ORIGINAL PAPER

Eileen K. Jaffe

Investigations on the metal switch region of humanporphobilinogen synthase

Received: 14 June 2002 /Accepted: 14 August 2000 / Published online: 28 September 2002� SBIC 2002

Abstract Porphobilinogen synthase (PBGS) is anancient and highly conserved protein that functions inthe first common step in tetrapyrrole biosynthesis. ThePBGS protein sequence contains a unique metal switchregion that has been postulated to dictate an exclusivecatalytic use of either zinc or magnesium, and perhapsalso potassium. In some PBGS, the cysteines of themetal switch sequence DXCXCX(Y/F)X3G(H/Q)CGhave been demonstrated to bind a catalytic zinc, andin other PBGS, the aspartic acid residues of the metalswitch sequence DXALDX(Y/F)X3G(H/Q)DG havebeen postulated to bind a catalytically essential magne-sium and/or potassium. The current work describeschimeric proteins that contain the aspartate-richsequences of pea PBGS and Pseudomonas aeruginosaPBGS in place of the naturally occurring cysteine-richsequence of human PBGS. The resultant chimeric PBGSproteins, peainhuman PBGS and psuinhuman PBGS, aresubstantially activated by both magnesium and potas-sium, but not by zinc. The specific activities of thechimeras are significantly lower than human PBGS.Detailed kinetic and inhibition data are presented forboth chimeric proteins and are discussed in terms of thisunique phylogenetic variation in metal ion usage. Theidentity of a basic residue, which is Arg221 in humanPBGS, strictly correlates with the presence or absenceof the cysteine-rich sequence. Those PBGS with theaspartate-rich metal switch sequence contain Lys in theanalogous position. The R221K mutation was insertedinto wild type and chimeric human PBGS and found tofurther reduce the activity of both, illustrating the subtlenature of the role of this residue.

Keywords Porphobilinogen synthase Æ Zinc ÆMagnesium Æ Potassium

Abbreviations ALA 5-aminolevulinic acid Æ PBGSporphobilinogen synthase bME 2-mercaptoethanol4-OSA, 4-oxosebacic acid 4,7-DOSA 4,7-dioxosebacicacid bis-tris propane 1,3-bis(tris[hydroxymethyl]meth-ylamino propane)

Introduction

The porphobilinogen synthase (PBGS) family of metal-loenzymes is unique in its phylogenetic variation inselectivity for required cations. PBGS is essential to thefirst common step in the biosynthesis of tetrapyrroles(e.g., heme, chlorophyll), is an ancient protein, and ishighly conserved throughout the archaea, eubacteria,and eukarya. The PBGS active site has been defined bymultiple crystal structures, e.g. [1, 2, 3] and the majorityof the active site amino acids are universally conserved.However, an essential catalytic zinc ion, which is seen insome of these active site structures, is not conserved.Although zinc is the only divalent metal ion that hasbeen seen in this region in the current crystal structures,kinetic data indicate that some PBGS do not requirezinc. Of these non-zinc-requiring PBGS, some have nostrictly required divalent metal ions, e.g. [4], othersrequire magnesium for activity [5, 6, 7], and some exhibitsubstantial activation by monovalent cations such aspotassium [4, 6]. Crystal structures have yet to revealhow magnesium binds at the PBGS active site, while avery recent publication shows an active site monovalentcation [8].

The region of PBGS sequence that binds the catalyticzinc includes three cysteine ligands in a continuousstretch of 11 amino acids in the context DXCXCX(Y/F)X3G(H/Q)CG as illustrated in Fig. 1. This unusualbinding motif for a catalytic zinc is present in PBGS fromanimals, yeast, archaea, Gram-positive bacteria, andsome other bacteria [9]. Prior to crystal structure

J Biol Inorg Chem (2003) 8: 176–184DOI 10.1007/s00775-002-0403-x

E.K. JaffeInstitute for Cancer Research,Fox Chase Cancer Center,7701 Burholme Avenue,Philadelphia, PA 19111, USAE-mail: EK_Jaffe@fccc.eduTel.: 215-728-3695Fax: 215-728-2412

determination, this cysteine- and histidine-rich sequencehad been proposed as a zinc-binding site in human PBGS[10]. Shortly thereafter, the sequence of PBGS fromgarden pea revealed the absence of this putative zinc-binding sequence and in its place a glutamic acid-richsequence proposed to bind magnesium [5], a divalentmetal that had previously been implicated as essential tothe plant PBGS activity [11]. Figure 1 includes peaPBGS, whose sequence in this region, DXALDX(Y/F)X3G(H/Q)CG, is representative of all green plants and

many bacteria [9]. Older literature had implicatedmonovalent cations as essential to the activity of PBGSfrom Rhodobacter spheroides [12]. More recent studies ofPBGS from Bradyrhizobium japonicum (B. japonicum)and Pseudomonas aeruginosa (P. aeruginosa) show adramatic response to both magnesium and monovalentcations [4, 6]. The sequences for the metal switch regionof these proteins are also included in Fig. 1. An overlayof the zinc-binding site of human PBGS (PDB code1E51) and the corresponding residues of P. aeruginosaPBGS (PDB code 1B4K) is presented in Fig. 2.

