Rearrangement reactions of 1,1-divinyl-2-phenylcyclopropanes.

1,1-Divinyl-2-phenylcyclopropanes are entry points to a rich area of rearrangement chemistry. With N,N-diallyl amide substrates, tandem radical cyclizations can be initiated at room temperature. Warming provides products of pure thermal rearrangements with acids, ester, and amides. These isomerizations give vinylcyclopentenes resulting from divinylcyclopropane rearrangements and more deeply rearranged tricyclic spirolactams resulting from aromatic Cope rearrangements followed by ene reactions. Conversion of the carbonyl group to an alcohol or ether opens retro-ene pathways followed by either tautomerization or Claisen rearrangement.

Introduction

Vinyl-substituted cyclopropanes are readily accessible intermediates that are commonly used in synthesis.[1] Release of the cyclopropane ring strain provides a driving force for a variety of radical[2] and metal-mediated[3] transformations. The archetypical transformation is the rearrangement of vinylcyclopropanes to cyclopentenes at high temperatures (often 300 °C or more),[4] hereafter called the vinylcyclopropane rearrangement (Figure 1a). Suitably substituted vinylcyclopropanes can also undergo other rearrangements including retro-ene reactions.[5]

Figure 1

Rearrangement reactions of vinylcyclopropanes and divinylcyclopropanes.

Among the various substituted vinylcyclopropanes, 1,2-divinyl-cyclopropanes are important precursors for 3,3-sigmatropic reactions like the Cope rearrangement (Figure 1b).[6] Such rearrangements typically occur at more accessible temperatures (<100 °C) than vinylcyclopropane rearrangements provided that the vinyl groups are disposed 1,2-cis. At higher temperatures, cis and trans isomers often equilibrate, opening a path from the trans isomer to the Cope product.

In contrast, relatively little is known about rearrangements of 1,1-divinylcyclopropanes.[7] The parent 1,1-divinylcyclopropane has been studied in detail by Dolbier,[7a,7b] and undergoes the vinylcyclopropane rearrangement to give vinylcyclopentene at about 250 °C (Figure 1c).

We recently used complex yet readily available 1,1-divinyl-cyclopropanes as key intermediates for tandem radical cyclizations to make meloscine and a variety of analogs.[8] In a typical example (Figure 2), treatment of 1 with tributyltin hydride produced tetracycle 2, which was further converted to the natural products epi-meloscine 3a (in 3 steps) and meloscine[9]3b (in one more step). While studying the radical-mediated rearrangements of 1,1-divinylcylopropanes with phenyl substituents, we began to encounter facile rearrangements that occurred without radical initiators.

Figure 2

Radical rearrangement of a 1,1-divinylcyclopropane is a key step in a short synthesis of epi-meloscine, meloscine, and analogs.

Here we report that suitably substituted 1,1-divinyl-2-phenyl-cylopropanes undergo a variety of thermal rearrangements in an accessible temperature regime. These include the vinylcyclopropane rearrangement, a tandem aromatic Cope-ene rearrangement, and a retro-ene reaction followed by either tautomerization or Claisen rearrangement. Taken together, the results suggest that substituted 1,1-divinylcylopropanes have a rich and controllable rearrangement chemistry.

Results and Discussion

We first encountered pure thermal rearrangements of 1,1-divinyl-2-phenylcylopropanes during study of the tandem radical cyclization of benzyl allyl amide 9a, whose synthesis and onward radical and thermal reactions are shown in Schemes 1 and 2. Amide 9a is readily made in four steps (Scheme 1). Regioselective cyclopropanation of the diethylphosphate ester of buta-2,3-dien-1-ol[10]4 with ethyl 2-phenyldiazoacetate was catalyzed by Rh2(esp)2[11] to provide stable methylenecyclopropane 5 in 70% yield.

Scheme 1

Synthesis of Divinylcyclopropane Precursor 9

esp is C6H4-m-(CH2CH2CO2)2; Ghosez reagent is (Me)2C=C(Cl)NMe2.

Scheme 2

Results of Rearrangement Reactions of 9a under Radical (a) and Pure Thermal (b) Conditions

Addition of vinyl magnesium bromide (C2H3MgBr) to a solution of CuCN (0.2 equiv), LiCl (0.2 equiv), and 5 followed by workup and chromatography provided a 62% yield of 1,1-divinylcyclo-propane ester 6 resulting from SN2′ displacement of the phosphate. Also isolated in 22% yield was regioisomer 7 resulting from SN2 displacement. Despite the minor SN2 product, this two-step route to the divinylcyclopropane 6 is more direct and more efficient than the five-step route used for the divinylcyclopropanes in the meloscine work.[8]

Base-promoted hydrolysis of hindered ester 6 in ether with excess potassium tert-butoxide and a limited amount of water (3 equiv) provided acid 8 after standard workup. This classic Gassman hydrolysis method[12] succeeds at room temperature. This is important because standard saponification of the hindered ester of 7 did not occur at room temperature. Heating gave multiple products, some of which we later understood to arise from thermal rearrangements (see below).

