Wittig/B─H insertion reaction: A unique access to trisubstituted Z-alkenes

Abstract

The Wittig reaction, which is one of the most effective methods for synthesizing alkenes from carbonyl compounds, generally gives thermodynamically stable E-alkenes, and synthesis of trisubstituted Z-alkenes from ketones presents notable challenges. Here, we report what we refer to as Wittig/B─H insertion reactions, which innovatively combine a Wittig reaction with carbene insertion into a B─H bond and constitute a promising method for the synthesis of thermodynamically unstable trisubstituted Z-boryl alkenes. Combined with the easy transformations of boryl group, this methodology provides efficient access to a variety of previously unavailable trisubstituted Z-alkenes and thus provides a platform for discovery of pharmaceuticals. The unique Z-selectivity of the reaction is determined by the maximum overlap of the orbitals between the B─H bond of the borane adduct and the alkylidene carbene intermediate in the transition state.

The Wittig/B–H insertion reaction was developed as a unique access to various previously unavailable trisubstituted Z-alkenes.

INTRODUCTION

Alkenes are important raw materials, and the development of methods for their synthesis has long been the focus of chemists. One such method is the Wittig reaction, that is, the reaction of a carbonyl compound with a nucleophile such as a phosphorus ylide, to generate an alkene by means of an addition-elimination process. This method has the advantages of a broad substrate scope, good functional group tolerance, high yields, and operational simplicity, making it one of the most reliable and practical methods for olefin synthesis (1, 2). However, this protocol is mainly improved toward a practical stereoselective synthesis of thermodynamically stable E-alkenes, while there are still some difficulties in achieving selectivity for the thermodynamically unstable Z configuration, especially for the synthesis of trisubstituted Z-alkenes with ketones as substrates (3–5). The control of Z-selectivity of this strategy is mainly achieved by changing the substituents of carbonyl substrates, activity of phosphorus ylides, reaction conditions, and ionic additives. Nevertheless, these methods are only well applicable to aldehyde substrates to obtain Z-disubstituted olefins, while the Z-selective Wittig reaction of ketones can be achieved only with active alkyl phospholipids (Fig. 1A) (3–10). Therefore, it is necessary to develop reliable strategies for the synthesis of Z-trisubstituted alkenes, which is of great importance in the synthesis, functionalization, and modification of natural products and drug molecules (11).

Fig. 1. Accesses to boryl alkenes.

(A) Wittig-type reaction. (B) This work: Wittig/B─H insertion reaction.

Here, we report a method for Wittig/B─H insertion reactions by which a series of thermodynamically unstable trisubstituted Z-boryl alkenes could easily be synthesized from readily available dialkyl ketones, diazomethyl phosphate or trimethylsilyldiazomethane, and Lewis base–borane adducts under mild reaction conditions with good functional group tolerance (Fig. 1B). Various previously unavailable trisubstituted Z-alkenes were prepared through the stereospecific transformations of boryl groups of the products. In addition, intramolecular Wittig/B─H insertion reactions smoothly produced five- to eight-membered B─N or B─P heterocyclic compounds, many of which cannot be synthesized by known methods.

RESULTS

Reaction development and optimization

We began our study by using estrone 3-methyl ether (1a) and trimethylamine-borane (3a) as model substrates for the Wittig/B─H insertion reaction. To our delight, dropwise addition of 1a and dimethyl (diazomethyl)phosphate (2) dissolved in dichloromethane (DCM) to the solution of 3a and ^tBuONa in DCM at −78°C and then reacting for 2 hours after warmed to −40°C gave target alkenyl borane 4aa in 70% isolated yield [73% by nuclear magnetic resonance (NMR) spectroscopy] and a Z/E ratio of >98:2. Having achieved the desired transformation, we systematically evaluated the reaction conditions (Table 1). In addition to DCM, other solvents [dichloroethane (DCE), tetrahydrofuran (THF), and 1,2-dimethoxyethane (DME)] were suitable; but DCM gave the highest yield and the best Z/E selectivity (entries 1 to 4). Attempts at other bases (KHMDS, ^tBuOK, ^tBuOLi, and MeONa) in the reaction did not lead to a better result (entries 5 to 8). Side reactions appeared to increase at higher temperatures, so the yield and selectivity of the reaction decrease when the temperature was increased (entries 9 to 12). The reaction was compatible with Lewis base–coordinated borane adducts 3b-3f but gave the products with different yields and selectivity (entries 13 to 17). Among these adducts, demethylamine-borane (3b, entry 13) and N-heterocyclic-carbene-borane (3f, entry 17) gave target products in approximately 60% yield but poor Z/E ratio, whereas reaction with dimethylphenylphosphine-borane (3e, entry 16) had a similar yield but higher selectivity (Z/E > 98:2). The reaction with N-methylpiperidine borane (3c) gave 51% yield and moderate selectivity (entry 14). Although the reaction with 3,5-dimethylpyridine-borane (3d) ran smoothly (entry 15), the yield was poor because the product 4ad was unstable and partially decomposed during separation. Unfortunately, the reactions with HBpin or B₂pin₂ instead of borane adducts failed to give desired boryl alkene products. When replacing N₂CHP(O)(OMe)₂ to TMSCHN₂ and using ⁿBuLi as the base, THF as the solvent, the target alkenyl borane was obtained in 63% yield with Z/E ratio of >98:2 (entry 18).

