FEBS Letters 589 (2015) 880–884
journal homepage: www.FEBSLetters.org
Unusual non-enzymatic flavin catalysis enhances understanding of flavoenzymes Erick A. Argueta, Amanda N. Amoh, Prapti Kafle, Tanya L. Schneider ⇑ Department of Chemistry, Connecticut College, 270 Mohegan Avenue, New London, CT 06320, USA
a r t i c l e
i n f o
Article history: Received 7 January 2015 Revised 24 February 2015 Accepted 25 February 2015 Available online 5 March 2015 Edited by Michael Ibba Keywords: Flavin Oxidase Non-ribosomal peptide synthesis Biosynthesis Thiazole
a b s t r a c t Flavin cofactors are central to many biochemical transformations and are typically tightly bound as part of a catalytically active flavoenzyme. This work indicates that naturally occurring flavins can act as stand-alone catalysts to promote the oxidation of biosynthetically inspired heterocycles in aqueous buffers. Flavin activity was compared with that of oxidases important in non-ribosomal peptide synthesis, providing a rare direct comparison between the catalytic efficacy of flavins alone and in the context of a full flavoenzyme. This study suggests that such oxidases are likely to possess an active site base, as oxidase activity was greater than that of flavins alone, particularly for less acidic substrates. These findings offer perspective on the development of robust and catalytically effective, designed miniature flavoenzymes. Ó 2015 Federation of European Biochemical Societies. Published by Elsevier B.V. All rights reserved.
1. Introduction Flavins are redox cofactors central to numerous metabolic processes [1]. As such, they are nearly always tightly associated with flavoenzymes, and this complex together is responsible for the catalysis of a wide range of biochemical transformations [2]. While flavoenzymes are well-known in primary metabolic pathways, they also play an important role in secondary metabolism [3]. Flavin-binding oxidases are located in biosynthetic pathways yielding nitrogen-containing heterocycles such as thiazoles and oxazoles. These heteroaromatic rings are formed from posttranslational modification of serine, threonine and cysteine in the ribosomal production of peptides like microcin B17 [4] and are also found in non-ribosomal peptides formed by non-ribosomal peptide synthetases (NRPS) [5]. Oxidases in such biosynthetic pathways facilitate the oxidation of thiazoline and oxazoline rings to the more stable, aromatic thiazole and oxazole rings, often necessary for the bioactivity of the natural product. Because these nitrogencontaining heteroaromatic rings are of interest, particularly from a medicinal chemistry perspective, their formation has been investigated.
Author contributions: E.A.A. and T.L.S. designed and performed catalyst assays. A.N.A. and P.K. synthesized and characterized small molecules. T.L.S. supervised the project and wrote the manuscript with input from E.A.A., A.N.A. and P.K. ⇑ Corresponding author. Fax: +1 860 439 2477. E-mail address: [email protected] (T.L. Schneider).
The NRPS oxidases important for the biosynthesis of the anticancer agents epothilone and bleomycin have been characterized, with catalytic activity for a series of thiazoline and oxazoline substrate mimics measured [6]. These oxidases were found to bind flavin mononucleotide (FMN) non-covalently but so tightly that denaturation is necessary to separate the two. Flavin cofactors are not usually exchanged after catalytic turnover and tend to be regenerated while enzyme-bound, requiring molecular oxygen or other oxidants. Because flavins can be simply regenerated by molecular oxygen, and are linked to the formation of bioactive small molecules, they are intriguing targets for stand-alone bioinspired catalysts. Some synthetic flavin analogs have been evaluated, suggesting their utility as effective and environmentally mild oxidants for synthetic transformations [7]. Hydroperoxyflavins recently have been shown to promote the formation of heteroaromatic pyridines and benzothiazoles at ambient temperature in methanol and in the presence of molecular oxygen [8]. This report joins other work describing synthetic flavins capable of catalyzing the oxidation of amines, sulfides and aldehydes [9–11]. These synthetic oxidation reactions take advantage of the redox potential of flavins as ‘‘green’’ catalysts while removing the enzyme portion of flavoenzymes which can be more susceptible to denaturation under varied solvent conditions. In this work, the potential of naturally occurring flavins as stand-alone catalysts mediating the oxidation of biosynthetically inspired small molecules under physiologically relevant conditions
http://dx.doi.org/10.1016/j.febslet.2015.02.034 0014-5793/Ó 2015 Federation of European Biochemical Societies. Published by Elsevier B.V. All rights reserved.
E.A. Argueta et al. / FEBS Letters 589 (2015) 880–884
was probed. FMN, flavin adenine dinucleotide (FAD), and riboflavin all catalyze the formation of heteroaromatic rings in thioesterlinked small molecules known to be substrates for NRPS oxidases. As these reactions occur in aqueous buffers mirroring the reaction conditions needed for oxidase activity, this study allows a rare direct comparison between the catalytic efficacy of flavins alone and in the context of the full flavoenzyme. This analysis provides mechanistic insights on the catalytic role of the oxidase enzyme beyond the bound flavin cofactor and offers perspective on the development of robust and catalytically effective, designed miniature flavoenzymes. 2. Materials and methods 2.1. Reagents and enzymes FMN, FAD and riboflavin were purchased from Sigma–Aldrich. Reagents for the synthesis of substrates and product standards were also purchased from Sigma–Aldrich. Other reagents were purchased from Fisher Scientific unless otherwise specified. Overexpression and purification to yield EpoB-Ox were performed as described [6]. To determine flavin-bound enzyme concentration, a sample of purified EpoB-Ox was heat denatured for 2 min at 90 °C, and the supernatant was subjected to UV–vis analysis. Concentration was calculated based on the known extinction coefficient for FMN, 12 200 cm1 M1 at 450 nm [12]. 