Acid 8 became a pivotal intermediate in synthesis of substrates for the rearrangement studies, so the crude product was purified by flash chromatography to provide a high quality sample in 77% yield. Reaction of 8 with the 1-chloro-N,N-2-trimethyl-1-propenylamine (Ghosez reagent),[13] 4-(N,N-dimethylamino)-pyridine (DMAP), and allylbenzylamine provided amide 9a in 69% yield.

In a typical tandem radical reaction experiment (Scheme 2a),[8,14] a benzene solution of 9a and diphenyl disulfide was irradiated with a UV lamp at room temperature. Evaporation and flash chromatography provided the target azabicyclooctane 10 in 50% yield. This product presumably results from the sequence of elementary steps summarized in Scheme 2a. Addition of PhS· to one of the vinyl groups of 9a induces cyclopropane opening to give dienyl sulfide 11. This undergoes two successive 5-exo radical cyclizations to give bicyclic β-thiophenyl radical 12, which in turn fragments to provide 10 and give back PhS·.

In reactions of 9a with various radical-generating species (PhSSPh, Bu3SnH) conducted above room temperature,[15] we consistently observed two new products by TLC analysis alongside 10. This was reminiscent of the above attempts at thermal saponification, which also gave unexpected products. So we hypothesized that background thermal chemistry was occurring.

In a control experiment shown in Scheme 2b, divinylcyclopropane 9a was simply heated alone in refluxing toluene. After 2 h, both precursor 9a and tandem radical product 10 were absent by TLC analysis; the spots for the two new thermally formed products were the only ones present. Evaporation and flash chromatography provided these two products in pure form.

The minor product, isolated in 16% yield, was the vinylcyclopentene 13a. This is the product of a vinylcyclopropane rearrangement in which one of the vinyl groups of the divinylcyclopropane participates and the other is a substituent.[7a] The rearrangement is regioselective with migration to the more-substituted cyclopropane carbon atom (the one bearing the amide and phenyl groups).

The major product was a more deeply rearranged tricyclic spirolactam 14a, isolated in 71% yield as a single stereoisomer. The structure of 14a was assigned by a series of 1D and 2D NMR experiments (see Supporting Information). The upfield chemical shift of the protons of the methyl substituent on the lactam ring (d, 0.55 ppm) shows that this group is cis to the adjacent phenyl ring. Spirolactam 14 forms by an aromatic Cope rearrangement followed by an ene reaction, as discussed in more detail below.

The conversion of 9a to 13a and 14a occurs at slower but still significant rates at temperatures as low as 40 °C (about 40% conversion after 36 h). Storage for a few days at ambient temperatures does not give much rearrangement, but a freezer is recommended for long-term storage of 9a and related divinylcyclopropanes.

We next studied both of these rearrangement pathways with the aid of readily available acid 8 as a common precursor for a dozen assorted substrates. (Here we focus on the rearrangement chemistry; see the Supporting Information for full details on substrate preparation and characterization.) To study the vinylcyclopropane rearrangement, we used precursors shown in Table 1 either without a pendant enophile (acid 8 itself and the saturated ester 16a and amide 9b) or precursors with an enophile that is held in an unfavorable geometry for an ene reaction (allyl ester 16b and cyclic allyl amides 9c and 9d).

Table 1

Vinylcyclopropane Rearrangement Products and Yields

After automated flash chromatography.

As shown by the results in Table 1, these substrates all provided solely the products of regioselective vinylcyclopropane rearrangements in good yields. For example, heating of parent acid 8 in refluxing toluene for 2 h, followed by cooling, evaporation, and flash chromatography, provided ring expanded vinylcyclopentenyl acid 15 in 86% yield (Table 1, entry 1). Likewise, the derived esters 16a,b and amides 9b–d provided the corresponding ring-expanded products 17a,b and 13b–d in spot-to-spot reactions and with uniformly good isolated yields (73–91%, entries 2–6).

Returning to the major spirolactam product 14a of Scheme 1b, we suggest that this forms by the back-to-back sigmatropic reactions shown in Figure 3. First, the phenyl ring and the adjacent cis-vinyl group combine in a 3,3-sigmatropic reaction that is a rare example of an aromatic Cope rearrangement[16] to give 18a (Figure 3a). This rearrangement may be endothermic, but the release of strain energy of the cyclopropane at least partially compensates for the loss of aromaticity of the phenyl ring.

Figure 3

Spirolactam 14a forms by a rare and probably reversible aromatic Cope rearrangement followed by an irreversible ene reaction.

