1
A Data Intensive Approach to Mechanistic Elucidation Applied to Chiral Anion Catalysis Andrew J. Neel †,* , Anat Milo ‡,* , Matthew S. Sigman ‡ , F. Dean Toste † † Department of Chemistry, University of California, Berkeley, CA 94720 ‡ Department of Chemistry, University of Utah, Salt Lake City, UT 84112 Summary: An understanding of the catalyst-substrate interactions that underlie selectivity (e.g. enantio-, diasterio-, or regioselectivity) in a given reaction is essential for the rational design of superior catalytic systems. Frequently however, it is difficult to discern how molecular features work in concert to achieve a desired outcome. We reasoned that in such instances, the outcome of every catalyst-substrate combination contains valuable information regarding their transition state interaction. Herein we describe a methodology for extracting this information in a meaningful way. This technique involves correlating various molecular descriptors with experimental enantioselectivity outcomes using linear regression techniques, and subsequently using the resulting predictive models to develop a mechanistic proposal that can be evaluated experimentally. Two case studies are presented from the field of chiral anion catalysis in which catalyst-substrate association is particularly challenging to discern. In each case, this data-intensive approach has yielded non-intuitive insights that have led to a deeper mechanistic understanding, enabling in one instance the rational design of a catalyst that displays the highest enantioselectivity observed for its system. Asymmetric Fluorination (cont’d) 2. Asymmetric Fluorination Acknowledgements 1. Asymmetric CDC Design/prepare library Collect descriptors Obtain experimental data Develop hypothesis Experimentally validate N N N R R Big Small Electron Rich Electron Poor N N N Geometry IR Vibrations N N O R R ee 1 = XX ee n = YY P O O O O N R ? ? O O N N N N N N P O OH tailored catalysts B 1 : σ : Hammett Charge 1. 2. 3. 5. 6. Identify correlations -0.5 0 0.5 1 1.5 2 -0.5 0 0.5 1 1.5 2 Measured DDG ‡ HkcalêmolL Predicted DDG ‡ HkcalêmolL 4. Background - 0.5 0 0.5 1 1.5 2 - 0.5 0 0.5 1 1.5 2 Measured DDG ‡ Hkcal ê mol L Predicted DDG ‡ Hkcal ê mol L 3 6 4 5 8 7 2 1 y = 0.92x + 0.07 R 2 = 0.91 9 10 11 N N N Br i Pr i Pr N N N single i Pr group = torsion w/o clash displays torsion angle close to 90 o Data-Intensive Approach: Outline Ee determined for every catalyst-substrate combination (132 total) Ee of each substrate with all catalysts modeled using only catalyst descriptors 4 and vice versa (23 small models total) Enantioselectivity trends not intuitive, even small modification to substrates or catalysts drastically affects enantioselectivity Hypothesis : Every data point is meaningful: even low selectivity values contain valuable information regarding substrate-catalyst interaction in the transition state Goal : Develop method to extract this information Multiple models explored N O H N 2-OMe H 2-Me H 2-Br H 2- i Pr H 2-OMe Br 2-OMe i Pr 2-OMe Ph 85 (85) 88 (90) 95 (93) 93 (94) 93 (92) 93 (90) 91 (92) 92 (89) substrate % ee R 2 R 1 4-Me 78 (77) 4-OMe 78 (75) 4-NO 2 87 (79) H H H 1 2 3 4 5 6 7 8 entry 9 10 11 R 1 R 2 H H N N N i Pr i Pr Br N N N i Pr References 1. Mahlau, M.; List, B. Angew. Chem. Int. Ed. 2013, 52, 518. 