Effect of pH on the structure and stability ofBacillus circulans ssp.alkalophilus phosphoserine aminotransferase: Thermodynamic and crystallographic studies

Effect of pH on the Structure and Stability of Bacilluscirculans ssp. alkalophilus PhosphoserineAminotransferase: Thermodynamic andCrystallographic StudiesEvangelia G. Kapetaniou,1 Angelos Thanassoulas,2 Anatoly P. Dubnovitsky,1 George Nounesis,2 andAnastassios C. Papageorgiou1*1Turku Centre of Biotechnology, University of Turku and Åbo Akademi University, Turku, Finland2National Centre for Scientific Research “Demokritos,” Aghia Paraskevi, Greece

ABSTRACT pH is one of the key parametersthat affect the stability and function of proteins. Wehave studied the effect of pH on the pyridoxal-5�-phosphate-dependent enzyme phosphoserine amino-transferase produced by the facultative alkaliphileBacillus circulans ssp. alkalophilus using thermody-namic and crystallographic analysis. Enzymatic ac-tivity assay showed that the enzyme has maximumactivity at pH 9.0 and relative activity less than 10%at pH 7.0. Differential scanning calorimetry andcircular dichroism experiments revealed variationsin the stability and denaturation profiles of theenzyme at different pHs. Most importantly, releaseof pyridoxal-5�-phosphate and protein thermal dena-turation were found to occur simultaneously at pH6.0 in contrast to pH 8.5 where denaturation pre-ceded cofactor’s release by approximately 3°C. Tocorrelate the observed differences in thermal dena-turation with structural features, the crystal struc-ture of phosphoserine aminotransferase was deter-mined at 1.2 and 1.5 Å resolution at two differentpHs (8.5 and 4.6, respectively). Analysis of the twostructures revealed changes in the vicinity of theactive site and in surface residues. A conforma-tional change in a loop involved in substrate bind-ing at the entrance of the active site has beenidentified upon pH change. Moreover, the number ofintramolecular ion pairs was found reduced in thepH 4.6 structure. Taken together, the presentedkinetics, thermal denaturation, and crystallographicdata demonstrate a potential role of the active sitein unfolding and suggest that subtle but structur-ally significant conformational rearrangements areinvolved in the stability and integrity of phospho-serine aminotransferase in response to pH changes.Proteins 2006;63:742–753. © 2006 Wiley-Liss, Inc.

Key words: phosphoserine aminotransferase; alka-line adaptation; protein stability; pyri-doxal-5�-phosphate; X-ray crystallogra-phy; thermal denaturation; circulardichroism

INTRODUCTION

Protein stability and activity are affected by variousparameters, such as temperature, salinity, and pH. De-tailed understanding of the effect of each of these factorson protein function is necessary for the successful use ofproteins in industrial and biotechnological processes.1

Although much is known about the adaptation of proteinsto elevated temperatures and salinity, similar studies onthe effect of pH are limited. Phosphoserine aminotransfer-ase is a PLP-dependent enzyme that catalyzes the reversibleamino group transfer from L-glutamate to 3-phosphohydroxy-pyruvate to form 2-oxoglutarate and L-phosphoserine, respec-tively (Fig. 1). This reaction is the second step of the“phosphorylated pathway” responsible for the biosynthesis ofL-serine,2,3 a key source of one-carbon units in the living cells.PSAT is a member of the large �-family of PLP-dependentenzymes that consists of four subgroups (I–IV4). PSATtogether with serine isopenicillin-N-epimerase and the nifSgene product belong to the subgroup IV characterized by lowsequence identity with members of any other subgroup ofaminotransferases. Furthermore, PLP-dependent enzymeshave been classified into five fold types based on sequenceand structural considerations.5 Phosphoserine aminotrans-ferase belongs to type I fold along with the other �-family

Abbreviations: AMG, �-methyl-L-glutamate; BCIR, Bacillus circu-lans ssp. alkalophilus; BALC, Bacillus alcalophilus; CD, circulardichroism; DSC, differential scanning calorimetry; PLP, pyridoxal-5�-phosphate; PSAT, phosphoserine aminotransferase; PDB, ProteinData Bank; PEG, polyethylene glycol; RMSD, root-mean-square devia-tion.

Grant sponsor: Academy of Finland; Grant number: 878699; Grantsponsor: Sigrid Juselius Foundation; Grant sponsor: EMBL/DESY;Grant number: HPRI-CT-1999-00017; Grant sponsor: Graduate Fel-lowship Program of NCSR “Demokritos.”

Anatoly P. Dubnovitsky’s present address is Department of Molecu-lar Biology, Swedish University of Agricultural Sciences, UppsalaBiomedical Centre, PO Box 590, SE-751 24 Uppsala, Sweden.

*Correspondence to: Anastassios C. Papageorgiou, Turku Centre forBiotechnology, University of Turku and Åbo Akademi University,Tykistokatu 6, BioCity, 20521 Turku, Finland.E-mail: [email protected]

Received 27 September 2005; Revised 7 December 2005; Accepted 4January 2006

Published online 10 March 2006 in Wiley InterScience(www.interscience.wiley.com). DOI: 10.1002/prot.20935

PROTEINS: Structure, Function, and Bioinformatics 63:742–753 (2006)

© 2006 WILEY-LISS, INC.

members of PLP-dependent enzymes despite the low se-quence identity. Aspartate aminotransferase is the beststructurally characterized representative of type I fold.

The crystal structure of PSAT from Escherichia coli (ECPSAT) and that from the obligate alkaliphile Bacillusalcalophilus (BALC PSAT) have been determined to 2.3 Å(PDB accession code 1BJN6) and 1.08 Å resolution (PDBaccession code 1W237), respectively. BALC PSAT has thesame overall architecture with EC PSAT that consists of ahomodimer characterized by a two-domain fold per sub-unit. A central seven-stranded �-sheet conserved amongall aminotransferases is present in each subunit of PSATs.EC PSAT exhibits a pH-dependent activity profile with abroad maximum between pH 7.5 and 8.5.8 In contrast, theenzymatic activity of BALC PSAT was found to exhibit apH optimum at pH 9.7 Structural comparison of BALCPSAT with EC PSAT showed a number of distinct featuresin BALC PSAT that may explain the enzyme’s adaptationto alkaline pH. Similar features were also identified inphosphoserine aminotransferase from Bacillus circulansssp. alkalophilus, a facultative alkaliphile able to grow atpHs up to 10.9

Investigation of the folding and stability mechanismsamong members of the same structural family coulduncover new determinants in the structure and function ofproteins, especially in cases where significant sequencevariations occur.10 Furthermore, the role of the cofactorPLP in structural stabilization has been reported to varyamong different PLP-dependent enzymes.11,12 Here, wehave studied Bacillus circulans ssp. alkalophilus phospho-serine aminotransferase (BCIR PSAT) to better under-stand its function and stability in response to pH. This isthe first report on the unfolding of a subgroup IV represen-tative. Thermal denaturation measurements were per-formed to establish the role of pH in the stability of theenzyme. In addition, structural analysis of BCIR PSAT tohigh resolution was performed. These studies were facili-tated by the availability of two crystal forms of BCIR PSATgrown at pH 8.5 and 4.6, respectively. Finally, the effect ofpH on the PLP cofactor was investigated.

