Rapid Report pubs.acs.org/biochemistry
Alteration of Polyketide Stereochemistry from anti to syn by a Ketoreductase Domain Exchange in a Type I Modular Polyketide Synthase Subunit Clara H. Eng,†,§ Satoshi Yuzawa,∥ George Wang,⊥,# Edward E. K. Baidoo,⊥,# Leonard Katz,§,∥ and Jay D. Keasling*,†,§,∥,⊥,#,‡ †
Department of Chemical and Biomolecular Engineering, ‡Department of Bioengineering, and ∥QB3 Institute, University of California, Berkeley, California 94270, United States § Synthetic Biology Engineering Research Center, 5885 Hollis Street, Emeryville, California 94608, United States ⊥ Joint BioEnergy Institute, 5885 Hollis Street, Emeryville, California 94608, United States # Biological Systems and Engineering Division, Lawrence Berkeley National Laboratory, Berkeley, California 94270, United States S Supporting Information *
(TE) domain, which typically hydrolyzes or cyclizes the nascent acyl donor chain to form an acid or a lactone, respectively. Products released from a PKS may undergo subsequent glycosylation, hydroxylation, methylation, or other modifications to form the final active compound.2 AT domains exclusively utilize α-substituted substrates with (S) stereochemistry.3,4 This α-substituent undergoes inversion of configuration upon KS-catalyzed condensation of an extender unit with the nascent acyl donor chain.3,5 When present, the KR conducts NADPH-dependent reduction of the β-ketone, exercising stereoselective control over β-hydroxyl stereochemistry mediated by different binding modes that control the face of keto group attack by the 4-pro-S hydride of the NADPH cofactor.6,7 The KR may also perform additional epimerization of the α-substituent,8−10 with stereospecific reduction conducted on only one epimer. Naming conventions have been established to describe β-hydroxyl stereochemical outcome (A or B) and α-substituent stereochemical outcome (1 nonepimerized, 2 epimerized),6,7,11 as illustrated in Figure 1. Earlier studies have demonstrated the successful substitution of both A1- and B1-type heterologous KR domains for A1-type Ery KR2 in bimodular and trimodular derivatives of the 6deoxyerythronolide B synthase (DEBS) with the TE from module 6 appended.12−14 In select cases, these KR domain exchanges inverted β-hydroxyl stereochemistry with no loss of activity. More recently, Weissman and co-workers replaced Ery KR2 with various A2-type KR domains in a bimodular derivative of DEBS, successfully altering α-methyl stereochemistry, though at the expense of greatly decreased titers of products.10 In this study, we use an alternative model system harboring an A2-type KR to alter stereochemical outcome from anti (A2/ B1) to syn (A1/B2), which to the best of our knowledge has not been previously achieved by any method, including directed mutagenesis15−18 and KR domain exchanges.10,12−14
ABSTRACT: Polyketide natural products have broad applications in medicine. Exploiting the modular nature of polyketide synthases to alter stereospecificity is an attractive strategy for obtaining natural product analogues with altered pharmaceutical properties. We demonstrate that by retaining a dimerization element present in LipPks1+TE, we are able to use a ketoreductase domain exchange to alter α-methyl group stereochemistry with unprecedented retention of activity and simultaneously achieve a novel alteration of polyketide product stereochemistry from anti to syn. The substrate promiscuity of LipPks1+TE further provided a unique opportunity to investigate the substrate dependence of ketoreductase activity in a polyketide synthase module context.
