- Email: [email protected]
Biosynthesis of vitamin B2: Structure and mechanism of riboflavin synthase
Available online at www.sciencedirect.com
ABB Archives of Biochemistry and Biophysics 474 (2008) 252–265 www.elsevier.com/locate/yabbi
Review
Biosynthesis of vitamin B2: Structure and mechanism of riboflavin synthase Markus Fischer a,*, Adelbert Bacher b b
a Institute of Food Chemistry, University of Hamburg, Grindelallee 117, D-20146 Hamburg, Germany Lehrstuhl fu¨r Biochemie, Technische Universita¨t Mu¨nchen, Lichtenbergstr. 4, D-85747 Garching, Germany
Received 19 December 2007, and in revised form 5 February 2008 Available online 14 February 2008
Abstract The biosynthesis of one riboflavin molecule requires one molecule of GTP and two molecules of ribulose 5-phosphate as substrates. GTP is hydrolytically opened, converted into 5-amino-6-ribitylamino-2,4(1H,3H)-pyrimidinedione by a sequence of deamination, side chain reduction and dephosphorylation. Condensation with 3,4-dihydroxy-2-butanone 4-phosphate obtained from ribulose 5-phosphate leads to 6,7-dimethyl-8-ribityllumazine. The final step in the biosynthesis of the vitamin involves the dismutation of 6,7-dimethyl-8-ribityllumazine catalyzed by riboflavin synthase. The mechanistically unusual reaction involves the transfer of a four-carbon fragment between two identical substrate molecules. The second product, 5-amino-6-ribitylamino-2,4(1H,3H)-pyrimidinedione, is recycled in the biosynthetic pathway by 6,7-dimethyl-8-ribityllumazine synthase. This article will review structures and reaction mechanisms of riboflavin synthases and related proteins up to 2007 and 122 references are cited. Ó 2008 Elsevier Inc. All rights reserved. Keywords: Vitamin B2; Riboflavin synthase; Lumazine synthase; Lumazine protein; X-ray structure; Reaction mechanism
In association with a wide variety of apoenzymes, the flavocoenzymes derived from vitamin B2 (riboflavin) are probably the most chemically versatile cofactors. They have long been known to enable a wide variety of redox reactions involving one- as well as two-electron transfer processes, whereas other coenzymes are typically limited to one of them (for example, pyridine nucleotides catalyze exclusively two-electron transfers) [1,2]. Moreover, they are involved in an impressive number of reactions that do not involve a net exchange of electrons with the respective substrate. For a comprehensive description of flavoenzymes that catalyze reactions with no net redox change the reader is directed to the review by Bornemann [3]. Flavins serve as primary and secondary emitters in bacterial luminescence. Lumazine proteins, which can bind riboflavin, FMN, 6,7-dimethyl-8-ribityllumazine or 6methyl-7-oxo-8-ribityllumazine, are believed to act as opti*
Corresponding author. Fax: +49 40428384342. E-mail address: markus.fi[email protected] (M. Fischer).
0003-9861/$ - see front matter Ó 2008 Elsevier Inc. All rights reserved. doi:10.1016/j.abb.2008.02.008
cal transponders in the bioluminescence emission of photobacteria [4–8]. The interposition of lumazine protein as an optical transponder between the energy generation and light emission steps has been proposed to result in shifts of the emission maximum and an increased quantum yield [9]. This extraordinary photochemical functionality reflects the enormous versatility of the isoalloxazine chromophore, whereas the side chains of the flavocoenzymes serve primarily as anchors that secure their binding to the cognate apoproteins, which typically form relatively tight complexes. Moreover, flavocoenzymes can form very complex catalytic sites involving more than one flavocoenzyme, modified flavins and/or additional cofactors such as iron sulfur clusters [10–13]. In some cases, flavocoenzymes become covalently linked to their apoenzymes [14]. Photoreceptors involved in processes like stem bending toward a light source (phototropism), chloroplast migration to places of appropriate light intensity (chloroplast photorelocation), and the
M. Fischer, A. Bacher / Archives of Biochemistry and Biophysics 474 (2008) 252–265
253
Fig. 1. Biosynthesis of riboflavin and flavocoenzymes. Step I, GTP cyclohydrolase II; step II, 2,5-diamino-6-ribosylamino-4(3H)-pyrimidinone 50 phosphate deaminase; step III, 5-amino-6-ribosylamino-2,4(1H,3H)-pyrimidinedione 50 -phosphate reductase; step IV, 2,5-diamino-6-ribosylamino-4(3H)pyrimidinone 50 -phosphate reductase; step V, 2,5-diamino-6-ribitylamino-4(3H)-pyrimidinone 50 -phosphate deaminase; step VI, hypothetical phosphatase; step VII, 3,4-dihydroxy-2-butanone 4-phosphate synthase; step VIII, 6,7-dimethyl-8-ribityllumazine synthase; step IX, riboflavin synthase; step X, riboflavin kinase; step XI, FAD synthetase; 1, GTP; 2, 2,5-diamino-6-ribosylamino-4(3H)-pyrimidinone 50 -phosphate; 3, 5-amino-6-ribosylamino2,4(1H,3H)-pyrimidinedione 50 -phosphate; 4, 2,5-diamino-6-ribitylamino-4(3H)-pyrimidinone 50 -phosphate; 5, 5-amino-6-ribitylamino-2,4(1H,3H)pyrimidinedione 50 -phosphate; 6, 5-amino-6-ribitylamino-2,4(1H,3H)-pyrimidinedione; 7, ribulose 50 -phosphate; 8, 3,4-dihydroxy-2-butanone 4phosphate; 9, 6,7-dimethyl-8-ribityllumazine; 10, riboflavin; 11, FMN; 12, FAD.
opening of stomatal guard cells to facilitate gas exchange are the plasma membrane-associated phototropins and homologs [15–21]. The common features of the phototropin photoreceptor family are two light-sensitive domains, LOV1 and LOV2, and a serine/threonine kinase domain. Each of the two LOV domains non-covalently binds a single flavin mononucleotide (FMN)1 as a chromophore. After illumination by blue light, the LOV photocycle comprises a light-induced addition of a thiol group to the C(4a) position of the FMN chromophore [20,22]. Flavins have also been shown to be involved in circadian and seasonal clocks [23–25]. The mechanistic complexity of flavoprotein catalysis is to some extent matched by the complexity of some of the reactions involved in the biosynthesis of the isoalloxazine moiety of riboflavin. Surprisingly, some of these mechanistically complex reactions can proceed without catalysis under relatively mild conditions [26–33]. Thus, it is conceiv-
1 Abbreviations used: FMN, flavin mononucleotide; MjaRS, M. jannaschii.
able that the formation of flavin could initially have occurred by spontaneous processes, prior to the evolution of macromolecular biocatalysis. The biosynthesis of riboflavin has been reviewed repeatedly and the reader is directed to these articles for an overview of the pathway (Fig. 1) [34–38]. This article will focus on the last step in the biosynthetic pathway, which continues to present fascinating mechanistic problems even after more than four decades of research. Briefly, the initial steps of the riboflavin pathway involve the conversion of GTP (1) into 5-amino-6-ribitylamino2,4(1H,3H)-pyrimidinedione (6). The intermediate is obtained via structurally different precursors in different organisms. More specifically, the first committed pathway intermediate, 2,5-diamino-6-ribosylamino-4(3H)-pyrimidinone 50 -phosphate (2), is reduced and then deaminated in fungi and in Archaea [39,40], whereas reduction of the side chain precedes deamination in eubacteria and plants [41–44]. It is still unknown how the phosphate residue in 5-amino-6-ribitylamino-2,4(1H,3H)-pyrimidinedione 50 phosphate (5) is released to generate 5-amino-6-ribitylamino-2,4(1H,3H)-pyrimidinedione (6). It is clear, however,
254
M. Fischer, A. Bacher / Archives of Biochemistry and Biophysics 474 (2008) 252–265
Fig. 2. Anionic species of 6,7-dimethyl-8-ribityllumazine and 6,7,8-trimethyllumazine.
that dephosphorylation is mandatory since the phosphate ester cannot serve as a substrate for the consecutive enzyme reaction catalyzed by 6,7-dimethyl-8-ribityllumazine synthase (VIII), which catalyzes the condensation of 6 with 3,4-dihydroxy-2-butanone 4-phosphate (8) that is obtained by a complex rearrangement/elimination sequence from ribulose 5-phosphate (7). The final step in the biosynthesis of riboflavin involves a most unusual dismutation. In a formal sense, two molecules of 6,7-dimethyl-8-ribityllumazine (9) exchange a four-carbon unit, thus transforming one of them into riboflavin and the other into 5-amino-6-ribitylamino2,4(1H,3H)-pyrimidinedione (6). The reaction is catalyzed by riboflavin synthase (IX) but can also proceed without catalysis in boiling aqueous solution at neutral or acidic pH [30,32,33,45]. The second product of the dismutation, 6, can again serve as substrate for lumazine synthase (VIII). The xylene ring of riboflavin (10) is generated from two molar equivalents of the carbohydrate precursor, 3,4-dihydroxy-2-butanone 4-phosphate (8). Since only one four-carbon unit can be condensed with the pyrimidine precursor 6 by lumazine synthase, the reaction stoichiometry requires that the four-carbon unit is exposed twice to the catalytic action of lumazine synthase. However, the random nature of that process deserves some emphasis. In closer detail, one half of the GTP molecules that enter the pathway make it all the way through to riboflavin in their first passage. The others must pass through at least one additional cycle of consumption/regeneration under the catalytic influence of lumazine synthase and riboflavin synthase. In fact, a fraction of pyrimidine molecules (6) will complete several cycles of consumption/regeneration before reaching the level of riboflavin (10). Formally, 6,7-dimethyl-8-ribityllumazine can be interpreted as a tetra-nor analog of riboflavin, and the chemical similarity between the two compounds is indeed quite close. Both show strong absorption of visible light and both are fluorescent with high quantum yields. Both can be easily reduced to the respective dihydro forms, and the 50 -phosphate of 6,7-dimethyl-8-ribityllumazine can serve as a catalytically competent flavocoenzyme analog with certain apoenzymes. Notably, the properties of the 7a methyl group of 6,7dimethyl-8-ribityllumazine and the vinylogous 8a methyl group of riboflavin differ dramatically in their acidity.
The 8a methyl group has a somewhat enhanced acidity, which enables selective deuteration, albeit under drastic reaction conditions. On the other hand, the 7a methyl group of 6,7-dimethyl-8-ribityllumazine and of other 8substituted 7-methyl lumazines represent extreme cases of CH acidity. Specifically, the pK for the dissociation of a proton from the 7a methyl group is about 8 for 6,7dimethyl-8-ribityllumazine [46] and about 10 for 6,7,8-trimethyl-8-ribityllumazine [47]. In the case of compounds carrying appropriately placed hydroxyl groups at the position 8 side chain, the exomethylene anion resulting from deprotonation is subject to nucleophilic attack of C-7 of the lumazine ring system; in the case of 6,7-dimethyl-8-ribityllumazine, the exomethylene anion species accounts only for a small percentage of the anion mixture, which is dominated by four isomeric tricyclic anion species in rapid equilibrium (Fig. 2) [48–50]. Recent studies have identified a covalent adduct of two 6,7-dimethyl-8-ribityllumazine molecules (9) as a key intermediate (Fig. 3, compound Q) in the enzymatic formation of riboflavin [51–54]. Not surprisingly, the fusion of the ring systems occurs in cis [51]. Surprisingly, however, the structurally unrelated riboflavin synthases of Archaea, on the one hand, and of plants, fungi and eubacteria, on the other, proceed via diastereomeric forms of the pentacyclic intermediate (Fig. 3, compound Q0 and compound Q, resp.) (the C-atoms 6 and 7 of the pentacyclic intermediate, without reference to the chiral centers of the ribityl side chains, are of course enantiomeric). It is obvious that a sequence of two elimination reactions can convert each of the two different pentacyclic intermediates into equimolar amounts of riboflavin (10) and 5-amino-6-ribitylamino-2,4(1H,3H)-pyrimidinedione (6). Each of the two different riboflavin synthases types is limited to cleaving its cognate reaction intermediate but not that produced by enzymes of the other type, whereas both diasteromeric pentacycles can afford the product mixture by uncatalyzed fragmentation. On the other hand, the formation of the covalent pentacyclic adduct by dimerization of two 6,7-dimethyl-8-ribityllumazine remains mechanistically enigmatic, even more so since it can obviously proceed in the absence of a catalyst. Although pentacyclic intermediates have only been documented experimentally for the enzyme-catalyzed process, it appears rather safe to assume that they are also involved
M. Fischer, A. Bacher / Archives of Biochemistry and Biophysics 474 (2008) 252–265
255
Fig. 3. Stereochemistry of 6,7-dimethyl-8-ribityllumazine conversion into riboflavin catalyzed by trimeric eubacterial and pentameric archaeal riboflavin synthase. Binding of the substrates in anti-parallel orientation occurs at two sites, one leading to acceptance of a four-carbon unit (9a, acceptor site) and the other to its donation (9b, donor site); Q and Q0 , pentacyclic reaction intermediates. R, ribityl.
256
M. Fischer, A. Bacher / Archives of Biochemistry and Biophysics 474 (2008) 252–265
in the case of the uncatalyzed reaction, which has been shown to proceed with exactly the same, strict regiospecificity as the enzyme-catalyzed process. Notably, both the uncatalyzed and the enzyme-catalyzed processes require a quasi-c2-symmetric arrangement of the substrates in the transition state for the dimerization. We will return to this point after an introduction to the structures and properties of riboflavin synthases. Riboflavin synthase activity was first observed in cell extracts of naturally flavinogenic ascomycetes [55–58]. Later, the enzymes from the yeasts Saccharomyces cerevisiae and Schizosaccharomyces pombe, the eubacteria Bacillus subtilis and Escherichia coli and from the plant, Arabidopsis thaliana, have been studied in some detail [51,59–68,122]. These enzymes are all homotrimers. However, as described below, the riboflavin synthases from Archaea are homopentamers with completely unrelated amino acid sequences. The subunits of all homotrimeric riboflavin synthases are all characterized by intramolecular sequence similarity.