The phylogenetic variation in essential cationsbetween a transition element (zinc), an alkaline earthmetal (magnesium), and a monovalent cation is uniqueto PBGS. Divalent zinc has unique properties thatcontribute to its widespread existence as a catalyticmetal at enzyme active sites [13]. Zn2+ is a soft metal,which readily accepts oxygen, nitrogen, and sulfurligands in a variety of geometries with little change inthe associated free energies [14]. This flexibility allowsZn2+ to readily undergo changes in ligation geometrythrough the course of an enzyme-catalyzed reaction.We have proposed that ligation changes occur on thecatalytic Zn2+ in the PBGS-catalyzed reaction [15].In contrast to Zn2+, Mg2+ is a hard metal ion thatgenerally prefers an octahedral coordination geometrywith predominantly oxygen ligands [14]. Since weknow that the zinc of PBGS is essential to catalysis,and we have proposed a flexible role for this zinc, it isnot at all obvious how magnesium could substitute forzinc in the PBGS-catalyzed reaction. Finally, potassi-um is a monovalent metal ion that is quite differentfrom both zinc and magnesium. Besides the obviouscharge difference, monovalent cations generally have alarger number of first coordination sphere ligandsrelative to divalent cations.

Fig. 1 The metal-binding determinants of PBGS sequences for avariety of PBGS proteins, including the chimeric peainhuman andpsuinhuman described herein. Amino acids are numbered accordingto the human PBGS sequence. The metal switch region is indicatedby a box. Catalytic zinc ligands are on yellow, probable magnesiumligands are on red. The first and second sphere allostericmagnesium ligands are on black background. Only the glutamateis a first coordination sphere ligand, along with five watermolecues. The ligands for a second, non-catalytic zinc are ongreen [1, 20]

Fig. 2 Stereo diagram overlay of the crystal structures of humanPBGS (PDB code 1E51) and P. aeruginosa PBGS (PDB code1B4K) showing the position of the catalytic zinc of human PBGS.The zinc and its ligands are shown as balls. Zinc is dark red; oxygen,nitrogen, and sulfur are colored as cpk. The four zinc ligands arethree sulfur atoms from cysteine and the amino nitrogen ofporphobilinogen. The carbon atoms of the amino acids of humanPBGS are in white, those of P. aeruginosa PBGS are in black. Thecarbons of porphobilinogen, which is bound to human PBGS, arecolored cyan. The carbons of levulinic acid, which is bound toP. aeruginosa PBGS are colored magenta. The three cysteineligands to the zinc are found in the human PBGS metal switchsequence shown in Fig. 1 and are arranged left to right as Cys122,Cys124, and Cys132. The overlay residues of P. aeruginosa PBGSare Ala129, Asp131, and Asp139. Human Asp120 and P. aerugin-osa Asp127 are included to show possible magnesium ligands inP. aeruginosa PBGS. Arg221 of human PBGS and the analogousLys229 of P. aeruginosa PBGS are also shown, far right

177

Despite the differences between zinc and magnesiumand based on sequence alignments, it is generallybelieved that the aspartate-rich metal switch sequencesreflect the presence of a catalytic magnesium. However,this hypothesis has not previously been rigorously test-ed. Dissection of the roles of cations in the PBGS-catalyzed reaction is complicated by the existence ofadditional metal ion binding sites elsewhere in theprotein structure. For instance, most PBGS contain anallosteric magnesium site, which is spatially distinct fromthe strictly required metal ions. The role of thismagnesium has been best described for E. coli PBGS[16]. The allosteric magnesium is apparent in the crystalstructures of E. coli and P. aeruginosa PBGS (e.g., PDBcodes 1B4K [2] and 1I8J [3]) such that the sequencedeterminants can be deduced and this information isalso included in Fig. 1. Based on sequence data from�130 different species, we conclude that the only PBGSthat do not use the allosteric magnesium are from ani-mals, yeast, and the bacterial genus Rhodobacter wherethe first coordination sphere glutamate is altered [9]. Inanimal and yeast PBGS, a positively charged arginine issituated at the site of this magnesium [17]. Interpretationof mutagenic studies of the metal switch region of PBGSare confounded if the parent PBGS contains the allos-teric magnesium. Hence, the current study uses humanPBGS, which does not contain the allosteric magnesium-binding site, to investigate how the metal switchsequence dictates the essential cations of PBGS. Thechimeric human PBGS variants described herein containthe metal switch region of pea PBGS and P. aeruginosaPBGS; the chimeric sequences are included in Fig. 1.A complementary study by O’Brian and coworkersincluded a cassette mutant of B. japonicum PBGS thathad a human-like metal switch region inserted [18]. Thisvariant was shown to have acquired a zinc requirement,but it was not reported how the variant responded toeither the allosteric magnesium or monovalent cationsthat have also been shown to affect native B. japonicumPBGS activity [6].