Table 1. Wittig/B─H insertion reaction: Optimization of the reaction conditions^a.

^a Reaction conditions: 1a (0.3 mmol), 2 (0.9 mmol), 3 (1.2 mmol), and base (0.9 mmol), 3 ml of solvent, −78° to −40°C. Reaction time: 2 hours. Conversions and yields were determined by ¹H NMR using 1,3,5-trimethoxybenzene as an internal standard. The data in parenthesis is isolated yield. ^b Performed at −35°C. ^c Performed at −60° to −40°C. ^d Performed at −40°C. ^e Performed at −20°C. ^f Performed at 0°C. ^g Performed at rt (room temperature). ^h Reaction conditions: 1a (0.3 mmol), TMSCHN₂ (0.9 equiv), 3a (1.2 mmol), and ⁿBuLi (0.9 mmol), 3 ml of THF, −78° to −40°C.

Entry	Adduct	Base	Solvent	Product	Conv. (%)	Yield (%)	Z/E
1	3a	tBuONa	DCM	4aa	>98	73 (70)	>98:2
2b	3a	tBuONa	DCE	4aa	94	55	95:5
3	3a	tBuONa	THF	4aa	>98	63	>98:2
4c	3a	tBuONa	DME	4aa	96	56	95:5
5	3a	KHMDS	DCM	4aa	80	47	98:2
6	3a	tBuOK	DCM	4aa	73	44	97:3
7	3a	tBuOLi	DCM	4aa	66	38	>98:2
8	3a	MeONa	DCM	4aa	70	45	>98:2
9d	3a	tBuONa	DCM	4aa	>98	70	97:3
10e	3a	tBuONa	DCM	4aa	>98	68	93:7
11f	3a	tBuONa	DCM	4aa	93	68	90:10
12g	3a	tBuONa	DCM	4aa	87	58	81:19
13	3b	tBuONa	DCM	4ab	>98	(57)	58:42
14	3c	tBuONa	DCM	4 ac	>98	(51)	83:17
15	3d	tBuONa	DCM	4ad	>98	28	63:37
16	3e	tBuONa	DCM	4ae	>98	(63)	>98:2
17	3f	tBuONa	DCM	4af	>98	(61)	53:47
18h	3a	nBuLi	THF	4aa	>98	(63)	>98:2

Evaluation of substrate scopes

Using the optimal conditions, we evaluated Wittig/B─H insertion reactions of various ketones with amine-borane adduct 3a (Fig. 2). At first, we made attempts on bioactive molecules including natural products and pharmaceuticals. In addition to estrone 3-methyl ether (1a), many other steroids and their derivatives such as androsterone (1b), dehydroepiandrosterone (1c), estradiene dione-3-keta (1d), desogestrel derivative (1e), progesterone (1f), hydroxyprogesterone acetate (1g), and methyl oleanonate (1h) were also successfully applied in the reaction, and all afforded good to high yields with excellent Z-selectivity. Besides, loxoprofen derivative (1i) afforded the target product in high yields and good Z-selectivity.

Fig. 2. Wittig/B─H insertion reactions of unsymmetric ketones.