2. Rauniyar, V.; Lackner, A. D.; Hamilton, G. L.; Toste, F. D. Science 2011, 334, 1681. 3. Neel, A. J.; Hehn, J. P.; Tripet, P. F.; Toste, F. D. J. Am. Chem. Soc. 2013, 135, 14044. 4. Milo, A.; Bess, E. N.; Sigman, M. S. Nature, 2014, 507, 210. 5. Neel, A. J.; Milo, A.; Toste, F. D.; Sigman, M. S. Science 2015, 347, 737. 6. Zi, W.; Wang, Y-M.; Toste, F. D. J. Am. Chem. Soc. 2014, 136, 12864. 7. Martínez-Aguirre, M. A.; Yatsimirsky, A. K. J. Org. Chem. 2015, 80, 4985. In 2014, Toste and coworkers disclosed a method for the enantioselective fluorination of allylic alcohols via chiral anion phase-transfer catalysis, using aryl boronic acids (BA) as transient directing groups (e.g. B). 6 Preliminary survey reveals ability to tune ee over 2.5 kcal/mol range (–78 - 65 % ee) by changing only BA Library of PAs and BAs prepared to cover large structural space then ee’s obtained for each combination Degree of asymmetric induction heavily dependent on BA and PA structure- better understanding of their interaction may facilitate improvement of scope. Dataset trends and model complexity suggest that different mechanisms may be operative depending on PA/BA structure. Catalyst-dependent nonlinear effect supports this notion. Catalyst speciation affected by PA/BA structure. Given propensity of phosphate anion to form –ate complex with boronic acids 7 , a plausible scenario may involve this species. B O O Ph OH P O O O * B O Ph P O O O * O O H R R N N Cl F X N N Cl F X -ate complex H-bonded complex Current efforts focused on modeling subsets of the data to identify structural features controlling selectivity within a given regime. M + – chiral anion + X – insoluble cationic reagent – + soluble chiral ion pair enantioenriched product prochiral substrate + The Toste group has a long standing interest in the use of chiral anions for asymmetric catalysis. 1 Recently, this strategy has been applied in the realm of phase transfer catalysis. 2 In 2013, Toste and coworkers reported an enantioselective oxidative coupling reaction enabled by a novel class of triazolyl phosphoric acids (PAs, e.g. 3a). 3 N H N O R 1 N N N R R 1 R 2 R H H (4a) 4-Me 4-OMe 4-NO 2 2-OMe 2-Br 2- i Pr 2-Me 2-OMe 2-OMe 2-OMe 2,6-(OMe) 2 H (4b) H (4c) H (4d) H (4e) H (4f) H (4g) H (4h) H (4l) Ph (4i) Br (4j) i Pr (4k) CH(Ph) 2 (3b) 2,4,6-(Me) 3 Ph (3c) 2,4,6-( i Pr) 3 Ph (3d) 2,4,6-(Cy) 3 Ph (3e) 1-adamantyl (3a) Ph (3f) 4-OMe-Ph (3g) 4-(NEt 2 )-Ph (3h) 4-(SO 2 Me)-Ph (3i) 2,6-(OMe) 2 -Ph (3j) 2,6-(F) 2 -Ph (3k) R 2 Conclusion : Succeeded in rationally modifying noncovalent interactions to improve an enantioselective reaction 5 3,5-(OMe) 2 3,5-Me 2 3,5-(CF 3 ) 2 3- OMe 3- Me 3- CF 3 X X XX X 2,6-(CF 3 ) 2 2- OMe 2,6-Me 2 2-CF 3 2- Me 4- t Bu 4-NO 2 4-Ph 4- OMe 4-CF 3 4-Me 4- Bn 4- i Bu B(OH) 2 R 4-R' 2-R 3-R -1.3 (-80) 1.0 (70) kcal/mol (% ee) Ph OH F (S) (R) Ph Me OH (S)-TRIP ( 10 mol %) Selectfluor (1.3 eq.) Na 2 HPO 4 (4.0 eq.) boronic acid (1.0 eq.) MS 4A toluene (0.1 M) rt, 16 h, 500 rpm Ph OH F (S)-6 5 O O P O OH R