Communication pubs.acs.org/JACS
Quantifying Possible Routes for SpnF-Catalyzed Formal Diels−Alder Cycloaddition Michael G. Medvedev,*,†,‡ Alexey A. Zeifman,‡ Fedor N. Novikov,‡,§ Ivan S. Bushmarinov,*,† Oleg V. Stroganov,‡,§ Ilya Yu. Titov,‡,§ Ghermes G. Chilov,*,‡,§ and Igor V. Svitanko‡ †
X-ray Structural Laboratory, A.N. Nesmeyanov Institute of Organoelement Compounds RAS, 119991 Moscow, Russian Federation N.D. Zelinsky Institute of Organic Chemistry RAS, 119991 Moscow, Russian Federation § MolTech Ltd., Leninskie gory, 1/75 A, 119992 Moscow, Russian Federation ‡
S Supporting Information *
For SpnF, different routes for the title cyclization (Scheme 1) were proposed and studied via quantum-chemical10−12 and kinetic isotope effect13 approaches, but the true mechanism remains ambiguous. Hess Jr. and Smentek11 have studied the cyclization of a truncated substrate (S in Scheme 1) by DFT and MP2 calculations. They concluded that SpnF catalyzes a concerted, highly asynchronous Diels−Alder reaction by folding the substrate into the proper conformation and lowering its activation energy by stabilization of the highly polarized TS structure.11 Gordeev and Ananikov have investigated10 the influence of substitutions and molecular contraction on the activation energy of the title reaction; they concluded that the main role of SpnF is the contraction of the substrate. The recent DFT gas-phase molecular dynamics simulation by Patel et al.12 suggests that the reaction may proceed via a bis-pericyclic TS, which can eventually either directly lead to the product (P) or to I-[6 + 4] undergoing a Cope rearrangement leading to the product P (Scheme 1). Patel and coauthors explored in the utmost detail the reaction paths leading to and from the lowest transition state they located. However, the flexible nature of the macrocycle suggests that a single TS, even a particularly stable one, may be insufficient to get the complete picture. The target of our study is to evaluate quantitatively and exhaustively the possible transition states leading from S to P and to determine the true nature of the title reaction in water, as the first step toward modeling of the SpnF-catalyzed process. We aim to map the whole landscape of possible transition states for the studied reaction. As was found by Black et al.14 and Patel et al.,12 the main free-energy bottlenecks for the Diels−Alder and bispericyclic mechanisms geometrically coincide with the potential energy saddle points, so the thermally corrected canonical transition state theory should provide a good description of the title reaction. So far, the proposed routes for the title cycloaddition are the Diels−Alder8 (DA), bis-pericyclic12 (BPC), and biradical13 (BR) ones. While the biradical mechanism can occur via two different transition states (see Scheme 1), the BR-1 ones should be lower in energy due to higher electron delocalization.15 We also tested a hypothetical mechanism that we will refer to as “altDA” since it starts with the alternative Diels−Alder
ABSTRACT: The Diels−Alder reaction is a cornerstone of modern organic synthesis. Despite this, it remains essentially inaccessible to biosynthetic approaches. Only a few natural enzymes catalyze even a formal [4 + 2] cycloaddition, and it remains uncertain if any of them proceed via the Diels−Alder mechanism. In this study, we focus on the [4 + 2] cycloaddition step in the biosynthesis of spinosyn A, a reaction catalyzed by SpnF enzyme, one of the most promising “true Diels−Alderase” candidates. The four currently proposed mechanisms (including the Diels−Alder one) for this reaction in water (as a first-order approximation of the enzymatic reaction) are evaluated by an exhaustive quantum mechanical search for possible transition states (728 were found in total). We find that the line between the recently proposed bis-pericyclic [J. Am. Chem. Soc. 2016, 138 (11), 3631] and Diels−Alder routes is blurred, and favorable transition states of both types may coexist. Application of the Curtin−Hammett principle, however, reveals that the bis-pericyclic mechanism accounts for ∼83% of the reaction flow in water, while the classical Diels−Alder mechanism contributes only ∼17%. The current findings provide a route for modeling this reaction inside the SpnF active site and inferring the catalytic architecture of possible Diels− Alderases.
