Exploration of Iron- and a-Ketoglutarate-Dependent Dioxygenases as Practical Biocatalysts in Natural Product Synthesis

Abstract

Catalytic C─H oxidation is a powerful transformation with enormous promise to streamline access to complex molecules. In recent years, biocatalytic C─H oxidation strategies have received tremendous attention due to their potential to address unmet regio- and stereoselectivity challenges that are often encountered with the use of small-molecule-based catalysts. This Account provides an overview of recent contributions from our laboratory in this area, specifically in the use of iron- and α-ketoglutarate-dependent dioxygenases in the chemoenzymatic synthesis of complex natural products.

Introduction

The ability to oxidize C─H bonds in a controlled and selective manner holds tremendous promise in organic synthesis. The possibility of forging new C─X bonds in this manner not only offers the possibility of minimizing functional group interconversions and protecting group use, but also the opportunity to develop alternative retrosynthetic logic that is based on late-stage functionalization of complex molecules.¹ Additionally, selective C─H oxidation also enables access to unique building blocks to enrich the pool of starting materials available for multi-step synthesis. Given this potential, catalytic C─H oxidation methods have witnessed tremendous developments in recent years. One of the first catalytic systems for such purpose was reported by Shilov² in the use of platinum-based catalysts for the oxidation of simple hydrocarbons. Since then, a variety of transition-metal-based systems (Figure 1A),³ including metal-porphyrin catalysts,⁴ and bioinspired iron-based systems pioneered by Que,⁵ White⁶ and Costas⁷ has been developed, and some of these systems prove to be mild enough to allow C─H oxidation of a broad range of substrates. Organocatalytic strategies have also recently been developed.⁸ Despite these advances, contemporary small-molecule-based catalytic systems often suffer from poor selectivity profile, especially when applied in the functionalization of complex scaffolds. As C─H bonds are highly ubiquitous in organic molecules, designing catalytic systems that can differentiate small differences in C─H bond energy is highly challenging and many small-molecule-based systems rely on the innate stereoelectronic properties of their substrates or the use of directing groups for regioselective functionalization.

Figure 1.

A. Selected examples of small-molecule-based C─H oxidation catalysts. B. A brief overview of cofactors used in C─H oxidation enzymes.

In contrast, Nature has evolved a myriad of enzymes for catalytic C─H oxidation of small molecules (Figure 1B) and this idea was explored in the past in the use of whole-cell microbes for such purpose.⁹ While this approach garnered some success, the inability to control the outcome of the reaction ultimately limited its synthetic utility. Conversely, advances in molecular biology have provided us with the ability to identify the genes that encode for the enzymes involved in many biosynthetic steps, as well as the ability to recombinantly express and even engineer the corresponding enzymes. Such advancements have provided a greater degree of control and versatility in the exploration of biocatalytic methods, in particular to address unmet challenges in organic synthesis.¹⁰ In this Account, we highlight recent contributions from our laboratory to the field of biocatalytic C─H oxidation, especially in the use of non-heme dioxygenases in the synthesis of complex natural products.

Overview of Natural Oxygenases

Millions of years of evolution have given rise to several major oxygenase superfamilies, such as cytochrome P450s,¹¹ Rieske oxygenases,¹² flavin-dependent monooxygenases (FMOs)¹³ and iron-and α-ketoglutarate dependent dioxygenases (Fe/αKGs)¹⁴ that are all capable of performing site-selective oxidations of small molecules. Among these superfamilies, the P450s are considered to be the most widely-investigated for synthetic applications. Pioneering efforts from Arnold,¹⁵ Reetz,¹⁶ Sherman¹⁷ and others¹⁸ in the past two decades have demonstrated the utility of these enzymes in the biotechnological production of useful small molecules. Similarly, the FMOs have also been broadly studied as biocatalysts for a wide range of reactions,¹⁹ including oxidation of heteroatoms, halogenation, Baeyer-Villiger oxidation, and more recently, for oxidative dearomatization of phenols.²⁰ The synthetic application of Rieske oxygenases has predominantly focused on the use of toluene dioxygenases for the cis-1,2-dihydroxylation of aromatics.²¹ While several new Rieske dioxygenases have recently been discovered and characterized,²² the general sensitivity of their [2Fe-2S] cluster pose a significant challenge for their synthetic application. Fe/αKGs are characterized by the presence of a His¹-X-Asp/Glu-X_n-His² iron-binding motif and the use of αKG as a stoichiometric co-substrate in their reaction.²³ During catalysis, the decarboxylative decomposition of αKG to succinate is coupled to molecular oxygen activation to generate the active Fe(IV)-oxo oxidizing species. Substrate oxidation takes place via hydrogen atom abstraction, followed by radical rebound to generate the oxidation product. This enzyme superfamily has been known since 1966 during the identification of hydroxylated proline residues in collagen.²⁴ Several other members are also known to play important roles in key physiological processes,²⁵ including epigenetic regulation and post-translational modification of proteins. Curiously, with the exception of several early examples in their use for the production of β-lactam analogs,²⁶ very little work has been done to investigate the utility of Fe/αKGs in complex molecule synthesis when we began our studies.