Materials and methods

All reagents were of the highest quality available. Human PBGSwas expressed from an artificial gene construct as previouslyreported [19] using the QuikChange method to incorporate varia-tions in the amino acid sequence. The primers for the peainhuman,psuinhuman, and R221K mutations were respectively GCCACT-GACGTCGCCCTGGATCCGTACTCTTCTGACGGTCACGA-CGGTCTCCTC, GCCACTGACGTCGCCCTGGATCCGTTC-ACTACTCACGGTCAGGACGGTCTCCTGAGCGAGAAC,and GGCGACAAGCGCTGCTATCAGCTGCC. The resultingplasmids were sequenced throughout the artificial gene in bothdirections. The wild type PBGS was the N59/C162A variant aspreviously reported [20]. The purification of the variants followedthe published protocol through the 45% ammonium sulfateprecipitation. The protein pellet was redissolved in 30 mM KPi,pH 7.5, 10 lM Zn2+, 10 mM bME, 0.1 mM PMSF, 15%ammonium sulfate and adsorbed to a 200 mL phenyl Sepharosecolumn equilibrated in the same buffer. The protein eluted at theend of a 1 L gradient to 2 mM KPi, pH 7.5, 10 lM Zn2+, 10 mM

bME, 0.1 mM PMSF or in a one column volume wash of the finalbuffer following the gradient. PBGS-containing fractions werepooled and pumped onto an 200 mL DEAE Biogel column equil-ibrated in 30 mM KPi, pH 7.5, 10 lM Zn2+, 10 mM bME,0.1 mM PMSF and eluted with a 5 column-volume salt gradient ofthe same buffer with a final KCl concentration of 0.4 M. HumanPBGS elutes about midway through the gradient. Inactive humanPBGS variants can be pooled without contamination fromchromosomally encoded E. coli PBGS by analyzing the activity ofcolumn fractions in the presence and absence of 1 mM Mg2+,which gives a two-fold stimulation to E. coli PBGS activity atpH 7. The DEAE separation of a low-activity human PBGS vari-ant from E. coli PBGS has been illustrated previously [21]. Thehuman PBGS (and variants)-containing fractions were pooled andconcentrated to �10 mg/mL prior to the final purification on a 1 mlong Sephacryl S300 in 0.1 M KPi, pH 7, 10 mM bME, 10 lMZn2+. Depending on the amount of protein recovered from theDEAE column, either a 270 mL or a 2 L Sephacryl column wasemployed.

Activity assays

PBGS activity was determined at 37 �C using fixed time assays at0.1 M bis-tris propane-HCl (pH as noted), 10 mM 2-mercapto-ethanol, 10 mM ALA-HCl. Sufficient PBGS (5–100 lg/ml) andincubation times (5 min–16 h) were used to ensure a reliable A555

reading (0.05–0.9) when porphobilinogen was detected throughcomplexation with Ehrlich’s reagent (2.0 g p-dimethylaminobenz-aldehyde, dissolved in 50 mL glacial acetic acid, to which is added20 mL concentrated perchloric acid, and finally brought to 100 mLwith glacial acetic acid) as previously described [7]. If necessary,stopped reaction mixtures were diluted prior to workup withEhrlich’s reagent. Mammalian PBGS has previously been shownto exhibit extraordinary stability during long incubations at 37 �Cin the presence of substrate, e.g. [22]. Detailed procedures forpH activity profiles, Km and Vmax determinations, and inactivationby 4-oxosebacic acid (4-OSA) and 4,7-dioxosebacic acid(4,7-DOSA) have been previously described [3, 15, 20]. Most assayconditions are listed in the results section or in the figure legends.

Results

Peainhuman PBGS

From 10 g of cell paste, the yield of peainhuman PBGSwas 67 mg, which is comparable to yields of wild-typeprotein. The chromatographic properties were much thesame as the wild type thus suggesting no gross structuralchanges. The maximum activity observed for peainhu-man PBGS was �0.1 lmol h–1 mg–1 (see below), whichcorresponds to about 0.25% and 0.05% of the maximalactivity of human or pea PBGS, respectively. The puri-fication buffers contained 10 lM zinc, as is used forwild-type human PBGS, which purifies with about 8 zincbound per octamer as determined by atomic absorptionspectroscopy [20]. Upon purification, peainhuman PBGSwas found to contain �2 zinc per octamer, consistentwith disruption of the zinc ligands.

In order to obtain a maximal activity for peainhumanPBGS, it was necessary to add metal ions to the assaybuffer. As an initial screen, peainhuman PBGS wasassayed for 4 h using 100 lg/mL enzyme in the bufferbis-tris propane-HCl at pH values of 7, 8, and 9 in thepresence and absence of added zinc, magnesium, and/or

178

potassium. Based on prior characterization of PBGSfrom a wide variety of species, the metal concentrationswere chosen to be 10 lM Zn2+, 1 mM Mg2+, and100 mM K+. These assays contained residual metal ionsfrom the protein purification buffer at �0.1 lM zinc and�2 mM potassium. The results at pH 8 are illustrated inFig. 3 and similar trends are seen at pH 7 and 9. Ashypothesized, the activity of peainhuman PBGS issignificantly enhanced by the addition of magnesium,but not by zinc. Potassium activation of peainhumanPBGS was unexpected since the activity of neither peanor human PBGS responds to potassium. The activityobserved in the absence of added metals is attributed tothe low level of potassium present in these assays. Themarginal stimulation of activity by zinc is reproducibleand ascribed to the presence of the second zinc site ofhuman PBGS that remains in the peainhuman protein(see Fig. 1). The small kinetic role of this zinc has beenpreviously described [20]. At the metal ion concen-trations used here, the maximal activity observed wasin the simultaneous presence of both magnesium andpotassium.