T
he Diels−Alder reaction is one of the most powerful tools of modern synthetic chemistry.1 It is of special importance in the synthesis of biologically active compounds and pharmaceuticals where construction of complex cyclic scaffolds is often needed.1 However, biosynthetic approaches to Diels− Alder reaction are not yet available.2 Among the very few enzymes catalyzing at least a formal [4 + 2] cycloaddition, some have been shown to act through a non-Diels−Alder mechanism,3 while the exact mechanisms for the others remain unknown.4−7 One of the promising candidates for a “true Diels−Alderase” is the SpnF enzyme,8 which catalyzes a [4 + 2] cycloaddition step (Scheme 1) in the biosynthesis of Spinosyn A, an effective and “green” insecticide.9 SpnF is remarkable since it was the first discovered natural enzyme catalyzing solely the cycloaddition step.10 Although SpnF was discovered five years ago,8 the exact mechanism remains unknown. © 2017 American Chemical Society
Received: December 24, 2016 Published: February 27, 2017 3942
DOI: 10.1021/jacs.6b13243 J. Am. Chem. Soc. 2017, 139, 3942−3945
Communication
Journal of the American Chemical Society Scheme 1. Formal [4 + 2] Cycloaddition Catalyzed by SpnFa
a
All routes proposed up to date are shown: the Diels−Alder (DA), bis-pericyclic (BPC), biradical (BR), and alternative Diels−Alder (altDA) ones.
cycloaddition, followed by two Cope rearrangements (Scheme 1). Quantum mechanical calculations were performed at M062X16/6-31+g(d)17 level of theory with ultrafine grid and implicit water modeled with PCM. This level of theory is known to be appropriate for organic chemistry calculations18 and has been shown to provide very accurate results for the Diels−Alder reaction.19 The modeled molecule was simplified by replacing the ethyl group at C21 with a methyl. As a cross-check, the energies of found stationary points were recalculated with the PBE020-D321 method recently shown to be well-grounded in theory;22 it provided qualitatively identical results (see SI and Figure S1). At the first step, we have generated possible conformations of P using MMBS23 algorithm. It provided 1971 structures, which were optimized with the MM3 force field. After removal of duplicates (see SI), 560 unique structures remained. Subsequent calculations were performed at the M06-2X/6-31+g(d) level as described above. Each structure was optimized, and a search for the [4 + 2] TS leading to the current conformation of P was performed (see SI for details). This generated 376 unique TSs with energies within 30 kcal mol−1 from the lowest one (TSs with higher energies were excluded from further consideration). The found TSs have similar geometries of the reacting fragment, with the C4−C7−C11−C12 torsion angle for 85% of them in the range from 0° to 30° (Figure S2). They represent all reasonable conformations leading to P via formation of C4−C12 and C7−C11 bonds, so we have used their geometries in searches for TSs of other routes. The 376 TSs generated on the previous step belong to either DA or BPC mechanisms; however, some TSs may have been missed in the case where both routes lead to the same product conformation. We searched for these additional TSs starting from the already located ones by shortening the C2−C14 bond and discovered eight additional unique TS structures. A search by increasing the C2−C14 distance did not yield any new TSs. The BR-1 mechanism should have smaller reaction barriers15 than BR-2, so it was studied first. Starting from initial 376 TS structures, 331 unique triplet biradical TSs (BR-1T) were located, all with much higher energies than the corresponding singlet TSs (Figure 1). The lowest BR-1T TS lies 19 kcal mol−1
Figure 1. Notched box plot24 of thermally corrected energies for bottleneck TSs (DA, BPC, BR-1S, BR-1T, and altDA) relative to the lowest S structure. Notches denote 95% confidence intervals of medians.