Sequence comparison between the proximal and distal domains of the homotrimeric riboflavin synthases from all completely sequenced genomes showed two non-overlapping branches. Hence, it is believed that both domains have evolved from a common ancestor by gene duplication. This domain structure has been observed for the first time using the enzyme from B. subtilis. This sequence showed marked internal homology encompassing 26 identical residues and 23 conservative replacements [69]. Later, this has been shown for enzymes of other organisms, like E. coli or S. pombe (Fig. 4) [51,69]. It was also correctly suggested on basis of the sequence data that two domains could form a pseudo-c2-symmetric pair that could form an active site at the domain interface. A pair of substrate molecules bound at that pseudo-c2-symmetric domain interface would have precisely the antiparallel topology required by the regiochemical features of the reaction. Notably, the intramolecular sequence similarity does not extend to the C-terminal segment of about two dozen residues. It is also relevant
Fig. 4. Internal sequence alignment of riboflavin synthase from S. pombe. (a) Alignment of the N barrel (red) and the C barrel (blue) domains; (b) stereodiagram of an internal superposition of the two b barrels with bound substrate analog inhibitor 6-carboxyethyl-7-oxo-8-ribityllumazine (18, PDB entry #1KZL). Reprinted from Gerhardt et al. [51], Copyright (2002) by Elsevier.
M. Fischer, A. Bacher / Archives of Biochemistry and Biophysics 474 (2008) 252–265
that the N-terminus of trimeric riboflavin synthases (conserved motif MFTG) shows a very high degree of evolutionary conservation. The ensuing hypothesis that the subunits fold into two closely similar domains was later validated by crystal structure analysis [51,64]. Surprisingly, the crystal structure analysis of the E. coli enzyme revealed a protein that is devoid of trigonal symmetry but shows two different types of pseudo-c2-symmetry [64]. The interaction of the subunits in the trimer is primarily mediated by the arrangement of the C-termini (which do not participate in the intrasubunit sequence similarity) in a triple helical motif (Fig. 5). The two folding domains of the E. coli subunit are related by ˚ , and the domains of the S. pombe enzyme an rmsd of 1.1 A ˚ (Fig. 4b). In each case, the are related by an rmsd of 0.97 A folding domains of a given subunit are related by pseudoc2-symmetry. A spurious c3 symmetry that had been proposed, incorrectly, on basis of a preliminary crystal
Fig. 5. Riboflavin synthase of E. coli (PDB entry #1I8D).
257
diffraction study was later explained by the fact that the three pseudo-c2-axes are related by angles of 124°, 85° and 151°; as already mentioned, the enzyme is really devoid of trigonal symmetry [64]. Last but not least, one specific N-terminal domain in the E. coli protein and the C-terminal domain of a second subunit are closely associated and form yet another pseudo-c2-symmetric pair (Fig. 5). As discussed in more detail below, this special pair forms the single active site of the protein, which is located at the domain interface and can impose on the two bound substrate molecules the quasi-c2-symmetric relationship that is required by the regiospecificity of the catalyzed reaction (cf. Fig. 6). The crystal structure of the E. coli enzyme was obtained without low molecular weight ligands, and arguments on catalysis had to remain speculative [64]. Nevertheless, a compelling model can be based on a combination of structural data obtained with riboflavin synthases of E. coli and S. pombe [51]. The crystal structure of the latter protein is clearly that of a monomer, despite the fact that the protein has been unequivocally shown to be trimeric in solution [62]. The monomer state observed in the crystal is believed to be induced by the organic solvent (methylpentane diol) used as a precipitant that is believed to break the hydrophobic interaction responsible for the cohesion of the Cterminal triple helix motif. On the other hand, the study with the S. pombe enzyme afforded a complex with the enzyme inhibitor 6-carboxyethyl-7-oxo-8-ribityllumazine (18) [70] (Fig. 7) that can be interpreted as a substrate analog. Both folding domains of the S. pombe enzyme bind that compound in shallow surface depressions with the ribityl side chains in extended conformations. Due to the closely similar folding topologies, the inhibitor molecules as observed in the monomeric S. pombe enzyme can be modeled into the trimeric E. coli enzyme in order to analyze the substrate topology at the putative single active site of that homotrimer. It was advantageous to omit the position 6 side chain of the inhibitor in order to avoid sterical hindrance (Fig. 8). As shown in Fig. 6, the modeled substrates come into close contact at the special pair of N-terminal domain of one subunit that forms an interface with the C-terminal domain of an adjacent subunit. The orientation of the modeled substrates correctly predicts the known regiochemistry of the reaction that
Fig. 6. Active site of riboflavin synthase of S. pombe formed by two adjacent monomers with bound 6-carboxyethyl-7-oxo-8-ribityllumazine (18). (a) The ligand bound to the N barrel (red) is drawn in yellow, whereas the 6-carboxyethyl-7-oxo-8-ribityllumazine in the adjacent C barrel (blue) is shown in dark yellow; (b) proposed binding of 6,7-dimethyl-8-ribityllumazine at the active site; (c) model for the pentacyclic reaction intermediate.
258
M. Fischer, A. Bacher / Archives of Biochemistry and Biophysics 474 (2008) 252–265
Fig. 7. Hydrogen-bonding topology of 6-carboxyethyl-7-oxo-8-ribityllumazine (18) bound to S. pombe riboflavin synthase (PDB entry #1KZL). (a) Nterminal domain; (b) C-terminal domain; (c) distances from the thiolate group of cysteine 48; (d) distances from the side chain oxygen of serine 146; R, ribityl chain. Reprinted from Gerhardt et al. [51], Copyright (2002) by Elsevier.
Fig. 8. X-Ray structures of riboflavin synthases from E. coli (PDB entry #1I8D) and S. pombe (PDB entry #1KZL). Left, Diagram of the superposition of one subunit of riboflavin synthase from E. coli (blue) and S. pombe (green); center, S. pombe riboflavin synthase monomer with bound 6-carboxyethyl-7oxo-8-ribityllumazine (18, yellow); right, trimeric model of S. pombe riboflavin synthase with bound 6-carboxyethyl-7-oxo-8-ribityllumazine (yellow). Reprinted from Gerhardt et al. [51], Copyright (2002) by Elsevier.
requires an antiparallel orientation. Moreover, the model shows that the ligands are in a virtually ideal position for the formation of the pentacyclic reaction intermediate (Figs. 3 and 6c). The topology leaves no doubt that the lumazine molecule originally bound to the N-terminal domain is the four-carbon acceptor arranged to end up as riboflavin and the substrate bound to the C-terminal domain is the four-carbon donor destined to end up as 6. As expected, the dimerization must result in a cis fusion of the two ring systems. It is even possible to predict the
absolute stereochemistry of the resulting pentacyclic moiety (cf. Fig. 3, compound Q) [51]. With special regard to the stereochemical features of the enzyme reaction, it is now in order to address the fact that an altogether different riboflavin synthase type is present in Archaea. An archaeal riboflavin synthase was first purified from the methanogenic bacterium Methanothermobacter thermoautotrophicus (Methanobacterium thermoautotrophicum) [71]. Later, the cognate gene was cloned by marker rescue and was found to encode a peptide of 153 amino
M. Fischer, A. Bacher / Archives of Biochemistry and Biophysics 474 (2008) 252–265
acid residues that was devoid of sequence similarity with the eubacterial riboflavin synthases described above [71]. Moreover, the M. thermoautotrophicus enzyme showed no intramolecular sequence similarity. However, sequence analysis identified the protein as a paralog of 6,7dimethyl-8-ribityllumazine synthase (Fig. 9a) [72]. Later, the enzyme from Methanocaldococcus jannaschii with similarity to riboflavin synthase of M. thermoautotrophicus was cloned and characterized. It was shown that this enzyme forms a pentamer in solution and in the crystal structure [73]. Moreover, the crystal structure revealed a c5-symmetric structure and a folding topology closely resembling that of 6,7-dimethyl-8-ribityllumazine synthases (Fig. 10). Studies with isotope-labeled substrate that were monitored by 13C NMR spectroscopy documented using the enzyme from M. jannaschii that the archaeal riboflavin syn-
259
thase obeys the same regiochemistry as observed without catalysis and under the catalytic influence of the homotrimeric riboflavin synthases from eubacteria [72]. Presteady state kinetic analysis suggested the formation of a pentacyclic intermediate Q0 that could be isolated in substance after rapid quenching of presteady state enzyme/substrate mixtures [53]. Interestingly, however, it turned out that the pentacyclic compound produced by the archaeal enzyme can only be converted by archaeal riboflavin synthase leading to two molecules of 9 (backward reaction) or one molecule each of riboflavin (10) and 6 (forward reaction) but not by riboflavin synthase of eubacterial origin. Vice versa, the pentacyclic intermediate produced by the E. coli could be catalytically converted by the E. coli but not by the archaeal enzyme [53]. These findings in conjunction with direct NMR studies leave no doubt
Fig. 9. Structural comparision of riboflavin synthase from M. jannaschii (PDB entry #2B99) and lumazine synthase from S. pombe (PDB entry #1KZL). (a) sequence alignment of riboflavin synthase from M. jannaschii (MjaRS) and lumazine synthase from S. pombe (SP-LS). Secondary structure elements (ahelices, rods; b -strands, arrows) corresponding to MJ-RS are in red, and the corresponding ones of SP-LS are in green. Active site residues are marked by asterisks in the corresponding colors; (b) structural alignment of the active sites formed by two adjacent monomers of SP-LS (dark green, residues Glu17 to Asp112; light green, residues Ser113 to Leu158) and MJ-RS (red, residues Thr2 to Met90; pink, residues Thr91 to Leu144). Bound riboflavin (green) in the case of SP-LS is in almost the same position as the bound inhibitor 6,7-dioxo-8-ribityllumazine A (red) in the case of MJ-RS. Secondary structure element labeling refers to MJ-RS; (c) 6,7-dioxo-8-ribityllumazine (19).
260
M. Fischer, A. Bacher / Archives of Biochemistry and Biophysics 474 (2008) 252–265
Fig. 10. Left: schematic ligand binding at the active sites of a pentameric riboflavin synthase; right: ribbon presentation of the M. jannaschii riboflavin synthase pentamer in complex with 6,7-dioxo-8-ribityllumazine (19, red) viewed along the 5-fold noncrystallographic symmetry axis (PDB entry #2B99). Individual subunits are shown in different colors.
whatsoever that the pentacyclic intermediates produced by the enzymes of E. coli and M. jannaschii are diastereomers. In both cases, the two lumazine moieties are connected with cis stereochemistry, and the heterocylic moieties, without consideration of the side chains would appear as enantiomers. However, due to the chiral properties of the ribityl side chains, which have D-configuration, the compounds Q and Q0 are diastereomers whose chemical shifts show minor but nevertheless significant differences [53]. In order to discuss catalysis by the pentameric, archaeal riboflavin synthase, it is necessary to address the paralogous lumazine synthases that have been studied in much more detail over a period of three decades, although detailed coverage of lumazine synthases is not the subject of this review. Briefly, lumazine synthases occur in three different quaternary structures, as pentamers, as icosahedral dodecamers of pentamers and, rarely, as c5-symmetric dimers of pentamers [65,67,74–85]. The icosahedral species form hollow capsids; in Bacillaceae, these capsids can contain a trimeric riboflavin synthase in the central core space [86] (Fig. 11). Invariably, the topologically equivalent active centers are located at the interfaces of adjacent subunits in the pentamer module. Inside the spacious active site cavity, the 5-amino group of 6 is believed to undergo an electrophilic attack by the carbonyl group of 3,4-dihydroxy-2-butanone 4-phosphate (8). The resulting Schiff base eliminates phosphate prior to ring closure. The active sites of the paralogous riboflavin synthase from Archaea have the same basic topology as that of lumazine synthase. X-ray structure analysis shows that two molecules of the inhibitor 6,7-dioxo-8-ribityllumazine (19) can be accommodated [73]. The topology of the two substrate analogs at the active site is in excellent agreement with the formation of the 6S/7R-form of the pentacyclic intermediate (Fig. 3). Sequence comparison indicates that the pentameric archaeal riboflavin synthases must have separated from the lumazine synthase branch at a very early time point in evolution.
Fig. 11. Computer generated model of a heavy riboflavin synthase complex. The capsid was generated using the coordinates from the lumazine synthase 60-mer of B. subtilis (PDB entry #1RVV) and from the riboflavin synthase trimer of E. coli (PDB entry #1I8D).
Apart from the stereochemical aspects, which have been described above, the expectation that protein structure analysis might clarify the details of the complex riboflavin synthase mechanism has been frustrated so far. A mutagenesis screen of the active site region of the trimeric E. coli enzyme failed to reveal any amino acid residues that might be involved in covalent catalysis [87]. However, the catalytic activity of the enzyme can be brought to an undetectably low level by alteration of the hydrophobic Nterminus. Since the N-terminal sequence motif MFTG is highly conserved over very large genetic distances, this is not particularly surprising. The most plausible function of the hydrophobic N-terminal residues consists in the accurate positioning of the acceptor substrate.