T
ype I modular polyketide synthases (PKSs) are responsible for the production of diverse polyketide natural products, including those that function as antibacterial, antiparasitic, and antitumor agents. The medicinal value of polyketides has fueled an interest in exploring the chemical space of natural product analogues. While chemical modification of existing polyketides has enabled the production of more active derivatives,1 the structural and stereochemical complexity of these natural products remain a substantial barrier to employing semisynthetic methods. The stringent chemo-, regio-, and stereospecificity of enzyme catalysts make protein engineering a promising alternative. Type I PKS modules are grouped into protein subunits that together constitute a single PKS. Each module is responsible for the addition of two carbon units to a growing polyketide backbone and is minimally comprised of a set of acyltransferase (AT), ketosynthase (KS), and acyl carrier protein (ACP) domains. Together, these three catalytic domains effect a single round of decarboxylative condensation between an acyl starter or intermediate and a dicarboxylic acid extender unit. The resulting β-ketone may be further reduced by optional ketoreductase (KR), dehydratase (DH), and enoyl reductase (ER) domains. Chain termination is catalyzed by a thioesterase © XXXX American Chemical Society
Received: February 12, 2016 Revised: March 3, 2016
A
DOI: 10.1021/acs.biochem.6b00129 Biochemistry XXXX, XXX, XXX−XXX
Biochemistry
Rapid Report
An expression vector for a variant lacking a DE, (−DE)LipPks1+TE, was also constructed. Escherichia coli strain K207-3,21 which has a substrate promiscuous surfactin phosphopantetheinyl transferase from Bacillus subtilis integrated into the genome, was used as the expression host for all LipPks1+TE variants. With the exception of (−DE)LipPks1+TE, all initial constructs could be well purified using Ni affinity chromatography followed by anion exchange chromatography. Each preparation resulted in 0.5−2 mg/L purified protein. While (−DE)LipPks1+TE could not be purified, sodium dodecyl sulfate−polyacrylamide gel electrophoresis (SDS−PAGE) immediately following cell lysis revealed that LipPks1+TE and (−DE)LipPks1+TE were expressed similarly, with half of the total protein recovered in the insoluble fraction, as shown in Supplementary Figure 2. This suggests that the DE may be important for protein stability, rendering the variant lacking the DE more susceptible to proteolysis outside of the cellular environment. To evaluate reductive ability, each LipPks1+TE variant was incubated overnight with isobutyryl-CoA (1a), methylmalonylCoA, and NADPH. The subsequent production of 3-hydroxy2,4-dimethylpentanoic acid (3a) was quantified using liquid chromatography and mass spectroscopy (LC−MS) with an authentic standard.19 To distinguish between variants that were not competent in condensation and those that were competent in condensation but incapable of reduction, each LipPks1+TE variant was also incubated overnight with 1a and methylmalonyl-CoA in the absence of NADPH to form a diketide product independent of the reductive ability of the KR. The diketide was then heated overnight to form a more stable ketone, 2-methylpentan-3-one (2a), with titers of products again determined by LC−MS with an authentic standard.22 (Amp KR2)LipPks1+TE produced 2a and 3a with activity comparable to that of wild-type LipPks1+TE, as shown in Table 1. This retention of activity represents a substantial improvement over previous work in which α-methyl stereochemistry has been altered by a KR domain exchange.10 By contrast, (Amp DE,KR2)LipPks1+TE exhibited greatly attenuated activity in both condensation and reduction assays. Neither Con KR2 variant yielded observable 3a, and only (Con KR2)LipPks1+TE was condensation competent. Keatinge-Clay and co-workers previously demonstrated that retention of the DE associated with the donor KR was required for activity after substitution into DEBS module 2, which does
Figure 1. β-Hydroxyl and α-methyl stereochemistry in a nascent acyl donor chain are determined by the KR domain. Optional epimerization is followed by reduction of the β-keto group.
Our chosen model system, LipPks1+TE,19 is a N-terminally hexahistidine-tagged recombinant polyketide synthase subunit that consists of the first module of the lipomycin synthase from Streptomyces aureofaciens Tü117 with the TE from DEBS appended to the C-terminus. As in DEBS module 2, the reductive cassette in LipPks1 has only a KR domain. However, unlike DEBS module 2, LipPks1 harbors an A2-type rather than A1-type KR and also contains a recently identified ∼55-amino acid dimerization element (DE) that lies between the AT and KR in approximately half of the PKS modules20 that contain a single reducing catalytic domain. While previous KR domain exchange experiments have treated the DE as part of the KR domain,10,14 we suspected that the DE may stabilize protein structure through interactions with the remainder of the PKS module and sought to probe the importance of the DE biochemically. As an initial test of the importance of the DE, we constructed expression vectors corresponding to variants in which the native Lip KR1 was substituted with one of two donor KR domains: an A1-type KR from the amphotericin pathway (Amp KR2) or a B1-type KR from the concanamycin pathway (Con KR2). For each of these donor KR domains, two variants were constructed differing only in the DE, which was taken either from the parent PKS module, LipPks1, resulting in (Amp KR2)LipPks1+TE and (Con KR2)LipPks1+TE, or from the PKS module that furnished the donor KR domain, yielding (Amp DE,KR2)LipPks1+TE and (Con DE,KR2)LipPks1+TE.
