Selenium heterocycles (e.g., selenazofurin) have a number of potential applications in medicinal chemistry; although isosteric with sulfur and oxygen, they have different chemical properties. As a route to exploring biological properties, this substitution is a valuable addition to the chemist's arsenal. The chemical synthesis of selenazolines and selenazoles involves the use of stoichiometric quantities of toxic selenium-containing precursors. However, we have incorporated selenocysteine into a substrate ribosomal peptide by feeding dilute solutions of selenocysteine to E. coli. We demonstrate that the dehydrating heterocyclases from the patellamide and the closely related trunkamide biosynthetic pathways process selenocysteine-containing substrate to yield selenazoline, thus establishing an enzymatic route to selenazolines. Such groups can also be incorporated into macrocyclic peptides. Although this remains unknown, it is possible that nature itself could utilise the same enzymes to produce selenium variants of natural products.
1,3-Selenazoles are intriguing structural motifs in medicinal chemistry1 as they serve as analogues of the corresponding thiazoles; arguably the best-known examples are the synthetic compounds selenazofurin2 (1) and amselamine3 (2; Scheme 1). Zhang et al.4 recently reported the chemical synthesis of selenazole-containing cyclic peptides (e.g., 3) and their application in X-ray diffraction studies of P-glycoprotein. 1,3-Selenazolines1a have also received attention in recent years,5 for example, Withers et al.5a have reported the synthesis of a GlcNAc-derived 1,3-selenazole that was evaluated as a potential O-GlcNAcase inhibitor. Most synthetic routes to 1,3-selenazoles are based on the Hantzsch synthesis,1a, 5c, 6 although other methods have been investigated.7
Cyanobactins are a family of ribosomally derived cyclic peptides containing six to 20 amino acids, they are produced from a larger precursor peptide and contain extensive chemical modifications including heterocycles, D-stereocentres and disulfide bonds. They are potent biological scaffolds that can cross cell membranes and are resistant to enzymatic degradation.8 Patellamides are members of the cyanobactin family that are produced by Prochloron didemni, an obligate symbiont of the sea squirt Lissoclinum patella found in Pacific reefs.9 The biosynthetic pathway involves the ATP-dependent heterocyclase PatD, protease PatA and oxidase/macrocyclase PatG, which convert the ribosomal precursor peptide PatE into the final macrocycle (Scheme 2). Patellamides and variants can be produced heterologously in E. coli by coexpressing a mutant PatE with the biosynthetic genes.10
Herein we describe the biosynthesis of a selenol-containing PatE from selenocysteine (SeCys). SeCys is commercially available and significantly less toxic than many other selenium precursors. We show that the selenium variant protein is a substrate for PatD (and its homologue from the trunkamide pathway, TruD),9, 10 which heterocyclises cysteine residues to thiazolines. This enzymatic route thus offers a mild and convenient laboratory method for the synthesis of selenazoline (Sezn) within the context of a peptide.
We purified a modified PatE9 precursor peptide that has a single product cassette containing the sequence ITASITFC and in which the cysteine is replaced by SeCys by using the medium supplementation method. We confirmed the incorporation of Se by mass spectrometry and denoted the product PatE(Se). We also observed PatE in which the selenol was hydrolysed to serine (PatE(O), 6828.6 Da, Figure 1); presumably this is a consequence of the growth and protein purification process. Selenol (like thiol) reacts with iodoacetamide (IAA) to give a covalent adduct, as can be seen in Figure 1. We reasoned that addition of IAA to TruD-treated PatE, which has heterocyclic Sezn, would not lead to a reaction, and thus no mass shift would be seen. We saw that PatE that had reacted with TruD (and PatD), and thus lost a water molecule, no longer reacts with IAA (Figure 1); this is consistent with loss of selenol reactivity as the result of formation of a heterocycle.
