Engineering of Taxadiene Synthase for Improved Selectivity and Yield

Oct 30, 2016 - Attempts at microbial production of the chemotherapeutic agent Taxol (paclitaxel) have met with limited success, due largely to a pathw...
1 downloads 13 Views 1MB Size
Download PDF
Letter pubs.acs.org/synthbio

Engineering of Taxadiene Synthase for Improved Selectivity and Yield of a Key Taxol Biosynthetic Intermediate Steven Edgar,† Fu-Shuang Li,§ Kangjian Qiao,† Jing-Ke Weng,*,§,‡ and Gregory Stephanopoulos*,† †

Department of Chemical Engineering and ‡Department of Biology, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139, United States § Whitehead Institute for Biomedical Research, 9 Cambridge Center, Cambridge, Massachusetts 02142, United States S Supporting Information *

ABSTRACT: Attempts at microbial production of the chemotherapeutic agent Taxol (paclitaxel) have met with limited success, due largely to a pathway bottleneck resulting from poor product selectivity of the first hydroxylation step, catalyzed by taxadien-5a-hydroxylase (CYP725A4). Here, we systematically investigate three methodologies, terpene cyclase engineering, P450 engineering, and hydrolase-enzyme screening to overcome this early pathway selectivity bottleneck. We demonstrate that engineering of Taxadiene Synthase, upstream of the promiscuous oxidation step, acts as a practical method for selectivity improvement. Through mutagenesis we achieve a 2.4-fold improvement in yield and selectivity for an alternative cyclization product, taxa-4(20)-11(12)-diene; and for the Taxol precursor taxadien-5α-ol, when coexpressed with CYP725A4. This works lays the foundation for the elucidation, engineering, and improved production of Taxol and early Taxol precursors.

T

tivity for 2 arises from aberrant catalytic cascades or incorrect substrate docking poses within the CYP725A4 active site.9,10 As a result, improvement of selectivity would only be correctable by engineering of the P450 itself. In contrast, in the more recently proposed mechanism, taxadiene is first epoxidized to 3 and then undergoes nonspecific degradation to a range of products, while an isomer of taxadiene, 5, still undergoes the traditional radical rebound in a product-specific manner. This epoxidation-based mechanism for 1 is supported by multiple lines of evidence including in vivo enzyme behavior and similarity of enzymatic products to those obtained using chemical epoxidation.7 Furthermore, chemical analogues of the active site heme moiety, such as FeIII(TPP)Cl, have been shown to catalyze both the epoxidation of 1, as well as its nonselective decomposition to other products.6 Under this mechanism, multiple locations for intervention are feasible, based on several plausible steps of chemical transformation from 4 toward 2 (Scheme 1). These include (A) engineering the cyclization step which generates 1 and 5 from 4, (B) improving CYP725A4 specificity, and (C) controlling the epoxide breakdown pathway via introduction of additional enzymes. We thus investigated each of these three approaches to enable high-level production of early Taxol pathway intermediates.

he chemotherapeutic drug Taxol has been the subject of extensive study1,2 due to its complex structure as well as its medical importance. However, despite immense effort, the complete biosynthetic pathway has not yet been fully elucidated, and heterologous production of even early pathway intermediates has remained elusive.3,4 Previous reports have demonstrated that taxadiene-5α-hydroxylase (CYP725A4), the cytochrome P450 catalyzing the first oxygenation of the taxadiene core (1), produces an array of nonspecific products in addition to the desired product taxadien-5α-ol (2) when expressed in E. coli. Indeed, less than 10% of taxadiene is converted to taxadien-5α-ol (2) by CYP725A4, presenting a significant bottleneck in practical production of early intermediates and downstream Taxol pathway elucidation, as only this product is able to be further transformed into Taxol.3−5 More recently, via both chemical and biological studies, it was demonstrated that this broad product profile is due to nonselective breakdown of an epoxide intermediate (3) rather than nonselective hydroxylation by CYP725A4 itself.3,6,7 We thus set out to leverage chemical and mechanistic information to alleviate this bottleneck by improving the product specificity of the early paclitaxel biosynthetic pathway from the geranylgeranyl pyrophosphate (GGPP) precursor (4) to the desired monohydroxylated diterpene product, 2. In this instance, differentiating between a radical-rebound-based mechanism and an epoxidation-based one for CYP725A4 is essential as it informs engineering efforts. In the originally proposed radical-rebound mediated mechanism,8 poor selec© XXXX American Chemical Society