Table 1. Concentrations of Ketones 2a−c and Hydroxyacids 3a−c Produced by LipPks1+TE Variants after Overnight Incubation with Substrates 1a−c in Micromoles per Litera 1a 2a Lip KR1 Amp KR2 Amp KR2b Con KR2 Con KR2b Ery KR1 Bor KR1 Spn KR3 Amp KR1 Ery KR6
A2 A1 A1 B1 B1 B2 B A1 A2 A1
42.2 35.0 8.0 8.7 nd 27.6 36.5 33.5 35.2 25.2
1b 3a
± ± ± ±
6.6 2.5 1.2 2.2
± ± ± ± ±
6.2 3.8 2.7 1.3 3.4
32.6 30.2 3.6 nd nd nd nd 11.4 12.8 9.2
2b
± 1.4 ± 3.5 ± 1.9
49.9 21.8 nd 4.6 nd 25.5 16.6 47.3 45.8 8.6
± 0.3 ± 0.2 ± 0.1
3b
± 7.2 ± 1.5 ± 0.4 ± ± ± ± ±
1c
3.8 2.3 6.9 3.9 2.0
13.6 9.8 1.8 nd nd nd nd 13.6 17.1 2.2
± 0.6 ± 1.2 ± 0.7
± 0.6 ± 2.7 ± 1.0
2c 46.5 33.4 6.7 11.5 nd 30.9 27.9 52.5 57.1 15.0
3c
± ± ± ±
1.5 3.8 1.1 0.2
± ± ± ± ±
1.3 2.9 3.9 3.8 2.0
24.5 24.1 3.9 nd nd nd nd 29.0 10.1 7.2
± 3.3 ± 1.5 ± 1.9
± 1.2 ± 0.6 ± 1.9
a Means and standard errors of experiments performed in triplicate are reported. nd indicates that no product was detected. bDE from the module furnishing the donor KR domain substituted for that of LipPks1.
B
DOI: 10.1021/acs.biochem.6b00129 Biochemistry XXXX, XXX, XXX−XXX
Biochemistry
Rapid Report
not natively harbor a DE.15 This suggested that retaining the DE from the module furnishing the donor KR is important in cases in which the parent PKS module lacks a DE. Our findings further imply that retaining the parent DE may be preferable in cases in which a DE is already present. To determine whether the strategy of retaining the parent DE can be applied to domain exchanges employing other donor KR domains, we cloned additional A1-type (Ery KR6 and Spn KR3 from the spinosyn pathway), A2-type (Amp KR1), B-type (Bor KR1 from the borreledin pathway), B1-type (Ave KR1 from the avermectin pathway), and B2-type (Ery KR1 and Pik KR1 from the pikromycin pathway) KR domains into LipPks1+TE, in each case keeping the parent DE intact. Amp KR1, Ave KR1, and Spn KR3 were taken from modules that naturally contain DEs, while Bor KR1, Ery KR1, Ery KR6, and Pik KR1 were taken from modules that lack DEs. Aside from variants utilizing Ave KR1 or Pik KR1, all constructs were readily purified. Intriguingly, with the sole exception of (Con KR2)LipPks1+TE, all LipPks1+TE variants in which the parent DE was kept intact exhibited condensation activity that was largely retained, no matter which donor KR was used. While the reason for the attenuated condensation activity exhibited by (Con KR2)LipPks1+TE remains unclear, Con KR2 is unusual in that it is a B1-type KR that contains a DE; most of the PKS modules that contain a DE have A-type rather than B-type KR domains. We further observe that all of the A-type domain exchange variants were capable of achieving at least moderate levels of reduction, though none of the B-type domain exchanges were successful. The robust success of our A1-type domain exchanges is of particular interest in light of recent work hypothesizing that anti products are thermodynamically favored.16
studies that have investigated how the structure of ACPtethered substrates impacts KR activity within a PKS module. Comparison of product concentrations resulting from incubation with different starter substrates revealed that the KR activity exhibited by each variant is substrate-dependent. While Amp KR2 was the most active variant when 1a was used as the starter substrate, when the starter substrate was changed to 1b or 1c, the most active variants utilized Amp KR1 or Spn KR3 and Amp KR2 or Spn KR3, respectively. Although it was not possible to identify clear, broadly applicable trends, some loose correlations were drawn between the native substrate structure, shown in Supplementary Table 1, and the outcome of reduction. Amp KR1, which natively has a very small substrate, was most active when the smallest starter unit, 1b, was used, while Spn KR3, which has a longer, less substituted native substrate, favored the starter unit 1c. Interestingly, the condensation ability of each variant relative to that of the wild type is also substrate-dependent, as shown in Supplementary Figure 23. This suggests that KR substitution alters the structure of the PKS module such that condensation is more favorable with specific substrates. We also sought to consider the impact of KR domain exchanges on stereochemical fidelity. Gas chromatography− mass spectrometry (GC−MS) was used to evaluate the stereochemical outcome of reduction catalyzed by the A-type LipPks1+TE variants when 1a was used as the starter substrate. Stereochemical assignments were made by comparison to authentic standards, which exhibited an elution order consistent with previous work.24 The results are summarized in Table 2. Table 2. Stereochemical Outcome of Ketoreduction Catalyzed by LipPks1+TE Variantsa LC−MS 1a syn Lip KR1 Amp KR2 Amp KR2b Spn KR3 Amp KR1 Ery KR6
A2 A1 A1 A1 A2 A1
anti
syn
● ● ● ●
1c anti
syn
● ● ● ●
● ●
GC−MS
1b
anti
syn (A1)
● ● ● ●
● ●
1a
● ● ● ●
● ●
anti (A2)
●
● ●
a
Observed stereoisomers for each combination of enzyme and starter substrate are marked with a bullet. The correspondence between KR type and stereochemical outcome is as follows. A1, syn (2R,3S); A2, anti (2S,3S); B1, anti (2R,3R); B2, syn (2S,3R). bDE from the module furnishing the donor KR domain substituted for that of LipPks1.