We digested TruD- and PatD-treated PatE(Se) with the protease GluC, which cleaves after acid residues, to produce smaller molecules for more precise analysis. GluC treatment yielded ASITASITF(Sezn)AYDGELE but did not cleave between C-terminal E and L, possibly because the preceding Sezn heterocycle inhibited proteolysis. For the TruD-treated PatE(Se), we observed a doubly charged fragment of m/z 910.8847 that has the expected isotope pattern for the incorporation of selenium into the molecule and matches the predicted mass of ASITASITF(Sezn)AYDGELE (+6.5 ppm; Figure 2 A).
For the PatD-processed PatE(Se), we observed a doubly charged fragment of 892.8717 Da with a selenium isotope pattern that matches ASITASITF(SeCys)AYDGELE with three heterocycles (+4.0 ppm; Figure 2 B). Proteolytic fragments were analysed by MS/MS, and the peptide sequences were determined by analysing the internal cleavage, b- and y-ion series (Figure S1 A and B). We were able to sequence the peptides to within one residue of the expected Sezn on both the N- and C-terminal side. The remaining three-residue fragment (F(Sezn)A) is 18 Da lighter than expected for an unmodified peptide containing SeCys. Although the TruD-treated peptide gave extensive ion series to the N terminus, the PatD-treated peptide with multiple heterocycles could only be sequenced by combining multiple ion series (Figure S1 A and B).
We repeated the experiment by using a different PatE(Se) with the cassette sequence ITACITFC and observed (mass spectrometry) Se incorporation. After heterocyclisation with TruD, we treated the resulting peptide with protease to yield the peptide ITA(Sezn)ITF(Sezn)AYDGE, which was then macrocyclised with purified PatG11 in vitro, as previously described.11 The macrocycle has the expected mass for two Se atoms (Figure 2 C), and fragmentation analysis (Figure S2) confirmed the macrocycle structure. PatG will only macrocyclise a peptide with a heterocycle (or Pro) immediately N-terminal to the AYDGE motif.12 Taken together, these data establish that TruD and PatD are capable of heterocyclising SeCys.
The heterocyclisation reaction of SeCys by PatD or TruD was complete in 30 min. This is a similar time frame to the reaction of the natural substrate, thus indicating that SeCys is a suitable substrate for PatD/TruD. The reaction has been proposed to proceed through nucleophilic attack of the XH on the adjacent amide to form a hemiorthoamide intermediate.13 This intermediate attacks the γ-phosphate of ATP to form a phosphorylated hemiorthoamide, which subsequently decomposes to thiazoline and inorganic phosphate. Se is considered to be more nucleophilic than S or O therefore it should be more reactive if hemiorthoamide formation is rate limiting. We lack a robust assay to detect changes in rate of less than an order of magnitude and suggest any effect is within this range for the SeCys variant.
Sezn compounds might have different biological as well as chemical properties from their oxygen and sulfur relatives. Their widespread testing and thus exploitation have been limited by the difficulty of producing them in the laboratory. An enzymatic route from commercially available SeCys is a relatively benign method for the production of Sezn-containing natural products. Such Sezn variants could be directly useful or form part of larger synthetic schemes. Furthermore, as many natural products contain thiazolines/thiazoles or oxazolines/ oxazoles and thus a heterocyclase is used during biosynthesis, one can envisage biosynthetic routes to Se variants of several natural products. Where the pathway can be reconstituted in E. coli, production of SeCys variants can be trivially accomplished by medium supplementation; this enables production by both ribosomal and nonribosomal peptide systems. Where microorganisms other than E. coli have to be used, either growth medium can be supplemented with SeCys (akin to E. coli) or SeCys-containing peptide substrate (made either synthetically or in E. coli) can be added to crude extract or purified proteins.
It has been reported that ∼25 % of sequenced prokaryotic genomes contain the genes necessary for selenocysteine biosynthesis.14 The demonstration that two heterocyclases can process SeCys raises the prospect of significant numbers of selenium-containing peptide-derived natural products to be discovered.