Brief Communication


Nature Chemical Biology 4, 739 - 741 (2008)
Published online: 26 October 2008 | doi:10.1038/nchembio.123

A concerted mechanism for berberine bridge enzyme

Andreas Winkler1, Andrzej L strokeyskowski2,4, Sabrina Riedl1,4, Martin Puhl1, Toni M Kutchan3, Peter Macheroux1 & Karl Gruber2


Berberine bridge enzyme catalyzes the conversion of (S)-reticuline to (S)-scoulerine by formation of a carbon-carbon bond between the N-methyl group and the phenolic ring. We elucidated the structure of berberine bridge enzyme from Eschscholzia californica and determined the kinetic rates for three active site protein variants. Here we propose a catalytic mechanism combining base-catalyzed proton abstraction with concerted carbon-carbon coupling accompanied by hydride transfer from the N-methyl group to the N5 atom of the FAD cofactor.


Berberine bridge enzyme (BBE) has a central role in benzophenanthridine biosynthesis and channels its substrate (S)-reticuline (1Compound 1) toward berberine (2Compound 2) and sanguinarine (3Compound 3) generation1. BBE catalyzes the stereospecific conversion of (S)-reticuline to (S)-scoulerine (4Compound 4), thereby forming the so-called berberine bridge (C8) by linking the isoquinoline ring with the benzyl ring system (Fig. 1a)1. This cyclization reaction has no known equivalent in organic chemistry and is proposed to proceed via a two-step process starting with oxidation of the N-methyl group to the corresponding iminium ion and followed by an ionic ring closure initiated by the deprotonated C3' OH group of the benzyl moiety2. This hydroxyl group apparently plays an important role for the enzymatic reaction, as substrate analogs lacking this group are not processed by the enzyme2. This finding suggests that an active site base may be involved in the reaction mechanism by deprotonation of the hydroxyl group, thereby facilitating the nucleophilic attack of the C2' carbon onto the N-methyl group. A similar reaction mechanism involving the intermediate formation of an iminium species is favored for amine oxidases, a diverse group of flavin-dependent enzymes (reviewed in refs. 3,4,5).

Figure 1: Reaction and structure of BBE.

Figure 1 : Reaction and structure of BBE.

(a) Overall reaction catalyzed by BBE1 (Enzyme Commission number 1.21.3.3). (b) Schematic representation of the protein structure. The N-terminal FAD-binding subdomains are shown in blue and green (including the C-terminal alpha-helical stretch in light green), and the central substrate binding domain is shown in magenta. N-linked sugar residues (blue) and the FAD cofactor (orange) are represented as stick models. The amino acids involved in the bicovalent linkage of FAD are shown in green. (c) Active site environments of the structures from the monoclinic crystals showing polar amino acids as dark green stick models. Alternate conformations were observed for Glu417. The flavin cofactor is shown in orange with its dual mode of attachment to the protein backbone via His104 and Cys166 represented in green. (d) Interactions between the substrate and active site amino acids (green). The substrate is shown in yellow, and FAD is shown in orange. Distances are indicated in Å.

Full size image (84 KB)

In order to better understand the mechanism of action for the BBE-catalyzed reaction, we elucidated the three-dimensional X-ray crystal structure of the enzyme from Eschscholzia californica in two different crystal forms (monoclinic and tetragonal) and in complex with the natural substrate (S)-reticuline (see Supplementary Methods, Supplementary Discussion and Supplementary Table 1 online). Based on this structure, we initiated a mutagenesis program involving three active site amino acids and analyzed their function.

The resulting molecular structure comprises two domains: a flavin adenine dinucleotide (FAD, 5Compound 5) binding domain and an alpha/beta domain with a seven-stranded, antiparallel beta-sheet forming the substrate binding site (Fig. 1b). This structure demonstrates that BBE belongs to the p-cresol methylhydroxylase (PCMH) superfamily (Supplementary Fig. 1 online). It also confirms the previously identified6 bicovalent attachment of the flavin cofactor to N1 of His104 (via C8alpha) and to Cys166 (via C6) (Fig. 1c and Supplementary Fig. 2 online), thus providing an additional example for the recently discovered group of bicovalently flavinylated enzymes6, 7, 8, 9, 10. The active site cavity close to the cofactor is lined by mainly hydrophobic residues, the only notable exceptions being Tyr106, Thr358, Asn390, Glu417 and His459 (Fig. 1c). The latter two residues show considerable conformational flexibility: Glu417 is present in two alternate conformations in the monoclinic structure (Fig. 1c), whereas His459 shows two conformations in the tetragonal structure (Supplementary Fig. 2). During the course of crystallization, we observed a bleaching of the yellow colored crystals. In line with that, we found flavin electron density indicating the (partial) presence of the colorless 4a-spirohydantoin degradation product11, 12 (6Compound 6), most notably in the much slower growing monoclinic crystals (see Supplementary Discussion and Supplementary Fig. 2).