Given this ‘lay of the land’, exploration of the biocatalytic potential of Fe/αKGs became the main focus of our laboratory early on. Several features of Fe/αKGs also render them attractive for further development as oxidation catalysts. The Fe/αKGs are self-sufficient enzyme and do not require any reductase partner for catalysis. While other oxygenases use NAD(P)H as reductant for the activation of molecular oxygen, reduction of the Fe center of Fe/αKGs from ferric to ferrous state can be effected by the use of inexpensive reductant such as sodium ascorbate. Furthermore, they are also generally smaller in size and exist as cytoplasmic proteins, which should lend to favorable recombinant expression. We were especially drawn to a subset of the superfamily that are capable of performing site-selective oxidation of free-standing amino acids (Figure 2).²⁷ As noted in the introduction, site-selective, remote C─H oxidation of organic small molecules remains an unaddressed challenge in chemical synthesis. C─H functionalization of amino acids in particular has been limited to methodologies that employ directing groups²⁸ or directed functionalization through the intermediacy of hydrogen atom abstraction.²⁹ Additionally, methods that allow remote C─H oxidation of amino acids will pave the way for subsequent synthesis of noncanonical amino acids by functional group interconversions of the newly-introduced hydroxyl group. Prior to our work, there have been numerous discoveries of amino acid hydroxylases belonging to the Fe/αKG superfamily,²⁷ though many of these discoveries were conducted mainly as part of mechanistic enzymology studies. Nevertheless, several reports, including those from the Ogawa group³⁰ and from a team at Bristol-Myers Squibb,³¹ suggest the potential utility of these enzymes as practical oxidation biocatalysts for large-scale reactions. It is also worth noting that more than 20,000 genes have been annotated as putative Fe/αKG-producing genes in both prokaryotes and eukaryotes and yet, only a small subset in prokaryotes has been validated as amino acid hydroxylases. This observation further motivated us to pursue a systematic exploration of the Fe/αKG superfamily as it will likely result in the discovery of novel amino acid hydroxylases.

Figure 2.

Selected examples of amino acid Fe/αKG hydroxylases.

C5 Hydroxylation of Aliphatic Amino Acids

At the outset, we were drawn to several Fe/aKGs that participate in the biosynthesis of 4-methyl-L-proline.³² Such motif arises in nature from a diastereoselective C5 oxidation of L-leucine, followed by intramolecular cyclization between the d-aldehyde and the α-amino groups and imine reduction. Of particular interest are the Fe/αKGs that catalyze the initial C5 oxidation due to the challenges associated with performing C─H functionalization at distal positions of amino acid structures with small-molecule catalysts or reagents. Furthermore, the unique capability of these enzymes to distinguish between the two diastereotopic methyl groups in L-leucine is unparalleled by any known contemporary chemical methods. Before we began our work, two leucine-5-hydroxylases, EcdK³³ and LdoA,³⁴ have previously been reported to catalyze the formation of (2S,4R)-5-(OH)-leucine and (2S,4S)-5-(OH)-leucine from L-leucine, respectively. A seminal study by Müller and co-workers also identified a putative leucine-5-hydroxylase from Streptomyces griseus, GriE that is predicted to produce (2S,4R)-5-(OH)-leucine, though this activity was not confirmed in the study.³⁵ As EcdK was reported to suffer from rapid inactivation, we became interested in functionally characterizing GriE to provide a better solution for the production of (2S,4R)-5-(OH)-leucine.

