The Molecular Mechanism of Lead Inhibition of Human Porphobilinogen Synthase*

Human porphobilinogen synthase (PBGS) is a main target in lead poisoning. Human PBGS purifies with eight Zn(II) per homo-octamer; four ZnA have predominantly nonsulfur ligands, and four ZnB have predominantly sulfur ligands. Only four Zn(II) are required for activity. To better elucidate the roles of Zn(II) and Pb(II), we produced human PBGS mutants that are designed to lack either the ZnA or ZnB sites. These proteins, MinusZnA (H131A, C223A) and MinusZnB (C122A, C124A, C132A), each become purified with four Zn(II) per octamer, thus confirming an asymmetry in the human PBGS structure. MinusZnA is fully active, whereas MinusZnB is far less active, verifying an important catalytic role for ZnB and the removed cysteine residues. Kinetic properties of the mutants and wild type proteins are described. Comparison of Pb(II) inhibition of the mutants shows that ligands to both ZnA and ZnB interact with Pb(II). The ZnB ligands preferentially interact with Pb(II). At least one ZnA ligand is responsible for the slow tight binding behavior of Pb(II). The data support a novel model where a high affinity lead site is a hybrid of the ZnA and ZnB sites. We propose that the lone electron pair of Pb(II) precludes Pb(II) to function in PBGS catalysis. mutants here called MinusZnA and MinusZnB. The former is designed to bind only four ZnB, and the latter is designed to bind only four ZnA. We present a kinetic characterization of MinusZnA in comparison to the wild type human PBGS proteins K59, N59, K59/C162A, and N59/C162A, the latter of which is the parent of MinusZnA and MinusZnB. The C162A mutation was introduced to avoid a non-native disulfide and is shown to be benign. Less comprehensive data are presented for the relatively inactive MinusZnB. This work shows that Pb(II) can interact with human PBGS in at least two different ways, consistent with the interpretation of the yeast PBGS crystal structure complexed with Pb(II). However, interaction of Pb(II) with a ZnA ligand could not have been predicted from the crystal structure.

Porphobilinogen synthase (PBGS, 1 EC 4.2.1.24, also known as 5-aminolevulinic acid dehydratase), an enzyme that functions in the first common step in tetrapyrrole biosynthesis (e.g. heme and chlorophyll), is a highly conserved protein throughout evolution but has significant phylogenetic variation in the number and types of metal ions that function in catalysis or at allosteric sites (1). The human PBGS protein and its metal binding properties are of particular interest, because human PBGS is a primary target for the environmental toxin lead (2). Past studies using yeast PBGS as a model for human lead poisoning are limited because of a phylogenetic difference in the metal binding stoichiometries of yeast and human PBGS (3,4). There are two codominant alleles, ALAD1 and ALAD2, which encode the human PBGS variants at position 59, K59 and N59, respectively. Recent data on the isolated K59 and N59 proteins do not support the epidemiological data that has correlated these alleles with a differential susceptibility to lead poisoning (3,5).
All PBGS proteins appear to be homo-octamers, and there is evidence for and against half-sites reactivity, including cases where any one type of metal is bound at four per octamer (e.g. Refs. 6 -10). Mammalian PBGS enzymes purify with Zn(II) bound at a stoichiometry of eight per octamer (6,7,11). Only four Zn(II) appear to be tightly bound and only four Zn(II) appear to be required for full catalytic activity of human PBGS (3), consistent with data on bovine PBGS (6,7,11,12). Preliminary data obtained on human PBGS show that Pb(II) can displace about half the enzyme-bound Zn(II) (3). There are two outstanding questions in the structure and mechanism of human PBGS: 1) Which of the two types of Zn(II) corresponds to the four Zn(II) that are essential for the activity of mammalian PBGS? and 2) What extent do the two Zn(II) sites play in the Pb(II) inhibition of enzyme activity?
