Surface exposed amino acid differences between mesophilic and thermophilic phosphoribosyl diphosphate synthase Bjarne Hove-Jensen 1 and James N. McGuire 2 1 Department of Biological Chemistry and 2 Center for Enzyme Research, Institute of Molecular Biology, University of Copenhagen, Denmark The amino acid sequence of 5-phospho-a- D -ribosyl 1-diphosphate synthase from the thermophile Bacillus caldolyticus is 81% identical to the amino acid sequence of 5-phospho-a- D -ribosyl 1-diphosphate synthase from the mesophile Bacillus subtilis. Nevertheless the enzyme from the two organisms possesses very different thermal properties. The B. caldolyticus enzyme has optimal activity at 60–65 °C and a half-life of 26 min at 65 °C, compared to values of 46 °Cand60sat65°C, respectively, for the B. subtilis enzyme. Chemical cross-linking shows that both enzymes are hexamers. V max is determined as 440 lmolÆmin )1 Æmg protein )1 and K m values for ATP and ribose 5-phosphate are determined as 310 and 530 l M , respectively, for the B. caldolyticus enzyme. The enzyme requires 50 m M P i as well as free Mg 2+ for maximal activity. Manganese ion substitutes for Mg 2+ , but only at 30% of the activity obtained with Mg 2+ . ADP and GDP inhibit the B. caldo- lyticus enzyme in a cooperative fashion with Hill coefficients of 2.9 for ADP and 2.6 for GDP. K i values are determined as 113 and 490 l M for ADP and GDP, respectively. At low concentrations ADP inhibition is linearly competitive with respect to ATP. A p redicted structure of t he B. caldolyticus enzyme based on homology m odelling with the structu re of B. subtilis 5-phospho-a- D -ribosyl 1-diphosphate synthase shows 92% of the amino acid differences to be on solvent exposed surfaces in the hexameric structure. Keywords: kinetics; mesophile; nucleotide metabolism; PRPP; thermophile. The compound 5-phospho-a- D -ribosyl 1-diphosphate (PRibPP) is a central intermediate in the de novo and salvage biosynthesis of pyrimidine, purine and pyridine nucleotides as well as in the biosynthesis of the amino acids histidine and tryptophan [1,2]. In addition, methanopterin, a folate analogue involved in C1 metabolism of methanogenic archaea, is synthesized with PRibPP as an inte rmediate [3]. PRib PP is the s ubstrate for a number of phosphoribosyl- transferases which catalyse the phosphoribosylation of a variety of nucleobases to the corresponding ribonucleoside monophosphates, i.e. the formation of N-glycosidic bonds. In methanopterin biosynthesis, a carbon–carbon bond is formed to C1 of the phosphoribosyl moiety of PRibPP [3,4]. Bacterial s pecies like Bacillus subtilis and Escherichia coli contain 10 enzymes, which utilize PRibPP as a substrate [5]. The s ynthesis of PRibPP is catalysed by PRibPP synthase, which transfers the b,c-diphosphoryl group of ATP to ribose 5-phosphate (Rib5P) to produce PRibPP and 5¢-AMP [6,7] (Scheme 1 ). The r eaction proceeds by a ttack of the b- phosphate by O-1 of Rib5P [7,8]. PRibPP synthase from E. coli [9,10], Salmonella enterica serovar Typhimurium [11,12] and B. subtilis [13] requires two Mg 2+ per subunit and a P i concentration of 50 m M . S. enterica and E. coli PRib PP synthases bind A TP (as MgÆATP) before Rib5 P. The E. coli enzyme furthermore b inds free Mg 2+ before binding MgÆATP in the catalytic cycle [14]. Regulation of the activity of PRibPP synthase is achieved primarily thro ugh the inhibition by ADP or GDP. It has been shown that ADP inhibits the enzyme by binding to the allosteric site in competition with P i as well as by competing w ith ATP for the active site [9,15,16]. GDP also inhibits PRibPP synthases from Gram-negative b acteria and mammals, but to a lesser extent and by binding at the allosteric site [13,17]. PRibPP synthase is active as a homomultimer with oligomerization states ranging from hexamer to higher st ates of aggregation depending on the detection method and the source of organism [18]. In the present work we describe the charac- terization of PRibPP synthase, which is encoded by the prs gene, from the thermophile Bacillus caldolyticus and compare i t with t he enzyme from the mesophile B. subtilis. Experimental procedures Materials Ribonucleotides were obtained from Pharmacia (Uppsala, Sweden), Sigma (St. Louis, MO, USA) or Roche (Mann- Correspondence to B. Hove-Jensen, Department of Biological Chem- istry, Institute of Molecular Biology, University of Copenhagen, 83H Sølvgade, DK-1307 Copenhagen K, Denmark. Fax: +45 3532 2040, Tel.: +45 3532 2027, E- m ail: hov e@mermaid.molbio.ku.dk Abbreviations: PRibPP, 5-phospho-a- D -ribosyl 1-diphosphate; Rib5P, ribose 5-phosphate. Enzyme: 5 -phospho-a- D -ribosyl 1-di phosphate synthase or A TP: D -ribose-5-phosphate p yrophosphotransferase ( EC 2. 