2024 Volume 72 Issue 3 Pages 241-247
Natural products from plants and microorganisms provide a valuable reservoir of pharmaceutical compounds. C–C bond formation and cleavage are crucial events during natural product biosynthesis, playing pivotal roles in generating diverse and intricate chemical structures that are essential for biological functions. This review summarizes our recent findings regarding biosynthetic enzymes that catalyze unconventional C–C bond formation and cleavage reactions during natural product biosynthesis.
Natural products synthesized by living organisms offer a rich source of pharmaceuticals.1–3) Enzymatic C–C bond formation and cleavage reactions are central to the construction of the intricate carbon frameworks that define the unique structures of natural products. These reactions involve the activation of specific substrates and their subsequent transformation into key biosynthetic intermediates through well-coordinated chemical mechanisms. The diversity of natural products arising from these biosynthetic pathways reflects the versatility of the enzymes catalyzing the formation and cleavage of various C–C bonds. This review summarizes our recent efforts to characterize unique enzymes catalyzing unusual C–C bond formation and cleavage reactions during the biosynthesis of natural products such as hormaomycins, belactosins, norlignans, and tolyporphins.
Hormaomycin (1) and belactosin A (2), distinctive peptide natural products from several Streptomycetes, feature unique structures with cyclopropane rings (Fig. 1a). Hormaomycin (1) contains uncommon (amino) acid units, including (1′R,2′R)-3-(2-nitrocyclopropyl)alanine (3, Ncpa).4–9) Belactosin A (2) has a β-lactone moiety conjugated with a (1′R,2′S)-3-(2-aminocyclopropyl)alanine (4, Acpa) residue.10–12) The structural similarity between (1′R,2′R)-Ncpa (3) and (1′R,2′S)-Acpa (4) suggests a shared biosynthetic pathway involving L-lysine as a precursor.13–16)
(b) Stereodivergent cyclopropanation of 6-nitronorleucine (5). (c) Possible reaction mechanisms. (d) Analysis using substrate analogues.
Based on the information about the hormaomycin (1) and belactosin A (2) biosynthetic gene clusters, hormaomycin (1) is proposed to be assembled by NRPSs, while ATP-grasp ligases construct the peptide bonds in belactosin A (2).13,14) Several homologous genes are present in the two biosynthetic gene clusters. Gene deletions in the belactosin A-producing strain Streptomyces sp. UCK14 and in vitro enzyme assays revealed that the diiron enzymes BelK and HrmI catalyze the oxidation of L-lysine to generate L-6-nitronorleucine (5).17–19) Subsequently, the nonheme iron enzymes BelL and HrmJ perform the dehydrogenative cyclization of the nitroalkane moiety of L-6-nitronorleucine (5) to produce Ncpa (3).17,18) Despite the moderate sequence identity between BelL and HrmJ (49% identity), BelL produces (1′S,2′S)-Ncpa, while HrmJ produces (1′R,2′R)-Ncpa, consistent with the stereoconfigurations found in belactosin A (2) and hormaomycin (1) (Fig. 1b). In belactosin A (2) biosynthesis, the reduction of the nitro group in (1′S,2′S)-Ncpa ((1′S,2′S)-3) is predicted to be catalyzed by the molybdenum-dependent reductase BelN.17,18,20)
Possible mechanisms for the cyclopropanation of L-6-nitronorleucine (5) are illustrated in Fig. 1c.17,21) The α-carbon (C6) of the nitro group in L-6-nitronorleucine (5) may undergo deprotonation by a base in the active site to generate the nitronate intermediate 6. Subsequently, an Fe(IV)–oxo species abstracts an H atom from the C4 position, forming the radical intermediate 7. Various pathways are conceivable, including radical addition across the double bond of the nitronate moiety to create the cyclopropane ring with the nitro anion radical 8. Further oxidation of this species could lead to the production of Ncpa (3). An alternative mechanism involving the cation intermediate 9 should also be considered. The radical in 7 could receive the hydroxyl group from the Fe(III) to form the C4–OH intermediate 10, which might undergo cyclization into 3 via an intramolecular substitution reaction. Although the C6-deprotonation of L-6-nitronorleucine (5) could occur non-enzymatically at pH 7.5, the addition of BelL accelerates this reaction. This deprotonation is crucial for the cyclization activity, as evidenced by the inability of HrmJ to accept the fluorinated substrate analogue 13 that cannot be deprotonated21) (Fig. 1d). In addition, the substrate analogue 11, containing a carboxylate moiety as a mimic of the nitro group in 5, is only hydroxylated without the formation of the cyclized product21) (Fig. 1d). Moreover, the enzymatic cyclization of L-6-nitronorleucine (5) is enhanced under basic conditions (pH 9–11), aligning with C6 deprotonation as a critical step.21)