2.2. Chemical synthesis of substrates and standards Phenylthiazoline carboxylic acid, phenylthiazole carboxylic acid, phenylthiazolinyl-S-NAC, phenylthiazolyl-S-NAC, phenyloxazolinyl-S-NAC, phenyloxazolyl-S-NAC, methylthiazolinyl-SNAC and methylthiazolyl-S-NAC were synthesized as described [6,13,14]. Hydrocinnamyl-S-NAC was prepared by combining hydrocinnamic acid (73.0 mg, 486 lmol), N-acetylcysteamine (NAC; 78 lL, 730 lmol), N-(3-dimethylaminopropyl)-N0 -ethylcarbodiimide hydrochloride (EDC; 140 mg, 730 lmol) and triethylamine (TEA; 102 lL, 732 lmol) in dichloromethane. After stirring at 25 °C for 18 h, the reaction was evaporated to dryness and resuspended in ethyl acetate. This mixture was washed once each with 10% hydrochloric acid, saturated sodium bicarbonate and brine. The remaining organic layer was evaporated to dryness to yield an oil (82%) that was used without further purification. 1H NMR (500 MHz, CDCl3): 7.26 (t, 2H), 7.18 (m, 3H), 6.00 (s, 1H), 3.37 (q, 2H), 2.98 (m, 4H), 2.87 (t, 2H), 1.91 (s, 3H). Cinnamyl-S-NAC was prepared by combining cinnamic acid (72.0 mg, 480 lmol), NAC (78 lL, 730 lmol), EDC (140 mg, 730 lmol) and TEA (102 lL, 732 lmol) in dichloromethane. After stirring at 25 °C for 18 h, the reaction was evaporated to dryness and resuspended in ethyl acetate. This mixture was washed once each with 10% hydrochloric acid, saturated sodium bicarbonate and brine. The remaining organic layer was evaporated to dryness to yield an oil (40%) that was used without further purification. 1H NMR (500 MHz, CDCl3): 7.59 (d, 1H), 7.51 (m, 2H), 7.35 (m, 3H), 6.69 (d, 1H), 6.31 (s, 1H), 3.46 (q, 2H), 3.13 (t, 2H), 1.95 (s, 3H). 2.3. FMN and oxidase activity assays Reactions containing FMN or EpoB-Ox were carried out in 75 mM sodium phosphate, pH 6.0, 7.0, or 8.0, or in 75 mM Tris, pH 8.8. Reaction mixtures also included 8% THF for substrate solubility. Assays comparing FMN and EpoB-Ox activity at varied pH were performed with 2 mM phenylthiazoline-S-NAC and 10 lM catalyst. Reactions, performed in triplicate, were quenched after
881
1 h upon HPLC injection. Phenylthiazolyl-S-NAC product was detected by analytical HPLC (Agilent 1100 Series) using a linear gradient of acetonitrile in water from 40% to 55% over 8 min, resolved by a 4.6 150 mm Eclipse XDB-C18 column (Agilent). In all cases, product formation was quantified by integrating peak area and comparing it with a standard curve prepared with synthetic phenylthiazolyl-S-NAC. Reported product formation in each case represents values corrected for the small amount of product formed when no catalyst was present. Product identity was confirmed by collecting the putative product peak and analyzing its mass by LC-MS (Thermo Scientific Surveyor LC system and LTQ XL mass spectrometer with electrospray ionization). Kinetic parameters for reactions with FMN were determined with 3 lM FMN, 75 mM sodium phosphate, pH 7.0, and a range of phenylthiazolinyl-S-NAC concentrations in a total volume of 50 lL. Reactions were quenched after 4 h, a time point within the linear range for this reaction, by HPLC injection. This experiment was carried out in triplicate, data were fit with the Michaelis– Menten equation, and the reported error indicates the standard deviation among the three trials. 2.4. Flavin activity comparison Assays to compare the initial rate of product formation with FMN, FAD and riboflavin were carried out in 75 mM sodium phosphate, pH 7.0 with 3 lM flavin and 3.1 mM phenylthiazolinyl-SNAC. Reactions were quenched at a series of time points by HPLC injection. A linear fit was used to determine initial velocity under these conditions for each flavin. Kinetic parameters for FAD were determined as described above for FMN. 3. Results and discussion While previous work with flavin-binding NRPS oxidases determined their catalytic efficacy in the biosynthesis of heteroaromatic rings [6], here, the possibility of flavin cofactors alone as catalysts in similar reactions was evaluated. FMN, the flavin bound by NRPS oxidases, was incubated with phenylthiazolinyl-S-NAC in sodium phosphate buffer, pH 8, at ambient temperature. After 1 h, a new peak was detected by HPLC that coeluted with chemically synthesized phenylthiazolyl-S-NAC (Fig. 1). The identity of the new peak was further confirmed by LC–MS ([M+H]+ observed, 307.2; calculated 307.1). Significantly less phenylthiazolyl-S-NAC product was observed in control reactions containing no FMN. This result indicates that FMN alone can act as an oxidation catalyst in the preparation of heteroaromatic small molecules under aqueous conditions. The catalytic activity of FMN alone was compared with that of the known NRPS oxidase EpoB-Ox, a flavoenzyme required for the biosynthesis of the thiazole ring within the anticancer agent epothilone [6]. Inspection of the catalytic activity of the two at 0.5 mol% revealed that both promote product formation at a range of pH values, as measured after 1 h at ambient temperature (Fig. 2A). Reactions with either catalyst yielded more product as buffer pH increased, supporting an oxidation mechanism that requires base-mediated deprotonation of the substrate in addition to hydride transfer to the flavin (Fig. 2B). Perhaps not surprisingly, this series of assays suggests that EpoB-Ox is the more robust catalyst, with a proportionately greater impact at lower pH than FMN alone. In order to further compare the activity of FMN and EpoB-Ox, kinetic parameters for FMN alone were determined in assays at pH 7 (Fig. 2C), matching the conditions used for the Michaelis–Menten analysis previously completed with EpoB-Ox [6]. While the Km in each case was similar (1.1 ± 0.1 mM for FMN; 2.1 ± 0.4 mM for EpoB-Ox), the rate constant measured for
882
E.A. Argueta et al. / FEBS Letters 589 (2015) 880–884
mAU 2000
A
1500 1000 500 0
0
2
4
6
2
4
6
2
4
6
B
2000 1500 1000 500 0
0
C
2000 1500 1000 500 0
0
min
Fig. 1. Oxidation catalyzed by FMN. Phenylthiazolinyl-S-NAC was combined with FMN (A) or no catalyst (B) and incubated for 1 h at 25 °C and pH 8.0 prior to HPLC analysis at 280 nm, the maximal absorbance for phenylthiazolyl-S-NAC. Authentic phenylthiazolyl-S-NAC (C) is also shown.