The crystal structure of yeast PBGS (Protein Data Bank code law5) revealed Zn(II) at two different sites with ligands that are consistent with prior chemical modification data (12,13). In the crystal structure, the two Zn(II) sites were not fully occupied (14). The most highly populated Zn(II) site shows the metal bound to a cluster of cysteine residues, which in the human protein correspond to Cys-122, Cys-124, and Cys-132. The Zn(II) that binds to this cysteine-rich site is defined as ZnB (8). Although ZnB is 8 Å from an active site lysine, its coordination environment more closely resembles structural rather than catalytic Zn(II). A lesser-populated Zn(II) site of yeast PBGS was only seen in a difference map and shows Zn(II) with an incomplete coordination shell consisting of one histidine and one cysteine, which in the human protein correspond to His-131 and Cys-223. The Zn(II) that binds to this site is known as ZnA (8). Although these ligands are typical of catalytic Zn(II), the low occupancy and 12-Å distance from a Schiff base-forming lysine indicate ZnA to be of questionable significance in the PBGS reaction mechanism (14).
Because yeast PBGS can bind 16 Zn(II) per octamer (4), which is twice as much as human PBGS (3), a homology model for human PBGS must accommodate this difference in stoichiometry. Our current model contains ZnA and ZnB bound to each of four subunits, whereas the other four subunits contain no metal ions (3). A picture of the Zn(II) sites of the human PBGS model is presented in Fig. 1, where adjacent amino acids are seen as ligands to ZnA (His-131) and ZnB (Cys-132). We propose that this is the basis for communication between the two metal sites.
To test this model, the current work addresses the stoichiometry and function of ZnA and ZnB by use of mutagenesis to remove (or incapacitate) each of the two sites to create the mutants here called MinusZnA and MinusZnB. The former is designed to bind only four ZnB, and the latter is designed to bind only four ZnA. We present a kinetic characterization of MinusZnA in comparison to the wild type human PBGS proteins K59, N59, K59/C162A, and N59/C162A, the latter of which is the parent of MinusZnA and MinusZnB. The C162A mutation was introduced to avoid a non-native disulfide and is shown to be benign. Less comprehensive data are presented for the relatively inactive MinusZnB. This work shows that Pb(II) can interact with human PBGS in at least two different ways, consistent with the interpretation of the yeast PBGS crystal structure complexed with Pb(II). However, interaction of Pb(II) with a ZnA ligand could not have been predicted from the crystal structure.

EXPERIMENTAL PROCEDURES
Materials-Most chemicals and buffers were obtained from Fisher Scientific or Sigma in the purest available form. 2-Mercaptoethanol (␤ME) was from Fluka and vacuum-distilled prior to use. High purity KOH was from Aldrich. Concentration devices, originally sold under the Amicon label, were obtained from Fisher as were Slide-A-Lyzer dialysis cassettes, originally sold under the Pierce label. Atomic absorption standards were SpecPure grade and obtained from Alfa AESAR.
Protein Expression and Purification-Human PBGS and mutants thereof were expressed and purified from an artificial gene as described previously (3). The mutations were achieved using the QuikChange technology of Stratagene. The sense strand primers were GCGTATG-CTAAGGCAGGTGCACAGGTGGTAGCCCCTTCC (for C162A), CCTG-TGACGTCGCCCTGGCTCCGTACACTTCTCACGGTCACGCCGGTC-TCC (for C122A, C124A, C132A), CGTACACTTCTCACGGTGCCTGC-GGTCTCCTGAGC (for H131A), and GGCGACCGCCGCGCCTATCAG-CTGCC (for C223A). All resultant plasmids were sequenced throughout the artificial gene in both directions. For protein production, 6-liter growths of BLR(DE3)pMVhum (3) (and mutants listed below) were started from single colonies of a fresh transformation at one colony per liter. The initial growth media was Luria Broth plus 0.4% glucose containing 12.5 g ml Ϫ1 tetracycline and 100 g ml Ϫ1 ampicillin. After 16 h at 37°C with shaking, the cells were harvested and resuspended in fresh broth without glucose to which tetracycline, ampicillin, and 20 M Zn(II) had been added. Human PBGS induction was carried out at 15°C with 100 M isopropyl-1-thio-␤-D-galactopyranoside and expression proceeded for an additional 45 h. The protein was purified from the soluble fraction of the lysed cells using a 25-45% ammonium sulfate fractionation, phenyl-Sepharose, DEAE-BioGel, and Sephacryl S300 columns as described previously (3). Care was taken to thoroughly regenerate all columns between preparations to prevent cross contamination of the proteins. Purifications of active and inactive human PBGS variants were carried out using two different batches of phenyl-Sepharose resin. Care was also taken to prevent contamination of the inactive proteins by chromosomally encoded Escherichia coli PBGS, the presence of which was determined by its ability to respond to Mg(II) as an allosteric activator (15,16). The E. coli PBGS elutes from the DEAE column in the tail end of human PBGS peak (3). Protein was isolated in amounts varying from 150 to 400 mg for N59, N59/C162A, N59/C162A/H131A/ C223A (hereafter called MinusZnA), and N59/C162A/C122A/C124A/ C132A (hereafter called MinusZnB). The yield for K59 and K59/C162A was ϳ50 mg of purified enzyme from 6 liters of growth. Purified proteins were concentrated to Ͼ5 mg/ml, aliquoted into portions, flashfrozen in liquid N 2 , and stored at Ϫ80°C.
