Email updates

Keep up to date with the latest news and content from BMC Biochemistry and BioMed Central.

Open Access Highly Accessed Research article

Riboflavin synthase of Schizosaccharomyces pombe. Protein dynamics revealed by 19F NMR protein perturbation experiments

Markus Fischer1*, Ann-Kathrin Schott1, Kristina Kemter1, Richard Feicht1, Gerald Richter1, Boris Illarionov2, Wolfgang Eisenreich1, Stefan Gerhardt3, Mark Cushman4, Stefan Steinbacher3, Robert Huber3 and Adelbert Bacher1

Author Affiliations

1 Lehrstuhl für Organische Chemie und Biochemie, Technische Universität München, Lichten-bergstr. 4, D-85747 Garching, Germany

2 The Institute for Biophysics, Krasnoyarsk, Russia

3 Department of Protein Crystallography, Max-Planck-Institut für Biochemie, Am Klopferspitz 18a, D-82512 Martinsried, Germany

4 Department of Medicinal Chemistry and Molecular Pharmacology, Purdue University, West Lafayette, Indiana, USA

For all author emails, please log on.

BMC Biochemistry 2003, 4:18  doi:10.1186/1471-2091-4-18

The electronic version of this article is the complete one and can be found online at: http://www.biomedcentral.com/1471-2091/4/18


Received:11 September 2003
Accepted:23 December 2003
Published:23 December 2003

© 2003 Fischer et al; licensee BioMed Central Ltd. This is an Open Access article: verbatim copying and redistribution of this article are permitted in all media for any purpose, provided this notice is preserved along with the article's original URL.

Abstract

Background

Riboflavin synthase catalyzes the transformation of 6,7-dimethyl-8-ribityllumazine into riboflavin in the last step of the riboflavin biosynthetic pathway. Gram-negative bacteria and certain yeasts are unable to incorporate riboflavin from the environment and are therefore absolutely dependent on endogenous synthesis of the vitamin. Riboflavin synthase is therefore a potential target for the development of antiinfective drugs.

Results

A cDNA sequence from Schizosaccharomyces pombe comprising a hypothetical open reading frame with similarity to riboflavin synthase of Escherichia coli was expressed in a recombinant E. coli strain. The recombinant protein is a homotrimer of 23 kDa subunits as shown by sedimentation equilibrium centrifugation. The protein sediments at an apparent velocity of 4.1 S at 20°C. The amino acid sequence is characterized by internal sequence similarity indicating two similar folding domains per subunit. The enzyme catalyzes the formation of riboflavin from 6,7-dimethyl-8-ribityllumazine at a rate of 158 nmol mg-1 min-1 with an apparent KM of 5.7 microM. 19F NMR protein perturbation experiments using fluorine-substituted intermediate analogs show multiple signals indicating that a given ligand can be bound in at least 4 different states. 19F NMR signals of enzyme-bound intermediate analogs were assigned to ligands bound by the N-terminal respectively C-terminal folding domain on basis of NMR studies with mutant proteins.

Conclusion

Riboflavin synthase of Schizosaccharomyces pombe is a trimer of identical 23-kDa subunits. The primary structure is characterized by considerable similarity of the C-terminal and N-terminal parts. Riboflavin synthase catalyzes a mechanistically complex dismutation of 6,7-dimethyl-8-ribityllumazine affording riboflavin and 5-amino-6-ribitylamino-2,4(1H,3H)-pyrimidinedione. The 19F NMR data suggest large scale dynamic mobility in the trimeric protein which may play an important role in the reaction mechanism.

Background

Flavocoenzymes derived from riboflavin (vitamin B2) act as cofactors for a variety of redox reactions and are indispensable in all cells. Whereas plants and many microorganisms obtain riboflavin by endogenous synthesis, animals depend on nutritional supply of the vitamin. Certain bacteria (e.g. Enterobacteriaceae) and yeasts are unable to incorporate riboflavin from the environment and are therefore absolutely dependent on the endogenous synthesis of the vitamin [1-4], and the enzymes of the riboflavin biosynthetic pathway are therefore potential targets for antiinfective drug design.

The absence of the riboflavin biosynthetic pathway in the human host is advantageous with regard to drug development. The final step in the riboflavin biosynthetic pathway is catalyzed by riboflavin synthase. The mechanistically unusual dismutation of the substrate, 6,7-dimethyl-8-ribityllumazine (Compound 3, Fig. 1) affording riboflavin (Compound 4) and 5-amino-6-ribitylamino-2,4(1H,3H)-pyrimidinedione (Compound 1) requires the presence of two substrate molecules serving as donor resp. acceptor of a 4-carbon moiety at the active site of the enzyme [5-7]. The regiochemistry of the enzyme-catalyzed reaction predicts an antiparallel arrangement of the two substrate molecules [8-10]. A pentacyclic dimer of 6,7-dimethyl-8-ribityllumazine (Compond 5, Fig. 2) has been identified as a kinetically competent intermediate in the complex reaction sequence [11]. The riboflavin synthases of Escherichia coli and Bacillus subtilis are homotrimeric proteins [12-14].

thumbnailFigure 1. Terminal reactions in the pathway of riboflavin biosynthesis. A, 3,4-dihydroxy-2-butanone 4-phosphate synthase; B, 6,7-dimethyl-8-ribityllumazine synthase; C, riboflavin synthase.

thumbnailFigure 2. Hypothetical reaction mechanism of riboflavin synthase. X, unkown nucleophile; R, ribityl. [11]

Intramolecular sequence similarity suggested that each subunit folds into two topologically similar domains, each of which can accommodate one substrate molecule. The active site was proposed to be located at the interface of adjacent domains; two substrate molecules were assumed to be sandwiched between adjacent subunits [14]. The structure of riboflavin synthase from E. coli as determined by X-ray crystallography at a resolution of 2.0 Å confirmed the predicted similarity in the folding patterns of the N-terminal and C-terminal domain (rms = 0.8 Å) [15] but gave no direct information on enzyme substrate interaction. The N-terminal domain of the E. coli enzyme (residues 1 – 97) could be expressed as a stable protein with a relatively high affinity for riboflavin.

Surprisingly, this artificial protein is a c2 symmetric homodimer, whereas the full length riboflavin synthase subunit forms a homotrimer [16]. The structure of the artificial N-terminal domain dimer in complex with riboflavin has been determined by NMR spectroscopy [17]. Protein perturbation studies monitored by 19F NMR with fluorine-substituted intermediate analogs had suggested that the binding sites of the homotrimeric riboflavin synthase are not topologically equivalent [18-20].