^aReaction conditions: 1 (0.3 mmol), 2 (0.9 mmol), 3a (1.2 mmol), and ^tBuONa (0.9 mmol), in 3 ml of DCM, −78° to −40°C. ^b Reaction conditions: 2 (0.42 mmol) and ^tBuONa (0.42 mmol), other things being equal. ^c Reaction conditions: 1A (5 mmol), 2 (15 mmol), 3a (20 mmol), and ^tBuONa (15 mmol), in 50 ml of DCM, −78° to −40°C, 2 hours.

Subsequently, we evaluated reactions of other unsymmetric aliphatic ketones (R¹ ≠ R²). When we used 4-methylpropiophenone (1j) as substrates, 4ja was obtained in 77% yield, and the Z/E ratio was 68:32. Increasing the difference in steric bulk between R¹ and R² (1j-1q) markedly improved the selectivity of the reaction; the product of the reaction of pinacolone (4pa) had a Z/E ratio of 95:5. Ketones with bicyclo[1.1.1]pentane (1m) and bicyclo[2.2.2]octane (1n) moieties, which were considered as saturated bioisosteres of benzenoids in medicinal chemistry, were also suitable for the reaction. When 3-methyl-3-phenylbutan-2-one (1r) was used, 4ra was obtained in 73% yield with a Z/E ratio of 95:5. Introducing substituents with various electronic properties on the aryl group of the ketone had no obvious effect on the yield or stereoselectivity of the reaction (4sa-4za). Good results were also obtained with ketones bearing a fused aryl group, a heteroaromatic group, or an alkyne (4Aa-4Ea). Even when the steric bulk of the ketone was increased further, the reaction still proceeded smoothly, giving product (4Fa) with high Z-selectivity, although the yield dropped to 41%. Unsymmetric cyclic ketones were also suitable substrates (4Ga-4Ja). When the synthesis of 4Aa was carried out on a gram scale, the yield and selectivity were well maintained, emphasizing the practicality of this method.

Various symmetric aliphatic ketones were also applied to this reaction (Fig. 3A). All the tested symmetric cyclic and acyclic dialkyl ketones smoothly afforded the target products (6aa-6ia) in moderate to high yields. Reactions of cyclic ketones with small, medium, or large rings exhibited high yields (6aa-6ga). Various functional groups were tolerated: ether (6ja), thioether (6ka), aliphatic amino group (6la), aniline (6ma), sulfonamide (6na), amide (6oa, 6pa), silyl (6qa), ester (6ra), hydroxy (6sa), sulfonate (6ta), and nitrile (6ua). The reaction of an aromatic ketone, acetophenone did not afford desired boryl alkene product but gave the prop-1-yn-1-ylbenzene with 87% yield as a major by-product.

Fig. 3. Wittig/B─H insertion reactions of symmetric ketones and intermolecular Wittig/B─H insertion reactions.

(A) Wittig/B─H insertion reactions with symmetric ketones. (B) Intermolecular Wittig/B─H insertion reactions.^a Reaction conditions: 5 (0.3 mmol), 2 (0.42 mmol), 3a (1.2 mmol), and ^tBuONa (0.42 mmol), in 3 ml of DCM, −78° to −40°C. ^b Reaction conditions: 2 (0.9 mmol) and ^tBuONa (0.9 mmol), other things being equal with a. ^c Reaction conditions: 2 (0.84 mmol) and ^tBuONa (0.84 mmol), other things being equal with a. ^d Reaction conditions: 7 (0.4 mmol) and BH₃•THF (0.48 mmol), in 0.5 ml of THF, −20°C, 2 hours; then, TMSCHN₂ (0.68 equiv), ⁿBuLi (0.68 equiv), and THF (0.1 M), −78° to −40°C. ^e Reaction conditions: 8 (0.3 mmol), N₂CHPO(OMe)₂ (0.42 equiv), ^tBuONa (0.42 equiv), and DCM (0.01 M), −78° to −40°C. ^f Reaction conditions: DCM (0.1 M), other things being equal with e.