MATERIALS AND METHODSProtein Purification

The pBCIR PSAT plasmid encoding PSAT from B.circulans ssp. alkalophilus (SwissProt accession codeQ59196) was expressed in E. coli strain XL1-Blue. The

recombinant protein is a mutant that contains a glutamateinstead of a lysine residue at position 3. The protein wasexpressed in bacteria by standard procedures. The cellswere then pelleted by centrifugation and resuspended in25 mM Tris-HCl pH 8, 100 �M PLP, 1 mM DTT, 1 mMPMSF, and 5 mM EDTA followed by sonication andcentrifugation at 20,000g. The supernatant was loadedonto a DEAE CL-6B column (1.5 � 12 cm, Econo-Pac,BioRad) equilibrated with 25 mM Tris-HCl pH 7.5. Afterextensive washing, proteins were eluted with a lineargradient of 0.1–0.4 M NaCl. Fractions containing PSATactivity were pooled, dialyzed against 50 mM Tris-HCl pH7.5, and loaded onto an AX 300 HPLC anion-exchangecolumn (25 � 0.46 cm, Eprogen, Darien, IL) equilibratedwith 50 mM Tris-HCl pH 7.5, 50 mM NaCl, and 0.01%�-mercaptoethanol. After washing with 0.2 M NaCl in 50mM Tris-HCl pH 7.5, protein was eluted with 0.5 M NaClin 50 mM Tris-HCl pH 7.5. Fractions containing �0.5mg/mL PSAT were pooled and concentrated using AmiconCentriprep (YM-30) filters. Sephacryl S-200 size-exclusionchromatography was performed with 50 mM Tris-HCl pH7.5 containing 0.15 M NaCl (column dimensions 100 � 1.5cm, BioRad). After gel filtration, the protein was concen-trated to �20 mg/mL.

Enzyme Activity Measurements

The enzymatic activity of BCIR PSAT in the direction ofphosphoserine biosynthesis at different pH values wasassayed by the UV method.13 Briefly, the enzyme reactionat 25°C was coupled with glutamate dehydrogenase andthe decrease of NADPH absorption at 340 nm was moni-tored in a recording spectrophotometer. The assay mixture(1 mL) contained 8 mM L-glutamic acid, 32 mM ammo-nium acetate, 100 mM appropriate buffer (pH range6.0–8.5, HEPES; pH 8.5–10.0, CHESS), 0.18 mM NADPH,20 �M PLP, 12 U glutamate dehydrogenase (SOR-ACHIM), and 0.5 �g of enzyme. The reaction was startedwith the addition of 0.5 mM phosphohydroxypyruvate andmonitored for 10 min. The pH of the mixture was mea-sured at the end of the reaction and the recorded pH wasused for the calculations. Km values for glutamate andphosphohydroxypyruvate were obtained by varying theconcentration of one substrate in the presence of saturat-ing concentrations of the other. Km and Vmax values werederived from Lineweaver-Burk plots.

Fig. 1. Schematic representation of the chemical reaction catalyzed by phosphoserine aminotransferase.

PROTEINS: Structure, Function, and Bioinformatics DOI 10.1002/prot

EFFECT OF PH ON PHOSPHOSERINE AMINOTRANSFERASE 743

CD Spectroscopy

CD measurements were conducted using a JASCO-715spectropolarimeter with a Peltier type cell holder fortemperature control. Wavelength scans in the far (190–260 nm) and the near (260–340 nm) UV regions wereperformed in Quartz SUPRASIL (HELLMA) precisioncells of 0.1- and 1-cm path length, respectively. Eachspectrum was obtained by averaging five to eight succes-sive accumulations with a wavelength step of 0.2 nm at arate of 20 nm min�1, response time 1 s, and band width 1nm. Buffer spectra were accumulated and subtracted fromthe sample scans. The absorption spectra were recorded byselecting the UV (single) mode of the instrument. CDexperiments involving thermal scanning at heating ratesof 1.5 K/min have been performed in the range from 20° to90°C, at 208 and 222 nm in order to continuously monitorstructural changes in BCIR PSAT secondary structure.Thermal CD scans have also been performed at 340 and415 nm to monitor the dissociation of PLP during thedenaturation process at pH 8.5 and 6.0, respectively.Analysis of the spectra was done using the CDNN pro-gram.14,15

DSC

High-sensitivity calorimetric measurements were per-formed on a VP DSC calorimeter (Microcal, Northampton,MA).16 Protein concentrations used in the DSC studiesvaried between 1.2 and 2.6 mg/mL. Protein samples andbuffer reference solutions were properly degassed undervacuum and constant stirring for a period of 20 min. Theywere then carefully loaded into the calorimeter cells toavoid bubble formation. Four to five reference scans withbuffer-filled cells (sample and reference cell volume was0.523 mL) preceded each sample run in order to accountfor the cell’s thermal history and achieve near perfectbaseline repeatability. A typical DSC experiment con-sisted of a heating scan at a programmed heating rate (u)of 1.5 K/min followed by a second heating scan to probe thereversibility of the thermal transitions. Whenever needed,the difference in the heat capacity between the initial andfinal states was modeled by a sigmoidal chemical base-line.17

Crystallization

BCIR PSAT was crystallized by the hanging-drop vapor-diffusion method in two conditions. Crystals were obtainedat 22°C after optimization of a previously described proto-col18 using 0.1 M sodium acetate buffer, pH 4.6, 4% (w/v)PEG 20000, and 5% (v/v) glycerol. These crystals (form A)belong to the monoclinic space group C2 with unit cellparameters 91.3 � 91.1 � 42.4 Å (� 111.3°) and amonomer in the crystallographic asymmetric unit. A sec-ond crystal form was obtained at 16°C using 0.1 MTris-HCl buffer, pH 8.5, 0.2 M sodium acetate, and 30%(w/v) PEG 4000. Optimization with addition of 8% (v/v)glycerol resulted in well-diffracting crystals (form B) thatbelong to the orthorhombic space group P212121 with unitcell parameters 44.8 � 90.5 � 157.8 Å and one dimer in thecrystallographic asymmetric unit.

