|Home | About | Journals | Submit | Contact Us | Français|
PPG Industries, Inc. Pittsburgh, PA 15222.
Jennifer M. Schomaker: 0000-0003-1329-950X
Site- and regiocontrolled Au-catalyzed allene carbocyclizations furnish highly substituted cyclopentenes in >1:1 dr. Significant substitution on the substrate is tolerated, with potential to install five contiguous stereocenters after alkene functionalization. Major challenges include identifying a Au/Cu catalyst that controls both the relative rates of allene epimerization/cyclization and the facial selectivity in addition of a metal enolate to the allene. Experiments to achieve stereodivergent cyclizations and transform key cyclopentenes into useful synthetic building blocks are described.
Methods to rapidly transform simple precursors into complex scaffolds are of continued interest to the synthetic community. The surge of new strategies for the preparation of allenes has stimulated their use as convenient precursors for the introduction of multiple functionalities into the products with high degrees of regio-, chemo-, and stereocontrol.1 In this context, we have recently described chemoselective, Rh(II)-catalyzed carbene insertions into allylic C–H bonds of 1.1 to yield allenes 1.2,2 which are well-suited for subsequent carbocyclization to furnish cyclopentenes of the form 1.3 (Scheme 1A). However, achieving diastereocontrolled reactions of 1.2 to 1.3 requires a catalyst that epimerizes the allene of 1.2 before C–C bond formation occurs. Once the stereochemistry of the strained bicycle 1.3 is established, transformations of the remaining alkene are expected to occur with high dr to deliver highly substituted cyclopentanes present in a host of bioactive molecules, including the antiviral peramivir,3 the neuroprotective compound biloba-lide,4 and the Chitinase inhibitor allosamidin5 (Scheme 1B).
Examples of Coniaene-like C–C bond formations that occur via intramolecular addition of a carbon nucleophile to a site of unsaturation are known using both Pd6 and Au catalysts.7 However, these approaches typically use prefunctionalized silyl enol ethers8 or do not contain axial chirality in the allene;8,9 reported catalyst systems show little or no reactivity with nonterminal alkynes10 or 1,3-disubstituted allenes. The substitution pattern available using a substrate such as 1.2 would yield more complex cyclopentene scaffolds, provided the dr of the metal-catalyzed cycloisomerization is controlled.
Initial investigations focused on Au catalysts known to activate alkynes8,10 toward cyclization using 2a (Table 1) as a 1:1 diastereomeric mixture between b and the allene (for an extensive list of catalysts studied in preliminary investigations, see Table S3). While early results utilizing Ph3PAuCl, AgOTf and TfOH as an additive (see the Supporting Information (SI) for details) were successful, delivering 3a in 90% yield and 2.5:1 dr, these results were not reproducible using a new bottle of Ph3PAuCl. This led us to consider an impurity in the original Au source might have prevented decomposition of the catalyst to Au(0) or inactive (Ph3P)2Au+ species.11 Inspired by Gandon and Lafollée’s report describing Cu(OTf)2 as a useful additive for Au catalysis,12 we found reactivity could be restored using a 1:1 ratio of Ph3PAuCl and Cu(OTf)2 in refluxing toluene (Table 1, entry 1 and Table S3) to deliver 3a-anti as the major product in moderate yield and a dr of 1.3. However, a ligand on the Au salt was necessary, as AuCl3 gave no conversion of 2a (entry 2). The use of a coordinating anion on Cu (entry 3) shut down the reaction, while Cu(I)OTf (see Table S3) was less effective than Cu(OTf)2. We propose one function of the Cu(OTf)2 is to ensure slow release of the active cationic Au species during the reaction,12 but it likely plays a dual role in aiding the formation of a Cu-enolate species. The impact of the Au:Cu ratio was briefly examined (see Table S4 in the SI for details). The loading of Ph3PAuCl could be decreased to 2.5 mol %, provided the Cu(OTf)2 was increased to 25 mol % (entry 4) to furnish 3a in slightly better dr and comparable yield to entry 1. The extended reaction time at the lower Ph3PAuCl loading likely led to catalyst decomposition, but the more electron-rich Cy3PAuCl catalyst proved more robust (entry 5), delivering 3a in a 93% yield and 1.8:1 dr.