Initial experimentations with GriE revealed a highly active enzyme that is capable of hydroxylating L-leucine with very high total turnover number as judged by LCMS.³⁶ Despite this promising start, several key issues still needed to be addressed, especially with regards to standardized reaction procedure, product isolation and purification. Initially, three different reaction conditions were examined: reaction with purified GriE, reaction with whole-cell E. coli expressing GriE and reaction with lysates of E. coli cells expressing the enzyme. Reaction with purified enzyme suffered from scalability issues as reaction conversion was found to decrease considerably on > 100 mg scale. We also found that reaction with lysates of E. coli expressing the enzyme proceeded with much higher conversion than the whole-cell counterpart and this quickly became our preferred procedure when running the reaction on preparative scale. From these experimentations, our standardized procedure for enzymatic reaction with Fe/αKGs consists of: (1) resuspension of host E. coli cells in reaction buffer following overnight protein expression, (2) lysis with sonication, (3) addition of substrate and reagents to initiate reaction, and (4) analysis and purification after overnight reaction (ca. 20 hours).

As the reactions were conducted in aqueous buffer or E. coli cell lysates, isolation of water-soluble hydroxylated amino acid products also became a non-trivial process. This issue was compounded further by the challenge of separating the desired product from unreacted starting material when reaction did not proceed to completion. Initial purification attempts using reverse-phase HPLC on C18 column proved unsuccessful as the isolated product after purification was found to contain co-eluting buffer and reaction components (αKG, ascorbic acid and succinic acid), or unidentified cellular components when reaction was run with cell lysates. A temporary solution was found by derivatizing the products as the Boc or Fmoc counterparts. However, this approach was found to be less than ideal for getting accurate isolated yields of the biocatalytic reaction due to material loss and yield decrease associated with the derivatization step. Eventually, we identified ion-exchange chromatography to be a suitable purification technique, allowing the isolation of amino acid components of the reaction away from other impurities. In certain cases where unreacted starting material was observed in the mixture, further separation of the hydroxylated product could be achieved by performing Boc or Fmoc derivatization. Using this workflow, we were able to identify >10 aliphatic L-amino acids that are accepted as substrate by GriE and also deduce some reactivity trends (Figure 5A). We noted that GriE hydroxylates its substrate at the C5 carbon with high regio- and diastereoselectivity. Both the α-carboxylate and α-amino group in the substrate are required for productive catalysis and the reaction is highly sensitive to steric changes at the C3 position of the substrate. Finally, the active site of the enzyme can also accommodate substrates containing longer alkyl chains, as well as L-leucine derivatives that contain additional substituent at the C4 position.

Figure 5.

A. Structure and proteosome inhibition activities of cepafungin I (29) and glidobactin A (30). B. Chemoenzymatic synthesis of 29 via biocatalytic hydroxylation of lysine with GlbB. C. Structure-activity relationship studies of cepafungin I, highlighting the importance of the 2º alcohol on the macrocycle and dienamide unit in the lipid tail for activity.

A search for synthetic application of GriE led to the identification of manzacidin C (8) as a target (Figure 5B). Our approach towards manzacidin C was inspired by a biosynthetic hypothesis put forth by Ohfune³⁷ that the manzacidins arise in nature from L-leucine via two C─H functionalization events at C4 and C5 respectively. Thus, we devised a general vision for our synthesis featuring a C─H amination event at C4 and a biocatalytic hydroxylation at C5 with GriE. Initial efforts to realize this strategy focused on first performing the native reaction of GriE on L-leu, and then leveraging the newly-introduced C5 hydroxyl group for subsequent intramolecular C─H amination using Du Bois’ or He’s protocol.³⁸ This approach led to our first generation formal synthesis of manzacidin C that proceeds in nine steps overall.³⁹ Despite this success, we reasoned that a direct C─H amination route at C4 could offer a more direct approach to manzacidin C. Based on prior structure-activity relationship studies, we hypothesized that GriE could accept as substrate a leucine derivative that contains an amino group surrogate at C4. A report on photocatalytic C─H fluorination of leucine by Britton and co-workers⁴⁰ inspired us to develop similar C─H azidation protocol employing tetrabutylammonium decatungstate (TBADT) as photocatalyst and 2,5-bis(trifluoromethyl)benzenesulfonyl azide as the azide source. Here, photoexcitation of TBADT would generate an activated species that performs C─H abstraction at C4 of leucine and the resulting tertiary radical would next undergo homolytic recombination with the azide source to afford the azidated product. Using this protocol, 4-azidoleucine (9) could be obtained from L-leucine in 49% isolated yield. Subjection of 9 to biocatalytic hydroxylation with GriE afforded 10 with >95% conversion. Finally, conversion of 10 to lactone 7 could be effected by hydrogenation under basic conditions, followed by in situ treatment with Boc₂O. This approach constitutes a five-step formal synthesis of manzacidin C and compares favorably to prior approaches to this target.