M. Fischer, A. Bacher / Archives of Biochemistry and Biophysics 474 (2008) 252–265
In the absence of significant guidance from structural biology, the pathway for the formation of the pentacyclic intermediates leaves ample room for speculation. All potential mechanisms should take into account that the reaction can take place in the complete absence of catalysts under relatively mild conditions (boiling aqueous solution). The fact that the non-enzymatic reaction and catalysis by both enzyme types invariably proceed with the same regiochemistry (although, as shown above, the stereochemical course is different for the two different enzyme classes) suggests that the non-enzymatic reaction proceeds via the same mechanism, although a formal proof for this is not available. Based on the unique CH acidity of the substrate 9, it has been proposed that a substrate molecule initially forms a covalent adduct by nucleophilic attack at C-7, either by a nucleophilic amino acid side chain or by a water molecule (Fig. 3) [88,89]. The resulting species was then proposed to be attacked by a second substrate molecule in the anion form. Following the discovery of the pentacyclic intermediates, it was not difficult to modify earlier mechanistic suggestions in order to accommodate the novel, experimentally documented intermediate. As mentioned above, the failure to identify, by systematic mutation studies, a nucleophilic amino acid side chain for the proposed C-7 adduct formation leaves only water as a plausible candidate for adduct formation. Some support for the covalent hydrate hypothesis can be deduced from the finding that the trifluoromethyl analogs of the substrate interact with trimeric riboflavin synthase in a strictly stereospecific manner; the diastereomer believed to have configuration 20 (epimer A) binds with relatively high affinity, whereas the diastereomer 21 (epimer B) shows no detectable binding (Fig. 12) [90]. Nevertheless, this cannot be considered as supporting evidence for a mechanistic role of a covalent hydrate. The observation that the uncatalyzed formation of riboflavin from 9 can proceed under acidic conditions does not precisely support the participation of an anionic species. It is however important to note that the pentacyclic intermediates have been rigorously shown to be kinetically competent intermediates of the
261
enzyme-catalyzed reaction. Last but not least, it should be noted that riboflavin synthase can process the pentacyclic intermediate in both directions, i.e. either cleavage under formation of two molecules of 6,7-dimethyl-8-ribityllumazine or, alternatively, one molecule each of riboflavin and 5-amino-6-ribitylamino-2,4(1H,3H)-pyrimidinedione (6). The N-terminal domain of E. coli riboflavin synthase can be expressed as a stable, recombinant protein [60]. The recombinant domains associate under formation of dimers with strict c2-symmetry (Fig. 13) [91,92]. The contact interface is different from the interfaces that mediate the intrasubunit contact and the intersubunit contact between N-terminal and C-terminal domains in the homotrimeric full-length protein. The recombinant N-terminal domain dimers bind riboflavin as well as 6,7-dimethyl-8ribityllumazine with relatively high affinity but are devoid of catalytic activity.
Fig. 13. The structure of the recombinant N-terminal domain dimer of riboflavin synthase from E. coli (PDB entry #1PKV) with bound riboflavin viewed along the 2-fold non-crystallographic symmetry axis.
Fig. 12. Fluorinated analogs of 6,7-bis(trifluoromethyl)-8-D-ribityllumazine. 20, Epimer A of 6,7-bis(trifluormethyl)-7-hydroxyl-8-ribityllumazine; 21, Epimer B of 6,7-bis(trifluormethyl)-7-hydroxyl-8-ribityllumazine; 22, 6-trifluoromethyl-7-oxo-8-ribityllumazine.
262
M. Fischer, A. Bacher / Archives of Biochemistry and Biophysics 474 (2008) 252–265
Fig. 14. Amino acid sequence similarity between the fluorescent protein of photobacteria and riboflavin synthase from S. pombe and E. coli. PL-LumP, lumazine protein of P. leiognathi (acc. no. Q06877); SP-RS, riboflavin synthase of S. pombe (acc. no. Q9Y7P0); EC-RS, riboflavin synthase of E. coli (acc. no. P0AFU8). Identical amino acid residues are in shadowed typeface.
Luminescent bacteria such as Photobacterium phosphoreum, Photobacterium leiognathi and certain Vibrio strains express a homotrimeric riboflavin synthase as well as proteins with sequence similarity to riboflavin synthase but devoid of catalytic activity [93,94]. These paralogs can bind various ligands such as 6,7-dimethyl-8-ribityllumazine, FMN and 6-methyl-7-oxo-8-ribityllumazine [5,6]. The complexes are highly fluorescent. Specifically, the fluorescence quantum yield of 6,7-dimethyl-8-ribityllumazine is increased by a factor of about 2 by complexation with lumazine protein from P. leiognathi [95]. The fluorescent proteins are all monomeric, in contrast to the paralogous, trimeric riboflavin synthase. The monomeric state is easily explained by the absence of the C-terminal sequence segment that plays the predominant role in the trimerisation of riboflavin synthase via triple helix formation (Fig. 14). It has been known for quite a while that the fluorescent proteins bind a single ligand molecule, despite the fact that sequence arguments (intramolecular sequence similarity and overall similarity to riboflavin synthases) predicts the presence of two similar folding domains in close analogy to trimeric riboflavin synthase [62,96,97]. Recently, the single binding site could be assigned to the N-terminal domain on basis of systematic mutagenesis studies in conjunction with NMR spectroscopy [98]. The luminescent proteins are all believed to play a role in light emission by the respective host organism. More specifically, energy is supposed to be transmitted from the excited state of bacterial luciferase to a luminescent protein by radiationless transition, and thus the fluorescent proteins may serve as an optical transponder that can modulate both the wavelength distribution and the quantum efficiency in that process [4,8,99,100]. Lumazine protein and the recombinant N-terminal homodimer domain of E. coli riboflavin synthase can both bind the bistrifluoromethyllumazine derivative 20 (Fig. 12)
with the same, strict diastereoselectivity as riboflavin synthase (only the A isomer is bound). Single 19F NMR signals are observed for each one of the methyl groups upon binding to either lumazine protein or recombinant N-terminal domain [60,62]. On the other hand, the compound in complex with riboflavin synthases from different organisms is characterized by complex 19F NMR spectra with multiple lines whose chemical shifts depend on ligand concentration. This is well in line with the absence of any strict symmetry in the homotrimeric native enzyme and has been interpreted tentatively as evidence for substantial conformational flexibility of the enzyme, although the details are not yet understood. A lumazine synthase/riboflavin synthase complex with an unusual quaternary structure has been observed in Bacillaceae and has been studied in more detail in B. subtilis [78,86,101–106]. The complex consists of a riboflavin synthase homotrimer enclosed in the central core space of the icosahedral lumazine synthase capsid (Fig. 11) [107]. The structural details of this complex are still unknown. Kinetic studies gave evidence of intermediate channelling; 6,7-dimethyl-8-ribityllumazine molecules that are generated by the lumazine synthase capsid are processed more rapidly by the resident riboflavin synthase module as compared to 6,7-dimethyl-8-ribityllumazine from the bulk solvent (Fig. 15) [108]. Confinement of the lumazine intermediate inside the icosahedral capsid has been proposed to be responsible for channeling, but the selective pressures that could have controlled the evolution of the complex structure remain unknown. Since several decades, it has been assumed that the enzymes of the riboflavin biosynthesis enzymes could be suitable targets for the development of novel antibacterial agents for the treatment of infections by Gram-negative bacteria and pathogenic yeasts. The hypothesis was based on the fact that riboflavin auxotrophs of E. coli and Salmo-
M. Fischer, A. Bacher / Archives of Biochemistry and Biophysics 474 (2008) 252–265
263
riboflavin synthase inhibitors has been reported recently and could serve as the basis for the screening libraries of drug-like compounds without apparent similarity with substrate and product structures that might have a better chance to penetrate the bacterial membrane barrier [121]. References
Fig. 15. Kinetic model of the reactions catalyzed by the heavy riboflavin synthase complex. a, a -subunits; b, b-subunits. Enzyme-bound ligands are shown in parentheses; ARP, 5-amino-6-ribitylamino-2,4(1H,3H)-pyrimidinedione (6), DHB, 3,4-dihydroxy-2-butanone 4-phosphate (8), LUM, 6,7-dimethyl-8-ribityllazine (9), RIB, riboflavin (10).
nella typhimurium require very high concentrations of exogenous riboflavin for growth, probably due to the absence of efficient uptake systems (by contrast, certain Gram-positive eubacteria can acquire riboflavin from the environment with outstanding efficacy). Recent studies have confirmed that riboflavin biosynthesis genes and proteins are virulence factors in S. enterica [109,110]. Indirect evidence further suggests that riboflavin biosynthesis enzymes could also be addressed as targets for novel drugs for tuberculosis and lepra. The argument is based on the comparison between the genomes of Mycobacterium tuberculosis and M. leprae [111]. The genome of the latter organism has passed through a very extensive process of gene fragmentation that has resulted in the destruction of about half of the gene endowment of typical Mycobacteriaceae. This process is believed to have accompanied the transition of M. leprae to a strictly intracellular lifestyle. Interestingly, the extensive gene fragmentation has consistently spared the genes of the riboflavin pathway, thus suggesting that this pathway is of vital importance even for a strictly intracellular lifestyle. Based on this hypothesis, extensive work on the synthesis of substrate analogs has resulted in a number of potent riboflavin synthase inhibitors [70,90,112–120]. Unfortunately, none of the known inhibitors shows in vivo activity against bacteria, probably because they all fail to enter the bacterial cells. A high throughput screening method for
[1] F. Mu¨ller (Ed.), Chemistry and Biochemistry of Flavoenzymes, vol. 2, CRC Press, Boca Raton, FL, 1991, p. 506. [2] B.A. Palfey, V. Massey, in: M. Sinnot (Ed.), Comprehensive Biological Catalysis, Academic Press, San Diego, 1998, pp. 83–154. [3] S. Bornemann, Nat. Prod. Rep. 19 (2002) 761–772. [4] J. Lee, Biophys. Chem. 48 (1993) 149–158. [5] V.N. Petushkov, J. Lee, Eur. J. Biochem. 245 (1997) 790–796. [6] P. Macheroux, K.U. Schmidt, P. Steinerstauch, S. Ghisla, P. Colepicolo, R. Buntic, J.W. Hastings, Biochem. Biophys. Res. Commun. 146 (1987) 101–106. [7] J. Lee, Biochemistry 11 (1972) 3350–3359. [8] J. Lee, I.B.C. Matheson, F. Mu¨ller, D.J. O’Kane, J. Vervoort, A.J.W.G. Visser, in: F. Mu¨ller (Ed.), Chem Biochem Flavoenzymes, vol. 2, CRC Press, Florida, Boca Raton, 1991, 109–151. [9] D.J. O’Kane, J. Lee, Biochemistry 24 (1985) 1484–1488. [10] V.W. Cheng, E. Ma, Z. Zhao, R.A. Rothery, J.H. Weiner, J. Biol. Chem. 281 (2006) 27662–27668. [11] J.M. Hudson, K. Heffron, V. Kotlyar, Y. Sher, E. Maklashina, G. Cecchini, F.A. Armstrong, J. Am. Chem. Soc. 127 (2005) 6977–6989. [12] G.A. Roberts, A. Celik, D.J. Hunter, T.W. Ost, J.H. White, S.K. Chapman, N.J. Turner, S.L. Flitsch, J. Biol. Chem. 278 (2003) 48914–48920. [13] H. O’Neill, S.G. Mayhew, G. Butler, J. Biol. Chem. 273 (1998) 21015–21024. [14] M. Mewies, W.S. McIntire, N.S. Scrutton, Protein Sci. 7 (1998) 7– 20. [15] W.R. Briggs, C.F. Beck, A.R. Cashmore, J.M. Christie, J. Hughes, J.A. Jarillo, T. Kagawa, H. Kanegae, E. Liscum, A. Nagatani, K. Okada, M. Salomon, W. Rudiger, T. Sakai, M. Takano, M. Wada, J.C. Watson, Plant Cell 13 (2001) 993–997. [16] T. Kagawa, J. Plant Res. 116 (2003) 77–82. [17] J.M. Christie, M. Salomon, K. Nozue, M. Wada, W.R. Briggs, Proc. Natl. Acad. Sci. USA 96 (1999) 8779–8783. [18] J.M. Christie, T.E. Swartz, R.A. Bogomolni, W.R. Briggs, Plant J. 32 (2002) 205–219. [19] S. Crosson, K. Moffat, Proc. Natl. Acad. Sci. USA 98 (2001) 2995– 3000. [20] S. Crosson, K. Moffat, Plant Cell 14 (2002) 1067–1075. [21] S. Crosson, S. Rajagopal, K. Moffat, Biochemistry 42 (2003) 2–10. [22] M. Salomon, W. Eisenreich, H. Du¨rr, E. Schleicher, E. Knieb, V. Massey, W. Ru¨diger, F. Mu¨ller, A. Bacher, G. Richter, Proc. Natl. Acad. Sci. USA 98 (2001) 12357–12361. [23] A. Sancar, Annu. Rev. Biochem. 69 (2000) 31–67. [24] T. Imaizumi, H.G. Tran, T.E. Swartz, W.R. Briggs, S.A. Kay, Nature 426 (2003) 302–306. [25] E. Silva, A.M. Edwards, in: D.P. Ha¨der, G. Jori (Eds.), Comprehensive Series in Photochemistry and Photobiology, vol. 6, The Royal Society of Chemistry, Cambridge, 2006, 328. [26] K. Kis, K. Kugelbrey, A. Bacher, J. Org. Chem. 66 (2001) 2555– 2559. [27] A. Eschenmoser, E. Loewenthal, Chem. Soc. Rev. 21 (1992) 1–16. [28] C.J. Strupp, Untersuchungen u¨ber die nicht-enzymatische Simulation des Biosyntheseweges zu Riboflavin. ETH Thesis Nr. 9832, Zu¨rich 1992. [29] D.H.G. Crout, J.A. Hadfield, J. Chem. Soc. Chem. Com. 10 (1987) 742–744. [30] R.L. Beach, G.W.E. Plaut, Tetrahedron Lett. 40 (1969) 3489–3492. [31] T. Paterson, H.C.S. Wood, J. Chem. Soc. Commun. (1969) 290–291.
