The ATP Binding Site on Rho Protein

THE JOURNAL OF BIOLOGICAL CHEMISTRY 0 19s by The American Society for Biochemistry and Molecular Biology, Inc.

Vol. 263, No. 35, Isme of December 15, pp. 18810-18815,1988 Printed in U.S.A.

The ATP Binding Site on Rho Protein AFFINITY LABELING OF LYS’” BY PYRIDOXAL 5’-DIPHOSPHO-5‘-ADENOSINE*

(Received for publication, May 24, 1988)

Alicia J. DombroskiS, James R. LaDineg, Richard L. Cross§, and Terry Platt$ From the $Department of Biochemistry, University of Rochester Medical Center, Rochester, New York 14642 and §Department of Biochemistry and Molecular Biology, Health Science Center, State University of New York, Syracuse, New York 13210

We have labeled the nucleoside triphosphate-binding domain of Escherichia coli rho factor with the ATP affinity analog [‘H]pyridoxal 5’-diphospho-5’-adenosine (PLP-AMP). PLP-AMP completely inactivates the RNA-dependent ATPase activity of rho upon incorporation of 3 mol of reagentlmol of hexameric rho protein. Although the potency of PLP-AMP is enhanced when an RNA substrate such as poly(C) is present, the stoichiometry for inhibition remains the same as in the absence of poly(C). The nucleotide substrate ATP com- petes very effectively for the binding site and protects against PLP-AMP inactivation. A domain of rho called N2, which comprises the distal two-thirds of the molecule (residues 152-419) and encompasses the region proposed to bind ATP, is labeled specifically in the presence of poly(C). Amino acid sequence analysis of the single [‘HIPLP-AMP labeled proteolytic fragment showed Lys”’ to be the site of modification, suggesting that this residue normally interacts with the 7-phosphoryl of bound ATP. These results agree with our proposed tertiary structure for the ATP-binding domain of rho that places this lysine residue in a flexible loop above a hydrophobic nucleotide-binding pocket comprised of several parallel &strands, similar to adenylate kinase, F1-ATPase, and related ATP-binding proteins. Parallel studies of rho structure and function by site-directed mutagenesis and chemical modification support this interpretation.

The Escherichia coli transcription termination factor rho is required for correct RNA 3‘-end formation at specific, rho- dependent sites (Platt and Bear, 1983; von Hippel et al., 1984; Platt, 1986). Rho has an RNA-dependent ATPase activity (Lowery and Richardson, 1977a, 1977b; Lowery-Goldhammer and Richardson, 1974; Howard and decrombrugghe, 1976) that is required for its ability to unwind RNA-DNA hybrids in vitro (Brennan et al., 1987). In uiuo, rho presumably binds to its recognition site on the nascent RNA transcript and, upon hydrolysis of ATP, releases the RNA from its DNA template (Richardson and Conaway, 1980; Shigesada and Wu, 1980). Rho undergoes clear conformational changes upon binding RNA and ATP (Engel and Richardson, 1984; Bear et al., 1985), and its RNA-binding region was initially shown to reside within the first 283 amino acids of the protein (Bear et al., 1985), but the specific sites of interaction for RNA and

Health Grants GM35658 (to T. P.) and GM23152 (to R. L. C.) and * These studies were supported in part by National Institutes of

National Institutes of Health Genetics Training Grant GM07102 (to A. J. D.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “aduertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

ATP have been no further defined until recently. We now know that the RNA-binding domain is confined

within the first 151 amino acids and have proposed that this RNA-binding domain must interact with a separate ATP- binding domain to facilitate nucleoside triphosphate hydrolysis (Dombroski and Platt, 1988). Very strong sequence ho- mology to the ATP-binding regions of other nucleotide-binding proteins led us to predict that the ATP-binding domain of rho is located in the region following amino acid 164 where five parallel p-strands can form a hydrophobic pocket for the ATP molecule (Dombroski and Platt, 1988). Oligonucleotide site-directed mutagenesis of rho within the first, and most highly conserved P-strand, at Lys’’l and Lys”, and at Aspz6‘ in the 04-strand suggests that these residues may be directly involved in interactions required for ATP hydrolysis and supports the proposed location for the ATP molecule within our tertiary structure model (Dombroski et al., 1988).