We obtained the structure of the complex with (S)-reticuline by soaking a monoclinic crystal in a cryosolution containing 20 mM of the substrate. Despite the lower resolution of the diffraction data (2.8 compared with 1.65 Å for the unbound structure), we observed clear difference electron density in the binding site (Supplementary Fig. 3 online), thus allowing the unequivocal positioning of the substrate (Fig. 1d). (S)-Reticuline is bound in a deep cleft with the phenolic ring pointing toward the bottom of the binding site (Supplementary Fig. 3) and is sandwiched between the flavin cofactor and amino acid residues extending from the beta-sheet of the central domain (Fig. 1b). The active site residue Glu417 is hydrogen bonded to the C3' OH group of the substrate, and the carboxamide group of Asn390 interacts with both the OH and the 4'-methoxy group of the phenolic ring (Fig. 1d). The corresponding polar substituents of the isoquinoline moiety are positioned close to Asp352 at the entrance to the active site. In the conformation observed in the complex structure, the N-methyl group of (S)-reticuline is sequestered between the C2' atom of the substrate and the N5 atom of the cofactor, with distances of 3.3 Å and 3.1 Å, respectively (Fig. 1d). The angle between these three atoms is approximately 130°.

Based on the structural information, we selected three amino acids (Tyr106, Glu417 and His459) for site-directed mutagenesis. In contrast to Glu417, Tyr106 and His459 are not directly interacting with the substrate. The estimated pKa of Tyr106, however, is in the range of the pH optimum of BBE1, and the proximity of His459 to the other selected residues (Fig. 1d) may modulate their properties. All three muteins could be expressed in roughly similar amounts as the wild-type protein6. Spectral characterization of the muteins indicated that the bicovalent flavin linkage was not affected by any of the three amino acid replacements6 and that the proteins were fully loaded with cofactor (see Supplementary Discussion).

We analyzed the catalytic efficiency of the protein variants by measuring steady state turnover rates as well as reductive and oxidative rates in comparison to those of the wild-type enzyme (Table 1). The most significant change in the observed rates was found for the E417Q mutein (see Supplementary Fig. 4 online), whereas Y106F and H459A showed only minor alterations (Supplementary Discussion). In E417Q, the reductive rate was reduced 1,500-fold compared with that of the wild-type enzyme. Cofactor reduction became the rate-limiting step during turnover, but no effect on reoxidation of the cofactor was observed (Table 1). This finding clearly supports the role of the glutamate side chain as the catalytic base necessary for carbon-carbon bond formation—which is consistent with the structure of the substrate complex (Fig. 1d)—and also demonstrates its involvement in the oxidative part of the reaction. The residual activity of this variant is most likely due to a fraction of deprotonated substrate molecules (estimated pKa of C3' OH: approx9.5) present at the pH of the measurement (Table 1).


The essential role of proton abstraction for substrate oxidation was further corroborated by the approx30,000-fold slower reduction of the cofactor by the substrate analog laudanosine (7Compound 7), which bears a methoxy instead of a hydroxyl group at the C3' position (Supplementary Fig. 5 online). The appearance of an absorbance maximum around 360 nm indicated the formation of a double bond in conjugation to the aromatic ring system of the isoquinoline moiety, which suggests that the C1-N bond rather than the exocyclic N-CH3 bond is oxidized. This type of reactivity was verified by NMR studies for the substrate analog 6,7-dimethoxy-2-methyl-1,2,3,4-tetrahydroisoquinoline (8Compound 8), the conversion of which is accompanied by similar spectral changes (Supplementary Discussion).

Small amounts of formaldehyde (9Compound 9, approx2% during prolonged incubation times) arise in the conversion of the substrate analog (S)-N-methylcoclaurine (10Compound 10)2, which also lacks the C3' OH group. This finding is considered as supporting evidence for the two-step mechanism2. In light of our results with laudanosine and 8Compound 8, however, it could also be explained by the formation of the intracyclic imine, followed by tautomerization and hydrolytic demethylation to yield formaldehyde and (S)-coclaurine (11Compound 11).