264
M. Fischer, A. Bacher / Archives of Biochemistry and Biophysics 474 (2008) 252–265
[32] T. Rowan, H.C. Wood, J. Chem. Soc., Perkin Trans. 1 4 (1968) 452– 458. [33] T. Rowan, H.C.S. Wood, Proc. Chem. Soc. (1963) 21–22. [34] M. Fischer, A. Bacher, Nat. Prod. Rep. 22 (2005) 324–350. [35] M. Fischer, A. Bacher, Physiol. Plant. 126 (2006) 304–318. [36] A. Bacher, S. Eberhardt, W. Eisenreich, M. Fischer, S. Herz, B. Illarionov, K. Kis, G. Richter, Vitam. Horm. 61 (2001) 1–49. [37] A. Bacher, S. Eberhardt, M. Fischer, K. Kis, G. Richter, Annu. Rev. Nutr. 20 (2000) 153–167. [38] A. Bacher, S. Eberhardt, G. Richter, in: F.C. Neidhardt et al. (Eds.), Escherichia coli and Salmonella, ASM Press, Washington, DC, 1996, pp. 657-664. [39] L. Chatwell, T. Krojer, A. Fidler, W. Ro¨misch, W. Eisenreich, A. Bacher, R. Huber, M. Fischer, J. Mol. Biol. 359 (2006) 1334–1351. [40] M. Graupner, H. Xu, R.H. White, J. Bacteriol. 184 (2002) 1952–1957. [41] M. Fischer, W. Ro¨misch, S. Saller, B. Illarionov, G. Richter, F. Rohdich, W. Eisenreich, A. Bacher, J. Biol. Chem. 279 (2004) 36299–36308. [42] G. Richter, M. Fischer, C. Krieger, S. Eberhardt, H. Lu¨ttgen, I. Gerstenschla¨ger, A. Bacher, J. Bacteriol. 179 (1997) 2022–2028. [43] S.C. Chen, Y.C. Chang, C.H. Lin, C.H. Lin, S.H. Liaw, J. Biol. Chem. 281 (2006) 7605–7613. [44] P. Stenmark, M. Moche, D. Gurmu, P. Nordlund, J. Mol. Biol. 373 (2007) 48–64. [45] R.L. Beach, G.W.E. Plaut, J. Am. Chem. Soc. 92 (1970) 2913–2916. [46] W. Pfleiderer, W. Hutzenlaub, Chemische Berichte 106 (1973) 3149– 3174. [47] R. Stewart, J.M. McAndless, J. Chem. Soc. Perkin Trans. 2 (1972) 376–379. [48] R.L. Beach, G.W.E. Plaut, J. Org. Chem. 36 (1971) 3937–3943. [49] D.H. Bown, P.J. Keller, H.G. Floss, H. Sedlmaier, A. Bacher, J. Org. Chem. 51 (1986) 2461–2467. [50] W. Pfleiderer, J.W. Bunting, D.D. Perrin, G. Nu¨bel, Chemische Berichte 99 (1966) 3503–3523. [51] S. Gerhardt, A.K. Schott, N. Kairies, M. Cushman, B. Illarionov, W. Eisenreich, A. Bacher, R. Huber, S. Steinbacher, M. Fischer, Structure 10 (2002) 1371–1381. [52] B. Illarionov, W. Eisenreich, A. Bacher, Proc. Natl. Acad. Sci. USA 98 (2001) 7224–7229. [53] B. Illarionov, W. Eisenreich, N. Schramek, A. Bacher, M. Fischer, J. Biol. Chem. 280 (2005) 28541–28546. [54] B. Illarionov, I. Haase, M. Fischer, A. Bacher, N. Schramek, Biol. Chem. 386 (2005) 127–136. [55] S. Kuwada, T. Masuda, T. Kishi, M. Asai, J. Vitaminol. (Kyoto) 4 (1958) 217–225. [56] G.F. Maley, G.W. Plaut, J. Biol. Chem. 243 (1959) 641–647. [57] T. Masuda, Pharm. Bull. 4 (1956) 71–72. [58] T. Masuda, Pharm. Bull. 5 (1957) 136–141. [59] S. Eberhardt, G. Richter, W. Gimbel, T. Werner, A. Bacher, Eur. J. Biochem. 242 (1996) 712–719. [60] S. Eberhardt, N. Zingler, K. Kemter, G. Richter, M. Cushman, A. Bacher, Eur. J. Biochem. 268 (2001) 4315–4323. [61] M. Fischer, I. Haase, R. Feicht, N. Schramek, P. Ko¨hler, P. Schieberle, A. Bacher, Biol. Chem. 386 (2005) 417–428. [62] M. Fischer, A.K. Schott, K. Kemter, R. Feicht, G. Richter, B. Illarionov, W. Eisenreich, S. Gerhardt, M. Cushman, S. Steinbacher, R. Huber, A. Bacher, BMC Biochem. 4 (2003) 18. [63] R.A. Harvey, G.W. Plaut, J. Biol. Chem. 241 (1966) 2120–2136. [64] D.I. Liao, Z. Wawrzak, J.C. Calabrese, P.V. Viitanen, D.B. Jordan, Structure (Camb.) 9 (2001) 399–408. [65] W. Meining, S. Mo¨rtl, M. Fischer, M. Cushman, A. Bacher, R. Ladenstein, J. Mol. Biol. 299 (2000) 181–197. [66] W. Meining, G. Tibbelin, R. Ladenstein, S. Eberhardt, M. Fischer, A. Bacher, J. Struct. Biol. 121 (1998) 53–60. [67] S. Mo¨rtl, M. Fischer, G. Richter, J. Tack, S. Weinkauf, A. Bacher, J. Biol. Chem. 271 (1996) 33201–33207. [68] G.W. Plaut, R.L. Beach, T. Aogaichi, Biochemistry 9 (1970) 771– 785.
[69] K. Schott, J. Kellermann, F. Lottspeich, A. Bacher, J. Biol. Chem. 265 (1990) 4204–4209. [70] M. Cushman, D. Yang, S. Gerhardt, R. Huber, M. Fischer, K. Kis, A. Bacher, J. Org. Chem. 67 (2002) 5807–5816. [71] S. Eberhardt, S. Korn, F. Lottspeich, A. Bacher, J. Bacteriol. 179 (1997) 2938–2943. [72] M. Fischer, A.K. Schott, W. Ro¨misch, A. Ramsperger, M. Augustin, A. Fidler, A. Bacher, G. Richter, R. Huber, W. Eisenreich, J. Mol. Biol. 343 (2004) 267–278. [73] A. Ramsperger, M. Augustin, A.K. Schott, S. Gerhardt, T. Krojer, W. Eisenreich, B. Illarionov, M. Cushman, A. Bacher, R. Huber, M. Fischer, J. Biol. Chem. 281 (2006) 1224–1232. [74] M. Fischer, I. Haase, R. Feicht, G. Richter, S. Gerhardt, J.P. Changeux, R. Huber, A. Bacher, Eur. J. Biochem. 269 (2002) 519– 526. [75] S. Gerhardt, I. Haase, S. Steinbacher, J.T. Kaiser, M. Cushman, A. Bacher, R. Huber, M. Fischer, J. Mol. Biol. 318 (2002) 1317–1329. [76] S. Klinke, V. Zylberman, H.R. Bonomi, I. Haase, B.G. Guimaraes, B.C. Braden, A. Bacher, M. Fischer, F.A. Goldbaum, J. Mol. Biol. 373 (2007) 664–680. [77] R. Ladenstein, K. Ritsert, R. Huber, G. Richter, A. Bacher, Eur. J. Biochem. 223 (1994) 1007–1017. [78] R. Ladenstein, M. Schneider, R. Huber, H.D. Bartunik, K. Wilson, K. Schott, A. Bacher, J. Mol. Biol. 203 (1988) 1045–1070. [79] E. Morgunova, W. Meining, B. Illarionov, I. Haase, G. Jin, A. Bacher, M. Cushman, M. Fischer, R. Ladenstein, Biochemistry 44 (2005) 2746–2758. [80] E. Morgunova, S. Saller, I. Haase, M. Cushman, A. Bacher, M. Fischer, R. Ladenstein, J. Biol. Chem. 282 (2007) 17231–17241. [81] K. Persson, G. Schneider, B.J. Douglas, P.V. Viitanen, T. Sandalova, Prot. Sci. 8 (1999) 2355–2365. [82] K. Ritsert, R. Huber, D. Turk, R. Ladenstein, K. Schmidt-Ba¨se, A. Bacher, J. Mol. Biol. 253 (1995) 151–167. [83] X. Zhang, W. Meining, M. Fischer, A. Bacher, R. Ladenstein, J. Mol. Biol. 306 (2001) 1099–1114. [84] V. Zylberman, P.O. Craig, S. Klinke, B.C. Braden, A. Cauerhff, F.A. Goldbaum, J. Biol. Chem. 279 (2004) 8093–8101. [85] I. Haase, S. Mo¨rtl, P. Kohler, A. Bacher, M. Fischer, Eur. J. Biochem. 270 (2003) 1025–1032. [86] A. Bacher, H. Schnepple, B. Maila¨nder, M.K. Otto, Y. Ben-Shaul, Flavins Flavoproteins, Proceedings of the 6th International Symposium (1980) 579–586. [87] B. Illarionov, K. Kemter, S. Eberhardt, G. Richter, M. Cushman, A. Bacher, J. Biol. Chem. 276 (2001) 11524–11530. [88] G.W. Plaut, R.L. Beach, in: W. Pfleiderer (Ed.), Chemistry and Biology of Pteridines, Proceedings of the International Symposium, 5th, de Gruyter, Berlin, 1975, pp. 101-124. [89] G.W.E. Plaut, R.L. Beach, in: T.P. Singer (Ed.), Flavins and Flavoproteins, Proceedings of the 5th International Symposium, Elsevier, Amsterdam, 1976, pp. 737–746. [90] M. Cushman, D.A. Patrick, A. Bacher, J. Scheuring, J. Org. Chem. 56 (1991) 4603–4608. [91] W. Meining, S. Eberhardt, A. Bacher, R. Ladenstein, J. Mol. Biol. 331 (2003) 1053–1063. [92] V. Truffault, M. Coles, T. Diercks, K. Abelmann, S. Eberhardt, H. Lu¨ttgen, A. Bacher, H. Kessler, J. Mol. Biol. 309 (2001) 949–960. [93] J. Lee, P. Koka, Methods Enzymol. 57 (1978) 226–234. [94] E.A. Meighen, FASEB J. 7 (1993) 1016–1022. [95] D.J. O’Kane, J. Lee, Methods Enzymol. 133 (1986) 149–172. [96] J. Scheuring, M. Fischer, M. Cushman, J. Lee, A. Bacher, H. Oschkinat, Biochemistry 35 (1996) 9637–9646. [97] J. Scheuring, J. Lee, M. Cushman, H. Patel, D.A. Patrick, A. Bacher, Biochemistry 33 (1994) 7634–7640. [98] B. Illarionov, W. Eisenreich, M. Wirth, C.L. Lee, Y.E. Woo, A. Bacher, M. Fischer, Biol. Chem. 388 (2007) 1313–1323. [99] J. Lee, D.J. O’Kane, A.J. Visser, Biochemistry 24 (1985) 1476–1483. [100] J. Lee, in: J.G. Burr (Ed.), Chemiluminescence and Bioluminescence, Marcel Dekker Inc., New York, 1985, 401-437.
M. Fischer, A. Bacher / Archives of Biochemistry and Biophysics 474 (2008) 252–265 [101] A. Bacher, S. Eberhardt, M. Fischer, S. Mo¨rtl, K. Kis, K. Kugelbrey, J. Scheuring, K. Schott, Methods Enzymol. 280 (1997) 389–399. [102] A. Bacher, B. Maila¨nder, J. Bacteriol. 134 (1978) 476–482. [103] A. Bacher, R. Baur, U. Eggers, H.D. Harders, M.K. Otto, H. Schnepple, J. Biol. Chem. 255 (1980) 632–637. [104] R. Ladenstein, H.C. Ludwig, J. Biol. Chem. 258 (1983) 11981– 11983. [105] A. Bacher, H.C. Ludwig, H. Schnepple, Y. Ben-Shaul, J. Mol. Biol. 187 (1986) 75–86. [106] R. Ladenstein, B. Meyer, R. Huber, H. Labischinski, K. Bartels, H.D. Bartunik, L. Bachmann, H.C. Ludwig, A. Bacher, J. Mol. Biol. 187 (1986) 87–100. [107] A. Bacher, R. Baur, U. Eggers, H. Harders, H. Schnepple, Flavins Flavoproteins, in: T.P. Singer (Ed.), Proceedings of the 5th International Symposium (1976) 729-732. [108] K. Kis, A. Bacher, J. Biol. Chem. 270 (1995) 16788–16795. [109] C. Rollenhagen, D. Bumann, Infect. Immun. 74 (2006) 1649–1660. [110] D. Becker, M. Selbach, C. Rollenhagen, M. Ballmaier, T.F. Meyer, M. Mann, D. Bumann, Nature 440 (2006) 303–307. [111] S.T. Cole, K. Eiglmeier, J. Parkhill, K.D. James, N.R. Thomson, P.R. Wheeler, N. Honore, T. Garnier, C. Churcher, D. Harris, K. Mungall, D. Basham, D. Brown, T. Chillingworth, R. Connor, R.M. Davies, K. Devlin, S. Duthoy, T. Feltwell, A. Fraser, N. Hamlin, S. Holroyd, T. Hornsby, K. Jagels, C. Lacroix, J. Maclean, S. Moule, L. Murphy, K. Oliver, M.A. Quail, M.A. Rajandream, K.M. Rutherford, S. Rutter, K. Seeger, S. Simon, M. Simmonds, J.
[112] [113] [114] [115] [116] [117] [118] [119]
[120] [121]
[122]
265
Skelton, R. Squares, S. Squares, K. Stevens, K. Taylor, S. Whitehead, J.R. Woodward, B.G. Barrell, Tuberculosis 81 (2001) 157–163. J. Chen, T. Sambaiah, B. Illarionov, M. Fischer, A. Bacher, M. Cushman, J. Org. Chem. 69 (2004) 6996–7003. M. Cushman, G. Jin, T. Sambaiah, B. Illarionov, M. Fischer, R. Ladenstein, A. Bacher, J. Org. Chem. 70 (2005) 8162–8170. M. Cushman, F. Mavandadi, K. Kugelbrey, A. Bacher, Bioorg. Med. Chem. 6 (1998) 409–415. M. Cushman, F. Mavandadi, D. Yang, K. Kugelbrey, K. Kis, A. Bacher, J. Org.Chem. 64 (1999) 4635–4642. M. Cushman, J.T. Mihalic, K. Kis, A. Bacher, Bioorg. Med. Chem. Lett. 9 (1999) 39–42. M. Cushman, T. Sambaiah, G. Jin, B. Illarionov, M. Fischer, A. Bacher, J. Org. Chem. 69 (2004) 601–612. M. Cushman, D. Yang, K. Kis, A. Bacher, J. Org. Chem. 66 (2001) 8320–8327. M. Cushman, D. Yang, J.T. Mihalic, J. Chen, S. Gerhardt, R. Huber, M. Fischer, K. Kis, A. Bacher, J. Org. Chem. 67 (2002) 6871–6877. A. Talukdar, B. Illarionov, A. Bacher, M. Fischer, M. Cushman, J. Org. Chem. 72 (2007) 7167–7175. J. Kaiser, B. Illarionov, F. Rohdich, W. Eisenreich, S. Saller, J. Van den Brulle, M. Cushman, A. Bacher, M. Fischer, Anal. Biochem. 365 (2007) 52–61. M.A. Santos, J.J. Garcı´a-Ramı´rez, J.L. Revuelta, J. Biol. Chem. 270 (1995) 437–444.