ATP-binding sites can be defined more precisely using purine nucleotide affinity analogs (Colman, 1983); however, the failure of some analogs to specifically target the ATP- binding site results in indiscriminate modification at a wide variety of locations. Bear et al. (1985) attempted to label the ATP-binding domain of rho with the photoaffinity analog 8- azidoadenosine 5”triphosphate and the chemical affinity label adenosine 2’,3‘-dialdehyde 5‘-triphosphate, but results were not consistent and both analogs bound nonspecifically in more than one tryptic fragment of rho. The labeling of the rho factor’s ATP-binding site had thus proven to be difficult. Recently, an ATP analog called pyridoxal 5’-diphospho-5’- adenosine (PLP-AMP)’ was synthesized for use in labeling ATP-binding sites. PLP-AMP was designed to resemble ATP structurally but to allow covalent modification of lysyl residues in adenine nucleotide binding sites (Tamura et al., 1986). The pyridoxal carbonyl reacts with the t-amino group of lysines to form a Schiff-base and upon borohydride reduction yields a stable pyridoxamine derivative. Labeling of other enzymes with PLP-AMP appears to be quite specific and to resemble normal ATP-binding behavior (Tamura et al., 1986, 1988; Tagaya et al., 1987; Rao et al., 1988). We present here the results of experiments to measure binding, competition, specificity, and stoichiometry of PLP-AMP interaction with the active rho hexamer. We further show that rho factor can be labeled with [3H]PLP-AMP uniquely at Lysl” within the flexible loop of the proposed ATP-binding domain, in good agreement with our predicted tertiary structure and with the results of oligonucleotide site-directed mutagenesis (Dom- broski et al., 1988).

The abbreviations used are: PLP-AMP, pyridoxal 5’-diphospho- 5”adenosine; HPLC, high performance liquid chromatography; SDS, sodium dodecyl sulfate; poly(C), polycytidylate.

18810

ATP Binding Site on Rho Protein 18811

EXPERIMENTAL PROCEDURES

Materiak-Rho protein was prepared and purified by the procedure of Wu et al. (1981). [3H]PLP-AMP was synthesized and purified as described by Tamura et al. (1986). ATP, polycytidylate (poly(C)), and Sephadex G-50 were from Pharmacia LKB Biotechnology Inc. Poly- ethyleneimine plates (Polygram CEL 300 PEI) were from Brinkmann Instruments. Tissue solubilizer (NCS) was from Amersham Corp. Aqueous scintillant (ScintiVerse 11) and nonaqueous scintillant (ScintiLine) were from Fisher. Endoproteinases were supplied by Boehringer Mannheim.

Modification of Rho by PLP-AMP--f3H]PLP-AMP was bound to rho in a reaction mixture containing 50 mM Tris-HC1, pH 8, 300 nM rho (hexamer), 0-500 p~ [3H]PLP-AMP (15,000 cpm/nmol), 1 mM magnesium acetate, and 60 p~ poly(C) when appropriate. Preparative labeling of rho was done using 10 p M rho (hexamer), 50 mM Tris- HC1, pH 8, 1 mM magnesium acetate, 1 mM poly(C), and 200 p M [3H] PLP-AMP. In the competition experiments, ATP and PLP-AMP were present a t 50 p M while rho (hexamer) was at 100 nM. In every case after 4 h at room temperature in the dark, NaBH4 was added at 5 mg/ml to 60 mM and let react for 15 min. Unincorporated label was removed using Sephadex G-50 centrifuge columns (Penefsky, 1977) with 50 mM Tris-HC1, pH 8. Samples taken before and after centrifugation were counted in aqueous scintillant and used to calculate the stoichiometry of binding.

Rho was labeled for cleavage with hydroxylamine (NHZOH) in a 100-pl reaction mixture containing 50 mM Tris-HC1, pH 8, 40 pg of rho protein at 5 p~ (hexamer), 50-200 pM [3H]PLP-AMP, and 1 mM magnesium acetate plus 1 mM poly(C) where indicated. After 4 h at room temperature in the dark, NaBH4 was added to 60 mM and the mixture was incubated at room temperature for an additional 15 min. A Sephadex centrifuge column was used as described above to separate unincorporated nucleotide. 4 p1 of 10% sodium dodecyl sulfate (SDS) and 250 pl of 2 M NHzOH (in 2 M LiOH and 0.8 M Tris-HC1) was added to the labeled rho and incubated for 4 h at 45 "C. The reaction mixture was neutralized with 400 pl of 1 N acetic acid, and the labeled protein was precipitated by addition of 1 ml ice-cold acetone. After 1 h at -20 "C, the mixture was centrifuged for 15 min at 4 "C in an Eppendorf microcentrifuge. The pellet was air-dried briefly and resuspended in 30 pl of SDS/urea sample buffer (Dom- broski and Platt, 1988). After heating for 3 min at 90 "C, the mixture was loaded onto a 0.1% SDS/10% polyacrylamide gel and electropho- resed at 25 mA until the bromphenol blue band was 2-3 cm from the bottom. Bands corresponding to the N1 and N2 fragments (Dom- broski and Platt, 1988) were visualized with Coomassie staining, excised, and solubilized by addition of 1 ml of Amersham NCS Tissue Solubilizer a t a 91 dilution (NCS:HzO) followed by heating at 50 "C for 2 h. Samples were counted in organic scintillant.