7.6.1). Note: A department website i s available at http://www.molbio.ku.dk Note: Dedicated to the memory of the late Professor Agnete M unch- Petersen, a fine colleagu e and a great mentor. (Received 4 August 2004, rev ised 17 September 2004, accepted 4 October 2004) Eur. J. Biochem. 271, 4526–4533 (2004) Ó FEBS 2004 doi:10.1111/j.1432-1033.2004.04412.x heim, Germany). Antibiotics, isopropyl thio-b- D -galactoside and EGTA were obtained from Sigma. Restriction endo- nucleases were obtained from Promega (Madison, WI, USA). Oligodeoxyribonucleotides were purchased from DNA Technology (A ˚ rhus, Denmark) or Hobolth DNA Syntese (Hillerød, Denmark). FPLC was performed using a Bio-Rad Bio Logic system with UV detection at 280 nm. Polyethyleneimine-cellulose coated TLC sheets were from Baker-flex (J. T. Baker, Phillipsburg, NJ, USA). Cloning and expression of the B. caldolyticus prs gene The prs gene was s ynthesized by PCR w ith pHO219 DNA [19] as the template, the oligodeoxyribonucleotides 5¢-AA GAAA GAATTC-TAGCGGAGGTCTATCATG-3¢ and 5¢-ATGTTT AAGCTTA-TTAGTCGAACAGGACGCT-3¢ as primers and DNA polymerase f rom Pyrococcus furiosus in the presence of the four deoxyribonucleoside triphos- phates. The nucleotides preceding hyphens indicate non- complementary extensions and recognition sites for the restriction endonuclease Eco RI and Hin dIII are underlined. Standard procedures were used for thermocycling in a T rio- Thermoblock (Biometra, Go ¨ ttingen, Germany). T he PCR product w as digested by Eco RI and HindIII, a nd ligated to EcoRI an d Hin dIII digested DNA of the expression vector pUHE23-2 [H. Bujard, University of Heidelberg, Germany, personal communication]. The nucleotide sequence of the insert of the r esulting plasmid, pJM1, was determined in an Abi Prism Genetic Analyser (model 310) with the Bigdye Terminator Cycle S equencing Ready Reaction Kit as recommended by the supplier (PE Applied Biosystems, Foster City, CA, USA). Purification of recombinant P Rib PP synthase of B. caldolyticus and B. subtilis The plasmid pJM1 was t ransformed i nto the PRibPP- less E. coli strain HO1986 (Dprs-4::Kan R araC am araD D(lac)U169 trp am mal am rpsL relA thi deoD gsk-3 udp supF ÔFÕ R /F lacI q zzf::Tn10), which contains no endogenous PRib PP synthase activity. HO1986 is a deriva tive of strain HO1088 [20] and was kindly p rovided b y B . N . K rath (t his institute). It is resistant to an unspecified nonlambdoid, non- P type bacteriophage. Cultures of s train HO1986/pJM1 were grown at 37 °C to an attenuance at 436 nm of 1.2–1.5 ( 3 · 10 11 cellsÆL )1 ), measured in an Eppendorf 6121 spectrophotometer. At t his time isopropyl thio-b- D -galacto- side was added to a final concentration of 50 l M ,and incubation continued for 16 h. Unless otherwise s tated the following steps were performed at 4 °C. Cells were harvested by centrifugation at 20 000 g for 2 0 min. Collected cells were resuspended in five volumes of 50 m M potassium phosphate buffer (pH 7.5), and sonicated f or 20 min (60 s bursts with 60 s pauses) followed by centrifugation at 20 000 g for 15 min. The supernatant fluid was 40 % saturated with ammonium sulphate. The precipitate was removed by centrifugation, and the supernatant fluid was 60% saturated with ammonium sulphate. The precipitate, collected by centrifugation, was redissolved in 50 m M potassium phos- phate buffer (pH 7.5) in half the original volume and dialyzed for 16 h against 2 L of 50 m M potassium phosphate buffer (pH 8.2). The dialysed enzyme preparation was applied to a Dyematrex Gel Green A column (Millipore, Bedford, MA, USA), and washed with five volumes of 50 m M potassium phosphate buffer (pH 8.2). Protein was eluted by u sing a linear gradient o ver s ix column volumes from 50 m M potassium phosphate buffer (pH 8.2) to 50 m M potassium phosphate, 300 m M potassium chloride (pH 8.2). PRib PP synthase activity eluted as two major peaks, which were pooled, dialyzed against 50 m M potassium phosphate buffer (pH 8.2), reapplied to the same column, and eluted under the same conditions as before. The larger of two activity peaks (fraction A) was further dialysed against 50 m M potassium phosphate buffer (pH 8.2) eluted isocrat- ically through a Pharmacia Superose 12 10/30 gel filtration column using an FPLC instrument at room temperature. PRib PP synthase activity eluted as three or four peaks. The largest was chosen for further study. The final enzyme fraction was greater than 95% pure as determined by SDS/ PAGE and staining i n Coomassie Brilliant Blue. T he enzyme was s tored i n 5 0% glycerol in aliquots at )80 °C. Recombinant B. subtilis PRibPP synthase was isolated from cells overexpressing the prs gene essentially as described previously [21] with a modification of the final anion exchange step as follows. The enzyme, dissolved in 50 m M potassium