Metal-Dependent Umpolung Reactivity of Carbenes Derived from Cyclopropenes.

Metal carbenes, divalent carbon species, are versatile intermediates that enable novel synthetic pathways. These species exhibit either electrophilic or nucleophilic character, depending on the carbene and metal fragments. Although the metal carbene reactivity is regulated by the metal, the umpolung of carbene reactivity by changing metal remains challenging. Here, we report a unique metal-induced de novo umpolung of carbene reactivity, wherein a carbene precursor can be transformed into either an electrophilic carbene or a nucleophilic carbenoid, depending on the metal promoters. Thus, a chemodivergent reaction of isatins and cyclopropenes is developed. Under the promotion of Zn2+ halides, a nucleophilic zinc carbenoid is formed and trapped by isatins to produce oxindole derivatives containing an alkenyl halide moiety. Using Rh2(esp)2 as a catalyst, the reaction delivers oxindoles carrying a dihydrofuran unit. This work provides a facile approach to harness the metal carbene reactivity and is critical for the development of diversity-oriented synthesis.

Introduction

Transition metal carbenes and carbenoids, which are highly reactive and versatile intermediates, have inspired and stimulated a number of research activities in chemistry (Dorwald, 1999, Moss and Doyle, 2014). These intermediates can participate in diverse chemical reactions, including C-H and X-H (X = O, N, Si, B, P, etc.) insertions, cyclopropanations, cycloadditions, and ylide formation and further transformations. Beyond the typical carbene reactions, many unique conversions have been reported in recent decades, for example, carbene migratory insertions (Xia et al., 2017) and gold-carbene-mediated annulations (Obradors and Echavarren, 2014). These studies can enable powerful synthetic pathways in diversity-oriented synthesis, total synthesis, and pharmaceutical process development, making this field dynamic for development purposes (Bertrand, 2002, Chiu, 2005, Bien et al., 2018).

The diverse reactivity profile of transition metal carbenes originates from their unique structures of a divalent carbon atom with two unshared valence electrons, paired or unpaired, with a broad range of different reactivities and diverse substituents (Grubbs et al., 2003). Typically, these complexes can be simply classified as Fischer carbenes and Schrock carbenes (alkylidenes), of which the former is often considered electrophilic and the latter is generally nucleophilic (Dötz and Stendel, 2009, Schrock, 2002, Mindiola and Scott, 2011). The borderline between traditional Fischer and Schrock carbenes is the non-heteroatom-stabilized carbene bound to late transition metals (Figure 1A) (de Frémont et al., 2009), which is usually electrophilic at the carbene center in contrast with the Schrock carbene. This kind of carbene, with intermediate characteristics and reactivity profiles, has emerged as one of most attractive research topics to discover new transformations (Dorwald, 1999, Moss and Doyle, 2014). Another reactive intermediate that exhibits the reaction characteristics of a carbene without the necessary divalent carbon center is the carbenoid (Figure 1B) (Closs and Moss, 1962, Gessner, 2016), which possesses a leaving group and a metal connected to the same carbon, displaying both electrophilic and nucleophilic characteristics. The unique and diverse structural characteristics of these carbene species comprises the foundation of diverse reactivity profiles.

Figure 1

Metal Carbene Intermediates and Their Reactivities

(A) Non-heteroatom-stabilized carbene: electrophilic.

(B) Carbenoid: ambiphilic.

(C) This work: metal-induced de novo umpolung of carbene reactivity.

Generally, due to the distinctly different structural features of different types of carbene and carbenoid species, it is quite challenging to generate more than one type of species from the same precursor, and different metals only modulate the level of electrophilicity (or nucleophilicity) rather than reversing the polarity (Cheng and Doyle, 2016). As an exceptional example, 3,3-diphenylcyclopropene was converted to an electrophilic rhodium carbene intermediate and a nucleophilic Schrock carbene complex by Wang (Zhang et al., 2015b) and Grubbs (Johnson et al., 1993), respectively, but the latter was not used as a synthetic intermediate for further transformations. Considering this fact, we envisioned the controllable formation of both electrophilic and nucleophilic carbene species from the same reactant via alteration of the metal, followed by divergent interception of these intermediates, which would enable the discovery of novel chemodivergent reactions. Herein, we report the metal-induced de novo umpolung of carbene reactivity (Figure 1C), in which a carbene precursor (cyclopropene) could be transformed to either an electrophilic carbene or a nucleophilic carbenoid, depending on the metal catalysts. Furthermore, trapping these electrophilic and nucleophilic carbene species affords structurally diverse molecules in a single step. This work provides an efficient strategy to harness the reactivity of metal carbenes and is critical for the development of diversity-oriented synthesis.