ABB Archives of Biochemistry and Biophysics 474 (2008) 252–265 www.elsevier.com/locate/yabbi
Review
Biosynthesis of vitamin B2: Structure and mechanism of riboflavin synthase Markus Fischer a,*, Adelbert Bacher b b
a Institute of Food Chemistry, University of Hamburg, Grindelallee 117, D-20146 Hamburg, Germany Lehrstuhl fu¨r Biochemie, Technische Universita¨t Mu¨nchen, Lichtenbergstr. 4, D-85747 Garching, Germany
Received 19 December 2007, and in revised form 5 February 2008 Available online 14 February 2008
Abstract The biosynthesis of one riboflavin molecule requires one molecule of GTP and two molecules of ribulose 5-phosphate as substrates. GTP is hydrolytically opened, converted into 5-amino-6-ribitylamino-2,4(1H,3H)-pyrimidinedione by a sequence of deamination, side chain reduction and dephosphorylation. Condensation with 3,4-dihydroxy-2-butanone 4-phosphate obtained from ribulose 5-phosphate leads to 6,7-dimethyl-8-ribityllumazine. The final step in the biosynthesis of the vitamin involves the dismutation of 6,7-dimethyl-8-ribityllumazine catalyzed by riboflavin synthase. The mechanistically unusual reaction involves the transfer of a four-carbon fragment between two identical substrate molecules. The second product, 5-amino-6-ribitylamino-2,4(1H,3H)-pyrimidinedione, is recycled in the biosynthetic pathway by 6,7-dimethyl-8-ribityllumazine synthase. This article will review structures and reaction mechanisms of riboflavin synthases and related proteins up to 2007 and 122 references are cited. Ó 2008 Elsevier Inc. All rights reserved. Keywords: Vitamin B2; Riboflavin synthase; Lumazine synthase; Lumazine protein; X-ray structure; Reaction mechanism
In association with a wide variety of apoenzymes, the flavocoenzymes derived from vitamin B2 (riboflavin) are probably the most chemically versatile cofactors. They have long been known to enable a wide variety of redox reactions involving one- as well as two-electron transfer processes, whereas other coenzymes are typically limited to one of them (for example, pyridine nucleotides catalyze exclusively two-electron transfers) [1,2]. Moreover, they are involved in an impressive number of reactions that do not involve a net exchange of electrons with the respective substrate. For a comprehensive description of flavoenzymes that catalyze reactions with no net redox change the reader is directed to the review by Bornemann [3]. Flavins serve as primary and secondary emitters in bacterial luminescence. Lumazine proteins, which can bind riboflavin, FMN, 6,7-dimethyl-8-ribityllumazine or 6methyl-7-oxo-8-ribityllumazine, are believed to act as opti*
Corresponding author. Fax: +49 40428384342. E-mail address: markus.fi[email protected] (M. Fischer).
0003-9861/$ - see front matter Ó 2008 Elsevier Inc. All rights reserved. doi:10.1016/j.abb.2008.02.008
cal transponders in the bioluminescence emission of photobacteria [4–8]. The interposition of lumazine protein as an optical transponder between the energy generation and light emission steps has been proposed to result in shifts of the emission maximum and an increased quantum yield [9]. This extraordinary photochemical functionality reflects the enormous versatility of the isoalloxazine chromophore, whereas the side chains of the flavocoenzymes serve primarily as anchors that secure their binding to the cognate apoproteins, which typically form relatively tight complexes. Moreover, flavocoenzymes can form very complex catalytic sites involving more than one flavocoenzyme, modified flavins and/or additional cofactors such as iron sulfur clusters [10–13]. In some cases, flavocoenzymes become covalently linked to their apoenzymes [14]. Photoreceptors involved in processes like stem bending toward a light source (phototropism), chloroplast migration to places of appropriate light intensity (chloroplast photorelocation), and the
M. Fischer, A. Bacher / Archives of Biochemistry and Biophysics 474 (2008) 252–265
253
Fig. 1. Biosynthesis of riboflavin and flavocoenzymes. Step I, GTP cyclohydrolase II; step II, 2,5-diamino-6-ribosylamino-4(3H)-pyrimidinone 50 phosphate deaminase; step III, 5-amino-6-ribosylamino-2,4(1H,3H)-pyrimidinedione 50 -phosphate reductase; step IV, 2,5-diamino-6-ribosylamino-4(3H)pyrimidinone 50 -phosphate reductase; step V, 2,5-diamino-6-ribitylamino-4(3H)-pyrimidinone 50 -phosphate deaminase; step VI, hypothetical phosphatase; step VII, 3,4-dihydroxy-2-butanone 4-phosphate synthase; step VIII, 6,7-dimethyl-8-ribityllumazine synthase; step IX, riboflavin synthase; step X, riboflavin kinase; step XI, FAD synthetase; 1, GTP; 2, 2,5-diamino-6-ribosylamino-4(3H)-pyrimidinone 50 -phosphate; 3, 5-amino-6-ribosylamino2,4(1H,3H)-pyrimidinedione 50 -phosphate; 4, 2,5-diamino-6-ribitylamino-4(3H)-pyrimidinone 50 -phosphate; 5, 5-amino-6-ribitylamino-2,4(1H,3H)pyrimidinedione 50 -phosphate; 6, 5-amino-6-ribitylamino-2,4(1H,3H)-pyrimidinedione; 7, ribulose 50 -phosphate; 8, 3,4-dihydroxy-2-butanone 4phosphate; 9, 6,7-dimethyl-8-ribityllumazine; 10, riboflavin; 11, FMN; 12, FAD.
opening of stomatal guard cells to facilitate gas exchange are the plasma membrane-associated phototropins and homologs [15–21]. The common features of the phototropin photoreceptor family are two light-sensitive domains, LOV1 and LOV2, and a serine/threonine kinase domain. Each of the two LOV domains non-covalently binds a single flavin mononucleotide (FMN)1 as a chromophore. After illumination by blue light, the LOV photocycle comprises a light-induced addition of a thiol group to the C(4a) position of the FMN chromophore [20,22]. Flavins have also been shown to be involved in circadian and seasonal clocks [23–25]. The mechanistic complexity of flavoprotein catalysis is to some extent matched by the complexity of some of the reactions involved in the biosynthesis of the isoalloxazine moiety of riboflavin. Surprisingly, some of these mechanistically complex reactions can proceed without catalysis under relatively mild conditions [26–33]. Thus, it is conceiv-
1 Abbreviations used: FMN, flavin mononucleotide; MjaRS, M. jannaschii.
able that the formation of flavin could initially have occurred by spontaneous processes, prior to the evolution of macromolecular biocatalysis. The biosynthesis of riboflavin has been reviewed repeatedly and the reader is directed to these articles for an overview of the pathway (Fig. 1) [34–38]. This article will focus on the last step in the biosynthetic pathway, which continues to present fascinating mechanistic problems even after more than four decades of research. Briefly, the initial steps of the riboflavin pathway involve the conversion of GTP (1) into 5-amino-6-ribitylamino2,4(1H,3H)-pyrimidinedione (6). The intermediate is obtained via structurally different precursors in different organisms. More specifically, the first committed pathway intermediate, 2,5-diamino-6-ribosylamino-4(3H)-pyrimidinone 50 -phosphate (2), is reduced and then deaminated in fungi and in Archaea [39,40], whereas reduction of the side chain precedes deamination in eubacteria and plants [41–44]. It is still unknown how the phosphate residue in 5-amino-6-ribitylamino-2,4(1H,3H)-pyrimidinedione 50 phosphate (5) is released to generate 5-amino-6-ribitylamino-2,4(1H,3H)-pyrimidinedione (6). It is clear, however,
254
M. Fischer, A. Bacher / Archives of Biochemistry and Biophysics 474 (2008) 252–265
Fig. 2. Anionic species of 6,7-dimethyl-8-ribityllumazine and 6,7,8-trimethyllumazine.
that dephosphorylation is mandatory since the phosphate ester cannot serve as a substrate for the consecutive enzyme reaction catalyzed by 6,7-dimethyl-8-ribityllumazine synthase (VIII), which catalyzes the condensation of 6 with 3,4-dihydroxy-2-butanone 4-phosphate (8) that is obtained by a complex rearrangement/elimination sequence from ribulose 5-phosphate (7). The final step in the biosynthesis of riboflavin involves a most unusual dismutation. In a formal sense, two molecules of 6,7-dimethyl-8-ribityllumazine (9) exchange a four-carbon unit, thus transforming one of them into riboflavin and the other into 5-amino-6-ribitylamino2,4(1H,3H)-pyrimidinedione (6). The reaction is catalyzed by riboflavin synthase (IX) but can also proceed without catalysis in boiling aqueous solution at neutral or acidic pH [30,32,33,45]. The second product of the dismutation, 6, can again serve as substrate for lumazine synthase (VIII). The xylene ring of riboflavin (10) is generated from two molar equivalents of the carbohydrate precursor, 3,4-dihydroxy-2-butanone 4-phosphate (8). Since only one four-carbon unit can be condensed with the pyrimidine precursor 6 by lumazine synthase, the reaction stoichiometry requires that the four-carbon unit is exposed twice to the catalytic action of lumazine synthase. However, the random nature of that process deserves some emphasis. In closer detail, one half of the GTP molecules that enter the pathway make it all the way through to riboflavin in their first passage. The others must pass through at least one additional cycle of consumption/regeneration under the catalytic influence of lumazine synthase and riboflavin synthase. In fact, a fraction of pyrimidine molecules (6) will complete several cycles of consumption/regeneration before reaching the level of riboflavin (10). Formally, 6,7-dimethyl-8-ribityllumazine can be interpreted as a tetra-nor analog of riboflavin, and the chemical similarity between the two compounds is indeed quite close. Both show strong absorption of visible light and both are fluorescent with high quantum yields. Both can be easily reduced to the respective dihydro forms, and the 50 -phosphate of 6,7-dimethyl-8-ribityllumazine can serve as a catalytically competent flavocoenzyme analog with certain apoenzymes. Notably, the properties of the 7a methyl group of 6,7dimethyl-8-ribityllumazine and the vinylogous 8a methyl group of riboflavin differ dramatically in their acidity.
The 8a methyl group has a somewhat enhanced acidity, which enables selective deuteration, albeit under drastic reaction conditions. On the other hand, the 7a methyl group of 6,7-dimethyl-8-ribityllumazine and of other 8substituted 7-methyl lumazines represent extreme cases of CH acidity. Specifically, the pK for the dissociation of a proton from the 7a methyl group is about 8 for 6,7dimethyl-8-ribityllumazine [46] and about 10 for 6,7,8-trimethyl-8-ribityllumazine [47]. In the case of compounds carrying appropriately placed hydroxyl groups at the position 8 side chain, the exomethylene anion resulting from deprotonation is subject to nucleophilic attack of C-7 of the lumazine ring system; in the case of 6,7-dimethyl-8-ribityllumazine, the exomethylene anion species accounts only for a small percentage of the anion mixture, which is dominated by four isomeric tricyclic anion species in rapid equilibrium (Fig. 2) [48–50]. Recent studies have identified a covalent adduct of two 6,7-dimethyl-8-ribityllumazine molecules (9) as a key intermediate (Fig. 3, compound Q) in the enzymatic formation of riboflavin [51–54]. Not surprisingly, the fusion of the ring systems occurs in cis [51]. Surprisingly, however, the structurally unrelated riboflavin synthases of Archaea, on the one hand, and of plants, fungi and eubacteria, on the other, proceed via diastereomeric forms of the pentacyclic intermediate (Fig. 3, compound Q0 and compound Q, resp.) (the C-atoms 6 and 7 of the pentacyclic intermediate, without reference to the chiral centers of the ribityl side chains, are of course enantiomeric). It is obvious that a sequence of two elimination reactions can convert each of the two different pentacyclic intermediates into equimolar amounts of riboflavin (10) and 5-amino-6-ribitylamino-2,4(1H,3H)-pyrimidinedione (6). Each of the two different riboflavin synthases types is limited to cleaving its cognate reaction intermediate but not that produced by enzymes of the other type, whereas both diasteromeric pentacycles can afford the product mixture by uncatalyzed fragmentation. On the other hand, the formation of the covalent pentacyclic adduct by dimerization of two 6,7-dimethyl-8-ribityllumazine remains mechanistically enigmatic, even more so since it can obviously proceed in the absence of a catalyst. Although pentacyclic intermediates have only been documented experimentally for the enzyme-catalyzed process, it appears rather safe to assume that they are also involved
M. Fischer, A. Bacher / Archives of Biochemistry and Biophysics 474 (2008) 252–265
255
Fig. 3. Stereochemistry of 6,7-dimethyl-8-ribityllumazine conversion into riboflavin catalyzed by trimeric eubacterial and pentameric archaeal riboflavin synthase. Binding of the substrates in anti-parallel orientation occurs at two sites, one leading to acceptance of a four-carbon unit (9a, acceptor site) and the other to its donation (9b, donor site); Q and Q0 , pentacyclic reaction intermediates. R, ribityl.
256
M. Fischer, A. Bacher / Archives of Biochemistry and Biophysics 474 (2008) 252–265
in the case of the uncatalyzed reaction, which has been shown to proceed with exactly the same, strict regiospecificity as the enzyme-catalyzed process. Notably, both the uncatalyzed and the enzyme-catalyzed processes require a quasi-c2-symmetric arrangement of the substrates in the transition state for the dimerization. We will return to this point after an introduction to the structures and properties of riboflavin synthases. Riboflavin synthase activity was first observed in cell extracts of naturally flavinogenic ascomycetes [55–58]. Later, the enzymes from the yeasts Saccharomyces cerevisiae and Schizosaccharomyces pombe, the eubacteria Bacillus subtilis and Escherichia coli and from the plant, Arabidopsis thaliana, have been studied in some detail [51,59–68,122]. These enzymes are all homotrimers. However, as described below, the riboflavin synthases from Archaea are homopentamers with completely unrelated amino acid sequences. The subunits of all homotrimeric riboflavin synthases are all characterized by intramolecular sequence similarity.
