All-Carbon [3+3] Oxidative Annulations of 1,3-Enynes by Rhodium(III)-Catalyzed C-H Functionalization and 1,4-Migration.

1,3-Enynes containing allylic hydrogens cis to the alkyne function as three-carbon components in rhodium(III)-catalyzed, all-carbon [3+3] oxidative annulations to produce spirodialins. The proposed mechanism of these reactions involves the alkenyl-to-allyl 1,4-rhodium(III) migration.

Transition metal-catalyzed oxidative annulations of n class="Chemical">alkynes[1] that proceed by directing group-promoted C(sp2)–H functionalization[1, 2] are versatile methods for heterocycle[3] and carbocycle[4] synthesis. Alkynes, including 1,3-enynes,[5] serve as two-carbon components in these reactions (Scheme 1 a). However, analogous reactions that result in three-carbon annulation are currently underdeveloped,[6] and addressing this shortcoming would expand the range of products accessible using C–H functionalization/oxidative annulation chemistry.

Scheme 1

Catalytic oxidative annulations of alkynes and 1,3-enynes.

Catalytic oxidative annulations of n class="Chemical">alkynes and 1,3-enynes.

Using rhodium(III) catalysis, we recently discovered a new mode of oxidative annulation of n class="Chemical">1,3-enynes that contain allylic hydrogens cis to the alkyne, in which they act as one-carbon components (Scheme 1 b).[7] The proposed mechanism[7] involves the 1,4-rhodium(III) migration[8, 9] of alkenylrhodium species A to give σ-allylrhodium(III) species C via rhodacycle B. Following isomerization of C into the electrophilic π-allylrhodium(III) species D, nucleophilic trapping by the directing group gives the product of [n+1] annulation. Given the isomerization of C into D, there exists the possibility for cyclization to occur at a different position of the extended π-system to give E, a product of [n+3] annulation (Scheme 1 b).[6]

Herein, we describe the realization of this possibility in rhodium(III)-catalyzed reactions of n class="Chemical">2-aryl cyclic 1,3-dicarbonyls 1 with 1,3-enynes 2 to give spirodialins 3 (Scheme 1 c). The majority of the products obtained are spirocyclic barbiturates, which are of interest given the well-established medicinal importance of the barbiturate motif, and the biological activity of structurally related spirocycles (Figure 1).[10]

Figure 1

Biologically active spirocyclic barbiturates.

Biologically active spirocyclic n class="Chemical">barbiturates.

During our studies of metal-catalyzed oxidative annulations of n class="Chemical">alkynes,[4a,b,e, 7] the reaction of 5-arylbarbituric acid 1 a with 1,3-enyne 2 a was performed using [{Cp*RhCl2}2] (2.5 mol %) and Cu(OAc)2⋅H2O (2.1 equiv) in dioxane/H2O (5:1) at 60 °C [Eq. (1)]. As well as providing the spiroindene 4 a through a standard two-carbon annulation,[4a,b,e] a [3+3] annulation occurred to give spirodialin 3 a as the major product. No one-carbon annulation product 5[7] was detected. Chromatographic purification gave a 72:28 mixture of 3 a and 4 a in 92 % yield. Without H2O, more side products were formed and the ratio of 3 a:4 a decreased to ca. 50:50. No reaction occurred without Cu(OAc)2⋅H2O.[11]

Further studies revealed the benzyloxy-containing 1,3-enyne 2 b to be superior to 2 a; the reaction of 2 b with 1 a gave n class="Chemical">spirodialin 3 b only, in 88 % yield as the E-isomer (Scheme 2). Reaction of 2 b with various 5-arylbarbituric acids[12] demonstrated compatibility with nitro (3 c), acetoxy (3 d), and halogen substituents (3 f–3 h) on the aryl group.[13] Spirodialin 3 e was not formed under the standard conditions,[14] but replacing dioxane/H2O with undried DMF enabled productive [3+3] annulation and isolation of 3 e in 37 % yield, along with several side products.[14] Free N–H groups on the barbituric acids were also tolerated (3 i and 3 j). In the latter case, 3 j was formed as a 1:1 inseparable mixture of diastereomers. The reaction of 2-phenyl Meldrum’s acid also gave [3+3] annulation, but the yield of 6 was only 32 % due to decomposition of the starting material and product under the acidic conditions.[15] Decreasing the loading of [{Cp*RhCl2}2] to 0.5 mol % in the reaction of 1 a with 2 b was well-tolerated and provided 3 b in 77 % yield.