The fact that Glu417 is an essential amino acid for substrate oxidation has further noteworthy mechanistic implications. Given that this amino acid does not interact with the N-methyl group but rather with the more distant C3' OH group of the substrate, it is reasonable to propose that deprotonation of the phenolic OH group increases the nucleophilicity of the C2' atom, thereby facilitating an SN2-type attack onto the N-methyl group. This leads to carbon-carbon bond formation and concomitant transfer of a hydride to the flavin. Hence, flavin reduction and berberine bridge formation are likely to be concerted processes initiated by proton abstraction of the phenolic hydroxyl group (Scheme 1). Even though the N-methyl group of the substrate is bound in proximity of the N5 atom of FAD (Fig. 1d) and is appropriately positioned for hydride transfer (as in other flavin-dependent amine oxidases13, 14), the E417Q mutant protein does not significantly catalyze flavin reduction (Table 1), which provides clear evidence for this concerted mechanism. Similarly, the inability of wild-type BBE to efficiently oxidize the substrate analog laudanosine also underlines the importance of the interaction of Glu417 with the C3' OH group of (S)-reticuline. Unless substantial protein dynamics are involved, the observed mode of substrate binding also excludes the proposed mechanism for amine oxidation observed in monoamine and polyamine oxidases3, 15, including the formation of a covalent adduct between flavin and substrate, as the amine function is too distant from the C4a atom (approx5 Å) of the cofactor for this type of mechanism (see Fig. 1d). Therefore, the BBE-catalyzed reaction cannot involve an iminium intermediate as in the two-step mechanism proposed in previous studies2, and hence BBE mechanistically diverges from other amine oxidases3, 4, 5.

Scheme 1: Proposed reaction mechanism for the BBE-catalyzed reaction.

Scheme 1 : Proposed reaction mechanism for the BBE-catalyzed reaction. Unfortunately we are unable to provide accessible alternative text for this. If you require assistance to access this image, or to obtain a text description, please contact npg@nature.com

Reduction of the cofactor is initiated by deprotonation of C3' OH by Glu417. This increases the nucleophilic behavior of the aromatic ring, thereby allowing carbon-carbon bond formation in a Friedel-Crafts–like alkylation reaction. The formal leaving group, a hydride, is at the same time incorporated into the flavin cofactor. Rearomatization of the intermediate substance leads to (S)-scoulerine and might occur spontaneously in solution or could be assisted by another active site base (depicted as B). Regeneration of the cofactor is accomplished by reaction with molecular oxygen and generation of hydrogen peroxide.

Full size image (52 KB)

We also found that replacement of Glu417 by glutamine leads to a loss of regiospecificity. Wild-type BBE and all other mutant proteins analyzed so far converted (S)-reticuline only to (S)-scoulerine2 and produced no potential side product, for example, via carbon-carbon coupling involving C6' instead of C2' after a rotation of the phenyl ring. Substrate conversion by the E417Q mutein, however, yielded an additional compound amounting to about one third of the total product. This side product coeluted with an authentic standard of (S)-coreximine (12Compound 12), the product analog obtained after a 180° rotation of the phenolic ring (Supplementary Fig. 6 online).

Through a combined structural and biochemical approach, we are now in a position to propose a new reaction mechanism (Scheme 1) that entails proton abstraction of the C3' OH group by the carboxyl group of Glu417, leading to SN2-type attack onto the N-methyl group and resulting in the formation of the berberine bridge (Fig. 1a). Carbon-carbon bond formation is associated with expulsion of a hydride from the methyl group to the flavin N5 position, leading to a two-electron reduction of the cofactor. In contrast to previous findings2, this mechanistic proposal incorporates both oxidation and ring closure into a single concerted step. The SN2-type mechanism is in agreement with earlier studies demonstrating inversion of configuration at the N-methyl group16. The proposed hydride transfer to the flavin is in line with the absence of any flavin radical intermediate during substrate-mediated reduction of the enzyme in stopped-flow experiments17 (see also Supplementary Fig. 4).

These new mechanistic ideas are potentially also relevant for two closely related enzymes: Delta1-tetrahydrocannabinolic acid (THCA) synthase and cannabidiolic acid (CBDA) synthase from Cannabis sativa18, 19. Both enzymes also catalyze an oxidative carbon-carbon coupling reaction, which is proposed to proceed via a two-step mechanism, where ring closure is an ionic process following the oxidation of the substrate18, 19. Owing to the similarity of the catalyzed reactions and the substantial sequence similarity to BBE (approx40% identity), it is conceivable that these two enzymes operate by an analogous one-step concerted mechanism. Sequence alignments indeed indicate Glu442 in these enzymes to be equivalent to the proposed active site base (Glu417) in BBE (Supplementary Fig. 7 online). The actual role of these residues, however, will have to be demonstrated by mutagenesis and/or structural studies.

Note: Supplementary information and chemical compound information is available on the Nature Chemical Biology website.



Top

Acknowledgments

We appreciate the support of staff scientists at the synchrotron beamlines at DESY/EMBL-Hamburg during diffraction data collection and of Hansjörg Weber (Institute of Organic Chemistry, Graz University of Technology) for recording the NMR spectra. Financial support was provided by the Austrian Science Fund (Fonds zur Förderung der wissenschaftlichen Forschung) through the Doktoratskolleg "Molecular enzymology" W901-B05 (to K.G. and P.M.).