phosphate buffer (pH 7.5) was a pp lied to a 20 mL anion exchange Hiload Q-Sepharose column (Pharmacia), previously equilibrated with t he same buffer. PRib PP synthase was e luted b y applying a s alt g radient of 0% Salt Buffer [50 m M potassium phosphate buffer (pH 7 .5)] to 100% Salt Buffer [1 M sodium chloride in 50 m M potassium phosphate buffer (pH 7.5)] at a rate of 2mLÆmin )1 over 60 min. The gradient w as an initial linear increase from 0 t o 20% Salt Buffer, f ollowed by a hold for 40 mL and an increase to 35% Salt Buffer over approxi- mately 120 mL and finally a raise to 100% Salt Buffer. PRib PP synthase eluted at a sodium chloride concentration of approximately 0.30 M . The fractions with highest purity evaluatedbyassayofPRibPP synthase activity and by SDS/PAGE were pooled and dialyzed against 50 m M potassium phosphate buffer (pH 7.5). The enzyme was stored refrigerated [22]. Protein content was determined by the bicinchoninic acid procedure (Pierce Chemical Company, Rockford, IL, USA) as described previously with BSA as the standard [23]. MALDI-TOF mass spectrometry analysis was performed by the School of Chemical Sciences Mass Spectrometry Center, University of Illinois, Urbana-Champaign, IL, Scheme 1. Reaction catal y sed by PRib PP. Ó FEBS 2004 Bacillus caldolyticus PRibPP synthase (Eur. J. Biochem. 271) 4527 USA. Amino acid sequencing by automated Edman degradation was performed by the Department of Protein Chemistry, Institute of Molecular Biology, University of Copenhagen, Denmark. Assay of P Rib PP synthase activity The standard reaction buffer consisted o f 50 m M Tris/HCl, 50 m M potassium phosphate, 2.0 m M EGTA (pH 8.5, adjusted at 65 °C). The standard reaction contained 2.0 m M (10 G BqÆmol )1 )[ 32 P]ATP[cP] (prepared a s des- cribed previously [24]), 5.0 m M Rib5P,5.0m M magnesium chloride. Unless otherwise indicated the Mg 2+ concentra- tion w as 3 .0 m M in excess of the r ibonucleoside t riphos- phate concentration. In analyses of inhibition by ADP and in determination of K m for ATP and R ib5P, a buffer without EGTA was used. For the P i or sulphate dependence analysis, the en zyme was d iluted in 5 0 m M Tris/HCl buffer (pH 8 .2) containing BSA (2 gÆL )1 ) without prior dialysis. The reaction bu ffer for these studies was 50 m M Tris/HCl (pH 8 .5, adjusted at 65 °C). In all cases, the assay buffer with ATP, Rib5P and magnesium chloride present was prewarmed for 2 m in at the desired temperature and reaction initiated by the addition of enzyme. The enzyme had b een previously diluted in 50 m M potassium phosphate buffer (pH 8.5, adjusted at 20 °C) containing BSA (2 mg ÆmL )1 ) and prewarmed for 2 min at 20 °C. Reaction was performed for 3 min at three different enzyme dilutions. The reaction was terminated by mixing the s ample ( 10 lL) with 0.33 M formic acid (5 lL) and applying the 1 5 lLtoa polyethyleneimine-cellulose coated TLC sheet. The chro- matogram was developed i n 0.85 M potassium phosphate, which had been previously titrated to pH 3.4 with 0.85 M phosphoric acid. The radioactive content in individual spots wasdeterminedinaPackardInstant Imager (model 2024). B. subtilis PRibPP synthase activity was assayed by the same p rocedure. Enzyme activity is expressed as lmolÆmin )1 Æmg protein )1 . Kinetic analysis Results of initial velocity determinations, which were averages of at least three d eterminations, were fitted t o the following equations using the program ULTRAFIT (version 3.0.5, Biosoft, Cambridge, UK). E quation 1 is the Micha- elis–Menten equation for hyperbolic substrate saturation kinetics, w hereas Eqn 2 is the rate equation for a sequential mechanism. For competitive and noncompetitive inhibition the initial velocities were fitted to Eqn 3 a nd 4, respectively [25]. Equation 5 was used to estimate the H ill coefficient in inhibition studies. v ¼ V app S K m þ S ð1Þ v¼ V max ½ATP½Rib5P K ATP ½Rib5Pþ K Rib5P ½ATPþ K iATP K Rib5P þ½ATP½Rib5P ð2Þ v ¼ V app S K m 1 þ I K is  þ S ð3Þ v ¼ V app S K m 1 þ I K is  þ S1þ I K ii  ð4Þ v ¼ V max 1 þ I K i  n ð5Þ where v is the initial v elocity, V app is the a pparent maximal velocity, K m is the a pparent Michaelis–Menten c onstant for the varied substrate S, V max is the maximal velocity, K ATP and K Rib5P are the Michaelis–Menten constants for ATP and R ib5P, respectively. K iATP is the dissociation constant for ATP, K is and K ii are inhibitor constants f or the inhibitor I obtained from t he effect on slopes a nd intercept, respect- ively, K i is the inhibitor constant for the substrate S, and n is the Hill coefficient. Chemical cross-linking Cross-linking was performed with bis(sulphosuccinimidyl) suberate (Pierce) at a concentration of 1.8 m M in 20 m M potassium phosphate buffer (pH 8.3) with a protein concentration range of 91–910 