More specifically, these experiments showed that certain ligands could be bound to the protein in at least four different binding states, whereas only two different states (one for the N-terminal and one for the C-terminal domain) would have been expected in a c3 symmetric homotrimer. These observations are well in line with the absence of trigonal molecular symmetry revealed by the X-ray structure analysis of riboflavin synthase of E. coli [15]. The structure of riboflavin synthase from S. pombe in complex with the substrate analog, 6-carboxyethyl-7-oxo-8-ribityllumazine has been determined at 2.1 Å resolution. In contrast to the homotrimeric solution state of native riboflavin synthase, the enzyme was found to be monomeric in the crystal structure. Structural comparison of the riboflavin synthases of S. pombe and E. coli suggested oligomer contact sites and delineated the catalytic site for dimerization of the substrate and subsequent fragmentation of the pentacyclic intermediate [21]. This paper reports 19F NMR protein perturbation studies using wild type and mutant riboflavin synthase of S. pombe in complex with fluorinated intermediate analogs. The data suggest large scale domain mobility which appear relevant for the complex reaction mechanism.

Results

A putative rib5+ gene of S. pombe (NCBI accession number, T40995) predicted a protein of 208 amino acid residues with similarity to riboflavin synthase (50 % identity with riboflavin synthase of S. cerevisiae and 36 % identity with riboflavin synthase of E. coli).

The open reading frame was amplified from S. pombe cDNA and was cloned into the plasmid vector pNCO113. The cDNA sequence as determined by dideoxynucleotide sequencing was identical with the chromosomal sequence reported earlier and has been deposited in the GenBank data base (accession number AF505789).

A recombinant E. coli strain carrying a plasmid with the S. pombe rib5+ gene under control of the T5 and the lac operator expressed a protein with an approximate mass of 23 kDa as judged by SDS electrophoresis. The recombinant protein accounting for about 20 % of cell protein was purified to apparent homogeneity by three chromatographic steps as described under Methods. Partial Edman degradation afforded the N-terminal sequence MFTGLVEAIGVVKDVQGTID in agreement with the open reading frame.

The protein shows internal sequence similarity. Twenty one amino acid residues are identical in the N-terminal and C-terminal section (Fig. 3).

thumbnailFigure 3. Intramolecular sequence similarity of S. pombe riboflavin synthase. N-terminal part, amino acids 1–97; C-terminal part, amino acids 98–208. Identical residues are coloured black. The residues subjected to mutagenesis in this study are marked by an arrow.

The recombinant protein had a specific activity of 158 nmol mg-1 min-1 and an apparent KM of 5.7 μM at pH 7.2 and 37°C (Table 1). Electrospray mass spectrometry afforded a relative mass of 22,863 Da in good agreement with the predicted mass of 22,861 Da. The recombinant protein sediments as a single, symmetrical boundary with an apparent velocity of 4.1 S. Sedimentation equilibrium experiments indicate a molecular mass of 74 kDa using an ideal monodisperse model for simulation. Sedimentation equilibrium analysis in the presence of 5 mM 6-carboxyethyl-7-oxo-8-D-ribityllumazine (Compound 8; Fig. 4) afforded an apparent molecular weight of 76 kDa. It follows that the enzyme is homotrimer like the riboflavin synthases of B. subtilis and E. coli [13,22] and that the quaternary structure is not affected by binding of the substrate analog.

Table 1. Bacterial strains and plasmids

thumbnailFigure 4. Intermediate analogs used for protein perturbation studies. 6,7-bis(trifluoromethyl)-8-ribityllumazine hydrate, epimer A (Compound 6a) and epimer B (Compound 6b). 6-trifluoromethyl-7-oxo-8-ribityllumazine (Compound 7). 6-carboxyethyl-7-oxo-8-ribitylluamzine (Compound 8), 5-nitro-6-ribitylamino-2,4(1H,3H)-pyrimidinedione (Compound 9).

Protein perturbation studies monitored by 19F NMR had been performed earlier with riboflavin synthases from B. subtilis and E. coli and had shown surprising multiplicities of signals for enzyme-bound ligands [18-20,23]. Thus, studies with the intermediate analog, 6-trifluoromethyl-7-oxo-8-ribityllumazine (Compound 7, Fig. 4) had suggested at least four different binding states characterized by different 19F signals spread over a range of about 9 ppm, and the covalent hydrate of 6,7-bis(trifluoromethyl)-8-ribityllumazine (Compound 6a, epimer A) could be bound in at least two different states. On the other hand, lumazine protein from Photobacterium phosphoreum, a monomeric paralog of riboflavin synthase without enzyme activity which serves as an optical transponder in bioluminescence showed single signals for each trifluoromethyl group of lumazine analogues [24,25]. Similarly, an artificial N-terminal domain of E. coli riboflavin synthase which forms a c2 symmetric homodimer showed single lines for each trifluoromethyl group of trifluoromethylsubstituted lumazine derivatives [16].

Figs. 5 and 6 show comparative protein perturbation experiments with various orthologs and paralogs of the S. pombe enzyme including lumazine protein (Figs. 5A, 6A), the N-terminal domain of E. coli riboflavin synthase (Figs. 5B, 6B) and full length riboflavin synthases of different organisms (Figs. 5C,5D,5E, 6C,6D,6E). All spectra were recorded in the present study under identical experimental conditions.

thumbnailFigure 5. 19F NMR spectra of 6-trifluoromethyl-7-oxo-8-ribityllumazine (Compound 7). A, lumazine protein of P. leiognathi (0.13 mM protein, 0.25 mM 7); B, recombinant N-terminal domain of E. coli riboflavin synthase (0.80 mM protein, 0.98 mM 7); C, riboflavin synthase of E. coli (0.37 mM protein, 1.43 mM 7); D, riboflavin synthase of B. subtilis (0.10 mM protein, 0.75 mM 7); E, riboflavin synthase of S. pombe (0.28 mM protein, 1.22 mM 7). All samples contained 20 mM phosphate buffer, 100 mM KCl and 10% D2O, pH 7. Signals of the free ligand are indicated by a dashed line. The data were processed with a line broadening of 20 Hz.

thumbnailFigure 6. 19F NMR signals of 6,7-bis(trifluoromethyl)-8-ribityllumazine hydrate (epimer A; Compound 6a). A, lumazine protein of P. leiognathi (0.14 mM protein, 0.23 mM 6a); B, recombinant N-terminal domain of E. coli riboflavin synthase (0.75 mM protein, 0.56 mM 6a); C, riboflavin synthase of E. coli (0.5 mM protein, 3.34 mM 6a); D, riboflavin synthase of B. subtilis (0.13 mM protein, 0.41 mM 6a); E, riboflavin synthase of S. pombe (0.26 mM protein, 0.60 mM 6a); x, impurities. For other details see legend to Fig. 5.