Scheme 2

[a] Conducted with 0.50 mmol of 1 a–1 j. [b] Yield of isolated products. [b] Conducted with 0.5 mol % of [{Cp*RhCl2}2]. [c] Conducted in undried DMF. Side products were also obtained; see Ref. [14]. [d] Conducted at 120 °C. [e] Conducted with 5 mol % of [{Cp*RhCl2}2].

[a] Conducted with 0.50 mmol of 1 a–1 j. [b] Yield of isolated products. [b] Conducted with 0.5 mol % of [{Cp*RhCl2}2]. [c] Conducted in undried n class="Chemical">DMF. Side products were also obtained; see Ref. [14]. [d] Conducted at 120 °C. [e] Conducted with 5 mol % of [{Cp*RhCl2}2].

Interestingly, Cu(OAc)2⋅n class="Chemical">H2O rapidly decomposed cyclic hydrazide 7, precluding its use as the oxidant in the reaction with 1,3-enyne 2 b [Eq. (2)]. However, reaction of 7 (2.0 equiv) with 2 b without Cu(OAc)2⋅H2O but with inclusion of NaOAc⋅3H2O (3.0 equiv) gave spirodialin 8 in 47 % yield, along with 2 b (30 % recovery). We speculate that the N–N bond of 7 could be serving as an oxidant to regenerate the catalyst,[16] but we were unable to isolate the reduced form of 7 to confirm this hypothesis.

Table 1 presents the results of oxidative annulations of 5-arylbarbituric acids with various n class="Chemical">1,3-enynes. No spiroindenes or benzopyrans from two- or one-carbon annulations, respectively, were detected. 1,3-Enynes 2 c and 2 d, containing protected or unprotected 2-hydroxyethyl groups on the alkyne were tolerated (entries 1 and 2). Use of a methoxy group in the 1,3-enyne in place of a benzyloxy group was also possible (entry 3). With a 5-(4-nitrophenyl)-substituted barbituric acid, oxidative annulations with 1,3-enynes 2 a, 2 f, and 2 g containing various groups trans to the alkyne proceeded efficiently to give spirodialins 3 n–3 p in 72–95 % yield (entries 4–6). As with the corresponding one-carbon annulations,[7] the 4-nitrophenyl group favors 1,4-rhodium(III) migration over the formation of spiroindenes [compare with Eq. (1)]. 1,3-Enynes 2 h and 2 i containing cyclic groups were also competent substrates (entries 7 and 8), and spirodialin 3 r was isolated in 57 % yield, despite containing a potentially acid-sensitive enol acetal.

Table 1

[3+3] Oxidative annulations of various 1,3-enynes[a]

Entry	1,3-Enyne	Product		Yield [%][b]
1 2	2 c 2 d		3 k R=TBS 3 l R=H	60 64
3	2 e		3 m	80
4 5 6	2 a 2 f 2 g		3 n R=Me 3 o R=Ph 3 p R=H	95 78 72
7	2 h		3 q	86
8	2 i		3 r	57

[a] Conducted with 0.50 mmol of 1. [b] Yield of isolated products.

Notably, the formation of a highly sterically hindered n class="Chemical">spirodialin 3 s containing contiguous all-carbon sp3 quaternary centers from 1,3-enyne 2 j occurred efficiently [Eq. (3)]. The reaction of 1 b with 1,3-enyne 9, which does not contain any cis-allylic hydrogens, led only to the formation of spiroindene 4 b in 82 % yield, thus highlighting the importance of this structural feature for [3+3] annulation [Eq. (4)].