Sequence comparison between the proximal and distal domains of the homotrimeric riboflavin synthases from all completely sequenced genomes showed two non-overlapping branches. Hence, it is believed that both domains have evolved from a common ancestor by gene duplication. This domain structure has been observed for the first time using the enzyme from B. subtilis. This sequence showed marked internal homology encompassing 26 identical residues and 23 conservative replacements [69]. Later, this has been shown for enzymes of other organisms, like E. coli or S. pombe (Fig. 4) [51,69]. It was also correctly suggested on basis of the sequence data that two domains could form a pseudo-c2-symmetric pair that could form an active site at the domain interface. A pair of substrate molecules bound at that pseudo-c2-symmetric domain interface would have precisely the antiparallel topology required by the regiochemical features of the reaction. Notably, the intramolecular sequence similarity does not extend to the C-terminal segment of about two dozen residues. It is also relevant
Fig. 4. Internal sequence alignment of riboflavin synthase from S. pombe. (a) Alignment of the N barrel (red) and the C barrel (blue) domains; (b) stereodiagram of an internal superposition of the two b barrels with bound substrate analog inhibitor 6-carboxyethyl-7-oxo-8-ribityllumazine (18, PDB entry #1KZL). Reprinted from Gerhardt et al. [51], Copyright (2002) by Elsevier.
M. Fischer, A. Bacher / Archives of Biochemistry and Biophysics 474 (2008) 252–265
that the N-terminus of trimeric riboflavin synthases (conserved motif MFTG) shows a very high degree of evolutionary conservation. The ensuing hypothesis that the subunits fold into two closely similar domains was later validated by crystal structure analysis [51,64]. Surprisingly, the crystal structure analysis of the E. coli enzyme revealed a protein that is devoid of trigonal symmetry but shows two different types of pseudo-c2-symmetry [64]. The interaction of the subunits in the trimer is primarily mediated by the arrangement of the C-termini (which do not participate in the intrasubunit sequence similarity) in a triple helical motif (Fig. 5). The two folding domains of the E. coli subunit are related by ˚ , and the domains of the S. pombe enzyme an rmsd of 1.1 A ˚ (Fig. 4b). In each case, the are related by an rmsd of 0.97 A folding domains of a given subunit are related by pseudoc2-symmetry. A spurious c3 symmetry that had been proposed, incorrectly, on basis of a preliminary crystal
Fig. 5. Riboflavin synthase of E. coli (PDB entry #1I8D).
257
diffraction study was later explained by the fact that the three pseudo-c2-axes are related by angles of 124°, 85° and 151°; as already mentioned, the enzyme is really devoid of trigonal symmetry [64]. Last but not least, one specific N-terminal domain in the E. coli protein and the C-terminal domain of a second subunit are closely associated and form yet another pseudo-c2-symmetric pair (Fig. 5). As discussed in more detail below, this special pair forms the single active site of the protein, which is located at the domain interface and can impose on the two bound substrate molecules the quasi-c2-symmetric relationship that is required by the regiospecificity of the catalyzed reaction (cf. Fig. 6). The crystal structure of the E. coli enzyme was obtained without low molecular weight ligands, and arguments on catalysis had to remain speculative [64]. Nevertheless, a compelling model can be based on a combination of structural data obtained with riboflavin synthases of E. coli and S. pombe [51]. The crystal structure of the latter protein is clearly that of a monomer, despite the fact that the protein has been unequivocally shown to be trimeric in solution [62]. The monomer state observed in the crystal is believed to be induced by the organic solvent (methylpentane diol) used as a precipitant that is believed to break the hydrophobic interaction responsible for the cohesion of the Cterminal triple helix motif. On the other hand, the study with the S. pombe enzyme afforded a complex with the enzyme inhibitor 6-carboxyethyl-7-oxo-8-ribityllumazine (18) [70] (Fig. 7) that can be interpreted as a substrate analog. Both folding domains of the S. pombe enzyme bind that compound in shallow surface depressions with the ribityl side chains in extended conformations. Due to the closely similar folding topologies, the inhibitor molecules as observed in the monomeric S. pombe enzyme can be modeled into the trimeric E. coli enzyme in order to analyze the substrate topology at the putative single active site of that homotrimer. It was advantageous to omit the position 6 side chain of the inhibitor in order to avoid sterical hindrance (Fig. 8). As shown in Fig. 6, the modeled substrates come into close contact at the special pair of N-terminal domain of one subunit that forms an interface with the C-terminal domain of an adjacent subunit. The orientation of the modeled substrates correctly predicts the known regiochemistry of the reaction that
Fig. 6. Active site of riboflavin synthase of S. pombe formed by two adjacent monomers with bound 6-carboxyethyl-7-oxo-8-ribityllumazine (18). (a) The ligand bound to the N barrel (red) is drawn in yellow, whereas the 6-carboxyethyl-7-oxo-8-ribityllumazine in the adjacent C barrel (blue) is shown in dark yellow; (b) proposed binding of 6,7-dimethyl-8-ribityllumazine at the active site; (c) model for the pentacyclic reaction intermediate.
258
M. Fischer, A. Bacher / Archives of Biochemistry and Biophysics 474 (2008) 252–265
Fig. 7. Hydrogen-bonding topology of 6-carboxyethyl-7-oxo-8-ribityllumazine (18) bound to S. pombe riboflavin synthase (PDB entry #1KZL). (a) Nterminal domain; (b) C-terminal domain; (c) distances from the thiolate group of cysteine 48; (d) distances from the side chain oxygen of serine 146; R, ribityl chain. Reprinted from Gerhardt et al. [51], Copyright (2002) by Elsevier.
Fig. 8. X-Ray structures of riboflavin synthases from E. coli (PDB entry #1I8D) and S. pombe (PDB entry #1KZL). Left, Diagram of the superposition of one subunit of riboflavin synthase from E. coli (blue) and S. pombe (green); center, S. pombe riboflavin synthase monomer with bound 6-carboxyethyl-7oxo-8-ribityllumazine (18, yellow); right, trimeric model of S. pombe riboflavin synthase with bound 6-carboxyethyl-7-oxo-8-ribityllumazine (yellow). Reprinted from Gerhardt et al. [51], Copyright (2002) by Elsevier.
requires an antiparallel orientation. Moreover, the model shows that the ligands are in a virtually ideal position for the formation of the pentacyclic reaction intermediate (Figs. 3 and 6c). The topology leaves no doubt that the lumazine molecule originally bound to the N-terminal domain is the four-carbon acceptor arranged to end up as riboflavin and the substrate bound to the C-terminal domain is the four-carbon donor destined to end up as 6. As expected, the dimerization must result in a cis fusion of the two ring systems. It is even possible to predict the
absolute stereochemistry of the resulting pentacyclic moiety (cf. Fig. 3, compound Q) [51]. With special regard to the stereochemical features of the enzyme reaction, it is now in order to address the fact that an altogether different riboflavin synthase type is present in Archaea. An archaeal riboflavin synthase was first purified from the methanogenic bacterium Methanothermobacter thermoautotrophicus (Methanobacterium thermoautotrophicum) [71]. Later, the cognate gene was cloned by marker rescue and was found to encode a peptide of 153 amino
M. Fischer, A. Bacher / Archives of Biochemistry and Biophysics 474 (2008) 252–265
acid residues that was devoid of sequence similarity with the eubacterial riboflavin synthases described above [71]. Moreover, the M. thermoautotrophicus enzyme showed no intramolecular sequence similarity. However, sequence analysis identified the protein as a paralog of 6,7dimethyl-8-ribityllumazine synthase (Fig. 9a) [72]. Later, the enzyme from Methanocaldococcus jannaschii with similarity to riboflavin synthase of M. thermoautotrophicus was cloned and characterized. It was shown that this enzyme forms a pentamer in solution and in the crystal structure [73]. Moreover, the crystal structure revealed a c5-symmetric structure and a folding topology closely resembling that of 6,7-dimethyl-8-ribityllumazine synthases (Fig. 10). Studies with isotope-labeled substrate that were monitored by 13C NMR spectroscopy documented using the enzyme from M. jannaschii that the archaeal riboflavin syn-
259
thase obeys the same regiochemistry as observed without catalysis and under the catalytic influence of the homotrimeric riboflavin synthases from eubacteria [72]. Presteady state kinetic analysis suggested the formation of a pentacyclic intermediate Q0 that could be isolated in substance after rapid quenching of presteady state enzyme/substrate mixtures [53]. Interestingly, however, it turned out that the pentacyclic compound produced by the archaeal enzyme can only be converted by archaeal riboflavin synthase leading to two molecules of 9 (backward reaction) or one molecule each of riboflavin (10) and 6 (forward reaction) but not by riboflavin synthase of eubacterial origin. Vice versa, the pentacyclic intermediate produced by the E. coli could be catalytically converted by the E. coli but not by the archaeal enzyme [53]. These findings in conjunction with direct NMR studies leave no doubt
Fig. 9. Structural comparision of riboflavin synthase from M. jannaschii (PDB entry #2B99) and lumazine synthase from S. pombe (PDB entry #1KZL). (a) sequence alignment of riboflavin synthase from M. jannaschii (MjaRS) and lumazine synthase from S. pombe (SP-LS). Secondary structure elements (ahelices, rods; b -strands, arrows) corresponding to MJ-RS are in red, and the corresponding ones of SP-LS are in green. Active site residues are marked by asterisks in the corresponding colors; (b) structural alignment of the active sites formed by two adjacent monomers of SP-LS (dark green, residues Glu17 to Asp112; light green, residues Ser113 to Leu158) and MJ-RS (red, residues Thr2 to Met90; pink, residues Thr91 to Leu144). Bound riboflavin (green) in the case of SP-LS is in almost the same position as the bound inhibitor 6,7-dioxo-8-ribityllumazine A (red) in the case of MJ-RS. Secondary structure element labeling refers to MJ-RS; (c) 6,7-dioxo-8-ribityllumazine (19).
260
M. Fischer, A. Bacher / Archives of Biochemistry and Biophysics 474 (2008) 252–265
Fig. 10. Left: schematic ligand binding at the active sites of a pentameric riboflavin synthase; right: ribbon presentation of the M. jannaschii riboflavin synthase pentamer in complex with 6,7-dioxo-8-ribityllumazine (19, red) viewed along the 5-fold noncrystallographic symmetry axis (PDB entry #2B99). Individual subunits are shown in different colors.
whatsoever that the pentacyclic intermediates produced by the enzymes of E. coli and M. jannaschii are diastereomers. In both cases, the two lumazine moieties are connected with cis stereochemistry, and the heterocylic moieties, without consideration of the side chains would appear as enantiomers. However, due to the chiral properties of the ribityl side chains, which have D-configuration, the compounds Q and Q0 are diastereomers whose chemical shifts show minor but nevertheless significant differences [53]. In order to discuss catalysis by the pentameric, archaeal riboflavin synthase, it is necessary to address the paralogous lumazine synthases that have been studied in much more detail over a period of three decades, although detailed coverage of lumazine synthases is not the subject of this review. Briefly, lumazine synthases occur in three different quaternary structures, as pentamers, as icosahedral dodecamers of pentamers and, rarely, as c5-symmetric dimers of pentamers [65,67,74–85]. The icosahedral species form hollow capsids; in Bacillaceae, these capsids can contain a trimeric riboflavin synthase in the central core space [86] (Fig. 11). Invariably, the topologically equivalent active centers are located at the interfaces of adjacent subunits in the pentamer module. Inside the spacious active site cavity, the 5-amino group of 6 is believed to undergo an electrophilic attack by the carbonyl group of 3,4-dihydroxy-2-butanone 4-phosphate (8). The resulting Schiff base eliminates phosphate prior to ring closure. The active sites of the paralogous riboflavin synthase from Archaea have the same basic topology as that of lumazine synthase. X-ray structure analysis shows that two molecules of the inhibitor 6,7-dioxo-8-ribityllumazine (19) can be accommodated [73]. The topology of the two substrate analogs at the active site is in excellent agreement with the formation of the 6S/7R-form of the pentacyclic intermediate (Fig. 3). Sequence comparison indicates that the pentameric archaeal riboflavin synthases must have separated from the lumazine synthase branch at a very early time point in evolution.
Fig. 11. Computer generated model of a heavy riboflavin synthase complex. The capsid was generated using the coordinates from the lumazine synthase 60-mer of B. subtilis (PDB entry #1RVV) and from the riboflavin synthase trimer of E. coli (PDB entry #1I8D).
Apart from the stereochemical aspects, which have been described above, the expectation that protein structure analysis might clarify the details of the complex riboflavin synthase mechanism has been frustrated so far. A mutagenesis screen of the active site region of the trimeric E. coli enzyme failed to reveal any amino acid residues that might be involved in covalent catalysis [87]. However, the catalytic activity of the enzyme can be brought to an undetectably low level by alteration of the hydrophobic Nterminus. Since the N-terminal sequence motif MFTG is highly conserved over very large genetic distances, this is not particularly surprising. The most plausible function of the hydrophobic N-terminal residues consists in the accurate positioning of the acceptor substrate.
M. Fischer, A. Bacher / Archives of Biochemistry and Biophysics 474 (2008) 252–265
In the absence of significant guidance from structural biology, the pathway for the formation of the pentacyclic intermediates leaves ample room for speculation. All potential mechanisms should take into account that the reaction can take place in the complete absence of catalysts under relatively mild conditions (boiling aqueous solution). The fact that the non-enzymatic reaction and catalysis by both enzyme types invariably proceed with the same regiochemistry (although, as shown above, the stereochemical course is different for the two different enzyme classes) suggests that the non-enzymatic reaction proceeds via the same mechanism, although a formal proof for this is not available. Based on the unique CH acidity of the substrate 9, it has been proposed that a substrate molecule initially forms a covalent adduct by nucleophilic attack at C-7, either by a nucleophilic amino acid side chain or by a water molecule (Fig. 3) [88,89]. The resulting species was then proposed to be attacked by a second substrate molecule in the anion form. Following the discovery of the pentacyclic intermediates, it was not difficult to modify earlier mechanistic suggestions in order to accommodate the novel, experimentally documented intermediate. As mentioned above, the failure to identify, by systematic mutation studies, a nucleophilic amino acid side chain for the proposed C-7 adduct formation leaves only water as a plausible candidate for adduct formation. Some support for the covalent hydrate hypothesis can be deduced from the finding that the trifluoromethyl analogs of the substrate interact with trimeric riboflavin synthase in a strictly stereospecific manner; the diastereomer believed to have configuration 20 (epimer A) binds with relatively high affinity, whereas the diastereomer 21 (epimer B) shows no detectable binding (Fig. 12) [90]. Nevertheless, this cannot be considered as supporting evidence for a mechanistic role of a covalent hydrate. The observation that the uncatalyzed formation of riboflavin from 9 can proceed under acidic conditions does not precisely support the participation of an anionic species. It is however important to note that the pentacyclic intermediates have been rigorously shown to be kinetically competent intermediates of the
261
enzyme-catalyzed reaction. Last but not least, it should be noted that riboflavin synthase can process the pentacyclic intermediate in both directions, i.e. either cleavage under formation of two molecules of 6,7-dimethyl-8-ribityllumazine or, alternatively, one molecule each of riboflavin and 5-amino-6-ribitylamino-2,4(1H,3H)-pyrimidinedione (6). The N-terminal domain of E. coli riboflavin synthase can be expressed as a stable, recombinant protein [60]. The recombinant domains associate under formation of dimers with strict c2-symmetry (Fig. 13) [91,92]. The contact interface is different from the interfaces that mediate the intrasubunit contact and the intersubunit contact between N-terminal and C-terminal domains in the homotrimeric full-length protein. The recombinant N-terminal domain dimers bind riboflavin as well as 6,7-dimethyl-8ribityllumazine with relatively high affinity but are devoid of catalytic activity.