The monotrifluoromethyl derivative, Compound 7, in complex with riboflavin synthase of S. pombe shows four signals at 8.8, 10.9, 15.6 and 16.0 ppm (Fig. 5E). The signal of the free ligand resonates at 7.7 ppm (indicated by a dashed line in Fig. 5) with a line width of 56 Hz, whereas the signals of the ligand bound in different states had apparent line widths extending from 90 to approximately 540 Hz. The riboflavin synthases of E. coli (Fig. 5C) and B. subtilis (Fig. 5D) show similar signal multiplicities, but the chemical shifts and line widths are markedly different. In stark contrast, lumazine protein (Fig. 5A) and the c2 symmetric N-terminal domains of E. coli riboflavin synthase (Fig. 5B) show single and much narrower lines for the protein bound ligand. It is immediately obvious that a given ligand can be bound in multiple, significantly different states by all riboflavin synthases studied, but only in a single state by lumazine protein or the artificial, c2 symmetric N-terminal domain.

A similar set of experiments with the bistrifluoromethyl analog 6a of 6,7-dimethyl-8-ribityl-lumazine is shown in Fig. 6. Due to the presence of six fluorine atoms, the lumazine derivative forms two stable diastereomeric hydrates 6a and 6b which differ by their configuration at the ring carbon atom 7 and are not subject to racemization or dehydration. The stereochemical assignments shown in Fig. 4 were suggested on basis of solid-state NMR studies [26,27].

As shown in Fig. 6A,6B, lumazine protein and the dimeric N-terminal domain of E. coli riboflavin synthase show single lines for each trifluoromethyl group of bound Compound 6a. In contrast to that, the riboflavin synthases (Figs. 6C,6D,6E) studied show multiple signals, and the signal patterns are highly variable for enzymes from different species. The enzyme of S. pombe (Fig. 6E) shows the most complex picture with three major signals for the position 7 trifluoromethyl group and two major signals for the position 6 trifluoromethyl group of the ligand in the bound state. The signals of the free epimer A appeared broadened to 80 Hz – 100 Hz, whereas the signals of the ligand bound in different states had apparent line widths of 170 – 300 Hz.

In an earlier study, we reported 19F protein perturbation studies obtained with a series of mutants of E. coli riboflavin synthase [18]. In that study, it was not possible to express mutant proteins characterized by replacement of the conserved cysteine 48 which is located close to the bound riboflavin in the artificial N-terminal domain dimer [17]. Since a catalytic role had been proposed for that residue [17], a renewed mutagenesis attempt was made with the S. pombe protein. In contrast to our earlier findings with the E. coli enzyme, cystein 48 of the S. pombe enzyme could be replaced by various amino acids, and the proteins could be expressed to high levels. Similarly, mutant proteins with the topologically equivalent S146 residue in the C-terminal domain (cf. Fig. 3) could be expressed.

The catalytic properties of the mutants are summarized in Table 3. Replacement of cystein 48 by serine reduced the catalytic rate of the enzyme by a factor of about 6; C48A and C48M mutants had no detectable enzyme activity. The replacement of serine 146 had only a minor influence on the catalytic rate, in close similarity with our earlier findings with E. coli riboflavin synthase [18]. The replacement of cystein 48 had major impact on ligand perturbation spectra (Fig. 7) with 6-trifluoromethyl-7-oxo-8-ribityllumazine (Compound 7, Fig. 4). When cystein was replaced by less bulky residues (alanine, serine), no 19F NMR signals were noted in the range of 14 – 17 ppm, whereas the wild type protein shows two intense, closely spaced signals at 15.6 and 16.0 ppm. When cystein 48 was replaced by the more bulky methionine, intense signals were observed at 13.8 and 14.0 ppm. Notably, the mutation in the N-terminal domain affects the entire spectrum rather than individual signals. Although a signal is consistently observed at about 8.8 ppm, at least its relative intensity is modulated substantially by the replacement of cystein 48.

Table 2. Oligonucleotide primers used for construction of rib5+ expression plasmid and for site-directed mutagenesis

thumbnailFigure 7. 19F NMR spectra of 6-trifluoromethyl-7-oxo-8-ribityllumazine (Compound 7) in the presence of riboflavin synthase mutants of S. pombe. Concentrations in the NMR samples were as follows: C48M, 0.28 mM protein, 1.10 mM 7; C48A, 0.15 mM protein, 0.57 mM 7; C48S, 0.15 mM protein, 0.80 mM 7; wild type, 0.28 mM protein, 1.22 mM 7; S146C, 0.20 mM protein, 0.92 mM 7; S146A, 0.26 mM protein, 1 mM 7. For other details see legend to Fig. 5.

In the C-terminal domain, serine 146 is the topological equivalent of cystein 48. Although this residue appears to be absolutely conserved, its replacement by the less bulky alanine or the more bulky cystein has relatively little influence on the catalytic rate and the Michaelis constant of the enzyme. However, the replacement of serine 146 had major impact on the protein perturbation spectra (Fig. 7). The signal at 8.8 ppm which is present in the wild type protein and all cystein 48 mutants has an increased intensity in case of the S146A mutant but is absent or shifted in the S146C mutant protein. Again, it should be noted that the replacement of serine 146 affects the entire spectrum rather than individual signals.