Similar to other αKG-dependent nonheme iron enzymes,22–26) BelL and its homologues utilize an Fe(IV)-oxo species to abstract an H atom from the substrate or 6.17,21) This has been confirmed through transient kinetic analyses using stopped-flow absorption spectroscopy and Mössbauer spectroscopy.21) The pro-S-H is selectively abstracted from C4, as indicated by the loss of the deuterium atom when (4S)-[4-2H]-5 is used as the substrate17) (Fig. 1c). Intriguingly, all tested homologues of BelL similarly abstract pro-S-H, irrespective of the diastereoselectivity of the reactions they catalyze. The radical 7 is likely cyclized directly into 3 via either a cation- or radical-based mechanism, as the synthetic 4-hydroxyl compound 14 cannot be converted to Ncpa (3), effectively excluding the possibility of the involvement of 14 (and 10) as an on-pathway intermediate during the cyclopropanation. Density functional theory (DFT) calculations suggested that the radical-based cyclization of 7 to 8 is energetically feasible.17) The delocalization of spin density over the three heteroatoms may contribute to the stabilization of 8, thus overcoming the inherent ring strain of the cyclopropane ring.17)
BelL and its homologues exhibit notable stereochemical control during the dehydrogenative cyclization of L-6-nitronorleucine (5) (Fig. 1b). The mechanism for stereochemical control was also investigated based on structural analyses of this enzyme class.21) The X-ray crystal structure of ScBelL from Streptomyces cavourensis NBRC 13026 revealed a jelly roll fold with two anti-parallel β-sheets. Sequence alignments and comparisons of AlphaFold2 models of the tested cyclopropanases identified conserved active site residues, with a few amino acid differences. For instance, BelL has histidine (His) at position 136, while the corresponding residue in HrmJ is arginine (Arg).21) These residues play a crucial role in controlling the product stereochemistry, as demonstrated by the variant analysis. The BelL-H136R variant produces (1′R,2′R)-3 without forming (1′S,2′S)-3.21) Similarly, the HrmJ-R136H variant generates (1′S,2′S)-3, while retaining the activity to produce (1′R,2′R)-3. However, further investigations are required to elucidate how these cyclopropanases control the selectivity by the subtle differences in their primary sequences.21,27)
1.1. Decarboxylative Rearrangement during Norlignan BiosynthesisNorlignans are a group of natural phenolic compounds characterized by a diphenylpentane carbon skeleton (C6–C5–C6), and are predominantly found in conifers and monocotyledons.28–30) Hinokiresinols (16 and 17), the structurally simplest norlignans, are synthesized by various plant species.31–33) For instance, the geometric isomers (Z)-hinokiresinol (16) and (E)-hinokiresinol (17) have been identified in Asparagus officinalis and Cryptomeria japonica, respectively34–36) (Fig. 2). Hinokiresinols are considered to function as protective agents for plants, due to their antifungal properties.37) These compounds originate through the coupling of two phenylpropane (C6–C3) units produced in phenylpropanoid biosynthesis.38–40) In the final step, the decarboxylative rearrangement of 4-coumaryl 4-coumarate (15) is catalyzed by hinokiresinol synthase in a stereoselective manner, yielding either (Z)- or (E)-hinokiresinols (16 or 17) with the diphenylpentane carbon skeleton (C6–C5–C6).41,42)
Hinokiresinol (HRS) synthase from A. officinalis comprises two subunits, HRSα and HRSβ, sharing moderate sequence identity of 54%.41,42) HRSα and HRSβ are essential for converting 4-coumaryl 4-coumarate (15) into (Z)-hinokiresinol (16) with near-complete diastereo- and enantioselectivity41,42) (Fig. 2). When one of the subunits is used, (E)-hinokiresinol (17) can be formed with compromised enantioselectivity. HRS is annotated as belonging to the phloem protein 2 (PP2) family,43,44) whose members are involved in phloem-based defense as structural proteins capable of binding carbohydrates displayed by plant pathogens.