FMN (kcat = 0.47 ± 0.04 min1) was lower by more than an order of magnitude than that measured for EpoB-Ox (kcat = 7.7 ± 0.8 min1). Taken together, these results suggest that the oxidase may possess an active site base important for this reaction, explaining its greater catalytic efficacy at neutral pH when compared with
FMN. It is also likely that the enzyme binds both FMN and substrate in proximity, facilitating the oxidation reaction, while reactions catalyzed by FMN rely simply on productive collisions between FMN and substrate in solution. To continue investigating flavin cofactors as catalysts in mild, aqueous buffers, riboflavin and flavin adenine dinucleotide (FAD) were evaluated in comparison to FMN. Both of these naturally occurring flavins were also able to catalyze the oxidation of phenylthiazolinyl-S-NAC. As seen in Fig. 2D, inspection of initial reaction rates for all three flavins revealed that FMN and riboflavin display nearly identical initial rates (0.032 nmol/min and 0.035 nmol/min, respectively) while the initial rate in the presence of FAD is about half of that of the other flavins (0.019 nmol/min). Following up on this result, kinetic parameters for FAD were determined, yielding Km = 0.69 ± 0.14 mM and kcat = 0.22 ± 0.05 min1. Again, the turnover number for FAD is only about half the value determined for FMN under the same reaction conditions. This result could be due to a difference in reduction potential between flavins, though reported values for riboflavin (200 mV) and FMN (205 mV) differ as much as FAD (209 mV) differs from FMN [15]. It seems more likely that the difference between FAD and the other flavins is due to its adenine ring and the possibility of intramolecular interactions between this moiety and the isoalloxazine ring. Prior studies have indicated that FAD predominantly forms this type of stacked, closed conformation, particularly in solution between pH 3.5 and 11 [16–18], although the high concentration of FAD needed for NMR studies can make it difficult to deconvolute intramolecular and intermolecular complexation [19]. Here, the ability to observe catalysis by FAD alone in solution at low micromolar concentrations provides results that further support intramolecular stacking within the molecule. Assuming that substrate oxidation occurs when the thiazoline ring is proximal to the isoalloxazine portion of the flavin, it is conceivable that intramolecular stacking between adenine and isoalloxazine blocks
A
B
C
D
Fig. 2. Comparison of EpoB-Ox and flavin-mediated oxidation. (A) FMN (open bars) or EpoB-Ox (solid bars) at 0.5 mol% was incubated with phenylthiazolinyl-S-NAC for 1 h at 25 °C at the indicated pH values prior to HPLC analysis to determine the extent of substrate conversion to product phenylthiazolyl-S-NAC. Error bars denote standard deviation for triplicate data. (B) Proposed mechanism for phenylthiazolinyl-S-NAC oxidation, requiring both base-catalyzed deprotonation at the a-carbon and hydride transfer to the isoalloxazine ring of the flavin. R varies for the three flavins investigated in this study. (C) Michaelis–Menten analysis used to determine kinetic parameters for FMN with phenylthiazolinyl-S-NAC. Error bars denote standard deviation for triplicate data. (D) Phenylthiazolyl-S-NAC product formation with time, as catalyzed by riboflavin (j), FMN (D), or FAD (). Error bars denote standard deviation for triplicate data but are masked by the symbols.
883
E.A. Argueta et al. / FEBS Letters 589 (2015) 880–884
the approach of substrate to one side of the FAD isoalloxazine ring system, resulting in the observed lower turnover in the presence of this flavin. These results also underscore the important role that FAD-binding flavoenzymes play in modulating the redox efficacy of this particular flavin through the binding conformation and rigidity of the enzyme–flavin complex. The likelihood of FMN to act more generally as a stand-alone catalyst in the formation of oxidized thioester-linked small molecules was tested with the series of substrates shown in Fig. 3A. These substrates included methylthiazolinyl-S-NAC, phenyloxazolinyl-S-NAC, hydrocinnamyl-S-NAC and phenylthiazoline carboxylic acid. Turnover was not detected for any of these molecules, even at pH 8.8, using assay conditions that led to ready oxidation of phenylthiazolinyl-S-NAC as discussed above. Higher pH values were not tested as thioester hydrolysis was observed at longer times even at pH 8.8. Increased temperature also led to hydrolysis rather than substrate oxidation. These results are not unexpected as deprotonation of these test molecules is likely to be more difficult than deprotonation of phenylthiazolinyl-S-NAC. Aromatic substituents in model compounds have been shown to increase acidity by several orders of magnitude [20] and the polarizability of sulfur is known to enhance the stability of carbanions [21]. Particularly relevant to this study, prior work in aqueous buffer similar to flavin assay conditions demonstrated that phenylthiazolinyl-S-NAC readily undergoes exchange at the a-carbon at neutral pH, while exchange for methylthiazolinyl-SNAC and phenyloxazolinyl-S-NAC is nearly undetectable [6]. Methylthiazolinyl-S-NAC and phenyloxazolinyl-S-NAC are known substrates for EpoB-Ox at pH 7, again indicating that the flavoenzyme has catalytic properties beyond FMN under these reaction conditions. Neither phenylthiazoline carboxylic acid nor hydrocinnamyl-S-NAC were turned over by the enzyme, though, demonstrating limitations to EpoB-Ox activity as well that likely also align with increased difficulty of substrate a-carbon
deprotonation. These results further delineate the differences between the flavoenzyme EpoB-Ox and FMN alone; the EpoB-Ox protein probably provides base catalysis at neutral pH that surpasses buffer pH and enables oxidation of some substrates with less acidic a-protons. This is important as the natural substrate for EpoB-Ox is a methylthiazoline ring, oxidized in the biosynthesis of epothilone (Fig. 3B). Work is in progress to identify the protein sidechain(s) acting as a base in this class of biosynthetic oxidases. Due to the ability to measure oxidation of phenylthiazolinyl-SNAC catalyzed by three naturally occurring flavins, conclusions can be inferred about the protein component of NRPS oxidases. While this model substrate is oxidized in the presence of flavins alone in solution, this appears to be an unusual occurrence that does not hold for substrates with less acidic a-protons, suggesting that NRPS oxidases are more than a flavin-binding scaffold and likely have basic residues necessary for catalysis in most cases. It seems that naturally occurring flavins alone are unlikely to be potent catalysts for a broad range of substrates, but can oxidize select substrates under mild, physiologically relevant conditions. This result also indicates the importance of including basic functionality in any future design of miniature flavoenzymes targeting this class of substrates. While designed flavoenzymes have been prepared [22–24], those with demonstrated catalytic function have mainly focused on oxidation of dihydronicotinamides, requiring only hydride transfer to the flavin cofactor for oxidized product formation. Substrate scope for designed flavoenzymes could be broadened with the inclusion of basic functionality. The finding that flavins alone can oxidize certain thioesterlinked heterocycles may also help to explain open questions about heteroaromatic ring formation in NRPS-catalyzed biosynthesis. For example, bleomycin features a bithiazole moiety (Fig. 3B) but its biosynthetic machinery contains only one oxidase, unusual for stepwise NRPS biosynthesis that would predict an oxidase for each thiazole formed [25]. Perhaps the second thiazole ring is oxidized
A
B
epothilone D
bleomycin
Fig. 3. (A) Additional possible substrates investigated for oxidation catalyzed by FMN. None of these molecules were oxidized as determined by HPLC analysis. (B) Structures of thiazole-containing natural products epothilone D and bleomycin.