lgÆmL )1 (equivalent to 3–30 l M PRibPP sy nthase subunit). The reaction (10 lL) was incubated at room temp erature f or 30 min f ollowed by quenching with an equal volume of 100 m M Tris/HCl (pH 8 .5). Samples were analysed by SDS/PAGE (10% acrylamide). Molecular modelling Molecular modelling was based on the coordinates of the crystal form of B. subtilis PRib PP synthase with sulphate present [26]. An unresolved loop, RPKPNVAEVM(199– 208), w as added t o this s tructure using HOMOLOGY software (Biosym/Msi, San Diego, C A, USA) and minimized using the manufacturer’s suggested settings. The resulting struc- ture was u sed as a template to build a m odel of B. caldo- lyticus P RibPP synthase by using the program HOMOLOGY . The residues that deviated from the B. subtilis sequence were minimized to remove any gross errors. The whole structure w as subjected to r epeated rounds of minimization and molecular dynamics using the DISCOVER module (Biosym/Msi) again using the manufacturer’s suggested settings. The final root-mean-square deviation between the two backbones was 0.005. Analysis of the structure with PROSTAT in HOMOLOGY and VERIFY 3- D [27] revealed only two problem areas. The first was the loop RQDRKAR- SRN(99–108), which had some non-ideal torsion angles, but they arose from the analogous loop in the original structure. The other problem was the constructed loop (amino acids residues 197–206), which i s flexible anyway , so s mall errors were of little consequence. Graphics were made by using the program INSIGHT (Biosym/Msi). Results Purification and characterization B. caldolyticus PRibPP synthase was purified to homo- geneity by ammonium sulphate precipitation, triazyl dye 4528 B. Hove-Jensen and J. N. McGuire (Eur. J. Biochem. 271) Ó FEBS 2004 chromatography and gel filtration. An approximate subunit mass was determined by MALDI-TOF mass spectrometry as 34 496.8 Da and agreed within 1% deviation with the value, 34 296 Da, calculated f rom the deduced amino a cid sequence. N-terminal sequencing r evealed the sequence Ser- Asp-Xaa-Gln-His-Gln-Leu-Lys-Leu-Phe, which is in agree- ment with the deduced amino acid sequence and shows t hat the initial methionine has been removed. Comparison of the nucleotide sequences of the insert o f p JM1 a nd the original insert of pHO219 (GenBank and EBI Data Bank accession number X83708) revealed three discrepancies. Lys289 and Arg294 were found to be glutamic acid and alanine, respectively. The c odon for Val292 was found to be GUG and not GUC as published originally [19]. Temperature and pH dependency Temperature d ependency of t he enzymatic activity of B. caldolyticus PRibPP synthase was determined in the range 40–75 °C using the standard reaction buffer. A bell shaped profile was obtained w ith maximal activity at 60 °C (data not shown). In all of the experiments reported here, the reactions we re initiated with enz yme that had been prewarmed at room temperature. Initiating the r eaction with Rib5P gave an optimum at 60–65 °C. This suggests that the presence of the substrate ATP prior to initiating the reaction may stabilize the enz yme. The optimal temperature a ppeared to vary between 60 and 6 5 °C among enzyme preparations. For comparison the tem- perature dependency o f t he enzymatic activity o f B. subtilis PRib PP synthase was determined as well and revealed an optimal temperature of 46 °C. The s tability of the two enzymes at 65 °C was determined. A dramatic difference was observed. The half-life of the B. caldoly ticus enzyme was 2 6 min, w hereas that of the B. subtilis enzyme was 60 s (data not shown). The optimal pH of B. caldolyticus P RibPP synthase was 8.25–8.75 when the activity was assayed at 6 5 °C. The activity dropped to 8 0% of maximal a t p H 9.5 and t o o nly about 25% at pH 6.5 compared to the activity at pH 8.50. At least in part this reduction in enzyme activ ity at higher pH may be caused by the formation of magnesium– phosphate complexes, and, thus, cause a depletion of Mg 2+ . An i dentical pH optimum was obtained w ith B. subtilis PRibPP synthase when activity was assayed at 37 °C. P i and metal ion requirements In the a bsence of added P i , which corresponds to a minimal P i concentration of 12.5 l M intheassay,theenzymewas weakly ac tiv e (4 .8% of maximum). As the P i concentration was raised, the enzyme gained activity and re ached a maximum at 50 m M , whereas it was slowly reduced to 58% at 120 m M and 17% at 200 m M . The enzyme could use sulphate ion in p lace of P i but on ly at about 30% of maximal activity at a concentration of 0.50 M .At50m M , the optimal concentration for P i , sulphate was hardly activating (5% of maximal activity), whereas 1 M sulphate was strongly inhibitory (5% of maximal activity). The enzyme clearly preferred Mg 2+ as