During substrate scope characterization of GriE, we observed formation of over-oxidation products such as imine 11 at high enough enzyme concentrations (Figure 5C). This observation led us to develop a one-pot chemoenzymatic cascade to prepare a range of 4-substituted proline derivatives with GriE. Following complete over-oxidation of the substrate with GriE, the reaction mixture was treated with NH₃•BH₃ to effect reduction of the cyclic imine. This approach allowed us to prepare five different 4-substituted proline derivatives with high yields and diastereoselectivity. Finally, we showed that the biocatalytic formation of (2S,4R)-4-MePro could conducted on preparative scale (100 mg) with moderate yield after Fmoc protection and that the resulting product could be readily incorporated into solid-phase peptide synthesis workflow to provide the first total synthesis of the antiviral lipopeptide cavinafungin B (14).⁴¹ As an extension of this strategy, we also developed a chemoenzymatic approach to (2S,3R)-3-hydroxy-3-methylproline, a rare amino acid motif found in two natural products polyoxypeptin A and FR225659 (Figure 5D).⁴² Our approach commenced with iterative oxidation of L-Ile with UcsF, an isoleucine 5-hydroxylase from UCS-1025A biosynthesis,⁴³ to the corresponding cyclic imine, followed by in situ treatment with NH₃•BH₃ to afford proline derivative 15. The final C3 hydroxylation was performed with GetF, a pipecolic acid hydroxylase from GE81112 biosynthesis that has previously been shown to exhibit promiscuous activity,⁴⁴ to complete the synthesis of the noncanonical amino acid target (16), following Boc protection for ease of isolation.

Chemoenzymatic Synthesis of Tambromycin

Based on the initial success described above, we were motivated to further extend such ‘hydroxylation-centric’ synthetic strategy to other alkaloidal or peptidic natural products. One particular nonribosomal peptide that caught our attention was tambromycin, a linear tetrapeptide isolated from several Streptomyces species through a metabologenomics approach in 2016.⁴⁵ Tambromycin contains several unusual structural motifs, namely a trisubstituted indole motif, two 2-methyl-L-serine units and a highly unusual pyrrolidine-containing amino acid called L-tambroline (Tam). Preliminary mechanistic investigations suggest that Tam arises via desaturation of L-Lys at its C2─C3 bond, followed by formation of the pyrrolidine ring via an intramolecular 1,4-addition of the e-amino group. While the enzymes that catalyze this series of reactions had not been characterized, we hypothesized that an analogous formation of protected Tam could be effected through an S_N2-type displacement at the C3 carbon of protected 3-hydroxy-L-lysine (23). In turn, 23 would be accessed via biocatalytic hydroxylation with KDO1, an Fe/αKG enzyme from Catenulispora acidiphila that was previously discovered by Zaparucha and co-workers through genome mining (Figure 4A).⁴⁶

Figure 4.

A. Structure and retrosynthetic analysis of tambromycin (17). B. Chemoenzymatic synthesis of tambromycin featuring biocatalytic C─H hydroxylation of lysine with KDO1.