GC−MS analysis demonstrated that almost all of the A-type LipPks1+TE variants exclusively produced the expected stereoisomer. Only A1-type (Ery KR6)LipPks1+TE, which produced A1 and A2 products in an approximate 9:2 ratio, generated any unexpected stereoisomers. One explanation is that because Ery KR6 has a larger native substrate, the consequent larger active site may relax the stereospecificity of reduction. While we were unable to assign exact stereochemistry for products resulting from incubation with starter substrates 1b and 1c, different LC retention times were observed for syn and anti products in both cases. This was verified by comparing the retention time of the wild-type Lip KR1 product, which was expected to be anti,25,26 to that of the authentic synthetic
Figure 2. Products generated by LipPks1+TE variants in this study.
Furthermore, LipPks1+TE was previously demonstrated to accept a set of diverse starter substrates that includes propionylCoA (1b) and butyryl-CoA (1c),19 in addition to the predicted starter substrate, 1a. 23 The substrate promiscuity of LipPks1+TE provided a unique opportunity to probe the effect of substrate structure on the activity of KR domain exchange variants. The results of condensation and reduction activity assays are listed in Table 1. We are aware of no other C
DOI: 10.1021/acs.biochem.6b00129 Biochemistry XXXX, XXX, XXX−XXX
Biochemistry
Rapid Report
(5) Weissman, K. J., Timoney, M., Bycroft, M., Grice, P., Hanefeld, U., Staunton, J., and Leadlay, P. F. (1997) Biochemistry 36, 13849− 13855. (6) Reid, R., Piagentini, M., Rodriguez, E., Ashley, G., Viswanathan, N., Carney, J., Santi, D. V., Hutchinson, C. R., and McDaniel, R. (2003) Biochemistry 42, 72−79. (7) Caffrey, P. (2003) ChemBioChem 4, 654−657. (8) Garg, A., Khosla, C., and Cane, D. E. (2013) J. Am. Chem. Soc. 135, 16324−16327. (9) Garg, A., Xie, X., Keatinge-Clay, A., Khosla, C., and Cane, D. E. (2014) J. Am. Chem. Soc. 136, 10190−10193. (10) Annaval, T., Paris, C., Leadlay, P. F., Jacob, C., and Weissman, K. J. (2015) ChemBioChem 16, 1357−1364. (11) Keatinge-Clay, A. T. (2007) Chem. Biol. 14, 898−908. (12) Kao, C. M., McPherson, M., McDaniel, R. N., Fu, H., Cane, D. E., and Khosla, C. (1998) J. Am. Chem. Soc. 120, 2478−2479. (13) McDaniel, R., Kao, C. M., Hwang, S. J., and Khosla, C. (1997) Chem. Biol. 4, 667−674. (14) Kellenberger, L., Galloway, I. S., Sauter, G., Bohm, G., Hanefeld, U., Cortes, J., Staunton, J., and Leadlay, P. F. (2008) ChemBioChem 9, 2740−2749. (15) Zheng, J., Piasecki, S. K., and Keatinge-Clay, A. T. (2013) ACS Chem. Biol. 8, 1964−1971. (16) Bailey, C. B., Pasman, M. E., and Keatinge-Clay, A. T. (2016) Chem. Commun. 52, 792−795. (17) Baerga-Ortiz, A., Popovic, B., Siskos, A. P., O’Hare, H. M., Spiteller, D., Williams, M. G., Campillo, N., Spencer, J. B., and Leadlay, P. F. (2006) Chem. Biol. 13, 277−285. (18) O’Hare, H. M., Baerga-Ortiz, A., Popovic, B., Spencer, J. B., and Leadlay, P. F. (2006) Chem. Biol. 13, 287−296. (19) Yuzawa, S., Eng, C. H., Katz, L., and Keasling, J. D. (2013) Biochemistry 52, 3791−3793. (20) Zheng, J., Taylor, C. A., Piasecki, S. K., and Keatinge-Clay, A. T. (2010) Structure 18, 913−922. (21) Murli, S., Kennedy, J., Dayem, L. C., Carney, J. R., and Kealey, J. T. (2003) J. Ind. Microbiol. Biotechnol. 30, 500−509. (22) Yuzawa, S., Katz, L., and Keasling, J. D. (2015) U.S. Patent Application 20150307855. (23) Bihlmaier, C., Welle, E., Hofmann, C., Welzel, K., Vente, A., Breitling, E., Muller, M., Glaser, S., and Bechthold, A. (2006) Antimicrob. Agents Chemother. 