A structural basis for half-of-the-sites metal binding revealed in Drosophila melanogaster porphobilinogen synthase.

Porphobilinogen synthase (PBGS) proteins fall into several distinct groups with different metal ion requirements. Drosophila melanogaster porphobilinogen synthase (DmPBGS) is the first non-mammalian metazoan PBGS to be characterized. The sequence shows the determinants for two zinc binding sites known to be present in both mammalian and yeast PBGS, proteins that differ in the exhibition of half-of-the-sites metal binding. The pH-dependent activity of DmPBGS is uniquely affected by zinc. A tight binding catalytic zinc binds at 0.5/subunit with a Kd well below microm. A second inhibitory zinc exhibits a Kd of approximately 5 microm and appears to bind at a stoichiometry of 1/subunit. A molecular model of DmPBGS suggests that the inhibitory zinc is located at a subunit interface using Cys-219 and His-10 as ligands. Zinc binding to this previously unknown inhibitory site is proposed to inhibit opening of the active site lid. As predicted, the DmPBGS mutant H10F is active but is not inhibited by zinc. H10F binds a catalytic zinc at 0.5/subunit and binds a second nonessential and noninhibitory zinc at 0.5/subunit. This result reveals a structural basis for half-of-the-sites metal binding that is consistent with a reciprocating motion model for function of oligomeric PBGS.

The porphobilinogen synthases (PBGS) 1 are a family of highly homologous homo-octameric proteins responsible for catalyzing the first common step in the biosynthesis of a broad range of tetrapyrrole pigments such as heme, vitamin B12, chlorophyll, and cofactor F430 of the methanogenic bacteria (1). PBGS is also known as 5-aminolevulinate dehydratase or ALAD. The most significant phylogenetic difference among PBGS proteins is in the constellation of metal ions at catalytic and allosteric sites (2,3). Yeast and mammalian PBGS share the sequence determinants for two zinc binding sites, one of which is absent from the PBGS of any archaea, bacteria, protist, or photosynthetic eucarya (3). The human and yeast PBGS both contain the sequence determinants for a catalytic active site zinc (also known as ZnB), as well as a second non-essential zinc (also known as ZnA) (4), but they differ in metal binding stoichiometry. For instance, in the case of human PBGS, the catalytic zinc shows half-of-the-sites binding at a stoichiometry of 4/homo-octamer (4) (PDB accession number 1E51), 2 whereas the fungal enzyme binds the catalytic zinc at 8/octamer (6) (PDB accession number 1AW5). Some bacterial PBGS show half-of-the-sites metal binding at catalytic and/or allosteric sites (e.g. Bradyrhizobium japonicum and Pseudomonas aeruginosa), whereas others (e.g. Escherichia coli) do not.
The current sequence databases contain PBGS sequences from ϳ130 different organisms. Of these, the PBGS of Drosophila melanogaster (DmPBGS) is the only complete non-mammalian and non-fungal PBGS sequence that shows the sequence determinants for the two zinc binding sites found in yeast and mammalian PBGS. The gene encoding PBGS is absent from the completed genome of Caenorhabditis elegans and not yet verified in other incomplete non-mammalian metazoan genomes such as that of Danio rerio. Hence, to help deduce the molecular determinants for expression of the half-of-the-sites metal binding phenomenon, we obtained the expressed sequence tag encoding DmPBGS and cloned, expressed, purified, and characterized the protein, as described herein. Unexpectedly, DmPBGS was found to interact with zinc somewhat differently from either mammalian or yeast PBGS. Investigation of these differences provides novel insight into the structural basis for the half-of-the-sites metal binding phenomenon.
Characterization of DmPBGS is also significant because insects are among the most abundant metazoan species on earth and can act as agricultural pests or human disease vectors. For many insects, flight is essential and depends upon aerobic respiration; consequently, tetrapyrroles (hemes) play an important role. Hence, species-specific inhibition of tetrapyrrole biosynthesis might prove useful in control of insects and or insectborne diseases. Because PBGS has recently been identified to have species-specific sensitivity to certain active site-directed inhibitors (7,8), we have characterized DmPBGS with these inhibitors.

