1.1. Porphobilinogen synthase: still a mechanistic enigma

Porphobilinogen synthase (PBGS, E.C. 4.2.1.24) is the second enzyme in the biosynthesis of the 'pigments of life' [1] and catalyses the formation of porphobilinogen (PBG) (1) starting from two molecules of 5-aminolevulinic acid (ALA) (2) (Scheme 1) [2,3].

Scheme 1: Biosynthesis of porphobilinogen (1)

PBG is transformed with the aid of two additional enzymes to the last common precursor - the uroporphyrinogen III (Uro III) [4,5]. This part of the biosynthesis is common to all the biochemical pathways leading to tetrapyrrolic natural products. In the absence of the porphobilinogen synthase, two molecules of ALA condense to a symmetric diimin, which is oxidized to a pyrazine derivative [6]. Despite intensive studies of the PBGS from different sources, the exact mechanism of formation of PBG is not yet clear. Several groups have published methods to obtain crystal of PBGS from different sources [7] and recently the first x-ray structure of PBGS from yeast with 2.3 Å resolution has been etablished [8]. Different mechanisms have been proposed for this seemingly simple Knorr type condensation [9]. The mechanism for the enzyme catalysed reaction must allow to explain the dichotomy between the chemical condensation and the biochemical transformation. The difference between the chemical reactivity and the biosynthesis is also an important issue in the context of the potential status of the ´pigments of lifeª as prebiotic compounds [10,11]. In the first proposal, Shemin postulated a C-C bond formation, via an aldol type reaction as the central step for the biosynthetic mechanism (Scheme 2) [12]. The aldol reaction creates the intermediate 3 which joins the two aminolevulinates for the first time together.

Scheme 2: Proposed mechanisms for the biosynthesis of porphobilinogen

Based on this mechanism, a biomimetic synthesis could be achieved in satisfactory yield in our group (Scheme 3) [13]. In this synthesis the connection of two analogues of ALA occurs first by a formation of C-C bond, via an aldol reaction. This intermediate spontaneously ring closes to form the desired pyrrole with all substituents in place.

Scheme 3: Biomimetic synthesis of porphobilinogen [13]

Several years later Jordan postulated a C-N bond formation connecting the two substrates as the first step [14]. In the same paper, he suggests also a mechanistic alternative, which follows more closely Shemin's proposal. However Jordan postulates that the first substrate recognised by the enzyme goes to the P-Site, as shown by his elegant pulse labelling experiment [5]. Until now, no mechanism could be proven. The lack of firm structural information does not allow to distinguish between the several mechanistic proposals for the PBG formation so far.

1.2. Biochemical aspects

The porphobilinogen synthase is a wide spread metallo-enzyme which has been found in plants, bacteria, animals and humans [15]. The enzyme was shown to be an octamer with 8 identical subunits. In same cases it has been shown the the dimer is the minimum active subunit of PBGS [3]. The primary structures derived from cDNAs suggest that all the Porphobilinogen synthase are quite similar to each other [16-18]. Early on it has been shown that PBGS is forming a Schiff base with at least one of the two substrates [19]. The identification of the Schiff base as a competent intermediate in the biosynthetic pathway influenced strongly the early mechanistic proposals [12]. The active site lysine has been identified using labeled substrate, which was irreversibly bound to the enzyme by reduction with NaBH4 [12]. The cDNA derived protein sequences in the vicinity of the active site lysine are highly conserved independent of the organism from which the enzyme has been isolated [20]. Also the overall homology of the sequences for PBGS from different organisms is high [16]. EXAFS-Studies (Extended X-ray Absorption Fine Structure) of bovine liver PBGS allowed to identify two different binding sites for Zn(II) [18]. Based on these studies it could be shown that one of the Zn-ion is complexed with four cysteines, and the other Zn (II) with cysteine, histidine, aspirigine and tyrosine (Scheme 4) [21].

Scheme 4: Postulated arrangement of the two Zn-ions in PBGS from bovine [18].

A systematic comparison of the primary structures leads to the prediction of characteristic metal binding sites [22]. PBGS has been divided into two categories based on the dependence of metal ions for their reactivity: 1) the Zn2+ dependent and 2) the Mg2+ dependent enzymes [3]. The first class shows a pH optima around 6.3 and 7.1. PBGS from Escherichia coli and yeast have been attributed to this category despite the more alkaline pH needed for activity (for E.coli, pH= 8.1) [23]. Jaffe proposed an enlarged categorisation [16,24]. She based her proposal on the hypothesis that hints for the presence of up to three different binding sites for divalent metals like zinc and magnesium have been accumulated by her group [16,20-22,24,25]. She classified the porphobilinogen synthase in 4 different classes depending of the occupation of the three sites by metal ions. The enzyme isolated from Escherichia coli, contains two sites (Site A and site B) for Zn (II) and one site (Site C) for Mg (II) and belongs therefore to class II. The two Zn(II) atoms are located close to the active site, although only one Zn (II) is necessary for activity (Scheme 4). The Mg 2+ has the role of an allosteric activator [26].