Towards this goal, we developed a gram-scale access to 23 by optimizing the biocatalytic hydroxylation with lysates of E. coli expressing KDO1 (Figure 4B). Key in this process is the discovery that the soluble expression of KDO1 could be improved considerably when chaperones GroES/EL are co-expressed.⁴⁷ Following Boc protection and benzyl ester formation, cyclic sulfamidate 25 was prepared via treatment with SOCl₂ and oxidation. Simple heating of this compound in DMF effected a clean formation of 26, which constitutes the protected version of Tam. It is worth noting that other approaches to effect the intramolecular displacement using Mitsunobu or Appel conditions led only to E1Cb elimination of the 3-OH group. Concurrently, we developed a concise synthetic access to the trisubstituted indole motif (19) by employing a C─H borylation strategy to install the 6-Cl group.⁴⁸ Two key fragments, 27 and 28, were next assembled via conventional peptide couplings of their respective building blocks. The final coupling between 27 and 28 could be effected cleanly to afford protected tambromycin despite the presence of various nucleophilic groups in the two precursors. Finally, a routine saponification removed the methyl ester protecting group to complete our synthesis of tambromycin. This strategy provides an improved access to tambroline as prior approaches to this motif either require the use of toxic or high-mass reagents (e.g. CH₂N₂, diethyl azodicarboxylate and diphenyl phosporyl azide)⁴⁹ or proceed with minimal stereocontrol.⁵⁰ In addition, it also demonstrates the advantage of combining new advances in biocatalytic and chemical C─H functionalization in streamlining the synthesis of complex natural product.

Chemoenzymatic Synthesis of Cepafungin I and Related Analogs

In addition to providing opportunity to further push the boundary of biocatalytic retrosynthetic logic, our ‘hydroxylation-centric’ chemoenzymatic approach also opens a new avenue for medicinal chemistry exploration of biologically-relevant natural products. This virtue is exemplified by our recent investigation into the syrbactin natural product family. The syrbactins are hybrid peptide-polyketide natural products that share a common 12-membered macrolactam ring and are known to be potent inhibitors of the 20S proteasome.⁵¹ Recently, there has been a surge of interests in proteasome inhibitors due to the central role that the proteasome plays in the regulation of cell cycle and apoptosis.⁵² Several proteasome inhibitors such as bortezomib and carfilzomib have also been approved for clinical use in the treatment of multiple myeloma.⁵³ Within the syrbactins, cepafungin I attracted our attention due to a report in 2012 that described it as a single-digit nanomolar inhibitor of purified yeast 20S proteasome.⁵⁴ Examination of the structural features of cepafungin I, as well as prior synthetic efforts towards the natural product family,⁵⁵ led to the recognition that a streamlined access to 4-hydroxylysine would pave the way for a concise and modular synthesis of the natural product and related analogs. Our investigation into the biosynthetic origin of cepafungin I/giodbactin A in Polyangium brachysporum revealed the involvement of an Fe/αKG enzyme, GlbB, in the hydroxylation of free L-lysine prior to the loading of the hydroxylated product into the hybrid polyketide synthase/nonribosomal peptide synthase module.^56,57 Moreover, GlbB was shown to be a highly efficient enzyme capable of hydroxylating L-lysine with very high total turnover number (5,900).

The aforementioned discovery enabled a rapid access to 4-hydroxylysine for our chemoenzymatic synthesis of cepafungin I.⁵⁸ Similar to what was observed in our synthesis of tambromycin, soluble expression of GlbB was dramatically improved by co-expression of GroES/EL, which enabled multi-gram preparation of 4-hydroxylysine (31) with E. coli lysates on a single pass. Compound 31 was protected as its Boc derivative and intramolecularly cyclized to the corresponding butyrolactone, which in turn was converted to dipeptide 33 via aminolysis with alanine derivative 32 in the presence of AlMe₃. Weinreb amide reduction and Wittig homologation afforded 35, which could be exhaustively deprotected, cyclized intramolecularly and coupled with the appropriate tail fragment (36) to complete the synthesis of cepafungin I (29). This route proceeded in nine steps (longest linear sequence) and ca. 8% overall yield, and provides one of the shortest synthetic access to the syrbactins.

