Enantioselective Synthesis of Spiroindenes by Enol-Directed Rhodium(III)-Catalyzed C-H Functionalization and Spiroannulation.

Chiral cyclopentadienyl rhodium complexes promote highly enantioselective enol-directed C(sp(2))-H functionalization and oxidative annulation with alkynes to give spiroindenes containing all-carbon quaternary stereocenters. High selectivity between two possible directing groups, as well as control of the direction of rotation in the isomerization of an O-bound rhodium enolate into the C-bound isomer, appear to be critical for high enantiomeric excesses.

Cyclopentadienyl rhodium(III) complexes are well-established as highly active and versatile precatalysts in a diverse array of C–H functionalization reactions.[1] However, enantioselective variants of these reactions only became possible with the development of chiral C2-symmetric cyclopentadienyl ligands by Ye and Cramer,[2] and an artificial RhIII-containing metalloenzyme by Ward, Rovis, and co-workers.[3] To date, a handful of catalytic enantioselective RhIII-catalyzed C–H functionalizations have been described,[2-5] but there is a compelling need to develop new processes to access novel classes of enantioenriched products.[6]

We recently reported Ru- and Pd-catalyzed oxidative annulations of α-aryl cyclic 1,3-dicarbonyl compounds (or their enol tautomers) with alkynes that provide achiral or racemic spiroindenes.[7] Given that indenes appear in several biologically active compounds,[8, 9] the ability to prepare chiral spiro-fused indenes 4 by asymmetric C–H functionalization would be valuable.[4d, 10] Because we also found that [{Cp*RhCl2}2] is an effective precatalyst,[7a, 11] chiral cyclopentadienyl rhodium complexes 3 appeared to be highly promising for investigation. However, in contrast to existing enantioselective RhIII-catalyzed C–H functionalizations, which all rely upon aryl C(sp2)–H activation of substrates containing a single directing group (Scheme 1 a),[2-5] the substrates 1 required for our proposed study contain two potential directing groups (Scheme 1 b). Within the accepted model for enantioinduction using complexes 3,[2b, 5] cyclorhodation can generate up to four species, which differ in which directing group participates in cyclometallation, and/or the orientation of the rhodacycle within the chiral pocket (Scheme 2). This situation contrasts with existing processes,[2-5] including the dearomatizing oxidative spiroannulations of You and co-workers,[4d] in which only two conformations of one rhodacycle need to be considered. Given the possibility of other reaction pathways with potentially different stereochemical outcomes, the development of a highly enantioselective process was far from certain. Herein, we report the successful realization of asymmetric [3+2] spiroannulations to give a diverse range of spiroindenes in up to 97 % ee.

Scheme 1

Enantioselective RhIII-catalyzed C–H functionalizations.

Scheme 2

Possible species to consider upon cyclorhodation.

Our investigations began with an evaluation of chiral cyclopentadienyl rhodium complexes 3 a–3 f[2b] in the reaction of 4-hydroxy-6-methyl-3-phenyl-2H-pyran-2-one (1 a) with 1-phenylpropyne (2 a, 1.5 equiv), using Cu(OAc)2 (2.1 equiv) in DMF[12] at 50 °C for 24 h (Table 1). Benzoyl peroxide, which was employed as an additive in previous enantioselective Rh-catalyzed C–H functionalizations,[2, 4] was unnecessary,[13] and in all cases, only one regioisomer of spiroindene 4 a was detected. The parent complex 3 a (R=H) gave 4 a in 93 % NMR yield, but the enantioselectivity was moderate (entry 1).[14] Higher selectivities were obtained with complexes 3b–3f containing larger groups at the 3,3′-positions (entries 2–6). The OTBDPS- containing complex 3 f was optimal, and provided 4 a in high NMR yield and 95 % ee (entry 6).

Table 1

Catalyst evaluation in the reaction of 1 a with 2 a[a]

Entry	Rh complex3	NMR Yield [%][b]	ee [%][c]
1	3 a R=H	93	58
2	3 b R=OMe	97	90
3	3 c R=OiPr	33	78
4	3 d R=Ph	41	88
5	3 e R=OTIPS	84	92
6	3 f R=OTBDPS	98	95

[a] Reactions were conducted with 0.05 mmol of 1 a. [b] Determined by 1H NMR spectroscopy using 1,3,5-trimethoxybenzene as an internal standard. [c] Determined by HPLC analysis on a chiral stationary phase. TIPS=triisopropylsilyl, TBDPS=tert-butyldiphenylsilyl.

With an effective chiral complex identified, the enantioselective spiroannulation of 1 a with various alkynes was explored (Scheme 3). With unsymmetrical alkynes, the regioselectivities of these reactions were excellent, and with the exception of spiroindene 4 d, which was formed as a 19:1 regioisomeric mixture, only single regioisomers were detected. With 1-phenylpropyne (2 a), spiroindene 4 a was isolated in 84 % yield and 95 % ee. The same reaction run at room temperature provided 4 a in 78 % yield and 97 % ee. Diphenylacetylene reacted to give spiroindene 4 b in 67 % yield and 93 % ee, whereas a symmetrical dialkyl alkyne gave spiroindene 4 c in moderate yield and enantioselectivity. However, other alkyl/(hetero)aryl alkynes were excellent reaction partners. For example, alkynes containing 5-indolyl, 3-indolyl, or 2-thienyl substituents provided spiroindenes 4 d–4 f in 74–93 % yield and 89–97 % ee.