Diazo compounds are the most convenient and widely used carbene precursors owing to their high reactivity and diverse structural features (Zollinger, 1995, Doyle et al., 1998). However, in the absence of an electron-withdrawing group adjacent to the diazo moiety, such as diazo alkanes or alkenes, the compound suffers severe stability and safety issues (Battilocchio et al., 2016, Greb et al., 2017), which greatly limit its access and applications. As an alternative strategy, non-diazo carbene precursors have attracted considerable interest over the last decades (Jia and Ma, 2016, Ma et al., 2016, Wang and Wang, 2019). Cyclopropene, a reliable and easy-to-handle precursor, could generate vinyl carbene in a safe, mild, and practical way with a 100% atom economy via transition metal-catalyzed ring-opening rearrangement (Rubin et al., 2007, Archambeau et al., 2015, Vicente, 2016, Benitez et al., 2009). Furthermore, the unique vinyl functionality could offer new opportunities to discover new transformations. Thus, we depict here the differentiated reactivities of vinyl carbene derived from cyclopropene with zinc or rhodium complexes as promoters (Figure 1C). For the zinc halide-promoted reaction, the generated ambiphilic zinc carbenoid (Pasco et al., 2013, Nishimura et al., 2015), which is the key intermediate in the Simmons-Smith (SS) reaction (Denmark et al., 1991, Denmark et al., 1992), shows a nucleophilic character and undergoes nucleophilic attack to isatins without elimination of the halogen atom, delivering oxindole derivatives 3 containing a synthetically valuable alkenyl halide moiety. Importantly, despite the theoretical nucleophilicity, the nucleophilic reactivity of the zinc carbenoid without elimination of halogen atoms has never been achieved (Knochel et al., 1989, Retherford et al., 1989), which provides unique access to alkenyl halides using inexpensive and non-toxic zinc halides as halogenating agents under very mild conditions. On the other hand, in the case of rhodium catalysis, the reaction forms an electrophilic Rh-carbene, followed by an ylide formation and trapping process (Guo and Hu, 2013, Zhang et al., 2015a) to give product 4. Remarkably, although we have developed various electrophilic trapping processes of active ylide intermediates (Guo and Hu, 2013, Zhang et al., 2015a), this is the first interception of the active ylide without an α-carbonyl group that is deemed essential for stabilization and trapping of the ylide. Overall, this controllable metal-induced de novo umpolung of carbene reactivity presents an efficient approach for chemodivergent synthesis.

Results and Discussion

Optimization of Reaction Conditions

We commenced our study by exploring the reaction of 3-hydroxymethyl-3-phenylcyclopropene (Rubina et al., 2004, Selvaraj et al., 2014) 1a with isatin 2a under the activation of various metal catalysts in different reaction conditions. When the reaction was conducted with catalytic ZnCl2 (0.1 equiv.) (González et al., 2015) in CH2Cl2, the reaction only resulted in a trace amount of 3a (Table 1, entries 1 and 2), but increasing the loading of ZnCl2 (2.0 equiv.) gave rise to 3a in 84% yield with a 92:8 diastereomeric ratio (dr) and complete E-selectivity in 10 min (Table 1, entries 3–6). Further optimizations did not improve the results (Table 1, entries 7–11). Interestingly, when catalytic Rh2(esp)2 was selected as the catalyst in CH2Cl2, dihydrofuryl 3-hydroxyl oxindole 4a, the trapping product of oxonium ylide, was obtained in 45% yield with 73:27 diastereomeric ratio (dr), whereas other metal catalysts, such as Rh2(OAc)4, Rh(COD)Cl2, or (PPh3)AuNTf2, did not provide detectable amounts of product (Table 1, entries 12–15). The yield of 4a was increased to 60% after screening the solvents, indicating methyl tert-butyl ether (MTBE) as the optimal solvent (Table 1, entries 16–20). The divergent reaction pathways switched by the catalyst or reagent will enhance the utility of this reaction in organic synthesis.