ATPase Assays-Labeled rho was diluted to 50 nM, and 1 p1 was added to 9 p1 of a reaction mixture containing ATPase buffer (50 mM KCl, 20 mM Tris acetate, pH 7.9,0.4 mM magnesium acetate, 0.1 mM EDTA, and 0.1 mM dithiothreitol), 20 pM ATP, 45 nCi [y-32P]ATP, and 5 p~ poly(C). Aliquots were removed after 5 min at 37 "C, and reaction products were separated by thin layer chromatography on polyethyleneimine plates as described (Dombroski et al., 1988). In other experiments, ATPase activity was determined by a spectrophotometric coupled enzyme assay which linked ADP formation to the oxidation of NADH. Rho samples were preincubated with 1 ml of assay medium containing 5 p~ poly(C). ATPase reactions were started by addition of ATP to 200 p ~ .

Proteolytic Digestions and Purification of Labeled Peptide-Limited tryptic digestion of PLP-AMP-labeled rho was done according to Dombroski and Platt (1988) with poly(C) bound to PLP-AMP-labeled rho and ATP plus poly(C) both bound to the rho control. In both cases 10 pg of rho was digested.

For sequence analysis, PLP-AMP-labeled rho (0.46 mg/ml) was precipitated by addition of %o volume of 0.15% deoxycholate followed 10 min later by %a volume of ice-cold 72% trichloroacetic acid. Protein pellets were collected by centrifugation, rinsed twice with acetone, and dissolved in 5 M urea, 100 mM Tris-HC1 (pH 8.0) to a final protein concentration of 0.46 mg/ml. Endoproteinase Lys-C was added to give a ratio of 1:40 (w/w) of endoproteinase Lys-C:rho. The mixture was incubated at 37 "C overnight. The next morning an equal aliquot of endoproteinase Lys-C was added, and digestion was continued for 3 h a t 37 "C. The protein solution was diluted 5-fold with 100 mM Tris-HC1 (pH 8.0) to reduce the urea concentration to 1 M. Endoproteinase Arg-C was then added to give a ratio of 1:40 (w/w) of endoproteinase Arg-C:rho. Overnight incubation at 37 'C was

carried out as before. A second identical aliquot of endoproteinase Arg-C was added, and incubation was continued for 3 to 5 h. The protein solution was filtered through a 0.22-pm luer-lock syringe filter (Millipore) to remove insoluble material before HPLC injection. The filter was placed in a scintillation vial filled with scintillation mixture and counted. The material collected on the filter contained no radioactivity.

Peptide fragments were separated in three reversed-phase HPLC steps using a Vydac C-4 column and a Waters system. The flow rate was 1.0 ml/min for all separations. In the first step, the digest was applied to a column equilibrated with 0.1% trifluoroacetic acid (solvent A). Linear gradients of acetonitrile (solvent B: 0.1% trifluoroacetic acid, 90% acetonitrile) were developed at 1% solvent B/min from 0-5% solvent B and at 0.5% solvent B/min from 5-55% solvent B. In the second step, the major tritium-containing fraction obtained in step 1 was applied to the same column equilibrated with 0.1% ammonium trifluoroacetate (pH 6.5) (solvent E: NH4 trifluoroacetic acid). Linear gradients of acetonitrile (solvent F: 0.1% NH4 trifluoroacetic acid, 90% acetonitrile) were developed at 1% solvent F/min from 0-50% solvent F. In step 3, the 3H-labeled peptide obtained in step 2 was applied to a column equilibrated with solvent A, and eluted with a linear gradient of solvent B at 1% solvent B/min from 10- 20%.

Sequence Analysis of the 3H-Labeled Peptide-An Applied Biosys- tems Model 477A liquid-pulsed sequencer was used for the automated Edman chemical degradations. The phenylthiohydantoin amino acid derivatives were detected and identified by reversed-phase HPLC (Applied Biosystems Model 120A) using a CIS column.

RESULTS

Characterization of PLP-AMP Modification of Rho-The rate of inactivation of ATPase activity by PLP-AMP was monitored both in the presence and absence of ATP. Reac- tions were stopped at the time indicated (Fig. 1) by addition of NaBH4, and the remaining activity was assayed as described under "Experimental Procedures." Fig. l shows the protection of the rho from PLP-AMP inactivation by ATP as a function of incubation time, where the concentrations of ATP and PLP-AMP were equimolar at 50 p ~ , with rho at 100 nM. The results suggest that PLP-AMP and ATP compete for the same binding site on rho.

Inactivation as a function of PLP-AMP concentration was then performed in the presence and absence of the RNA

I I I I I

\

\

I I I I I

1 0 20 30 4 0 50

TIME, min.

FIG. 1. ATP protection of rho from inactivation. Rho was incubated at 100 nM with or without 50 p~ ATP for 15 min in 50 mM Tris-HC1 (pH 8) and 1 mM magnesium acetate. [3H]PLP-AMP was added to 50 p~ and NaBH4 added to 60 mM at various times vp to 60 min. Unincorporated analog was removed using centrifuge columns. ATPase assays were performed as described under "Exper- imental Procedures." Inactivations: E!, +ATP; and 0, -ATP.