the metal ion, but could use Mn 2+ ,Zn 2+ ,Cd 2+ or Cu 2+ . The activity in the presence of Mn 2+ was about 30% of the activity determined inthepresenceofMg 2+ , while the activity in the presence of Zn 2+ ,Cd 2+ or Cu 2+ was only 5–10% of the activity determined in the presence of Mg 2+ .Itislikelythattwo Mg 2+ were bound per subunit, one in complex with ATP and one bound at the active site, because activity increased as the Mg 2+ concentration w as raised above the ribo- nucleoside triphosphate concentration. No activity was observed in the presence of Ca 2+ ,Fe 2+ ,Co 2+ or Ni 2+ . Kinetic analysis It was necessary to use an excess of Mg 2+ over ATP, similar to what has been observed for other PRibPP synthases. Even under these conditions ATP exerted substrate inhibi- tion at concentrations above 1 m M . However, results of initial v elocity v s. the concentration of A TP or Rib5P were found to follow Michaelis–Menten kinetics a t ATP con- centrations below 0.8 m M . I n double reciprocal plots of the data, intersecting lines indicated that the reaction followed a sequential mechanism (Fig. 1). The data were fitted to Eqn 2 and the following values were obtained: K ATP 310 ± 110 l M , K Rib5P 530 ± 140 l M and V max 440 ± 69 lmolÆmin )1 Æmg protein )1 . Assay of enzyme activity in the presence of a variety of nucleotides showed that 5¢-AMP, GTP, 5¢-GMP and CTP, each at a concentration of 5.0 m M , had little or no Fig. 1. Reaction mechanism o f PRibPP synthase and determination o f kinetic constants. Activity was determined as described in Experimental procedures. The magnesium ch loride c oncentration was 3.0 m M over the ATP concentratio n. 1/v is expressed as lmol )1 ÆminÆmg protein. Double reciprocal plots of initial velocity vs. Rib5P at five concen- trations of ATP are shown. The concentration of Rib5P was varied from 0.2 to 0.8 m M in the presence of different concentrations of ATP: e,0.1m M ; n,0.2m M ; h,0.4m M ; ·,0.6m M ;ors,0.8m M .Lines represent fitting of the data t o Eqn 2. Ó FEBS 2004 Bacillus caldolyticus PRibPP synthase (Eur. J. Biochem. 271) 4529 effect on the enzyme activity, as activity varied from 92 to 109% of the activity obtained in the absence of these nucleotides. The activity in the presence of 5.0 m M UTP was only 20% of that in the absence of UTP, indicating significant inhibition. Only ADP and GDP showed significant inhibition at physiologically relevant concen- trations, less than 1% residual activity in the presence of 1m M ADP or 5 m M GDP. As e xpected from these results, GDP was a less efficient inhibitor (Fig. 2). Inhibition by ADP as well as by GDP was strongly cooperative, with Hill coefficients for ADP and GDP determined as 2.9 ± 0.1 and 2.6 ± 0.1, respectively. The apparent K i values determined under these assay condi- tions (3.0 m M ATP) were 113 ± 1 l M for ADP and 490 ± 9 l M for GDP. Inhibition with ADP at various ATP c oncentrations was analysed. In the inhibitor concentration range employed here, 0.06–0.18 m M , ADP was a linear competitive inhibitor of ATP saturation (Fig. 3). Analysis of the data with respect to noncompetitive inhibition (Eqn 4) failed t o give a satisfying fit. Quaternary structure Chemical cross-linking of PRibPP synthase followed by SDS/PAGE revealed two major bands of M r 220 000 and 100 000 (Fig. 4). The monomer behaved as a 36 000 M r polypeptide. This result indicates the formation of hexa- mers and trimers. In addition some higher order oligomers were seen. I nterestingly, n o o r very little dimer was observed. Higher order oligomers o f B. caldolyticus PRib PP synthase were consistently seen by gel fi ltration, and they possessed significant a ctivity but not as high as the hexamer (data not shown). Identical results, i.e. Fig. 4. The quaternary structure of PRibPP synthase. Cross-linking was performed as described in Experimental procedures. Lanes 1 and 7 contain M r standards (Bio-Rad): I, M r 208 000; II, M r 115 000; III, M r 79 500; IV, M r 49 500; V, M r 34 800. Lane 2 contains untreated enzyme (0.9 lg app lied in gel). Lanes 3–6 contain cross-linked enzyme. The amount of protein loaded in e ach lane of t he ge l: lane 3, 4.5 lg applied in gel; l ane 4, 2 .3 lg; lane 5, 1.1 lg; lane 6, 0.5 lg. Fig. 3. Inhibition of B. caldolyticus PRibPP synthase activity by ADP. Activity was determined as described in Experimental procedures. The magnesium chloride c once ntration exce eded total nucleotid e con cen- tration by 3.0 m M .1/v is expressed as lmol )1 ÆminÆmg protein. Double reciprocal plots of initial velocity vs. ATP at six concentrations of ADP are shown. The concentration of ATP was varied from 0.05 to 0.80 m M in the presence of different concentrations of ADP: ,,0m M ; s,0.06m M ; h,0.09m M ; n,0.12m M ; e,0.15m M ,or· ,0.18m M . Lines represent fitting o f the data to Eqn 3. Fig. 2. Inhibition by ADP and GDP of B. caldo lyticus PRibPP syn- thase activity. A ctivity was determined as described in E xperimental procedures wit h ATP a nd Rib 5P c onc entrations o f 3.0 and 5.0 m M , respectively, a nd Mg 2+ exceeding the total ribon ucleotid e co ncentra- tion by 3.0 m M . The specific activity of the enzyme was 400 lmolÆ min )1 Æmg protein )1 (determined at 65 °C). Ribonucleoside diphos- phate varied from 0 to 5 m M . Curves represent fitting of the entire data sets to Eq n 5 . h,ADP;s,GDP. 4530 B. Hove-Jensen and J. N. McGuire (Eur. J. Biochem. 