above the lowest overall TS. Starting from the located BR-1T TSs, we have performed a search for singlet biradical TSs (BR1S), which were shown to be feasible for a similar reaction.25 Nine BR-1S TSs were located, the lowest of them being 11 kcal mol−1 above the lowest nonbiradical TS (Figure 1). Therefore, biradical mechanisms are energetically disfavored and do not contribute to the reaction in water. In addition, we looked for altDA TSs starting from the 376 initial TS structures; only four TSs have been located. The lowest of them is 34 kcal mol−1 higher than the lowest overall TS (Figure 1). Thus, the only possible routes for the title reaction in the absence of an enzyme are DA and BPC for which a total of 384 TSs were located. Steepest descent from these TSs along the imaginary frequency normal modes resulted in 384 substrates and the same number of products. All substrates correspond to S, while the products are of two kinds: 371 correspond to P and the other 13 are I-[6 + 4]. For the latter, the Cope TSs were located (see SI), which turned out to be lower than the preceding cycloaddition TSs by at least 7 kcal/mol. Therefore, the BPC TS is the bottleneck on the S → [BPC]‡ → I-[6 + 4] → [Cope]‡ → P route. Due to the complicated nature of the BPC path,12 TSs cannot be attributed to DA or BPC mechanisms considering only the products they lead to. Moreover, assignment of TS types for 384 geometries with differing conformations and bond lengths by 3943
DOI: 10.1021/jacs.6b13243 J. Am. Chem. Soc. 2017, 139, 3942−3945
Communication
Journal of the American Chemical Society
Figure 2. Dependence of the TS energy on the C4−C12 and C2−C14 interatomic distances. The TS found by Patel et al.12 is denoted by a cross. Black line shows the average energy for a given bond length, with the light gray area denoting its 95% confidence interval.
This result provides an explanation for the experimental kinetic isotope effect study,13 which demonstrated that C7−C11 bond in P is formed before the C4−C12 one in the absence of the enzyme. The behavior of a reaction in water as a whole is guided by the Curtin−Hammett principle,27 i.e., determined by relative TS energies due to fast interconversion of substrates. We applied the corresponding equations to our data (see SI for details) and found that ∼95% of the reaction should proceed through 14 transition states summarized in Table 1. Thus, while DA TSs are
hand would be ambiguous; a robust classification approach was required. For this task, we employed the quantum theory of atoms in molecules (QTAIM),26 which identifies chemical bonding as a bridge of electron density (“bond path”) linking the participating atoms. We hypothesized that the bond paths corresponding to specific not fully formed bonds will allow identifying different types of TSs. In addition, we compared the relative TS energies of molecules having a given intramolecular bond path to the overall distribution (see SI, Figure S3). The C7−C11 and C4−C12 bond paths were found in all TSs except for a few high-energy ones; what came as a surprise was the fact that the presence of C2−C14 bond path was strongly associated with a lowered TS energy. Since the formation of this bond is a feature of the BPC transition state, we could immediately conclude that the BPC transition states are on the average more stable than DA. Further, we classify the TSs with bond path linking C2 to C14 atom or one of them to the second’s immediate neighbor as BPC. Thus, among the singlet TSs found, 144 correspond to the BPC mechanism and the remaining 240 belong to the DA mechanism (see Figure S4 for the lowest lying ones). A box plot for the thermally corrected energies of the located TSs (240 DA, 144 BPC, 9 BR-1S, 331 BR-1T, and four altDA) is shown in Figure 1. This presentation, however, simplifies the actual distribution of transition states associated with the title reaction. From an explicit consideration of the C4−C12 and C2−C14 distances (Figures 2, S5) we can see that there is no sharp border between BPC and DA transition states. Actually, two nearly independent trends affect the transition state energy: first, shortening of the C4−C12 bond is disfavorable, with its optimum length being ∼2.75 Å; second, transition states with partially formed C2−C14 bond are more favorable on average. The actual ranges of C4− C12 and C2−C14 distances only partially overlap, with the BPC transition state found by Patel et al. (located independently by our search as TS #7) being more of an exception with its nearly equal C4−C12 and C2−C14 bond lengths. Most notably, there exist relatively favorable TSs corresponding to DA path (with C2−C14 bond distance >4 Å), but the C4−C12 bond in them is still not fully formed, being longer than C7−C11 by at least 0.7 Å and making the corresponding Diels−Alder reaction highly asynchronous.