18812 ATP Binding Site on Rho Protein

I

8 o F ' i + 5 0 100 150 200 250

3 [ H-PLP-AMP]. pM

FIG. 2. The effect of poly(C) on the inactivation of rho by PLP-AMP. The protein was incubated at 300 nM with varying concentrations of [3H]PLP-AMP in 50 mM Tris-HC1 (pH 8) for 240 min, both with and without 60 p~ poly(C). NaBH4 was added to 60 mM to stop the reaction, and after 15 min the unincorporated analog was removed using Sephadex G-50 spin columns. ATPase assays were performed as described under "Experimental Procedures." Inactiva- tions: 0, +poly(C); 0, -poly(C).

substrate poly(C). Rho protein was treated with PLP-AMP from 0 to 500 pM in reaction mixes containing 300 nM rho for 4 h. Aliquots were removed, treated with NaBH4, and analyzed for ATPase activity. The percentage of initial ATPase activity was plotted as a function of PLP-AMP concentration (Fig. 2). In the absence of poly(C), half-maximal inhibition is obtained at 26 PM PLP-AMP with maximal inhibition not occurring until 300 PM PLP-AMP. When poly(C) is added to the reaction, half-maximal inhibition is reached at 18 PM PLP-AMP with full inhibition by 100 p~ PLP-AMP. The increased potency of PLP-AMP observed in the presence of poly(C) most likely reflects a conformational change in the rho upon binding poly(C) that increases the affinity of the ATP site for ligand. We have seen a similar increase in the rate of inactivation of ATPase for the binding of the affinity analog 5'-p-fluorosulfonylbenzoyl adenosine to rho.*

Further evidence for direct competition between ATP and PLP-AMP was provided by probing the conformation of rho with bound nucleotide substrates using limited tryptic digestion. The conformation of rho has been shown to change upon binding to ATP and poly(C) and upon ATP hydrolysis (Engel and Richardson, 1984; Bear et al., 1985). ATP protects rho from total degradation by trypsin with a strong characteristic band appearing at 31 kDa. The same band is seen in digestions of rho containing both ATP and poly(C) (active hydrolysis conditions). We compared the pattern of digestion for rho plus ATP and rho labeled with PLP-AMP in the presence of poly(C). As shown in Fig. 3, both digestions resulted in the same pattern of proteolytic cleavage with the appearance of the characteristic 31-kDa band. The intensity of bands in the rho-PLP-AMP lane is reduced, presumably because this protein is unable to catalyze hydrolysis of ATP and may therefore be in a slightly different conformational state. When unmod- ified rho is digested in the absence of ATP, no band is seen at 31 kDa (Bear et al., 1985).

The stoichiometry of covalent PLP-AMP binding to rho was determined by binding [3H]PLP-AMP to rho in the

* A. J. Dombroski and T. Platt, unpublished results.

1 2

Rho -w

FIG. 3. Trypsin analysis of protein conformation. Reactions were done in 10 pl total volume and contained TDMK buffer (40 mM Tris-HC1, pH 7.5, 0.1 mM dithiothreitol, 5 mM MgClz, and 50 mM KCl), 10 pg of rho ([3H]PLP-AMP-treated rho was from the preparative labeling batch; see "Experimental Procedures"), 2 mM poly(C) and in the control reaction, 1 mM ATP. Trypsin was added to 0.01 mg/ml and the mixtures incubated for 60 min at 37 "C. The reactions were stopped by addition of 10 pl of SDS/urea sample buffer and heating to 90 "C for 5 min. Products were analyzed on a 0.1% SDS/

I is rho + ATP + poly(C). Lune 2 is PLP-AMP-labeled rho + poly(C). 10% polyacrylamide gel stained with Coomassie Brilliant Blue. Lune

Bands corresponding to rho and the 31-kDa cleavage product are marked.

"

"

2 4 0 " CC z

20"

0

1 2 3

PLP-AMP/RHO (mol/mol)

FIG. 4. Stoichiometry of ['HIPLP-AMP labeling. The residual ATPase activity was determined as in Fig. 1 and plotted as a function of moles of PLP-AMP bound per rho hexamer. Extrapola- tion to zero activity gives 3.3 mol of PLP-AMP/hexamer. 0, points from TLC assays; A, points from spectrophotometric coupled enzyme assays.

presence and absence of poly(C). The labeled protein was reduced and separated from unbound ligand using centrifuge columns. Scintillation counting of the reaction mixture, both before and after separation of free analog, allowed calculation of the moles of [3H]PLP-AMP bound per rho hexamer. The residual ATPase activity of rho, inactivated at various concentrations of PLP-AMP (0-500 PM) was plotted as a function of moles of PLP-AMP incorporatedper rho hexamer as shown in Fig. 4. The stoichiometry of covalently bound PLP-AMP at complete inactivation extrapolates to 3.3 mol/mol of enzyme. The same stoichiometry is observed when poly(C) is bound to rho. A stoichiometry of approximately three analog molecules per hexamer agrees with the results of Stitt (1988), who found that only 3 mol of ATP could bind to each rho hexamer in the presence or absence of poly(C). It has also been demonstrated that the presence of two to three mutant subunits in a hybrid hexamer blocks ATPase activity (Rich- ardson and Ruteshouser, 1986).