The pH vs. activity profiles of human PBGS and peaPBGS differ substantially [7, 20]. Fig. 4A illustrates thepH rate profiles of human PBGS, pea PBGS, and pea-inhuman PBGS, each carried out at a substrate concen-tration of 10 mM ALA, a buffer concentration of 0.1 Mbis-tris propane-HCl, and with the addition of an opti-mal configuration metal ions. The metal ion concentra-tions were 10 lM Zn2+ for human PBGS, 10 mMMg2+ for pea PBGS, and 1 mM Mg2+ with 0.1 M K+

for peainhuman PBGS. For the human protein, optimalactivity is seen between pH values of �6.5–7.2 and thisactivity is dependent upon a mildly acidic deprotonationphenomenon (pKa�5.3) which has previously beenassociated with binding of the catalytic zinc and adramatic rise in the Km for ALA below this pKa [20].Human PBGS activity is reduced upon a second

deprotonation, pKb of 8.1, the basis of which has not beexperimentally determined. In contrast, the pH rateprofile of pea PBGS shows an optimum betweenpH�7.8–9.0 with an apparent pKa�7.3 [7]. Since peaPBGS does not have a catalytic zinc, the chemical basisfor the pKa of pea PBGS must be different from that ofhuman PBGS. Like pea PBGS, peainhuman PBGSshows a mildly basic pH optimum, but the apparentpKa value for the activating deprotonation is aboutmidway between that of human PBGS and pea PBGS(pKa�6.6). The basic pKb�9.6 is also significantlyhigher than seen for human PBGS. Hence, it appearsthat the metal switch sequence and the identity of theactive site metal ion have a dramatic effect on the pHrate profile.

Activation profiles were obtained for peainhumanPBGS as a function of both magnesium and potassiumat a fixed pH value of 8 and a fixed concentration ofALA (10 mM). In this case, the protein was extensivelydialyzed against bis-tris propane, pH 8, 10 mM bME, inorder to minimize the amount of preassociated metalions. Figure 4B illustrates the magnesium activationprofile in the presence and absence of 0.1 M KCl.In both cases there is small but measurable activity inthe absence of added magnesium (vo) and a simplethree parameter binding curve v ¼ vo þ V 0 � Mg2þ� ����

tðKact Mg2þ½ � þ Mg2þ� �ÞÞÞ gives a good fit with a Kact Mg2þ½ �

of 110 lM. For the data obtained in the presence ofpotassium, a Hill equation gave a significantly better fitwith a Kact Mg2þ½ � of 170 lM and a Hill coefficient of 0.57.The residual activity observed in the absence of addedmetal ions is found to be an intrinsic property ofpeainhuman PBGS. This activity is not sensitive toinhibition by 4-oxosebacic acid or by 1,10-phenanthro-line, both of which would inhibit a trace (0.025%)contamination by E. coli PBGS, if present. Figure 4Cillustrates the potassium activation curve in the absenceand presence of 10 mM Mg2+. In the absence of Mg2+,theKact Kþ½ � ¼ 7:2 mM. At the saturating concentration of10 mM Mg2+, K+ does not provide any additionalactivation. Km and Vmax values were determined forpeainhuman PBGS at pH 8 in 10 mM Mg2+ and foundto be 68 lM and 0.156 lmol h–1 mg–1. The Km and Vmax

at 1 mM Mg2+ and 0.1 M K+ were quite similar at66 lM and 0.138 lmol h–1 mg–1. The Km values arecomparable to those observed for human PBGS at itspH optimum (�100 lM) [20].

The human PBGS variant R221K

There is a correlation between the amino acid sequenceof the metal switch region and a basic residue thatcorresponds to position 221 for human PBGS, which isincluded in turquoise in Fig. 1 and illustrated in Fig. 2.Those PBGS that have the cysteine-rich metal switchregion contain arginine in this position, and those thatdo not have the cysteine-rich sequence contain lysine inthis position [9]. Hence, we posited that an R221K

Fig. 3 The activity of purified peainhuman PBGS at pH 8 in bis-tris propane as a function of added metal ions. These were 4-hassays containing �100 lg of peainhuman PBGS

179

mutation might enhance the poor activity of peainhumanPBGS. The R221K mutation was incorporated into thesequences of human PBGS and peainhuman PBGS. Infact, the optimal activity of both R221K variants wasreduced by approximately an order of magnitude rela-tive to the parent. The maximal activity for R221Khuman PBGS is �4.5 lmol h–1 mg–1 and that of R221Kpeainhuman PBGS is �0.0025 lmol h–1 mg–1. TheR221K human PBGS variant (at 10 lM Zn2+) had analtered pH rate profile relative to the parent (pKb�7.5),as is shown in Fig. 4A. Replacement of the Zn2+ with1 mM Mg2+ or 0.1 M K+ did not effect the pH rateprofile. Because the protein is isolated in 10 lM Zn2+,omission of Zn2+ also did not affect the pH rate profile.The activity of R221K peainhuman PBGS was too low toallow meaningful interpretation of the pH rate profile(data not shown). Hence, the simplistic hypothesis aboutthe catalytic contribution of the basic residue at position221 of human PBGS was not supported.