Author Contributions

A.W., P.M. and K.G. designed experiments and analyzed results. A.W., P.M. and K.G. wrote the manuscript. A.W., A.L. and K.G. determined the crystal structures. A.W., S.R. and M.P. generated, expressed and purified mutant proteins. A.W. and M.P. performed enzyme kinetic experiments. T.M.K. prepared and provided special compounds and helped in editing the manuscript.

Received 6 June 2008; Accepted 7 October 2008; Published online 26 October 2008.

Top

References

  1. Steffens, P., Nagakura, N. & Zenk, M.H. Phytochem. 24, 2577–2583 (1985). | Article | ChemPort |
  2. Kutchan, T.M. & Dittrich, H. J. Biol. Chem. 270, 24475–24481 (1995). | Article | PubMed | ChemPort |
  3. Edmondson, D.E., Binda, C. & Mattevi, A. Arch. Biochem. Biophys. 464, 269–276 (2007). | Article | PubMed | ChemPort |
  4. Scrutton, N.S. Nat. Prod. Rep. 21, 722–730 (2004). | Article | PubMed
  5. Fitzpatrick, P.F. Bioorg. Chem. 32, 125–139 (2004). | Article | PubMed | ChemPort |
  6. Winkler, A., Hartner, F., Kutchan, T.M., Glieder, A. & Macheroux, P. J. Biol. Chem. 281, 21276–21285 (2006). | Article | PubMed | ChemPort |
  7. Huang, C.-H. et al. J. Biol. Chem. 280, 38831–38838 (2005). | Article | PubMed | ChemPort |
  8. Alexeev, I., Sultana, A., Mantsala, P., Niemi, J. & Schneider, G. Proc. Natl. Acad. Sci. USA 104, 6170–6175 (2007). | Article | PubMed | ChemPort |
  9. Heuts, D.P.H.M., Winter, R.T., Damsma, G.E., Janssen, D.B. & Fraaije, M.W. Biochem. J. 413, 175–183 (2008). | Article | PubMed | ChemPort |
  10. Rand, T., Qvist, K.B., Walter, C.P. & Poulsen, C.H. FEBS J. 273, 2693–2703 (2006). | Article | PubMed | ChemPort |
  11. Harayama, T., Tezuka, Y., Taga, T. & Yoneda, F. J. Chem. Soc. [Perkin 1] 75–83 (1987). | Article
  12. Iwata, M., Bruice, T.C., Carrell, H.L. & Glusker, J.P. J. Am. Chem. Soc. 102, 5036–5044 (1980). | Article | ChemPort |
  13. Settembre, E.C. et al. Biochemistry 42, 2971–2981 (2003). | Article | PubMed | ChemPort |
  14. Malito, E., Coda, A., Bilyeu, K.D., Fraaije, M.W. & Mattevi, A. J. Mol. Biol. 341, 1237–1249 (2004). | Article | PubMed | ChemPort |
  15. Miller, J.R. & Edmondson, D.E. Biochemistry 38, 13670–13683 (1999). | Article | PubMed | ISI | ChemPort |
  16. Bjorklund, J.A. et al. J. Am. Chem. Soc. 117, 1533–1545 (1995). | Article | ChemPort |
  17. Winkler, A., Kutchan, T.M. & Macheroux, P. J. Biol. Chem. 282, 24437–24443 (2007). | Article | PubMed
  18. Sirikantaramas, S. et al. J. Biol. Chem. 279, 39767–39774 (2004). | Article | PubMed | ChemPort |
  19. Taura, F. et al. FEBS Lett. 581, 2929–2934 (2007). | Article | PubMed | ChemPort |
  1. Institute of Biochemistry, Graz University of Technology, Petersgasse 12/II, 8010 Graz, Austria.
  2. Institute of Molecular Biosciences, University of Graz, Humboldtstras zlige 50/III, 8010 Graz, Austria.
  3. Donald Danforth Plant Science Center, 975 North Warson Road, St. Louis, Missouri 63132, USA
  4. Present addresses: Institute of Biotechnology, PO Box 65, Viikinkaari 1, FIN-00014 University of Helsinki, Finland (A.L.) and Institute of Biophysics and Nanosystems Research, Austrian Academy of Sciences, Schmiedlstras zlige 6, 8042 Graz, Austria (S.R.).

Correspondence to: Sabrina Riedl1,4 e-mail: peter.macheroux@tugraz.at

Correspondence to: Karl Gruber2 e-mail: karl.gruber@uni-graz.at

MORE ARTICLES LIKE THIS

These links to content published by NPG are automatically generated.

NEWS AND VIEWS

Cyclization in concert

Nature Chemical Biology News and Views (01 Dec 2008)