Analytical Methods-Enzyme assays (3), Zn(II) analysis by atomic absorption spectroscopy, and Zn(II) binding by equilibrium dialysis (3,12) were carried out according to procedures described previously. For MinusZnB activity, assays contained as much as 2 mg ml Ϫ1 protein and were allowed to proceed for between 2 and 40 h. All enzyme assays were carried out in metal-free plastic test tubes at 37°C. Continuous assays monitoring the formation of porphobilinogen at 236 nm (17) were carried out using a Cary 50 spectrometer fitted with an epoxy dip probe. Matrix-assisted laser desorption ionization time-of-flight (MALDI-TOF) mass spectroscopy was performed on a PerSeptive Biosystems Voyager DE mass spectrometer in linear mode at an accelerating voltage of 20,000 V.
The Stoichiometry of Zn(II) Required for Optimal Activity-To determine the stoichiometry of Zn(II) required for maximal activity, K59, N59, N59/C162A, and MinusZnA were dialyzed against 50 mM sodium acetate, pH 5.0, 10 mM ␤ME overnight to prepare the Zn(II)-free apoenzyme (3). Following a brief (2 h) dialysis back into neutral pH buffer (0.1 M TES-KOH, 10 mM ␤ME, pH 7.0), the proteins were analyzed for Zn(II) content by atomic absorption spectroscopy and assayed at 2 M total subunit concentration in the presence of 0, 0.1, 0.2, 0.3, 0.4, and 0.5 M Zn(II) added to the assay. For all data points, A 555 values were Ն0.10 and thus of high confidence.
pH Activity Relationships-The pH rate profiles were determined using buffers containing 0.1 M bis-tris propane-HCl (initial pH varied from 6 to 9), 10 mM ␤ME, 10 M Zn(II), and a fixed amount of recombinant human PBGS (K59 at 4.3 g/ml, N59 at 10 g/ml, N59/C162A at 15 g/ml, K59/C162A at 15 g/ml, MinusZnA at 15 g/ml, and Mi-nusZnB at 1.0 mg/ml). In all cases the enzyme-buffer mixture was preincubated for 10 min at 37°C prior to the addition of ALA-HCl to a final concentration of 10 mM. The final pH was measured in mock assays at room temperature. The active proteins were assayed using a 5-min fixed time assay; MinusZnB was assayed for 5 h. For determination of K m and V max values, the ALA-HCl concentrations were 10 M, 30 M, 100 M, 300 M, 1 mM, 3 mM, and 10 mM with stock dilutions made in 0.1 M HCl to keep the assay pH and ionic strength constant. All assays were terminated with a one-half volume of STOP reagent (20% trichloroacetic acid, 0.1 M HgCl 2 ). K m and V max determinations for MinusZnB used 0.5 mg ml Ϫ1 and 18-h assays. Because the catalytic activity of MinusZnB was reduced by five orders of magnitude, K m and V max determinations were carried out in the presence and absence of Mg(II) to control for activity due to trace contamination by E. coli PBGS. Mg(II) causes a pH-dependent activation of the E. coli protein (16), but not the human protein. Porphobilinogen formed was determined by absorbance at 555 nm about 8 min after the addition of a one and one-half volume of modified Ehrlich's reagent. The extinction coefficient of the pink complex formed (⑀ 555 ) was 62,000 cm Ϫ1 M Ϫ1 .