884
E.A. Argueta et al. / FEBS Letters 589 (2015) 880–884
by the same oxidase, or slowly in the presence of molecular oxygen, or perhaps oxidation is facilitated by free flavin. This work suggests that a thiazoline ring next to an aromatic ring, as seen with phenylthiazolinyl-S-NAC, is readily oxidized in the presence of FMN alone. The intracellular concentration of free flavin in Escherichia coli, while not the same bacterium as the Streptomyces verticillus producer of bleomycin, is estimated at approximately 4 lM [26], similar to concentrations used in the in vitro assays reported here. It is possible that there is at least a historic role for free flavins as catalysts. Some have linked coenzymes to the prebiotic world, noting that FAD, as well as others, contains a nucleotide and thus may be related to early, possibly more catalytically active, nucleic acid enzymes [27,28]. Additional work will be needed in order to determine if free flavins are relevant in modern day natural biosynthesis. Acknowledgements E.A.A. and A.N.A. were supported by NSF-S-STEM 1153371. A.N.A. and P.K. thank the Keck Foundation for summer research fellowships. This work was supported in part by a Cottrell College Science Award from Research Corporation for Science Advancement. References [1] Walsh, C. (1980) Flavin coenzymes: at the crossroads of biological redox chemistry. Acc. Chem. Res. 13, 148–155. [2] Mansoorabadi, S.O., Thibodeaux, C.J. and Liu, H.-W. (2007) The diverse roles of flavin coenzymes – nature’s most versatile thespians. J. Org. Chem. 72, 6329– 6342. [3] Walsh, C.T. and Wencewicz, T.A. (2013) Flavoenzymes: versatile catalysts in biosynthetic pathways. Nat. Prod. Rep. 30, 175–200. [4] Li, Y.-M., Milne, J.C., Madison, L.L., Kolter, R. and Walsh, C.T. (1996) From peptide precursors to oxazole and thiazole-containing peptide antibiotics: microcin B177 synthase. Science 274, 1188–1193. [5] Walsh, C.T., Malcolmson, S.J. and Young, T.S. (2012) Three ring posttranslational circuses: insertion of oxazoles, thiazoles, and pyridines into protein-derived frameworks. ACS Chem. Biol. 7, 429–442. [6] Schneider, T.L., Shen, B. and Walsh, C.T. (2003) Oxidase domains in epothilone and bleomycin biosynthesis: thiazoline to thiazole oxidation during chain elongation. Biochemistry 42, 9722–9730. [7] Imada, Y. and Naota, T. (2007) Flavins as organocatalysts for environmentally benign molecular transformations. Chem. Rec. 7, 354–361. [8] Chen, S., Hossain, M.S. and Foss, F.W. (2013) Bioinspired oxidative aromatizations: one-pot syntheses of 2-substituted benzothiazoles and pyridines by aerobic organocatalysis. ACS Sustainable Chem. Eng. 1, 1045– 1051. [9] Murahashi, S.-I. (2011) Development of biomimetic catalytic oxidation methods and non-salt methods using transition metal-based acid and base ambiphilic catalysts. Proc. Jpn. Acad. Ser. B Phys. Biol. Sci. 87, 242–253.
[10] Imada, Y., Kitagawa, T., Wang, H.-K., Komiya, N. and Naota, T. (2013) Flavincatalyzed aerobic oxidation of sulfides in aqueous media. Tetrahedron Lett. 54, 621–624. [11] Murray, A.T., Matton, P., Fairhurst, N.W.G., John, M.P. and Carbery, D.R. (2012) Biomimetic flavin-catalyzed aldehyde oxidation. Org. Lett. 14, 4656–4659. [12] Cerletti, P. (1959) Properties of riboflavin phosphates. Anal. Chim. Acta 20, 243–250. [13] Hoveyda, H.R., Karunaratne, V. and Orvig, C. (1992) Design and synthesis of multidentate 2-(20 -hydroxyphenyl)-2-thiazolines for biomedical application. Tetrahedron 48, 5219–5226. [14] Chen, H.W., O’Connor, S.E., Cane, D.E. and Walsh, C.T. (2001) Epothilone biosynthesis: assembly of the methylthiazolylcarboxy starter unit on the EpoB subunit. Chem. Biol. 8, 899–912. [15] Williamson, G. and Edmondson, D.E. (1985) Effect of pH on oxidationreduction potentials of 8a-N-imidazole-substituted flavins. Biochemistry 24, 7790–7797. [16] Sarma, R.H., Dannies, P. and Kaplan, N.O. (1968) Investigations of inter- and intramolecular interactions in flavin–adenine dinucleotide by proton magnetic resonance. Biochemistry 7, 4359–4367. [17] Kainosho, M. and Kyogoku, Y. (1972) High-resolution proton and phosphorus nuclear magnetic resonance spectra of flavin–adenine dinucleotide and its conformation in aqueous solution. Biochemistry 11, 741–752. [18] Islam, S.D.M., Susdorf, T., Penzkofer, A. and Hegemann, P. (2003) Fluorescence quenching of flavin adenine dinucleotide in aqueous solution by pH dependent isomerisation and photo-induced electron transfer. Chem. Phys. 295, 137–149. [19] Grininger, M., Seiler, F., Zeth, K. and Oesterhelt, D. (2006) Dodecin sequesters FAD in closed conformation from the aqueous solution. J. Mol. Biol. 364, 561– 566. [20] Bordwell, F.G. (1988) Equilibrium acidities in dimethyl sulfoxide solution. Acc. Chem. Res. 21, 456–463. [21] Bernasconi, C.F. and Kittredge, K.W. (1998) Carbanion stabilization by adjacent sulfur: polarizability, resonance, or negative hyperconjugation? Experimental distinction based on intrinsic rate constants of proton transfer from (phenylthio)nitromethane and 1-nitro-2-phenylethane. J. Org. Chem. 63, 1944–1953. [22] Distefano, M.D., Kuang, H., Qi, D. and Mazhary, A. (1998) The design of proteinbased catalysts using semisynthetic methods. Curr. Opin. Struct. Biol. 8, 459– 465. [23] Imperiali, B., McDonnell, K.A. and Shogren-Knaak, M. (1999) Design and construction of novel peptides and proteins by tailored incorporation of coenzyme functionality in: Implementation and Redesign of Catalytic Function in Biopolymers (Schmidtchen, F.P., Ed.), pp. 1–38, Springer, Berlin. [24] Butterfield, S.M., Goodman, C.M., Rotello, V.M. and Waters, M.L. (2004) A peptide flavoprotein mimic: flavin recognition and redox potential modulation in water by a designed b hairpin. Angew. Chem. Int. Ed. 43, 724–727. [25] Shen, B., Du, L., Sanchez, C., Edwards, D.J., Chen, M. and Murrell, J.M. (2002) Cloning and characterization of the bleomycin biosynthetic gene cluster from Streptomyces verticillus ATCC15003. J. Nat. Prod. 65, 422–431. [26] Wilson, A.C. and Pardee, A.B. (1962) Regulation of flavin synthesis by Escherichia coli. J. Gen. Microbiol. 28, 283–303. [27] White, H.B. (1976) Coenzymes as fossils of an earlier metabolic state. J. Mol. Evol. 7, 101–104. [28] Nguyen, K.V. and Burrows, C.J. (2012) Whence flavins? Redox-active ribonucleotides link metabolism and genome repair to the RNA world. Acc. Chem. Res. 45, 2151–2159.