In line with the other proteins described in this study, riboflavin synthase of S. pombe binds epimer A believed to represent the 7-R diasteromer, whereas epimer B (7-S) showed no detectable interaction (data not shown). The 6-trifluoromethyl groups of 6-trifluoromethyl-7-oxo-8-ribityl-lumazine and 6,7-bis(trifluoromethyl)-8-ribityllumazine hydrate (epimer A) are topologically equivalent. The 19F NMR signals assigned to that group in the bistrifluoromethyl compound 6a in Figs. 6 and 8 are spread out over a range of only about 4 ppm, as compared to a range of 9 ppm in case of the 7-oxo compound 7, but the impact of mutations is similar. Again, mutation of either cystein 48 or serine 146 affects all signals attributed to bound ligands (Fig. 8).

thumbnailFigure 8. 19F NMR spectra of 6,7-bis(trifluoromethyl)-8-ribityllumazine hydrate (epimer A, Compound 6a) in the presence of riboflavin synthase mutants of S. pombe. Concentrations in the NMR samples were as follows: C48M, 0.30 mM protein, 0.94 mM 6a; C48A, 0.19 mM protein, 0.33 mM 6a; C48S, 0.20 mM protein, 0.42 mM 6a; wild type, 0.26 mM protein, 0.60 mM 6a; S146C, 0.26 mM protein, 0.65 mM 6a; S146A, 0.34 mM protein, 1 mM 6a; x, impurities. For other details see legend to Fig. 5.

6,7-Bis(trifluoromethyl)-8-ribityllumazine hydrate (epimer A) could be displaced from S. pombe protein by addition of riboflavin or the product analog, 5-nitro-6-ribitylamino-2,4(1H,3H)-pyrimidinedione (Compound 9, Fig. 4), thus suggesting that each of the signals, assigned to bound ligand, represented binding sites with specificity for substrate/intermediate analogs rather than non-specific protein surface interactions (data not shown). In the mixture, epimer A displaced from the enzyme by the competing ligands had a linewidth of about 20 Hz.

The complexity of the interaction of the proteins with the ligands is further illustrated by the titration experiments shown in Figs. 9 and 10. In the spectra of the wild type enzyme as well as the C48A mutant, the protein/ligand ratio affects not only the intensity and the line width of the different bands assigned to compound 7 in the protein bound state, but also the chemical shifts of individual signals. Remarkably, some of the peaks appeared with smaller line widths in later stages of the titration experiment (cf. peak at about 9 ppm in Fig. 9). This effect could indicate a combination of substrate induced conformational changes affecting the dynamics of the protein.

thumbnailFigure 9. 19F NMR spectra of 6-trifluoromethyl-7-oxo-8-ribityllumazine (Compound 7) in titration experiments with riboflavin synthase of S. pombe (wild type). The initial concentration of the protein was 0.36 mM. Compound 7 was added to the concentrations indicated. The spectra are displayed with equal intensities of the signals at 15.6 ppm (indicated by *). For other details see legend to Fig. 5.

thumbnailFigure 10. 19F NMR spectra of 6-trifluoromethyl-7-oxo-8-ribityllumazine (Compound 7) in titration experiments with C48A mutant of riboflavin synthase of S. pombe. The initial concentration of the protein was 0.24 mM. Compound 7 was added to the concentrations indicated in the spectra. The spectra are displayed with equal intensities of the signals as indicated by *. For other details see legend to Fig. 5.

Discussion

Lumazine proteins are paralogs of riboflavin synthase occurring in luminescent marine bacteria where they are believed to serve as optical transponders which absorb energy from bacterial luciferase by Förster transfer and are conducive to enhanced luminescence quantum yield as well as modulation of the emission frequency [28,29]. In contrast to the homotrimeric riboflavin synthase which can bind up to 6 substrate molecules, lumazine proteins have been reported to be monomers which bind only one molecule of 6,7-dimethyl-8-ribityllumazine. Although they are likely to share the two-domain architecture of the riboflavin synthase subunit, it appears that only one domain engages in ligand binding. Notably, all riboflavin synthases studied as well as lumazine protein are strictly stereoselective and bind epimer A (Compound 6a, Fig. 4) but not epimer B (Compound 6b) of the diastereomeric 6,7-bis(trifluoromethyl)-8-ribityllumazine hydrate.

Not surprisingly, when the fluorinated lumazines 6a or 7 are bound to lumazine protein, a single 19F NMR signal is observed for each respective trifluoromethyl group (Figs. 5A, 6A). Similarly, the c2 symmetric dimer obtained by recombinant expression of the N-terminal domain of riboflavin synthase shows a single 19F NMR signal for each trifluoromethyl group of a bound ligand (Figs. 5B, 6B).

Since both domains of riboflavin synthase can bind ligand molecules, one would expect to observe separate sets of signals for ligands bound to each respective domain. Beyond that expectation, however, all riboflavin synthases studied in this paper show multiple signals for each ligand studied (Figs. 5C,5D,5E, 6C,6D,6E). Moreover, the signal patterns depend on the ratio of protein and ligand in the solution in unexpected ways (Figs. 9, 10), and the replacement of an amino acid at the C-terminal or N-terminal domain ligand binding site affects virtually all signals attributed to bound ligands. It follows that riboflavin synthase is not c3 symmetric in solution. The asymmetry revealed by the X-ray structure of the E. coli enzyme [15] is thus an inherent property of the enzyme and is not due to crystal contacts.

Based on the NMR experiments with the mutant proteins obtained by replacement of cystein 48 or serine 146, it is now possible to assign 19F signals of the bound ligand in Fig. 7 to the N- or C-terminal domain. Comparison of the spectra in Fig. 7 suggests that the 19F signals shifted to lower field represent ligand molecules which are bound to the N-terminal domain and are therefore predominantly affected by the modification of cystein 48, whereas the signals shifted to higher field represent molecules bound to the C-terminal domain which are more susceptible to the modification of serine 146. However, it must be emphasized that all signals are affected at least to some extent by the mutations.

The X-ray structure of the E. coli enzyme suggests that direct contact exists only between the ligand binding sites of one N-terminal domain and the C-terminal domain of the adjacent subunit [15], whereas the ligand binding sites of two N-terminal and two C-terminal domains are exposed to solvent but not to an adjacent protein subunit. Since a number of studies has indicated that riboflavin synthase can bind certain ligands at a stoichiometry of 1:6 [18,20,22], we assume that each N-terminal and each C-terminal domain can participate in ligand binding.

The major difference is a cysteine residue (position 48 in the N-terminal domain) being replaced by a serine residue (position 146 in the C-terminal domain). Both amino acids are absolutely conserved in all putative riboflavin synthase paralogs. Nevertheless, serine 146 of the S. pombe enzyme can be replaced by alanine with only a minor impact on enzyme activity. Replacement of cysteine 48 by serine reduces the activity by a factor of five, but replacement by alanine affords a soluble protein whose activity, if any, is below the level of detection. It should also be noted that mutant genes specifying E. coli riboflavin synthase mutants carrying alanine or serine instead of cysteine 48 (corresponding to cysteine 48 of the S. pombe enzyme) could not be expressed in recombinant E. coli strains [18]. As seen in the X-ray structure of riboflavin synthase of S. pombe [21] the closest neighbor of the thiol group at position 48 is the 6-methylene group of the substrate analog, 6-carboxyethyl-7-oxo-8-ribityllumazine with a distance of 3.7 A° indicating an essential role of C48 in the catalytic mechanism of riboflavin synthase.