PBGS catalyses the transformation of two molecules of the same substrate. PBGS must therefore contain two sites of recognition for two identical substrates. Jordan determined the sequence of recognition for the enzymes isolated from human and bovine liver by pulse labelling experiments [14,27]. Using [5-14C]ALA Jordan determined that the first ALA recognised by the enzymes will form the propionic acid side chain in the final product (P-side ALA recognised by the P-site of the enzyme) and not in the A-site (Scheme 5).

Scheme 5: Results of the pulse labelling experiment by Jordan [14,27].

Dialysis experiments allowed to determine the dissociation constants of the P-side ALA (KD = 3.8 µM) and the dissociation constant of the A-side ALA (KD = 242 µM) [21]. The P-side ALA is recognised 64 times better than the A-side ALA. Labelling experiments showed, that the active site lysine of the PBGS from human erythrocytes forms a Schiff base with the P-side ALA [28]. Based on binding studies with covalently modified PBGS it was postulated that the A-side ALA is stabilised in the active site by complexation with Zn(II) ion [21]. In analogy to the X-ray of Zn(II) enzyme carboxypeptidase with glycylglycine [29] a chelate between the amino group and the carbonyl group of the substrate and the Zn-ion is postulated. The Zn(II) situated close to the A-site is called ZnA and is essential for the activity of PBGS. The second Zn (II) - called ZnB - is proposed to be near to the bound P-side substrate. ZnB is not necessary for the activity of the enzyme. The P-side ALA is the tighter bound substrate of PBGS and forms a protonated Schiff Base between a lysine and the carbonyl group of ALA as could be shown by highly elegant NMR experiments [30-32].

1.3. Site selective inhibition studies

Since several years our group investigates the mechanism of PBGS by systematic inhibition studies [15,33,34]. To test the recognition sites of PBGS from E. coli [34] and R. spheroides [33], a series of analogues of the substrate, of the product and also of postulated intermediates have been synthesised and their inhibition potency has been determined. The goal of these investigations is to contribute to our knowledge on the active site of the enzyme, to elucidate possible mechanisms for the transformation and to analyse differences between enzymes from different sources.

Since the publication of the review article dedicated to the studies of the mechanism of PBGS and to the results of inhibition experiments, several groups have contributed significantly to our knowledge in this field [23,35,36]. Due to the availability of several recombinant enzymes in large quantities more systematic studies have become possible in recent years. The behaviour of the enzymes from Escherichia coli, Saccharomyces cerevisiae and the plant enzyme from Pisum sativum could be compared systematically [23]. The inhibition potency of the inhibitors studied with these three enzymes was rather similar with one major exception: the enzyme from P. sativum was much more sensitive to succinyl acetone than the enzymes from the two other sources. In this study a series of diacids were studied as well in view of the fact that the product PBG is bound relatively tightly to the enzyme [30]. The inhibition constants determined were so high, that no clear conclusion could be drawn from the studies of these compounds (see chapter 1.6.). In a second study from the same group a lower homologue of the substrate and two diketones were studied [35]. The diketone formed adducts with the enzyme, but these adducts could not be identified with electrospray mass spectroscopy (ES-MS). To obtain unequivocal proof of the incorporation of the diketones the adducts had to be reduced with NaBH4.

A systematic study of the inhibition potency of analogues of 5-aminolevulinic acid has been reported very recently [36]. The enzyme used in this study stems from Bacillus subtilis, whose gene had been cloned and overexpressed in E. coli [37]. Most of the results of the inhibition studies with B. subtilis are quite similar to the results obtained from enzymes from other sources. The most important difference is the marked sensitivity of PBGS from B. subtilis towards the presence and absence of the 5-amino group. With some of the substrate analogues inactivation of PBGS could be observed. ES-MS proofed to be the technique of choice to determine the inactivation. Important experiments in this context were dedicated to the study of the inactivation with 5-chlorolevulinic acid [21,36,37]. The results of the ES-MS at different concentrations of the inhibitor are indicative for a non-specific alkylating agent. At low concentrations clean monoalkylation was detected in the ES-MS but virtually no inactivation could be observed [36]. In the same paper the synthesis and the inhibition study of a close analogue of the intermediate postulated by Jordan is reported. The inhibition potency of this compound is very low indeed [36].

Finally two groups have reported a mechanism based inactivation of PBGS using a thioester analogue of the substrate [36,38]. One group detected the bond formation between the active site of the enzyme and the glycyl part of the inhibitor with the help of NMR spectroscopy [38] whereas the second group used ES-MS for the identification of the covalent adduct [36]. As a consequence of these beautiful studies our knowledge about PBGS has been considerably increased. Despite this effort a clear answer concerning the mechanism of PBGS could not be obtained yet.

