今天推送的文章发表在ACS Catalysis的“Functional Application of the Single-Module NRPS-like d-Alanyltransferase in Maytansinol Biosynthesis”，通讯作者为山东大学的朱德裕副教授和沈月毛教授。
In the biosynthesis of polyketone natural products, certain post-PKS modifications greatly enhance the bioactivity potential, but present unique challenges in the synthesis of target compounds. Maytansinol is a direct precursor of maytansine-derived antibody-drug conjugates and a probe to study microtubule dynamics. The current process of producing Maytansinol mainly relies on reductive hydrolysis to remove the 3-O-acyl group of ansamitocins, and the reaction conditions are harsh and by-products are inevitably formed. Non-ribosomal peptide synthetase (NRPS) serves as a multi-module assembly line to catalyze the biosynthesis of natural products of various peptides with potent biological activity. AstC, a unique A-T-TE three-domain NRPS-like enzyme found in Streptomyces spp. (Figure 1A), is an O-d-alanyltransferase responsible for the specific loading of d-alanyl groups onto the 11-OH of ansatrinein. However, in the function of canonical NRPS, the incorporation of d-configuration amino acids rarely occurs through direct activation of the a-domain, but instead relies on E-domain-mediated epimerization.
In this study, the authors elucidated the function of the TE domain of AstC in intermolecular esterification and revealed its substrate promiscancy for acyl donors and polyketone acceptors. Through genome mining, a novel AstC homolog, SmAstC, was identified and showed that SmAstC exhibited the highest catalytic activity for the d-alanylylation of the ansamitocins precursor DDM obtained from actinomycete HGF052. Taking advantage of the broad substrate selectivity of the TE domain and the propensity for spontaneous hydrolysis of the d-alanylation products, the PKS post-modification of ansamitocins biosynthesis was reprogrammed by introducing the gene encoding SmAstC into HGF052, so that the ansamitocins biosynthetic pathway in this engineered strain was shifted to the production of maytansinol.
1
The TE domain of AstC catalyzes intermolecular esterification reactions
The role of AstC in the transfer of d-alanine to 11-OH in ansatriens has previously been revealed. However, the precise function of its domains remains unclear, especially the role of the TE domain in catalyzing intermolecular esterification. In the present study, the authors purified the TE domain of AstC (AstC-TE) and tested its activity using d-alanyl N-acetylcysteine thioelate (d-alanyl SNAC) as an acyl donor. Incubation of AstC-TE with d-alanyl SNAC and trienylmethylmethanol (1) successfully formed 11-O-d-alanyltrienylmethylol (2), consistent with the enzymatic assay performed with full-length AstC (Figure 1B).
Phylogenetic analysis highlighted AstC-TE as a unique catalyst for catalyzing intermolecular esterification between aminoacyl donors and polyketone backbones (Figure 1C). Two other TEs within the same branch as AstC (Sln9 TE and FrsA TE) are derived from comparable single-module NRPS with a C-A-T-TE structure responsible for modifying the intermolecular esterification of amino acids and peptides. The sequence alignment of AstC-TE with representative type I TE shows a conserved GXSXG motif but differs in the composition of the catalytic triplet. In the TE domain of NRPS, the classical Ser-Asp-His triplet is usually conserved, but the AstC-TE is composed of Ser-Asn-His catalytic triplets. To assess the effect of this substitution on catalytic activity, the authors mutated the N117 residue of AstC-TE. The N117D mutant of AstC-TE exhibited a 5-fold decrease in activity, while the N117A mutant completely eliminated activity (Figure 1D). This unique structural feature observed in AstC-TE distinguishes it from other TEs and helps in the identification of related cognate enzymes.
2
Substrate range for the AstC-TE domain
Since AstC-TE has been shown to independently catalyze intermolecular esterification between acyl donors and acceptors, the authors set out to explore the substrate range of AstC-TE for ansamycins with different acyl SNACs and with different backbones. Previously, some of the incorporation of d-amino acids was achieved with full-length AstC, but the incorporation of l-amino acids could not be achieved. The authors chemically synthesized several l-aminoacyl SNACs and used them as substrates for the enzymatic assay of AstC-TE. The results showed that the l-alanine, l-methoxy, l-leucine, and l-phenylalanine moieties, which were not previously utilized by full-length proteins, could be completed by independent AstCTE (Figure 2A). AstC-TE does exhibit broad compatibility with different acyl donors, and the strict preference of full-length AstC for d-amino acids with small side chains will be attributed to its A-domain.
Based on suboptimal acyl donors provided in the form of SNACs, full-length AstC was subsequently utilized to screen for compatibility with a range of ansamycins as acyl acceptors. Detection of d-alanylation products by ESI-HRMS analysis showed that AstC was able to d-alanylation of two different types of pentaketo-ansamycin and two octa-keto-ansamicin, exhibiting a high degree of promiscyanity for acyl acceptors (Figure 2B).