The riboflavin synthase subunit folds into two domains with closely similar folding topology. This had been anticipated on the basis of sequence arguments [14] and has been confirmed by X-ray structure analysis of riboflavin synthase of E. coli (without bound ligand) [15] and of S. pombe (with bound ligand). The monomeric S. pombe riboflavin synthase model arranged in a distinct N-terminal barrel (residues 1–90), an almost identically folded distinc C-terminal barrel (residues 91–184), and a C-terminal α-helix [21].

On basis of these structural arguments, one could expect that ligands bound to C-terminal domains without close interaction with N-terminal domains should not be influenced by conservative changes in the N-terminal domain. More specifically, the replacement of sulfur by oxygen in the C48S mutant should not influence the 19F resonance of ligands bound at topologically remote C-terminal domain. The same argument holds vice versa for the influence of mutations in the C-terminal domain, e.g. the replacement of oxygen by sulfur in the C-terminal domain on ligands bound to distant N-terminal domains.

On basis of the finding that, contrary to that hypothesis, minor modifications of either N-terminal or C-terminal binding sites can modulate all aspects of the 19F signatures of the bound ligands, we propose that riboflavin synthase is subject to large scale dynamic motions occurring on the time scale of the NMR experiments. More specifically, we propose that the pair of N-terminal and C-terminal domain in direct contact as observed in the crystal structure is not permanently associated in solution where the respective domains of different subunits can temporarily assume that special relationship by major domain motions. Previous structural analysis [15] and molecular modeling revealed a markedly nonsymmetric trimeric state of riboflavin synthase. The modeling experiments, based on the inhibitor-bound protein [21], strongly suggest that the close intersubunit contact of the trimer as seen in the crystals of the E. coli protein is indeed in an active conformation and capable of supporting the dismutation reaction. Therefore, this trimer very likely represents a distinct functional state. As there is no reason to suggest a static asymmetric ensemble, it appears likely that thermal fluctuation or substrate binding converts the three different intersubunit contacts into one another, retaining an asymmetric trimer.

Such a dynamic model would fit the wide variation of line widths found for the different protein bound states as well as the unexpected dependence of the 19F NMR spectra on the molar ratio between protein and ligand as revealed in Figs. 9 and 10.

Conclusions

The observations on protein dynamics are highly relevant for the reaction mechanism. Conceivably, the ligand binding sites of N- and C-terminal domains can be loaded with substrate while they are out of physical contact with each other. In order for the dismutation to occur, the domains could then form the local N-terminal/C-terminal domain complex by motion of the entire protein. The products could be unloaded after that domain pair has been separated again by major protein motions, and contacts between other domains could be established which could also result in product formation. In other words, all six domains would participate in the dismutation of the substrate, but they would do so at different times.

A dynamic model involving configuration changes of riboflavin synthase in the context of the catalytic reaction has been proposed earlier on basis of competitive ligand binding [23]. Since the reaction catalyzed by riboflavin synthase occurs on the time of seconds (the turnover number of the S. pombe enzyme is 11 min-1 per trimer), relatively slow dynamic motions occurring on the millisecond time scale would amply suffice the dynamic requirements of that slow catalytic reaction.

Methods

Materials

6,7-Dimethyl-8-ribityllumazine [30], 6-trifluoromethyl-7-oxo-8-ribityllumazine [19], 6-carboxy-ethyl-7-oxo-8-ribityllumazine [31,32] and 6,7-bis(trifluoromethyl)-8-ribityllumazine hydrate (epimers A and B) [19] were prepared by published procedures. Oligonucleotides were custom synthesized by Interactiva, Ulm, Germany. Restriction enzymes were from New England Biolabs, Schwalbach, Germany. T4 DNA ligase and reverse transcriptase (SuperScriptII) were from Gibco BRL, Eggenstein, Germany. Taq Polymerase was from Finnzyme, Epsoo, Finnland. Vent DNA po-lymerase was purchased by New England Biolabs, Schwalbach, Germany. DNA fragments were purified with the QIAquick PCR Purification Kit from Qiagen, Hilden, Germany.

Microorganisms and plasmids

Microorganisms and plasmids used in this study are summarized in Table 1.

Proteins

Recombinant riboflavin synthase of E. coli and B. subtilis were prepared as described earlier [18,23].

Preparation of cDNA

The isolation of RNA from S. pombe var. pombe Lindner (ATCC 16491) and the preparation of cDNA was performed as described earlier [33].

Construction of the rib5+ expression plasmid

A hypothetical open reading frame (accession number T40995) was amplified by PCR using S. pombe cDNA as template and the oligonucleotides SSP-RIB5-1 and SSP-RIB5-2 (Table 2) as primers. The amplificate was digested with EcoRI and BamHI and was ligated into the expression vector pNCO113 which had been digested with the same enzymes yielding the plasmid designated pNCO-SSP-Rosy-WT.

Table 3. Steady state kinetic analysis of wild type and mutant riboflavin synthases

Site directed mutagenesis

A procedure described earlier was used for site-directed mutagenesis [18]. Plasmid pNCO-SSP-Rosy-WT was used as template. The primers used for mutagenesis are shown in Table 2.

Transformation of E. coli cells

E. coli XL-1 Blue cells were transformed with ligation mixtures by published procedures [34]. Transformants were selected on LB agar plates supplemented with ampicillin (170 mg/l). All plasmid constructs were sequenced by the automated dideoxynucleotide method [35] using a 377 Prism DNA sequencer from Applied Biosystems (Weiterstadt, Germany).

pNCO113 type plasmids reisolated from XL-1 Blue cells were transformed into E. coli M15 [pREP4] [36] cells carrying the pREP4 repressor plasmid for the overexpression of lac repressor protein. Kanamycin (15 mg/l) and ampicillin (170 mg/l) were added to secure the maintenance of both plasmids in the host strain. In the pNCO113 expression plasmids, the riboflavin synthase variants are under control of the T5 promotor and the lac operator.

pT7-7 type plasmids reisolated from XL-1 Blue cells were transformed into E. coli BL21 (DE3) cells. The host strain contains a T7 RNA polymerase gene under the control of a lac operator, whereas the riboflavin synthase genes are under control of the T7 promotor.