Received: July 23, 2016 Published: October 30, 2016 A

DOI: 10.1021/acssynbio.6b00206 ACS Synth. Biol. XXXX, XXX, XXX−XXX

Letter

ACS Synthetic Biology Scheme 1. Early Steps of Paclitaxel Biosynthesisa

isomers, taxa-4(5)-11(12)-diene (1) and taxa-4(20)-11(12)diene (5)11 at 92% and 8%, respectively,11 and CYP725A4 has been shown to act differentially upon the two isomers.7 While 1 is transformed into a range of products through an epoxide intermediate, 5 is hydroxylated with high specificity to the desired product, 2, exclusively.7 Thus, by engineering of TXS for increased production of 5, the selectivity of the overall pathway can be improved as the P450 then behaves in a selective manner on the alternative substrate. Our effort to selectively alter the product profile of TXS was facilitated by the well-established TXS catalytic mechanism (Figure 1B).12−14 Close examination of this mechanism suggests that the final step of the catalytic cycle dictates which of the two isomers is generated; abstraction of H5β results in 1, whereas abstraction of H20 gives rise to 2. This aids engineering as fewer perturbations must be made than if it occurred early in catalysis. We next employed computational docking of 1 into the TXS active site. In agreement with previous literature, we obtained two poses which indicate the PPi (in a substrate-assisted catalytic regime) or residue Y688

a

The known native pathway (blue) catalyzed by TXS and CYP725A4 terminates in an epoxide intermediate (3), which then undergoes nonspecific degradation to compounds such as OCT (10). Alternatively, three engineered pathways (black) are proposed originating from catalytic branch points (A, B, and C) in order to improve selectivity for the desired product, 2.

We selected the first catalytic branch point as an initial point of investigation with the goal of engineering the terpene cyclase step, catalyzed by Taxadiene Synthase (TXS). Although it is not intuitive to focus efforts “upstream” of the pathway bottleneck, this approach has many advantages. TXS natively produces two

Figure 1. (A) Tertiary diagram depicting selectivity obtained from saturation mutagenesis of selected residues. Three primary isomers are depicted along each axis, with color scale indicating total titer. Points representing Y688L mutants are denoted with an arrow. (B) Proposed reaction scheme generating taxadiene isomers from geranylgeranyl pyrophosphate (4). Black arrows denote the known, native, pathway; blue arrows indicate a proposed pathway to the desired product, and red arrows indicate a prosed pathway to the novel product, 9. (C) Observed selectivity for taxadieneisomers from alanine scanning and (D) selected mutants. (E) Observed selectivity for hydroxylation products obtained from coexpression of CYP725A4 with select TXS mutants. B

DOI: 10.1021/acssynbio.6b00206 ACS Synth. Biol. XXXX, XXX, XXX−XXX

Letter

ACS Synthetic Biology

was achieved primarily through a reduction in the production of 1 with the titer of compound 5 remaining constant. Despite residue Y688 not appearing responsible for H5β abstraction from alanine-scanning, this residue appears to play an important role in selectivity, possibly due to its proximity to the PPi. The most successful mutant, Y688L, displayed markedly improved selectivity and yield of 5 (2.4-fold) with a concurrent decrease in production of 1. Consistent with the alanine-scanning mutagenesis results, substitution of residue S587 with any other amino acid abolished production of the typically observed isomers (Figure S11). Additionally, mutant Q609G produced 9 as its sole product, yielding the first known terpene cyclase producing this product as its primary product. S587G also yielded a previously undetected product with a titer of less than 0.5 mg/L (Figure S11). Due to low titer and the fact that this was not a desired product, we did not pursue to resolve its structure. Furthermore, mutations in the N-terminal region primarily resulted in reduced enzymatic activity, and were less likely to generate 9 than core active site residues (Figure S5). We next aimed to test whether TXS mutants could be used to improve the production of the desired hydroxylated product, 2, when coexpressed with CYP725A4. With this aim, we transformed four of the mutants obtained above (Y688F, Y688L, Y688Q and Y609G) into a strain containing a chromosomal copy of the upstream MEP pathway and CYP725A4 and its cytochrome P450 reductase partner.7 As seen in Figure 1E, TXS mutants with improved selectivity for 5 displayed an increased production of 2 following coexpression with CYP725A4 due to the high selectivity observed for hydroxylation of 5. The most successful TXS mutant, Y688L yielded a 2.4-fold improvement in titer as predicted. We next evaluated oxidation by CYP725A4 as a target for selectivity improvement although the epoxidation of taxadiene is proposed to be substrate driven,7,17 we noted CYP725A4 natively produces low levels of a dihydroxylated product, proposed to be taxadien-5α-10β-diol.4,10 Thus, we set out instead to expand this dihydroxylating capability of the enzyme though mutagenesis to circumvent the poor specificity of the first oxidation step. Due to the high frequency at which mutations disrupt P450 maturation, we employed a structural and phylogenetically based approach to both conserve enzymatic activity and enhance this native dihydroxylating ability (Figure S12 through S14). Although a majority of the mutants failed to alter product ratios, mutant T380S displayed dramatically altered selectivity and produced a dihydroxylated product as its primary product (Figure 2A, S15). Although the retention time and mass spectra resembled the proposed Taxadien-5α-10β-diol (Figure S16), upon investigation this was found to be a novel product, 5(12)-oxa-3(11)-cyclo-taxan-10-ol (11), which would arise via hydroxylation after epoxide degradation (Figure 2B). This result highlights that although dihydroxylated products are obtainable through this engineering approach, the proposal to exploit this dihydroxylating ability must inherently compete with native epoxide breakdown. Our final point of investigation focused on engineering the breakdown of the epoxide intermediate (3) for increased production of 2. We hypothesized that the epoxide intermediate (3) may undergo a one-step concerted reaction yield 2 (Scheme 1).8 Indeed, reactions of this sort have precedent within isoprenoid biosynthesis, exemplified by limonene epoxide hydrolase.18,19 Moreover, a recent study identified a number of Taxus transcripts with high sequence