journal homepage: www.FEBSLetters.org
Unusual non-enzymatic flavin catalysis enhances understanding of flavoenzymes Erick A. Argueta, Amanda N. Amoh, Prapti Kafle, Tanya L. Schneider ⇑ Department of Chemistry, Connecticut College, 270 Mohegan Avenue, New London, CT 06320, USA
a r t i c l e
i n f o
Article history: Received 7 January 2015 Revised 24 February 2015 Accepted 25 February 2015 Available online 5 March 2015 Edited by Michael Ibba Keywords: Flavin Oxidase Non-ribosomal peptide synthesis Biosynthesis Thiazole
a b s t r a c t Flavin cofactors are central to many biochemical transformations and are typically tightly bound as part of a catalytically active flavoenzyme. This work indicates that naturally occurring flavins can act as stand-alone catalysts to promote the oxidation of biosynthetically inspired heterocycles in aqueous buffers. Flavin activity was compared with that of oxidases important in non-ribosomal peptide synthesis, providing a rare direct comparison between the catalytic efficacy of flavins alone and in the context of a full flavoenzyme. This study suggests that such oxidases are likely to possess an active site base, as oxidase activity was greater than that of flavins alone, particularly for less acidic substrates. These findings offer perspective on the development of robust and catalytically effective, designed miniature flavoenzymes. Ó 2015 Federation of European Biochemical Societies. Published by Elsevier B.V. All rights reserved.
1. Introduction Flavins are redox cofactors central to numerous metabolic processes [1]. As such, they are nearly always tightly associated with flavoenzymes, and this complex together is responsible for the catalysis of a wide range of biochemical transformations [2]. While flavoenzymes are well-known in primary metabolic pathways, they also play an important role in secondary metabolism [3]. Flavin-binding oxidases are located in biosynthetic pathways yielding nitrogen-containing heterocycles such as thiazoles and oxazoles. These heteroaromatic rings are formed from posttranslational modification of serine, threonine and cysteine in the ribosomal production of peptides like microcin B17 [4] and are also found in non-ribosomal peptides formed by non-ribosomal peptide synthetases (NRPS) [5]. Oxidases in such biosynthetic pathways facilitate the oxidation of thiazoline and oxazoline rings to the more stable, aromatic thiazole and oxazole rings, often necessary for the bioactivity of the natural product. Because these nitrogencontaining heteroaromatic rings are of interest, particularly from a medicinal chemistry perspective, their formation has been investigated.
Author contributions: E.A.A. and T.L.S. designed and performed catalyst assays. A.N.A. and P.K. synthesized and characterized small molecules. T.L.S. supervised the project and wrote the manuscript with input from E.A.A., A.N.A. and P.K. ⇑ Corresponding author. Fax: +1 860 439 2477. E-mail address: [email protected] (T.L. Schneider).
The NRPS oxidases important for the biosynthesis of the anticancer agents epothilone and bleomycin have been characterized, with catalytic activity for a series of thiazoline and oxazoline substrate mimics measured [6]. These oxidases were found to bind flavin mononucleotide (FMN) non-covalently but so tightly that denaturation is necessary to separate the two. Flavin cofactors are not usually exchanged after catalytic turnover and tend to be regenerated while enzyme-bound, requiring molecular oxygen or other oxidants. Because flavins can be simply regenerated by molecular oxygen, and are linked to the formation of bioactive small molecules, they are intriguing targets for stand-alone bioinspired catalysts. Some synthetic flavin analogs have been evaluated, suggesting their utility as effective and environmentally mild oxidants for synthetic transformations [7]. Hydroperoxyflavins recently have been shown to promote the formation of heteroaromatic pyridines and benzothiazoles at ambient temperature in methanol and in the presence of molecular oxygen [8]. This report joins other work describing synthetic flavins capable of catalyzing the oxidation of amines, sulfides and aldehydes [9–11]. These synthetic oxidation reactions take advantage of the redox potential of flavins as ‘‘green’’ catalysts while removing the enzyme portion of flavoenzymes which can be more susceptible to denaturation under varied solvent conditions. In this work, the potential of naturally occurring flavins as stand-alone catalysts mediating the oxidation of biosynthetically inspired small molecules under physiologically relevant conditions
http://dx.doi.org/10.1016/j.febslet.2015.02.034 0014-5793/Ó 2015 Federation of European Biochemical Societies. Published by Elsevier B.V. All rights reserved.
E.A. Argueta et al. / FEBS Letters 589 (2015) 880–884
was probed. FMN, flavin adenine dinucleotide (FAD), and riboflavin all catalyze the formation of heteroaromatic rings in thioesterlinked small molecules known to be substrates for NRPS oxidases. As these reactions occur in aqueous buffers mirroring the reaction conditions needed for oxidase activity, this study allows a rare direct comparison between the catalytic efficacy of flavins alone and in the context of the full flavoenzyme. This analysis provides mechanistic insights on the catalytic role of the oxidase enzyme beyond the bound flavin cofactor and offers perspective on the development of robust and catalytically effective, designed miniature flavoenzymes. 2. Materials and methods 2.1. Reagents and enzymes FMN, FAD and riboflavin were purchased from Sigma–Aldrich. Reagents for the synthesis of substrates and product standards were also purchased from Sigma–Aldrich. Other reagents were purchased from Fisher Scientific unless otherwise specified. Overexpression and purification to yield EpoB-Ox were performed as described [6]. To determine flavin-bound enzyme concentration, a sample of purified EpoB-Ox was heat denatured for 2 min at 90 °C, and the supernatant was subjected to UV–vis analysis. Concentration was calculated based on the known extinction coefficient for FMN, 12 200 cm1 M1 at 450 nm [12]. 