EXPERIMENTAL PROCEDURES
Cloning and Expression of DmPBGS-The expressed sequence tag containing the gene for DmPBGS (catalogue number 98002) was purchased from ResGen. The gene was amplified by PCR with the addition of the NdeI and BamHI sites, using the primers 5Ј-GCTAAGCGAAC-CATATGGAGCGGAAACTGC-3Ј and 5Ј-CGCATGTACGGATCCACAT-GGTATCAAGACATCGG-3Ј, where the restriction sites are underlined. The amplified gene was digested with NdeI and BamHI and ligated directly into pET17b, which had also been digested with NdeI and BamHI.
The sequences of several resulting plasmids were determined * This work was supported by NIEHS, National Institutes of Health (NIH) Grant ES03654 (to E. K. J.), by NCI, National Institutes of Health Grant CA06927 awarded to the Institute for Cancer Research, and by an appropriation from the Commonwealth of Pennsylvania. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 1 The abbreviations used are: PBGS, porphobilinogen synthase; 4-OSA, 4-oxosebacic acid; 4,7-DOSA, 4,7-dioxosebacic acid; DmPBGS, D. melanogaster porphobilinogen synthase; ␤-ME, 2-mercaptoethanol; ALA, 5-aminolevulinic acid; o-phe, 1,10-phenanthroline. throughout the gene in both directions using a series of internal and external primers. An error-free plasmid was denoted pJMPBGS and was used for further study. pJMPBGS was transformed into E. coli strain BLR(DE3) and the CodonPlus variant BL21(DE3)-RIL for expression using the procedure we had previously used for pMVhum (9). The DmPBGS mutant H10F was prepared by the QuikChange method (Stratagene) using the sense strand primer 5Ј-GCACAGTGGAATGT-TCCATGCCACGCTGCGGC-3Ј. All oligonucleotides were synthesized in the Fannie E. Rippel Biochemistry and Biotechnology Facility at Fox Chase Cancer Center (FCCC). DNA sequence determination was in the FCCC DNA Sequencing Core Facility.
DmPBGS purification initially followed the protocol for human PBGS expressed from pMVhum (9). The protocol was altered as differences were observed. The final purification protocol followed the published procedures up to the ammonium sulfate cut, which was altered to use the protein that precipitated between 30 and 50% ammonium sulfate. This protein was redissolved in 30 mM KPi, pH 7.5, 10 M Zn 2ϩ , 10 mM 2-mercaptoethanol (␤-ME), 0.1 mM phenylmethylsulfonyl fluoride, and 10% ammonium sulfate, adsorbed to a 100-ml phenyl-Sepharose column, and eluted in 2 mM KPi, pH 7.5, 10 M Zn 2ϩ , 10 mM ␤-ME, 0.1 mM phenylmethylsulfonyl fluoride following an 800-ml linear gradient to the final buffer. DmPBGS was pooled and applied directly to a 100-ml DEAE Biogel A column that was equilibrated in 2 mM KPi, pH 7.5, 10 M Zn 2ϩ , 10 mM ␤-ME, 0.1 mM phenylmethylsulfonyl fluoride. The protein was eluted with a 1-liter gradient to 0.4 M KCl in 30 mM KPi, pH 7.5, 10 M Zn 2ϩ , 10 mM ␤-ME, 0.1 mM phenylmethylsulfonyl fluoride. DmPBGS elutes in the first third of the gradient and is baseline separated from chromosomally encoded E. coli PBGS activity that elutes in the second half of the gradient. DmPBGS was pooled, concentrated to ϳ10 mg/ml, and further purified on a 1-meter long 270-ml Sephacryl S300 column in 0.1 M KPi, pH 7.0, 10 M Zn 2ϩ , 10 mM ␤-ME. Dry weight analysis was used to determine the extinction coefficient of DmPBGS as previously described (10).
