Differential Reactivities of Enyne Substrates in Ruthenium- and Palladium-Catalyzed Cycloisomerizations

Abstract

Complementary methods for the transition metal-catalyzed enyne cycloisomerizations of cyclic olefins have been developed. By using distinct ruthenium and palladium catalysts, decalins and 7,6-bicycles can be obtained with dichotomous stereochemical outcomes. The change in mechanism that accompanies the change in metal affords trans-fused 1,4-dienes with ruthenium and their cis-fused diastereomers under palladium catalysis. In the reactions under ruthenium catalysis, a coordinating group is required, and acts to direct the metal to the same side of the carbocycle, resulting in the observed trans diastereoselectivity. Subtle changes in the carbocyclic substrate led to the discovery of a heretofore-unobserved mechanistic pathway, providing bicyclic cycloisomerization products under palladium catalysis and tricyclic products under ruthenium catalysis in DMA. The differential effect of DMA supports a mechanism in which the coordination requirements of the two paths differ, allowing for the reaction to be shuttled through the metallacycle pathway (generating tricyclic products) when DMA is used as a solvent.

Introduction

A primary focus in our laboratory is the invention of transition metal-catalyzed reactions for the rapid construction of molecular complexity from simple starting materials.¹ Of these, atom-economic reactions² - transformations that only use catalytic reagents and in which each atom of the starting materials is contained in the product - exemplify the ideal of efficient chemical synthesis. Developing methods that meet these criteria is desirable not only because such processes generate few byproducts, but because they make more efficient use of raw materials in an increasingly resource-conscious world.

(1)

A representative example of such a transformation is the cycloisomerization of enynes.³ Extending our previous work in this area to enynes containing cyclic olefins would allow access to fused carbocyclic ring systems (eq 1), chemical motifs that are abundant in naturally occurring, biologically active molecules. Recognizing, therefore, that the ability to rapidly construct fused 5-, 6-, and 7-membered ring systems would be an important synthetic tool, and that control over the relative stereochemistry of fused ring junctions is a longstanding challenge in the synthetic community, we began a research program to study the reaction of cyclic olefin substrates (1) with a variety of transition metal catalysts.

(2)

A preliminary account of these studies, detailing palladium- and ruthenium-catalyzed cycloisomerizations of 6- and 7-membered carbocyclic enynes, was recently disclosed.⁴ Herein we present a full account of this work, which has resulted in the discoveries of 1) diastereoselective enyne cycloisomerizations to form bicycles that, depending on the metal catalyst, exclusively form either the cis- or trans-fused ring system and 2) reaction conditions that, depending on the metal catalyst, selectively yield either bi- or tricyclic products. These studies provide new insights into the mechanisms of these reaction manifolds, the implications of which are discussed.

Preparation of Enyne Substrates

(3)

The requisite enyne substrates can be readily prepared via a general synthetic plan that relies on the deconjugative alkylations of cyclic enoates.⁵ In the event, commercially available cyclohexene 6 was alkylated with iodides 7, 8, and 9 using LDA/DMPU to provide esters 10, 11, and 12, which vary only in their tether lengths (eq 3).

Substrates containing both larger and smaller ring sizes were also synthesized (eqs 4 and 5) following a similar protocol. Deconjugative alkylations of cyclopentene 18 and cycloheptene 19 were less effective than the alkylation of cyclohexene 6, but worked reasonably well using LDA/HMPA to afford enynes 20 and 21 from iodide 7, and enynes 26 and 27 from iodide 8.⁶

(4)

(5)

The silylated esters obtained from the deconjugative alkylation (10–12, and 26–27) and their desilylated analogs (13–15, 22–23, and 28–29) were used to access an array of substrates via conventional methods. The synthetic routes to enynes 34–38 from precursor 11 are illustrated in Scheme 1; the cyclopentene- and cycloheptene-derived congeners were made analogously.⁷

Scheme 1.

Substrate Syntheses

Enynes containing an amide at the quaternary were synthesized via standard conditions (eq 7).

(6)

(7)

(8)

Cyclooctene substrate 50 could not be synthesized using deconjugative alkylation, likely owing to the conformational restrictions caused by transannular interactions in a cyclooctyl ring. Instead, β,γ-unsaturated ester 49 was synthesized using a ninhydrin ene reaction reported by Gill, et al.,⁸ followed by oxidative cleavage⁹ and subsequent Fischer esterification (eq 8). Alkylation and functional group manipulation provided enyne 51.

Cycloisomerizations

6,6- and 7,6- Bicyclic Structures

We began by studying the cycloisomerization of enyne 1 using our standard conditions for ruthenium-catalyzed alkene–alkyne couplings (5 mol % CpRu(MeCN)₃PF₆ in acetone at ambient temperature). The reaction afforded bicycle 2 in 90% yield (Table 1, entry 1). Remarkably, the transformation was completely diastereoselective for the trans-fused ring system.

Table 1.

6,6- and 7,6-Bicycles via Ru-Catalyzed Cycloisomerizations

entry	substrate	product	yield, time (%, h)a	drb	entry	substrate	product	yield, time (%, h)a	drb
1			90, 3	>19 : 1	6			70, 3d,f	>19 : 1
			82, 3c	>19 : 1
2			88, 2	>19 : 1	7			70, 6e	>19 : 1
3			95, 3	>19 : 1	8			99, 3	>19 : 1
4			86, 6d	>19 : 1	9			87, 5e	>19 : 1
5			79, 3	>19 : 1	10			99, 4	>19 : 1

^aIsolated yield.

^bDetermined by ¹H NMR.

^cUsing 3 mol % catalyst.

^dUsing 10 mol % catalyst.

^eUsing 20 mol % catalyst.

^fTemp: 40 °C.

This result is particularly noteworthy because more classical ways of making decalins, such as cycloalkylations, generally lead to cis-fused bicycles.¹⁰ The more commonly employed protocols for the synthesis of trans decalins are the Birch reduction of Wieland-Miescher ketones, or intramolecular enolate alkylation to generate the cis-decalin core, which, in a second step, is equilibrated to the trans isomer.¹¹ Both of these methods are limited to decalins containing an epimerizable ring junctions. Additionally, they are indirect, requiring cyclization to an intermediate bicycle that is then transformed to the desired trans isomer.