Pb(II) Inhibition Studies-Fixed-time lead inhibition assays included the holoenzymes as purified and were carried out with no added metals, with 20 M Pb(II), and with both 10 M Zn(II) and 20 M Pb(II) added to the assay preincubation mixture. Because of the ability of phosphate to buffer metal ions, these assays were carried out in TES-KOH at an initial pH of 7.0. Free Pb(II) is Ͻ20 M due to complexation by ␤ME. After the addition of ALA to 10 mM, the active proteins were assayed for 5 min. For MinusZnB, 2 mg ml Ϫ1 of protein was used and the assays were monitored from 1 to 6 h, removing aliquots at 0.5-h intervals for porphobilinogen determination. Continuous assays of the active enzymes were allowed to proceed for 1 h with A 236 readings taken at 20 per minute. The rate of approach to equilibrium was calculated from ⌬A 236 between the actual reading and the steady-state rate extrapolated back to the point of substrate addition. The ⌬A 236 data were fitted to a simple exponential decay model and results are reported as t1 ⁄2 values.

RESULTS
Protein Expression and Purification-Previous work showed that human PBGS corresponding to the N59 allele partitioned into the soluble fraction of E. coli 3-fold better than K59 (3). The C162A mutation was prepared with the intention of removing an internal disulfide bond with Cys-119. This disulfide was apparent in a model of human PBGS based on the yeast PBGS crystal structure (3) and is seen in preliminary crystal- lographic data on bovine and human PBGS. 2 The resulting protein, which is named N59/C162A, partitioned almost completely into the soluble fraction of the E. coli host cell and exhibited purification properties as described for Asn-59 (3). The mutant proteins MinusZnA and MinusZnB also expressed very well, partitioned into the soluble fraction of the host cell, and purified like the wild type proteins. K59/C162A was found to partition like K59 where the predominant fraction was in inclusion bodies (3). As before, all proteins were purified from the soluble fraction. The final purified pools from the S-300 column were analyzed for Zn(II) content by atomic absorption spectroscopy. Wild type variants N59, K59, N59/C162A, and K59/C162A contained ϳ8 Zn per octamer, MinusZnA purified with 3.9 Ϯ 0.6 Zn/octamer, and MinusZnB purified with 4.0 Ϯ 0.5 Zn/octamer. DEAE fractions of MinusZnB that contained trace activity that can be activated by Mg(II) indicated E. coli PBGS contamination, and this MinusZnB was further purified by a second pass through the DEAE column, and once again, the activity activated by Mg(II) eluted solely in the tail end of the protein peak. Because high level heterologous expression in E. coli can lead to incomplete N-terminal processing, the purified protein was digested with trypsin and subjected to MALDI-TOF mass spectral analysis. No evidence was found for aberrant N-terminal processing. The N-terminal methionine was present with complete removal of the formyl group (data not shown).
Catalytic Activities of Human PBGS Variants-The N59/ C162A and K59/C162A variants of PBGS as well as MinusZnA are all enzymatically active with specific activities comparable with N59 and K59 (3) under standard assay conditions (0.1 M potassium phosphate, final pH 6.8, 10 M Zn(II), 10 mM ␤ME). The specific activities at pH 6.8 are constant from 1 to 30 M Zn(II). To maintain consistency with our prior published work on mammalian PBGS, 10 M Zn(II) was used for most kinetic studies. The specific activities of Asn-59, Lys-59, N59/C162A, K59/C162A, and Minus ZnA were determined at 10 mM ALA over the pH range of 6 -9 using the buffer bis-tris propane. All showed broad roughly bell-shaped curves with an apparent pH optimum between 6.8 and 7.3, similar to bovine and other mammalian PBGS (18 -20). The dependence of specific activity on pH was also determined for MinusZnB, which has a much reduced catalytic potency. In this case there was a nonlinear rise of specific activity with increasing pH from pH 6 to 9. The detailed analysis of MinusZnB is the subject of ongoing work.