Possible mechanisms for the rearrangement reaction of 15 are shown in Fig. 3.41,45) In Path A, the ester enolate 18 is first formed by deprotonation from the phenolic oxygen at C4, and then undergoes a Claisen rearrangement to generate 19 (path A). Alternatively, deprotonation of the phenolic oxygen at C4 in 15, C8–C7′ bond formation, and C–O bond cleavage may occur in a concerted manner to generate 19 (path B). Another possibility is that the deprotonation of the phenolic oxygen at C4′ may trigger the fragmentation of 15 into two components, the coumarate 20 and the p-quinone methide 21 (path C). Intermediate 19 may then be formed through the recombination between 20 and 21, producing a bond between C8 and C7′. Following the formation of 19 via one of the three possible mechanisms, decarboxylation of 19 would subsequently result in the production of 16.
Substrate analogs (22, 23, and 26) were designed and utilized to investigate the potential mechanisms.45) Compound 22 lacking the C4′–OH was not consumed in the incubation with HRS, contradicting paths A and B (Fig. 4). Conversely, 23 lacking the C4–OH was accepted by HRS and cleaved into 24 and 25, although no C–C bond formation occurred. Product 25 could have resulted from the hydration of the extended quinone methide 21 (Fig. 3). Compound 26, containing a C7–C8 single bond, was also cleaved by HRS to form 27 and 25. These observations collectively support path C, highlighting the crucial role of C4′–OH deprotonation.
X-ray crystal structures of the heterodimeric HRS were pursued to elucidate the structural basis of the rearrangement reaction.45) The dimer of the apo HRS contains a cavity between the two subunits (Fig. 5a). The structure of the heterodimer in complex with the inhibitor 29 was also solved (Fig. 5b). The inhibitor 29 is located within the pseudo C2 symmetric cavity of the HRS dimer, adopting a bent conformation with a C8–C7′ distance of 4.0 Å. The conformation of 29 in the heterodimer cavity seems to be influenced by both hydrophobic and hydrophilic residues of HRS, and may represent that of 15 before catalysis.
(b) Active site of HRS in complex with the inhibitor 29. Adapted with permission from ref 45. Copyright 2023 American Chemical Society.
In vitro assays were performed with the HRS heterodimer variants to analyze the functions of the polar active site residues.45) The α-D165A and α-D165N mutants produced 17 and 28, respectively (Fig. 4). The formation of 28 likely resulted from the nucleophilic attack of C8 in 20 on the terminal methylene C9′ of 21 instead of C7′, followed by decarboxylation. Thus, α-Asp165 contributes to both the regioselectivity and stereoselectivity of the HRS reaction by dictating the positioning and conformation of 20, as well as those of 21, by hydrogen bonding to the phenol oxygen at C4′. The substitution of α-Lys164 with Ala abolished the activity toward 15, consistent with the proposed function of α-Lys164 as a base to deprotonate the oxygen at C4′ and initiate the fragmentation of 15 into 20 and 25. In contrast, the β-K168A variant was still active toward 15, but generated only trace amounts of the rearrangement products (Fig. 4). Instead, the levels of 20/25 were significantly higher than those in the control reaction without enzyme, implying the enzymatic cleavage of 15. This may be initiated by the deprotonation of the phenol oxygen at C4′ by α-Lys164. The intermediate 20/21 could be released from the cavity in the absence of β-Lys168, and then hydrated to generate 25. Therefore, β-Lys168 is proposed to facilitate the efficient coupling of 20 and 21 by deprotonating the phenol oxygen at C4 of 20. The reaction of 15 with β-D169A or β-D169N generated the rearrangement products 16, 17, and 28 with reduced catalytic activity. Thus, the polar interactions between the phenol oxygen at C4 and β-Asp169/β-Lys168 are crucial for the binding and selective conversion of 15.
Our findings revealed that, instead of the concerted Claisen rearrangement mechanism, substrate 15 undergoes initial cleavage into the coumarate 20 and the extended p-quinone methide 2145) (Fig. 6). Subsequently, these fragments recombine to generate the intermediate 19 with a newly formed C8–C7′ bond. This stepwise fragmentation and recombination process is facilitated by precisely positioned acidic and basic amino acid residues within the cavity. Prior to the decarboxylation step, a conformational change in 19 may be necessary to produce the hinokiresinol (16) product with the desired cis geometry. Hydrophobic residues in the active site also play crucial roles in controlling the regioselectivity of the coupling of fragmented intermediates 20 and 21, as well as the stereoselectivity of the decarboxylation step.