Using the conditions described in entries 1 and 4–5 of Table 1, a variety of 1,3-disubstituted allenes were explored. The n-pentyl-substituted allene 2a (Table 2, entries 1–2) delivered 3a in 93% yield; the stereochemistry at the ring juncture was syn, while the 1.8:1 dr between the two starred carbons favored the anti product. Switching the ester from Et in 2a to Me in 2b (entry 3) had little effect, while a less bulky Me-substituted allene 2c (entries 4–5) furnished 3c in good yield (entry 5) and poor dr, irrespective of the catalyst. A tethered phenyl substituent in 2d (entry 6) was tolerated; switching to Cy3PAuCl resulted in a doubling of the dr (entry 7) for reasons that were not clear at the time. Changing the nature of the ester in entries 1–3 had little effect, although the bulkier ester in 2e (entry 8) did improve the dr. Branching in the tether adjacent to the allene in 2f (entry 9) resulted in 3f with a 6.5:1 dr; however, the extra bulk at C3 gave a lower yield. Methyl substitution at the ring junction in 2g was well-tolerated (entry 10), yielding two adjacent quaternary, all-carbon stereocenters in 3g.
A series of 1,1,3-trisubstituted allenes were also subjected to Au(I)-catalyzed carbocyclization. Di-n-propyl allene 2h was converted to 3h in 62% yield (Table 2, entry 11), while allene 2i, containing additional branching in the side chain, gave 3i in 32% yield and 1.1:1 dr (entry 12). Moving the third substituent on the allene from C3 to C1 in 2j (entry 13) gave a 54% yield of 3j in 2.0:1 dr using Ph3PAuCl. Attempts to increase yield and dr with a higher catalyst loading, longer reaction time, or Cy3PAuCl (entries 14–15) were not successful. However, replacing the C1Me of 2j with a bulky tert-butyldimethylsilyl (TBS) group in 2k (entry 16) furnished 3k in 51% yield and a good dr of 4.5:1. In contrast, trimethylsilyl (TMS) substitution on the C3 distal allene double bond resulted in rapid protodesilylation to deliver 3a when subjected to the standard reaction conditions with both Ph3PAuCl and Cy3PAuCl (not shown).
At first glance, the dr of the cyclopentene products obtained in Table 2 do not appear to respond in any predictable way to changes in the ligand or the allene substitution pattern. However, the ability to obtain a significantly higher dr in the product than the starting 1:1 dr of the substrate indicates the rate of allene epimerization is faster than carbocyclization. Thus, it is theoretically possible to obtain a single cyclopentene diastereomer if better insight into the important features of this complex system can be obtained.
A partial mechanistic pathway to rationalize the observed stereochemical outcomes in Table 2 is illustrated in Figure 1. In this scenario, 2a (1:1 dr) engages Cu(OTf)2 to form two Cu-enolate species, A and B. There are multiple ways in which Au(I) can coordinate to A and B, as the metal may bind to the C1–C2 or the C2–C3 double bond from either face, depending on allene substitution pattern and substituent size. Nonetheless, literature precedent suggests eventual formation of achiral Au(I) η1-allyl cations as intermediates,13 represented by species C–F. Experimentally, the observation that the juncture between the two five-membered rings is always syn has two important implications. First, it supports our proposal that epimerization of Au-coordinated allene13 to stereodefined and achiral Au(I) η1-allylic cationic intermediates C–F occurs at a faster rate than ring closure. Second, it tells us that the copper enolate approaches the Au(I) η1-allyl cation on the same face as the proton shown in red in C–F. Thus, intermediates C and F will lead to the 3a-anti product, while D and E furnish the 3a-syn diastereomer. Reversibility in the stereochemical-determining C–C bond-forming step was ruled out by subjecting isolated 3a-anti and 3a-syn to the reaction conditions (Ph3PAuCl and Cu(OTf)2 in refluxing toluene, Figure 1, bottom) and noting the absence of epimerization.
The key to stereocontrolled reaction is to understand how the allene substitution pattern, Au catalyst, and Lewis acid additive work together to influence which of the four proposed intermediates C–F (Figure 1) are favored. C and E are favored when the allene is monosubstituted at C3 in order to minimize A1,3 strain. However, steric clashing between [Au] and the lactone must also be considered; C and D minimize this interaction, especially as the size of [Au] increases. As C minimizes both A1,3 strain and [Au]/substrate interactions, we propose this is the lowest energy species leading to anti product, while D is the lowest energy intermediate expected to deliver the syn product. According to this model, increasing catalyst bulk with 1,3-disubstituted allenes should have little impact on the dr. Indeed, only a slight increase in dr resulted when moving from Ph3PAuCl to Cy3PAuCl (Table 2, entries 1–2 and 4–5). However, as C3 becomes larger, preference for both C and the anti product should increase. This is indeed the case for 3d–f (Table 2, entries 6–9); however, if the bulk is too far removed, this effect is lost. The increase in dr was particularly noteworthy in 2f, where additional A1,3 strain due to the iPr group resulted in a 6.5:1 dr of 3f. Disubstitution at C3 of 2h–i (Table 2, entries 11–12) disables the ability to relieve A1,3 strain, and essentially no dr is observed. In contrast, additional substitution at C1, especially when R is bulky (compare entries 13–15 vs 16), is expected to further increase preference for C over D, resulting in increased preference for the anti product. This is what is observed, with reaction of 2k (entry 16) yielding 3k in 4.5:1 dr. The effect of the Au(I) ligand is still under investigation.