We were pleased to find that a variety of functionality at the quaternary carbon, including aldehydes, amides, and carboxylic acids, proved amenable to the reaction (Table 1, entries 2–5). Primary alcohol 35 was slightly less reactive under the standard conditions, but was cycloisomerized to bicycle 60 in 86% yield by increasing the catalyst loading and reaction time (entry 4).

The functionality at the alkyne terminus was then examined. Substrates in which the alkyne is substituted with either a trimethylsilyl group (11) or a proton (14) failed to undergo the desired transformation (Figure 1). Enynes 52 and 55, in which the alkyne is substituted with an amide and a ketone, respectively, underwent the cycloisomerization, albeit with longer reaction times than alkynoate 1. Gratifyingly, in the case of ynamide 52, this could be overcome by increasing the temperature to 40 °C, which provided bicycle 53 in 70% yield (Table 1, entry 6) as a single diastereomer. While significant decomposition was observed when the reaction with ynone 55 was heated, complete conversion was obtained at room temperature when the catalyst loading was increased to 20 mol % (entry 7).

Figure 1.

Substrates unreactive to ruthenium catalysis.

The cycloheptene-based enynes also underwent the cycloisomerization in excellent yield and with complete diastereoselectivity (Table 1, entries 8–10). As was observed for the cyclohexene-containing substrates, the high functional group tolerance of this reaction allowed for the formation of bicyclic products containing esters (59, 99% yield), aldehydes (62, 87% yield), and amides (65, 99% yield).

The limits of this transformation are illustrated in Figure 1. Silyl ether 37 and carbonate 38 were inert to these reaction conditions, even with higher catalyst loadings and temperatures. Nitrile 66, ynal 67, ynoic acid 68 and cyclooctene 51 also did not react when treated with the ruthenium catalyst, allowing for the quantitative recovery of the starting enynes.

To investigate the potential differences between products arising from these ruthenium-catalyzed reaction conditions and our previously-reported “ligandless” palladium-catalyzed cycloisomerization,¹² ester 1 was examined. Highest conversions were obtained when HCO₂H was used in combination with DCE as the solvent. Variations to this system (e.g., AcOH in place of HCO₂H or benzene instead of DCE as the solvent) led to diminished reactivity.

In our initial studies, concomitant formation of olefin isomers prevented us from obtaining the desired 1,4-diene in acceptable yields. Gratifyingly, the addition of 2 vol % acetonitrile prevented the formation of these byproducts (Scheme 3).¹³ These optimized conditions (4 mol % Pd₂dba₃·CHCl₃, 2 equiv HCO₂H, DCE, 2 vol % acetonitrile, 40 °C, Scheme 3), were applied to all of our subsequent palladium-catalyzed cycloisomerizations.

Scheme 3.

Optimized Pd-Catalyzed Cycloisomerization

The cyclic enynes used to illustrate the scope of the ruthenium-catalyzed reaction (Table 1) were also cycloisomerized under palladium catalysis. The corresponding 6,6- and 7,6-cis-fused bicycles were obtained in high yields and, in contrast to the ruthenium-catalyzed process, in this palladium-catalyzed reaction they were obtained with complete diastereoselectivity for the cis isomers (Table 2).

Table 2.

6,6- and 7,6-Bicycles via Pd-Catalyzed Cycloisomerizations

entry	substrate	Product	yield time (%, h)a	drb	entry	substrate	product	yield, time (%, h)a	drb
1			92, 4	>19 : 1	6		–	NR	–
2			73, 5	>19 : 1	7			67, 12	>19 : 1
3		–	NR	–	8			95, 3	>19 : 1
4			83, 2.5	>19 : 1	9			92, 4	>19 : 1
5			68, 15	>19 : 1	10			95, 2.5	>19 : 1

^aIsolated yield.

^bDetermined by ¹H NMR.

A variety of functional groups are tolerated by the palladium catalyst, with the exception of acid 57 (entry 3) and ynamide 52 (entry 6). Aldehydes (entries 2 and 9), amides (entries 5 and 10) and ynones (entry 7) underwent the cycloisomerization in excellent yields and diastereoselectivity. In contrast to the slow reactivity observed under ruthenium catalysis, under palladium catalysis alcohol 35 underwent facile cyclization to generate bicycle cis-73 in less than three hours (entry 4).

The relative stereochemistry of the products was determined by examination of key coupling constants and NOE enhancements (Figures 2 and 3). In all cases, the double-bond geometry was secured by observing NOEs between H^a and H^b. COSY spectra were used to correlate chemical shift with connectivity. The differences between the NOEs observed for trans-2 and those observed for cis-69 constitute strong evidence for our stereochemical assignment.

Figure 2.

Stereochemical assignment of trans-2

Figure 3.

Representative stereochemical assignments of cis-bicycles

For the trans-fused series, a strong NOE enhancement was measured between the doubly allylic proton (H^c) and the other axial protons (illustrated for trans-69 in Figure 2). In contrast, strong NOE enhancements were observed between H^c and the vinyl protons (H^a and H^b) for the cis-fused bicycles. Representative examples are illustrated for cis-69 and cis-74 (Figure 3). In addition to H^b–H^c enhancements, the strong NOE enhancement between the aldehydic proton H^d and H^c of cis-74 are highly diagnostic and further confirm our assignments.

6,5-Bicyclic System

Decreasing the tether length by one carbon allowed us to examine the formation of five-membered rings from cyclohexene and cycloheptene-containing enynes, which would produce 6,5- and 7,5-bicycles, respectively.

The formation of 6,5-bicycles via ruthenium-catalyzed cycloisomerization was probed using enyne 16. In the event, 6,5-bicycle 77 was isolated in 86% yield as a single diastereomer (eq 9). Notably, ¹H NMR studies revealed the product contained a cis ring fusion (Figure 4). Unfortunately, attempts to form analogous 7,5-bicycles were not successful.¹⁴

Figure 4.

Stereochemical NOE analysis of 51

(9)

6,7-System

Attempts to form seven-membered rings via ruthenium-catalyzed cycloisomerization revealed a limit of this catalytic system. While small amounts of 6,7-bicycle 79 could be observed, cycloisomerization did not occur catalytically, even under forcing conditions (Table 3). After increasing both the temperature to 60 °C and the catalyst loading to 30 mol %, at the end of 24 hours the reaction had only reached 33% conversion (entry 4), representing a reaction with minimal catalyst turnover.

