ribosomes present in the reaction we examined hydrolysis of 32 P-labeled GTP using thin layer chromatography as described above. Because nucleotide exchange in SR␤⌬TM is very inefficient (see Fig. 5) we incubated SR␤⌬TM with [␥-32 P]GTP and EDTA. In other low molecular weight GTPases this incubation step allows rapid nucleotide exchange. The exchange reaction was stopped by adding excess Mg 2ϩ (25,26). By monitoring the degree of exchange by nitrocellulose filter binding it was discovered that even in the presence of nucleotide and EDTA only 6% of SR␤⌬TM could exchange GTP for [␥-32 P]GTP (data not shown), in agreement with the results obtained from time-dependent nucleotide exchange (Fig. 5). Unbound nucleotide was removed by repurifying SR␤⌬TM on CM-Sepharose, SR␤⌬TM was incubated alone or in the presence of Ͼ2-fold molar excess of 80 S ribosomes, and GTP hydrolysis over time was monitored by TLC. As expected, we detected no intrinsic GTPase activity in SR␤⌬TM alone. Moreover, the presence of ribosomes did not stimulate the GTPase activity of SR␤⌬TM (data not shown).
In addition to proposing that a ribosomal component acts as an SR␤ GAP, Bacher et al. (18) measured a decreased affinity between SR␤ and guanine nucleotides in the presence of ribosomes, suggesting that a ribosomal component also behaves as a guanine nucleotide releasing factor. However, if SR␤ displays a lower affinity for GTP when incubated with ribosomes, it is likely that some GTP would dissociate from SR␤ during the incubation and become available for hydrolysis by the ribosome. Because we did not observe any GTP hydrolysis we conclude that the GTP remains tightly bound to SR␤ throughout the incubation with ribosomes.
Chemical cross-linking experiments have yielded a specific cross-link between SR␤ and a protein within the 60 S ribosomal subunit, suggesting that a physical association does occur (11). We were unable to detect a cross-link between SR␤ and a ribosomal protein using conditions that result in cross-links between SR␤⌬TM molecules and that in previous publications supported cross-linking between ribosomes and SR␣/SR␤ (see Supplemental Fig. 2). This raised the possibility that we do not detect an influence of the ribosome on SR␤, because the ribosome is unable to bind SR␤ in the absence of SR␣. To test this possibility SR␤ binding to 80 S ribosomes and RNCs was assessed by sedimentation in sucrose density gradients. SR␤⌬TM or SR␤ C71D ⌬TM was incubated with purified 80 S ribosomes or RNCs, and ribosome-bound SR␤ was separated from unbound SR␤ by centrifugation on a 10 -40% sucrose gradient. Fractions were collected and analyzed by Western blotting with an antibody against SR␤ (Fig. 8). Both SR␤⌬TM and SR␤ C71D ⌬TM formed a stable complex with both untranslating ribosomes and RNCs, as revealed by their comigration in sucrose (Fig. 8,  b-e). Prolactin, which is not expected to interact with ribosomes, remained at the top of the gradient (Fig. 8f). These data demonstrate that although the ribosome is unable to stimulate the SR␤ GTPase, a stable interaction between the two can still occur. It should be noted that SR␤ is present in excess over ribosomes in these experiments, so it is not possible to estimate the percentage of SR␤ that is able to bind to ribosomes from this figure. By performing the experiment with equimolar amounts of SR␤⌬TM and ribosomes, we determined that 22% of SR␤ is recovered in the ribosome-containing fractions (data not shown). Although it is not possible to estimate the amount of SR␤ that can initially form a complex with ribosomes, the fact that 22% of SR␤ remains bound to ribosomes throughout a 16-h centrifugation step provides evidence that the interaction between ribosomes and SR␤ is stable. DISCUSSION To examine the properties of the isolated SR␤ GTPase, we expressed SR␤⌬TM and SR␤ C71D ⌬TM in E. coli. The transmembrane domain was deleted from both SR␤ molecules to increase the solubility and yield in the expression system. The deleted region is not believed to contribute to the activity of SR␤, because SR␤⌬TM has already been shown to rescue translocation function in vivo in yeast containing two disrupted SR␤ alleles (38).