Fig. 13. The structure of the recombinant N-terminal domain dimer of riboflavin synthase from E. coli (PDB entry #1PKV) with bound riboflavin viewed along the 2-fold non-crystallographic symmetry axis.
Fig. 12. Fluorinated analogs of 6,7-bis(trifluoromethyl)-8-D-ribityllumazine. 20, Epimer A of 6,7-bis(trifluormethyl)-7-hydroxyl-8-ribityllumazine; 21, Epimer B of 6,7-bis(trifluormethyl)-7-hydroxyl-8-ribityllumazine; 22, 6-trifluoromethyl-7-oxo-8-ribityllumazine.
262
M. Fischer, A. Bacher / Archives of Biochemistry and Biophysics 474 (2008) 252–265
Fig. 14. Amino acid sequence similarity between the fluorescent protein of photobacteria and riboflavin synthase from S. pombe and E. coli. PL-LumP, lumazine protein of P. leiognathi (acc. no. Q06877); SP-RS, riboflavin synthase of S. pombe (acc. no. Q9Y7P0); EC-RS, riboflavin synthase of E. coli (acc. no. P0AFU8). Identical amino acid residues are in shadowed typeface.
Luminescent bacteria such as Photobacterium phosphoreum, Photobacterium leiognathi and certain Vibrio strains express a homotrimeric riboflavin synthase as well as proteins with sequence similarity to riboflavin synthase but devoid of catalytic activity [93,94]. These paralogs can bind various ligands such as 6,7-dimethyl-8-ribityllumazine, FMN and 6-methyl-7-oxo-8-ribityllumazine [5,6]. The complexes are highly fluorescent. Specifically, the fluorescence quantum yield of 6,7-dimethyl-8-ribityllumazine is increased by a factor of about 2 by complexation with lumazine protein from P. leiognathi [95]. The fluorescent proteins are all monomeric, in contrast to the paralogous, trimeric riboflavin synthase. The monomeric state is easily explained by the absence of the C-terminal sequence segment that plays the predominant role in the trimerisation of riboflavin synthase via triple helix formation (Fig. 14). It has been known for quite a while that the fluorescent proteins bind a single ligand molecule, despite the fact that sequence arguments (intramolecular sequence similarity and overall similarity to riboflavin synthases) predicts the presence of two similar folding domains in close analogy to trimeric riboflavin synthase [62,96,97]. Recently, the single binding site could be assigned to the N-terminal domain on basis of systematic mutagenesis studies in conjunction with NMR spectroscopy [98]. The luminescent proteins are all believed to play a role in light emission by the respective host organism. More specifically, energy is supposed to be transmitted from the excited state of bacterial luciferase to a luminescent protein by radiationless transition, and thus the fluorescent proteins may serve as an optical transponder that can modulate both the wavelength distribution and the quantum efficiency in that process [4,8,99,100]. Lumazine protein and the recombinant N-terminal homodimer domain of E. coli riboflavin synthase can both bind the bistrifluoromethyllumazine derivative 20 (Fig. 12)
with the same, strict diastereoselectivity as riboflavin synthase (only the A isomer is bound). Single 19F NMR signals are observed for each one of the methyl groups upon binding to either lumazine protein or recombinant N-terminal domain [60,62]. On the other hand, the compound in complex with riboflavin synthases from different organisms is characterized by complex 19F NMR spectra with multiple lines whose chemical shifts depend on ligand concentration. This is well in line with the absence of any strict symmetry in the homotrimeric native enzyme and has been interpreted tentatively as evidence for substantial conformational flexibility of the enzyme, although the details are not yet understood. A lumazine synthase/riboflavin synthase complex with an unusual quaternary structure has been observed in Bacillaceae and has been studied in more detail in B. subtilis [78,86,101–106]. The complex consists of a riboflavin synthase homotrimer enclosed in the central core space of the icosahedral lumazine synthase capsid (Fig. 11) [107]. The structural details of this complex are still unknown. Kinetic studies gave evidence of intermediate channelling; 6,7-dimethyl-8-ribityllumazine molecules that are generated by the lumazine synthase capsid are processed more rapidly by the resident riboflavin synthase module as compared to 6,7-dimethyl-8-ribityllumazine from the bulk solvent (Fig. 15) [108]. Confinement of the lumazine intermediate inside the icosahedral capsid has been proposed to be responsible for channeling, but the selective pressures that could have controlled the evolution of the complex structure remain unknown. Since several decades, it has been assumed that the enzymes of the riboflavin biosynthesis enzymes could be suitable targets for the development of novel antibacterial agents for the treatment of infections by Gram-negative bacteria and pathogenic yeasts. The hypothesis was based on the fact that riboflavin auxotrophs of E. coli and Salmo-
M. Fischer, A. Bacher / Archives of Biochemistry and Biophysics 474 (2008) 252–265
263
riboflavin synthase inhibitors has been reported recently and could serve as the basis for the screening libraries of drug-like compounds without apparent similarity with substrate and product structures that might have a better chance to penetrate the bacterial membrane barrier [121]. References
Fig. 15. Kinetic model of the reactions catalyzed by the heavy riboflavin synthase complex. a, a -subunits; b, b-subunits. Enzyme-bound ligands are shown in parentheses; ARP, 5-amino-6-ribitylamino-2,4(1H,3H)-pyrimidinedione (6), DHB, 3,4-dihydroxy-2-butanone 4-phosphate (8), LUM, 6,7-dimethyl-8-ribityllazine (9), RIB, riboflavin (10).
nella typhimurium require very high concentrations of exogenous riboflavin for growth, probably due to the absence of efficient uptake systems (by contrast, certain Gram-positive eubacteria can acquire riboflavin from the environment with outstanding efficacy). Recent studies have confirmed that riboflavin biosynthesis genes and proteins are virulence factors in S. enterica [109,110]. Indirect evidence further suggests that riboflavin biosynthesis enzymes could also be addressed as targets for novel drugs for tuberculosis and lepra. The argument is based on the comparison between the genomes of Mycobacterium tuberculosis and M. leprae [111]. The genome of the latter organism has passed through a very extensive process of gene fragmentation that has resulted in the destruction of about half of the gene endowment of typical Mycobacteriaceae. This process is believed to have accompanied the transition of M. leprae to a strictly intracellular lifestyle. Interestingly, the extensive gene fragmentation has consistently spared the genes of the riboflavin pathway, thus suggesting that this pathway is of vital importance even for a strictly intracellular lifestyle. Based on this hypothesis, extensive work on the synthesis of substrate analogs has resulted in a number of potent riboflavin synthase inhibitors [70,90,112–120]. Unfortunately, none of the known inhibitors shows in vivo activity against bacteria, probably because they all fail to enter the bacterial cells. A high throughput screening method for
[1] F. Mu¨ller (Ed.), Chemistry and Biochemistry of Flavoenzymes, vol. 2, CRC Press, Boca Raton, FL, 1991, p. 506. [2] B.A. Palfey, V. Massey, in: M. Sinnot (Ed.), Comprehensive Biological Catalysis, Academic Press, San Diego, 1998, pp. 83–154. [3] S. Bornemann, Nat. Prod. Rep. 19 (2002) 761–772. [4] J. Lee, Biophys. Chem. 48 (1993) 149–158. [5] V.N. Petushkov, J. Lee, Eur. J. Biochem. 245 (1997) 790–796. [6] P. Macheroux, K.U. Schmidt, P. Steinerstauch, S. Ghisla, P. Colepicolo, R. Buntic, J.W. Hastings, Biochem. Biophys. Res. Commun. 146 (1987) 101–106. [7] J. Lee, Biochemistry 11 (1972) 3350–3359. [8] J. Lee, I.B.C. Matheson, F. Mu¨ller, D.J. O’Kane, J. Vervoort, A.J.W.G. Visser, in: F. Mu¨ller (Ed.), Chem Biochem Flavoenzymes, vol. 2, CRC Press, Florida, Boca Raton, 1991, 109–151. [9] D.J. O’Kane, J. Lee, Biochemistry 24 (1985) 1484–1488. [10] V.W. Cheng, E. Ma, Z. Zhao, R.A. Rothery, J.H. Weiner, J. Biol. Chem. 281 (2006) 27662–27668. [11] J.M. Hudson, K. Heffron, V. Kotlyar, Y. Sher, E. Maklashina, G. Cecchini, F.A. Armstrong, J. Am. Chem. Soc. 127 (2005) 6977–6989. [12] G.A. Roberts, A. Celik, D.J. Hunter, T.W. Ost, J.H. White, S.K. Chapman, N.J. Turner, S.L. Flitsch, J. Biol. Chem. 278 (2003) 48914–48920. [13] H. O’Neill, S.G. Mayhew, G. Butler, J. Biol. Chem. 273 (1998) 21015–21024. [14] M. Mewies, W.S. McIntire, N.S. Scrutton, Protein Sci. 7 (1998) 7– 20. [15] W.R. Briggs, C.F. Beck, A.R. Cashmore, J.M. Christie, J. Hughes, J.A. Jarillo, T. Kagawa, H. Kanegae, E. Liscum, A. Nagatani, K. Okada, M. Salomon, W. Rudiger, T. Sakai, M. Takano, M. Wada, J.C. Watson, Plant Cell 13 (2001) 993–997. [16] T. Kagawa, J. Plant Res. 116 (2003) 77–82. [17] J.M. Christie, M. Salomon, K. Nozue, M. Wada, W.R. Briggs, Proc. Natl. Acad. Sci. USA 96 (1999) 8779–8783. [18] J.M. Christie, T.E. Swartz, R.A. Bogomolni, W.R. Briggs, Plant J. 32 (2002) 205–219. [19] S. Crosson, K. Moffat, Proc. Natl. Acad. Sci. USA 98 (2001) 2995– 3000. [20] S. Crosson, K. Moffat, Plant Cell 14 (2002) 1067–1075. [21] S. Crosson, S. Rajagopal, K. Moffat, Biochemistry 42 (2003) 2–10. [22] M. Salomon, W. Eisenreich, H. Du¨rr, E. Schleicher, E. Knieb, V. Massey, W. Ru¨diger, F. Mu¨ller, A. Bacher, G. Richter, Proc. Natl. Acad. Sci. USA 98 (2001) 12357–12361. [23] A. Sancar, Annu. Rev. Biochem. 69 (2000) 31–67. [24] T. Imaizumi, H.G. Tran, T.E. Swartz, W.R. Briggs, S.A. Kay, Nature 426 (2003) 302–306. [25] E. Silva, A.M. Edwards, in: D.P. Ha¨der, G. Jori (Eds.), Comprehensive Series in Photochemistry and Photobiology, vol. 6, The Royal Society of Chemistry, Cambridge, 2006, 328. [26] K. Kis, K. Kugelbrey, A. Bacher, J. Org. Chem. 66 (2001) 2555– 2559. [27] A. Eschenmoser, E. Loewenthal, Chem. Soc. Rev. 21 (1992) 1–16. [28] C.J. Strupp, Untersuchungen u¨ber die nicht-enzymatische Simulation des Biosyntheseweges zu Riboflavin. ETH Thesis Nr. 9832, Zu¨rich 1992. [29] D.H.G. Crout, J.A. Hadfield, J. Chem. Soc. Chem. Com. 10 (1987) 742–744. [30] R.L. Beach, G.W.E. Plaut, Tetrahedron Lett. 40 (1969) 3489–3492. [31] T. Paterson, H.C.S. Wood, J. Chem. Soc. Commun. (1969) 290–291.