Psuinhuman PBGS

Psuinhuman PBGS (74 mg) was isolated from 13.1 g ofcell paste. In this case, the major peak of PBGS proteinwas identified by SDS PAGE and found to be almostcompletely inactive. The psuinhuman PBGS proteineluted from the phenyl sepharose column following the1 L gradient and preceded a peak containing PBGSactivity that was ascribed to chromosomally encodedE. coli PBGS. The main peak of protein eluting from theDEAE column was >98% pure by SDS PAGE and theprotein was not further purified. The low activity presentin psuinhuman PBGS protein was investigated as afunction of pH in the presence and absence of 10 lMzinc, 1 mM magnesium, and 0.1 M potassium. In allcases the maximum observed activity was on the orderof 0.005 lmol h–1 mg–1 as illustrated in Fig. 5A. The pHrate data in the presence of magnesium or potassiumwas fit to a bell curve yielding an acidic pKa=6.6±0.1and a basic pKb=9.9±0.1. The pH rate data, in thepresence or absence of zinc, fit similarly to an acidic

pKa=6.8±0.1 and a basic pKb=9.7±0.1. Although thepH rate profiles resembled those previously character-ized for E. coli PBGS, this activity is not ascribed to acontamination by native E. coli PBGS because it is in-sensitive to inhibition by 4-OSA under conditions whereE. coli PBGS would be inactivated by this inhibitor.

A magnesium activation profile for psuinhuman PBGSat 0.1 Mbis-tris propane, pH 8.1, is illustrated in Fig. 5B,where it is seen that Mg2+ provides about a 60% stimu-lation of the intrinsic activity with a remarkably tightKact Mg2þ½ � of �16.3 lM. A three-parameter hyperbolic fit

Fig. 4A–C The kinetic characterization of peainhuman PBGS.Each data point represents an individual assay. A Activity as afunction of pH in bis-tris propane-HCl for human PBGS at 10 lMZn2+ (circles, maximal activity �40 lmol h–1 mg–1); pea PBGS at10 mM Mg2+ (squares, maximal activity �200 lmol h–1 mg–1);peainhuman PBGS at 1 mM Mg2+, 100 mM K+ (upside downtriangles, maximal activity �0.1 lmol h–1 mg–1); and R221Khuman PBGS at 10 lM Zn2+ (diamonds, maximal activity�4.5 lmol h–1 mg–1). All assays were carried out in duplicate; inmost cases the variation between duplicates is smaller than the sizeof the symbols. B Activity of peainhuman PBGS (upside-downtriangles) at pH 8 as a function of Mg2+ in the presence (gray-filled) and absence (white-filled) of 0.1 M potassium. The solid linesrepresent a simple three-parameter activation curve (see text). Thedashed line is to a four-parameter Hill equation. C Activity ofpeainhuman PBGS (upside-down triangles) as a function of K+ inthe presence (gray-filled) and absence (black) of 10 mM Mg2+. Theblack line represents a three-parameter activation curve

c

180

shows vo=0.0065 lmol h–1mg–1 andVmax=0.0094 lmolh–1 mg–1. Stimulation of a similar magnitude is also seenby the addition of 0.1 M K+.

Inhibition by 4-OSA and 4,7-DOSA

4,7-DOSA is a potent time-dependent inactivator ofhuman PBGS (Ki 100m �1 lM) and a poor inhibitor ofpea PBGS (Ki 100m ‡2 mM) [3]. The peainhuman PBGSwas evaluated following a 100 min preincubation with4,7-DOSA over the concentration range of 0.1–1000 lM. Under the conditions of a 6-h assay at 1 lMprotein, the Ki 100m for peainhuman PBGS was deter-mined to be �0.3 mM (data not shown). Controlexperiments, where 4,7-DOSA was added with thesubstrate, showed no inhibition. The low sensitivity ofpeainhuman PBGS supports our prior conclusion that4,7-DOSA is a much more potent inhibitor of PBGS

that use a catalytic zinc than those that do not [3]. Incontrast, 4-OSA inactivation of PBGS has been found tobe specific for E. coli PBGS [15]. 4-OSA is not a potentinhibitor of either human PBGS or pea PBGS. 4-OSAinhibition was tested using an overnight incubation ofprotein with 3 mM 4-OSA prior to addition of sub-strate, which would completely inactivate E. coli PBGS.Under their optimal conditions, peainhuman PBGSretained 69% activity and psuinhuman PBGS retained84% activity when preincubated with 4-OSA, which iscomparable to the results with human PBGS [15].