3
The discovery of homologous enzymes and the structural basis of their specific D-alaination
Observed in the SSN network, AstC is not a segregated entity, but rather belongs to a cluster of homologous enzymes, including AnsC and MycC, which have previously been identified in other ansatreinin biosynthetic pathways, but remain uncharacterized. The authors constructed a phylogenetic tree of these AstC homologs using maximum likelihood. Six homologs from different clades, namely SyAstC, SmAstC, AnsC, MycC, SlAstC, and KbAstC (Figure 2C), were selected as native mutants to identify potential candidates with enhanced activity against DDM. The results showed that these enzymes exhibited a strong preference for D-alanine as an acyl donor (Figure S9A).
In the assembly of NRPS, the incorporation of d-amino acids is often dependent on the epimeric domain. However, the deletion of the E domain in the A-T-TE tridomain d-alanine transferase suggests that the A domain directly activates d-alanine. In order to elucidate the configuration preference, the authors performed a crystallization screen of the A domain of these six d-alyltransferases, and successfully obtained the complex eutectic structure of the A region of SyAstC and the mimetic of d-alaminodenylic acid intermediate, d-Ala-SA (Fig. 2D). The cavities formed by the residues F211, D212, G304, V310 and W311 are postulated as binding sites for the d-alanyl moiety. The d-alanyl amino group forms hydrogen bonds with the carboxyl group of D212, the carbonyl group of G304 and the backbone of V310, respectively. This arrangement is in perfect agreement with that observed in the d-alanyl carrier protein ligase DltA (PDB ID: 3DHV).
The C269 residue in DltA, which is considered to be a key determinant of the preference for d-enantiomers, is accordingly replaced by A280 in the SyAstCa domain, while the side chain of W311 occupies a similar space in DltA to C269, with its indole group located at the position of α-carbon 3.4Å from the d-alanyl moietyre. This suggests that W311 dominates enantiomer preference, which is supported by the increased activity against l-alanine observed in the W311A mutant.
4
SmAstC催化DDM的3-O-d亚烷基化反应
Using the optimal substrate d-alanine as an acyl donor, six selected d-alanineyltransferases (SyAstC, SmAstC, AnsC, MycC, SlAstC, and KbAstC) showed significant differences in their ability to catalyze the conversion of DDM(3) to 3-O-d-alanine-based DDM(4). Among them, SmAstC has the highest catalytic activity for 3, which is increased by 8-fold compared to AstC (Figure 3A).
SmAstC is a homolog with 89.6% sequence identity with AstC, and detailed substrate-specific kinetic analysis shows that SmAstC exhibits a clear preference for d-alanine as its acyl donor and significantly lower catalytic efficiency for d-serine, glycine, and l-alanine. In addition, this observation suggests that the accumulation of D-alarosinated product 4 does not persist over time, but rather decreases gradually. A series of stability assessments across the pH range revealed a tendency to spontaneous hydrolysis4, significantly accelerating under alkaline conditions (Figure 3C). This spontaneous hydrolysis is preliminarily thought to occur through the participation of adjacent groups of the D-alanine amino group.
Kinetic parameters of SmAstC were measured at the initial velocity phase under optimized enzymatic pH and temperature conditions, showing a conversion of 8.8±0.9 μM for d-alanine and 11.9±0.5 min-1 for kcat (Figure 3D). The apparent Km and kcat values of 3 were 0.24±0.03 mM, and 0.28±0.02 min-1, respectively, while the apparent Km and kcat values of 1 were 44±5 μM and 24.0±0.8 min-1, respectively (Figure 3D). These characterizations reveal significant changes in substrate affinity and catalytic efficiency of SmAstC for both natural and non-native substrates.
5
合成Maytansinol
To assess the activity of SmAstC against DDM(3) in vivo, the authors integrated the SmAstC gene on the chromosome of the HGF052 strain, and the resulting strain HGFS01 showed significant changes in the metabolite profile, significantly accumulating maytansinol (5) (Fig. 4A). Disruption of the 3-O-acyltransferase gene asm19 in HGF052 hinders the effective catalysis of DDM by downstream modifier enzymes, so only trace amounts of maytansinol are detected. In the HGFS01 strain, the production of maytansinol is attributed to the 3-O-alanylation of DDM catalyzed by SmAstC, and the subsequent 4,5-epoxidation and N-methylation reactions catalyzed by Asm11 and Asm10, respectively, as evidenced by the non-detection of 3-O-propaneminol DDM (4).