ATP Binding Site on Rho Protein 18813

Identification of the PLP-AMP Binding Site-Earlier studies suggested that the ATP-binding site of rho resides in the region following amino acid 164, where a high degree of sequence similarity is observed between rho and other nucleotide-binding proteins (Dombroski and Platt, 1988). Hydrox- ylamine cleavage of rho at asparagine 151 generates an amino- terminal RNA-binding fragment, N1, and a fragment, N2, containing the putative ATP-binding site (Dombroski and Platt, 1988). Our initial efforts to identify the PLP-AMP binding site focused on determining if PLP-AMP bound predominately in the N2 fragment, as expected from our predictions. We labeled pure rho with [3H]PLP-AMP, and following cleavage with hydroxylamine at Asn’” to yield the N1 (RNA binding fragment) and N2 (putative ATP-binding domain) peptides, the fragments were resolved with a 0.1% SDS/lO% polyacrylamide gel. The excised bands corresponding to N1 and N2 were solubilized and scintillation-counted to locate the region of PLP-AMP binding. This analysis was done, both with and without poly(C) present, since the conformation of the protein is dependent upon the binding of both ATP and poly(C). In the absence of poly(C), 14-35% of the label was found in N1 and 65-86% in N2. Fig. 5 shows the result of a typical hydroxylamine cleavage of rho gener- ating the 151-amino acid N1 peptide and the 268-amino acid N2 peptide. When poly(C) was present, the specificity of labeling improved such that 93% of the label appeared bound to N2.

The specific site of labeling by [3H]PLP-AMP was then determined. Preparative amounts (5-10 nmol at a time) of rho were labeled with [3H]PLP-AMP in the presence of poly(C) to assure the highest specificity. The reactions were designed so that an average of 1-1.5 mol of PLP-AMP were incorporated per rho hexamer. This ratio allows good effi- ciency of labeling while minimizing background. Unincorpor- ated reagent was separated from labeled rho using centrifuge columns. The labeled rho was digested sequentially with endoproteinase Lys-C and endoproteinase Arg-C. The peptides were separated by reversed-phase HPLC using a C4 column. In the first step, the column was equilibrated with 0.1% trifluoroacetic acid (solvent A) and eluted with a linear gradient of solvent B (0.1% trifluoroacetic acid, 90% acetonitrile). Fig. 6 shows the elution of peptides derived from the initial digest where the 3H profile is designated by the solid circles. The single labeled rho peptide eluted at 17% solvent B. Two

Amino acids

Rho+ a 419

N2 + 268

N1+ - 151

65-86 % 93 %

14-35 % 7 0%

FIG. 5. Hydroxylamine cleavage of labeled rho. Rho labeled with [3H]PLP-AMP was cleaved with hydroxylamine as described under “Experimental Procedures.” The mobilities of the resulting N1 (151 amino acids) and N2 (268 amino acids) fragments on a 0.1% SDS/10% polyacrylamide gel are indicated. The percent of label incorporated into the cleaved protein, both in the presence and absence of poly(C), is shown for N1 and N2 to the right of the gel.

additional HPLC separations were required to obtain peptide suitable for sequencing (see “Experimental Procedures”). In the second step, a single labeled peptide eluted at 20% solvent F, and in the third step, a single symmetrical peak eluted at 20% solvent B.

The amino acid sequence of the purified 3H-labeled peptide is shown in Table I. The sequence corresponds to amino acids 174-184 of rho in the proposed ATP-binding domain. Since no phenylthiohydantoin derivative was detected at cycle 8 and the derivatives at every other position were consistent with the known sequence of this region, we infer that the lysine at position 181 was the sole site of PLP-AMP binding. Identical results were obtained in two independent experiments that included PLP-AMP labeling and peptide isolation and sequencing.

DISCUSSION

We have predicted, on the basis of a high degree of sequence similarities between rho and other nucleotide-binding proteins (Dombroski and Platt, 1988), that the nucleotide-binding site for rho protein resides within a 200-amino acid domain that begins at about residue 164. In the present study, we show that an adenine nucleotide analog, PLP-AMP, can be covalently coupled to the postulated nucleotide-binding region of rho at a single site in the ATP-binding domain. The PLP- AMP molecule features a pyridoxal phosphate moiety replac- ing the y-phosphate of ATP, where the aldehyde group on the pyridoxal ring is tritiated. This analog was designed so that the reactive pyridoxal group could form a Schiff-base with lysine residues that normally interact with the phosphoryl groups of ATP (Tamura et al., 1986). Borohydride reduction of the Schiff-base yields a pyridoxamine derivative, which can then be located using peptide sequence analysis. PLP-AMP acts as a potent inhibitor of a number of enzymes that bind ATP with a high degree of specificity for modification of active site residues (Tamura et aZ., 1986, 1988; Tagaya et al., 1987; Rao et al., 1988).