The intermediate 18a was not observed. The exomethylene cyclohexadiene in 18a is a highly reactive enophile, and the ensuing intramolecular ene reaction provides the aromatized spirolactam 14a (Figure 3b). The geometry of the ene reaction necessitates that the new CH3 group in the product is cis to the aromatic ring.

To further study the aromatic Cope-ene path, we prepared the four amides 9e–h shown in Scheme 3. Like 9a, each of these substrates has an accessible ene component of an ene reaction (alkene or alkyne) present on the amide N-substituent. As usual, these precursors were made in good yields from the pivotal acid 8. Heating of N,N-diallylamide 9e at reflux in toluene for 2 h, followed by evaporation and chromatography, provided spirolactam 14e in 65% again as a single isomer. Isolated alongside this was the vinyl cyclopentene 13e in 16% yield. Likewise, the phenyl-substituted diallyl amide 9f provided 14f as a single stereoisomer in 45% yield. In this and the following examples, we stopped targeting isolation of the minor vinylcyclopentene products, but these were present in small amounts prior to the chromatography.

Scheme 3

Additional Examples of Formation of Sequential Cope-Ene Spirolactam Products

Thermal reactions of the two N-propargyl amides 9g and 9h gave similar ratios of spirolactams-to-vinylcyclopentenes, roughly 6/1 according to 1H NMR integration of the crude products. The major spirolactams 14g and 14h were isolated in 73% and 53% yield, respectively. In the case of propargyl silane 9h, the alkenyl silane product 14h was a single Z-isomer, again resulting from the geometry of the intramolecular ene reaction.

To learn about chirality transfer in these rearrangements, we selected diallyl amide 9e because two reactions (aromatic Cope and vinylcyclopentene rearrangements) can be probed in one experiment. Racemic 9e was resolved into its component enantiomers by chiral HPLC (see Supporting Information), and then these enantiomers were heated individually under the usual conditions. The results of these experiments are also shown in Scheme 3.

Starting from highly enriched precursors 9e (er 96/4 and 1/99), the Cope-ene products 14e showed low levels of enantioenrichment (58/42 and 35/65) while the vinylcyclopentene products 13e were racemic (50/50). Further, chiral HPLC analysis of starting material at partial conversion showed that racemization of 9e competed efficiently with its onward reactions. In a typical experiment, the conversion of 9e at 60 min was 76% and the er of remaining 9e was 54/46 (initial er 96/4).

Figure 4 shows a plausible interpretation of these results. The starting divinylcyclopropane 9e or ent-9e opens to a diradical[17] (boxed intermediate) that is either achiral or more likely chiral but racemic due to rapid σ-bond rotations. Both components of the diradical are well stabilized by conjugation, accounting for the relatively low temperature of bond cleavage. Reclosure of the diradical in a 1,1′-fashion racemizes the precursor while closure in a 1,3′-fashion provides the fully racemic vinylcyclopropane rearrangement product 13e.[6a]

Figure 4

Possible mechanistic scenario for competing vinylcyclopropane and Cope-ene rearrangement paths with a focus on stereochemistry. Reaction conditions, toluene reflux, 2 h; R = allyl.

The aromatic Cope rearrangement could be concerted and stereospecific with 9e giving 18e and ent-9e giving ent-18e. (A boat transition state with both π-groups endo to the cyclopropane ring is expected.[6a]) However, this rearrangement is probably reversible and either it or the ensuing ene reaction competes ineffectively with the racemization of the precursor 9e in refluxing toluene.

It is also possible that the aromatic Cope rearrangement is not concerted, but instead, product 18e arises by 3,3′-closure of the diradical. However, recall that the vinylcyclopropane product 13e is racemic but that the Cope-ene product 14e is not. This suggests that the diradical may not be a common intermediate in their formation. Instead, the partial (but not complete) racemization of the precursor may better account for the low (but not zero) level of chirality transfer from 9e to 14e.

The types of products that we observe provide an interesting contrast to recent observations by Stephenson and co-workers.[16b] They generated vinyl phenylcyclopropanes like 19 in situ by radical cyclizations and observed that these underwent rearrangements to 21, presumably by an aromatic Cope rearrangement to 20 followed by rearomatization by 1,3-hydrogen shift. The reaction conditions were mild: DMF, 40 °C for several hours. Stephenson observed no vinylcyclopropane rearrangement products at all, and our substrates never gave products of aromatic Cope rearrangements followed by hydrogen shift. This is surprising because rearomatization by 1,3-shift is a common reaction of exomethylene cyclohexadienes.[16a,18] The concerted shift is thermally forbidden, but deprotontation/reprotonation or other stepwise mechanisms are possible.