Table 1. Important TS IDs, Types, Thermally Corrected Relative Energies (Erel), Lengths of Forming Bonds, and Contributions TS #
TS type
Erel(TS) (kcal mol−1)
dC4−C12 (Å)
dC2−C14 (Å)
contriba (%)
120 7b 69 231 197 85 3 309 43 232 0 35 36 113
BPC BPC BPC BPC DA BPC BPC BPC DA BPC BPC BPC DA DA
0.00 0.29 0.46 0.58 0.73 1.07 1.11 1.11 1.15 1.17 1.17 1.20 1.29 1.48
2.822 2.883 2.709 2.672 2.819 2.960 2.815 2.739 2.713 2.761 2.688 2.691 2.811 2.719
3.151 3.034 3.496 3.476 3.722 2.986 3.117 3.399 3.553 2.948 3.452 3.388 4.576 4.779
24.0 14.6 11.0 9.0 7.0 3.9 3.7 3.7 3.4 3.3 3.3 3.2 2.7 2.0
a Calculated from the Curtin−Hammett principle. bTS studied by Patel et al.12
more numerous, BPC TSs have larger contribution due to their lower relative energies. At 298 K, ∼83% of the reaction proceed through BPC TSs and ∼17% through DA TSs. This result is well reproduced by PBE0-D3 calculations (see SI) even despite the difference between M06-2X and PBE0 functionals.22 This is likely caused by a favorable error cancellation arising from the fact that all DA and PBC TSs are similar in nature, so the functionals’ systematic errors largely cancel out.28 3944
DOI: 10.1021/jacs.6b13243 J. Am. Chem. Soc. 2017, 139, 3942−3945
Communication
Journal of the American Chemical Society
(2) Preiswerk, N.; Beck, T.; Schulz, J. D.; Milovník, P.; Mayer, C.; Siegel, J. B.; Baker, D.; Hilvert, D. Proc. Natl. Acad. Sci. U. S. A. 2014, 111, 8013−8018. (3) Guimarães, C. R. W.; Udier-Blagović, M.; Jorgensen, W. L. J. Am. Chem. Soc. 2005, 127, 3577−3588. (4) Klas, K.; Tsukamoto, S.; Sherman, D. H.; Williams, R. M. J. Org. Chem. 2015, 80, 11672−11685. (5) Minami, A.; Oikawa, H. J. Antibiot. 2016, 69, 500−506. (6) Byrne, M. J.; Lees, N. R.; Han, L.-C.; van der Kamp, M. W.; Mulholland, A. J.; Stach, J. E. M.; Willis, C. L.; Race, P. R. J. Am. Chem. Soc. 2016, 138, 6095−6098. (7) Li, L.; Yu, P.; Tang, M.-C.; Zou, Y.; Gao, S.-S.; Hung, Y.-S.; Zhao, M.; Watanabe, K.; Houk, K. N.; Tang, Y. J. Am. Chem. Soc. 2016, 138, 15837−15840. (8) Kim, H. J.; Ruszczycky, M. W.; Choi, S.; Liu, Y.; Liu, H. Nature 2011, 473, 109−112. (9) Yano, B. L.; Bond, D. M.; Novilla, M. N.; McFadden, L. G.; Reasor, M. J. Toxicol. Sci. 2002, 65, 288−298. (10) Gordeev, E. G.; Ananikov, V. P. PLoS One 2015, 10, e0119984. (11) Hess, B. A., Jr.; Smentek, L. Org. Biomol. Chem. 2012, 10, 7503− 7509. (12) Patel, A.; Chen, Z.; Yang, Z.; Gutiérrez, O.; Liu, H.; Houk, K. N.; Singleton, D. A. J. Am. Chem. Soc. 2016, 138, 3631−3634. (13) Kim, N. H. Ph.D. Dissertation, The University of Texas at Austin, Austin, TX, 2013. (14) Black, K.; Liu, P.; Xu, L.; Doubleday, C.; Houk, K. N. Proc. Natl. Acad. Sci. U. S. A. 2012, 109, 12860−12865. (15) Rogers, D. W.; Matsunaga, N.; Zavitsas, A. A. J. Org. Chem. 2006, 71, 2214−2219. (16) Zhao, Y.; Truhlar, D. G. Theor. Chem. Acc. 2008, 120, 215−241. (17) Hehre, W. J.; Ditchfield, R.; Pople, J. A. J. Chem. Phys. 1972, 56, 2257−2261. (18) Peverati, R.; Truhlar, D. G. Philos. Trans. R. Soc., A 2014, 372, 20120476. (19) Linder, M.; Brinck, T. Phys. Chem. Chem. Phys. 2013, 15, 5108− 5114. (20) Adamo, C.; Barone, V. J. Chem. Phys. 1999, 110, 6158−6170. (21) Grimme, S.; Antony, J.; Ehrlich, S.; Krieg, H. J. Chem. Phys. 2010, 132, 154104. (22) Medvedev, M. G.; Bushmarinov, I. S.; Sun, J.; Perdew, J. P.; Lyssenko, K. A. Science 2017, 355, 49−52. (23) Watts, K. S.; Dalal, P.; Tebben, A. J.; Cheney, D. L.; Shelley, J. C. J. Chem. Inf. Model. 