The route described above also facilitated access to several analogs for chemoproteomics studies and preliminary structure-activity relationships between cepafungin I and 20S proteasome. Firstly, we synthesized 38, a structural analog of 29 bearing a terminal alkyne moiety to profile the cellular targets of the natural product scaffold. Gel-based profiling with this probe suggested that cepafungin I is highly selective for the 20S proteasome subunits in multiple myeloma cells. LC-MS/MS-based in situ competitive pull-down experiment further validated this observation, identifying only the 20S proteasome subunits as the high affinity targets of cepafungin I. To inform future medicinal chemistry optimization of the scaffold, we next synthesized several analogs to investigate the importance of several key subunits for proteasome binding. Approximately 10-fold reduction in potency was observed upon removal of the macrocyclic secondary alcohol and approximately 14-fold loss in potency was observed upon replacement of the unsaturated lipid tail with its saturated counterpart. This data provides key insights into the importance of these functional groups, especially the macrocyclic alcohol, for proteasome inhibition. Furthermore, this report lays the groundwork of further medicinal chemistry development of cepafungin analogs as proteasome inhibitors through structure-based design.

Chemoenzymatic Synthesis of GE81112 B1 and Related Analogs

We have also recently shown the synthetic utility of Fe/αKG hydroxylation in the preparation of the tetrapeptide antibiotic GE81112 B1.⁵⁹ The GE81112s are three tetrapeptide congeners that were isolated in 2006 from Streptomyces sp. L-49973 and shown to be inhibitors of prokaryotic translation initiation.⁶⁰ Further mechanistic studies⁶¹ reported in 2016 show that the GE81112s exhibit an inhibitory mechanism that is distinct from other known translation initiation inhibition, suggesting its potential utility for further development as a novel antibacterial agent. Preliminary investigation by Müller and co-workers in 2010 established the general framework of GE81112 biosynthesis and revealed several enzymes likely responsible for the tailoring of the amino acid monomers prior to assembly by the NRPS module.⁶² Two Fe/αKGs, GetF and GetI, could be identified in the biosynthetic gene cluster. Annotated as a putative pipecolic acid hydroxylase, the function of GetF was validated and characterized by Hüttel and co-workers in 2017.⁴⁴ Conversely, GetI was initially predicted to be the hydroxylase involved in the biosynthesis of the hydroxychlorohistidine subunit of GE81112 though no functional characterization was performed to support this hypothesis. In 2019, we determined that GetI functions as citrulline-4-hydroxylase, instead of chlorohistidine hydroxylase, and is likely responsible for the biosynthesis of the second monomer of all three GE81112s.⁶³ Furthermore, sequence-based rational mutagenesis identified a variant of GetI containing four mutations away from wild-type that exhibits a complete switch in substrate preference from citrulline to arginine.

The functional characterization of GetI provided an opportunity to chemoenzymatically construct the first two amino acid monomers of GE81112 B1. Biocatalytic hydroxylation of L-Pip and L-Cit was performed on preparative scale with GetF and GetI, respectively to provide the desired hydroxylated amino acid products with high reaction conversions and yields. Following protecting group manipulations, fragment coupling between the two monomers in the presence of EDC and oxyma completed the synthesis of our first key fragment towards GE81112 B1 (50). The remaining two monomers were prepared via conventional synthetic chemistry. Extensive optimization of an azo coupling procedure⁶⁴ on Boc-L-His-OMe (44) revealed azo compound 45 to be an optimal coupling partner to provide 48% isolated yield of 46, which constitutes a protected surrogate of the desired 2-amino-L-histidine fragment. Several asymmetric aldol variants were examined for the preparation of the final monomer.⁶⁵ Among the conditions tested, a variant employing thiazolidinone chiral auxiliary (47) was able to provide the desired aldol adduct in 59% yield as a single diastereomer, which in turn was reduced to the free amine (49) in the presence of (NH₄)₂S. Standard peptide coupling and selective Boc deprotection furnished dipeptide 51, which could be combined with 50 to provide the protected form of GE81112 B1. Three consecutive deprotection steps completed the synthesis of GE81112 B1, which proceeded in eleven steps (longest linear sequence) from known compounds. This route was adapted for the preparation of five simplified analogs of GE81112 and one analog containing 3-OH-Pro in place of 3-OH-Pip at AA1, which in turn allowed us to elucidate the contributions of various peripheral modifications on the tetrapeptide towards its antimicrobial activity. We identified the importance of the syn-β-OH amino acid unit at AA1 and the β-OH moiety at AA4 for antimicrobial activity. Rather surprisingly, the γ-OH moiety at AA2 was found to provide virtually no contribution towards antimicrobial activity. Taken together, our chemoenzymatic approach provided a tractable synthetic access to GE81112 B1 and related analogs and provided the first structure-activity relationship data on GE81112’s antimicrobial activity.