Protein purification

Recombinant E. coli strains were grown in LB medium containing ampicillin (170 mg/l) and kana-mycin (20 mg/l) at 37°C with shaking overnight. Erlenmeyer flasks containing 500 ml of medium were inoculated at a ratio of 1:50 and were incubated at 37°C with shaking. At an optical density of 0.6 (600 nm), IPTG was added to a final concentration of 2 mM, and incubation was continued for 5 h. The cells were harvested by centrifugation, washed with 0.9 % NaCl and stored at -20°C. Frozen cell mass (6 g) was thawed in 35 ml of 20 mM potassium phosphate, pH 7.0. The suspension was subjected to ultrasonic treatment and was then centrifuged (20 min, 15,000 rpm, 4°C). The supernatant was placed on top of a Q-Sepharose column (2.0 × 7.5 cm, Amersham Phar-macia Biotech, Freiburg, Germany) (flow rate, 5 ml min-1) which had been equilibrated with 200 mM potassium phosphate, pH 7.0. The column was developed with a linear gradient of 0–1.0 M potassium chloride in 20 mM potassium phosphate, pH 7.0 (total volume, 700 ml). Fractions were combined and concentrated by ultrafiltration using Amicon 10 kDa membranes. The solution was placed on top of a Superdex-200 column (2.6 × 60 cm, Amersham Pharmacia Biotech, Freiburg, Germany) (flow rate, 3 ml min-1) which was developed with 20 mM potassium phosphate, pH 7.0, containing 70 mM potassium chloride. Fractions were combined and placed on top of a column of Phenyl-Sepharose CL4B (1.5 × 5.5 cm, Amersham Pharmacia Biotech, Freiburg, Germany) (flow rate, 2 ml min-1) which had been equilibrated with 20 mM potassium phosphate, pH 7.0, containing 500 mM ammonium sulfate. The column was developed with a linear gradient of 0–10 % glycerol containing 20 mM potassium phosphate, pH 7.0 (total volume, 100 ml).

Isolation and purification of lumazine protein

Inclusion bodies were isolated from the recombinant E. coli strain BL21(DE3) [pPHL36] and were solubilized as described earlier [37]. A solution (25 ml) containing 50 mM Tris hydrochloride, pH 7.2, 0.5 mM DTT and solubilized protein was passed through a column of Q Sepharose Fast Flow (2 × 18 cm, Amersham Pharmacia Biotech, Freiburg, Germany) (flow rate, 1 ml min-1). The column was washed with 100 ml of 50 mM Tris hydrochloride, pH 7.2, containing 0.5 mM DTT and was then developed with a linear gradient of 0 – 0.5 M NaCl in the same buffer (total volume, 280 ml). Lumazine protein was eluted from 220 to 260 ml. The enzyme fraction was dialyzed against 50 mM sodium/potassium phosphate, pH 7.0, containing 5 mM 2-mercaptoethanol. The solution was centrifuged for 30 min at 26,000× g, and the supernatant was passed through a column of Superdex 75 (2.6 × 60 cm, Amersham Pharmacia Biotech, Freiburg, Germany) (flow rate, 3 ml min-1). The column was developed with 360 ml of 50 mM phosphate, pH 7.0, containing 5 mM DTT. The enzyme was eluted from 180 to 200 ml. Fractions were combined and concentrated by ultrafiltration.

Estimation of protein concentration

Protein concentration was estimated by a modified Bradford procedure [38].

SDS polyacrylamide gel electrophoresis

Sodium dodecyl sulfate polyacrylamide gel electrophoresis was performed with 16 % polyacryl-amide gels by published procedures [39]. Molecular weight standards were supplied by Sigma (Munich, Germany).

Protein sequencing

Sequence determination was performed by the automated Edman method using a 471 A Protein Sequencer (Perkin Elmer).

Steady state kinetics

Assay mixtures containing 100 mM phosphate, pH 7.2, 10 mM EDTA, 10 mM sodium sulfite, 0.6 mM 6,7-dimethyl-8-ribityllumazine, and protein were incubated at 37°C. Formation of riboflavin was monitored photometrically at 470 nm. The absorbance coefficient for riboflavin is 9,100 M-1 cm-1 at 470 nm [7,40].

Analytical ultracentrifugation

Boundary sedimentation and sedimentation equilibrium experiments were performed using an analytical ultracentrifuge Optima XL-I from Beckman Instruments (Palo Alto, CA) equipped with absorbance and interference optics. Aluminum double sector cells equipped with sapphire windows were used throughout. Protein solutions were dialyzed against 20 mM potassium phosphate, pH 7.0, containing 100 mM potassium chloride. The partial specific volume was estimated from the amino acid composition yielding a value of 0.73 ml g-1 [41]. Sedimentation equilibrium experiments were performed with a solution containing 20 mM potassium phosphate, pH 7.0, 100 mM potassium chloride, and protein (0.7 mg ml-1). Samples were centrifuged at 11,000 rpm and 4°C for 72 h. Protein concentration was monitored photometrically at 280 nm. For boundary sedimentation experiments the protein (1.6 mg ml-1) was centrifuged at 55,000 rpm and 20°C for 2 h. Protein concentration was monitored interferometrically at intervals of 10 min.

Electrospray mass spectrometry

Experiments were performed with a triple quadrupol ion spray mass spectrometer API365 (SciEx, Thornhill, Ontario, Canada) [42].

NMR spectroscopy

19F NMR spectra were recorded at 470 MHz using an AVANCE 500 MHz spectrometer from Bruker Instruments (Karlsruhe, Germany) equipped with a 19F probehead. Experimental parameters were as follows: pulse angle, 70° (10 μsec); repetition rate, 3 sec; 64 K data set; 500 to 20,000 scans. Chemical shifts were referenced to an external standard containing 50 mM sodium trifluoroacetate, pH 7.0. Measurements were performed at 11 °C. Samples contained 20 mM phosphate, pH 7.0, and 100 mM potassium chloride in 10 % D2O.