2.2. Chemical synthesis of substrates and standards Phenylthiazoline carboxylic acid, phenylthiazole carboxylic acid, phenylthiazolinyl-S-NAC, phenylthiazolyl-S-NAC, phenyloxazolinyl-S-NAC, phenyloxazolyl-S-NAC, methylthiazolinyl-SNAC and methylthiazolyl-S-NAC were synthesized as described [6,13,14]. Hydrocinnamyl-S-NAC was prepared by combining hydrocinnamic acid (73.0 mg, 486 lmol), N-acetylcysteamine (NAC; 78 lL, 730 lmol), N-(3-dimethylaminopropyl)-N0 -ethylcarbodiimide hydrochloride (EDC; 140 mg, 730 lmol) and triethylamine (TEA; 102 lL, 732 lmol) in dichloromethane. After stirring at 25 °C for 18 h, the reaction was evaporated to dryness and resuspended in ethyl acetate. This mixture was washed once each with 10% hydrochloric acid, saturated sodium bicarbonate and brine. The remaining organic layer was evaporated to dryness to yield an oil (82%) that was used without further purification. 1H NMR (500 MHz, CDCl3): 7.26 (t, 2H), 7.18 (m, 3H), 6.00 (s, 1H), 3.37 (q, 2H), 2.98 (m, 4H), 2.87 (t, 2H), 1.91 (s, 3H). Cinnamyl-S-NAC was prepared by combining cinnamic acid (72.0 mg, 480 lmol), NAC (78 lL, 730 lmol), EDC (140 mg, 730 lmol) and TEA (102 lL, 732 lmol) in dichloromethane. After stirring at 25 °C for 18 h, the reaction was evaporated to dryness and resuspended in ethyl acetate. This mixture was washed once each with 10% hydrochloric acid, saturated sodium bicarbonate and brine. The remaining organic layer was evaporated to dryness to yield an oil (40%) that was used without further purification. 1H NMR (500 MHz, CDCl3): 7.59 (d, 1H), 7.51 (m, 2H), 7.35 (m, 3H), 6.69 (d, 1H), 6.31 (s, 1H), 3.46 (q, 2H), 3.13 (t, 2H), 1.95 (s, 3H). 2.3. FMN and oxidase activity assays Reactions containing FMN or EpoB-Ox were carried out in 75 mM sodium phosphate, pH 6.0, 7.0, or 8.0, or in 75 mM Tris, pH 8.8. Reaction mixtures also included 8% THF for substrate solubility. Assays comparing FMN and EpoB-Ox activity at varied pH were performed with 2 mM phenylthiazoline-S-NAC and 10 lM catalyst. Reactions, performed in triplicate, were quenched after
881
1 h upon HPLC injection. Phenylthiazolyl-S-NAC product was detected by analytical HPLC (Agilent 1100 Series) using a linear gradient of acetonitrile in water from 40% to 55% over 8 min, resolved by a 4.6 150 mm Eclipse XDB-C18 column (Agilent). In all cases, product formation was quantified by integrating peak area and comparing it with a standard curve prepared with synthetic phenylthiazolyl-S-NAC. Reported product formation in each case represents values corrected for the small amount of product formed when no catalyst was present. Product identity was confirmed by collecting the putative product peak and analyzing its mass by LC-MS (Thermo Scientific Surveyor LC system and LTQ XL mass spectrometer with electrospray ionization). Kinetic parameters for reactions with FMN were determined with 3 lM FMN, 75 mM sodium phosphate, pH 7.0, and a range of phenylthiazolinyl-S-NAC concentrations in a total volume of 50 lL. Reactions were quenched after 4 h, a time point within the linear range for this reaction, by HPLC injection. This experiment was carried out in triplicate, data were fit with the Michaelis– Menten equation, and the reported error indicates the standard deviation among the three trials. 2.4. Flavin activity comparison Assays to compare the initial rate of product formation with FMN, FAD and riboflavin were carried out in 75 mM sodium phosphate, pH 7.0 with 3 lM flavin and 3.1 mM phenylthiazolinyl-SNAC. Reactions were quenched at a series of time points by HPLC injection. A linear fit was used to determine initial velocity under these conditions for each flavin. Kinetic parameters for FAD were determined as described above for FMN. 3. Results and discussion While previous work with flavin-binding NRPS oxidases determined their catalytic efficacy in the biosynthesis of heteroaromatic rings [6], here, the possibility of flavin cofactors alone as catalysts in similar reactions was evaluated. FMN, the flavin bound by NRPS oxidases, was incubated with phenylthiazolinyl-S-NAC in sodium phosphate buffer, pH 8, at ambient temperature. After 1 h, a new peak was detected by HPLC that coeluted with chemically synthesized phenylthiazolyl-S-NAC (Fig. 1). The identity of the new peak was further confirmed by LC–MS ([M+H]+ observed, 307.2; calculated 307.1). Significantly less phenylthiazolyl-S-NAC product was observed in control reactions containing no FMN. This result indicates that FMN alone can act as an oxidation catalyst in the preparation of heteroaromatic small molecules under aqueous conditions. The catalytic activity of FMN alone was compared with that of the known NRPS oxidase EpoB-Ox, a flavoenzyme required for the biosynthesis of the thiazole ring within the anticancer agent epothilone [6]. Inspection of the catalytic activity of the two at 0.5 mol% revealed that both promote product formation at a range of pH values, as measured after 1 h at ambient temperature (Fig. 2A). Reactions with either catalyst yielded more product as buffer pH increased, supporting an oxidation mechanism that requires base-mediated deprotonation of the substrate in addition to hydride transfer to the flavin (Fig. 2B). Perhaps not surprisingly, this series of assays suggests that EpoB-Ox is the more robust catalyst, with a proportionately greater impact at lower pH than FMN alone. In order to further compare the activity of FMN and EpoB-Ox, kinetic parameters for FMN alone were determined in assays at pH 7 (Fig. 2C), matching the conditions used for the Michaelis–Menten analysis previously completed with EpoB-Ox [6]. While the Km in each case was similar (1.1 ± 0.1 mM for FMN; 2.1 ± 0.4 mM for EpoB-Ox), the rate constant measured for
882
E.A. Argueta et al. / FEBS Letters 589 (2015) 880–884
mAU 2000
A
1500 1000 500 0
0
2
4
6
2
4
6
2
4
6
B
2000 1500 1000 500 0
0
C
2000 1500 1000 500 0
0
min
Fig. 1. Oxidation catalyzed by FMN. Phenylthiazolinyl-S-NAC was combined with FMN (A) or no catalyst (B) and incubated for 1 h at 25 °C and pH 8.0 prior to HPLC analysis at 280 nm, the maximal absorbance for phenylthiazolyl-S-NAC. Authentic phenylthiazolyl-S-NAC (C) is also shown.