R B HO OH R (7b) 2,4,6-(Me) 3 (7c) 2- i Pr (7d) H (7e) 2,6-( i Pr) 2 (7f) 4- i Pr (7g) 3,5-(Me) 2 (7h) 3,5- (CF 3 ) 2 R = 4-Me = 4-NO 2 = 2-Me = 2-OMe = 3,5-(OMe) 2 = 3,5-(Me) 2 = 3,5-(CF 3 ) 2 B OH HO B OH HO B OH HO B OH HO Me Me NO 2 MeO 7a 7b 7c 7d 7e 7f 7g 7h 66 91 29 47 -15 -8 72 11 24 55 70 33 19 -2 0 -8 -61 7 28 2 4 -15 -6 -57 -48 41 -2 8 B OH HO MeO OMe B OH HO Me Me -77 20 23 22 12 21 -23 -31 63 37 42 34 31 -16 36 38 18 13 -12 10 cat R = (7a) 2,4,6-( i Pr) 3 0 10 20 30 40 50 60 70 0 10 20 30 40 50 60 70 80 90 100 Product Ee (%) Catalyst Ee (%) (S)444iPr O O P O OH i Pr i Pr Ph OH F + o-tolylboronic acid 7f (S)-6 0 10 20 30 40 50 60 70 0 10 20 30 40 50 60 70 80 90 100 Product Ee (%) Catalyst Ee (%) (S)4TRIP Ph OH F O O P O OH i Pr i Pr i Pr i Pr i Pr i Pr + p-tolylboronic acid 7a (R)-6 Specific goals: accurately predict new PA/BA outcomes via understanding of underlying mechanism. Ph HO B OR HO B O O B HO OH Me Me (R)-Na-7f (1 equiv) (1.3 equiv) Ph OH P O O O * benzene-d 6 11 B NMR: δ = 28.7 ppm 11 B NMR: δ = 1.45 ppm 11 B NMR: δ = 30.4 ppm The authors acknowledge the University of California and the Lawrence Berkeley National Laboratory for funding. A.J.N. gratefully acknowledges an Amgen Fellowship in Organic Chemistry for funding and an ACS DOC grant for assistance with travel costs. A.M. would like to acknowledge the Center for High Performance Computing at the University of Utah. 18 examples up to 94 % ee up to 93% yield N H N O R 1 N N O R 1 triazole PA (5 mol%) N O AcHN BF 4 Na 3 PO 4 (2.4 equiv.) p-xylene, rt, 24 h R 2 R 2 N H N O R 1 R O P O O O * via chiral ion pair catalyst conversion (%) ee (%) 1 3a 86 91 8 -84 R 2 R 2 R 2 X N Y pyr-3a imid-3a 88 -41 93 -45 entry 1 3 4 5 Ar = O O Ar Ar C 8 H 17 C 8 H 17 pyr-3a: X=CH, Y=N imid-3a: X=N, Y=CH 3a: X=N, Y=N 1: R 2 = i Pr 2 95 16 2 2: R 2 = Cy PO 2 H R 1 = Bn R 2 = H A OMe 74 82 38 77 43 84 77 75 NO 2 Me 69 80 49 78 MeO 41 70 82 70 N N N N N N Cy Cy Cy N N N N N N ‡ ‡ ‡ ‡ ‡ ‡ ‡ ‡ ‡ ‡ ‡ CHPh2 2,4,6- MePh 2,4,6- iPrPh 2,4,6- CyPh 1- Ad Ph 4- OMe - Ph 2,6- OMe - Ph 4- H NEt2L Ph 2,6- F - Ph 4- SO2Me- Ph 0.5 1 1.5 2 0.5 1 1.5 2 Measured DDG ‡ Hkcal ê mol L Predicted DDG ‡ Hkcal ê mol L 2-Me-Bn * * * * * * * * * N N O Me cat 3h 3g 3f ee 1 3i 62 33 3a 3b 91 63 57 23 R 2 = 0.95 y = 0.95x + 0.05 3j 3k 53 63 Torsion Angle = –0.14 - 0.18νN=N + 1.39 νRingD + 0.33 νN=N:νRingD R 2 = 0.91 ΔΔG = 1.06 + 0.36 νN=N - 0.42tor R 2 = 0.96 ΔΔG = 1.18 + 0.67 νN=N – 0.61 νRingD – 0.18 νRingD:νN=N * * * * * * N N N νN=N νRing Deformation R torsion angle 4f 3d 3e 3c 88 93 92 * * * Ph Me OH (S)-AdDIP ( 10 mol %) Selectfluor (1.3 eq.) Na 2 HPO 4 (4.0 eq.) p-tolylboronic acid (1.0 eq.) MgSO 4 p-xylene/ethylcyclohexane (1:1) rt, 16 h Ph O B Ar O [F] H P O O O O * H n Ph OH n F 6, n =1, 72 % yield, 93 % ee n = 2, 48 % yield, 0 % ee 5, n =1 B n =2 up to 86 % yield up to 94 % ee O O P O OH Ad Ad i Pr i Pr i Pr i Pr (S)-AdDIP 15 examples B(OH) 2 B(OH) 2 B(OH) 2 B(OH) 2 F F F F F Me Me t-Bu B(OH) 2 Me with (S)-TRIP 35 0 –30 25 50 ee of (S)-6 4-Me 4-NO 2 2-Me 2-OMe –tor (cat) –B 1 (cat) B 5 (BA) 2,4,6 -Me (7b) 2,4,6- i Pr (7a) AdDIP 2- i Pr (7c) H (7d) 4- i Pr (7f) 2-BA with 2,4,6-cat