Table 3.

Behavior of 6,7-System

entry	conditions	catalyst loading (mol %)	resultsa
1	2.5 h, 23 °C	5	no reaction
2	3 h, 45°C	5	17% conv.
3	6 h, 45°C	20	25% conv.
4	24 h, 60 °C	30	33% conv.

^aDetermined by ¹H NMR of crude reaction.

5,n-Bicycles

The excellent yields and diastereoselectivities with which the 6,6- and 7,6-ring systems were obtained led us to investigate the formation of 5,n-ring systems. Toward this end, enynes 24 and 80 were subjected to the ruthenium-catalyzed cycloisomerization reaction conditions. Ester 24 underwent smooth cyclization to the 5,5-bicyclic product (81), with complete selectivity for the cis-diastereomer (Table 4, entry 1). Enyne 80, containing a silyl ether at the quaternary carbon, also underwent cyclization to generate the cis-fused product (82) under ruthenium catalysis (entry 2). The relative stereochemistry of the bicyclic products was secured by an analysis of the COSY spectrum along with the NOE enhancements illustrated in Figure 5.

Table 4.

Ru-Catalyzed Cycloisomerizations to Form 5,5-Bicycles

entry	substrate	product	conversion (%)a	yield (%)b	dra
1			100	81	>19:1
2			100	85	>19:1

^aDetermined by ¹H NMR;

^bIsolated yield.

Figure 5.

Representative stereochemical assignment

Extending the alkyne tether by one carbon in order to form the 5,6-bicycle had a profound impact on the reaction. When substrate 30 was submitted to the optimized palladium-catalyzed reaction conditions, the expected bicycle 83 was formed in excellent yield as a single diastereomer (Scheme 4). To our surprise, when 30 was submitted to the ruthenium-catalyzed reaction conditions, bicycle 83 was formed in equal amounts along with tricycle 84. Both products were formed as single diastereomers.

Scheme 4.

Novel Tricyclic Product Obtained Under Ruthenium Catalysis

A similar [2+2] cycloaddition with iodoenynes was recently reported by Fürstner, in which the penta-methyl-cyclopentene (Cp*) analog of our catalyst, Cp*Ru(MeCN)₃PF₆, was employed.¹⁵ When enyne 30 was submitted to the conditions reported by Fürstner (catalytic Cp*Ru(MeCN)₃PF₆ in DMF at room temperature), no reaction was observed.

The nickel-catalyzed [2+2] cycloaddition of 1,2-diphenylethyne and 2,5-norbornadiene (NBD) reported by Schrauzer in 1964¹⁶ was the first transition metal-catalyzed alkyne–olefin [2+2] addition. Since Schrauzer’s initial report, palladium,¹⁷ platinum,¹⁸ cobalt,¹⁹ gold,²⁰ rhodium,²¹ rhenium,²² ruthenium,²³ iridium,²⁴ and molybdenum²⁵ have all been shown to catalyze the cycloaddition of an alkyne and an olefin. However, despite the array of metal complexes that have been employed for this transformation, this reactivity is unprecedented for the CpRu(MeCN)₃PF₆ catalyst.

Under the standard conditions (5 mol % CpRu(MeCN)₃PF₆, acetone, 23 °C, 2 h), a 1:1 mixture of bicycle and tricycle was obtained in high yields, for all of the substrates examined (Table 5).

Table 5.

Cycloisomerizations Under Standard Conditions

entry	R1	R2	conversion (%)	ratio tricycle/ bicyclea	yield (tricycle, bicycle, %)b
1	CO2Me	CO2Me	100	~ 1:1	89b
2	CH2OH	CO2Me	100	~ 1:1	43, 53
3	CH2OTBS	CO2Me	100	~ 1:1	46, 42
4	CH2OAc	CO2Me	100	~ 1:1	30, 38
5	CHO	CO2Me	100	~ 1:1	49, 50
6	CH2OTBDPS	CHO	100	~ 1:1	98c
7	CH2OBn	CHO	100	~ 1:1	90c

^aDetermined by ¹H NMR of crude reaction.

^bIsolated yield.

^cIsolated as a mixture of tricycle and bicycle.

It is believed that cyclobutene 84 as drawn is the product characterized. Correlations between chemical shift and connectivity were deduced by COSY analysis. Proton NMR studies show an NOE enhancement between H^a and H^b, as would be expected for two cis protons (Scheme 5). An NOE enhancement is also observed between the methylene proton H^c and H^b. Importantly, the chemical shifts of these diagnostic cyclobutene protons correlate with the values reported for similar tricyclic cyclobutenes formed under platinum catalysis.²⁶ Moreover, if the ester was unsaturated in the other direction of the cyclobutene, that compound (85) would contain a highly strained olefin.²⁷

Scheme 5.

Given that we could access the bicycle exclusively via palladium catalysis, we then sought reaction conditions to provide the tricycle exclusively under ruthenium catalysis (Scheme 6). We began our optimization studies with enyne 30.

Scheme 6.

Desired Catalyst-Controlled Product Selectivity

Reasoning that the introduction of another coordinating solvent with different ligating properties might affect the selectivity of the reaction, a variety of solvents were examined. Unfortunately, none of the additives (acetic acid, dioxane, ethyl acetate, methanol) increased the selectivity for either product (Table 6). Toluene (Table 6, entry 4) and DMSO (Table 6, entry 6) completely shut down the reaction, while the introduction of acetonitrile decreased the reaction conversion to 44%. The addition of nitromethane proved to be differential, providing tricycle 84 in a slight excess relative to bicycle 83 (Table 6, entry 10).

Table 6.

Solvent Screen

entry	solvent additive	conversion (%)a	ratio tricycle/bicyclea
1	2.5 vol % AcOH	100	1:1
2	20 vol % AcOH	100	1:1
3	20 vol % dioxane	100	1:1
4	20 vol % toluene	NR	—
5	20 vol % EtOAc	100	1:1
6	20 vol % DMSO	NR	—
7	20 vol % MeCN	44	1:1
8	2.5 vol % MeOH	100	1:1
9	50 vol % CH2Cl2	100	1:1
10	20 vol% MeNO2	100	1.2:1

^aDetermined by ¹H NMR of crude reaction.