may act as the base for abstraction (Figure S1). However, we also obtained a number of alternative poses with similar energetics and predicted poses varied significantly by choice of A or B chain and presence or absence of PPi in the model. This is likely due to the highly hydrophobic nature of taxadiene and large active site volume, making computational docking unreliable and requiring a broader mutagenesis approach for improvement. To determine which, if any, residues were directly responsible for the abstraction of H5β, we performed alaninescanning across the six polar active site residues (S587, Q609, Y684, Y688, C719, and C830), and examined the biochemical function of the resulting mutant enzymes when expressed in a previously engineered E. coli strain with increased GGPP supply.3 As seen in Figure 1C, no mutant displayed abolished production of 1 while retaining production of 5, indicating that no single residue appears to be solely responsible for abstraction of H5β. Alternatively, the PPi may act as the abstracting base, as shown by our docking and proposed previously,12 or a single residue is responsible for the formation of both isomers. Several other results can also be gleaned from this experiment. First, the C719A and C830A mutants display reduced in vivo cyclase activity, but retain the original selectivity (Figure S2), indicating these residues may affect enzyme turnover or stability. Furthermore, despite its distance from the magnesium cluster, the S587A mutant displays no activity and thus still appears to be essential for catalysis and/or overall protein stability. Interestingly, mutants of residues Q609, Y684, and Y688 yielded an unknown product with retention time and mass spectra resembling that of 1 and 5. Following production via fermentation in bioreactors, purification, and characterization, the product was determined to be Verticilla-3,7,12-triene (9). This product has been previously isolated from Bursera suntui plant tissue,15 and its derivatives have also been isolated from various Taxus species,16 although the terpene synthase responsible for its synthesis has not been identified to date. Based on the structure of this compound, we can make a reasonable mechanistic proposal for its production, as shown in Figure 1B. For these mutants, it is likely that the change to a less bulky amino acid perturbs the active site geometry favoring alternative carbocation rearrangement to produce alternative terpene products. To achieve a broader range of active-site perturbations, we performed saturation mutagenesis targeting select active-site residues of TXS, which include the six polar residues as discussed above, as well as a number of additional residues in proximity to these polar residues or the cleaved PPi (S713, V714, G715, W753, V831, Y835). Since the Taxadiene Synthase crystal structure has an exposed active site due to N-terminal truncation; whereas the active form is likely capped by the N-terminus as in other terpene synthases,12 we selected 8 residues from the N-terminal region which we predicted to reside in proximity to the active site cavity (R84 to G91) for mutagenesis as well. Enzymatic activity was then assayed using a whole-cell in vivo system as above. As shown in Figure 1A, the selectivity and activity of the mutant library varied widely, typically consisting of a mixture of the three isomers previously obtained (1, 5, and 9), with a majority of mutants clustered around the original selectivity with reduced activity (Figure S4 to S10). Although a large number of residues improved the apparent selectivity for 5, this C

DOI: 10.1021/acssynbio.6b00206 ACS Synth. Biol. XXXX, XXX, XXX−XXX

Letter

ACS Synthetic Biology

here to reduce the formation of byproducts and drive the carbon flux toward Taxol precursors will enable both heterologous production and pathway elucidation.

