Studies on the selectivity of proline hydroxylases reveal new substrates including bicycles

Graphical abstract


Introduction
Hydroxylated amino acids are common starting materials for the synthesis of pharmaceuticals and agrochemicals. They are also intermediates in the biosynthesis of natural products with interesting biomedicinal properties, including many antimicrobials [1][2][3]. Recent work has also expanded the set of identified hydroxylated protein residues, which, although not as large as that of hydroxylated amino acids, is extensive [1].
The ferrous iron-and 2-oxoglutarate (2OG)-dependent dioxygenases (ODDs) possibly catalyse the widest range of oxidations of amino acids and proteins of any identified enzyme family [1][2][3]. Prolyl hydroxylation of collagen was the first identified reaction catalysed by 2OG oxygenases and occurs to give C-3-and C-4-hydroxyprolyl protein products. Analogous prolyl hydroxylations of many collagen-like proteins have been identified [1]. Trans-C-4-prolyl hydroxylation of the hypoxia inducible transcription factor plays a key role in the adaptive response to hypoxia in animals [4]; related modifications have been identified in lower eukaryotes and prokaryotes [5,6]. Trans-C-3-prolyl hydroxylation also occurs in a ribosomal protein (RPS23) in eukaryotes ranging from yeasts to humans [7,8]. Interestingly, the yeast, but not the human, ribosomal prolyl hydroxylase catalyses a second hydroxylation of the same residue to give a dihydroxylated product [7].
In connection with the work on prolyl hydroxylation aimed at defining regio-and stereo-selectivities, we are interested in exploring the selectivity of proline (1) hydroxylases and using them to prepare standards for amino acid analyses, associated with the functional assignment of protein mono and di-hydroxylases. Aside from their use in preparing standards for structure validation, 2OG oxygenases have considerable potential as industrial scale biocatalysts, as demonstrated by the use of a recombinant L-proline (1) trans-4-hydroxylase (trans P4H) [9].
products were detected for cisP3H and one for transP4H (we were unable to assign the stereochemistry of these by NMR) (Fig. 5d).

Conclusions
2OG dependent amino acid hydroxylases have considerable value as industrial catalysts [9]. Our results substantially expand the biocatalytic potential of pH catalysed hydroxylations, including with the demonstration that certain N-alkylated amino acids and bicyclic ring systems are accepted as PH substrates. The products produced, including di-and tri-hydroxylated prolines should be useful in ongoing  (Table S7).
The observation that substrate analogues with four, six, and seven membered ring containing substrate analogues undergo PH catalysed oxidations, with dihydroxylated products being produced in some cases, is of interest from a biocatalytic perspective ( Fig. 4 and 5). Some substituted prolines, including N-methylated prolines, were also found to be PH substrates giving triply functionalised pyrrolidine rings (Fig. 4). However, N-acylated or (2R) -amino acids were not PH substrates (Table S6).
The observation of two products with some bicyclic substrate/PH combinations is mechanistically interesting, suggesting that the ferryl intermediate in 2OG-dependent PH catalysis [1] may be positioned approximately equidistant between different potentially oxidised CeH bonds (Fig. 6). However, a mechanism involving hydrogen abstraction at one site, followed by hydrogen atom transfer and hydroxylation at a second site cannot be ruled out. There is a possibility of such processes occurring in 2OG oxygenase catalysis in carbapenem antibiotic biosynthesis [34]. It is possible that such processes occur in 2OG oxygenase catalysis involving protein/nucleic oxidations.   (Table S7).
2OG oxygenases have been shown [1][2][3] to be amenable to structure based protein engineering to produce enzymes with improved product selectivities with unnatural substrates [1,9]. Although this work is at an early stage with PHs, it has been reported that a directed-evolution approach has led to a modified cisP4H (with 3 substitutions) with improved activity with respect to hydroxylation of L-pipecolic acid to give (2S,5S)-5-hydroxy-L-pipecolic acid [27]. The relaxed substrate/product selectivities of PHs, and some, but not all, other 2OG oxygenases (including some catalysing protein modifications, e.g. factor inhibiting HIF) suggests that they may be suitable for 'late stage modifications' of valuable chemicals; following the initial identification of a desired product, its oxygenase catalysed production may be optimised by protein engineering. There would, thus, appear to be considerable, as yet largely unexplored, potential for (engineered) PHs, and other 2OG oxygenases, in biocatalysis.

Materials and methods
DNA encoding for Streptomyces sp. (strain TH1) cisP3H type I, Sinorhizobium meliloti cisP4H, and Dactylosporangium sp. transP4H (with 5′-NdeI and 5′-HindIII sites) was synthesised by Life Technologies' GeneART, and codon-optimised for E. coli expression. The PH genes were then subcloned into a Takara pCOLD TM I expression vector to produce N-terminally His 6 -tagged PHs. Sequence-verified clones (by The Gene Service, Source BioScience, Oxford, UK) were transformed into Stratagene BL21-Gold (DE3) competent cells for protein production. Reagents, casting equipment, and electrophoresis tanks for SDS-PAGE were obtained from Bio-Rab Laboratories; molecular-mass markers (Thermo PageRuler Plus) were from Thermo Scientific. Reagents for PH reactions were from Sigma-Aldrich and Fluka. Full details are given in the Supplementary Information.

Declaration of Competing Interest
There is no conflict of interest.