Authors contributions

The study was designed and coordinated by MF and AB. The study was performed as follows: AKS, BI, KK, MF, WE contributed to the NMR part of the manuscript. MF cloned and overexpressed the protein. RF, AKS purified the protein. MC synthezised the ligands used in this study. BI constructed the mutant proteins. SS, SG, RH, GR participated in the discussion of the study.

Acknowledgements

This work was supported by grants from the Deutsche Forschungsgemeinschaft and the Fonds der Chemischen Industrie. We thank Angelika Werner for expert help with the preparation of the manuscript, Peter Köhler for protein sequencing, Lars Schulte for skillful assistance and Adolphus van Loon for plasmids. Mark Cushman is grateful for the support afforded by NIH Grant GM51469.

References

  1. Bandrin SV, Beburov MI, Rabinovich PM, Stepanov A: Riboflavin auxotrophs of Escherichia coli.

    Genetika 1979, 15:2063-2065. PubMed Abstract OpenURL

  2. Wang A: Isolation of vitamin B2 auxotrophs and preliminary genetic mapping in Salmonella typhimurium.

    Yi Chuan Xue Bao 1992, 19:362-368. PubMed Abstract OpenURL

  3. Oltmanns O, Lingens F: Isolation of riboflavin-deficient mutants of Saccharomyces cerevisiae.

    Z Naturforsch 1967, B22:751-754. OpenURL

  4. Logvinenko EM, Shavlovskii GM, Koltun LV: Preparation and properties of riboflavin-dependent mutants of Pichia guilliermondii Wickerham yeasts.

    Mikrobiologiia 1972, 41:1103-1104. PubMed Abstract OpenURL

  5. Plaut GWE: Studies on the nature of the enzymic conversion of 6,7-dimethyl-8-ribityllumazine to riboflavin.

    J Biol Chem 1963, 238:2225. OpenURL

  6. Harvey RA, Plaut GWE: Riboflavin synthase from yeast: Properties of complexes of the enzyme with lumazine derivatives and riboflavin.

    J Biol Chem 1966, 241:2120-2136. PubMed Abstract | Publisher Full Text OpenURL

  7. Plaut GWE., Beach RL, Aogaichi T: Studies on the mechanism of elimination of protons from the methyl groups of 6,7-dimethyl-6-ribityllumazine by riboflavin synthetase.

    Biochemistry 1970, 9:771-785. PubMed Abstract OpenURL

  8. Beach RL, Plaut GWE: Stereospecificity of the enzymic synthesis of the o-xylene ring of riboflavin.

    J Am Chem Soc 1970, 92:2913-2916. PubMed Abstract OpenURL

  9. Paterson T, Wood HC: Deuterium exchange of C-methyl protons in 6,7-dimethyl-8-D-ribityllumazine, and studies of the mechanism of riboflavin biosynthesis.

    J Chem Soc Chem Commun 1969, 290-291. OpenURL

  10. Sedlmaier H, Müller F, Keller PJ, Bacher A: Enzymatic synthesis of riboflavin and FMN specifically labeled with 13C in the xylene ring.

    Z Naturforsch 1987, 42c:425-429. OpenURL

  11. Illarionov B, Eisenreich W, Bacher A: A pentacyclic reaction intermediate of riboflavin synthase.

    Proc Natl Acad Sci USA 2001, 98:7224-7229. PubMed Abstract | Publisher Full Text | PubMed Central Full Text OpenURL

  12. Bacher A, Baur R, Eggers K, Harders H, Otto MK, Schnepple H: Riboflavin synthases of Bacillus subtilis. Purification and properties.

    J Biol Chem 1980, 255:632-637. PubMed Abstract | Publisher Full Text OpenURL

  13. Eberhardt S, Richter G, Gimbel W, Werner T, Bacher A: Cloning, sequencing, mapping and hyperexpression of the ribC gene coding for riboflavin synthase of Escherichia coli.

    Eur J Biochem 1996, 242:712-719. PubMed Abstract OpenURL

  14. Schott K, Kellermann J, Lottspeich F, Bacher A: Riboflavin synthases of Bacillus subtilis. Purification and amino acid sequence of the α subunit.

    J Biol Chem 1990, 265:4204-4209. PubMed Abstract | Publisher Full Text OpenURL

  15. Liao DI, Wawrzak Z, Calabrese JC, Viitanen PV, Jordan DB: Crystal structure of riboflavin synthase.

    Structure 2001, 9:399-408. PubMed Abstract | Publisher Full Text OpenURL

  16. Eberhardt S, Zingler N, Kemter K, Richter G, Cushman M, Bacher A: Domain structure of riboflavin synthase.

    Eur J Biochem 2001, 268:4315-4323. PubMed Abstract | Publisher Full Text OpenURL

  17. Truffault V, Coles M, Diercks T, Abelmann K, Eberhardt S, Lüttgen H, Bacher A, Kessler H: The solution structure of the N-terminal domain of riboflavin synthase.

    J Mol Biol 2001, 309:949-960. PubMed Abstract | Publisher Full Text OpenURL

  18. Illarionov B, Kemter K, Eberhardt S, Richter G, Cushman M, Bacher A: Riboflavin synthase of Escherichia coli. Effect of single amino acid substitutions on reaction rate and ligand binding properties.

    J Biol Chem 2001, 276:11524-11530. PubMed Abstract | Publisher Full Text OpenURL

  19. Cushman M, Patrick DA, Bacher A, Scheuring J: Synthesis of epimeric 6,7-bis(trifluoromethyl)-8-ribityllumazine hydrates. Stereoselective interaction with the light riboflavin synthase of Bacillus subtilis.

    J Org Chem 1991, 56:4603-4608. OpenURL

  20. Cushman M, Patel HH, Scheuring J, Bacher A: F NMR Studies on the mechanism of riboflavin synthase. Synthesis of 6-(Trifluoromethyl)-7-oxo-8-(D-ribityl)lumazine and 6-(Trifluoromethyl)-7-methyl-8-(D-ribityl)lumazine.

    J Org Chem 1992, 57:5630-5643. OpenURL

  21. Gerhardt S, Schott A, Kairies N, Cushman M, Illarionov B, Eisenreich W, Bacher A, Huber R, Steinbacher S, Fischer M: Studies on the reaction mechanism of riboflavin synthase. X-ray crystal structure of a complex with 6-carboxyethyl-7-oxo-8-ribityllumazine.