Tolyporphins, exemplified by compounds 30 and 31, represent structurally unique tetrapyrrole secondary metabolites produced by Nostocales cyanobacterium HT-58-2 cultures46–51) (Fig. 7a). Tolyporphin A (30), the most abundant member, features a dioxobacteriochlorin macrocycle functionalized by two units of 2-O-acetyl-abequose via C-glycosidic bonds. Notably, two pyrrole β sites (C2 and C13) in the macrocycle are unsubstituted, distinguishing them from typical tetrapyrrole cofactors.
(b) Biosynthetic pathway of tolyporphin A. (c) A possible mechanism of the TolI-catalyzed vinyl cleavage reaction.
HemF1, encoded by the tol gene cluster, shares high similarity with the canonical HemF, catalyzing decarboxylations of the two propionate side chains on rings A and B of coproporphyrinogen III (35)52) (Fig. 7b). The tol gene cluster also encodes another HemF-like protein, HemF2, which exhibits moderate similarity to HemF. In vivo and in vitro assays demonstrated that HemF1 functions similarly to HemF, converting coproporphyrinogen III (35) to protoporphyrinogen IX via double decarboxylations52) (Fig. 7b). In contrast, HemF2 decarboxylated the two remaining propionate side chains on rings C and D to generate the tetravinyl intermediate 36. Structural comparisons indicated that the active site pocket for substrate binding in HemF2 is smaller and more hydrophilic than that of HemF1, suggesting that HemF2 is a noncanonical protoporphyrinogen IX oxidase that evolved from HemF to catalyze multiple oxidative decarboxylations.
The iron-sulfur cluster-binding enzyme TolI was recently shown to remove all four vinyl groups from 36 to form 37 with unsubstituted pyrrole β sites.52) A plausible mechanism for TolI-catalyzed vinyl group cleavage involves the tautomerization of a vinyl-substituted pyrrole moiety in 36 to the conjugated imine form 3852) (Fig. 7c). The subsequent addition of water to 38 produces the alcohol 39, followed by tautomerization to generate the β-hydroxyl imine 40. Retro-aldol type C–C bond cleavage then yields 41 and acetaldehyde, consistent with the detection of acetaldehyde as a byproduct and the oxygen-independence of TolI catalysis. It has also been speculated that similar vinyl group cleavage reactions occur during the biosynthesis of the related natural products, bonellin (32) from the female green spoonworm Bonellia viridis,53–55) corallistins (such as 33) from the coral sea demosponge Corallistes sp.,56,57) and isabelline A (34) from the Western Australian sponge Isabela sp.58) (Fig. 7a). However, these biosynthetic pathways remain unexplored, and further studies are necessary to test these hypotheses.59)
Understanding enzymatic C–C bond formation and cleavage during natural product biosynthesis is not only integral to unraveling the molecular intricacies of these pathways but also holds immense potential for biotechnological applications. In this review, we have described our recent efforts in the discovery and characterization of enzymes catalyzing pathways related to Ncpa, norlignans, and tolyporphins. Harnessing the enzymatic machinery involved in these processes could offer innovative strategies for the synthesis of bioactive compounds with pharmaceutical, agricultural, or industrial relevance.
I express my sincere gratitude to Prof. Ikuro Abe at the University of Tokyo and Prof. Hung-wen Liu at the University of Texas at Austin for their direction, continuous support, and encouragement. I sincerely thank my co-workers whose names appear in the reference section of this review. The work described in this review was supported by grants from the Ministry of Education, Culture, Sports, Science and Technology, Japan (JSPS KAKENHI Grant Numbers JP20KK0173, JP22H05123, JP23K13847, JP23H02641, and JP23H00393), Kobayashi Foundation, Koyanagi Foundation, Astellas Foundation, Mochida Memorial Foundation, Naito Foundation, Japan Foundation for Applied Enzymology, Amano Enzyme Foundation, Takeda Science Foundation, Sumitomo Foundation, Kishimoto Foundation, and Suzuken Memorial Foundation.
The author declares no conflict of interest.
This review of the author’s work was written by the author upon receiving the 2024 Pharmaceutical Society of Japan Award for Young Scientists.