To test whether our model has predictive power, we reasoned incorporation of a gem-dimethyl group on the lactone ring of 2l (Table 3) should disfavor E and F due to increased steric interactions between the ring and the Au catalyst. Product formation should proceed more readily from C, as opposed to D, due to decreased steric effects as the enolate approaches the Au(I) η1-allyl cation to form the new C–C bond. In the event, treatment of 2l with Ph3PAuCl furnished 3l in 70% yield and 2.6:1 dr (entry 1). Switching to the more sterically demanding Cy3PAuCl (entry 2) gave 3l in excellent 89% yield and 6.6:1 dr, a marked increase over the analogous allene precursor 2a. These reactions could be carried out on gram scale, with isolated yields closely tracking with NMR yields (see the SI for details), highlighting another strategy to exploit interactions between substrate and catalyst to influence dr. Computational studies are in progress to inform other strategies to achieve highly stereocontrolled allene carbocyclizations.
Previous reports14 have shown that the additive identity in Au-catalyzed Coniaene type cyclizations may influence stereo-selectivity. With this in mind, we were curious if changing the metal additive could select for the syn diastereomer in reactions of 2l (Table 4). All reactions were complete in 24 h. Interestingly, although In(OTf)3 (entry 1) was less effective than Cu(OTf)2, a switch in the dr from 6.6:1 anti to 1.2:1 syn was noted. Sn(OTf)2 (entry 2) gave a higher yield and similar dr, while Zn(OTf)2 further improved the preference for 3l-syn in good yield (entry 3). Eu(OTf)3 (entry 5) gave the best balance of dr and yield in this limited screen, producing 3l-syn as the major diastereomer in 83% yield and 1.9:1 dr. Presumably, the nature of the Lewis acid additive shifts the equilibrium between the cationic η1-allylic intermediates C–F; further investigations into the exact role of the metal additive and its impact on stereoselectivity are currently being pursued.
Conversion of the cyclopentene products to highly substituted cyclopentanes was briefly explored (Scheme 2). Substrate 3m was prepared in the same manner as 3l (Table 4), with the exception that E = CO2Me. Both 3l–m could be enriched to >19:1 dr by standard purification by column chromatography. The concave nature of the 5,5-bicyclic rings of 3l and 3m were leveraged to selectively functionalize one face of the alkene over the other to furnish substituted cyclopentanes in a highly diastereocontrolled manner. Treatment of 3m with mCPBA delivered epoxide 4 in excellent dr, while diol 6 was obtained in >19:1 dr under the Upjohn dihydroxylation conditions. Hydroboration of 3l with BH3·THF, followed by oxidation with NaBO3·4H2O, produced the alcohol 8 in 64% yield and >19:1 dr. The pendant methyl ester moiety of 3m could be removed by heating in DMSO with LiCl and H2O to furnish lactone 5. Using conditions described previously by Du Bois, treatment of 5 with catalytic Rh2(tfacam)4 (tfacam4 = CF3CONH), PhI(OAc)2, and 1,1,1-trichlorosulfamate ester (TcesNH2) provided aziridine 7 in 66% yield and >19:1 dr.15
In conclusion, we have demonstrated that highly complex cyclopentane scaffolds can be accessed in only a few steps from diastereomeric mixtures of allene precursors. Equilibrium control over the addition of a metal enolate to an activated allene drives a diastereocontrolled carbocyclization reaction. The success of this transformation hinges on the slow delivery of the fragile cationic gold species via triflate exchange with a Cu(OTf)2 or other Lewis acidic triflate source. Factors including minimization of A1,3 strain and steric interactions between the substrate and the Au(I) catalyst are proposed to be the primary factors driving the stereoselectivity. Subsequent transformations of the alkene products show the variety of functionalization patterns that can be accessed via this carbocyclization methodology.
This work was funded by the University of Wisconsin and NIH R01GM111412. The NMR facilities at UW Madison are funded by the NSF (CHE-1048642, CHE-0342998) and NIH S10 OD012245. The Q Exactive Plus Orbi mass spectrometer is supported by NIH S10 Grant 1S10OD020022-1.