■

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acssynbio.6b00206. Experimental procedures, full-length sequences, chemical characterization of products, and supporting figures and schemes. (PDF)

■

AUTHOR INFORMATION

Corresponding Authors

*E-mail: [email protected]. *E-mail: [email protected]. Author Contributions

S.E. and G.S. conceived the project. S.E., K.Q., F.S.L., J.K.W., and G.S. designed and executed the experiments, analyzed the results and wrote the manuscript. Notes

The authors declare no competing financial interest.

■

Figure 2. (A) Selectivity observed from wild-type and T380S mutant CYP725A4, isoOCT7 is a derivative of 8. (B) Proposed pathway from taxadiene to newly observed dihydroxylated product.

ACKNOWLEDGMENTS We would like to acknowledge Dr. Deyou Qiu for providing Taxus transcriptome data. This research is funded in part by grants from the Pew Scholars Program in the Biomedical Sciences and the Searle Scholar Program to J.K.W., by the National Institutes of Health (1-R01-GM084323−01A1) to G.S., and by an NSF GRFP grant supporting S.E.

similarity to known degradative enzymes, making them attractive candidates for engineering purposes.20 We obtained full-length sequences for 23 of these candidate genes by querying the initial partial transcripts against several other available Taxus transcriptomes and by Rapid Amplification of cDNA Ends (RACE).20−22 Expression of each of the candidate genes in an E. coli strain3 previously engineered to produce 2 failed to yield significant alteration in product profiles, indicating that these genes are unlikely to guide epoxide degradation. Enzyme discovery is further hindered as poor stability of epoxide 3 precludes activity-guided fractionation from Taxus protein crude extract, and the necessary rearrangement may occur through simple acid−base catalysis (Figure S17). Furthermore, the broad product range not only occurs in heterologous systems such as in E. coli, but is also observed, at least to a degree, in Taxus, indicating such epoxide degradative enzymes may not exist.7 As a result, this approach was deemed infeasible as a route to overcome the Taxol biosynthetic bottleneck. In summary, we investigated three possible approaches to improve the selectivity and throughput of early Taxol biosynthesis in E. coli. This was done by taking a holistic approach to selectivity improvement, focusing not only on the oxidation catalyzed by CYP725A4, but also upstream and downstream of it. Our results demonstrate TXS, acting upstream of the nonspecific oxidation step, is a key enzyme in directing pathway flux toward Taxol precursors, such as taxadien-5α-ol (2). We further showed that engineering of TXS is capable of improving the production of 2 through an unexpected metabolic route, whereas engineering of CYP725A4 and screening of hydrolase genes were ineffective at decreasing the production of undesirable byproducts. As only single mutants were generated here, further engineering is likely to enable near complete shifts in selectivity for the desired product. Furthermore, application of the approaches utilized

■

REFERENCES

(1) Nicolaou, K. C., Yang, Z., Liu, J. J., Ueno, H., Nantermet, P. G., Guy, R. K., Claiborne, C. F., Renaud, J., Couladouros, E. a, and Paulvannan, K. (1994) Total synthesis of taxol. Nature 367, 630−634. (2) Wilde, N. C., Isomura, M., Mendoza, A., and Baran, P. S. (2014) Two-Phase Synthesis of (−)-Taxuyunnanine D. J. Am. Chem. Soc. 136, 4909. (3) Ajikumar, P. K., Xiao, W.-H., Tyo, K. E. J., Wang, Y., Simeon, F., Leonard, E., Mucha, O., Phon, T. H., Pfeifer, B., and Stephanopoulos, G. (2010) Isoprenoid pathway optimization for Taxol precursor overproduction in Escherichia coli. Science 330, 70−74. (4) Zhou, K., Qiao, K., Edgar, S., and Stephanopoulos, G. (2015) Distributing a metabolic pathway among a microbial consortium enhances production of natural products. Nat. Biotechnol. 33, 377− 383. (5) Rontein, D., Onillon, S., Herbette, G., Lesot, A., Werck-Reichhart, D., Sallaud, C., and Tissier, A. (2008) CYP725A4 from yew catalyzes complex structural rearrangement of taxa-4(5),11(12)-diene into the cyclic ether 5(12)-oxa-3(11)-cyclotaxane. J. Biol. Chem. 283, 6067−75. (6) Barton, N. a, Marsh, B. J., Lewis, W., Narraidoo, N., Seymour, G., Fray, R., and Hayes, C. J. (2016) Accessing low-oxidation state taxanes: Is taxadiene-4(5)-epoxide on the taxol biosynthetic pathway? Chem. Sci. 00, 1−6. (7) Edgar, S., Zhou, K., Qiao, K., King, J. R., Simpson, J. H., and Stephanopoulos, G. (2016) Mechanistic Insights into Taxadiene Epoxidation by Taxadiene-5α-Hydroxylase. ACS Chem. Biol. 11, 460. (8) Jennewein, S., Long, R. M., Williams, R. M., and Croteau, R. (2004) Cytochrome P450 taxadiene 5??-hydroxylase, a mechanistically unusual monooxygenase catalyzing the first oxygenation step of taxol biosynthesis. Chem. Biol. 11, 379−387. (9) Yadav, V. G. (2014) Journal of Molecular Catalysis B. J. Mol. Catal. B: Enzym. 110, 154−164.