This promising result led us to examine DMF, a solvent that we have previously shown is compatible for reactions with CpRu(MeCN)₃PF₆. In this case, when DMF was employed as a solvent the reaction did not proceed at all after 3 hours at 23 °C. The addition of 2 vol % DMF to an acetone solution of substrate and catalyst (Table 7, entry 1) slowed the reaction, but cyclobutene formation occurred in more than twofold excess relative to bicycle formation. Further increasing the amount of DMF relative to acetone led to a moderate increase in selectivity (Table 7, entry 2), but use of greater than 20 vol % DMF decreased the conversion significantly (Table 7, entry 3).

Table 7.

Nitrogen-Based Solvent Screen

Entry	solvent additive	time (h)	conversion (%)d	ratio tricycle/bicycled
1	2 vol% DMF	15	100	2.4:1
2	20 vol% DMF	13	80	3:1
3	50 vol% DMF	13	67	3:1
4	5 vol% NMP	4	100	3.6:1
5a	10 vol% NMP	3	100	5:1
6a	20 vol% NMP	3	100	3:1
7	50 vol% DMA	15	48	10:1
8	80 vol% DMA	18	63	14.6:1
9	100 vol% DMA	18	74	15.2:1
10b,c	100 vol% DMA	12	93	12.5:1
11b	20 vol% NMP, 80 vol% DMA	16	78	7:1
12b	10 vol% NMP, 80 vol% DMA	12	40	10.3:1

^aUsing 20 mol % catalyst.

^bUsing 15 mol % catalyst.

^cReaction conducted at 40 °C.

^dDetermined by ¹H NMR of crude reaction

Hoping that an alternative nitrogen-based solvent might similarly favor formation of cyclobutene 84, N-methylpyrrolidone (NMP), nitromethane, and N,N-dimethylacetamide (DMA) were tested as additives (Table 7, entries 4–12). The use of NMP resulted in faster reaction rates, but with modest product selectivity (entries 5 & 6). The highest product selectivity was obtained with 100 vol % DMA, at room temperature (entry 9), however, the reaction stopped at 74% conversion. Raising the temperature to 40 °C increased the conversion to 93% while only slightly decreasing the selectivity (entry 10). These conditions (15 mol % CpRu(MeCN)₃PF₆, DMA, 40 °C), the best compromise between conversion and selectivity, were then adopted as the optimized conditions for tricycle formation.

Using these optimized conditions, a variety of substrates were examined (Table 8). Free alcohols (Table 8, entry 2) and aldehydes (Table 8, entry 3) are tolerated, generating the corresponding cyclobutenes in excellent yields. In contrast to the cycloisomerization reaction to form 6,6- and 7,6- bicycles, this system tolerates a wider variety of functional groups, such as silyl ethers (Table 8, entries 4, 6, and 7) and ynals (Table 8, entries 4, 5, and 8).

Table 8.

Cycloisomerizations Under Tricycle-Optimized Conditions

entry	substrate	product	yield(%)a	tricycle/ bicycleb	entry	substrate	product	yield(%)a	tricycle/ bicycleb
1			91	12.5:1	5			81c	8.3:1
2			71	2.7:1	6			91	10.4
3			89	14.2:1	7			91	10.7:1
4			61	>19:1	8			66	10.8:1

^aYield of isolated tricycle.

^bDetermined by ¹H NMR.

^cIsolated as an inseparable, 8.3:1 mixture of tricycle and bicycle.

Complementary product selectivity can be obtained simply by changing the catalyst from ruthenium to palladium. To demonstrate this, the substrates were also subjected to the palladium-catalyzed cycloisomerization conditions. We were pleased to find that the reaction produced the desired cis-fused bicycles in good to excellent yields with complete diastereoselectivity (Table 9).

Table 9.

5,6-Cis-Fused Bicycles via Pd-Catalyzed Cycloisomerizations

entry	substrate	product	yield(%)a	dra	entry	substrate	product	yield(%)a	drb
1			80	>19:1	5			47c	>19:1
2			97	>19:1	6			91	>19:1
3			81	>19:1	7			99	>19:1
4			57	>19:1	8			80	>19:1

^aIsolated yield.

^bDetermined by ¹H NMR.

^cIsolated along with 23% of olefin-isomerized 1,5-diene.

Carbonate 111, silyl ether 112, and alcohol 113 (Figure 6) were unreactive as well. In separate experiments, increasing the temperature of the standard reactions (in acetone) to 50 °C did not improve the conversion.

Figure 6.

Unproductive enynes with electron-deficient alkynes.

Not all of the substrates examined cyclized successfully under ruthenium catalysis. Curiously, Weinreb amide 43, the cyclopentene analog of 63 and 44, was not reactive. Under the standard reaction conditions (acetone, 23 °C), the reaction with amide 43 failed to go to complete conversion after 3.5 hours (standard reaction time 2 hours). Additionally, under these standard conditions, amide 43 was the only substrate to give an unequal mixture of bi- and tricyclic products, producing the two in a 1:4 ratio, respectively (Table 10, entry 1). In DMA, conversion was only 42% after 11 hours at 40 °C (entry 2), and after 24 hours only 50% (entry 3). Repeated attempts to separate the tricycle from the starting material were not successful, and the isolation of tricycle 109 was ultimately abandoned.

Table 10.

Behavior of Weinreb Amide

entry	conditions	time (h)	conversion (%)a	ratio 109/110
1	acetone, 23 °C	3.5	71	4 : 1
2	DMA, 40 °C	11	42	8 : 1
3	DMA, 40 °C	24	50	ndb

^aDetermined by 1H NMR.

^bNot determined.

As had been observed previously, enynes containing an alkyne terminated by a group other than an electron-withdrawing group were completely unreactive (Figure 7). Ynamide 52 and ynone 55 (Table 1, entries 6 and 7), used as substrates for generating the 6,6-bicyclic products, required more forcing conditions. We reasoned that if the (unknown) factors that slowed the cycloisomerization of the these enynes were present in the 5,6-system, a change in product ratio might result, favoring the cyclobutene product. Instead, reactions with 5,6-ynone 114 and ynamide 115 were even more sluggish than their 6,6-analogs (Scheme 7).

Figure 7.