Analysis of the EIC peaks of the fermentation products of the HGFS01 strain revealed the presence of different signals corresponding to d-alanylated products, such as N-desmethyl-3-O-alyl-maytansinol (6) and 3-O-alanyl-maytansinol (7), as well as their corresponding hydrolysates, N-desmethyl-maytansinol (8) and maytansinol 5 (Figure 4A). These results confirm the role of SmAstC in vivo and the biosynthesis process of Maytansinol (Figure 4B). SmAstC-catalyzed 3-O-d-alanylation of DDM restores the disruption of the ansamitocin post-PKS modification step caused by the knockout of the 3-O-acyltransferase gene asm19, allowing downstream 4,5-epoxidation and amide N-methylation. Spontaneous hydrolysis of 3-O-d-alanyl maytansinol leads to the production of maytansinol as the main metabolite.
6
Rational design of the SmAstC-TE domain
Previous studies have shown that overexpression of the pathway-specific positive transcriptional regulator gene ASM18 effectively increases DDM production (3). However, in the asm18-overexpressing mutant HGFS04, the overexpression of the SmastC gene failed to significantly enhance the production of Maytansinol. This suggests that high levels of DDM are not efficiently converted to Maytansinol, and that SmAstC-catalyzed d-alanylation of DDM is a rate-limiting step in this biosynthesis process. Therefore, the authors improved the catalytic efficiency of SmAstC for DDM by focusing on the SmAstC-TE domain (SmAstC-TE) through a rational design approach (Fig. 5A). Due to the difficulty of obtaining protein crystals in the SmAstC-TE domain or cocrystals with their substrates, the authors used the structural model predicted by AlphaFold2 to make a reasonable design. The structural alignment of the SmAstC-TE and holo-EntF-T-TE domains elucidated the arrangement of acyl donor channels in SmAstC-TE and the corresponding binding pockets of acyl acceptors. A covalent docking simulation was then performed using a Cov_Dox server using the predicted catalytic intermediate and SmAstC TE to simulate the transition state of d-alanylation (Figure 5B).
To assess the effect of substrate binding to DDM-related residues within the pocket, alanine scanning mutagenesis was performed on residues within ligand 4Å (Figure S21). W89 and F91 mutate to alanine in the conserved GXSXG motif, resulting in a sharp decrease in enzyme activity. Residues L26, N119, I140, V212, and E239 were identified as hotspot residues for multi-point mutagenesis design using the FuncLib server. The results showed that the M2 mutant exhibited the highest activity, with a 41% increase in activity compared to the wild-type enzyme (Figure 5C). In addition, residues K125, L136, and F190 located on the outer surface of the pocket were selected for mutagenesis to evaluate their potential effects on the catalytic efficiency of DDM. The K125A mutant showed a significant increase in enzymatic activity, while the L136A mutant had a slight effect on activity, and the W190A mutation completely eliminated the formation of the product (Figure 5D). To further investigate the effects of residue K125, site-specific mutagenesis was performed with focused and rational iteration, followed by an assessment of the activity changes exhibited by the resulting mutants. The decrease in activity observed when K125 mutates to a bulky residue suggests that the spatial influence exerted by K125 appears to be a major factor influencing the observed change in activity. The K125G mutation of the M2 mutant (SmAstCM) ultimately results in a 3.2-fold increase in in vitro yield. The kcat/Km value of SmAstCM is 2.8 times higher than that of SmAstC, indicating an improved substrate specificity for DDM (Figure 5E).
To probe the in vivo catalytic activity of SmAstCM mutants, the authors constructed HGFS06 by overexpressing asm18 and SmAstCM genes. It was found that the yield of Maytansinol in HGFS06 was 2.4 times higher than that in HGFS05 (Fig. 6A), demonstrating the effectiveness of rational design to improve the catalytic efficiency of DDM by improving SmAstC. The accumulation of 8 indicates a potential limitation of the subsequent modification steps on the overall transformation of DDM to Maytansinol (Figure 6A). Further overexpression of asm11 and asm10 genes contributed to the conversion of 8 to Maytansinol, but failed to significantly increase the yield of Maytansinol, possibly due to affecting the overall metabolic status of this strain. A recently developed strategy to mimic the DEBS enzyme assembly line, which enables the orderly assembly of key cascades and improves cocatalytic efficiency, has been used to assemble SmAstCM, Asm11, and Asm10 with the aim of reducing the accumulation of intermediate metabolites and accelerating the conversion to metachlorinol (Figure 6B). The resulting HGFS08 strain significantly increased the yield of Maytansinol by solid-state fermentation (Figure 6C). To verify the effectiveness of the engineered strains for scale-up fermentation, a 3L fermentation was performed using a laboratory-scale bioreactor to evaluate the performance of HGFS08. Maytansinol was obtained after 10 days of fermentation with a yield of 25±2 mg/L.