X-ray Data Collection, Structure Determination,and Quality of the Models

Complete data sets for both crystal forms were col-lected from single crystals at the EMBL X11 beamline atthe DORIS storage ring, DESY, Hamburg. Crystals wereflash-cooled to 100 K using 15% glycerol as cryopro-tectant for both crystal forms. Diffraction data wererecorded on a 165-mm MAR CCD detector. Completedata sets to 1.5 and 1.2 Å resolution were collected fromcrystal forms A and B, respectively. The data wereintegrated, merged, and scaled using the HKL programsuite.19 The TRUNCATE program20 was used to convertintensities to amplitudes. The final statistics for thedata collection are shown in Table I. In both cases,initial phases were obtained by molecular replacementwith AMoRe.21 For crystal form A, the structure of BCIRPSAT at 2.3 Å (PDB accession code 1BT4) without PLPand water molecules was used as a search model. Forcrystal form B, a transformation matrix based on the ECPSAT dimer structure was applied to locate the secondmolecule in the crystallographic asymmetric unit ofBCIR PSAT. A subset of the total number of reflections(5%) was randomly selected and set aside for cross-validation analysis to monitor the progress of refine-ment using the Rfree factor.22 The structures wererefined initially using CNS v.1.123 and SHELX-9724 atthe later stages. The models were visualized with theprograms O25 and XtalView.26 The crystal form B dataset is complete up to 1.5 Å resolution (overall complete-ness 96%). Inclusion of data to 1.2 Å resolution datadecreased the overall completeness to 86%. Neverthe-less, because the resulting map was of superior quality,the resolution was extended from 1.5 to 1.2 Å in 0.1 Åsteps. Multiple conformations for several side-chainswere taken into account and modeled. Anisotropic refine-ment of temperature factors for all nonhydrogen atomsof crystal form B model remarkably improved the elec-tron density map and resulted in a decrease of Rcryst andRfree values by 1.3% and 3%, respectively. In the case ofcrystal form A, anisotropic refinement of temperaturefactors was applied for N, S, O, and P atoms and resultedin a decrease of Rcryst and Rfree values by 2% and 0.5%,respectively. The working and reference sets weremerged only in the final refinement run. The quality ofthe models was checked using PROCHECK.27 The RMSDfrom ideal geometry and Luzzati plots were calculatedusing SHELX.24 LSQMAN28 was used for structuralsuperposition and calculation of RMSD for particularloops. Hydrogen bonds were assessed by CONTACT20

using ANGLE instruction and 3.3 Å distance limit.Secondary structure elements were analyzed withDSSP.29 Structure-based alignment was performed us-ing the protein structure comparison server SSM30 atthe European Bioinformatics Institute (http://www.ebi-.ac.uc/msd-srv/ssm) and was inspected on graphics. Forthe analysis of the dimer interface, the Protein-Proteininteraction server (PPIS) was used.31


744 E.G. KAPETANIOU ET AL.

Data Deposition

Atomic coordinates and structure factors have beendeposited in the Rutgers Protein Data Bank under theaccession codes 1W3U (form A) and 2C0R (form B).

RESULTS AND DISCUSSIONpH Optimum

BCIR PSAT was first studied to establish whether itsenzymatic function is pH dependent. The enzymatic activ-ity of BCIR PSAT in the direction of phosphoserine synthe-sis was measured at various pHs using saturated concen-trations of substrates. BCIR PSAT shows a bell-shapedactivity profile with a maximum at pH 9.0 (Fig. 2). At pH9.5, the enzyme retains 60% of its maximum activity.However, the enzyme activity is lost rapidly upon pHchange with no activity observed below pH 6.0. Theapparent Km values for glutamate and phosphohydroxy-pyruvate at pH 9.0 were 0.20 and 0.09 mM, respectively(Table II). pH lowering clearly affects the binding ofL-glutamate to BCIR PSAT with Km increased to 1.0 mMat pH 7.0. The effect on 3-phosphohydroxypyruvate bind-ing is less dramatic with only a small (�10%) increase ofKm (Km 0.1 mM). The intracellular pH of alkaliphilicBacillus species growing at pH around 11 has been shownto deviate by two full pH units from the environmentalpH.32,33 The pH optimum obtained for BCIR PSAT ap-

pears therefore to be in good agreement with the cytoplas-mic pH expected for BCIR during growth at pH �10.

Secondary Structure Changes

To explore whether the biological activity profile of BCIRPSAT is directly related to thermodynamic stability at

TABLE I. Data Collection and Refinement Statistics

Form A (pH 4.6) Form B (pH 8.5)

Space group C2 P212121Cell dimensions (Å) a 91.3 a 44.8

b 91.1 b 90.5c 42.4 c 157.8� 111.3°

Resolution range (Å) 15–1.50 (1.55–1.50) 15–1.20 (1.24–1.20)Wilson B-factor (Å2) 13.99 13.24No. of observations 393,963 4,268,848No. of unique reflections 49,679 171,731Completeness (%) 96.2 (90.8) 86.0 (74.5)Rsym (%) 3.5 (19.9) 8.1 (48.3)I/(I) 20 (4) 19 (3)No. of molecules/a.u. 1 2Refinement statistics

Resolution range (Å) 15.0–1.50 12.0–1.20No. of reflections in working set 47,144 162,944No. of reflections in test set (5%) 2,481 8,578Protein atoms 2,771 5,608Water molecules 340 809Cofactor molecules 1 2Glycerol molecules 1 —Rcryst/Rfree (%) 14.4/20.8 15.2/21.1

RMSDBond lengths (Å) 0.009 0.011Bond angles (°) 0.028 0.030

Average B-factors (Å2)Main-chain 15.3 15.8Side-chain 20.2 23.2Waters 26.9 30.8PLP 11.8 13.1

Fig. 2. pH-dependence of BCIR PSAT enzymatic activity. The assaywas performed at different pH values using 0.1 M HEPES (squares),CHES (circles), or CABS (triangles).



various pHs, thermal unfolding experiments were per-formed. First, CD was used to study the heat-induceddenaturation of BCIR PSAT at pH values 6.0, 8.5, and10.0. At lower pHs, the enzyme formed aggregates thatprevented the recording of the CD spectra. The pI of BCIRPSAT is �4.55,7 thus explaining the formation of aggre-gates at low pHs. In the native state (room temperature),the far-UV CD spectra for all three pHs are identical [Fig.3(a)]. Analysis of the spectra reveals 34.6 � 1.0% �-helicalcontent and 14.2 � 1.4% �-sheets. However, in the dena-tured stage, the far-UV CD spectra at various pHs showdistinct differences especially between alkaline and acidicpHs [Fig. 3(b)] suggesting that the denatured states arestructurally different from each other. Indeed, the spec-trum at pH 6.0 is indicative of extended structural changesoccurring during thermal denaturation. Analysis of thedenatured spectrum at 80°C temperature shows an ap-proximately 50% loss in the �-helical content with respectto the native. However, the spectra at pH 8.5 and 10.0demonstrate that the denatured BCIR PSAT moleculeretains an adequate amount of the native’s secondarystructure. However, the two spectra are not similar,suggesting that the two denatured states have structuraldifferences as well. To exclude the possibility of kineticcontamination of the denatured spectra, CD data werecollected for the pH 6.0 sample at T 75°C and T 80°Cafter an incubation period of approximately 1 h. The twospectra were identical indicating minimal kinetic effects.Analogously, CD data for the high pH samples have beenobtained at temperatures 70°, 80°, and 90 °C after 1-hincubation periods at each temperature. Kinetic contami-nation has been minimal for these samples as well.