Unreactive substrates with electron-rich alkynes.

Scheme 7.

Relatively slower reactivity of 115 and 114.

^a Using 10 mol % CpRu(MeCN)₃PF₆ in acetone at 40 °C. ^b Reaction time was 6 h using 20 mol % CpRu(MeCN)₃PF₆ in acetone at 23 °C.

To determine if these optimized conditions (15 mol % CpRu(MeCN)₃PF₆, DMA, 40 °C) were unique to 5,6-congeners, substrates from the 6,6- and 6,7- systems (esters 52 and 31) were submitted to the reaction under the same conditions. No reaction was observed after 12 hours (eq 10), indicating that the formation of tricyclic products is indeed unique to the 5,6-system.

(10)

Mechanistic Considerations: Palladium Catalysis

The cis-selectivity observed in the palladium-catalyzed cycloisomerization can be explained by a mechanism in which the metal is directed to the same side of the ring as the alkyne (Scheme 6). The active catalyst is generated by oxidative insertion of palladium(0) into HCO₂H to generate a palladium(II) species, that then hydrometallates the alkyne to form vinyl-palladium species 117. Then, intramolecular binding and insertion of the olefin provides intermediate 118. Subsequent β-hydride elimination produces the 1,4-diene product, which is still bound to the palladium through η²-coordination of the newly formed olefin (119). The concomitant formation of olefin isomers observed in our initial studies are hypothesized as arising from palladium hydride insertion–elimination pathways at this step. The introduction of acetonitrile facilitates decomplexation of the product via ligand exchange (119→69).

Mechanistic Considerations: Ruthenium Catalysis

Electronic requirements

The sum of our studies on this catalyst system –both these current studies, and the studies of non-cyclic enynes²⁸ previously reported– suggest that there are two operative mechanisms in enyne cycloisomerizations catalyzed by CpRu(MeCN)₃PF₆: C–H insertion and metallacycle formation. The operable, productive mechanism for a given enyne is dictated by its olefin geometry. The data suggest that for enynes containing (E)-olefins, cycloisomerization proceeds via a ruthenacycle, while enynes containing (Z)-olefins undergo an allylic C–H insertion. The two reaction manifolds are illustrated in Scheme 9. In path A, initial binding is followed by reversible metallacycle formation (120a → 121a). β-hydride elimination from the pseudo-equatorial substituent generates a vinyl ruthenium-hydride (122) that undergoes reductive elimination, yielding the cyclized product (123) and regenerating the catalyst.

Scheme 9.

Possible mechanistic manifolds

Evidence that β-hydride elimination occurs toward the trans-substituent (CH₂R¹ in metallacycle 121a, corresponding to the pseudo-equatorial position) was provided by the cycloisomerizations of geranyl-derived enyne 128 and neryl-derived enyne 129 (eq 11). β-Hydride elimination occurred regioselectively from the (E)-substituent in to provide the corresponding cyclopentanones 130 and 131. The product ratios indicate that while elimination toward the cis-substituent is possible, it is the higher-energy pathway.

In the case of (Z)-olefins (120b, Scheme 9), the unfavorable orbital overlap in metallacycle 121b between the metal and H^b, the hydrogen required for β-elimination from the intermediate ruthenacyclopentene, prevent the reaction from proceeding forward along this pathway. (Z)-Olefins thus react according to an alternative mechanism (path B). In this case, initial allylic C–H activation provides ruthenium-allyl species 124 that undergoes an intramolecular 1,4-addition to the alkyne, forming vinyl ruthenium 125. Reductive elimination affords diene 127 and regenerates the active catalyst (126).

(11)

Evidence for this dichotomy was provided by deuterium-labeling studies, wherein a substrate (132) sterically prohibited from undergoing metallacycle formation underwent allylic C–H insertion exclusively into the cis-methyl substituent (eq 12).

(12)

Lastly, careful examination of the dependence of reactivity on the alkyne substitution pattern of the following enynes reveals an interesting pattern. The series of enynes containing (Z)-olefins that were examined in our previous study and the yields of the corresponding diene products are depicted in Figure 8. The series of enynes, also containing (Z)-olefins, examined as part of this current study are illustrated in equation 13. The difference in reactivity between terminal alkyne 134a and TMS alkyne 134c are intriguing. Specifically, the failure of cyclohexenyl 14 which, like enyne 134a, is also a 1,6-enyne containing a terminal alkyne and a cis-olefin, is difficult to explain. If the metallacycle derived from terminal alkyne 134a is somehow able to achieve the requisite metal–β-hydride orbital overlap at higher temperatures, it is odd that neither silyl-enyne 134c nor cyclic 14 do.

Figure 8.

Reactivity of (Z)-olefin-containing enynes

One possible explanation for these observations is that the terminal alkyne in 134a is somehow facilitating an in situ olefin isomerization, to the corresponding trans-olefin, which then cycloisomerizes via the metallacycle pathway. If the terminal alkyne, via vinylidene formation, for instance, aids in this geometrical isomerization process, it would explain inability the silyl alkyne to undergo cyclization. Additionally, it would explain why the cyclic olefin-containing enynes, whose olefins are geometrically constrained, do not react.

As noted previously, silyl alkyne 11 and terminal alkyne 14 were completely unreactive (eq 13) while analogous substrate 1, featuring a methyl ester at the alkyne terminus, afforded the desired 1,4-diene in excellent yield as a single diastereomer (eq 13 and Table 1, entry 1). The reactivities of both these cyclic substrates and of linear enynes 134a–c can be rationalized by a C–H insertion mechanism (path B, Scheme 9), wherein the nature of the carboruthenation step (124 → 125) is akin to the conjugate addition of a metal–alkyl species to an unsaturated carbonyl. The reaction cannot occur, therefore, if the alkyne is not a 1,4-acceptor (i.e., terminated by an electron-withdrawing group).²⁹

(13)

The failure enynes 111 and 112 (Figure 6) to react is likely due to the steric demand of the functional groups (a tertiary carbonate in 111 and a tertiary ether in 112) at the quaternary center in both enynes. Van der Waals interactions between the tertiary ether and one of the methylene groups in the side chain might prevent the pendant alkyne from achieving the orientation required for orbital overlap that would lead to further reaction.

trans-Diastereoselectivity (6,6- and 7,6-Bicycles)

Although the aforementioned facets of this chemistry explain reactivity they do not explain diastereoselectivity. We had anticipated generation of the cis-fused decalin system, expecting that a coordinated alkyne would direct allylic C–H activation. However, carbon–carbon bond formation occurs syn to the ester, which implies that the C–H insertion must also occur on the same face. It appears, therefore, that the carbonyl moiety is acting to direct the formation of the intermediate allyl ruthenium species (Scheme 10). Carbonyl-directed η³-allyl formation (135) is followed by ligand exchange to form alkynyl-coordinated 136, in which the metal is bound to the carbocycle in an η¹ manner; subsequent carboruthenation and reductive elimination produce the observed diastereomer (trans-2).