Scheme 3 depicts a possible catalytic cycle for these reactions, using representative substrates 1 a and 2 a. This cycle is similar to that proposed for the one-n class="Chemical">carbon annulations we described previously.[7] Cyclorhodation of 1 a with rhodium diacetate 10 would give rhodacycle 11. Migratory insertion of 1,3-enyne 2 a then provides rhodacycle 12, which upon reductive elimination would give spiroindene 4 a. However, reversible protonolysis of 12 forms alkenyrhodium species 13, which can then undergo 1,4-rhodium(III) migration to form σ-allylrhodium(III) species 14. This intermediate can lead to π-allylrhodium(III) species 15 by a series of σ–π–σ interconversions and E/Z isomerization. Outer sphere nucleophilic attack of the π-allylrhodium(III) moiety[17, 18] of 15 by C5 of the barbituric acid then gives spirodialin 3 a and rhodium(I) species 16, which undergoes Cu(OAc)2-promoted oxidation to regenerate 10. The preference of 5-monosubstituted barbituric acids for C-allylation over O-allylation has been observed previously in Pd-catalyzed asymmetric allylic alkylations.[19] However, an alternative pathway involving an inner-sphere reductive elimination cannot be excluded.

Scheme 3

Possible catalytic cycle.

The reaction of 1 a with 1,3-enyne 2 k gave n class="Chemical">spiroindene 4 c only (Scheme 4), a result that differs from the formation of spirodialins 3 b (Scheme 2) and 3 m (Table 1, entry 3) from 1,3-enynes 2 b and 2 e, respectively. A possible explanation for this contrasting behavior might be coordination of the acetoxy group to rhodium, resulting in stabilization of 18-electron intermediates such as rhodacycles 17 and 18 (analogous to 12 in Scheme 3, but the σ-haptomers) or alkenylrhodium species 19. This stabilization likely disfavors 1,4-rhodium(III) migration and leads instead to reductive elimination from 18 to give 4 c.

Scheme 4

Formation of spiroindene 4 c from 1,3-enyne 2 k.

Formation of spiroindene 4 n class="Gene">c from 1,3-enyne 2 k.

The reaction of 1 b with the hexadeuterated 1,3-enyne [D]6-2 a gave traces of a n class="Chemical">spiroindene [D]6-4 d (<5 %), and spirodialin [D]6-3 n in 88 % yield (Scheme 5 a), in which incomplete deuterium transfer (91 % D) from the cis-allylic position of [D]6-2 a to the alkenyl position of the dialin ring of [D]6-3 n was observed. Furthermore, the reaction of 1 b with 2 a in 5:1 dioxane/D2O led to 10 % deuteration at the same position of [D]-3 n, with no spiroindene detected (Scheme 5 b). These results are similar to the corresponding experiments with [D]6-2 a in the one-carbon annulations reported previously,[7] and are consistent with 1,4-rhodium(III) migration occurring by a concerted metalation-deprotonation/reprotonation sequence (similar to A to C in Scheme 1 b).[7]

Scheme 5

Oxidative annulation with a hexadeuterated 1,3-enyne.

Although the spirocyclic n class="Chemical">barbiturates prepared in this study are themselves of interest, they can be transformed into other compounds. For example, treatment of 3 o with aqueous NaOH in THF gave the highly functionalized naphthalene 20 in 90 % yield [Eq. (5)].

In conclusion, we have reported rhodium(III)-catalyzed, all-n class="Chemical">carbon [3+3] oxidative annulations of 5-arylbarbituric acids and related compounds with 1,3-enynes containing allylic hydrogens cis to the alkyne. This new mode of oxidative annulation further demonstrates the power of alkenyl-to-allyl 1,4-rhodium(III) migration in generating electrophilic allylrhodium species for the construction of polycyclic systems. Other applications of this method of allylmetal generation will be reported in due course.