271) Ó FEBS 2004 chemical cross-linking products with M r of 220 000 and 100 000 were obtained with B. subtilis P RibPP synthase as well. Model structure An alignment of B. caldolyticus and B. subtilis PRibPP synthases is shown in Fig. 5. The amino acid sequences of the two polypeptides are 81% identical. The crystal structure of B. subtilis PRibPP synthase has been solved with two ADP molecules per monomer, one bound at the active site and one bound in an allosteric cleft. The structure has also been solved with sulphate bound in the allosteric cleft and in place of the phosphate group of Rib5P in the a ctive site [26]. A model based on the sulphate structure was constructed using a homology- based method (Fig. 6). All of t he amino acids of the active sites as well as those of the monomer–monomer contact surfaces were identical in the two proteins. The only exceptions were Leu70 and Lys199, which are isoleucine and a rginine, respec tively, i n t he B. subtilis enzyme. In addition, all of the amino acids involved in allosteric regulation by ADP were conserved [28]. Inter- estingly, of the 59 altered amino acid residues, 54 (i.e. 92%) were solvent exposed in the hexameric struc- ture. The five buried residues of B. caldolyticus PRibPP synthase were as follows, with the corresponding amino acid of B. subtilis PRibPP synthase given in parenthesis: Ile43 (Val), Val56 (Cys), Leu70 (Ile), Asn108 (Glu) and Val115 (Phe). Consistent with the surface lo cation of the altered amino acids were hydrophobicity surface maps of monomers from the two Bacillus PRibPP synthases. These revealed a n increase in polar surface area in the B. caldolyticus enzyme compared to that of B. subtilis (data not shown). Discussion It is apparent that the thermophilic version of the Bacillus enzyme possesses the sam e basic s tructure as its m esophilic relative and that both enzymes function by the same mechanism. In particular all of t he residues identified as important in c atalysis a nd allosteric regulation as well as in monomer–monomer contact of the B. subtilis PRibPP synthase were retained in the B. caldolyt icus enzyme with the t wo exceptions of conservative replacements mentioned above [22,26,28–30]. T hus, the mechanism of catalysis and regulation appe ars to b e s imilar for the two enzymes. The two enzymes differed primarily in their thermal properties. The origin of this d ifference is at present unknown. In general, the number of individual amino a cids varied little among the two enzymes. Exceptions were asparagine, alanine, glycine a nd methionine. Analysis of the number o f asparagine and glutamine residues revealed a bias against these thermolabile amino acids. Both enzymes contained 10 glutamine residues. B. subtilis PRibPP synthase con- tained 17 asparagines c ompared to 11 of the B. caldolyticus enzyme. Curiously, however, four of these 17 asparagines of the B. subtilis enzyme were replaced by glutamines in the B. caldolyticus enzyme. T hus, the A sn + G ln content may Fig. 5. Alignment of B. ca ldolyticus and B. subtilis PRibPP synthase amino acid sequences. Bc, B. caldolyticus; Bs, B. subtilis. b-Sheets are shown as yellow letters, a- helices as blue letters. Residues that are different among the two sequenc es, are shown as red letters in the B. caldolyticus sequence. Fig. 6. Model structure of hexameric B. c aldo lyticus PRibPP synthase. One dimer is shown with grey shading, a second dimer with green and purple shading and a third dimer with blue a nd yellow shading. Red atoms i ndicate amino acids that differ among B. c aldolyticus and B. subt ilis PRibPP s ynthases (detailed i n Fig. 5). Ó FEBS 2004 Bacillus caldolyticus PRibPP synthase (Eur. J. Biochem. 