Building a Model of DmPBGS-Two models were prepared to approximate the structure of octameric DmPBGS. These were based on the crystal structures of yeast PBGS with aminolevulinic acid bound (PDB accession number 1H7P, 1.67 Å resolution; Ref. 11) and human PBGS with the product porphobilinogen bound (PDB accession number 1E51, 2.83 Å resolution). The unit cell of the former is a monomer making a symmetric octamer, whereas the latter unit cell is an asymmetric dimer, thus making an asymmetric octamer; the models were originally built as monomer and dimer, respectively. Sequence alignments were performed with the BLAST program (12) with some manual adjustment upon examination of the sequences in light of the structures. All insertions and deletions in the alignment occurred in coil regions located between regular secondary structure segments. For both models, loop modeling in insertion and deletion regions in the alignment was performed with the Modeler program (version 6) (13), and side chain conformations were predicted with the program SCWRL (version 2.95) (14 -16). Side chains for residues that were identical in the DmPBGS sequence and the template proteins were kept fixed in their crystallographic Cartesian coordinates. The octamers were built using the symmetry transformations provided in the PDB entries with in-house custom software. The side chain calculations were repeated following formation of the octamer from the fundamental asymmetric units. Again, conserved amino acids were kept in their original positions.
PBGS Activity Assays-Determinations of PBGS activity used a fixed-time assay at 37°C in 0.1 M bis-tris propane-HCl, 10 mM ␤-ME, and the substrate 5-aminolevulinic acid (ALA), with variations in assay pH, concentrations of metal ions such as Zn 2ϩ , Mg 2ϩ , K ϩ , and the Zn 2ϩ chelator, 1,10-phenanthroline (o-phe). Procedures for pH versus activity profiles, K m and V max determinations, and inhibition by 4,7-dioxosebacic acid and 4-oxosebacic acid were as previously described (4,7). DmPBGS (10 g/ml) activity was determined as a function of pH using a fixed assay time of 5 min, a fixed ALA concentration of 10 mM, with and without 10 M Zn 2ϩ , 1 mM Mg 2ϩ , or 0.1 M K ϩ . Reported pH values are final assay pH following addition of ALA-HCl. The o-phe inhibition studies were carried out such that each assay contained the appropriate amount of o-phe in 10 l of ethanol (1% final volume).
Equilibrium Dialysis Studies-Equilibrium dialysis was carried out at 37°for 4 h under gentle agitation in an air shaker. Protein solutions were ϳ1 ml at ϳ1 mg ml Ϫ1 , and the dialysis buffer volume was 200 ml. Dialysis was carried out in 0.1 M bis-tris propane-HCl, 10 mM ␤-ME at pH 8 in the presence of 0.01-10 mM o-phe (Ϯ1 mM ALA) and in the presence of 0 -30 M Zn 2ϩ (Ϯ1 mM ALA). Enzyme-bound zinc was determined by atomic absorption spectroscopy as previously described (9).

Protein Expression and Purification-The expression of
DmPBGS from pJMPBGS was attempted in E. coli strains BLR(DE3) and BL21-CodonPlus-RIL. Based on SDS gels following a small scale growth, the CodonPlus host did not appear to greatly influence expression. However, because the native gene encoding DmPBGS contains clusters of rare codons, the CodonPlus strain was selected for expression in order to minimize the possibility of translational errors. From each 1-liter growth we obtained ϳ10 g of cells, from which ϳ45 mg of DmPBGS was purified to homogeneity. The final purification buffer was 0.1 M potassium phosphate, pH 7, containing 10 mM ␤-ME, and 10 M Zn 2ϩ , which was originally selected because it corresponds to optimal assay conditions for mammalian PBGS. Under these conditions, purified DmPBGS has a specific activity of ϳ2.5 mol h Ϫ1 mg Ϫ1 , which is an order of magnitude lower than human PBGS. Under optimal conditions for DmPBGS (see below), its specific activity is ϳ17 mol h Ϫ1 mg Ϫ1 . Freshly purified DmPBGS contains 1.7 ϩ 0.2 zinc/subunit as determined by atomic absorption spectroscopy. Based on a dry weight analysis, DmPBGS has a 1% A 280 ϭ 0.79.