50, 2113−2121. (24) Piasecki, S. K., Taylor, C. A., Detelich, J. F., Liu, J., Zheng, J., Komsoukaniants, A., Siegel, D. R., and Keatinge-Clay, A. T. (2011) Chem. Biol. 18, 1331−1340. (25) Hartmann, O., and Kalesse, M. (2014) Angew. Chem., Int. Ed. 53, 7335−7338. (26) Hofferberth, M. L., and Bruckner, R. (2014) Angew. Chem., Int. Ed. 53, 7328−7334. (27) Siskos, A. P., Baerga-Ortiz, A., Bali, S., Stein, V., Mamdani, H., Spiteller, D., Popovic, B., Spencer, J. B., Staunton, J., Weissman, K. J., and Leadlay, P. F. (2005) Chem. Biol. 12, 1145−1153. (28) Hackh, M., Muller, M., and Ludeke, S. (2013) Chem. - Eur. J. 19, 8922−8928.
standard, which was determined by NMR to contain a mixture of diastereomers (26:1 syn:anti). The syn/anti stereochemistries of the products corresponding to starter substrates 1b and 1c are listed in Table 2. It is notable that in each LC−MS analysis, only a single peak corresponding to the expected syn/anti outcome was observed for all combinations of starter substrates and LipPks1+TE variants. This suggests that stereochemical outcome in the context of a PKS module is relatively robust to alterations in substrate structure, compared to observations in studies investigating the activity of isolated KR domains on diketide-SNACs16,24,27 and β-keto esters.28 The broad substrate specificity of LipPks1+TE allowed us to demonstrate that the condensation and reduction activity of KR domain exchange variants are both dependent on substrate structure. Moreover, by using LipPks1+TE as a PKS model system, we were able to harness a strategy of retaining a native DE to alter α-methyl stereochemistry with unprecedented retention of activity and further achieve a novel alteration of KR stereochemical outcome from anti to syn. Identifying structural and molecular recognition elements that have previously remained underappreciated has enormous potential to improve the efficacy of catalytic domain swapping. By utilizing such insights, we hope to pave the way for the exploitation of PKSs as a platform for combinatorial biosynthesis.
■
ASSOCIATED CONTENT
S Supporting Information *
The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.biochem.6b00129.
■
Experimental procedures and additional data (PDF)
AUTHOR INFORMATION
Corresponding Author
*E-mail:
[email protected]. Funding
This work was supported by the Joint BioEnergy Institute, which is funded by the Office of Science, Office of Biological and Environmental Research, of the U.S. Department of Energy (Contract DE-AC02-05CH11231), by the National Science Foundation (Grants EEC-0540879 and MCB-1341894), and by the National Science Foundation Graduate Research Fellowship Program (Grant DGE 1106400 to C.H.E.). Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS The authors thank Kosan Biosciences for E. coli strain K207-3. We are also grateful to Constance Bailey for helpful discussion and to Jim Kirby for assistance with the GC−MS experiments.
■
REFERENCES
(1) Weissman, K. J., and Leadlay, P. F. (2005) Nat. Rev. Microbiol. 3, 925−936. (2) Khosla, C., Tang, Y., Chen, A. Y., Schnarr, N. A., and Cane, D. E. (2007) Annu. Rev. Biochem. 76, 195−221. (3) Cane, D. E., Liang, T. C., Taylor, P. B., Chang, C., and Yang, C. C. (1986) J. Am. Chem. Soc. 108, 4957−4964. (4) Marsden, A. F., Caffrey, P., Aparicio, J. F., Loughran, M. S., Staunton, J., and Leadlay, P. F. (1994) Science 263, 378−380. D
DOI: 10.1021/acs.biochem.6b00129 Biochemistry XXXX, XXX, XXX−XXX