Table 1

Optimization of Reaction Conditions for the Divergent Reaction of 1a and 2a

Entry	Metal Complex	Solvent	Time	Yield of 3a (%)a,b	drc	Yield of 4a (%)a,b	drc
1	ZnCl2 (0.1 equiv.)	CH2Cl2	5 h	<5	–	–	–
2	ZnCl2 (0.5 equiv.)	CH2Cl2	5 h	18	92:8	–	–
3	ZnCl2 (1.0 equiv.)	CH2Cl2	1 h	47	92:8	–	–
4	ZnCl2 (1.5 equiv.)	CH2Cl2	10 min	57	92:8	–	–
5	ZnCl2 (2.0 equiv.)	CH2Cl2	10 min	87 (84d)	92:8	–	–
6	ZnCl2 (3.0 equiv.)	CH2Cl2	10 min	76	92:8	–	–
7	ZnCl2 (2.0 equiv.)	CHCl3	10 min	74	94:6	–	–
8	ZnCl2 (2.0 equiv.)	(CH2Cl)2	10 min	71	94:6	–	–
9	ZnCl2 (2.0 equiv.)	toluene	10 h	41	–	–	–
10	ZnCl2 (2.0 equiv.)	n-haxane	10 h	<5	–	–	–
11e	ZnCl2 (2.0 equiv.)	CH2Cl2	10 min	67	92:8
12	Rh2(OAc)4 (5.0 mmol%)	CH2Cl2	12 h	–	–	<5	–
13	Rh2(COD)Cl2 (5.0 mmol%)	CH2Cl2	12 h	–	–	<5	–
14	Rh2(esp)2 (5.0 mmol%)	CH2Cl2	2 h	–	–	45	73:27
15	AuPPh3NTf2 (5.0 mmol%)	CH2Cl2	2 h	–	–	<5	–
16	Rh2(esp)2 (5.0 mmol%)	CHCl3	2 h	–	–	38	70:30
17	Rh2(esp)2 (5.0 mmol%)	(CH2Cl)2	2 h			28	70:30
18	Rh2(esp)2 (5.0 mmol%)	THF	20 h	–	–	41	75:25
19	Rh2(esp)2 (5.0 mmol%)	MTBE	2 h	–	–	62	75:25
20	Rh2(esp)2 (2.5 mmol%)	MTBE	3 h	–	–	60	75:25

dr, diastereomeric ratio; COD: 1,5-cyclooctadiene; esp: α,α,α′,α'-tetramethyl-1,3-benzenedipropionic acid; THF, tetrahydrofuran; NMR, nuclear magnetic resonance.

Ratio of substrates, 1a:2a = 2:1.

Yields are determined by 1H NMR spectroscopy using 1,3,5-trimethoxybenzene as the internal standard.

Determined by 1H NMR analysis of the crude mixture.

Isolated yield.

Ratio of substrates, 1a:2a = 1.5:1.

Substrate Scope

We then investigated the scope of substrates under the promotion of zinc halides (Figure 2). First, ZnCl2, ZnBr2, and ZnI2 were tested, and all gave corresponding halide three-component products 3a-3c in good yields (82%–89%) with dr up to 94:6. Notably, the introduction of a vinyl halide moiety greatly improved the synthetic utility of the desired products 3 because they are beneficial for further transformations through coupling reactions to prepare more functionalized molecules.

Figure 2

Scope of the Reactions Induced by Zinc Halides (top) or Rh2(esp)2 (bottom)

a Yields are determined by 1H NMR spectroscopy using 1,3,5-trimethoxybenzene as the internal standard.