Kinetic and Metal Binding Properties of DmPBGS-DmPBGS activity varies as a function of pH as illustrated in Fig. 1 and quantified in Table I. The maximal activity is observed with no added metals and peaks at ϳpH 8. Under these conditions, the pH activity profile shows a good fit to a simple bell curve with pK a ϭ 7.2, pK b ϭ 9.1 and a maximal velocity of 21 mol h Ϫ1 mg Ϫ1 . The addition of 0.1 M K ϩ or 1 mM Mg 2ϩ , which are required for, or stimulate, the activity of most other PBGS, has little effect on the pK a and pK b values, though Mg 2ϩ appears to afford some inhibition based on V max (see Table I). The addition of 10 M Zn 2ϩ , however, dramatically alters the profile to reveal two pK a values of 6.1 and 10.0, both of which fit best to a two-proton model. The net effect of 10 M Zn 2ϩ at pH values below the optimal pH of 8 is a dramatic inhibition. For comparison, at its optimal pH of ϳ7, human PBGS is not inhibited by Zn 2ϩ at concentrations below ϳ30 M (9, 17). The pH activity profile of human PBGS at 10 M Zn 2ϩ is included in Fig. 1 and in Table I.
The K m for ALA was determined for DmPBGS at pH 8 with no added metal ions and found to be 108 ϩ 12 M, which is in the range of all other PBGS at their conditions of optimal pH and metal ions (e.g. 4). The V max was found to be 16. mol h Ϫ1 mg Ϫ1 . The kinetic parameters K m and V max were not quantified under other conditions of pH or buffer metal ion concentrations. Fig. 2 quantifies the effect of adding Zn 2ϩ on the activity of DmPBGS at pH 8 and at 10 mM ALA. The data show inhibition to about 10% of maximal activity, which fit well to Equation 1.
The resulting IC 50 ϭ 4.24 ϩ 0.49 M Zn 2ϩ , and the Hill coefficient is 1.51 ϩ 0.23. It is interesting to note that the best fit to V min is 1.89 ϩ 0.61 mol h Ϫ1 mg Ϫ1 , in excellent agreement with the plateau rate observed between pH 6.5 and 7.5 at 10 M Zn 2ϩ (see Fig. 1 and V max 1 in Table I). To quantitatively correlate the inhibitory effect of zinc on enzyme activity with the zinc content of the protein, equilibrium dialysis studies were carried out under assay conditions in the presence and absence of 1 mM ALA in the dialysis buffer. These data are shown in Fig. 3A and quantified in Table II. The data fit well to the following simple binding in Equation 2, where Zn TB is total bound zinc, Zn cat is the stoichiometry of a required zinc, Zn Inh is the stoichiometry of an inhibitory zinc, [Zn 2ϩ ] is the free Zn 2ϩ concentration, and K dZnInh is the dissociation constant for the inhibitory zinc. Correlation of Fig. 3 with Fig. 2 indicates that under conditions where zinc does not inhibit, there are ϳ0.5-0.7 zinc/subunit, which is consistent with half-of-the-sites metal binding. Under conditions where zinc shows significant inhibition of DmPBGS, the protein contains ϳ1.5 zinc/subunit, suggesting that the inhibitory zinc has a stoichiometry of 1/subunit. The K dZnInh value for binding the inhibitory zinc in the presence of substrate, ϳ7 M Zn 2ϩ , is in good agreement with the kinetically determined IC 50 of ϳ4 M Zn 2ϩ (see Fig. 2).