In our substrates, the vinylcyclopropane rearrangement competes with the Cope-ene rearrangement of 9a and 9e–h as indicated by the consistent observation minor cyclopentene products 13a and 13e–h (Schemes 2 and 3). With the substrates in Table 1, where onward ene reactions are either impossible or disfavored by geometry, we isolate only the vinylcyclopropane rearrangement products. For example, the reaction of 16a did not produce any rearomatized aromatic Cope product 23 (Figure 5b), only vinylcyclopentene 17a.

Figure 5

Contrasts with the results of Stephenson.

Still, the reversible aromatic Cope rearrangement must be occurring at least to some extent with 16a and related substrates because they are so structurally similar to amides 9a and 9e–g (only the amide or ester group differs). As they reflect this similarity, all the precursors react under the same conditions independent of which products are finally formed. Evidently then, rearomatization of the transient aromatic Cope rearrangement products like 22 cannot compete with either the onward ene reaction (when available) or the vinylcyclopropane rearrangement.

In an effort to induce rearomatization by hydrogen-shift, we heated ester 16a in toluene with either acid (CF3CO2H, 2 equiv) or base (iPr2NEt, 2 equiv); however, no new rearomatization product 23 was observed (Figure 5b). Instead, the usual vinylcyclopentene 17a was the only apparent product, and it was formed at about the same rate as in the experiment with no additive (Table 1, entry 2).

Finally, to learn about the role of the carbonyl group in these acid, ester, and amide substrates, we prepared alcohol 24 by reduction of ester 6 with lithium aluminum hydride (91% yield, see Supporting Information), then converted this to allyl ether 28 with NaH and allyl bromide (67% yield). The rearrangement chemistry of these two substrates is summarized in Scheme 4.

Scheme 4

Thermal Rearrangements of Reduced Alcohol 24 (a) and Allyl Ether 28 (b)

To start, both 24 and 28 were stable in refluxing toluene for several hours. So, the carbonyl group is an important activator in the acid, ester, and amide substrates even though it is not a direct participant in either the vinylcyclopropane or aromatic Cope rearrangements.

The reduced substrates 24 and 28 both underwent clean rearrangements at 250 °C in DMF in a microwave apparatus to give new types of products. Rearrangement of alcohol 24 (Scheme 4a) provided dienyl aldehyde 25 in 59% yield as a 6/1 mixture of E/Z isomers alongside 21% of the vinylcyclopropane rearrangement product 26. Aldehyde 25 probably arises from the retro-ene reaction[5] of 24 via TS-24 to form enol 27, which then tautomerizes.

Allyl ether 28 was then prepared because it has the requisite functionality to undergo all three types of thermal rearrangements observed so far. These are (1) the aromatic Cope-ene rearrangement (the allyl group of the ether is the potential ene component of 28); (2) the vinylcyclopropane rearrangement; and (3) the retro-ene reaction.

In the event, heating of allyl ether 28 at 250 °C provided a mixture of a major aldehyde product 29a and a minor vinylcyclopropane rearrangement product 30. These products could not be separated by direct flash chromatography, so the mixture was exposed to sodium borohydride. This reduced the aldehyde 29a to the more polar alcohol 29b, which was then isolated in 50% yield by flash chromatography. The less polar vinylcyclopropane rearrangement product 30 survived the NaBH4 reduction, and was isolated in 21% yield.

Aldehyde 29a presumably arises from an initial retro-ene reaction through TS-28 that gives allyl vinyl either 31. Subsequent 3,3-sigmatropic rearrangement of this intermediate (a Claisen rearrangement) provides the major product 29a. The aromatic Cope-ene product from this substrate would be a spiro ether, but there is no evidence for its formation.

Conclusions

These results suggest that 1,1-divinyl-2-phenylcyclopropanes are a class of substrates that have an especially rich array of rearrangement reactions. Figure 6 summarizes the different pathways observed herein. Starting from derivatives of a single divinylcyclopropane acid 8, we have isolated the products of tandem radical reactions, vinylcyclopropane rearrangements, aromatic Cope-ene rearrangements, and retroene rearrangements followed by either tautomerization or Claisen rearrangement.

Figure 6

Common acid precursor 8 provides a diverse set of rearrangement products in 1–3 steps.

Each of the structure types can be formed as the major (though not always exclusive) product. The default thermal reaction seems to be the vinylcyclopropane rearrangement, which gives vinylcyclopentene products in high yields when the other paths are disfavored or impossible. However, the other paths are easily dialed in by choice of R group and reaction conditions.

The radical reaction path regenerates a 1,1-divinyl substituent in the product. In contrast, all three types of thermal reactions produce 1,3-dienes that directly result from the divinylcyclopropane. However, the structural setting of these dienes is very different depending on the reaction. The versatility of the reactions and the diversity of the products suggest that substituted 1,1-divinylcyclopropanes could be useful intermediates in synthesis, perhaps even on par someday with their much more well studied monovinyl- and 1,2-divinylcyclopropane relatives.