Discussion

The hypothesis that the cysteine-rich metal switch regionof PBGS dictates a kinetic requirement for zinc [10] wassomewhat controversial at first because catalytic zincions generally do not contain a cysteine-rich coordina-tion environment [23, 24]. However, at this point it isunequivocally established that there is a required cyste-ine-rich zinc at the active site of many PBGS [1]. Thiszinc has the three cysteines as ligands, in the generalsequence DXCXCX(V/F)X3G(H/Q)CG [9]. The active-site zinc is proposed to be pentadentate when bound tothe substrate 5-aminolevulinate (ALA) [15]. The coor-dination is to A-side ALA, which is the substrate mol-ecule where C3 and C4 are incorporated into the pyrrolering, as illustrated in Fig. 6. C1 and C2 become theacetyl moiety of porphobilinogen and C5 retains theamino moiety. Zinc ligation of A-side ALA is proposedto be through the C4 carbonyl oxygen and the C5 aminogroup. The C4 carbonyl oxygen is eventually lost aswater. Some PBGS crystal structures (e.g., PDB code1I8J) show the C4-derived oxygen as a zinc ligand [3].However, in the enzyme-product complex (e.g., PDBcode 1E51), this water molecule is no longer associatedwith the zinc, which is seen in a tetrahedral geometrywith the three cysteine ligands and the amino group ofporphobilinogen as the fourth ligand, as illustrated inFig. 2. This and other data implicate the active site zincof PBGS as functional in the binding and reactivity ofthis substrate molecule.

The complementary hypothesis that an aspartate-richmetal switch region of PBGS dictates a kinetic require-ment for magnesium was first proposed upon deducingthe sequence of a plant PBGS [5]. This reasonable

Fig. 5A,B Activity of psuinhuman PBGS. A As a function of pH inthe absence of added metals (triangles), presence of 10 lM Zn2+

(upside-down triangles), presence of 10 mM Mg2+ (squares), andpresence of 0.1 M K+ (circles). Each symbol represents onedetermination. The lines indicate a non-linear best fit to a two pKa

bell curve v ¼ Vmax�

1þ 10 pKa�pHð Þ þ 10 pH�pKbð Þ� �� �where the data

are pooled (dashed line indicates the presence and absence of Zn2+,solid line indictes the presence of Mg2+ or K+). B Mg2+ activationat pH 8. The solid line represents a hyperbolic fit to the data

Fig. 6 The PBGS-catalyzed reaction. The red substrate is A-sideALA, the blue substrate is P-side ALA, and the black atomsillustrate the source of the product water

181

hypothesis was put forth without an appreciation for theadditional, more loosely binding, allosteric magnesium,which binds elsewhere in the sequence, and is now seento exist in most PBGS [9]. Later kinetic characterizationof PBGS from B. japonicum, P. aeruginosa, and P. sat-ivum PBGS, which each contain a variation of theaspartate-rich metal switch region (see Fig. 1), showed avariety of metal ion responses. The B. japonicum proteinappeared to require and bind a catalytic magnesium at astoichiometry of four per octamer, while binding theallosteric magnesium at a stoichiometry of eight peroctamer [6]. The pH vs. activity profile of B. japonicumPBGS was dependent upon monovalent cations, suchthat there appeared to be a potassium requirement whenassays were performed at neutral pH [6]. The P. aeru-ginosa PBGS contains the allosteric magnesium at astoichiometry of four per octamer and does not bindmagnesium at the active site [2, 4]. P. aeruginosa PBGSactivity also responds to monovalent cations [4]. Acti-vation of pea (P. sativum) PBGS by magnesium suggeststhe presence of both a tight binding catalytic magnesiumand a looser binding allosteric magnesium at a totalstoichiometry of 16 magnesium per octamer at 10 mMmagnesium [7]. Pea PBGS activity does not respond tomonovalent cations. These results suggested thatmonovalent cations might activate when there are fewerthan 16 magnesium sites per octamer, but did not es-tablish the relationship between the aspartate-rich metalswitch region and a requirement for an active sitemagnesium. Of particular concern was the structure andkinetic data obtained for P. aeruginosa PBGS [2, 4],which does not contain magnesium bound to theaspartate-rich metal switch region at the active site, asillustrated in Fig. 2.

Coordination of the amino group of porphobilinogento the active-site zinc was originally proposed on thebasis of 13C and 15N chemical shifts for enzyme-boundisotopically labeled porphobilinogen [25]. The enzymesfor which these chemical shifts were observed were thePBGS of bovine and E. coli, both of which contain thecysteine-rich metal switch sequence and both of whichrequire a catalytic zinc. The 13C chemical shifts of[3,5-13C]-porphobilinogen have also been obtained forthe aspartate-rich metal switch containing PBGS fromB. japonicum and P. aeruginosa PBGS [4, 26]. In the caseof B. japonicum and P. aeruginosa PBGS, the chemicalshifts were the same as each other, but distinctly differ-ent from that of the zinc-containing PBGS. The differentchemical shifts most probably reflect the fact that theamino group of the product is not coordinated to ametal ion for the PBGS with the aspartate-rich metalswitch sequence.

The current study addresses the function of the metalswitch sequences of pea PBGS and P. aeruginosa PBGSin the absence of the confounding kinetic phenomenoncontributed by the allosteric magnesium. The resultsclearly show that removing the cysteine-rich metalswitch sequence abrogates the zinc requirement. Theresults also show that the aspartate-rich metal switch

region can confer a requirement for a catalytic magne-sium, with a Kd similar to that of pea PBGS(Kreq=35 lM) [7]. In the context of the human PBGS,these two aspartate-rich metal switch sequences bothconfer a somewhat sloppy metal ion response, whereineither magnesium or potassium is equally effective, withthe Kd for potassium being about an order of magnitudeweaker than that of magnesium. The factors that de-termine the presence (e.g., P. aeruginosa and B. japoni-cum PBGS) or absence (pea PBGS) of the potassiumresponse in naturally occurring PBGS are not clear fromthe metal binding determinants included in Fig. 1, andare not further elucidated through characterization ofthe chimeric human proteins peainhuman or psuinhumanPBGS.