To demonstrate that DmPBGS requires at least some of the tightly bound zinc that is found to co-purify with the protein, DmPBGS activity was assessed as a function of the zinc chelator, o-phe, as illustrated in Fig. 2. Maximal activity is observed at o-phe concentrations below 0.1 mM, and the inhibition profile fits best to a cooperative model where IC 50 ϭ 0.84 ϩ 0.03 mM o-phe with a Hill coefficient of 1.9 ϩ 0.1. To determine enzyme-bound Zn 2ϩ as a function of o-phe equilibrium dialysis, experiments were carried out in the presence and absence of 1 mM ALA as illustrated in Fig. 3A and quantified in Table II. At low o-phe the maximal catalytic activity seen in Fig. 2 is again associated in Fig. 3A with Zn 2ϩ bound at a stoichiometry of 0.5/subunit, confirming half-of-the-sites binding of the catalytic Zn 2ϩ . In this case, the binding data fit best to a cooperative model for o-phe inhibition (Table II) The data presented in Figs. 2 and 3 show that DmPBGS binds Zn 2ϩ differently from either human or yeast PBGS. Like the ZnB of human PBGS (4), the catalytic Zn 2ϩ of DmPBGS shows half-of-the-sites metal binding. However, Fig. 3A suggests that the inhibitory Zn 2ϩ of DmPBGS does not show the half-of-the-sites binding that has been seen for the non-essential ZnA of human PBGS (4). In support of the notion that the inhibitory zinc of DmPBGS is somehow related to ZnA of yeast and mammalian PBGS, the K dZnInh of DmPBGS is of comparable magnitude to the K d for ZnA of mammalian PBGS (ϳ5 M), which binds at 0.5 zinc/subunit (4,9,17).
The DmPBGS Mutant H10F-Cys-219 of DmPBGS is analogous to the cysteine that has been shown to bind the nonessential (but also non-inhibitory) ZnA of human (4) and yeast (6) PBGS. In our model of DmPBGS, shown in Fig. 4A, Cys-219 lies very near to His-10 of a neighboring subunit, such that Cys-219 and His-10 could form part of a previously uncharacterized zinc binding site. Various PBGS crystal structures suggest that the position of Cys-219 is dependent on whether the active site lid is opened or closed. Hence, zinc binding to this site in the closed lid conformation shown in Fig. 4A could inhibit lid opening and thus inhibit catalysis. To test the hypothesis that His-10 is involved in binding the inhibitory zinc, this residue was altered to Phe, which is found in the analogous position of human PBGS. The behavior of DmPBGS mutant H10F during purification was indistinguishable from wild type DmPBGS. The yield was 75 mg from 17 g of cell paste. The specific activity of H10F at pH 8 is seen to be ϳ12 mol h Ϫ1 mg Ϫ1 , only marginally lower than wild type, regardless of a The fitted equation is velocity ϭ V max /(1 ϩ 10 (pKa Ϫ pH) ϩ 10 (pH Ϫ pKb )). b The fitted equation is velocity ϭ (V max 1 /(1 ϩ 10 (2 ϫ (pKa1 Ϫ pH)) )) ϩ (V max 2 /(1 ϩ 10 (2 ϫ (pKa2 Ϫ pH)) )). Reported pK values are pK a 1 and pK a 2 . c The fitted equation is velocity ϭ V max /(1 ϩ 10 (2 ϫ (pKa Ϫ pH)) ϩ 10 (pH Ϫ pKb) ).

FIG. 2. The inhibition of DmPBGS by Zn 2؉ and o-phenanthro-
line. The activity of DmPBGS at pH 8.0 is illustrated as a function of added Zn 2ϩ (E) and as a function of o-phenanthroline (Ⅺ). Both data fits are cooperative as described under "Results."