Effective use of nucleotide affinity analogs for labeling ATP-binding sites requires that the analog compete with the normal nucleotide substrate for binding and that this binding does not disrupt the normal nucleotide-induced conformation of the protein. In the case of rho protein, we used competition experiments and limited proteolysis to show that PLP-AMP binds in a manner analogous to ATP. The binding of PLP- AMP to rho causes inactivation of the RNA-dependent AT- Pase activity, presumably by blocking the ATP-binding site. Competition experiments showed that the presence of ATP dramatically slows the inactivation of ATPase by PLP-AMP, suggesting direct competition between ATP and PLP-AMP for the same site. The binding of ATP to rho is known to induce a trypsin-resistant conformation with a characteristic pattern of fragments (Bear et al., 1985). Since PLP-AMP appears to compete with ATP for the ATP-binding site, we used limited trypsin digestions to show that rho, which has been modified by PLP-AMP, retains the ATP-induced conformation recognized by trypsin. Taken together, the evidence supports the contention that PLP-AMP is inactivating rho by binding to its normal nucleotide-binding site rather than to random sites in a nonspecific manner.

The specificity and rate of nucleotide binding can often be influenced by the binding of a second substrate to the protein. In the case of yeast hexokinase, PLP-AMP becomes a much more potent inhibitor when the substrate glucose is included in the reaction mixture (Tamura et aZ., 1986). The ATPase activity of rho was inactivated by PLP-AMP in both the presence and absence of the RNA substrate poly(C). However,

18814 ATP Binding Site on Rho Protein

- I I 0.6 -

I 1000 D -

0.4 5 r

0- 0 0.2 500

10 20 30 40

% E

FIG. 6. HFLC elution profile of ‘H-labeled peptides. Peptides were separated by HPLC on a C-4 column using a linear gradient^ developed at 1% solvent B/min from 0-5% solvent B and at 0.5% solvent B/min from 5- 55% solvent B. Solvent A was 0.1% trifluoroacetic acid, and solvent B was 0.1% trifluoroacetic acid in 90% acetonitrile. The recovered peptide is designated by the hatched peak. This fraction was further purified through two more columns before being sequenced (see “Experimental Procedures”). In other identical experiments, no significant radioactivity was recovered beyond 40% solvent B.

TABLE I Amino acid sequence analysis of the 3H-labekd peptide

The yields of phenylhydantoin amino acid residues in picomoles are indicated. The analysis was started with 200 pmol of 3H-labeled peptide. The actual rho sequence in this region from position 174- 184 is given. The preceding residue is ArglT3, consistent with the digestion conditions used. The site of labeling is denoted by X where no phenylhydantoin derivative was detected.

Cycle Amino acid pmol Rho 174-184

1 GlY 117 GlY 2 Leu 124 3

Leu Ile 100 Ile

4 Val 134 Val 5 Ala 147 6

Ala Pro 114

7 Pro

Pro 77 8

Pro X LY s

9 Ala 69 Ala 10 G ~ Y 52 GlY 11 LY s 18 LYS

the affinity of rho for PLP-AMP was enhanced in the presence of poly(C) such that maximal inhibition was achieved at less than 100 PM PLP-AMP when poly(C) was bound to rho but required approximately 300 PM PLP-AMP when poly(C) was omitted. We have made similar observations for the nucleotide analog 5’-fluorosulfonylbenzoyl adenosine where the binding of poly(C) appears to make the nucleotide-binding site of rho much more accessible to the analog.‘ Other evidence that the binding of poly(C) changes the rho conformation comes from studies of limited proteolytic cleavage of rho with trypsin. Rho by itself is somewhat resistant to total trypsinolysis but, if poly(C) is bound to the protein, it becomes much more accessible and is totally degraded.

The stoichiometry of analog binding can also provide an indication of a specific interaction between an affinity probe and the catalytic site of an enzyme. In the case of PLP-AMP we found that, upon complete inactivation of ATPase activity, each rho hexamer had approximately three molecules of the analog bound. The same result was obtained in the presence of poly(C). An analysis of ATP hydrolytic sites on rho has shown that the rho hexamer can bind only three molecules of ATP in the absence of poly(C) and that all three molecules are hydrolyzed upon addition of RNA. Three molecules of ADP or nonhydrolyzable ATP analogs also bind in the presence of poly(C) (Stitt, 1988). Also, subunit mixing experiments demonstrate that the incorporation of two to three

mutant subunits in a hybrid hexamer inactivates ATPase activity (Richardson and Ruteshouser, 1986). Our results agree with a stoichiometry of three ATP sites per rho hexamer, and the PLP-AMP binding characteristics for rho appear to be very similar and possibly identical to those for ATP. Since the enzyme consists of six identical subunits, the ATP sites show apparent negative cooperativity in binding ligand. An alternative explanation is that a single binding site is shared between subunits resulting in three ATP sites per hexamer.