For all pHs, the PLP molecule dissociates from thedenatured BCIR PSAT molecule. This is clearly demon-strated by CD difference spectra (Fig. 4). Indeed, becauseunbound PLP is optically inactive, the resultant difference(nature-denatured) spectra show the characteristic posi-tive bands at either around 345 nm or around 415 nmdepending on the protonation state of PLP.7,34,35

DSC

Further characterization of the thermal denaturation ofBCIR PSAT was performed by using high-precision DSC.As in the case of the CD experiments, the calorimetric datademonstrate substantial differences in PSAT’s behaviorbetween alkaline and acidic environments. The DSC Cp

(heat capacity under constant pressure) versus T profilesobtained at various pH conditions are presented in Figure5. At pH 6.0 and protein concentration (Ct) of 1.2 mg/mL,

the thermal unfolding is described by a sharp Cp peak witha Tm (temperature at maximum Cp) of 71°C for heatingscan rate of 1.5 K/min. The thermal transition is irrevers-ible, possibly because of aggregation phenomena in thedenatured state (visible aggregates can be observed in thedenatured state). A complete thermodynamic analysis ofthe unfolding is thus not possible because the DSC profilesmay well be kinetically contaminated. Nevertheless, thecalorimetric enthalpy change (�H) associated with theunfolding can be calculated from the Cp versus T trace. AtpH 6.0, �H is 288 � 27 kcal/mol, which corresponds to0.80 � 0.07 kcal/mol-residue. This result is in good agree-ment with the findings concerning globular proteins ofapproximately the same size with BCIR PSAT undergoing

TABLE II. Kinetic Parameters for BCIR PSAT

Glutamate Phosphohydroxypyruvate

Km(mM)

Vmax(�NADPH�mol/min)

Km(mM)

Vmax(�NADPH�mol/min)

pH 7.0 1 0.006 0.10 0.006pH 9.0 0.2 0.01 0.09 0.01

Fig. 3. A: Far-UV CD spectra in the native BCIR PSAT at 25°C. B: Inthe thermally denatured state of BCIR PSAT at various pHs: pH 6, 80°C(solid line); pH 8.5, 90°C (dashed line), and pH 10, 90°C (dotted line). Thespectra have been obtained after 1-h incubation period at each tempera-ture.



complete thermal denaturation (�H 0.77 � 0.09 kcal/mol-residue).36

The DSC profiles are considerably different in alkalinepHs. At pH 8.5, Tm is 58.4°C (for scan rate 1.5 K/min, Ct 1.2mg/mL), almost 13° lower than the corresponding Tm atpH 6.0. The DSC trace at pH 10.0 is similar to that at pH8.5 with a Tm of 58°C (scan rate 1.5 K/min, Ct 1.1mg/mL). Moreover, the Cp peaks at high pH are broaderthan the ones at pH 6.0. This, in turn, is indicative of a lesscooperative transition at high pH. At a first glance, it

would appear that in the acidic environment the BCIRPSAT molecule acquires more thermostability. The situa-tion is more complex though. The enthalpic content of theCp peaks at pHs 8.5 and 10.0 is 121 � 11 and 124 � 11kcal/mol, respectively, or approximately 0.34 kcal/mol-residue. This accounts for only half of the enthalpy changemeasured at pH 6.0 and is clearly smaller than thatexpected for a complete thermal denaturation to an un-structured final state. Moreover, aggregation phenomenain the denatured states at the alkaline pHs are consider-ably milder than those observed at pH 6.0. Thus, BCIRPSAT may follow a different unfolding pathway dependingon the pH conditions. One possible explanation mightconcern the role of PLP. Indeed, from the CD experimentsit has become evident that PLP dissociates from thedenatured BCIR PSAT molecule. The PLP dissociationcould therefore contribute to the observed Ct-dependenceof Tm in the calorimetric results (Fig. 5). However, if onlythe dissociation of the PLPs was to take place, the experi-

Fig. 6. A: The temperature dependence of the fractional population ofdenatured BCIR PSAT molecules as depicted by the CD signal at 222 nm(dotted line) and at 340 nm (dashed line) at pH 6.0, scan rate 1.5 K/min,and Ct 0.4 mg/mL. B: PSAT molecules as depicted by the CD signal at208 nm (dotted line) and at 415 nm (dashed line) at pH 8.5, scan rate 1.5K/min, and Ct 0.4 mg/mL.

Fig. 4. CD difference spectra (native—denatured) from 280–500 nmfor three different pH values: pH 6 (solid line), denatured state spectrumobtained at T 80°C; pH 8.5 (dashed line), denatured state spectrumobtained at T 90°C; pH 10 (dotted line), denatured state spectrumobtained at T 90°C.

Fig. 5. Irreversible DSC profiles obtained at scan rate 1.5 K/min atvarious pH and Ct values: pH 6 (solid line), pH 8.5 (dashed line), and pH10 (dotted line). The DSC peaks at pH 10 are shown at three differentconcentrations which depict the characteristic dependence of Tm on Ct:For lowest Tm, Ct 1.2 mg/mL, for middle Tm, Ct 2.0 mg/mL, and forhighest Tm, Ct 2.6 mg/mL.