Structural Basis for Half-of-the-sites Metal Binding to PBGS
whether or not 10 M Zn 2ϩ was added to the assay mixture. The pH rate profile of H10F does not exhibit the zinc inhibition phenomenon but is otherwise the same as the wild type protein (see Fig. 1). The fitted pH rate profile data is detailed in Table  I; however, for clarity of presentation only the combined fit line is illustrated in Fig. 1 (R 2 ϭ 0.96). The behavior of H10F supports the hypothesis that inhibition by zinc is because of zinc binding through His-10 of DmPBGS.
DmPBGS mutant H10F provides an independent tool for determining the stoichiometry of the inhibitory zinc of DmPBGS. To do so, the H10F protein was evaluated for its ability to bind zinc in an equilibrium dialysis experiment, the results of which are illustrated in Fig. 3B. In this case the looser binding zinc of H10F retains the K d of ϳ5 M but binds at a reduced stoichiometry of ϳ0.5/subunit. Thus, H10F has lost the inhibitory zinc, which appears to have bound at a stoichiometry of 0.5/subunit. The conclusion is that DMPBGS contains the catalytic Zn 2ϩ (akin to ZnB) at 4/octamer, contains the non-essential Zn 2ϩ (akin to ZnA) at 4/octamer, and contains an inhibitory Zn 2ϩ at 4/octamer. Because Cys-219 is implicated in binding both ZnA and the inhibitory zinc, the binding of these two metal ions must be mutually exclusive (see "Discussion").
Sensitivity of DmPBGS to the Species-selective Inhibitors 4,7-DOSA and 4-OSA-The differences in identity and stoichiometry of active site and allosteric metal ions for the different species of PBGS have been found to correlate with a differential susceptibility to various active site-directed inhibitors (7,8). This is most notable for the inhibitor 4,7-dioxosebacic acid (4,7-DOSA), which we have previously characterized with PBGS from human, E. coli, P. aeruginosa, Pisum sativum (pea), and B. japonicum, each of which contains a unique constellation of active site and allosteric metal ions. We find that a 100-min preincubation of DmPBGS with 4,7-DOSA over the concentration range 0.1 M-1 mM reveals an IC 50 of 9.7 Ϯ 0.5 M 4,7-DOSA, with a Hill coefficient of 1.5. This fit is significantly better than the fit to a simple hyperbolic equation, which gives IC 50 of 9.1 Ϯ 0.9 M 4,7-DOSA. The DmPBGS mutant H10F exhibits a comparable IC 50 of 9.6 Ϯ 0.5 but fits quite well to a non-cooperative model. Fig. 5A puts 4,7-DOSA inhibition of DmPBGS in the context of prior studies with other species of PBGS (7). The prior work suggested that sensitivity of PBGS to 4,7-DOSA is significantly enhanced by the active site zinc, and results with DmPBGS are consistent with that observation.
The species-dependent inhibition pattern observed for 4-OSA was significantly different from that of 4,7-DOSA (8). 4-OSA was found to be an effective inhibitor only of E. coli PBGS, and this inhibition was significantly less potent than that seen with 4,7-DOSA. 4-OSA was found to be virtually ineffective at inhibiting human, P. aeruginosa, P. sativum, or B. japonicum PBGS, even at an inhibitor concentration of 3 mM and an inhibitor-enzyme preincubation time of 24 h. In contrast, a 16-h incubation of DmPBGS with 3 mM 4-OSA showed only 60% retention of activity relative to a control incubation without 4-OSA. Fig. 5B shows a more quantitative comparison of 4-OSA inhibition of DmPBGS relative to E. coli PBGS. As described previously, the E. coli PBGS gives a good fit to the Copeland approximation (18), revealing an IC 50 ϭ 0.22 ϩ 0.01 mM 4-OSA. In comparison, the inhibition data for DmPBGS fits less well but gives a reasonable approximation for the IC 50 of 2.7 ϩ 0.5 mM 4-OSA, which is significantly more sensitive than for any of the others that have been tested, except for E. coli PBGS.