FMN (kcat = 0.47 ± 0.04 min1) was lower by more than an order of magnitude than that measured for EpoB-Ox (kcat = 7.7 ± 0.8 min1). Taken together, these results suggest that the oxidase may possess an active site base important for this reaction, explaining its greater catalytic efficacy at neutral pH when compared with
FMN. It is also likely that the enzyme binds both FMN and substrate in proximity, facilitating the oxidation reaction, while reactions catalyzed by FMN rely simply on productive collisions between FMN and substrate in solution. To continue investigating flavin cofactors as catalysts in mild, aqueous buffers, riboflavin and flavin adenine dinucleotide (FAD) were evaluated in comparison to FMN. Both of these naturally occurring flavins were also able to catalyze the oxidation of phenylthiazolinyl-S-NAC. As seen in Fig. 2D, inspection of initial reaction rates for all three flavins revealed that FMN and riboflavin display nearly identical initial rates (0.032 nmol/min and 0.035 nmol/min, respectively) while the initial rate in the presence of FAD is about half of that of the other flavins (0.019 nmol/min). Following up on this result, kinetic parameters for FAD were determined, yielding Km = 0.69 ± 0.14 mM and kcat = 0.22 ± 0.05 min1. Again, the turnover number for FAD is only about half the value determined for FMN under the same reaction conditions. This result could be due to a difference in reduction potential between flavins, though reported values for riboflavin (200 mV) and FMN (205 mV) differ as much as FAD (209 mV) differs from FMN [15]. It seems more likely that the difference between FAD and the other flavins is due to its adenine ring and the possibility of intramolecular interactions between this moiety and the isoalloxazine ring. Prior studies have indicated that FAD predominantly forms this type of stacked, closed conformation, particularly in solution between pH 3.5 and 11 [16–18], although the high concentration of FAD needed for NMR studies can make it difficult to deconvolute intramolecular and intermolecular complexation [19]. Here, the ability to observe catalysis by FAD alone in solution at low micromolar concentrations provides results that further support intramolecular stacking within the molecule. Assuming that substrate oxidation occurs when the thiazoline ring is proximal to the isoalloxazine portion of the flavin, it is conceivable that intramolecular stacking between adenine and isoalloxazine blocks
A
B
C
D
Fig. 2. Comparison of EpoB-Ox and flavin-mediated oxidation. (A) FMN (open bars) or EpoB-Ox (solid bars) at 0.5 mol% was incubated with phenylthiazolinyl-S-NAC for 1 h at 25 °C at the indicated pH values prior to HPLC analysis to determine the extent of substrate conversion to product phenylthiazolyl-S-NAC. Error bars denote standard deviation for triplicate data. (B) Proposed mechanism for phenylthiazolinyl-S-NAC oxidation, requiring both base-catalyzed deprotonation at the a-carbon and hydride transfer to the isoalloxazine ring of the flavin. R varies for the three flavins investigated in this study. (C) Michaelis–Menten analysis used to determine kinetic parameters for FMN with phenylthiazolinyl-S-NAC. Error bars denote standard deviation for triplicate data. (D) Phenylthiazolyl-S-NAC product formation with time, as catalyzed by riboflavin (j), FMN (D), or FAD (). Error bars denote standard deviation for triplicate data but are masked by the symbols.
883
E.A. Argueta et al. / FEBS Letters 589 (2015) 880–884
the approach of substrate to one side of the FAD isoalloxazine ring system, resulting in the observed lower turnover in the presence of this flavin. These results also underscore the important role that FAD-binding flavoenzymes play in modulating the redox efficacy of this particular flavin through the binding conformation and rigidity of the enzyme–flavin complex. The likelihood of FMN to act more generally as a stand-alone catalyst in the formation of oxidized thioester-linked small molecules was tested with the series of substrates shown in Fig. 3A. These substrates included methylthiazolinyl-S-NAC, phenyloxazolinyl-S-NAC, hydrocinnamyl-S-NAC and phenylthiazoline carboxylic acid. Turnover was not detected for any of these molecules, even at pH 8.8, using assay conditions that led to ready oxidation of phenylthiazolinyl-S-NAC as discussed above. Higher pH values were not tested as thioester hydrolysis was observed at longer times even at pH 8.8. Increased temperature also led to hydrolysis rather than substrate oxidation. These results are not unexpected as deprotonation of these test molecules is likely to be more difficult than deprotonation of phenylthiazolinyl-S-NAC. Aromatic substituents in model compounds have been shown to increase acidity by several orders of magnitude [20] and the polarizability of sulfur is known to enhance the stability of carbanions [21]. Particularly relevant to this study, prior work in aqueous buffer similar to flavin assay conditions demonstrated that phenylthiazolinyl-S-NAC readily undergoes exchange at the a-carbon at neutral pH, while exchange for methylthiazolinyl-SNAC and phenyloxazolinyl-S-NAC is nearly undetectable [6]. Methylthiazolinyl-S-NAC and phenyloxazolinyl-S-NAC are known substrates for EpoB-Ox at pH 7, again indicating that the flavoenzyme has catalytic properties beyond FMN under these reaction conditions. Neither phenylthiazoline carboxylic acid nor hydrocinnamyl-S-NAC were turned over by the enzyme, though, demonstrating limitations to EpoB-Ox activity as well that likely also align with increased difficulty of substrate a-carbon
deprotonation. These results further delineate the differences between the flavoenzyme EpoB-Ox and FMN alone; the EpoB-Ox protein probably provides base catalysis at neutral pH that surpasses buffer pH and enables oxidation of some substrates with less acidic a-protons. This is important as the natural substrate for EpoB-Ox is a methylthiazoline ring, oxidized in the biosynthesis of epothilone (Fig. 3B). Work is in progress to identify the protein sidechain(s) acting as a base in this class of biosynthetic oxidases. Due to the ability to measure oxidation of phenylthiazolinyl-SNAC catalyzed by three naturally occurring flavins, conclusions can be inferred about the protein component of NRPS oxidases. While this model substrate is oxidized in the presence of flavins alone in solution, this appears to be an unusual occurrence that does not hold for substrates with less acidic a-protons, suggesting that NRPS oxidases are more than a flavin-binding scaffold and likely have basic residues necessary for catalysis in most cases. It seems that naturally occurring flavins alone are unlikely to be potent catalysts for a broad range of substrates, but can oxidize select substrates under mild, physiologically relevant conditions. This result also indicates the importance of including basic functionality in any future design of miniature flavoenzymes targeting this class of substrates. While designed flavoenzymes have been prepared [22–24], those with demonstrated catalytic function have mainly focused on oxidation of dihydronicotinamides, requiring only hydride transfer to the flavin cofactor for oxidized product formation. Substrate scope for designed flavoenzymes could be broadened with the inclusion of basic functionality. The finding that flavins alone can oxidize certain thioesterlinked heterocycles may also help to explain open questions about heteroaromatic ring formation in NRPS-catalyzed biosynthesis. For example, bleomycin features a bithiazole moiety (Fig. 3B) but its biosynthetic machinery contains only one oxidase, unusual for stepwise NRPS biosynthesis that would predict an oxidase for each thiazole formed [25]. Perhaps the second thiazole ring is oxidized
A
B
epothilone D
bleomycin
Fig. 3. (A) Additional possible substrates investigated for oxidation catalyzed by FMN. None of these molecules were oxidized as determined by HPLC analysis. (B) Structures of thiazole-containing natural products epothilone D and bleomycin.