K m and V max values were determined for the human PBGS proteins at pH 6, 7, and 8 in bis-tris propane; the results are presented in Table I. As previously reported in terms of specific activities for wild type forms K59 and N59 (3), K59/C162A is 1.5-to 2-fold more active than N59/C162A. Table I shows that there are no significant phenomenological differences between K59, N59, K59/C162A, and N59/C162A. Thus, N59/C162A is an appropriate parent for our mutants. The general trends described below have been seen for other PBGS (21). Table I illustrates that V max values for any one protein in the pH range from 6 to 8 vary by factors of 2 to 5, whereas K m values show a more significant variation with pH. At pH 6, the reciprocal relationship between [ALA] and enzyme activity is linear from 10 M to 10 mM and the K m values are fairly large, Ͼ1 mM. The high K m at below optimal pH may be related to a pH-dependent decrease in enzyme-bound Zn(II) (3), because Zn(II) has been shown to be essential for binding and reactivity of the ALA substrate that is not involved in Schiff base formation to Lys-252 (22). At pH 7, the reciprocal relationships were again linear over three orders of magnitude variation in [ALA] and the K m values, which are all on the order of 0.1 mM, resembled those previously published for all PBGS at their optimum pH when their divalent metal ion requirements are met. However, above the optimal pH, at pH 8, the reciprocal relationships deviate from linearity and reveal what might be interpreted as a substrate activation phenomenon. Fig. 2A, which shows the kinetic data at pH 6, 7, and 8 for N59, illustrates the magnitude of this phenomenon. The apparent K m at pH 8, considering only [ALA] below 300 M, was in the range of 15-50 M and is reported in Table I. However, above 300 M ALA all the active human PBGS showed pronounced downward curvature at pH 8 (for N59, triangles in Fig. 2A), and the apparent K m values were about twice that seen at pH 7. The human PBGS mutant MinusZnA shows a relatively unremarkable pH dependence for the K m and V max values with the exception of the 3-to 6-fold elevated K m for ALA at pH 6. This may indicate a role for ZnA, His-131, or Cys-223 in substrate binding. The comparable V max values of MinusZnA relative to wild type PBGS suggest that ZnA, His-131, and Cys-223 do not function in a rate-determining step of catalysis.
Determination of kinetic constants for MinusZnB, which exhibits very low activity, required 0.5 mg ml Ϫ1 enzyme and assay times of ϳ18 h. The observed activities were corrected by subtracting the Mg(II)-responsive activity derived from trace levels of copurified E. coli PBGS. At pH 7, MinusZnB exhibited a K m of 20.4 mM and a V max of 0.006 mol h Ϫ1 mg Ϫ1 giving a 10 6 -fold reduction in V max /K m relative to wild type. The K m value for MinusZnB was more than two orders of magnitude larger than wild type at neutral pH, suggesting that ZnB is the zinc ion that is required for binding the substrate that determines the K m . We have previously proposed that the catalytically essential Zn(II) coordinates the K m -determining ALA in a bidentate fashion (12,23). Also at neutral pH, the V max value for MinusZnB was nearly 5000-fold smaller than wild type, indicating that ZnB and its cysteine ligands play a significant role in catalysis. Nevertheless, it is intriguing that MinusZnB continues to be capable of catalyzing the formation of porphobilinogen from ALA, albeit at a dramatically reduced rate. Clearly the PBGS-catalyzed reaction is favorable enough to proceed slowly through some alternate mechanism that is in-2 H. L. Carrell, unpublished results. dependent of the catalytic Zn(II) and its cysteine ligands. Stoichiometry of Zn(II) Required for Optimal Activity-Bovine and human PBGS (K59) require at least four Zn(II) per octamer for full activity (3,6,7,11). The activity is not increased when eight Zn(II) per octamer are present. To determine the Zn(II) dependence for N59/C162A and Minus ZnA, substoichiometric amounts of Zn(II) were added to assays containing 2 M PBGS subunits, which first had been depleted in Zn(II) by pH 5.0 dialysis (Ͻ0.2 Zn(II)/octamer). Back-dialysis to pH 7 resulted in an enzyme with one to two Zn(II)/octamer. For this reason, the rates with no additional Zn(II) were in the range of 25-50% of the maximal activity as illustrated in Fig.  2B. In all three cases, near maximal activity was obtained after addition of approximately two Zn(II) per octamer. The slope of the line through these low [Zn(II)] points shows that four Zn(II) per octamer corresponds to 100% activity. The Zn(II) in the assay components was determined to be negligible. These data are presented in Fig. 2B alongside a complete Zn(II) titration for human PBGS N59 (3).