1.4. Kinetics of the inhibition studies of PBGS

PBGS catalyses the asymmetric condensation of two identical substrates and belongs to the class of the two-substrate enzymes [39]. The kinetics of this class of enzymes is necessarily more complex and the understanding of the inhibition is often not straight forward. However the observed kinetics follows without any doubt the laws of a Michaelis-Menten kinetics [40]. Granick was the first to discuss this problem [41]. He proposed the following three conditions in order to explain this kinetic behaviour: the two substrates are binding to the enzyme in a predetermined sequence; the dissociation constant of the first enzyme-substrate complex is considerably lower compared to the second and the final step -the formation of PBG - is the rate-determining step. Accepting these three conditions, the formation of PBG follows a Michaelis-Menten kinetics. Using arguments based only on the kinetics of the process the Michaelis-Menten behaviour is due to the fact, that the first more tightly bound substrate is forming a Schiff base with the enzyme. This covalently bound enzyme-first substrate complex (= E') is formed fast and reversibly and the enzyme kinetics we observe under our experimental conditions is the formation of the Michaelis-Menten complex (S.E') between E' and the second substrate (= A-side ALA) and the transformation to the product [41].

Scheme 6: Schematised mechanism for PBGS showing the intermediates relevant for the kinetic behaviour.

Inhibitors can interact in different steps of the reaction [42]. During the inhibition tests carried out in our group it became clear that different kinetic behaviours are observed for different inhibitors. Very often the inhibition behaviour - competitive, uncompetitive, mixed, slow binder or irreversible - stays the same for a series of compounds. The most interesting results are those where one slight change in the structure completely changes the inhibition behaviour. In view of these results it is an important to find out if the difference of the inhibition behaviour can be interpreted as difference of the site of interaction with the enzyme? The large number of inhibitors studied in our group allow us to arrive at a few general conclusions.

The analogues of the substrate as well as the analogues of the product show a pure competitive inhibition [33]. Competitive inhibition is attributed to a direct competition between the substrate and the inhibitor. In our case a direct competition with the second substrate has to be assumed. The competitive inhibitors have therefore to interact with the A-site of the enzyme. The analogues of the substrate are obviously well recognised in this site. All variations in inhibition potency of competitive inhibitors have to be attributed to difference in recognition with the A-site of the enzyme.

A large number of inhibition studies were done in our group and by other groups [8,35,36,42-44] with substances containing 1,3-diketones. The first compound of this series detected and described as a potent inhibitor of PBGS was the succinyl acetone (5) [42-45]. The succinyl acetone still is the best inhibitor of the PBGS known (Ki=1.4 mM) and shows an uncompetitive or mixed inhibition. Almost all compounds containing this structural element manifest an uncompetitive inhibition. Only two of them could be shown to be competitive inhibitors. Comparing the Ki of 6 and 7 indicates that the replacement of one carboxylate by one nitro group induces the switch from uncompetitive to competitive inhibition. The Ki-value changes from 318 mM to a value around 16.5 mM.

Scheme 7: Succinyl acetone (5) and related compounds

The interpretation of the uncompetitive behaviour of our inhibitors is less straight forward than the interpretation of the competitive behaviour. The hypothesis we use at the moment to explain the uncompetitive behaviour is the following. We assume that the uncompetitive inhibitors interact as well with the A-site of the enzyme as with the P-site. In order that the interaction with the P-site becomes kinetically competent the Schiff base formed between the inhibitor and the P-site lysine has to be more stable or at least as stable as the Schiff base formed between the substrate and the active site lysine. Under these conditions we will observe the mixed inhibition of the free enzyme (=E) and the competitive inhibition with the enzyme linked with the first substrate. This double interaction leads kinetically to uncompetitive or mixed inhibition behaviour. In general the inhibition constants determined for the uncompetitive inhibitors are often lower than those of the corresponding competitive inhibitors. This observation is in accordance with our hypothesis. The higher inhibition efficiency of the uncompetitive inhibitors can be explained by the difference between the two Km values ( KmP lower than KmA) for the two substrates (P-side substrate and A-side substrate). The two compounds 6 and 7 are on the border between competitive and uncompetitive inhibitors. The introduction of the nitro changes the type of inhibition. In this case however the competitive inhibitor has a lower dissociation constant than the uncompetitive inhibitor. This experiment is in accordance with the hypothesis that the nitro group is better recognised if it is used as a substitute of the carboxylate group of the A-Side substrate. The replacement of a carboxylate by a nitro group has not the same beneficial effect if the exchange is done on the P-side of the molecule. In this case the complement on the enzyme active site for the A-side carboxylate is already occupied and the introduction of the nitro group is not as valuable as indicated by the change in the Ki-values.