Nonaromatic boron–containing heterocycles are an important class of cyclic structures; however, only a few types are known, and good synthetic methods are lacking (12–18). As a result, their properties and applications have rarely been studied (19). We found that an intramolecular version of the Wittig/B─H insertion reaction reported herein could be used to directly construct nonaromatic boron–containing heterocyclic compounds (Fig. 3B). Using ketone 7a as the substrate, trimethylsilyldiazomethane as the diazo source, and ⁿBuLi as the base, we obtained target compound 9a in 51% yield. When the carbonyl substrate or the N substituent was changed, product 9b or 9c was obtained in moderate yield. When there was a substituent on the carbon atom between the N atom and the carbonyl group, the reaction also proceeded smoothly: Target compounds 9d and 9e were obtained in moderate yields. Substrates with two carbon atoms between the carbonyl group and the N atom also underwent the desired reaction, affording six-membered B─N heterocycles 9f-9j. Notably, in addition to six-membered ring product 9f, we could also synthesize products with spirocyclic and fused-ring structures 9g-9j by means of this method. Products with a seven- or eight-membered ring (9k and 9l) could also be obtained, albeit in only modest yields. When substrates with both a carbonyl group and a phosphino-borane moiety were used, various B─P heterocycles with five- or six-membered rings (9m-9p) were generated in moderate to good yields. It is worth mentioning that boron heterocycles 9d-9p have not previously been reported. Our synthesis of boron-containing heterocycles 9d-9p lays the foundation for the discovery of boron-containing functional molecules. For example, the B─N heterocyclic compounds synthesized as described in this paper are isosteres of cycloalkenes. Because cycloalkenes are commonly found in natural products and drugs, our method has potential utility for the development of drug molecules.

Transformations of the products

The boryl alkenes obtained by the Wittig/B─H insertion reaction were stable to common purification procedures (e.g., chromatography and recrystallization) and were also stable during long-term storage (no degradation was observed over the course of months). They could undergo various transformations (Fig. 4). To simplify the experimental operation and the difficulty of characterization, we demonstrated the transformations using 4Aa as a substrate. For instance, 4Aa could be easily transformed into pinacol borate 10, boronate N-methyliminodiacetic ester (BMIDA) (11), or boron chloride (16) with no change in the double bond configuration. We obtained a single crystal of 11 and confirmed its configuration by means of x-ray diffraction analysis and nuclear Overhauser effect spectroscopy. Reaction of 4Aa with CuCl₂ or CuBr₂ afforded alkenyl chloride 13 or alkenyl bromide 14, respectively, with well-retained configuration. Reaction with I₂ under basic conditions produced alkenyl iodide 15. Product 4Aa also smoothly underwent Pd-catalyzed Suzuki coupling with an alkenyl bromide or aryl iodide to afford trisubstituted Z-alkene 17 (90%) or 18 (65%). In addition, 4Aa underwent a C─N coupling reaction with imidazole to produce enamine compound 19 and a C─C coupling to produce Z-alkenyl ester compound 21. After 4Aa was converted to the corresponding borate ester, a homologation reaction afforded trisubstituted Z-allyl borate ester 12. Product 4Aa also underwent Rh-catalyzed conjugate addition with a conjugated ketene to afford alkenyl ketone 20. These transformations demonstrate the potential applications of this Wittig/B─H insertion reaction in the synthesis of previously unavailable trisubstituted Z-alkenes.

Fig. 4. Transformations of product 4Aa.

Mechanistic studies

Last, we studied the reaction mechanism, with a focus on accounting for its unique Z-selectivity. We began by carrying out a series of control experiments. A deuterium labeling experiment indicated that a B─H insertion (20–22) process was involved in the transformation (fig. S1A). In addition, a kinetic isotope effect of 1.66 was determined (Fig. 5A and fig. S1, A and B); this value is similar to that for B─H insertion reactions reported in the literature (23–27). The presence of 2,2,6,6-tetramethyl-1-piperidinyloxy free radical (TEMPO), a free radical scavenger, in the reaction system had no obvious effect on the product yield, a result that excludes the involvement of a free radical process (Fig. 5A and fig. S1C). When both deuterated trimethylamine-borane and an N-heterocyclic-carbene-borane were present in the system at the same time, we observed no hydrogen-deuterium exchange in the two products (fig. S1D). Therefore, we speculate that carbene insertion into the B─H bond is concerted.

Fig. 5. Proposed mechanism and computational studies on Z/E selectivity.

(A) Control experiments. (B) Proposed reaction pathway of the Wittig/B─H insertion reaction. (C) Calculations on the Z/E selectivity.