271) 4531 be of significance for the enhanced thermostability of B. caldolyticus PRibPP synthase, similar to what has been shown for ce rtain enzymes from hyperthermophilic organ- isms [31]. Furthermore, the B. caldolyticus enzyme con- tained 33 alanines compared to 28 in the B. subtilis enzyme as well as one additional change to alanine. The amino acids of the B. subtilis enzyme at positions corresponding to these six alanines were serine, glutamate, valine, lysine and two glycines. It is possible therefore th at these alanines contribute compactness to t he thermophilic enzyme. T he glycine content of the B. caldolyticus enzyme was three less than that of the B. subtilis enzyme. In the former enzyme the corresponding amino acids were cysteine, alanine and serine. Therefore, it is possible that the thermophilic enzyme is more rigid in structure than the mesophilic enzyme. Finally, t he B. caldolyticus enzyme contains four more methionines than the B. subtilis enzyme, corresponding to proline, valine, isoleucine and glutamine in the latter e nzyme. The signifi- cance of this difference, if any, remains unknown. It is possible that subtle changes along the primary structure together contribute to the increased thermostability [32]. Altogether the modelling of B. caldolyticus P RibPP syn- thase indicated that the altered amino acids were primarily located o n t he surface of t he hexameric protein. Apart f rom t he thermal properties, the two enzymes also differ widely in their regulation. We determined K i values for ADP and GDP, in the presence of 3.0 m M ATP and 5.0 m M Rib5P, as 113 and 490 l M , respectively, for the B. caldolyticus enzyme. In comparison, the concentration of ADP and GDP resulting in 50% inhibition, and determined at identical s ubstrate c oncentrations as before, w ere g reater than 1 m M and greater than 5 m M , respectively, for the B. subtilis enzyme [12]. Similarly, UTP inhibited the B. caldolyticus to a higher extent, 20% residual activity, than the B. subtilis enzyme, 80% residual activity. Again, determined under identical assay conditions, other kinetic values differed by approximately two-fold or less. A summary of the properties of t he two enzymes is given in Table 1. Acknowledgements We are grateful to M. Willem oe ¨ s for discussions and for carefully reading the manu script, to B . N . Krath for providin g strain H O1986 and for assistance with analysis of kinetic data. We wish to thank T. D. Hansen for excellent technic al assistance. Financ ial support w as obtained from the Danish Natural Science R esearch Council. References 1. Hove-Jensen, B. (1988) Mutation in the phosphoribosylpyro- phosphate synthetase gene (prs) th at results in simultaneous requirements for purine and pyrimidine nucleosides, nicotinamide nucleotide, histidine and tryptophan in Escherichia coli. J. Bacte- riol. 170 , 1148–1152. 2. Hove-Jensen, B. (1989) Phosphoribosylpyroph osphate (PRPP)- less mutants of Escherichia coli. Mol. Microbiol. 3, 1487– 1492. 3. White, R.H. (1996) Biosynthesis of methanopterin. B ioc h emi stry 35, 3 447–3456. 4. Scott, J.W. & Rasche, M .E. (2002 ) P urification, o verproduction, and partial characterization of b-RFAP synthase, a key enzyme in the m ethanopterin bio synthesis p athway. J. Bacteriol. 184 , 4442– 4448. 5. Jensen, K.F. (1983) Metabolism of 5-phosphoribosyl 1-pyro- phosphate (PRPP) in Escherichia coli and Salmonella typhimu- rium.InMetabolism of Nucleotides, Nucleosides and Nucleobases in Microorganisms (Munch-Petersen, A., ed.), pp. 1–25. Academic Press, Lo ndon. 6. Kornberg, A., Lieberman, I. & Simms, E.S. (1955) Enzymatic synthesis and properties of 5- phospho ribosylpyrophosph ate. J. Bi ol. Chem. 215, 389–402. 7. Khorana, H.G., Fernandes, J.F. & Kornberg, A. (1958) Pyr- ophosphorylation of ribo se 5-phosphat e in the en zymatic synt h- esis of 5-phosphorylribose 1-pyrophosphate. J. Biol. Chem. 230, 941–948. 8. Miller, G.A. Jr, Rosenzweig, S. & Switzer, R.L. (1975) Oxygen-18 studies of the mechanism of pyrophosphoryl group transfer cata- lyzed by phosphoribosylpyropho sphate synthetase. Arch. Bio- chem. B iophys. 171, 732–736. 9. Hove-Jensen, B., Harlow, K.W., King, C.J. & Switzer, R.L. (1986) Phosphoribosylpyrophosphate synthetase of Escherichia coli. Properties of the purified enzyme and primary structure of the prs gene. J. Biol. Chem. 261, 6 765–6771. 10. Willemoe ¨ s, M. & Hove-Jensen, B . (1997) Binding of divalent magnesium by Escherichia coli phosphoribosyl diphosphate syn- thetase. Biochemistry 36, 507 8–5083. 11. Switzer, R.L. (1969) Regulation and mechanism of phospho- ribosylpyrophosphate synthetase. I. Purification and properties of the enzyme from Salmonella typhimurium. J. Biol. Chem. 244, 2854–2863. 12. Switzer, R.L. (1971) Regulation and mechanism of phospho- ribosylpyrophosphate s ynthetase. III. Kinetic studies of the reac- tion me chanism. J. Biol. Chem. 246 , 2447–2458. 13. Arnvig, K., Hove-Jensen, B. & Switzer, R.L. (1990) Purification and properties of phosphoribosyl-diphosphat e synthetase from Bacillus subtilis. Eur. J. Biochem. 192, 195–200. 14. Willemoe ¨ s, M., Hove-Jensen, B . & La rsen, S. (2000) Steady s tate kinetic m odel for the bin ding of substrates an d allosteric effectors to Escherichia coli ph osphoribosyl-diphosphate synthase. J. Biol. Chem. 275 , 35408–35412. 