D

DOI: 10.1021/acssynbio.6b00206 ACS Synth. Biol. XXXX, XXX, XXX−XXX

Letter

ACS Synthetic Biology (10) Biggs, B. W., Rouck, J. E., Kambalyal, A., Arnold, W., Lim, C. G., Mey, M., De Neil-johnson, M. O., Starks, C. M., Das, A., and Ajikumar, P. K. (2016) Orthogonal Assays Clarify the Oxidative Biochemistry of Taxol P450 CYP725A4. ACS Chem. Biol. 11, 1445. (11) Koepp, A., Hezari, M., and Zajicek, J. (1995) Cyclization of geranylgeranyl diphosphate to taxa-4 (5), 11 (12)-diene is the committed step of taxol biosynthesis in Pacific yew. J. Biol. Chem. 270, 8686. (12) Köksal, M., Jin, Y., Coates, R. M., Croteau, R., and Christianson, D. W. (2011) Taxadiene synthase structure and evolution of modular architecture in terpene biosynthesis. Nature 469, 116−20. (13) Lin, X., Hezari, M., Koepp, a. E., Floss, H. G., and Croteau, R. (1996) Mechanism of taxadiene synthase, a diterpene cyclase that catalyzes the first step of taxol biosynthesis in Pacific yew. Biochemistry 35, 2968−77. (14) Jin, Q., Williams, D. C., Hezari, M., Croteau, R., and Coates, R. M. (2005) Stereochemistry of the macrocyclization and elimination steps in taxadiene biosynthesis through deuterium labeling. J. Org. Chem. 70, 4667−4675. (15) Hernandez-Hernandez, J., and Roman-Marin, L. U. (2005) Verticillane Derivatives from Bursera suntui and Busera kerberi. J. Nat. Prod. 68, 1598−1602. (16) Itokawa, H. (2003) Taxoids occurring in the genus Taxus. In Taxus: The Genus Taxus, CRC Press, Boca Raton, FL. (17) Kumar, D., Karamzadeh, B., Sastry, G. N., and De Visser, S. P. (2010) What factors influence the rate constant of substrate epoxidation by compound i of cytochrome P450 and analogous iron(IV)-oxo oxidants? J. Am. Chem. Soc. 132, 7656−7667. (18) Hopmann, K. H., Hallberg, B. M., and Himo, F. (2005) Catalytic mechanism of limonene epoxide hydrolase, a theoretical study. J. Am. Chem. Soc. 127, 14339−14347. (19) Brodkorb, D., Gottschall, M., Marmulla, R., Lüddeke, F., and Harder, J. (2010) Linalool dehydratase-isomerase, a bifunctional enzyme in the anaerobic degradation of monoterpenes. J. Biol. Chem. 285, 30436−30442. (20) Sun, G., Yang, Y., Xie, F., Wen, J.-F., Wu, J., Wilson, I. W., Tang, Q., Liu, H., and Qiu, D. (2013) Deep sequencing reveals transcriptome re-programming of Taxus × media cells to the elicitation with methyl jasmonate. PLoS One 8, e62865. (21) Hao, D. C., Ge, G., Xiao, P., Zhang, Y., and Yang, L. (2011) The first insight into the tissue specific taxus transcriptome via Illumina second generation sequencing. PLoS One 6, e21220. (22) Li, S., Zhang, P., Zhang, M., Fu, C., Zhao, C., Dong, Y., Guo, A., and Yu, L. (2012) Transcriptional profile of Taxus chinensis cells in response to methyl jasmonate. BMC Genomics 13, 295.

E

DOI: 10.1021/acssynbio.6b00206 ACS Synth. Biol. XXXX, XXX, XXX−XXX