884
E.A. Argueta et al. / FEBS Letters 589 (2015) 880–884
by the same oxidase, or slowly in the presence of molecular oxygen, or perhaps oxidation is facilitated by free flavin. This work suggests that a thiazoline ring next to an aromatic ring, as seen with phenylthiazolinyl-S-NAC, is readily oxidized in the presence of FMN alone. The intracellular concentration of free flavin in Escherichia coli, while not the same bacterium as the Streptomyces verticillus producer of bleomycin, is estimated at approximately 4 lM [26], similar to concentrations used in the in vitro assays reported here. It is possible that there is at least a historic role for free flavins as catalysts. Some have linked coenzymes to the prebiotic world, noting that FAD, as well as others, contains a nucleotide and thus may be related to early, possibly more catalytically active, nucleic acid enzymes [27,28]. Additional work will be needed in order to determine if free flavins are relevant in modern day natural biosynthesis. Acknowledgements E.A.A. and A.N.A. were supported by NSF-S-STEM 1153371. A.N.A. and P.K. thank the Keck Foundation for summer research fellowships. This work was supported in part by a Cottrell College Science Award from Research Corporation for Science Advancement. References [1] Walsh, C. (1980) Flavin coenzymes: at the crossroads of biological redox chemistry. Acc. Chem. Res. 13, 148–155. [2] Mansoorabadi, S.O., Thibodeaux, C.J. and Liu, H.-W. (2007) The diverse roles of flavin coenzymes – nature’s most versatile thespians. J. Org. Chem. 72, 6329– 6342. [3] Walsh, C.T. and Wencewicz, T.A. (2013) Flavoenzymes: versatile catalysts in biosynthetic pathways. Nat. Prod. Rep. 30, 175–200. [4] Li, Y.-M., Milne, J.C., Madison, L.L., Kolter, R. and Walsh, C.T. (1996) From peptide precursors to oxazole and thiazole-containing peptide antibiotics: microcin B177 synthase. Science 274, 1188–1193. [5] Walsh, C.T., Malcolmson, S.J. and Young, T.S. (2012) Three ring posttranslational circuses: insertion of oxazoles, thiazoles, and pyridines into protein-derived frameworks. ACS Chem. Biol. 7, 429–442. [6] Schneider, T.L., Shen, B. and Walsh, C.T. (2003) Oxidase domains in epothilone and bleomycin biosynthesis: thiazoline to thiazole oxidation during chain elongation. Biochemistry 42, 9722–9730. [7] Imada, Y. and Naota, T. (2007) Flavins as organocatalysts for environmentally benign molecular transformations. Chem. Rec. 7, 354–361. [8] Chen, S., Hossain, M.S. and Foss, F.W. (2013) Bioinspired oxidative aromatizations: one-pot syntheses of 2-substituted benzothiazoles and pyridines by aerobic organocatalysis. ACS Sustainable Chem. Eng. 1, 1045– 1051. [9] Murahashi, S.-I. (2011) Development of biomimetic catalytic oxidation methods and non-salt methods using transition metal-based acid and base ambiphilic catalysts. Proc. Jpn. Acad. Ser. B Phys. Biol. Sci. 87, 242–253.
[10] Imada, Y., Kitagawa, T., Wang, H.-K., Komiya, N. and Naota, T. (2013) Flavincatalyzed aerobic oxidation of sulfides in aqueous media. Tetrahedron Lett. 54, 621–624. [11] Murray, A.T., Matton, P., Fairhurst, N.W.G., John, M.P. and Carbery, D.R. (2012) Biomimetic flavin-catalyzed aldehyde oxidation. Org. Lett. 14, 4656–4659. [12] Cerletti, P. (1959) Properties of riboflavin phosphates. Anal. Chim. Acta 20, 243–250. [13] Hoveyda, H.R., Karunaratne, V. and Orvig, C. (1992) Design and synthesis of multidentate 2-(20 -hydroxyphenyl)-2-thiazolines for biomedical application. Tetrahedron 48, 5219–5226. [14] Chen, H.W., O’Connor, S.E., Cane, D.E. and Walsh, C.T. (2001) Epothilone biosynthesis: assembly of the methylthiazolylcarboxy starter unit on the EpoB subunit. Chem. Biol. 8, 899–912. [15] Williamson, G. and Edmondson, D.E. (1985) Effect of pH on oxidationreduction potentials of 8a-N-imidazole-substituted flavins. Biochemistry 24, 7790–7797. [16] Sarma, R.H., Dannies, P. and Kaplan, N.O. (1968) Investigations of inter- and intramolecular interactions in flavin–adenine dinucleotide by proton magnetic resonance. Biochemistry 7, 4359–4367. [17] Kainosho, M. and Kyogoku, Y. (1972) High-resolution proton and phosphorus nuclear magnetic resonance spectra of flavin–adenine dinucleotide and its conformation in aqueous solution. Biochemistry 11, 741–752. [18] Islam, S.D.M., Susdorf, T., Penzkofer, A. and Hegemann, P. (2003) Fluorescence quenching of flavin adenine dinucleotide in aqueous solution by pH dependent isomerisation and photo-induced electron transfer. Chem. Phys. 295, 137–149. [19] Grininger, M., Seiler, F., Zeth, K. and Oesterhelt, D. (2006) Dodecin sequesters FAD in closed conformation from the aqueous solution. J. Mol. Biol. 364, 561– 566. [20] Bordwell, F.G. (1988) Equilibrium acidities in dimethyl sulfoxide solution. Acc. Chem. Res. 21, 456–463. [21] Bernasconi, C.F. and Kittredge, K.W. (1998) Carbanion stabilization by adjacent sulfur: polarizability, resonance, or negative hyperconjugation? Experimental distinction based on intrinsic rate constants of proton transfer from (phenylthio)nitromethane and 1-nitro-2-phenylethane. J. Org. Chem. 63, 1944–1953. [22] Distefano, M.D., Kuang, H., Qi, D. and Mazhary, A. (1998) The design of proteinbased catalysts using semisynthetic methods. Curr. Opin. Struct. Biol. 8, 459– 465. [23] Imperiali, B., McDonnell, K.A. and Shogren-Knaak, M. (1999) Design and construction of novel peptides and proteins by tailored incorporation of coenzyme functionality in: Implementation and Redesign of Catalytic Function in Biopolymers (Schmidtchen, F.P., Ed.), pp. 1–38, Springer, Berlin. [24] Butterfield, S.M., Goodman, C.M., Rotello, V.M. and Waters, M.L. (2004) A peptide flavoprotein mimic: flavin recognition and redox potential modulation in water by a designed b hairpin. Angew. Chem. Int. Ed. 43, 724–727. [25] Shen, B., Du, L., Sanchez, C., Edwards, D.J., Chen, M. and Murrell, J.M. (2002) Cloning and characterization of the bleomycin biosynthetic gene cluster from Streptomyces verticillus ATCC15003. J. Nat. Prod. 65, 422–431. [26] Wilson, A.C. and Pardee, A.B. (1962) Regulation of flavin synthesis by Escherichia coli. J. Gen. Microbiol. 28, 283–303. [27] White, H.B. (1976) Coenzymes as fossils of an earlier metabolic state. J. Mol. Evol. 7, 101–104. [28] Nguyen, K.V. and Burrows, C.J. (2012) Whence flavins? Redox-active ribonucleotides link metabolism and genome repair to the RNA world. Acc. Chem. Res. 45, 2151–2159.