Consistent with previous data, we have shown that recombinant SR␤ binds GTP with a K d of ϳ2 M, similar to detergentsolubilized SR (12), but much higher than the K d of SR␣-SR␤ dimers reconstituted into lipid vesicles (18). It must be noted that for all of these reports the K d measurement is relevant only to the small fraction of SR␤ that is able to bind solution GTP during the assay. In our experiments greater than 70% of the SR␤⌬TM was already bound to GTP, and less than 10% of the SR␤⌬TM added to the reaction binds GTP prior to crosslinking (Fig. 5). Furthermore, the short incubation time used in cross-linking studies favors the loose binding conformation (Fig. 4). The affinity for GTP of the larger population of SR␤ that purifies bound to GTP is unknown, but it is presumably much higher than 2 M, because there is no appreciable ex- change with solution GTP during an 8-h incubation (Fig. 5), and bound GTP is removed during purification of the protein extremely slowly or not at all, despite the absence of solution GTP in the purification buffers.
Previous attempts to measure the K d of SR␤ (12) did not account for the possibility that much of the SR␤ used in the assay may not be able to accept exogenous GTP. It is likely that these K d measurements reflect the affinity of the same small population of SR␤ measured here that can bind to (or exchange with) exogenous GTP and do not reflect the affinity of the majority of SR␤ in the assay. It is perhaps significant that 2-3% of SR␤ purified from E. coli is bound to GDP. Our estimates of the fraction of SR␤ that binds GTP, 3-6%, are similar enough to 2-3% that we speculate that the GDP bound form of SR␤ is responsible for the binding activity that we measured. Bacher et al. (18), by incorporating purified SR into proteoliposomes, have measured a K d in close agreement with other Ras-type GTPases. It is possible that lipid binding by SR␤ leads to a conformational change that permits exchange of bound GTP with exogenous GTP.
The GTP binding site in SR␤⌬TM differs from other low molecular weight GTPases in that it contains a cysteine at a position within the G1 GTPase consensus sequence that is highly conserved as either glycine or aspartic acid in other family members of Ras-type GTPases (Table I). Structural analysis of Ras (39) and ARF-1 (40) does not provide insight into the functional role of the amino acid at this position, but mutation of this residue is invariably detrimental to the function of the protein (16,17,41,42).
The identity of the amino acid at this position in SR␤ is somewhat less conserved, suggesting that the side chain at this position is less important for the function of the protein than it is for other GTPases (Table II). In yeast, SR␤ contains a glutamine at this position that simultaneously binds SR␣ and protrudes into the SR␤ GTP binding pocket (13). SR␤ shows greater sequence homology to Arf GTPases than to Ras (12); therefore we mutated the Cys to Asp in an attempt to convert SR␤ into an Arf-type GTPase. Compared with wild type, SR␤ C71D ⌬TM appears to bind GTP with faster kinetics than SR␤⌬TM, suggesting that the conformation of the protein has changed such that GTP has easier access to the binding pocket. The cysteine normally at this position may contribute to the stability of the protein, because at 24°C the mutant protein exhibits a gradual loss of nucleotide binding (see Figs. 4 and 5). Nucleotide preference was not affected by this mutation, because fluorescence assays failed to detect binding of mant-XTP to SR␤ C71D ⌬TM (data not shown). We were also unable to distinguish a difference in nucleotide affinity between SR␤⌬TM and SR␤ C71D ⌬TM for the small fraction of protein that binds exogenous nucleotide. Unlike the Asn in yeast SR␤⌬TM, the Cys in the canine protein is not likely to be involved in binding of SR␤⌬TM to SR␣, because the C71D mutation does not appear to interfere with coimmunoprecipitation of SR␣ with SR␤ C71D ⌬TM (Fig. 1). Finally, we were unable to detect GTPase activity arising from SR␤ C71D ⌬TM (Fig. 8) (data not shown), but because we could not detect GTPase activity from wild-type SR␤⌬TM the role of the cysteine in catalysis remains uncertain.
The use of fluorescence to study GTP binding to SR␤⌬TM revealed a two-step process (Fig. 4). The first step is rapid and was detected by monitoring energy transfer between Tyr (and Phe) residues within SR␤ and a mant fluorophore incorporated into GTP. The second step is slower and was detected by monitoring energy transfer between a Trp residue located at the carboxyl terminus of SR␤ and the mant fluorophore. This second step occurs on the same time scale as GTP binding moni-tored by a nitrocellulose filter binding assay, a technique that captures tightly bound protein-nucleotide complexes (Fig. 5). Taken together, these data suggest a two-step model for GTP binding to SR␤.