The Effect of Lead-Pb(II) inhibition of PBGS has historically been attributed to Pb(II) replacing Zn(II). The preparation of MinusZnA and MinusZnB variants allowed us to determine whether these Zn(II) sites are related to Pb(II) inhibition. We compared the ability of Pb(II) to inhibit the activities of wild type human PBGS variants, MinusZnA, and MinusZnB using a variety of different assays. Table II shows that, for the holoenzymes of K59, N59, K59/C162A, and N59/C162A (eight Zn(II)/ octamer), with no Zn(II) added to the assay, 20 M Pb(II) in the preincubation mixture caused significant inhibition in a 5-min assay (15-37% activity). This can be effectively counteracted by the addition of 10 M Zn to the preincubation mixture (80 -99% activity). MinusZnA was also sensitive to inhibition by Pb(II) (56% activity) but significantly less than the wild type variants as measured using the fixed time 5-min assay. This reduced inhibition for MinusZnA suggests the ligands to ZnA play a role in the inhibition of human PBGS by lead. Continuous assays, as described below, revealed the basis for the decreased sensitivity of MinusZnA to lead inhibition relative to the other active proteins.
Pb(II) inhibition of human PBGS cannot be thoroughly assessed with simple fixed-time assays, because continuous assays with Pb(II) show a nonlinear time dependence as seen before for bovine PBGS (24). This suggests that the way in which Pb(II) binds to PBGS varies depending on the presence or absence of substrate or product as seen for other slow binding inhibitors. The nonlinearity of human PBGS activity in the presence of lead is shown using continuous assays (carried out at 20 M and 50 M Pb(II)) following the accumulation of porphobilinogen by its absorbance at 236 nm. For the wild type human PBGS variants studied, K59, N59, and N59/C162A, the nonlinear time dependence was observed. The absorbance data for N59 is illustrated in Fig. 2C. The initial velocities, steadystate velocities, and half-times are reported in Table III; the velocity data are reported as percentage activity relative to a control assay without Pb(II). For each variant, with Pb(II) the apparent t1 ⁄2 for K59 Ͼ N59 Ͼ N59/C162A. MinusZnA did not show the time-dependent equilibration phenomenon, and the steady-state rates were established immediately. This strongly suggests that His-131 or Cys-223 plays a role in the Pb(II)⅐enzyme equilibrium in the absence of substrate and during the approach to steady state. However, when steady-state velocities are considered, MinusZnA is equally sensitive as the wild type proteins to Pb(II) inhibition. Thus, His-131 and Cys-223 do not participate in the Pb(II)⅐enzyme equilibrium during the steady-state turnover. From this we conclude that the ligands to ZnB most likely act as ligands to Pb(II) during the less profound steady-state inhibition. Supporting this conclusion is the result of a 6-h time course assay of MinusZnB, which showed no significant rate difference for assays done with or without Pb(II) or Zn(II) (see Table II  tions of Pb(II) cause a progressive increase in the K m value for ALA as well as a progressive decrease in the V max value. If one presumes that the Pb(II) replaces Zn(II), then this supports a role for Zn(II) in both substrate binding and catalysis.
Pb(II) Competition for the Zn(II) Sites of Human PBGS-Purified human and bovine PBGS contain eight Zn/octamer (3,7). When dialyzed against phosphate buffer at neutral pH at 4°C in the absence of additional Zn(II), four tightly bound Zn(II) remain (3,12). Assuming these Zn(II) are available to bulk solution, the estimated K d for this Zn(II) is Ͻ0.1 M; a second group of four Zn(II) are shown to be in equilibrium with free Zn(II) and bind with a K d of about 5 M. If instead, the protein is first brought to pH 5, all enzyme-bound Zn(II) are free to dissociate (3). Low pH appears to remove the enzyme's asymmetry; a subsequent equilibrium dialysis at room temperature in neutral pH phosphate buffer revealed the binding of eight Zn(II) per octamer with a single K d of about 1 M (3). Preliminary data on the ability of Pb(II) to displace the Zn(II) of human PBGS suggested that Pb(II) preferentially displaces Zn(II) from only half of the Zn(II) sites (3). Here, equilibrium dialysis experiments were performed on N59/C162A, Mi-nusZnA, and MinusZnB using asymmetric "as purified" protein in the absence of substrate. These data reflect Pb(II) and Zn(II) occupancy at t ϭ 0 of the kinetic studies, prior to the hysteresis shown in Fig. 2C. Fig. 3 illustrates Zn(II) binding to N59/ C162A, MinusZnA, and MinusZnB at 10 M Zn(II) and 0.5 M Zn(II) in the presence and absence of 20 M Pb(II). These data show that, at 0.5 M Zn(II), Pb(II) competes favorably for the ZnB sites (see MinusZnA results). Results with MinusZnB show far less Zn(II) binding to the ZnA site, and this enzymebound Zn(II) is not displaced by Pb(II). Fig. 1 contains two different ligand environments for Zn(II) and pertains to one-half the human PBGS subunits. An alternative model would have one ZnA on each of four of the subunits and one ZnB on each of the other four subunits. It is not yet possible to differentiate between these two models for the asymmetric eight Zn(II)-containing forms of human PBGS. Nevertheless, the Zn(II) stoichiometry data presented here on MinusZnA and MinusZnB establishes important aspects of our model; one-half the Zn(II) purified with human PBGS are ZnA, the other half are ZnB, and the ligands to these Zn(II) are likely as depicted in Fig. 1.