It is interesting to note that, in the context of thehuman PBGS, the aspartate-rich metal switch domainsof both pea PBGS and P. aeruginosa PBGS cause adramatic reduction in Vmax relative to human PBGS.The reduced activity of these chimeric human PBGSmay be ascribed to the different inherent lability of zincligands and coordination geometry relative to magne-sium. Zinc ligands are generally more labile and themetal ion changes coordination geometry during catal-ysis. The human protein is optimized to respond to theensuing structural changes. In contrast, magnesium li-gands are far less labile and magnesium does not readilyundergo the geometric changes for which the humanprotein is optimized. In addition, the metal ion con-stellation, wherein an aspartate-rich metal switch se-quence exists in the absence of the allosteric magnesium,is very rare in naturally occurring PBGS. In contrast arethe studies done by O’Brian and coworkers wherein theplacement of a cysteine-rich metal switch domain inB. japonicum PBGS created a zinc-requiring enzymewhose activity at optimal pH was reduced by only afactor of two relative to the parent enzyme [18]. In thiscase, we postulate that the protein is optimized for theinflexible magnesium ion and propose that zinc canreadily adopt the environment that the protein dictates.In addition, the chimeric B. japonicum PBGS resemblesmany naturally occurring PBGS that contain the cyste-ine-rich metal switch domain in addition to the allostericmagnesium determinants. For example, see the crystalstructures of E. coli PBGS (e.g., PDB code 1I8J). In thisregard, it is important to note that of 133 differentspecies of PBGS recently culled from the databases, onlythe bacterial genus Rhodobacter has a PBGS with anaspartate-rich metal switch domain in the absence of theallosteric magnesium binding determinants [9]. It isironic that R. spheroides PBGS was one of the first to becharacterized as representative of all photosyntheticorganisms [12].

The pH rate profiles for PBGS from different speciesshow substantial variation, as illustrated for human andpea PBGS in Fig. 4A. In light of the pH rate profiles forpeainhuman and psuinhuman PBGS, as well as the pre-viously reported chimeric B. japonicum PBGS [18], it istempting to ascribe the observed differences in the pKa

182

to the different active site metal ions. Unfortunately,insufficient data yet exist to definitively assign moleculardetails to the observed pH rate profiles of human or peaPBGS. The order of bond-making and bond-breakingsteps and/or the rate-determining step may be differentfor PBGS that use zinc relative to those that do not. Theobservation that the R221K mutation on human PBGScauses a shift in pKb is the first indication of thechemical basis for this ionization. One speculation isthat pKb is related to the role of Arg221 in catalysis, andthat a lower pH is required to generate the appropriatecharge on the replacement by lysine. Several PBGSstructures (human, yeast, and E. coli) show hydrogenbonding between this arginine and the carboxyl groupthat derives from A-side ALA. This interaction wouldorient A-side ALA for catalysis and contribute to theKm, which is dominated by the binding of A-side ALA.The factors affecting the pH rate profile for pea PBGSare much less well understood. In pea PBGS there arecomplex interactions between the active site, the allos-teric magnesium ion, and the oligomerization state ofthe protein, which have as yet confounded elucidation ofthe chemical and kinetic mechanism of the magnesium-utilizing PBGS [7]. The intrinsic activity of peainhumanPBGS is much lower than the parent proteins and henceit does not serve as a reliable model for the PBGS thatdo not utilize an active-site zinc.

Evolutionary rationale for the metal switch of PBGS

A recent phylogenetic analysis of PBGS sequences re-vealed that the archaeal kingdom of cellular organismscontain the cysteine-rich metal switch sequence, whichsuggests that the cysteine-rich PBGS is the ancestralprecursor to the aspartate-rich version [9]. The latteris found in most photosynthetic eubacteria and allphotosynthetic eukaryotes, and we have presented athermodynamic rationale for the evolution of theaspartate-rich PBGS in photosynthetic organisms [27].In photosynthetic eukaryotes, PBGS functions in thechloroplasts along with the other enzymes that functionin chlorophyll biosynthesis. One necessary step in chlo-rophyll biosynthesis is the thermodynamically difficultinsertion of magnesium into the planar and nitrogen-rich tetrapyrrole core protoporphyrin IX. This unfa-vorable reaction is in direct competition with the spon-taneous insertion of zinc into protoporphyrin IX [28].Hence, to help counteract this thermodynamic dilemma,one might argue that chlorophyll biosynthesis might bestbe carried out in the presence of relatively high con-centrations of free magnesium and relatively low con-centrations of labile zinc. In fact, the concentration ofmagnesium in chloroplasts during active chlorophyllbiosynthesis is about 10 mM. Thus, the thermodynamicdifficulty of inserting magnesium into protoporphyrinIX is cited as the evolutionary pressure against a chlo-rophyll biosynthetic enzyme (e.g., PBGS) whose activityis dependent upon zinc. The work presented herein

demonstrates that the activity of PBGS containing theaspartate-rich metal switch sequence do not dependupon zinc, but rather respond favorably to the presenceof magnesium and/or potassium.