Next, we assessed the substituents on cyclopropenes. Electron-withdrawing groups at the para-position (4-F, 4-Cl, and 4-Br) and meta-position (3-Br) of the aryl group were tolerated and afforded the desired products (3d∼3g, 3k) with equally good results (76%–91%, up to 94:6 dr) except for the compound with a 2-Br functionality (3h). The substrates with a dichloro-substituted phenyl or p-Tol also worked well to provide the corresponding product 3i-3j in 93%–95% yield with 95:5 diastereoselectivity. Moreover, when the free hydroxyl group of cyclopropene was capped by a methyl group, the reaction proceeded smoothly as well (3l). In addition, blocking the hydroxyl group with an acetyl (Ac) or removal of the oxygen functionality from cyclopropene had no deleterious effect on the yield and selectivities of 3m-3n, although an extension of the reaction time to 12 h was required. The remarkable rate acceleration of alcohol and ether substrates should be attributed to a complex-induced proximity effect (CIPE) (Denmark et al., 1992, Beak and Meyers, 1986).

Subsequently, the scope of isatins was also examined. Delightedly, various substituents on the aromatic ring of isatins, regardless of whether the substituent was chloride, bromide, fluoride, methyl, or methoxyl at the C4-, C5-, C6-, or C7-positions, were tolerated to afford 3o-3w in excellent yields with high diastereoselectivity (up to 94% yield and 95:5 dr, respectively). With regard to the N-substitution, both N-methyl and N-acetyl isatins were transformed into the corresponding halide alkenyl oxindoles 3x or 3y in good yield with a high dr value. Moreover, this transformation was also tolerant of the N-unprotected isatin to produce the desired product 3z in 78% yield with a diminished dr of 76:24.

Finally, we also studied the scope of the Rh2(esp)2-catalyzed reaction of cyclopropenes with isatins. As the retro-aldol reaction of 4 occurred during silica chromatography isolation, the reaction yield was determined by crude 1H nuclear magnetic resonance imaging. As shown in Figure 2, 4a and 4b were obtained in moderate yields with acceptable dr values. To stabilize the product, crude 4a, 4c, and 4d were methylated using MeI/NaH in a one-pot manner to provide the stable products 4a’ (52%, 73:27 dr), 4c’ (62%, 78:22 dr), and 4d’ (77%, 62:38 dr), respectively. Moreover, the relative configuration of 4 was determined by single-crystal X-ray diffraction analysis of 4b. To examine the electrophilic reactivity of the rhodium vinyl carbene, cyclopropanation, a classical reaction of electrophilic metal carbenes, was conducted to afford the corresponding cyclopropane 5 in 65% yield. Furthermore, treatment of 1a with Rh2(esp)2 in CH2Cl2 resulted in dimer 6 in 78% yield via a C-H insertion/cyclopropanation sequence, which was suppressed by using MTBE as the solvent in the reaction of 1 and 2.

Transformation of Products

To demonstrate the synthetic utility of this reaction, a gram-scale synthesis of 3b and 3c was achieved in 76% (96:4 dr) and 90% yield (55:45 dr), respectively. The alkenyl halides 3b and 3c were then used as coupling partners for further transformations (Figure 3). For example, Pd-catalyzed cross-coupling of 3c with 4-methoxyphenylboronic acid, n-C4H9ZnBr, TMSC≡CH, or vinyltributyltin at 25°C gave cross-coupling products 7a-7d in moderate to excellent yields of 60%–90%.