The Two Zn(II) Sites of Human PBGS-The model for Zn(II) bound to human PBGS presented in
The activity data for MinusZnA and MinusZnB unequivocally establish ZnB as the catalytically essential four Zn(II). This agrees with popular dogma, but was not obvious on the basis of common Zn(II)-containing enzymes. ZnA, inferred from the EXAFS to be pentacoordinate with predominantly oxygen and nitrogen ligands (8), appeared to be a good candidate for a catalytic Zn(II), because its coordination environment resembles many well characterized catalytic Zn(II) (25). Furthermore, an EXAFS study showed four ZnA to bind more tightly than four ZnB to bovine PBGS (8). This agreed with the notion that ZnA is the essential Zn(II) but is contrary to chemical intuition and to the crystallographic data on yeast PBGS (14). ZnB, which is inferred from the EXAFS to have a tetrahedral array of cysteine ligands, closely resembles structural Zn(II) such as is found in zinc fingers (25). There are, however, emerging examples of catalytically essential Zn(II)-thiolates (26). Based on a comparative analysis of the structures of yeast and Pseudomonas aeruginosa PBGS, the locations of both ZnA and ZnB appear to be within the active site when a mobile flap covering the active site is closed (1). The yeast PBGS crystal structure shows ZnB to be at high occupancy, and the low activity of the MinusZnB variant of human PBGS establishes ZnB as the one that is of catalytic importance. The role of ZnA in protein function, if any, remains unclear.
Lead Interactions with PBGS-Mammalian PBGS has long been known as a target enzyme for lead poisoning, as a Zn(II) metalloenzyme, and as an enzyme in which cysteine residues in a reduced state are important for activity. An attractive hypothesis developed, which presumes that there is a catalytically essential Zn(II) bound to these cysteine residues and that Pb(II) inhibits the protein through replacement of this Zn(II) (27). We now see that there are two types of Zn(II) in human PBGS, and at least one cysteine functions in binding each of the Zn(II) types. The current study defines the relationships of these Zn(II) and their ligands to inhibition by Pb(II).