2014, 54, 2680−2696. (24) McGill, R.; Tukey, J. W.; Larsen, W. A. Am. Stat. 1978, 32, 12. (25) Yu, P.; Patel, A.; Houk, K. N. J. Am. Chem. Soc. 2015, 137, 13518− 13523. (26) Bader, R. F. W. Atoms in Molecules: A Quantum Theory; Oxford University Press: Oxford, 1994. (27) Seeman, J. I. J. Chem. Educ. 1986, 63, 42. (28) Medford, A. J.; Wellendorff, J.; Vojvodic, A.; Studt, F.; AbildPedersen, F.; Jacobsen, K. W.; Bligaard, T.; Nørskov, J. K. Science 2014, 345, 197−200. (29) Sadovnichy, V.; Tikhonravov, A.; Voevodin, V.; Opanasenko, V. “Lomonosov”: Supercomputing at Moscow State University. In Contemporary High Performance Computing: from Petascale toward Exascale; Chapman & Hall/CRC Computational Science; CRC Press: Boca Raton, FL, 2013; pp 283−307.
We can conclude that only two mechanisms, namely, the Diels−Alder and the bis-pericyclic ones, can contribute to the reaction in water, while the energy barriers for alternative Diels− Alder and biradical paths are too high. Bis-pericyclic transition states are less numerous but tend to have lower energies and account for ∼83% of the nonenzymatic reaction at 298 K in water. Further theoretical modeling of this reaction in the active site of the protein will require all substrates’ and transition states’ geometries located in the current study to get the final picture of this first pure-Diels−Alderase workflow. According to our results, Diels−Alder, bis-pericyclic, and singlet biradical mechanisms are plausible in the SpnF active center. Still, given that BPC TSs are more compact and stable on average, this route is more likely for the enzymatic reaction.
■
ASSOCIATED CONTENT
S Supporting Information *
The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/jacs.6b13243. Figures S1−S9. Computational data, including a detailed description of transition states search methodology, details on QTAIM analysis, and Curtin-Hammett calculations (PDF) Energies of all located stationary points at M06-2X level of theory. Energies of important stationary points at PBE0D3 level of theory, as well as their thermally corrected energies at M06-2X level of theory (XLSX) Cartesian coordinates of all located stationary points (S: XYZ; Non-BR TSs: XYZ; BR-1S TSs: XYZ; BR-1T TSs: XYZ; P: XYZ)
■
AUTHOR INFORMATION
Corresponding Authors
*
[email protected] *
[email protected] *
[email protected] ORCID
Michael G. Medvedev: 0000-0001-7070-4052 Ivan S. Bushmarinov: 0000-0002-6534-4133 Author Contributions
All authors contributed equally. Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS M.G.M., F.N.N., O.V.S., G.G.C., and I.V.S. are grateful to Russian Science Foundation grant # 17-13-01526 for financial support. I.S.B. is grateful to Russian Foundation for Basic Research grant # 17-03-00907 for financial support. A.A.Z. is grateful to Russian Foundation for Basic Research grant # 16-33-60220 for financial support. M.G.M. acknowledges the computational resources provided by the Moscow State University’s Faculty of Computational Mathematics and Cybernetics: IBM Blue Gene/P supercomputer; by the Supercomputing Center of Lomonosov Moscow State University: the Lomonosov29 supercomputer; and by the federal center for collective usage at NRC “Kurchatov Institute”, http://computing.kiae.ru/.
■
REFERENCES
(1) Nishiwaki, N. Methods and Applications of Cycloaddition Reactions in Organic Syntheses; John Wiley & Sons: Hoboken, NJ, 2013. 3945
DOI: 10.1021/jacs.6b13243 J. Am. Chem. Soc. 2017, 139, 3942−3945