Figure 3

Gram-scale Synthesis and Derivatization of Products

Mechanistic Discussion

In a seminal study of zinc halide-catalyzed transformations with cyclopropene by López and Vicente (González et al., 2015) and Doyle (Deng et al., 2016), an electrophilic zinc vinylcarbene A (Figure 5) was proposed as the key intermediate, whereas for the diazo-involved SS reaction (Wittig and Schwarzenbach, 1959, Goh et al., 1969, Crumrine et al., 1975, Lévesque et al., 2014) and a theoretical study by Bernardi and Bottoni (Bernardi et al., 2000), the halogen of zinc carbene would further transfer to the carbon atom to form ambiphilic zinc carbenoid, the actual SS reagent (Denmark et al., 1991, Denmark et al., 1992). To obtain further insight into the properties of the proposed intermediate of our reaction, a competing reaction was conducted in which 1.0 equiv. of 4-methylstyrene was added to a mixture of 1a and 2a under the standard conditions of zinc catalysis (Figure 4A). The reaction yielded 63% 3b, accompanied by 32% yield of cyclopropane Z-5. Although the cyclopropane product is considered to be generated from the electrophilic zinc vinylcarbene A according to López and Vicente (González et al., 2015), it should be the ambiphilic zinc vinylcarbenoid, which presents unique nucleophilic reactivity in this reaction, that leads to adduct 3. Furthermore, the treatment of 4a with ZnBr2 under the standard conditions was not able to give 3e′, negating the possible pathway that 3 was derived from ZnBr2-promoted ring-opening or bromination of 4 (Figure 4B). According to this observation and the formation of 3l-3n, we hypothesized that the ambiphilic zinc vinylcarbenoid C is the key intermediate in our research. As for the rhodium-catalyzed process, the formation of 5 and 6 (Figure 2), as well as the reported process by Cossy (Archambeau et al., 2015), supported rhodium vinylcarbene as the intermediate.

Figure 5

Mechanistic Proposals

(A) Zinc-promoted reaction.

(B) Rh2(esp)2-catalyzed reaction.

Figure 4

Control Reactions

(A) Competing reaction.

(B) Conversion of 4a under the standard conditions of zinc catalysis.

Mechanistic Proposal

Based on the control reactions and the discussions above, a proposed reaction pathway is depicted in Figure 5. For the zinc promotion process (Figure 5A), zinc halide coordinates to cyclopropene 1 and induces ring-opening rearrangement to generate a zinc vinyl carbene A or the cyclic B, in which the oxygen functionality coordinates to zinc(II) and greatly accelerates the rate of the subsequent process via the CIPE. Subsequently, halogen migration of B results in the ambiphilic zinc carbenoid C, which undergoes nucleophilic addition to isatins 2 via a six-membered transition state (Vabre et al., 2015) TS-1 to afford alkenyl halide adduct D that gives rise to the final product 3 during workup with water. For the rhodium-catalyzed process (Figure 5B), Rh2(esp)2 promotes cyclopropene 1 to generate carbene E, which converts to the cyclic oxonium ylide F or the more stable G. Finally, nucleophilic addition of intermediate G to isatins 2 leads to trapping of the product 4 along with the regeneration of the rhodium(II) catalyst. This is the first report on the trapping of an active ylide without an α-carbonyl group that is considered indispensable for stabilizing the proposed intermediate (Guo and Hu, 2013, Zhang et al., 2015a).

Limitations of Study

Zinc fluoride is not effective for the zinc-promoted process.

Conclusion

We reveal a unique de novo umpolung of carbene reactivity via alteration of the metal. Based on this process, a unique chemodivergent aldol-type reaction of isatins with 3-hydroxymethyl-3-arylcyclopropenes is achieved, wherein cyclopropene as a carbene precursor can be converted to either an electrophilic rhodium carbene or a nucleophilic zinc carbenoid. Trapping of these carbene species allows for the facile, rapid, and efficient synthesis of structurally diversified oxindole derivatives with a synthetically important alkenyl halide moiety or a dihydrofuran unit in good yields with high stereo- and chemoselectivities. Significantly, the ambiphilic zinc vinyl carbenoid generated from cyclopropene and zinc halides undergoes a rare nucleophilic addition to electrophiles, which provides an efficient approach to E-selective alkenyl halides from inexpensive and non-toxic zinc halides under mild conditions. Moreover, electrophilic trapping of gem-halovinylzinc can extend the utilities of SS intermediates. This study provides an efficient approach to harness the reactivity of metal carbenes, therefore enriching the versatile carbene chemistry.

Methods

All methods can be found in the accompanying Transparent Methods supplemental file.