Conclusion and Future Direction

This account provides a brief summary of our efforts in the past four years to demonstrate the utility of Fe/αKG biocatalysts in complex molecule synthesis. Though these efforts have resulted in an encouraging outlook, there are still important points to be addressed to further advance the field. Firstly, we still have relatively limited toolkit available at our disposal for the biocatalytic C─H functionalization/hydroxylation of amino acids. A quick survey of the literature suggests a significant discrepancy between the number of functionally characterized amino acid hydroxylases to date and the number of discrete enzymes that are theoretically needed to hydroxylate every unique unactivated C(sp³)─H bond in canonical amino acids assuming completely orthogonal substrate specificity among the enzymes (Figure 7A). A quick survey of the literature shows that there are numerous bioactive peptide natural products containing unusual hydroxylated amino acid components (Figure 7B).⁶⁶ We believe that the syntheses of these molecules could be greatly simplified if the appropriate amino acid hydroxylases could be identified. Based on these observations, we expect that discovery of new Fe/αKGs with novel reactivity profile and substrate specificity will continue to be a fruitful endeavor. Given the sheer number of enzymes that have been annotated as Fe/αKGs in public databases, efforts in this area will likely need to incorporate elements of genome mining and bioinformatics analyses⁶⁷ to significantly reduce the screening burden.

Figure 7.

A. Comparison between the number of functionally characterized amino acid hydroxylases to date and the number of discreet enzymes that are theoretically needed to hydroxylate every unique unactivated C(sp³)─H bond in canonical amino acids assuming completely orthogonal substrate specificity. Note that aromatic C─H bonds and C─H bonds adjacent to heteroatoms are not considered in this tabulation. B. Representative examples of peptide natural products that contain unusual hydroxylation pattern. C. Chemoenzymatic synthesis of podophyllotoxin featuring oxidative coupling with 2-ODD-PH. D. Chemoenzymatic synthesis of fujenoic acid featuring biocatalytic hydroxylation with PtmO6.

In order to illustrate more broadly the biocatalytic potential of Fe/αKGs, it is also imperative that we show their utility in the synthesis of other natural product families as well. To this end, our laboratory has recently made headway in showcasing the application of two Fe/αKGs, 2-ODD-PH and PtmO6, in the chemoenzymatic total synthesis of the aryltetralin lignans (Figure 7C)⁶⁸ and oxidized ent-kauranes (Figure 7D),⁶⁹ respectively. Efforts from other laboratories⁷⁰ have also asserted the general utility of this enzyme superfamily in organic synthesis. Based on these precedents, further efforts in characterizing new Fe/αKGs and exploration of their biocatalytic potential promise to significantly expand the pool of useful biocatalysts available at our disposal and the types of complex scaffolds that can be accessed in a chemoenzymatic fashion.

Figure 3.

A. Selective C5 hydroxylation of aliphatic amino acids with GriE. B. Application of biocatalytic hydroxylation with GriE in the formal synthesis of manzacidin (8). C. Chemoenzymatic synthesis of 4-methylproline analogs through a combination of C─H oxidation with GriE and in-situ imine reduction and its utility in the synthesis of cavinafungin B (14). D. Chemoenzymatic synthesis of protected (2S,3R)-3-hydroxy-3-methylproline (16) through C─H oxidation with UcsF and GetF.

Figure 6.

A. Structure of GE81112 B1 (41) and organization of its biosynthetic gene cluster. B. Synthesis of four key monomers to GE81112 B1 featuring biocatalytic hydroxylation of L-pipecolic acid and L-citrulline with GetF and GetI, respectively. C. Completion of our chemoenzymatic total synthesis of GE81112 B1 (41). D. Minimal inhibitory concentration measurements of GE81112 B1 and six related analogs against E. coli MG1655.

Biography

Hans Renata received his B.A. degree from Columbia University in 2008 and earned his Ph.D. from The Scripps Research Institute in 2013 under the guidance of Prof. Phil S. Baran. After postdoctoral studies with Prof. Frances H. Arnold at Caltech, he began his independent career at The Scripps Research Institute in 2016. His research focuses on synthetic and biosynthetic studies of natural products and biocatalytic reaction developments.