Our two-domain model for rho tertiary structure suggests that ATP binds in a region called N2, which extends from amino acid 164 toward the carboxyl terminus (Dombroski and Platt, 1988). We would therefore expect to find all of the PLP-AMP incorporated in the N2 peptide. In order to deter- mine the specificity and general location of PLP-AMP binding, we cleaved [3H]PLP-AMP-labeled rho with hydroxylamine, both in the presence and absence of poly(C), to yield N1 (the RNA-binding domain) and N2. Without poly(C) present, 65% of the 3H was found in N2 and 35% in N1. This may be due to partial labeling of a lysine that normally interacts with the polynucleotide chain. If poly(C) is added to prevent such a reaction and enhance binding of the probe at the ATP site, the specificity is improved such that 93% of the PLP-AMP was bound to the N2. These results support our proposal that the N2 fragment contains the ATP-binding site and show that the RNA substrate, poly(C), induces a conformation that produces greater specificity of PLP-AMP binding. We have proposed a model for the conformational changes that rho goes through upon binding its substrates, based in part on these observations with PLP-AMP (Platt et al., 1988).

A number of ATP-binding domains contain a consensus sequence described by Walker et al. (1982) consisting of the peptide G-X-X-X-X-G-K-T. A similar sequence is found in rho at positions 178-185 (Fig. 7A, line I). PLP-AMP modification of rho in the presence of poIy(C) results in the specific labeling of Lyslal in this sequence (designated by an asterisk in line 1 of Fig. 7A). In other studies, PLP-AMP has been used to label the ATP-binding site on adenylate kinase (Ta- gaya et al., 1987). Although adenylate kinase lacks a lysine equivalent to Lys’*l of rho, it does have one 3 residues away in a position equivalent to Lys”. This residue (Lys*l) is the one labeled by PLP-AMP (Fig. 7A, line 2). The nucleotide binding site of the a-subunit of E . coli F1-ATPase (EcF1-a) can also be labeled with PLP-AMP (Rao et al., 1988). Al-

ATP Binding Site on Rho Protein 18815 A.

B.

FIG. 7. Homologous sequences contributing to the ATP- binding domain on rho and other nucleotide-binding proteins. AK, adenylate kinase; EcFl-a, a-subunit of E. coli F1-ATPase; MFI- 6, @-subunit of beef heart mitochondrial F1-ATPase; V-ATPase, carrot vacuolar ATPase; a-subunit of sulfolobus acidocaldarius Fl- type ATPase, archaebacteria. Identical sequences are boxed. Asterisks indicate residues that have been labeled by nucleotide affinity analogs.

though this protein also has the consensus sequence shown in Fig. 7A (see line 3), a lysine located 25 residues beyond this sequence (Lys'O') is labeled. In the ,&subunit of the beef heart mitochondrial F1-ATPase (MFl), Lyd6*, which is homologous to Lys'" of rho, is labeled by another nucleotide affinity analog, Nbf-C1 (Andrews et al., 1984) (Fig. 7A, line 4). Taken together, these results suggest that more than 1 lysyl residue may be present at the nucleotide binding sites on these proteins. The one that reacts with PLP-AMP may be the one that is best positioned to form a Schiff-base complex. Evi- dence supporting the participation of at least 2 lysyl residues at the nucleotide site of rho comes from site-directed mutagenesis studies, which show that both LyslS1 and Lys'" are important for interactions with ATP (Dombroski et al., 1988). From our model for rho's tertiary structure and from its reactivity towards PLP-AMP, we would predict that Lys'" resides at a position of potential interaction with the y- phosphoryl of ATP.

Additional evidence, Fig. 7B, compares the rho sequence 345-349 to homologous sequences in the mitochondrial p- subunit, the vacuolar ATPase, and the archaebacterial a- subunit. In this consensus sequence, Phe355 of rho corresponds to a tyrosyl residue of MFl-p that is affinity-labeled by both 2-azido-ATP (Garin et al., 1986; Cross et al., 1987) and 5'-fluorosulfonylbenzoyl inosine (Bullough et al., 1987) and is present at the catalytic site on the p-subunit of MF1 (Cross et al., 1987), and finally in archaebacterial AF1-a a tyrosine (Tyr-426) is present (Denda et al., 1988). In the carrot vacuolar ATPase sequence (Zimniak et al., 1988), a phenylalanine is also present at this position. Phenylalanine is a conservative replacement for tyrosine with regard to interaction with an adenine ring.

In summary, our results demonstrate the following: 1) PLP- AMP inactivates the ATPase activity of rho by binding in the nucleotide-binding site; 2) the stoichiometry of three sites per rho hexamer agrees with ATP-binding data; 3) the affinity for PLP-AMP is increased in the presence of the RNA substrate poly(C), supporting the notion that poly(C) induces a conformational change that enhances the binding of ligand at the catalytic site; and 4) the exact position of labeling at Lys"' occurs within the highly conserved ATP-binding consensus region and agrees with our predictions for the location and structure of rho's ATP-binding domain (Dombroski et al., 1988). In addition, the identification of a second consensus sequence suggests that the ATP site on rho may closely

resemble the model proposed for the catalytic site of FI- ATPases (Cross et al., 1987). Interactions between this ATP- binding domain and the separate RNA-binding domain are required for ATP-hydrolytic activity, but the details of such interactions remain obscure. Elucidating the mechanism of transcription termination and RNA release will likely require a better understanding of these domain interactions.