Scheme 10.

Source of trans-Diastereoselectivity

Further insights into our understanding of this reaction were provided by other substrates. Silyl ether 37 (see Figure 1), bearing a non-coordinating moiety at the quaternary carbon, was completely unreactive to these enyne cycloisomerizations, even with higher catalyst loadings and at elevated temperatures. Nitrile 66 was also unreactive, suggesting that C–H insertion to form the metal allyl species was not solely due to an inductive effect of the key functional group, but a specific Lewis basicity. The lack of reactivity observed for carbonate 38 is also understood in terms of this proposed mechanism. While the carbonate group possesses the requisite Lewis basicity, the position of the carbonyl oxygen in carbonate 38 is two atoms further away from the cyclohexyl ring than it is ester 1. As a result, the coordination leading to ruthenacycle 136 (Scheme 10) involves a ring of six-to-seven members, whereas the corresponding transition state derived from carbonate 38 would require the formation of an eight-to-nine-membered ring, which should be significantly less facile. Additionally, cyclooctene 51 was found to be inert to these reaction conditions, possibly due to the conformational rigidity often seen in cyclooctyl systems.³⁰

cis-Diastereoselectivity (6,5-, 5,5-, and 5,6-Bicycles)

The cis-bicyclic products formed under ruthenium catalysis can derive from one of the two mechanisms illustrated in Scheme 9: a metallacycle pathway (path A), where the intermediate 121 undergoes a β-hydride elimination of H^a prior to the final reductive elimination, or a C–H insertion pathway (path B), via 124 and 125. The cis selectivity observed for the formation of 6,5-, 5,5- and 5,6-bicycles could be explained by a catalytic cycle which proceeds through path A. As discussed, the body of evidence gathered up to this point suggests that for cis-olefinic enynes, the high energy barrier associated with the β-elimination of the pseudo-equatorial H^b is prohibitively high. Thus, while a mechanism which proceeds via a ruthenacycle intermediate is possible and cannot be ruled out at this time, rationales within the C–H insertion manifold have been primarily considered.

It is possible that increased strain present in the 1,2-disubstituted five-membered ring (138, Scheme 11), as compared to the six-membered cyclohexyl intermediate 136, causes the carboruthenation to occur on the same side of the ring as the alkyne, yielding the observed cis-fused systems.

Scheme 11.

Mechanistic Rationale for cis-Diastereoselectivity of Cyclopentenyl Enynesa,b,c

^a E = CO₂Me. ^b n = 0,1. ^c [Ru] = RuL_n, ligands removed for clarity.

Alternatively, the 1,2-disubstituted intermediate 138 could form, but the higher level of strain³¹ involved in forming the trans-fused 5,6-, 6,5-, and 5,5-ring systems might force the reaction through a minor intermediate that ultimately leads to the observed cis-fused bicycle (Scheme 12). By rendering the carbonyl-directed insertion step (139→138) non-productive, the reaction is forced to proceed via allyl-ruthenium intermediate 140, in which the carbonyl group is not bound. In this scenario, strongly-coordinating groups might be expected to retard the reaction rate, as they would cause the equilibrium to lie heavily toward formation of 138. This is consistent with the behavior of 5,6-enyne 43, containing a strongly-ligating Weinreb amide. Enyne 43 did not undergo complete conversion (see Table 10), even under extended reaction times and increased temperatures.

Scheme 12.

Alternative Rationale for cis-Diastereoselectivity of 5,n-Systemsa,b,c

^a E = CO₂Me. ^b m = 1,2; n = 0,1. ^c [Ru] = RuL_n, ligands removed for clarity.

Lastly, it is possible that the cis-fused 6,5-, 5,6- and 5,5-bicyclic products are formed in the same manner that the tricyclic products are– via metallacycle formation (path A, Scheme 13). In this case, the cis-diastereomer would necessarily form. While it does not account for the requirement that the alkynes be terminated by an electron-withdrawing group, it would explain why the coordinating group is no longer required.

Scheme 13.

Proposed Mechanistic Pathways for 5,6-Enyes

^aE = CO₂Me.

Product selectivity under ruthenium catalysis: bi- versus tricycle formation

Formation of tricycles is thought to occur via a metallacycle mechanism (path A, Scheme 13). It is likely that formation of the tricyclic product is seen only in the 5,6-system and not the 5,5-system because formation of the 5,5,4-tricycle is too high in energy. Interestingly, calculations³² of the relative energy differences between the cis-bicyclic products and the analogous tricycles revealed that the difference is smallest for 5,6-system. The difference in energy between 5,6-bicycle 83 and tricycle 84 is calculated to be 11.2 kcal. In comparison, the tricycle that would be produced from a cycloaddition of 16 is 30.1 kcal higher in energy than the observed bicycle, cis-77. The tricycle that would derive from 1 is 18.4 kcal higher in energy than cis-2, and 9.74 kcal higher in energy than the observed trans-2.

The precise role that the DMA plays in determining the product selectivity deserves comment. It is possible that DMA, a stronger ligand than acetone, occupies the coordination site otherwise used for β-hydride elimination, thereby facilitating the C–C reductive elimination step (145 → 84) of the cycloaddition (path A, Scheme 13). An alternative hypothesis is that the open coordination site required for the initial C–H insertion step of path B is occupied by a molecule of solvent. That is, the bound molecule of solvent outcompetes the agostic C–H bond as a ligand.