On the basis of our control experiments and previous mechanistic work on alkylidene carbenes (28–31), we propose the reaction mechanism shown in Fig. 5B with 5c as an example substrate. First, ^tBuONa deprotonates the dimethyl (diazomethyl)phosphonate to generate a carbanion, which attacks the carbonyl carbon with the assistance of Na⁺. The resulting species undergoes deoxygenation through a Wittig process to afford an alkylidene diazonium compound (32). Then, dediazonation affords an alkylidene carbene intermediate. The electron-rich hydrogen atom of the borane adduct approaches the empty orbital of the carbene to generate a three-membered ring transition state that leads to the B─H insertion product via the flow of electrons from the B─H bond to the C─H bond with simultaneous transfer of the boron atom to the carbon atom.

The formation of a free alkylidene carbene from the carbonyl compound does not account for the stereoselectivity; instead, the Z/E selectivity arises during insertion of the carbene into the B─H bond. Therefore, we performed density functional theory calculations on 4ra as a model compound by using the M06-2X/def2-TZVP//B3LYP/6–31+G(d) method in DCM solution [using the solvent model based on density (SMD)] with the Gaussian 09 program package (Fig. 5C). The calculations showed that the alkylidene carbene participates in the reaction in a singlet state and that its lone pair of electrons is distributed on the opposite site of the carbene double bond (33–35). Therefore, the lone pair and the carbene double bond are far from each other, and the electronic repulsion between them is minimal. When the electron-rich hydrogen of the borane adduct is close to the carbene carbon, the steric bulk of the two substituents of the alkylidene carbene can be expected to make the energy of the two processes to give Z and E products different. The electron-rich hydrogen of the borane adduct can approach from the methyl side of the double bond (via three-membered ring transition state Ts-2-Z) or from the side with the tertiary carbon substituent (via three-membered ring transition state Ts-2-E). Subsequent completion of the B─H insertion process generates the Z or E product, respectively. Our calculations indicated that the activation energy for the Z-selective process is about 4 kcal/mol lower than that of the E-selective process. This result is consistent with the experimental results (Fig. 2, 4ra).

To better understand the underlying mechanism for this energy difference, we took the three-membered ring transition state (Ts-1) afforded by the insertion of the 4-phenylcyclohexyl-substituted alkylidene carbene into the B─H bond as a reference to analyze the degrees of orbital overlap in the two different transition states (Fig. 5C) (36–38). Because the substituents on the carbene double bond of Ts-1 offer less steric hindrance to the borane adduct than in the other cases (Fig. 2, 4la-4ra), the overlap between the carbene carbon and the frontier orbitals of boron and hydrogen in this transition state is greater than that in any other transition state. To quantify the degree of overlap, we chose the plane defined by the carbene double bond and the carbon atoms of the substituents on each side (C₁─C₂─C₃─C₄) as the reference, and we used the average dihedral angles (θ_B and θ_H) between the plane of the boron or hydrogen atom with the plane of the carbene double bond (B─C₁─C₂ and H─C₁─C₂) and the reference plane, respectively, as indicators of the degree of overlap. Calculations revealed that in Ts-1, the average dihedral angle θ_H-Ref between the negative hydrogen of the borane and the carbene plane is 11.99°, and the average dihedral angle θ_B-Ref between the boron atom and the carbene plane is 18.90° [Fig. 5, C (I)]. The difference between the two average dihedral angles in Ts-2-Z [Fig. 5C (II)] and those in reference transition state Ts-1, Δθ_B-Z and Δθ_H-Z (Δθ_B-Z = 18.60° − 18.90° = −0.30° and Δθ_H-Z = 12.43° − 11.99° = 0.44°) are substantially smaller than the difference between the two average dihedral angles Δθ_B-E and Δθ_H-E in Ts-2-E [Fig. 5C (III)] and those in reference transition state Ts-1 (Δθ_B-E = 16.43° − 18.90° = −2.47° and Δθ_H = 6.44° − 11.99° = −5.55°). Therefore, we concluded that the bonding orbital of the E-selective transition state has a lower degree of overlap and that a higher energy barrier needs to be overcome for bonding. Through interaction region indicator analysis, we also found that the resistance to the insertion process is due mainly to steric repulsion between the carbene carbon and the Lewis base of the borane adduct [Fig. 5C (IV), area a] and to steric repulsion between the borane and the alkyl substituents of the alkylidene carbene [Fig. 5C (IV), area b]. In Ts-2-Z and Ts-2-E, there is a stronger repulsive force due to steric effects in area a of Ts-2-E, and there are obvious van der Waals interactions and a stronger repulsive force in area b of Ts-2-E. Consequently, a higher energy barrier has to be overcome to complete the insertion process through the E-selective transition state. These computational results are in good agreement with our experimental results.