Fig. 7. Ribbon representation of the BCIR PSAT structure at pH 8.5(A) dimer (B) monomer. Secondary-structure elements were calculatedusing DSSP.29 The �-helices are shown in yellow and the �-strands incyan. N- and C-termini are shown as spheres and PLP molecules inspace filling (A) and ball-and-stick (B). The small domain contains anextended five-stranded �-sheet, composed of a two-stranded parallel�-sheet (�1 and �13) and a three-stranded antiparallel �-sheet ( �11,��12, ��14). The large domain contains the conservative amongPLP-dependent enzymes44 seven-stranded �-sheet (�2–�5 and �7–�9)with all �-strands parallel except �9. Two additional �-strands (�6 and�10), six �-helices (�1–�6), and six 310-helices complete the largedomain. Helix �7 connects the large domain with the small domain. Thefive-stranded �-sheet is flanked by helix �7 and two additional helices (�8and �9) close to the C-terminus. The Figure was produced usingMOLSCRIPT45 and Raster3D.46



mental values of the Tm shifts with increasing Ct wouldhave been larger than the observed ones.37 This is indica-tive that both cofactor and dimer dissociation take placeduring the denaturation process. Thus, the heat-induceddenaturation could be best described by the followingmodel37:

N2(PLP)2 3 2D 2PLP (1)

where N is the native and D the denatured state of BCIRPSAT.

It is interesting at this point to investigate the kineticsof the thermal unfolding of BCIR PSAT as unveiled bythermal CD scans at heating rates of 1.5 K/min. Figure 6displays the fractional population of BCIR PSAT mol-ecules in the denatured state as a function of temperature

derived by i) the CD signal at 222 nm versus T, depictingthe changes in the helical structures of the native state,and ii) the CD signal at 415 nm versus T, depicting thedissociation of PLP molecules from BCIR PSAT. At pH 6.0[Fig. 6(a)], the two processes are remarkably synchronous.In contrast, at pH 8.5 [Fig. 6(b)], changes in the secondarystructure clearly precede (by almost 3°C) the dissociationof PLP. At pH 10.0 (data not shown), only a small (lessthan 1°C) temperature interval separates the signals atthe two wavelengths.

Based on the calorimetric and spectroscopic results, itseems that at pH 8.5 the dissociation of PLP from BCIRPSAT takes place once the changes in the secondarystructure have considerably progressed. The moleculedestabilizes into a substantially structured monomericstate, which remains stable for up to higher temperatures(90°C). Such behavior is likely pointing toward a multistepdenaturation process, which includes denatured intermedi-ate states containing PLP. In fact, this has also beenobserved during chemical denaturation experiments of thedimeric mitochondrial aspartate aminotransferase(mAAT).34 A common unfolding mechanism between mem-bers of the same structural family but with low sequencehomology is not unusual and has been suggested before forthe �-family of aminotransferases.10 However, a different

Fig. 9. A: Superposition of the BCIR PSAT pH 4.6 (green) and pH 8.5 (red) substrate binding loop. B: Superposition of the EC-AMG PSAT (green),EC PSAT (red), and BALC PSAT (black) substrate binding loop. AMG is shown.

Fig. 8. A: Superposition of crystal form A and B BCIR PSATmonomers. Three hundred eighteen C� atoms were found to be conforma-tionally invariant at 3.5 level (lower limit for error-scaled differencedistance matrix analysis). These residues are indicated as blue backbonetrace whereas the remaining (conformationally variant) are in red. Analy-sis was made with ESCET.38 The Figure was produced using MOL-SCRIPT45 and Raster3D.46 B, C: Distribution of B-factors for the main-chain and side-chain atoms. The black line corresponds to the high pHstructure and the red line to the low pH structure.


altogether scenario occurs at pH 6. Here, thermally in-duced changes in the secondary structure are not observeduntil higher temperatures. Interestingly, these changesoccur simultaneously with the release of PLP and thecooperative destabilization of BCIR PSAT into unstruc-tured monomers. The latter are characterized by highpropensity for aggregation. Finally, at pH 10.0, whereBCIR PSAT is biologically inactive, a denaturation processto a stable unfolding intermediate state is likely, almostsimultaneously with the release of PLP and the dissocia-tion of the homodimer. The proposed scenario at pH 10appears therefore to combine elements of both the pH 6.0and pH 8.5 unfolding processes. It is, however, strikingthat release of PLP is slightly postponed at pH 8.5 which isclose to the pH of optimum activity. At pH 6.0 and 10.0,BCIR PSAT enzymatic activity is diminished suggestingthat changes affecting active site’s functionality may wellcontribute to the PLP release and the kinetics of theunfolding.

Crystallographic Analysis and Quality of theStructures

In an attempt to explain the differences observed inBCIR PSAT enzymatic activity and thermal denaturationat different pHs, a detailed analysis of the crystal struc-ture of the enzyme was performed. Two different crystalforms (A, pH 4.6; B, pH 8.5) were grown. Based on theenzymatic activity measurements, BCIR PSAT at pH 4.6has no activity. In contrast, at pH 8.5, BCIR PSAT retainsapproximately 80% of its maximum activity. Both struc-tures were refined to high resolution (1.5 Å, pH 4.6; 1.2 Å,pH 8.5) with good stereochemistry. The final model at pH4.6 comprises 360 amino acid residues, 1 PLP molecule, 1glycerol molecule, and 340 water molecules. The finalmodel from form B crystals contains a homodimer in thecrystallographic asymmetric unit with a total of 722residues, 2 PLP molecules, and 809 water molecules. TableI shows the final refinement statistics for the atomicmodels of BCIR PSAT. Description and analysis of theBCIR PSAT structure is based on the pH 8.5 model (Fig. 7).Subunit A will be used throughout the remainder of thediscussion except where noted otherwise. BCIR PSATmodel at pH 4.6 will be used for comparison studies.

Structural Differences Upon pH Change

Form A and form B PSAT monomers were compared toidentify structural differences as a result of the differentpH used in crystallization. Least-squares superpositionyielded an RMSD of 0.5 Å for 357 C� positions. Thesecondary structure is almost identical with limitedchanges between the two forms as suggested also by theCD analysis of the native structure. The program ES-CET38 was used to identify conformational flexible regionsof the protein by comparing the two structures. Theanalysis was based on 352 residues. Of them, 9.4% werecategorized as flexible (2–4, 90–92, 133–143, 197–201,306–309, 310–312, 327–334, 359–362) (Fig. 8). Significantchanges in the flexible loop 133–143 (RMSD 2.7 Å), whichis situated close to the domain’s interface, and in the loop

327–334 (RMSD 2.3 Å) which harbors two substrate-binding residues (His328 and Arg329) were found. His328and Arg329 of the small domain of the same subunit andHis42 and Arg43 of the large domain of the second subunitare fully conserved residues in all PSAT structures. Theseresidues are oriented in two pairs and opposite to eachother in the active site. In addition, they have beensuggested to interact with the substrate in the EC PSATstructure in complex with �-methyl-L-glutamate.6 At pH8.5, loop 328–333 moves toward the entrance of thesubstrate-binding pocket (Fig. 9). Furthermore, His328 istilted �30° compared with its plane at pH 4.6. Notably, theB-factors for the 328–333 loop are higher than those at thepH 4.6 structure suggesting some flexibility that couldpossibly facilitate substrate binding at alkaline pH. Thedistribution of the B-factors for the main- and side-chainatoms is shown in Figure 8(B, C).