    Structure (Camb) 2002, 10:1371-1381. PubMed Abstract | Publisher Full Text OpenURL

  22. Otto MK, Bacher A: Ligand-binding studies on light riboflavin synthase from Bacillus subtilis.

    Eur J Biochem 1981, 115:511-517. PubMed Abstract OpenURL

  23. Scheuring J, Fischer M, Cushman M, Lee J, Bacher A, Oschkinat H: NMR analysis of site-specific binding in oligomeric proteins. Dynamic studies on the interaction of riboflavin synthase with trifluoromethyl-substituted intermediates.

    Biochemistry 1996, 35:9637-9646. PubMed Abstract | Publisher Full Text OpenURL

  24. Scheuring J, Lee J, Cushman M, Oschkinat H, Bacher A: 19F NMR studies on lumazine protein from Photobacterium phosphoreum.

    In In: Flavins and Flavoproteins Edited by Yagi K, Walter de Gruyter. 1994, 75-78. OpenURL

  25. Scheuring J, Lee J, Cushman M, Oschkinat H, Patel H, Patrick DA, Bacher A: Trifluoromethyllumazine derivatives as 19F NMR probes for lumazine protein.

    Biochemistry 1994, 33:7634-40. PubMed Abstract OpenURL

  26. Goetz JM, Poliks B, Studelska DR, Fischer M, Kugelbrey K, Bacher A, Cushman M, Schäfer J: Investigation of the binding of fluorolumazines to the 1-MDa capsid of lumazine synthase by 15N{19F} REDOR NMR.

    J Am Chem Soc 1999, 121:7500-7508. Publisher Full Text OpenURL

  27. Metha AK, Studelska DR, Fischer M, Giessauf A, Kemter K, Bacher A, Cushman M, Schaefer J: Investigation of the binding of epimer A of the covalent hydrate of 6,7-bis(trifluoromethyl)-8-D-ribityllumazine to a recombinant F22W Bacillus subtilis lumazine synthase mutant by 15N{19F}REDOR NMR.

    J Org Chem 2002, 67:2087-2092. PubMed Abstract | Publisher Full Text OpenURL

  28. Gast R, Neering IR, Lee J: Separation of a blue fluorescence protein from bacterial luciferase.

    Biochem Biophys Res Commun 1978, 80:14-21. PubMed Abstract OpenURL

  29. O'Kane DJ, Lee J: Physical characterization of lumazine proteins from Photobacterium.

    Biochemistry 1985, 24:1484-1488. PubMed Abstract OpenURL

  30. Bacher A, Neuberger G, Volk R: Enzymatic synthesis of 6,7-dimethyl-8-ribityllumazine.

    In In: Chemistry and Biology of Pteridines Edited by Cooper BA, Whitehead VM, Walter de Gruyter, Berlin. 1986, 227-230. OpenURL

  31. Wood HC, Wrigglesworth R, Yeowell DA, Gurney FW: Specific enzyme inhibitors in vitamin biosynthesis. II. Revised structures for some 8-substituted pyrido(2,3-d)-pyrimidines.

    J Chem Soc 1974, 11:1225-30. OpenURL

  32. Al-Hassan SS, Kuhol RJ, Livingstone DB, Suckling CJ, Wood HCS, Wrigglesworth R, Ferone R: Specific enzyme inhibitors in vitamin biosynthesis. Part 3. The synthesis and inhibitory properties of some substrate and transition state analogs of riboflavin synthase.

    J Chem Soc Perkin Trans I 1980, 2645-2656. OpenURL

  33. Fischer M, Haase I, Feicht R, Richter G, Gerhardt S, Changeux JP, Huber R, Bacher A: Biosynthesis of Riboflavin. 6,7-Dimethyl-8-ribityllumazine synthase of Schizosaccharomyces pombe.

    Eur J Biochem 2002, 269:519-26. PubMed Abstract | Publisher Full Text OpenURL

  34. Bullock WO, Fernandez JM, Short JM: XL1-blue: a high efficiency plasmid transforming recA Escherichia coli strain with β-galactosidase selection.

    BioTechniques 1987, 5:376-379. OpenURL

  35. Sanger F, Niklen S, Coulson AR: DNA sequencing with chain-terminating inhibitors.

    Proc Natl Acad Sci USA 1977, 74:5463-5467. PubMed Abstract OpenURL

  36. Stüber D, Matile H, Garotta G: System for high-level production in Escherichia coli and rapid purification of recombinant proteins: application to epitop mapping, preparation of antibodies, and structure-function analysis.

    In In: Immunological Methods Edited by Lefkovits I, Pernis P. 1990, IV:121-152. OpenURL

  37. Illarionov B, Illarionova V, Lee J, van Dongen W, Vervoort J: Expression and properties of the recombinant lumazine (riboflavin) protein from Photobacterium leiognathi.

    Biochem Biophys Acta 1994, 1201:251-258. PubMed Abstract | Publisher Full Text OpenURL

  38. Read SM, Northcote DH: Minimization of variation in the response to different proteins of the Coomassie blue G dye-binding assay for protein.

    Anal Biochem 1981, 116:53-64. PubMed Abstract OpenURL

  39. Läemmli UK: Cleavage of structural proteins during the assembly of the head of bacteriophage T4.

    Nature 1970, 227:680-685. PubMed Abstract OpenURL

  40. Plaut GWE, Harvey RA: The enzymic synthesis of riboflavin.

    In: Methods Enzymol 1971, 18B:515-538. OpenURL

  41. Laue TM, Shah BD, Ridgeway TM, Pelletier SL: In: Analytical ultracentrifugation in biochemistry and polymer science.. Edited by Harding SE, Rowe AJ, Horton JC. Royal Society of Chemistry, Cambridge; 1992:90-125. OpenURL

  42. Mann M, Wilm M: electrospray mass spectrometry for protein characterization.

    Trends Biochem Sci 1995, 20:219-224. PubMed Abstract | Publisher Full Text OpenURL

  43. Zamenhof PJ, Villarejo M: Construction and properties of Escherichia coli strains exhibiting α-complementation of β-galactosidase fragments in vivo.

    J Bacteriol 1972, 110:171-178. PubMed Abstract | PubMed Central Full Text OpenURL

  44. Studier FW: Use of bacteriophage T7 lysozyme to improve an inducible T7 expression system.

    J Mol Biol 1991, 219:37-44. PubMed Abstract OpenURL

  45. Tabor S: Expression using the T7 RNA polymerase/promoter system. In Current Protocols of Molecular Biology. Green Publishing and Wiley Interscience, NY; 1990:1621. OpenURL