DISCUSSION

In this work, we combined the Wittig reaction with the B─H insertion reaction to develop a convenient method for constructing alkenylboron compounds by adding a carbon atom to carbonyl compounds while also introducing a boron group. Compared with the traditional Wittig reaction, this method has the advantages of using stable, readily available borane substrates, of being operationally simple and convenient, and of affording unusual Z-alkenylborons, a selectivity that has not previously been reported. Moreover, we also realized intramolecular Wittig/B─H insertion reactions, thereby constructing a variety of unreported nonaromatic boron–containing heterocyclic compounds. The boryl alkene products could undergo various transformations and thus have promising utility for synthetic applications. Density functional theory calculations revealed that the unique Z-selectivity arises from the need for maximum overlap of transition-state orbitals during B─H insertion.

MATERIALS AND METHODS

Typical procedure for intermolecular Wittig/B─H bond insertion reaction

Under an argon atmosphere, trimethylamine-borane adduct 3a (4.0 equiv, 1.2 mmol) and ^tBuONa (3.0 equiv, 0.9 mmol) were introduced into a Schlenk tube. A cryogenic coolant circulating pump was used to cool the reaction system to −78°C. Then dichloromethane (2 ml) was injected into the Schlenk tube in one portion. Ketone 1 (0.3 mmol) and dimethyl (diazomethyl)phosphonate 2 (3.0 equiv, 0.9 mmol) were dissolved in dichloromethane (1 ml) in a vial under an argon atmosphere, then added into this reaction mixture dropwise. The resulting mixture was naturally warmed to −40°C and stirred for 2 hours. Petroleum ether was added to quench the reaction. After being warmed to room temperature, the reaction system was filtered and evaporated. The product was purified by flash column chromatography on silica gel with petroleum ether/ethyl acetate as eluent to give the product 4.

Typical procedure for intramolecular Wittig/B─H bond insertion reaction

Procedure A

Under an argon atmosphere, compound 7 (0.4 mmol) was introduced into a vial and 0.4 ml of THF was added to dissolve it. The solution was cooled to −20°C with a cryogenic coolant circulating pump and BH₃·THF (1.2 equiv, 0.48 mmol) was added dropwise. Then, it was stirred at −20°C for 2 hours to obtain 8 without further separation. Meanwhile, under an argon atmosphere ⁿBuLi (1.7 equiv, 0.68 mmol) and THF (1.5 ml) were introduced into a Schlenk tube. A cryogenic coolant circulating pump was used to cool the reaction system to −78°C, and trimethylsilyldiazomethane (1.0 M in THF, 1.7 equiv, 0.68 mmol) was added dropwise. After 2 hours the reaction system was warmed to −40°C and the solution of 8 was added dropwise. The resulting mixture was stirred at −40°C for 2 hours. Petroleum ether was added to quench the reaction. After being warmed to room temperature, the reaction system was filtered and evaporated. The product was purified by flash column chromatography on silica gel with petroleum ether/ethyl acetate as eluent to give the product 9.

Procedure B

Under an argon atmosphere, ^tBuONa (1.4 equiv, 0.42 mmol) were introduced into a Schlenk tube. A cryogenic coolant circulating pump was used to cool the reaction system to −78°C. Then dichloromethane (2 ml) was injected into the Schlenk tube in one portion. Compound 8 (0.3 mmol) and dimethyl (diazomethyl)phosphonate 2 (1.4 equiv, 0.42 mmol) were dissolved in dichloromethane (1 ml) in a vial under an argon atmosphere, then added into this reaction mixture dropwise. The resulting mixture was naturally warmed to −40°C and stirred for 2 hours. Petroleum ether was added to quench the reaction. After warmed to room temperature, the reaction system was filtered and evaporated. The product was purified by flash column chromatography on silica gel with petroleum ether/ethyl acetate as eluent to give the product 9.

Supplementary Materials

This PDF file includes:

Supplementary Text

Figs. S1 to S5

Table S1

NMR Spectra

References