Iron‐Catalyzed Cycloisomerization and C−C Bond Activation to Access Non‐canonical Tricyclic Cyclobutanes

Abstract

Cycloisomerizations are powerful skeletal rearrangements that allow the construction of complex molecular architectures in an atom‐economic way. We present here an unusual type of cyclopropyl enyne cycloisomerization that couples the process of a cycloisomerization with the activation of a C−C bond in cyclopropanes. A set of substituted non‐canonical tricyclic cyclobutanes were synthesized under mild conditions using [(Ph₃P)₂Fe(CO)(NO)]BF₄ as catalyst in good to excellent yields with high levels of stereocontrol.

A cationic Fe−NO complex catalyses the cycloisomerization of cyclopropyl enyne to fused non‐canonical cyclobutanes with up to four consecutive stereocenters with exclusive diastereoselectivity. Only Ag(+I) was found to show comparable activity, albeit significantly lower reactivity.

Cycloisomerizations allow the construction of complex annelated ring systems from readily available simple starting materials. The significance of the research activities is reflected in the number of reports on this particular type of skeletal rearrangements. ^[1] While Rh‐ ^[2] and Ir‐complexes ^[3] clearly dominated this field of chemistry, the advent of defined Au^I‐complexes as π‐Lewis acid catalyst opened up a totally new direction in this field. ^[4] Recently, we reported that the cationic 16‐electron Fe⁰‐complex [(Ph₃P)₂Fe(CO)(NO)]BF₄ 2 acts as a π‐Lewis acid catalyst redox‐neutral cycloisomerizations of enyne acetates or aryl allenyl ketones [Eq. (1), Scheme 1]. ^[5] Cycloisomerizations of cyclopropyl‐substituted enynes such as 5 were pioneered through a couple of landmark reports by Wender[ ^6a , ^6b , ^6c ] or Shintani/Hayashi ^[6d] using Rh‐complexes to provide products of a formal [5+2]‐cycloaddition [Eq. (2), Scheme 1]. The fact that Au^I‐complexes act as π‐Lewis‐acid was used by Echavarren in a remarkable study on cycloisomerizations of cyclopropyl‐substituted enynes such as 7 [Eq. (3), Scheme 1]. ^[7] In sharp contrast to the Rh‐catalysis the cycloisomerization does not deliver the [5+2]‐cycloaddition product but rather the product of a cycloisomerization‐Prins‐cyclization to give fused 5,5,4‐tricyclic product 8 [Eq. (3), Scheme 1].

Scheme 1.

Transition metal catalysed cyclizations of enynes and cyclopropyl‐substituted enynes.

We envisioned a comparative study on cyclopropyl‐substituted enynes to be a perfect litmus test to verify or falsify our understanding of the catalytic activity of [(Ph₃P)₂Fe(CO)(NO)]BF₄ toward enynes. Herein, we report that the cationic Fe‐complex [(Ph₃P)₂Fe(NO)(CO)]BF₄ catalyzes the cycloisomerization‐C−C‐bond activation of cyclopropyl‐substituted enynes to complex tricyclic n,5,4‐scaffolds that are present in natural products like sulcatine G or kelsoene [Eq. (3), Scheme 1].

Based on our previous work, we initiated this study by treating cyclopropyl enyne 12 with the Fe‐complex under the conditions that proved successful in our previous studies (Table 1). While initial attempts to employ simple unactivated cyclopropanes (R⁴=H, Me) or alkoxide‐substituted cyclopropanes (R⁴=OTBS) did not show any conversion, aryl‐substituted cyclopropanes like 12 showed good reactivity already under mild conditions in dichloromethane at 50 °C leading to a mixture of allylic acetate 14 or diene 13 (entry 1, Table 1).

Table 1.

Catalyst optimization.^[a]

Entry	Catalyst [mol‐%]	Solvent/T [°C]	13 [b]/14 [b]
1	2 (10)	CH2Cl2/50	80 %/–
2	2 (2)	CH2Cl2/50	40 %/35 %
3	2 (3)	CH2Cl2/50	43 %/33 %
4	2 (3)	CH2Cl2/30	60 %/34 %
5	–	CH2Cl2/50	–
6	HBF4 (2)	CH2Cl2/50	decomp.

[a] Reaction conditions: All reactions were performed on a 0.2 mmol scale in dry solvent (1 mL) under a N₂‐atmosphere for 22 h. [b] Yields determined by ¹⁹F NMR using 4,4′‐difluorobenzophenone as internal standard.

Diene 13 was isolated in good yields as the only product. Upon decreasing the catalyst loadings down to 3 mol‐% good conversions even at a temperature of 30 °C were possible, however, diene 13 was formed as a by‐product alongside with allylic acetate 14 [entries 3 and 4, Table 1, Eq. (2), Scheme 3]. ^[8] Control experiments showed that the reaction is indeed Fe‐catalyzed (entries 5 and 6, Table 1).

Scheme 3.

Fe‐catalyzed cycloisomerization‐elimination—influence of the ester and olefinic moiety. [a] Partial hydrolysis of the ester to the corresponding tertiary alcohol 40 was observed. Combined yield after acetylation of the crude product. ^[9]

The cycloisomerization‐elimination proved to be broadly applicable (Scheme 2). A variety of different cyclopropyl‐substituted enynes were transformed into the corresponding fused tricyclic cyclobutanes in good to high yields. While in all reported cases the yields based on recovered starting materials are exceeding 90 %, in some cases the formation of the allylic alcohol due to hydrolysis of the allylic ester moiety was observed to a minor amount. Importantly, halides, ethers, and amides are tolerated. Propargyl‐ but also homopropargylamides can be employed and thus allow for variation of the ring sizes, or upon use of higher substituted homopropargylic amides the formation of even tetracyclic structures like 29 in good yields. The nature of the p‐substituent Y has a significant impact on the yield (products 13–19 vs. products 20–24, Scheme 2). Whereas +M‐substituents increase the yield of products 13–19, which is indicative for the need to stabilize partial positive charge during the activation of the propargylic acetate, the opposite is true for the p‐substitution of the aryl‐group at the cyclopropane moiety (products 20–24, Scheme 2). This points into the direction of a catalyst‐controlled C−C‐bond activation of the cyclopropane rather than a carbocationic (Wagner–Meerwein‐type) ring‐opening reaction.