Characteristics of the Models of DmPBGS-The two models for the homodimer of homo-octameric DmPBGS, which are based on either yeast or human PBGS, differ predominantly in the surface residue configurations. Fig. 4A illustrates a DmPBGS model in stereoview based on the higher resolution yeast PBGS structure, PDB accession number 1H7P. This and other PBGS structures reveal that the enzyme active site components are in the center of the ␣ 8 ␤ 8 barrel that is comprised of the C-terminal 300 amino acids of each peptide. The N-terminal ϳ30 amino acids form an arm structure that hugs the neighboring barrel and interacts with other subunits of the octamer. The N-terminal arm is the most variable part of the PBGS protein and consequently is the region where the two models differ most. The illustrated model includes the three cysteine ligands to the active site Zn 2ϩ (ZnB), the histidine previously seen to bind ZnA, as well as the proposed inhibitory Zn 2ϩ ligands His-10 and Cys-219.

DISCUSSION
DmPBGS is the first insect PBGS for which a sequence became available and the first to be purified and characterized. The DmPBGS sequence shows the binding determinants for two zinc ions that have been previously characterized with both mammalian and yeast PBGS (2). Like mammalian PBGS, but unlike yeast PBGS, DmPBGS shows half-of-the-sites binding for the catalytically essential zinc at 0.5/subunit (4/octamer). Catalytic zinc binding to DmPBGS is presumed to be to the  Table II.

Structural Basis for Half-of-the-sites Metal Binding to PBGS
cysteine residues of the sequence DVCICPYSSHGHCG at the active site. Half-of-the-sites binding to this site is seen in the crystal structure of human PBGS (PDB accession number 1E51). Yeast PBGS structures show the catalytically essential zinc bound to the corresponding three-cysteine site, but there is no evidence for half-of-the-sites binding (6). Mutagenesis studies on human PBGS complement the crystal structures in as-serting the essential nature of the zinc ion bound to this (ZnB) site (4). Both human and yeast PBGS bind a second zinc ion, which for human PBGS is at a stoichiometry of 0.5/subunit (4). Chemical modification studies on bovine PBGS implicate Cys-223 of human PBGS in binding this second (ZnA) zinc ion (19). As noted above, the analogous residue is Cys-219 of DmPBGS. The ZnA binding site is confirmed by a crystal structure of yeast PBGS that shows weak binding of zinc to the analogous cysteine with a second ligation through the italicized His residue of the cysteine-rich catalytic zinc-binding sequence de-  The lines represent a non-linear best fit to the Copeland approximation as previously described (18). The species depicted in black contain the active site zinc; the species depicted in gray do not. Each of these species has a unique stoichiometry of active site and allosteric metal ions (3,4,5,10). B, inhibition of DmPBGS () and E. coli PBGS (f) by 4-OSA depicting a 16-h preincubation with the inhibitor. scribed above. A mutation to these ligands in human PBGS results in an active protein, which binds only the catalytic zinc at 0.5/subunit and which has a somewhat mildly elevated K m for ALA (4).