Acknowledgments-We thank Dr. James Tamura for synthesis of the PLP-AMP and Dr. Ellis Bell and Evelyn Bell for sequencing the labeled rho peptide. We also thank the members of our laboratories for helpful discussions and comments on the manuscript.

REFERENCES Andrews, W. W., Hill, F. C., and Allison, W. S. (1984) J. Biol. Chem.

259, 14378-14382 Bear, D. G., Andrews, C. L., Singer, J. D., Morgan, W. D., Grant, R.

A., von Hippel, P. H., and Platt, T. (1985) Proc. Natl. Acad. Sci.

Brennan, C. A,, Dombroski, A. J., and Platt, T. (1987) Cell 48,945-

Bullough, D. A., Verburg, J. G., Yoshida, M., and Allison, W. S.

Colman, R. F. (1983) Annu. Reu. Biochem. 52, 67-91 Cross, R. L., Cunningham, D., Miller, C. G., Xue, Z., Zhou, J.-M.,

and Boyer, P. D. (1987) Proc. Natl. Acad. Sci. U. S. A. 84, 5715- 5719

Denda, K., Konishi, J., Oshima, T., Date, T., and Yosbida, M. (1988) J. Biol. Chem. 263,6012-6015

Dombroski, A. J., and Platt, T. (1988) Proc. Nutl. Acad. Sci. U. S. A .

Dombroski, A. J., Brennan, C. A., Spear, P., and Platt, T. (1988) J.

Engel, D., and Richardson, J. P. (1984) Nucleic Acids Res. 12,7389-

Garin, J., Boulay, F., Issartel, J. P., Lunardi, J., and Vignais, P. V.

Howard, B. H., and de Crombruggbe, B. (1976) J. Biol. Chern. 261,

Lowery, C., and Richardson, J. P. (1977a) J. Biol. Chem. 252,1375-

Lowery, C., and Richardson, J. P. (1977b) J. Biol. Chem. 252, 1381-

Lowery-Goldhammer, C., and Richardson, J. (1974) Proc. Natl. Acud.

Penefsky, H. S. (1977) J. Biol. Chem. 252,2891-2899 Platt, T. (1986) Annu. Reu. Biochem. 55, 339-372 Platt, T., and Bear, D. G. (1983) in Gene Function in Prokaryotes

(Beckwith, J., Davies, J., and Gallant, J. A., eds) pp. 123-161, Cold Spring Harbor Laboratory, Cold Spring Harbor, NY

Platt, T., Brennan, C. A., Dombroski, A. J., and Spear, P. (1988) in Molecular Biology of RNA (Cech, T., ed) Vol. 94, New Series, pp. 325-334, Alan R. Liss, Inc., New York

Rao, R., Cunningham, D., Cross, R. L., and Senior, A. E. (1988) J. Biol. Chem. 263, 5640-5645

Richardson, J. P., and Conaway, R. (1980) Biochemistry 19, 4293- 4299

Richardson, J. P., and Ruteshouser, E. C. (1986) J. Mol. Biol. 189,

Shigesada, K., and Wu, C-W. (1980) Nucleic Acids Res. 8,3355-3367 Stitt, B. L. (1988) J. Biol. Chem. 263, 1130-1137 Tagaya, M., Yagami, T., and Fukui, T. (1987) J. Biol. Chem. 262,

Tamura, J. K., Rakov, R. D., and Cross, R. L. (1986) J. Biol. Chem.

Tamura, J. K., LaDine, J. R., and Cross, R. L. (1988) J. Biol. Chern.

von Hippel, P. H., Bear, D. G., Morgan, W. D., and McSwiggen, J.

Walker, J. E., Saraste, M., Runswick, M. J., and Gay, N. J. (1982)

Wu, A. M., Christie, G. E., and Platt, T. (1981) Proc. Natl. Acad. Sci.

Zimniak, L., Dittrich, P., Gogarten, J. P., Kibak, H., and Taiz, L.

U. S. A. 82, 1911-1915

952

(1987) J. Bi01. Chem. 262, 11675-11683

85,2538-2542

Biol. Chem. 263, 18802-18809

7400

(1986) Biochemistry 25, 4431-4437

2520-2524

1380

1385

Sci. U. S. A. 71, 2003-2007

413-419

8257-8261

261,4126-4133

263,7909-7912

A. (1984) Annu. Reu. Biochem. 53, 389-446

EMBO J. 1, 945-951

U. S. A. 78, 2913-2917

(1988) J. Biol. Chem. 263, 9102-9112

Documents

The ATP Binding Site on Rho Protein