The lead inhibition studies presented here and previously indicate that the interaction of Pb(II) with mammalian PBGS is altered in the presence of substrate (24). Furthermore, the steady-state inhibited rate was attained slowly. This type of phenomenon has been referred to as "slow binding inhibition" and indicates that the interaction of the inhibitor, here Pb(II), with the enzyme changes as a consequence of protein motion during catalysis. One precedent for a substrate-induced change in metal binding properties of a PBGS is the substrate-induced disproportionation of Mn(II) between two sites of Bradyrhizobium japonicum PBGS (28). We propose that the structural   Table III) implicates the mutated ZnA ligands in one of the Pb(II)-binding sites. Because the steady-state rates are not altered, the ZnA ligands are not implicated in the complex formed during the steady-state inhibition but rather in a tight-binding inhibitory Pb(II) site, which preferentially exists in the absence of substrate. This site is disfavored in the presence of substrate; it is missing in MinusZnA. We modeled the Pb(II)-binding sites in human PBGS (Fig. 4), drawing on published observations of Pb(II) bound to yeast PBGS in two alternate positions (2). One Pb(II) site appears to use the three ZnB ligands. This Pb(II) is called PbB and is shown in Fig. 4A. The three ligands are Cys-122, Cys-124, and Cys-132, and the lone pair of electrons on Pb(II) occupies the fourth position of a tetrahedral coordination geometry. Because the distance between Zn(II) and sulfur is typically around 2.2 Å and Pb(II) to sulfur distances are ϳ2.9 Å (29), PbB does not occupy precisely the same location as ZnB but is slightly further away from the cysteines. The second site is called PbAB; it is proposed to use Cys-124 and Cys-132 but not Cys-122 (numbered as in human PBGS), where the side chains are in a different configuration not seen in the yeast PBGS structure (2,30). We assume that Cys-223 is the ZnA ligand most likely to interact with Pb(II) in the alternate configuration. PbAB is illustrated in Fig. 4B. PbAB uses as ligands the side chains used by both ZnA and ZnB, Cys-124, Cys-132, and Cys-223, and the lone pair from Pb(II) sits at the fourth position of the tetrahedron. To model the PbAB site, the side-chain rotamer configurations of Cys-122, Cys-124, Cys-132, and Cys-223 were sampled while the backbone coordinates were held fixed. It is important to notice that Pb(II) occupancy of these two sites is mutually exclusive, that is, they cannot exist on the same subunit at the same time. This is because both use Cys-124 and Cys-132 as ligands, but the side-chain configurations of these amino acids are different for PbB versus PbAB. Pb(II) is apparently inhibitory in both the PbB and the PbAB sites for two reasons. First, in both sites the lone electron pair on Pb(II) would interfere sterically with substrate or product binding. Second, both positions for Pb(II) disallow the binding of ZnB, which this study shows is catalytically important perhaps through interaction with its cysteine ligands.
One might ask why turnover disfavors the PbAB site. The cysteine residue analogous to Cys-223 of human PBGS showed itself in the yeast PBGS crystal structure (Protein Data Bank file law5) to be the first ordered residue after a disordered region (14). The P. aeruginosa PBGS crystal structure identified this disordered region as a flap that can isolate the active site from solvent during catalysis (31). Our comparison of the two available structures of yeast PBGS (Protein Data Bank codes law5 and 1ylv) using the combinatorial extension method (32) shows the cysteine to be positioned differently when the lid is ordered relative to when it is disordered. Consequently, we propose that motion of the active site lid during catalysis moves Cys-223 and disfavors the population of the PbAB. Thus, at steady state, the enzyme-bound Pb(II) is preferentially bound to the PbB site. Variants K59 and N59 in Lead Poisoning-A body of epidemiological literature correlates the allelic variation between K59 and N59 with a differential susceptibility to lead poisoning (e.g. Ref. 33). A smaller number of epidemiological studies do not support this hypothesis (e.g. Ref. 34). Prior studies on purified K59 and N59 did not reveal a molecular basis for a difference nor did they support the existence of a difference (3). The current study reveals a small difference in the half-time for recovery from Pb(II) inhibition during turnover conditions in vitro. However, because most PBGS in circulating erythrocytes is left over from the massive heme biosynthesis that occurs in reticulocytes (that is, catalysis is not occurring), this difference is unlikely to affect the interaction of Pb(II) with blood PBGS, which is the major site of Pb(II) binding in circulating blood (35).
Conclusion-We have prepared the mutant human PBGS proteins MinusZnA and MinusZnB, which each purify with four Zn(II) per octamer, thus supporting the model of native human PBGS as normally purifying with four ZnA and four ZnB. This metal stoichiometry is fundamentally different from that of yeast PBGS, which was previously used as a model system for the human enzyme in lead poisoning. The mutagenesis studies show that ZnB and its ligands are catalytically important, whereas ZnA and its ligands are not. This provides biochemical characterization and confirmation of an unusual catalytic Zn(II) site. Pb(II) bound at a site analogous to the ZnB site (PbB) is shown to be a site of lead inhibition during steadystate turnover. Pb(II) is proposed alternatively to bind to a PbAB site, which uses ligands normally involved in the binding of both ZnA and ZnB. The PbAB site is disfavored during turnover. In both environments, it is the lone election pair of Pb(II) that is proposed to alter the active site environment and disfavor substrate binding and subsequent catalysis.