The first step involves GTP bound to SR␤ in a loose conformation. The filter binding assay does not detect these complexes as loosely bound GTP is washed away. We detect these complexes by energy transfer between SR␤ and mant-GTP. The time scale of this loose binding interaction is similar to GTP binding by other Ras-type GTPases assayed in their GDPbound state (30,35). The second step represents a tight binding conformation, detected by both filter binding and energy transfer between Trp and mant-GTP. Our data suggest that in SR␤⌬TM the carboxyl terminus of SR␤ reorients such that energy transfer occurs between Trp and mant. We propose that this conformational change stabilizes the tight binding GTP-SR␤⌬TM complex. Comparison of the data in Fig 4, b and c suggests that the increase in RET between the Trp and mant-GTP occurs subsequent to binding. The simplest explanation for the increase in RET with the Trp is that SR␤⌬TM undergoes a conformational change that moves the Trp closer to the GTP binding site. Consistent with a role for the carboxyl terminus of SR␤ in stabilizing the structure of the protein we have shown previously (21) that the carboxyl-terminal six amino acids (including the one Trp in SR␤) are required for folding of the protease-resistant core of SR␤. SR␤ C71D ⌬TM exhibits tight binding of GTP but does not undergo the conformational change that stabilizes the complex, because the rate of GTP binding is the same whether it is measured by RET or filter binding. It may be that access to the GTP binding site is altered in SR␤ C71D ⌬TM but the mechanism of GTP binding is not changed.
We have shown that the bulk (ϳ70%) of both SR␤⌬TM and SR␤ C71D ⌬TM are tightly bound to GTP. In contrast other Rastype GTPases are all purified in the GDP-bound state (28 -30). Even ARF-1 purifies GDP-bound, yet it exhibits no measurable GTPase activity in vitro (29,35). Ran purifies bound to both GTP and GDP, reflecting the equilibrium between the two populations within the cell (43). Thus, SR␤ is the only Ras-type GTPase confirmed to purify predominantly in the GTP-bound state. This result is not specific for isolated SR␤, because SR␤-SR␣ complexes also purified in the GTP-bound state (13).
The finding that SR␤ defaults to the GTP-bound state whereas other Ras-type GTPases default to the GDP-bound state makes some biological sense. GTP-bound SR␤ binds to SR␣ more tightly than when SR␤ is loaded with other nucleotides (21). Because SR␤ is required to anchor SR␣ to the ER membrane, and SR␣ is found almost exclusively in the membrane fraction (38), unlike other Ras-like GTPases it makes sense to keep SR␤ bound to GTP.
It has been reported previously (18) that ribosomes both stimulate the SR␤ GTPase and decrease the affinity of SR␤ for nucleotides. Furthermore, an interaction between SR␤ and a 21-kDa ribosomal protein has been detected, which may provide the basis for the effect of the ribosome on nucleotide binding by SR␤ (11). However, in the absence of SR␣, we were unable to detect an influence of ribosomes or ribosomal subunits on the SR␤ GTPase. Attempts to detect a cross-link between a ribosomal protein and SR␤ incubated with GTP or GDP were also unsuccessful (see Supplemental Fig. 2). Because SR␤ can bind to ribosomes in the absence of SR␣ these results suggest that SR␣ changes the quality of the interaction between the ribosome and SR␤, either by regulating the structure of SR␤ to facilitate GTP hydrolysis or by directly contributing residues that are required for GTP hydrolysis. This suggests an additional role for SR␣ in translocation as an effector of SR␤ and is consistent with structural data that demonstrate that SR␣ binds SR␤ predominantly through the SR␤ switch I region (13).
Ribosomes are unlikely to interact with SR␤ in the absence of SR␣ in vivo. Therefore, an SR␣-SR␤-ribosome-nascent chaintranslocon complex may be required for SR␤ to hydrolyze GTP. Hydrolysis of GTP may then lead to dissociation of SR␣ from SR␤, contributing to transfer of the RNC from SR to the translocon (21).
We have shown previously that SR␣ forms a tight physical association with SR␤, and this interaction is nucleotide-dependent and requires the intact GTPase domain of SR␤ and is facilitated by the unique loop sequence located between the G4 and G5 boxes (20,21). These characteristics are consistent with SR␣ assuming the role of an SR␤ effector molecule. Our finding that the ribosome does not stimulate the GTPase activity of isolated SR␤, together with previous data demonstrating GTPase activity of SR␤ only in the context of an SR␣-SR␤ribosome complex (18), leads us to speculate that SR␣ may regulate the GTPase activity of SR␤. A detailed structural analysis of SR␤ alone and in a complex with SR␣ will be required to assess the extent that binding of SR␣ regulates SR␤ function.