264
M. Fischer, A. Bacher / Archives of Biochemistry and Biophysics 474 (2008) 252–265
[32] T. Rowan, H.C. Wood, J. Chem. Soc., Perkin Trans. 1 4 (1968) 452– 458. [33] T. Rowan, H.C.S. Wood, Proc. Chem. Soc. (1963) 21–22. [34] M. Fischer, A. Bacher, Nat. Prod. Rep. 22 (2005) 324–350. [35] M. Fischer, A. Bacher, Physiol. Plant. 126 (2006) 304–318. [36] A. Bacher, S. Eberhardt, W. Eisenreich, M. Fischer, S. Herz, B. Illarionov, K. Kis, G. Richter, Vitam. Horm. 61 (2001) 1–49. [37] A. Bacher, S. Eberhardt, M. Fischer, K. Kis, G. Richter, Annu. Rev. Nutr. 20 (2000) 153–167. [38] A. Bacher, S. Eberhardt, G. Richter, in: F.C. Neidhardt et al. (Eds.), Escherichia coli and Salmonella, ASM Press, Washington, DC, 1996, pp. 657-664. [39] L. Chatwell, T. Krojer, A. Fidler, W. Ro¨misch, W. Eisenreich, A. Bacher, R. Huber, M. Fischer, J. Mol. Biol. 359 (2006) 1334–1351. [40] M. Graupner, H. Xu, R.H. White, J. Bacteriol. 184 (2002) 1952–1957. [41] M. Fischer, W. Ro¨misch, S. Saller, B. Illarionov, G. Richter, F. Rohdich, W. Eisenreich, A. Bacher, J. Biol. Chem. 279 (2004) 36299–36308. [42] G. Richter, M. Fischer, C. Krieger, S. Eberhardt, H. Lu¨ttgen, I. Gerstenschla¨ger, A. Bacher, J. Bacteriol. 179 (1997) 2022–2028. [43] S.C. Chen, Y.C. Chang, C.H. Lin, C.H. Lin, S.H. Liaw, J. Biol. Chem. 281 (2006) 7605–7613. [44] P. Stenmark, M. Moche, D. Gurmu, P. Nordlund, J. Mol. Biol. 373 (2007) 48–64. [45] R.L. Beach, G.W.E. Plaut, J. Am. Chem. Soc. 92 (1970) 2913–2916. [46] W. Pfleiderer, W. Hutzenlaub, Chemische Berichte 106 (1973) 3149– 3174. [47] R. Stewart, J.M. McAndless, J. Chem. Soc. Perkin Trans. 2 (1972) 376–379. [48] R.L. Beach, G.W.E. Plaut, J. Org. Chem. 36 (1971) 3937–3943. [49] D.H. Bown, P.J. Keller, H.G. Floss, H. Sedlmaier, A. Bacher, J. Org. Chem. 51 (1986) 2461–2467. [50] W. Pfleiderer, J.W. Bunting, D.D. Perrin, G. Nu¨bel, Chemische Berichte 99 (1966) 3503–3523. [51] S. Gerhardt, A.K. Schott, N. Kairies, M. Cushman, B. Illarionov, W. Eisenreich, A. Bacher, R. Huber, S. Steinbacher, M. Fischer, Structure 10 (2002) 1371–1381. [52] B. Illarionov, W. Eisenreich, A. Bacher, Proc. Natl. Acad. Sci. USA 98 (2001) 7224–7229. [53] B. Illarionov, W. Eisenreich, N. Schramek, A. Bacher, M. Fischer, J. Biol. Chem. 280 (2005) 28541–28546. [54] B. Illarionov, I. Haase, M. Fischer, A. Bacher, N. Schramek, Biol. Chem. 386 (2005) 127–136. [55] S. Kuwada, T. Masuda, T. Kishi, M. Asai, J. Vitaminol. (Kyoto) 4 (1958) 217–225. [56] G.F. Maley, G.W. Plaut, J. Biol. Chem. 243 (1959) 641–647. [57] T. Masuda, Pharm. Bull. 4 (1956) 71–72. [58] T. Masuda, Pharm. Bull. 5 (1957) 136–141. [59] S. Eberhardt, G. Richter, W. Gimbel, T. Werner, A. Bacher, Eur. J. Biochem. 242 (1996) 712–719. [60] S. Eberhardt, N. Zingler, K. Kemter, G. Richter, M. Cushman, A. Bacher, Eur. J. Biochem. 268 (2001) 4315–4323. [61] M. Fischer, I. Haase, R. Feicht, N. Schramek, P. Ko¨hler, P. Schieberle, A. Bacher, Biol. Chem. 386 (2005) 417–428. [62] M. Fischer, A.K. Schott, K. Kemter, R. Feicht, G. Richter, B. Illarionov, W. Eisenreich, S. Gerhardt, M. Cushman, S. Steinbacher, R. Huber, A. Bacher, BMC Biochem. 4 (2003) 18. [63] R.A. Harvey, G.W. Plaut, J. Biol. Chem. 241 (1966) 2120–2136. [64] D.I. Liao, Z. Wawrzak, J.C. Calabrese, P.V. Viitanen, D.B. Jordan, Structure (Camb.) 9 (2001) 399–408. [65] W. Meining, S. Mo¨rtl, M. Fischer, M. Cushman, A. Bacher, R. Ladenstein, J. Mol. Biol. 299 (2000) 181–197. [66] W. Meining, G. Tibbelin, R. Ladenstein, S. Eberhardt, M. Fischer, A. Bacher, J. Struct. Biol. 121 (1998) 53–60. [67] S. Mo¨rtl, M. Fischer, G. Richter, J. Tack, S. Weinkauf, A. Bacher, J. Biol. Chem. 271 (1996) 33201–33207. [68] G.W. Plaut, R.L. Beach, T. Aogaichi, Biochemistry 9 (1970) 771– 785.
[69] K. Schott, J. Kellermann, F. Lottspeich, A. Bacher, J. Biol. Chem. 265 (1990) 4204–4209. [70] M. Cushman, D. Yang, S. Gerhardt, R. Huber, M. Fischer, K. Kis, A. Bacher, J. Org. Chem. 67 (2002) 5807–5816. [71] S. Eberhardt, S. Korn, F. Lottspeich, A. Bacher, J. Bacteriol. 179 (1997) 2938–2943. [72] M. Fischer, A.K. Schott, W. Ro¨misch, A. Ramsperger, M. Augustin, A. Fidler, A. Bacher, G. Richter, R. Huber, W. Eisenreich, J. Mol. Biol. 343 (2004) 267–278. [73] A. Ramsperger, M. Augustin, A.K. Schott, S. Gerhardt, T. Krojer, W. Eisenreich, B. Illarionov, M. Cushman, A. Bacher, R. Huber, M. Fischer, J. Biol. Chem. 281 (2006) 1224–1232. [74] M. Fischer, I. Haase, R. Feicht, G. Richter, S. Gerhardt, J.P. Changeux, R. Huber, A. Bacher, Eur. J. Biochem. 269 (2002) 519– 526. [75] S. Gerhardt, I. Haase, S. Steinbacher, J.T. Kaiser, M. Cushman, A. Bacher, R. Huber, M. Fischer, J. Mol. Biol. 318 (2002) 1317–1329. [76] S. Klinke, V. Zylberman, H.R. Bonomi, I. Haase, B.G. Guimaraes, B.C. Braden, A. Bacher, M. Fischer, F.A. Goldbaum, J. Mol. Biol. 373 (2007) 664–680. [77] R. Ladenstein, K. Ritsert, R. Huber, G. Richter, A. Bacher, Eur. J. Biochem. 223 (1994) 1007–1017. [78] R. Ladenstein, M. Schneider, R. Huber, H.D. Bartunik, K. Wilson, K. Schott, A. Bacher, J. Mol. Biol. 203 (1988) 1045–1070. [79] E. Morgunova, W. Meining, B. Illarionov, I. Haase, G. Jin, A. Bacher, M. Cushman, M. Fischer, R. Ladenstein, Biochemistry 44 (2005) 2746–2758. [80] E. Morgunova, S. Saller, I. Haase, M. Cushman, A. Bacher, M. Fischer, R. Ladenstein, J. Biol. Chem. 282 (2007) 17231–17241. [81] K. Persson, G. Schneider, B.J. Douglas, P.V. Viitanen, T. Sandalova, Prot. Sci. 8 (1999) 2355–2365. [82] K. Ritsert, R. Huber, D. Turk, R. Ladenstein, K. Schmidt-Ba¨se, A. Bacher, J. Mol. Biol. 253 (1995) 151–167. [83] X. Zhang, W. Meining, M. Fischer, A. Bacher, R. Ladenstein, J. Mol. Biol. 306 (2001) 1099–1114. [84] V. Zylberman, P.O. Craig, S. Klinke, B.C. Braden, A. Cauerhff, F.A. Goldbaum, J. Biol. Chem. 279 (2004) 8093–8101. [85] I. Haase, S. Mo¨rtl, P. Kohler, A. Bacher, M. Fischer, Eur. J. Biochem. 270 (2003) 1025–1032. [86] A. Bacher, H. Schnepple, B. Maila¨nder, M.K. Otto, Y. Ben-Shaul, Flavins Flavoproteins, Proceedings of the 6th International Symposium (1980) 579–586. [87] B. Illarionov, K. Kemter, S. Eberhardt, G. Richter, M. Cushman, A. Bacher, J. Biol. Chem. 276 (2001) 11524–11530. [88] G.W. Plaut, R.L. Beach, in: W. Pfleiderer (Ed.), Chemistry and Biology of Pteridines, Proceedings of the International Symposium, 5th, de Gruyter, Berlin, 1975, pp. 101-124. [89] G.W.E. Plaut, R.L. Beach, in: T.P. Singer (Ed.), Flavins and Flavoproteins, Proceedings of the 5th International Symposium, Elsevier, Amsterdam, 1976, pp. 737–746. [90] M. Cushman, D.A. Patrick, A. Bacher, J. Scheuring, J. Org. Chem. 56 (1991) 4603–4608. [91] W. Meining, S. Eberhardt, A. Bacher, R. Ladenstein, J. Mol. Biol. 331 (2003) 1053–1063. [92] V. Truffault, M. Coles, T. Diercks, K. Abelmann, S. Eberhardt, H. Lu¨ttgen, A. Bacher, H. Kessler, J. Mol. Biol. 309 (2001) 949–960. [93] J. Lee, P. Koka, Methods Enzymol. 57 (1978) 226–234. [94] E.A. Meighen, FASEB J. 7 (1993) 1016–1022. [95] D.J. O’Kane, J. Lee, Methods Enzymol. 133 (1986) 149–172. [96] J. Scheuring, M. Fischer, M. Cushman, J. Lee, A. Bacher, H. Oschkinat, Biochemistry 35 (1996) 9637–9646. [97] J. Scheuring, J. Lee, M. Cushman, H. Patel, D.A. Patrick, A. Bacher, Biochemistry 33 (1994) 7634–7640. [98] B. Illarionov, W. Eisenreich, M. Wirth, C.L. Lee, Y.E. Woo, A. Bacher, M. Fischer, Biol. Chem. 388 (2007) 1313–1323. [99] J. Lee, D.J. O’Kane, A.J. Visser, Biochemistry 24 (1985) 1476–1483. [100] J. Lee, in: J.G. Burr (Ed.), Chemiluminescence and Bioluminescence, Marcel Dekker Inc., New York, 1985, 401-437.
M. Fischer, A. Bacher / Archives of Biochemistry and Biophysics 474 (2008) 252–265 [101] A. Bacher, S. Eberhardt, M. Fischer, S. Mo¨rtl, K. Kis, K. Kugelbrey, J. Scheuring, K. Schott, Methods Enzymol. 280 (1997) 389–399. [102] A. Bacher, B. Maila¨nder, J. Bacteriol. 134 (1978) 476–482. [103] A. Bacher, R. Baur, U. Eggers, H.D. Harders, M.K. Otto, H. Schnepple, J. Biol. Chem. 255 (1980) 632–637. [104] R. Ladenstein, H.C. Ludwig, J. Biol. Chem. 258 (1983) 11981– 11983. [105] A. Bacher, H.C. Ludwig, H. Schnepple, Y. Ben-Shaul, J. Mol. Biol. 187 (1986) 75–86. [106] R. Ladenstein, B. Meyer, R. Huber, H. Labischinski, K. Bartels, H.D. Bartunik, L. Bachmann, H.C. Ludwig, A. Bacher, J. Mol. Biol. 187 (1986) 87–100. [107] A. Bacher, R. Baur, U. Eggers, H. Harders, H. Schnepple, Flavins Flavoproteins, in: T.P. Singer (Ed.), Proceedings of the 5th International Symposium (1976) 729-732. [108] K. Kis, A. Bacher, J. Biol. Chem. 270 (1995) 16788–16795. [109] C. Rollenhagen, D. Bumann, Infect. Immun. 74 (2006) 1649–1660. [110] D. Becker, M. Selbach, C. Rollenhagen, M. Ballmaier, T.F. Meyer, M. Mann, D. Bumann, Nature 440 (2006) 303–307. [111] S.T. Cole, K. Eiglmeier, J. Parkhill, K.D. James, N.R. Thomson, P.R. Wheeler, N. Honore, T. Garnier, C. Churcher, D. Harris, K. Mungall, D. Basham, D. Brown, T. Chillingworth, R. Connor, R.M. Davies, K. Devlin, S. Duthoy, T. Feltwell, A. Fraser, N. Hamlin, S. Holroyd, T. Hornsby, K. Jagels, C. Lacroix, J. Maclean, S. Moule, L. Murphy, K. Oliver, M.A. Quail, M.A. Rajandream, K.M. Rutherford, S. Rutter, K. Seeger, S. Simon, M. Simmonds, J.
[112] [113] [114] [115] [116] [117] [118] [119]
[120] [121]
[122]
265
Skelton, R. Squares, S. Squares, K. Stevens, K. Taylor, S. Whitehead, J.R. Woodward, B.G. Barrell, Tuberculosis 81 (2001) 157–163. J. Chen, T. Sambaiah, B. Illarionov, M. Fischer, A. Bacher, M. Cushman, J. Org. Chem. 69 (2004) 6996–7003. M. Cushman, G. Jin, T. Sambaiah, B. Illarionov, M. Fischer, R. Ladenstein, A. Bacher, J. Org. Chem. 70 (2005) 8162–8170. M. Cushman, F. Mavandadi, K. Kugelbrey, A. Bacher, Bioorg. Med. Chem. 6 (1998) 409–415. M. Cushman, F. Mavandadi, D. Yang, K. Kugelbrey, K. Kis, A. Bacher, J. Org.Chem. 64 (1999) 4635–4642. M. Cushman, J.T. Mihalic, K. Kis, A. Bacher, Bioorg. Med. Chem. Lett. 9 (1999) 39–42. M. Cushman, T. Sambaiah, G. Jin, B. Illarionov, M. Fischer, A. Bacher, J. Org. Chem. 69 (2004) 601–612. M. Cushman, D. Yang, K. Kis, A. Bacher, J. Org. Chem. 66 (2001) 8320–8327. M. Cushman, D. Yang, J.T. Mihalic, J. Chen, S. Gerhardt, R. Huber, M. Fischer, K. Kis, A. Bacher, J. Org. Chem. 67 (2002) 6871–6877. A. Talukdar, B. Illarionov, A. Bacher, M. Fischer, M. Cushman, J. Org. Chem. 72 (2007) 7167–7175. J. Kaiser, B. Illarionov, F. Rohdich, W. Eisenreich, S. Saller, J. Van den Brulle, M. Cushman, A. Bacher, M. Fischer, Anal. Biochem. 365 (2007) 52–61. M.A. Santos, J.J. Garcı´a-Ramı´rez, J.L. Revuelta, J. Biol. Chem. 270 (1995) 437–444.