Analysis of the dimer interface for the pH 8.5 structurewith the Protein-Protein Interaction server (PPIS31,39)showed that approximately 15% of the solvent-accessiblesurface area of each subunit (2315 Å2) is buried upondimer formation. An approximately similar surface area(2207 Å2) is buried in form A BCIR PSAT dimer. Interac-tions between the two monomers in form B calculated byCONTACT20 are quite extensive and include 159 hydropho-bic contacts (cut-off distance 5.0 Å) and 44 polar contacts(cut-off distance 3.8 Å), of which 26 correspond to hydrogenbonds and two hydrogen-bound ion pairs between Arg4and Glu39 in each subunit. The number of intersubunitinteractions is the same as in the BCIR PSAT model at pH4.6.7 Thus, the dimer interface seems not to be affected bythe pH changes.

The total number of intramolecular interactions permonomer in the two structures is similar. However, smallvariations in the distances were found that may contributeto the stability profile at different pHs. A significantdifference between the two structures was found regardingthe number of ion pairs. The structure at pH 8.5 exhibits16 ion pairs compared with 11 found in the pH 4.6structure (Table III). As the number of ion pairs seems tobe influenced by the pH, the additional ion pairs at pH 8.5may contribute to the different behavior of BCIR PSAT atalkaline pH. Moreover, the fact that at pH 8.5 the proteinmolecule still retains a structured state before the releaseof PLP also points to a potential role of the intramolecularinteractions. Detailed analysis of the ion pairs revealsinteresting features. Three of the 11 ion pairs found in thepH 4.6 model are not present in the pH 8.5 model. Incontrast, a higher proportion of ion pairs (8 of 16) are foundin the pH 8.5 but not in the pH 4.6 structure. TheGlu18-Arg258 ion pair is located near the hinge regionconnecting the small and large domains of the monomer.As such, the orientation of the two domains may bedisrupted more easily in the pH 4.6 structure, thus contrib-uting to the collapse of the molecule. The region 134–142that undergoes a large movement during transition fromlow to high pH could also have a role to maintain thedomain orientation at high pH. This loop is located close tothe domain’s interface and is better defined at the pH 8.5



structure. The presence of factors that stabilize the do-mains or domain–domain interactions may affect thekinetics of the BCIR PSAT unfolding. For large proteins,the existence of several folding domains/subunits oftenresults in multiple unfolding paths.40 From the remainderof the ion pairs, it is worthy to mention the stronginteraction (2.49 Å) in the large domain between Glu53and Lys253 at pH 8.5 from neighboring helices �2 and �6,respectively. This interaction may again have a role in thestability of the large domain and subsequently the unfold-ing pathway.

Active Site

Close inspection of the active site was performed toanalyze changes in the chemical environment of PLPbetween the two structures. The two active sites in theBCIR PSAT dimer are located at the dimer interface [Fig.7(a)]. PLP is connected to Lys197 through a Schiff base.Table IV shows the hydrogen-bond interactions betweenPLP and active site atoms in crystal forms A and B of BCIRPSAT. Further contacts include aromatic stacking interac-tions between the pyridine ring of the cofactor and theindole ring of Trp103 (�3.6 Å separation distance) on there-face of PLP, and van der Waals interactions with theside-chains of Ala77 and Ser175 on the si-face. Moreover,the pyridine nitrogen atom (N1) makes a strong contact(2.7 Å) with OD2 atom of Asp173, a residue strictlyconserved in the aminotransferase superfamily. Rotationof Asp173 side-chain is prevented by hydrogen bonds withthe main-chain N and OG of Ser173 and a water-mediatedcontact with Thr149. No changes were observed in theorientation of the PLP aromatic ring upon pH change,hence the interactions of the cofactor are remarkablyconserved regardless of the pH.

The O3� atom of the cofactor forms a hydrogen bond withOG1 of Thr153. No similar interaction has been observedin the unligated EC PSAT. It was therefore suggested thatThr153 OG1 may contribute to the increased pKa of theimine nitrogen of the Schiff base in BALC and BCIRPSATs.7 The pKa for BCIR PSAT imine nitrogen is �9.0,hence at lower pHs is expected to be protonated. Theprotonation state of the imine nitrogen may also have arole in the release of PLP during thermal unfolding. Infact, the differences observed at pH 6.0 and 8.5 could berelated to the activity profile of the enzyme. However, atpH 10.0, the PLP release occurs in a manner similar tothat at pH 6.0 but in a nonsynchronous mode reminiscentof that at pH 8.5. It is therefore possible that other factorscould influence the release of PLP but at this stage it isdifficult to speculate. Nevertheless, the presence of inter-mediate stages during the thermal release of PLP at a pHclose to the pH optimum for enzymatic activity imposes aninteresting question for further experiments. In otherwords, the recorded differences seem to suggest a possibledependence on the cofactor of the unfolding kinetics andthe formation of unfolded species. Similar suggestionshave been put forward for AAT.41

As mentioned above, the loop 328–333 moves closer tothe entrance of the active site at pH 8.5 (Fig. 9). In its newposition, it seems to reduce the accessibility of the solventto the active site. Moreover, Arg329 from being highlyflexible at pH 4.6 becomes ordered at pH 8.5 and able tointeract with water molecules at the vicinity of the activesite. It is thus possible that BCIR PSAT Arg329 may havea stabilizing role for the active site through a waternetwork. In support of the stabilizing effect of the 328–333loop, stability studies in Cu, Zn superoxide dismutase haveshown that higher flexibility and solvent accessibility inthe active site result in pH sensitivity in the E. colienzyme.42 Finally, this loop may also serve for the betterbinding of substrate in alkaline environments close to thepH optimum of BCIR PSAT as deduced by the Km valuesobtained in the kinetic assays.

TABLE III. Potential Ion-Pair Intramolecular Interactionsin the BCIR PSAT Monomer at Basic and Acidic pH†

Residue 1 Residue 2 pH 8.5 pH 4.6

Glu18 OE1 Arg22 NH2 2.97Arg258 NH2 2.49

Arg22 NH2 Glu257 OE1 3.36 2.71Arg43 NE Glu48 OE2 3.40Glu53 OE2 Arg57 NH2 2.48 2.49

Lys253 NZ 3.67Arg57 NH1 Glu256 OE1 3.61Lys160 NZ Asp293 O 3.60Asp177 OD2 Arg181 NH1 2.73 2.67Arg210 NE Asp212 OD1 2.92Arg226 NE Asp228 OD1 2.91 2.90Glu250 OE1 Lys253 NZ 2.82 2.87Glu257 OE1 Arg258 NH1 3.42 3.19Arg269 NE Asp297 OD1 2.79 2.78Asp280 OD1 Arg295 NH2 2.78 2.70Lys311 NZ Glu312 OE1 3.39Glu312 OE1 Arg361 NE 2.84Glu355 OE1 Arg359 NH1 3.10Glu355 OE2 Lys358 NZ 2.74†Residue pairs in which negatively and positively charged atoms existwithin 4.0 Å.