This hypothesis, that DMA is effective because its occupation of a coordination site on the metal inhibits one of the pathways, implies that the pathways leading to tricyclic and bicyclic products have different coordination requirements. The oxidative metallacycle formation (enyne→145, path A) requires that two of ruthenium’s six coordination sites are unoccupied (Scheme 14). The initial coordination and insertion step (enyne→143) of path B must then require more than two open sites if it is inhibited by DMA. Indeed, species 146, in which the enyne occupies three coordination sites on the metal is a likely intermediate. As depicted, substrate–metal interactions occur through the alkyne, the η²-bound olefin, and the agostic C–H bond.

Scheme 14.

Coordination Requirements

This general hypothesis –that the relatively-stronger ligating properties of DMA shunt the reaction through the metallacycle pathway– is in agreement with the selectivity observed for Weinreb amide 43, a substrate with a strong ligating group. While it failed to undergo the reaction in full conversion, enyne 43 was the only substrate that displayed any inherent product selectivity in acetone; at 71% conversion the ratio of tricyclic to bicyclic products was 4:1 (Table 10, entry 1). The reduced reactivity of 43 would still be explained by unproductive ligation of the amide moiety to the catalyst. The role of DMA could take place while it is bound to the metal center, or simply present at a non-bonding distance around the metal. If its contribution requires the ability to alternate between the two, the presence of a strong coordinating group bound so close to the reaction site could hinder the reactivity.

Finally, a unifying mechanism in which the 6,5-, 5,6-, and 5,5-bicyclic products as well as the tricyclic products are all formed from the same metallacycle intermediate (145, path A of Scheme 13) cannot be ruled out. If both products derive from ruthenacyclopentene 145, the subsequent step –β-hydride elimination or reductive elimination– is product-determining. Exclusive formation of the tricyclic product by the 5,6-system is then understood in terms of the conformational differences between metallacycle 145 and the metallacycle formed from the cyclohexene-based substrates (e.g., 121b, Scheme 9). As a result of the larger H–C–H bond angles in cyclopentane rings compared to those in cyclohexane rings, the β-hydrogen (H^b) and the metal center are placed further apart in intermediate 145 (Scheme 13) than in metallacycle 121b. As a consequence of the greater distance between the metal and H^b in 145, β-hydride elimination would be kinetically disfavored, in turn allowing for reductive elimination to become a competitive process.

Conclusion

In conclusion, we have developed complementary methods for the transition metal-catalyzed enyne cycloisomerizations of cyclic olefins. By using distinct ruthenium and palladium catalysts, decalins and 7,6-bicycles can be obtained with dichotomous stereochemical outcomes. The change in mechanism that accompanies the change in metal affords trans-fused 1,4-dienes with ruthenium and their cis-fused diastereomers under palladium catalysis. In the reactions under ruthenium catalysis, a coordinating group is required, and acts to direct the metal to the same side of the carbocycle, resulting in the observed trans diastereoselectivity.

Subtle changes in the carbocyclic substrate led to the discovery of a heretofore-unobserved mechanistic pathway, providing bicyclic cycloisomerization products under palladium catalysis, tricyclic products under ruthenium catalysis in DMA, and a 1:1 mixture of the two under ruthenium catalysis in acetone. The different coordination requirements of the two paths allow for the reaction to be shuttled through the metallacycle pathway (generating tricyclic products) when DMA is used as a solvent. This method of obtaining these cyclobutene products complements the existing methods for catalyzing the formal [2+2] cycloaddition of alkynes. While other methods have been demonstrated for larger ring sizes with palladium, and for specific alkyne termini with platinum and gold, this method is unique in its substrate scope. To the best of our knowledge, this is the first example of a formal [2+2] cycloaddition catalyzed by CpRu(MeCN)₃PF₆.

Current efforts are directed at gaining a deeper understanding of some of the fundamental principles dictating the reactivity of these systems, as well as extending their utility by expanding the substrate scope.

Experimental Section

General procedure for the ruthenium-catalyzed cycloisomerizations in acetone

To a solution of the enyne substrate in acetone (0.1 M) at 23 °C under argon was added CpRu(MeCN)₃PF₆ (3–20 mol %, in general 5 mol %, all at once). The resulting mixture was stirred at room temperature for the time provided, at which point the solvent was removed via rotary evaporation. The residue was diluted in the minimum amount of CH₂Cl₂/Et₂O (1:1) and passed through a small plug of silica gel (Et₂O eluent). The filtrate was concentrated in vacuo to provide the cyclized adducts, which were further purified by flash chromatography.

Synthesis of bicycle 54

According to the general procedure, 20.0 mg 36 (0.0854 mmol) and 1.9 mg CpRu(CH₃CN)₃PF₆ (0.00427 mmol) were stirred in 854 µl acetone for 2 h. Purification of the residue by flash chromatography (11:1 hexanes/EtOAc eluent) afforded bicycle 54 (17.5 mg, 88% yield, R_F = 0.45 in 4:1 hexanes/EtOAc) as a colorless oil. Bicycle 54: ¹H NMR (400 MHz, C₆D₆) δ 9.58 (s, 1H), 5.84 (s, 1H), 5.56 (d, J = 10.4 Hz, 1H), 5.48−5.42 (m, 1H), 4.26 (br d, J = 12.7 Hz, 1H), 3.37 (s, 3H), 2.46 (br s, 1H), 1.77 (br d, J = 13.0 Hz, 1H), 1.73−1.54 (comp m, 3H), 1.50 (dt, J = 5.5, 13.1 Hz, 1H), 1.07−0.98 (m, 1H), 0.78 (dt, J = 5.5, 13.0 Hz, 1H); ¹³C NMR (100 MHz, C₆D₆) δ 206.0, 166.4, 160.6, 130.0, 124.9, 113.3, 51.3, 50.5, 47.9, 35.2, 33.0, 30.4, 24.7, 22.6; IR (film) 2934, 1717, 1640, 1193, 1161 cm⁻¹; HRMS (ESI⁺) m/z calc’d for [C₁₄H₁₈O₃ + Na]⁺: 257.1154, found 257.1153.

Prepatory scale example

According to the general procedure, 230 mg 36 (0.981 mmol) and 21.3 mg CpRu(CH₃CN)₃PF₆ (0.0491 mmol) were stirred in 9.81 mL acetone for 2 h. Purification of the residue by flash chromatography (11:1 hexanes/EtOAc eluent) afforded bicycle 54 (189 mg, 82% yield).