Scheme 3

Enantioselective oxidative annulations of 1 a with various alkynes. Reactions were conducted with 0.30 mmol of 1 a. Yields are of isolated products. Enantiomeric excesses were determined by HPLC analysis on a chiral stationary phase. [a] Conducted with 0.20 mmol of 1 a at room temperature for 24 h. [b] Formed as a 19:1 mixture of regioisomers as determined by 1H NMR of the unpurified reaction mixture. The isolated product was also a 19:1 mixture of regioisomers.

Various other substrates also underwent the spiroannulation with a range of alkynes to give spiroindenes containing ketoesters (4 g–4 l, 4 q, and 4 r), ketolactams (4 m–4 p and 4 t–4 v), a diketone (4 s), or a barbiturate (4 w) with generally high enantioselectivities (Scheme 4). Although complex 3 f was generally effective, in some cases the less sterically hindered complex 3 b gave superior yields and enantioselectivities (4 t–4 w). The reason for the superiority of complex 3 b in these cases is not currently known. Substitution at the meta- or para-position of the α-phenyl group was tolerated (4 g–4 i). With a meta-CF3 group, C–H functionalization occurred at the more sterically accessible site (4 g).[14] In our previous oxidative annulation work,[7] only six-membered cyclic 1,3-dicarbonyl compounds were employed. Therefore, it is notable that, for the first time, five- and seven-membered substrates could be employed (4 l, 4 m, and 4 o). The low yield of 4 l is attributed to its instability under the reaction conditions. Products containing the 1,3-dicarbonyl component within various polycyclic ring systems were also prepared (4 p–4 r and 4 v), although the enantioselectivities of 4 p and 4 q were more modest. A substrate in which the two possible directing groups are almost identical electronically, but sterically well-differentiated, gave spiroindene 4 s in 77 % yield and a reasonable 78 % ee. 1-Methyl-5-phenylbarbituric acid, in which the two carbonyl groups adjacent to the phenyl group are electronically and sterically similar, gave spiroindene 4 w with low enantioselectivity. Finally, several of the reactions could be carried out in dimethyl carbonate, a significantly more environmentally friendly solvent than DMF (4 i, 4 t, and 4 u).[15]

Scheme 4

Reactions were conducted with 0.20 or 0.30 mmol of 1 (see Supporting Information for details). Yields are of isolated products. Enantiomeric excesses were determined by HPLC analysis on a chiral stationary phase. [a] Dimethyl carbonate was used as the solvent. [b] The absolute stereochemistry of the major enantiomer of 4 w is not known.

To gain further insight into these annulations, deuteration reactions were conducted. Treatment of 1 c under the standard conditions in the absence of an alkyne but with the addition of D2O for 4 h led to recovery of [D]-1 c with 5 % deuteration at the ortho-positions of the arene only (Scheme 5 a). Furthermore, reaction of 1 c with alkyne 2 f under the same conditions led to recovered 1 c with no observable deuteration, and spiroindene [D]-4 h that was partially deuterated at the pyran-2,4-dione ring[16] but not at the arene (Scheme 5 b). Interestingly, the presence of D2O decreased the regioselectivity of this reaction compared to the one conducted in DMF only (Scheme 4), and [D]-4 h was isolated as an 8:1 mixture of inseparable regioisomers. The experiments shown in Scheme 5 suggest that cyclorhodation is largely irreversible under these conditions.

Scheme 5

Deuteration experiments.

A proposed catalytic cycle and stereochemical model[14] for these reactions is shown in Scheme 6 a, using 1 a and 2 a as representative substrates. After formation of rhodium diacetate complex 5, we assume that cyclorhodation of 1 a is promoted by the most enolizable of the two possible directing groups, which is the enol derived from the ketone rather than the ester, to give rhodacycle 6 a. Coordination and migratory insertion of the alkyne would then give rhodacycle 7 a. The alternative conformations 6 b and 7 b appear to be disfavored because of steric interactions between the side wall of the cyclopentadienyl ligand with the metallated arene of 6 b or the phenyl substituent of 7 b (Scheme 6 b). The next step is the isomerization of the O-bound rhodium enolate 7 a into the C-bound isomer 8 a, presumably through an oxa-π-allylrhodium species, which requires a rotation of the 4-alkoxypyran-2-one moiety. Because the rhodium alkoxide of this moiety is in closest proximity to the chiral ligand, it experiences the greatest steric interactions, and we propose there is a preference for this group to rotate away from the ligand to give 8 a. Reductive elimination of 8 a gives spiroindene 4 a and RhI species 9, which is oxidized by Cu(OAc)2 to regenerate 5. The formation of the minor enantiomer from 7 a requires an unfavorable rotation of the rhodium alkoxide towards the back wall of the chiral ligand (Scheme 6 c).

Scheme 6

Proposed catalytic cycle and stereochemical model.

An alternative explanation that cannot be excluded is that migratory insertion of 6 a with the alkyne directly produces a rhodacycle with a conformation closely related to that of 7 a, but with the rhodium alkoxide already partially rotated away from the chiral ligand. Continued rotation of the 4-alkoxypyran-2-one moiety in the same direction, according to the principle of least motion,[17] would then give 8 a.

In conclusion, we have developed an enantioselective synthesis of spiroindenes from the oxidative annulation of α-aryl cyclic 1,3-dicarbonyl compounds (or their enol tautomers) with alkynes, using chiral cyclopentadienyl rhodium catalysts. The process tolerates a wide range of substrates to give diverse products containing all carbon-quaternary stereocenters with high enantioselectivities. Application of these chiral complexes in other classes of C–H functionalization/oxidative annulation is underway, and these results will be reported in due course.