The data presented here suggest that for DmPBGS zinc also binds to a site that results in an incomplete inhibition of activity. The possibility that this inhibitory site is spatially equivalent to the previously described ZnA site was considered. Like other TIM-barrel enzymes, PBGS contains an active site lid, which is a continuous mobile stretch of amino acids that serves to gate access to the active site. Based on other PBGS structures, the DmPBGS lid is comprised of amino acids 205-223, as illustrated in Fig. 4A, and the putative ZnA ligand Cys-219 lies within this lid. As a part of the mobile lid, this cysteine has been seen in more than one location in various PBGS crystal structures. In yeast PBGS structure 1AW5, where much of the lid is disordered, the analogous cysteine is seen pointing in toward the active site and is ligated to an atom of ZnA, at least part of the time, as illustrated in Fig. 4B. The second ZnA ligand is seen in structure 1AW5 as a histidine, analogous to His-129 of DmPBGS, which is italicized within the catalytic zinc binding sequence above and is included in Fig. 4A. The ZnA seen in structure 1AW5 resides at the periphery of the active site, about 7 Å from ZnB and about 9 Å from the nearest active site catalytic lysine. Crystal structures of yeast PBGS containing heavy metals such as Hg and Pb (PDB accession numbers 1QML and 1QMV; Ref. 20) also show the cysteine pointed in toward the barrel near this histidine, and in these structures much of the active site lid is also disordered. On the other hand, there are many other yeast PBGS structures (e.g. PDB accession number 1H7P) where the lid is closed and the lid cysteine is pointing away from the active site, as illustrated in the DmPBGS model in Fig. 4A. In these cases, the cysteine in question is in close contact with the N-terminal arm of an adjacent subunit. In DmPBGS, (Fig. 4A), the nearest neighbor is His-10, potentially forming an inhibitory zinc site wherein zinc binding to this site would stabilize the closed lid conformation, thus inhibiting active site lid opening for product release. To test this hypothesis, we prepared the H10F variant of DmPBGS, choosing to alter the putative Zn 2ϩ ligand His-10 to the cognate found in human PBGS, which is Phe-12. The behavior of H10F, which is very similar to DmPBGS but is not inhibited by zinc, supports this hypothesis. Hence the characterization of DmPBGS and H10F reveals the location of a previously unknown inhibitory zinc binding site. Although tetrapyrrole biosynthesis is generally tightly regulated, the physiologic significance of the regulation of DmPBGS activity by zinc remains unclear.
As would be expected, removal of zinc inhibition is found to correlate with a loss of binding of the inhibitory zinc. The interesting result is the stoichiometry of that inhibitory zinc; H10F binds 0.5 zinc/subunit less than the wild type DmPBGS, which suggests that binding through Cys-219 and His-10 is at a stoichiometry of 0.5/subunit. In addition to the catalytic zinc, H10F retains the binding of 0.5 zinc/subunit, which does not inhibit activity and which has a comparable dissociation constant of ϳ5 M. Thus, zinc binding to DmPBGS mutant H10F is akin to the binding of zinc to human PBGS, which binds the catalytic ZnB at 4/octamer and the nonessential ZnA at 4/octamer (4). DmPBGS appears to bind these two zinc ions at a half-of-the-sites stoichiometry and also binds an inhibitory zinc at a half-of-the-sites stoichiometry. One attractive model is that both the inhibitory zinc and ZnA are bound through Cys-219 but that the spatial location of this zinc is a function of whether or not the active site lid is opened or closed. When the lid is closed, the ligands are Cys-219 and His-10, as in Fig. 4A; when the lid is open, the ligands are Cys-219 and His-129, analogous to the yeast PBGS conformation illustrated in Fig.  4B. The fact that both can be bound at the same time dictates that for the PBGS octamer, half of the lids must be open and half of the lids must be closed, as is seen in the asymmetric crystal structures of P. aeruginosa PBGS (PDB accession number 1B4K) and human PBGS (PDB accession number 1E51).
The stoichiometry of zinc binding to DmPBGS is in keeping with a reciprocating motion model for the function of oligomeric PBGS. When the active site lid is closed, the inhibitory zinc is bound to the same subunit as the catalytic zinc. The other subunit of the asymmetric dimer would contain zinc bound at the ZnA site. With every catalytic turnover the lid would open and the metal ions would move to the positions seen in the adjacent monomer. The reciprocating motion model provides a structural basis for half-of-the-sites metal binding and half-ofthe-sites reactivity for homo-octameric PBGS. Multiple studies show that mammalian PBGS is fully active when zinc is present at a total stoichiometry of 0.5/subunit (9,17,21). When this is put in the context of the reciprocating motion model, it dictates that with every turnover the zinc ion must dissociate from one subunit and reassociate to the adjacent subunit or move between subunits along the protein surface. Because turnover for PBGS is generally quite slow, on the order of 1/s, this is entirely feasible. The outstanding question remains as to how events at the active site of one subunit are communicated to the other subunit in the dimer pair.