General procedure for the ruthenium-catalyzed cycloisomerizations in DMA

To a solution of the enyne substrate in DMA (0.12 M) at 23 °C under argon was added CpRu(MeCN)₃PF₆ (3–20 mol %, in general 15 mol %). The resulting mixture was stirred in a pre-heated oil bath at 40 °C for the time provided (in general 12 h), at which point the reaction was diluted with diethyl ether and washed with water once, then washed once with brine. The aqueous layers were then combined and extracted once more with ether. The organic phases were combined, dried over MgSO₄, and concentrated in vacuo. The residue was further purified by flash chromatography to afford the cyclobutene adduct.

Synthesis of tricycle 101

According to the general procedure, 21.0 mg 100 (0.0741 mmol) and 4.7 mg CpRu(MeCN)₃PF₆ (0.0107 mmol) were stirred in 618 µl DMA for 16 h. Purification of the residue by flash chromatography (9:1 hexanes/EtOAc eluent) afforded tricycle 101 (13.8 mg, 66% yield, R_f = 0.45 in 17:3 hexanes/EtOAc) as a colorless oil. Tricycle 101: ¹H NMR (500 MHz, CDCl₃) δ 9.64 (s, 1H), 7.37−7.27 (comp m, 5H), 4.55−4.49 (comp m, 2H), 3.32−3.30 (br m, 1H), 3.20 (d, J = 9.0 Hz, 1H), 3.19 (d, J = 9.0 Hz, 1H), 2.84 (dm, J = 13.1 Hz, 1H), 2.72 (s, 1H), 2.11−2.05 (m, 1H), 2.00−1.96 (m, 1H), 1.78−1.59 (comp m, 6H), 1.45−1.40 (m, 1H); ¹³C NMR (125 MHz, CDCl₃) δ 185.2, 171.6, 138.7, 136.8, 128.4, 127.6, 127.5, 76.2, 73.4, 50.9, 45.7, 44.6, 29.0, 27.5, 27.4, 25.1, 24.3; IR (film) 2938, 2851, 1665, 1100 cm⁻¹; HRMS (ESI⁺) m/z calc’d for [C₁₉H₂₂O₂ + Na]⁺: 305.1520, found 305.1517.

General procedure for the palladium-catalyzed cycloisomerizations

To a solution of the enyne substrate and formic acid (2 equiv) in 49:1 DCE/MeCN (0.025 M) under argon was added Pd₂dba₃·CHCl₃ (4 mol%). The resulting mixture was heated to 40 °C and stirred for the time provided. Upon completion of the reaction, the solution was cooled to room temperature and partitioned between EtOAc and saturated NaHCO₃. The aqueous phase was extracted with EtOAc, and the combined organic phases were washed with brine, dried over Na₂SO₄, and concentrated in vacuo. Purification of the residue by flash chromatography provided the bicyclic adduct.

Synthesis of bicycle 70

According to the general procedure, 32.0 mg 36 (0.137 mmol), 10.3 µl HCO₂H (0.274 mmol), and 5.6 mg Pd₂dba₃·CHCl₃ (.00548 mmol) were stirred in 5.37 mL DCE and 110 µl CH₃CN for 5 h. Purification of the residue by flash chromatography (3:1 PhH/CHCl₃ eluent) afforded bicycle 70 (23.2 mg, 73% yield, R_F = 0.27 in 9:1 hexanes/EtOAc) as a colorless oil. Bicycle 70: ¹H NMR (500 MHz, CDCl₃) δ 9.44 (s, 1H), 5.83−5.78 (m, 1H), 5.73 (s, 1H), 5.57−5.53 (m, 1H), 3.68 (s, 3H), 3.16 (br s, 1H), 2.86−2.74 (comp m, 2H), 2.14−2.08 (comp m, 2H), 1.76−1.65 (comp m, 2H), 1.63−1.52 (comp m, 4H); ¹³C NMR (100 MHz, CDCl₃) δ 204.6, 167.1, 162.5, 129.0, 127.0, 116.3, 51.2, 51.1, 44.9, 28.2, 27.5, 25.3, 23.0, 22.2; IR (film) 2934, 1719, 1643, 1435, 1155 cm⁻¹; HRMS (ESI+) m/z calc’d for [C₁₄H₁₈O₃ + Na]+: 257.1154, found 257.1148.

Synthesis of bicycle 108

According to the general procedure, 20.0 mg 100 (0.0706 mmol), 5.3 µl HCO₂H (0.141 mmol), and 2.9 mg Pd₂dba₃·CHCl₃ (0.00282 mmol) were stirred in 2.77 mL DCE and 56.5 µl MeCN for 2 h. Purification of the residue by flash chromatography (7:3 benzene/CHCl₃ eluent) afforded bicycle 108 (16.0 mg, 80% yield, R_f = 0.25 in 17:3 hexanes/EtOAc) as a colorless oil. Bicycle 108: ¹H NMR (500 MHz, CDCl₃) δ 10.0 (d, J = 8.9 Hz, 1H), 7.36−7.28 (comp m, 5H), 5.88 (d, J = 8.2 Hz, 1H), 5.82 (app. dt, J = 2.6, 8.1 Hz, 1H), 5.49−5.46 (m, 1H), 4.52 (s, 2H), 3.40 (d, J = 9.0 Hz, 1H), 3.35 (d, J = 9.0 Hz, 1H), 3.26 (s, 1H), 3.11−3.05 (m, 1H), 2.49 (app. dq, J = 2.6, 16.5 Hz, 1H), 2.35−2.28 (m, 1H), 2.13 (app. dq, J = 2.2, 16.5 Hz, 1H), 1.77−1.65 (comp m, 2H), 1.60−1.49 (comp m, 2H); ¹³C NMR (125 MHz, CDCl₃) δ 168.0, 132.2, 128.4, 127.6, 127.5, 127.3, 75.4, 73.3, 57.4, 49.9, 43.7, 30.8, 26.0, 22.2; IR (film) 2929, 2851, 1671, 1101 cm⁻¹; HRMS (ESI⁺) m/z calc’d for [C₁₉H₂₂O₂ + Na]⁺: 305.1512, found 305.1514.

Scheme 2.

Undesired Olefin Isomerization

Scheme 8.

Proposed Pd-Cycloisomerization Catalytic Cycle