TABLE IV. Hydrogen-Bond Interactions for PLP†

PLP

pH 4.6 pH 8.5

Distance(Å)

Distance(Å)

O3� A-Thr153 OG1 2.52 A-Thr153 OG1 2.54O3� A-Lys197 N 2.49 A-Lys197 N 2.49N1 A-Asp173 OD2 2.68 A-Asp173 OD2 2.65O1P B-Thr239 OG1 2.67 B-Thr239 OG1 2.65O1P B-Thr239 N 2.83 B-Thr239 N 2.89O2P A-Ser78 N 2.79 A-Ser78 N 2.84O2P A-Ser78 OG 2.58 A-Ser78 OG 2.60O2P B-Asn238 ND2 2.86 B-Asn238 ND2 2.89O3P A-Gln196 NE2 2.80 A-Gln196 NE2 2.83O3P A-Ala77 N 3.20 A-Ala77 N 3.21O3P B-Wat6074 2.63 B-Wat25 2.73O4P A-Wat6208 2.82 A-Wat349 2.82O1P A-Wat6086 2.74 A-Wat177 2.70†Cut-off 3.2 Å. A and B refer to the two subunits of the BCIR PSAThomodimer.



The region 199–202 was found to be conformationallyflexible. Indeed, the conformation of this region is signifi-cantly different between the two subunits and adopts analternate geometry in subunit A. Interestingly, this regionis located near the active site, diametrically opposite to328–333 with PLP being in the center. Residues 200 and201 are Gly and Pro, respectively. Isomerization of Proresidues has been associated with slow folding phases.43

Because of their proximity to the active site, residues199–202 may also affect the kinetics of PLP release duringthermal unfolding.

CONCLUSIONS

Crystallographic and thermal unfolding studies wereperformed to characterize the stability and denaturationprofile of PSAT from the facultative alkaliphile B. circu-lans ssp. alkalophilus in response to pH changes. Kineticanalysis clearly established a maximum enzymatic activ-ity at pH 9.0 and no activity below pH 6.0. Thermalunfolding analysis revealed a major difference betweenpHs 6.0, 8.5, and 10.0 concerning the release of the PLPcofactor from the thermally unfolded protein molecule.This release was found to occur either synchronously orasynchronously with the thermally induced changes in thesecondary structure. High-resolution crystal structure anal-ysis showed subtle changes between BCIR PSAT struc-tures at pH 4.6 and 8.5, respectively. A conformationalchange of a loop at the entrance of the active site wasidentified that may be responsible for the enhanced sub-strate binding at pH 8.5. Additional regions at the surfaceof the protein were also found to adopt different conforma-tions at the two pHs and may well affect the unfoldingprofile as revealed by thermal denaturation experiments.Two prominent ion-pair interactions not present in the lowpH structure were identified in the pH 8.5 structure thatmay contribute to domain stabilization within the PSATmonomer. No major changes were found for PLP as aresult of the pH changes. Thus, the release of PLP mightbe affected by conformational rearrangements in the vicin-ity of the active site or distantly from it. Further experi-ments could shed light on the mechanism and the exactrole of pH in the unfolding of PSAT whereas mutagenesisstudies could establish a more precise role of variousresidues in the stability of the enzyme. In addition, the pHanalysis presented here may form the basis to experimentsfor a better understanding of the role of PLP in thestabilization of PSAT or other PLP-dependent enzyme.Moreover, comparative studies between alkaliphilic andneutrophilic PSATs may also provide further insights intopH stability.

ACKNOWLEDGMENTS

The authors thank N. Battchikova, M. Koivulehto, andT. Korpela, University of Turku, for generously providingthe BCIR PSAT plasmid, and the staff at EMBL/DESY,Hamburg for help and advice during data collection.E.G.K. is a graduate student of the Finnish NationalGraduate School in Informational and Structural Biology.A.T. acknowledges support from the Graduate FellowshipProgram of NCSR “Demokritos.”

REFERENCES

1. Jaenicke R, Bohm G. The stability of proteins in extreme environ-ments. Curr Opin Struct Biol 1998;8:738–748.

2. Ichihara A, Greenberg DM. Further studies on the pathway ofserine formation from carbohydrate. J Biol Chem 1957;224:331–341.

3. Walsh DA, Sallach HJ. Comparative studies on the pathway forserine biosynthesis in animal tissues. J Biol Chem 1966;241:4068–4076.

4. Mehta PK, Christen P. Homology of pyridoxal-5�-phosphate-dependent aminotransferases with the cobC (cobalamin synthe-sis), nifS (nitrogen fixation), pabC (p-aminobenzoate synthesis)and malY (abolishing endogenous induction of the maltose sys-tem) gene products. Eur J Biochem 1993;211:373–376.

5. Grishin NV, Phillips MA, Goldsmith EJ. Modeling of the spatialstructure of eukaryotic ornithine decarboxylases. Protein Sci1995;4:1291–1304.

6. Hester G, Stark W, Moser M, Kallen J, Markovic-Housley Z,Jansonius JN. Crystal structure of phosphoserine aminotransfer-ase from Escherichia coli at 2.3 Å resolution: comparison of theunligated enzyme and a complex with alpha-methyl-L-glutamate.J Mol Biol 1999;286:829–850.

7. Dubnovitsky AP, Kapetaniou EG, Papageorgiou AC. Enzymeadaptation to alkaline pH: atomic resolution (1.08 Å) structure ofphosphoserine aminotransferase from Bacillus alcalophilus. Pro-tein Sci 2005;14:97–110.

8. Kallen J, Kania M, Markovic-Housley Z, Vincent MG, JansoniusJN. Crystallographic and solution studies on phosphoserine ami-notransferase from Eschericia coli. In: Christen P, Korpela T,editors. Biochemistry of vitamin B6. Basel: Birkhauser Verlag;1987. p 157–160.

9. Horikoshi K, Akiba T. Alkalophilic microorganisms. A new micro-bial world. New York: Springer-Verlag; 1982.

10. Bhatt AN, Prakash K, Subramanya HS, Bhakuni V. Differentunfolding pathways for mesophilic and thermophilic homologuesof serine hydroxymethyltransferase. Biochemistry 2002;41:12115–12123.