Scheme 2.

Scope of the Fe‐catalyzed cycloisomerization‐elimination. ^[9]

As the formation of dienes 13–29 follows a two‐step mechanism of cycloisomerization and subsequent elimination [Eq. (2), Scheme 3] ^[8] we were wondering whether a variation of the ester moiety in the starting material might allow to slow down the elimination process and hence would provide an access to the corresponding allylic acetate (Scheme 3). Consequently, various propargylic esters 30–37 were prepared and subjected to the reaction conditions. Gratifyingly, in all cases the starting material was converted indicating that apart from pivalate these ester groups are compatible with the cycloisomerization conditions, however, in all cases a fast elimination to diene 13 or 17 was observed [Eq. (1), Scheme 3]. At this point we were wondering whether the Fe‐complex actually catalyzes the fast elimination. Indeed, subjecting allylic acetate 14 to the reaction conditions in the presence or absence of catalyst 2 led to either quantitative elimination to 13 or to no reaction, respectively [Eq. (2), Scheme 3]. Consequently, trisubstituted olefin 35 which is not prone toward elimination undergoes the cycloisomerization in 75 % yield to diastereomerically pure acetate 36 [Eq. (3), Scheme 3].

In transition metal catalyzed Tsuji–Trost‐type allylations allylic alkylcarbonates are highly reactive. In contrast to the use of allylic acetates, the leaving group can undergo a fast decarboxylation to give the corresponding alkoxide which acts as a base or nucleophile. Gratifyingly, the use of propargylic alkylcarbonates as starting materials slowed down the elimination process. Employing propargylic carbonate 37 we were able to isolate ether 38 in 52 % yield as a single diastereomer along with diene 17 as by‐product [Eq. (1), Scheme 4]. The use of trisubstituted olefin 39 led to the formation of the respective allylic ether 40 in 82 % yield and with exclusive diastereoselectivity [Eq. (2), Scheme 4].

Scheme 4.

Fe‐catalyzed decarboxylative cycloisomerization. ^[9]

Although a detailed description of mechanistic scenarios at the current state of research is not possible and needs further investigations, we performed additional experiments to gain some further information on potential pathways (Scheme 5). Different catalysts were tested [Eq. (1), Scheme 5]. While (Rh(CO)₂Cl)₂ did not work, we were surprised to find Echavarren's Au^I‐complex to be inactive, too [Eq. (1), Scheme 5]. The use of AgBF₄ did provide access to product 13 in moderate yields. ^[10] The use of AuCl₃ led to conversion of starting material, however, allene 44 and cyclopropane 43, in which the cyclopropyl‐moiety remained unaffected, were isolated in moderate yields [Eq. (2), Scheme 5]. As we considered both products to be potential intermediates in this transformation, we subjected them to the reaction conditions but did not find any conversion [Eq. (3) and (4), Scheme 5]. Based on these findings, our previous experimental results in related cycloisomerizations[ ^5a , ^5c ] and the fact that +M‐substituents at the propargylic aromatic unit promote the reaction whereas +M‐substituents in the arylcyclopropane moiety decrease the conversion (Scheme 2) we propose the following mechanistic scenario as a working hypothesis (Figure 1).

Scheme 5.

Control experiments.

Figure 1.

Mechanistic model for Fe‐catalyzed cycloisomerization.

The cationic Fe‐complex coordinates to the alkyne moiety leading to the cyclopropyl‐substituted metallcyclobutane intermediate II. The fact that +M‐substituents in the arylcyclopropane moiety decrease the conversion does not support the hypothetical formation of a free carbocationic intermediate, hence, it seems as if the cationic Fe‐center fosters II to undergo a metal‐centered C−C‐bond activation ring expansion process with release of ring‐strain to give ferracyclobutane III. 1,3‐Metallotropic shift within the exocyclic vinylacetate unit provides access to the ferracycloheptane IV that upon 1,2‐acetoxy migration and deferration leads to the observed product VI and recycled catalyst 2.

We report here a cycloisomerization‐C−C‐bond activation of enyneacetates featuring a vinylcyclopropane moiety. Catalytic amounts of the readily accessible cationic Fe‐complex [(Ph₃P)₂Fe(CO)(NO)]BF₄ mediate this transformation under mild conditions to give fused tricyclic cyclobutanes in good to high yields. Depending on the substitution pattern either 1,3‐dienes or allylic acetates were observed. Interestingly, amongst the different noble metal type catalysts tested only AgBF₄ showed a similar reactivity albeit at significantly lower yields.

Conflict of interest

The authors declare no conflict of interest.

Supporting information

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer reviewed and may be re‐organized for online delivery, but are not copy‐edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.

Supporting Information

In memory of Professor Klaus Hafner

F. Kramm, F. Ullwer, B. Klinnert, M. Zheng, B. Plietker, Angew. Chem. Int. Ed. 2022, 61, e202205169; Angew. Chem. 2022, 134, e202205169.

Data Availability Statement

The data that support the findings of this study are available in the Supporting Information of this article.