15. Switzer, R.L. & Sogin, D.C. (1973) Regulation and mechanism of phosphoribosylpyrophosphate synthetase. V. Inhibition by end product and regu lation by adenosine d iphosp hate. J. Biol . Chem. 248, 1 063–1073. 16. Gibson, K.J., Schubert, K.R. & Switzer, R.L. (1982) Binding of the substrates and the allosteric inhibitor adenosine 5¢-diphos- Table 1. Comparison of properties of Bacillus PRibPP synthases. Values for pH optimum, K m , V max and K i of the B. caldolyticus enzyme we re determined at 65 °C. Values for K m , V max and K i of the B. subtilis enzyme were determined at 37 °C[13]. Source of PRibPP synthase No. of amino acids Optimal K m V max (lmolÆ min )1 Æmg protein )1 ) K ADP i (m M ) Predominant oligomer Temp. (°C) pH ATP (m M ) Rib5P (m M ) B. caldolyticus 314 60–65 8.5 0.31 0.53 440 0.113 Hexamer B. subtilis 316 46 8.5 0.66 0.48 250 > 1 Hexamer 4532 B. Hove-Jensen and J. N. McGuire (Eur. J. Biochem. 271) Ó FEBS 2004 phate t o phosphoribo sylpyropho sphate synthet ase fro m Salm on- ella typhimurium . J. Bi ol. Chem. 257, 2391–2396. 17. Sonoda, T., Ishiharu, T., Ishijima, S., Kita, K., Ahmad, I. & Tatibana, M. (1998) Rat liver pho sphoribosylpyro phosphate synthetase is activated by free Mg 2+ in a manner that overcomes its inhibition by nucleotides. Biochem. Biophys. A cta 1387, 32–40. 18. Becker, M.A. (2001) Phosphoribosylpyrophosphate synthetase and the regulation of ph osph oribosylpyrop hosphate p rodu ction in human c ells. Prog. Nucle ic Acid R es. Mol. Biol. 69, 115–148. 19. Krath, B.N. & Hove-Jensen, B. (1996) Bacillus caldolyticus prs gene encoding phospho ribosyl-diphosph ate synthase. Gene 176, 73–79. 20. Krath, B.N. & Hove-Jensen, B. (2001) Class II recombinant phosphoribosyl diphosphate synthase from spinach. Phosphate- independence and diphosphoryl donor s pecificity. J. Biol. Chem. 276, 17851–17856. 21. Bentsen, A K., Larsen, T.A., Kadziola, A., Larsen, S. & Harlow, K.W. (1996) Overexpression of Bacillus subtilis phospho- ribosylpyrophosphate synthetase and crystallization and pre- liminary x-ray characterization of t he free enzyme and its substrate-effector complex. Proteins 24 , 238 –246. 22. Bentsen, A K. (1999) Structure–function relationships in Bacillus subtilis PRPP synthetase. PhD Thesis, University of Copenhagen, Denmark. 23. Smith, P.K., Krohn, R.I., Hermanson, G.T., Mallia, A.K., Gartner, F.H., Provenzano, M.D., Fujimoto, E.K., Goeke, N.M., Olson, B .J. & Klenk, D.C. (1985) Measurement of protein using bicinchoninic acid. Anal . Biochem. 150 , 76–85. 24. Jensen, K.F., Houlberg, U. & Nygaard, P. (1979) Thin-layer chromatographic methods to isolate 32 P-labeled 5-phospho - ribosyl-a-1-pyrophosphate (PRPP): Determination of cellular PRPP pools and assay of PRPP synthetase activity. Anal. Bio- chem. 98, 2 54–263. 25. Cleland, W.W. (1963) The kinetics of e nzyme-catalyzed reactions with two or more s ubstrates or products. II. Inhibition: N o men- clature an d theory. Bio chim. Biophys. A cta 67, 1 73–187. 26. Eriksen, T.A., Kadziola, A., Bentsen, A K., Harlow, K.W. & Larsen, S. (2000) Structural basis for the function of Bacillus subtilis phosphoribosyl-pyrophosphate synthetase. Nat. Struct. Biol. 7, 303–308. 27. Luthy, R., Bowie, J.U. & Eisenberg, D. (1992) Assessment of protein models with t hree -dimensional profi les. Nature 356, 83–85. 28. Nygaard, F.B. (2001) The molecular mechanism of catalysis and allosteric regulation in the phosphoribosyldiphosphate synthase from Bacillus subtilis. PhD Thesis, University of Copenhagen, Denmark. 29. Eriksen, T.A., Kadziola, A. & Larsen, S. (2002) Binding of cations in Bacillus subtilis phosphoribosyldiphosphate synthetase an d their role in c atalysis. Protein Sci. 11, 271–279. 30. Hilden, I., Hove-Jensen, B. & Harlow, K .W. (1 995) Ina ctivation of Escherichia coli phosphoribosylpyrophosphate synthetase by the 2¢,3¢-dialdehyde derivative of ATP. Identification of active site lysines. J. Biol. Chem. 270, 20730–20736. 31. Vieille, C., Epting, K.L., Kelly, R.M. & Zeikus, J.G. (2001) Bivalent cations and amino-acid composition contribute to the thermostability of Bacillus licheniformes xylose isomerase. Eur. J. Biochem. 26 8, 6291–6301. 32. Petsko, G.A. (2001) St ructural basis of thermostability in hyper- thermophilic proteins, or ÔThe re is m ore than one way to skin a catÕ. Methods En zymol. 334, 469– 478. Ó FEBS 2004 Bacillus caldolyticus PRibPP synthase (Eur. J. Biochem. 271) 4533 . Surface exposed amino acid differences between mesophilic and thermophilic phosphoribosyl diphosphate synthase Bjarne Hove-Jensen 1 and James. 5-phospho-a- D -ribosyl 1 -diphosphate synthase from the thermophile Bacillus caldolyticus is 81% identical to the amino acid sequence of 5-phospho-a- D -ribosyl 1 -diphosphate synthase