11. Cai K, Schirch D, Schirch V. The affinity of pyridoxal 5�-phosphatefor folding intermediates of Escherichia coli serine hydroxymethyl-transferase. J Biol Chem 1995;270:19294–19299.

12. Bettati S, Benci S, Campanini B, et al. Role of pyridoxal 5�-phosphate in the structural stabilization of O-acetylserine sulfhy-drylase. J Biol Chem 2000;275:40244–40251.

13. Hirsch H, Greenberg DM. Studies on phosphoserine aminotrans-ferase of sheep brain. J Biol Chem 1967;242:2283–2287.

14. Bohm G, Muhr R, Jaenicke R. Quantitative analysis of protein farUV circular dichroism spectra by neural networks. Protein Eng1992;5:191–195.

15. Dalmas B, Hunter GJ, Bannister WH. Prediction of proteinsecondary structure from circular dichroism spectra using artifi-cial neural network techniques Biochem Mol Biol Int 1994;34:17–26.

16. Plotnikov VV, Brandts JM, Lin LN, Brandts JF. A new ultrasensi-tive scanning calorimeter. Anal Biochem 1997;250:237–244.

17. Takahashi K, Sturtevant JM. Thermal denaturation of Streptomy-ces subtilisin inhibitor, subtilisin BPN�, and the inhibitor-subtilisin complex. Biochemistry 1981;20:6185–6190.

18. Moser M, Muller R, Battchikova N, Koivulehto M, Korpela T,Jansonius JN. Crystallization and preliminary X-ray analysis ofphosphoserine aminotransferase from Bacillus circulans ssp. alka-lophilus. Protein Sci 1996;5:1426–1428.

19. Otwinowski Z, Minor W. Processing of X-ray diffraction datacollected in oscillation mode. Methods Enzymol 1997;276:2283–2287.

20. Collaborative Computational Project Number 4. The CCP4 suite:programs for protein crystallography. Acta Crystallogr D BiolCrystallogr 1994;50:760–763.

21. Navaza J. AMoRe: an automated package for molecular replace-ment. Acta Crystallogr A 1994;50:157–163.

22. Brunger AT. Free R value: cross-validation in crystallography.Methods Enzymol 1997;277:366–396.

23. Brunger AT, Adams PD, Clore GM, et al. Crystallography andNMR system: a new software suite for macromolecular structuredetermination. Acta Crystallogr D Biol Crystallogr 1998;54:905–921.



24. Sheldrick G, Schneider T. SHELXL: high-resolution refinement.Methods Enzymol 1997;277:319–343.

25. Jones TA, Zou JY, Cowan SW, Kjeldgaard M. Improved methodsfor building protein models in electron density maps and thelocation of errors in these models. Acta Crystallogr A 1991;47:110–119.

26. McRee DE. XtalView/Xfit: a versatile program for manipulatingatomic coordinates and electron density. J Struct Biol 1991;125:156–165.

27. Laskowski RA, MacArthur MW, Moss DS, Thornton JM. PRO-CHECK: a program to check the stereochemical quality of proteinstructures. J Appl Crystallogr 1993;26:283–291.

28. Kleywegt GJ, Jones TA. A super position. CCP4/ESF-EACBMNewslett Protein Crystallogr 1994;31:9–14.

29. Kabsch W, Sander C. Dictionary of protein secondary structure:pattern recognition of hydrogen-bonded and geometrical features.Biopolymers 1983;22:2577–2637.

30. Krissinel E, Hendrick K. Secondary-structure matching (SSM), anew tool for fast protein structure alignment in three dimensions.Acta Crystallogr D Biol Crystallogr 2004;60:2256–2268.

31. Jones S, Thornton JM. Protein-protein interactions: a review ofprotein dimer structures. Prog Biophys Mol Biol 1995;63:31–65.

32. Krulwich TA, Ito M, Gilmour R, Guffanti AA. Mechanisms ofcytoplasmic pH regulation in alkaliphilic strains of Bacillus.Extremophiles 1997;1:163–169.

33. Horikoshi K. Alkaliphiles: some applications of their products forbiotechnology. Microbiol Mol Biol Rev 1999;63:735–750.

34. Wu T-H, Oses-Prieto JA, Iriarte A, Martinez-Carrion M. Releaseof pyridoxal 5�-phosphate upon unfolding of mitochondrial aspar-tate aminotransferase. Biochim Biophys Acta 2003;1647:315–320.

35. Kelly SM, Price NC. The use of circular dichroism in the investiga-tion of protein structure and function. Curr Protein Pept Sci2000;1:349–384.

36. Benitez-Cardoza CG, Rojo-Dominguez A, Hernandez-Arana A.Temperature-induced denaturation and renaturation of triosephos-phate isomerase from Saccharomyces cerevisiae: evidence of dimer-ization coupled to refolding of the thermally unfolded protein.Biochemistry 2001;40:9049–9058.

37. Sanchez-Ruiz JM. Theoretical-analysis of Lumry-Eyring modelsin differential scanning calorimetry. Biophys J 1992;61:921–935.

38. Schneider TR. Objective comparison of protein structures: error-scaled difference distance matrices. Acta Crystallogr D BiolCrystallogr 2000;56:714–721.

39. Jones S, Thornton JM. Principles of protein-protein interactions.Proc Natl Acad Sci USA 1996;93:13–20.

40. Dinner AR, Sali A, Smith LJ, Dobson CM, Karplus M. Understand-ing protein folding via free-energy surfaces from theory andexperiment. Trends Biochem Sci 2000;25:331–339.

41. Oses-Prieto JA, Bengoechea-Alonso MT, Artigues A, Iriarte A,Martinez-Carrion M. The nature of the rate-limiting steps in therefolding of the cofactor-dependent protein aspartate aminotrans-ferase. J Biol Chem 2003;278:49988–49999.

42. Battistoni A, Folcarelli S, Cervoni L, et al. Role of the dimericstructure in Cu, Zn superoxide dismutase. J Biol Chem 1998;273:5655–5661.

43. Wedemeyer WJ, Welker E, Scheraga H. Proline cis-trans isomer-ization and protein folding. Biochemistry 2002;41:14637–14644.

44. John RA. Pyridoxal phosphate-dependent enzymes. Biochim Bio-phys Acta 1995;1248:81–96.

45. Kraulis PJ. MOLSCRIPT: a program to produce both detailed andschematic plots of protein structures. J Appl Crystallogr 1991;8:730–737.

46. Merritt EA, Murphy MEP. Raster3D Version 2.0. A program forphotorealistic molecular graphics. Acta Crystallog D Biol Crystal-logr 1994;50:760–763.



Effect of pH on the structure and stability ofBacillus circulans ssp.alkalophilus phosphoserine aminotransferase: Thermodynamic and crystallographic studies

Documents