It is interesting to note that the phylogenetic analysisof PBGS from �130 species currently in the sequencedatabases revealed only one species containing twogenes encoding PBGS. This is the cyanobacterial species,Nostoc sp. PCC 7120, which has one copy of a PBGSwith the cysteine-rich metal switch and one copy ofa PBGS with the aspartate-rich PBGS [29]. The orien-tation of these genes suggests a gene duplication event,wherein the aspartate-rich sequence was able to evolvethrough functionally impotent forms in the presence of afunctional gene with the cysteine-rich domain. Thepresence of this gene duplication in cyanobacteria issignificant as the cyanobacteria are cited as the firstorganisms to develop photosynthesis and are believedto be the endosymbiotic precursors to chloroplasts.

Acknowledgements Mr. Jake Martins and Ms. Linda Stith areacknowledged for their outstanding technical support. The DNASynthesis Facility and the DNA Sequencing Facility were used inthe preparation of this manuscript. This work was supported bygrant number ES03654 from the National Institute of Environ-mental Health Sciences, NIH, by NIH grant CA06927 (ICR), andby an appropriation from the Commonwealth of Pennsylvania.

References

1. Erskine PT, Senior N, Awan S, Lambert R, Lewis G, Tickle IJ,Sarwar M, Spencer P, Thomas P, Warren MJ, Shoolingin-Jordan PM, Wood SP, Cooper JB (1997) Nat Struct Biol4:1025–1031

2. Frankenberg N, Erskine PT, Cooper JB, Shoolingin-JordanPM, Jahn D, Heinz DW (1999) J Mol Biol 289:591–602

3. Kervinen J, Jaffe EK, Stauffer F, Neier R, Wlodawer A,Zda\nov A (2001) Biochemistry 40:8227–8236

4. Frankenberg N, Jahn D, Jaffe EK (1999) Biochemistry38:13976–13982

5. Boese QF, Spano AJ, Li JM, Timko MP (1991) J Biol Chem266:17060–17066

6. Petrovich RM, Litwin S, Jaffe EK (1996) J Biol Chem271:8692–8699

7. Kervinen J, Dunbrack RL Jr, Litwin S, Martins J, ScarrowRC, Volin M, Yeung AT, Yoon E, Jaffe EK (2000) Biochem-istry 39:9018–9029

8. Frere F, Schubert WD, Stauffer F, Frankenberg N, Neier R,Jahn D, Heinz DW (2002) J Mol Biol 320:237–247

9. Jaffe EK (2003) Chemistry and Biology (submitted)10. Wetmur JG, Bishop DF, Cantelmo C, Desnick RJ (1986) Proc

Natl Acad Sci USA 83:7703–770711. Liedgens W, Lutz C, Schneider HAW (1983) Eur J Biochem

135:75–7912. Nandi DL, Shemin D (1968) J Biol Chem 243:1236–124213. Kimura E, Kikuta E (2000) J Biol Inorg Chem 5:139–15514. Bock CW, Katz AK, Glusker JP (1995) J Am Chem Soc

117:3754–376515. Jaffe EK, Kervinen J, Martins J, Stauffer F, Neier R, Wloda-

wer A, Zdanov A (2002) J Biol Chem 277:19792–1979916. Jaffe EK, Ali S, Mitchell LW, Taylor KM, Volin M, Markham

GD (1995) Biochemistry 34:244–25117. Jaffe EK (2000) Acta Crystallogr D Biol Crystallogr 56:115–12818. Chauhan S, Titus DE, O’Brian MR (1997) J Bacteriol

179:5516–5520

183

19. Jaffe EK, Volin M, Bronson-Mullins CR, Dunbrack RL Jr,Kervinen J, Martins J, Quinlan JF Jr, Sazinsky MH, Stein-house EM, Yeung AT (2000) J Biol Chem 275:2619–2626

20. Jaffe EK, Martins J, Li JM, Kervinen J, Dunbrack RL Jr(2001) J Biol Chem 276:1531–1537

21. Mitchell LW, Volin M, Martins J, Jaffe EK (2001) J Biol Chem276:1538–1544

22. Jaffe EK, Markham GD (1987) Biochemistry 26:4258–426423. Vallee BL, Auld DS (1990) Biochemistry 29:5647–564924. Jaffe EK (1993) Comments Inorg Chem 15:67–93

25. Jaffe EK, Markham GD, Rajogopalan JS (1990) Biochemistry29:8345–8350

26. Petrovich RM, Jaffe EK (1997) Biochemistry 36:13421–1342727. Jaffe EK (1995) J Bioenerg Biomembr 27:169–17928. Baum SJ, Plane RA (1966) J Am Chem Soc 88:910–91329. Kaneko T, Nakamura Y, Wolk CP, Kuritz T, Sasamoto S,

Watanabe A, Iriguchi M, Ishikawa A, Kawashima K, KimuraT, Kishida Y, Kohara M, Matsumoto M, Matsuno A, MurakiA, Nakazaki N, Shimpo S, Sugimoto M, Takazawa M,Yamada M, Yasuda M, Tabata S (2002) DNA Res (in press)

184