Enantioselective Synthesis of Complex Carbocycles by C–H Insertion of Aryl/Aryl Carbenes

Abstract

A range of benzene and indole-fused carbocyclic molecules were accessed by the enantioselective C–H insertion of rhodium aryl/aryl carbenes. 1,4, 1,5, and 1,6 C–H insertion reactions are chemoselective for insertion into carbon centers appended with an electron-donating heteroatom. Examples include a range of ethers, a free hydroxyl group, and a range of nitrogen substituents, including basic amines. DFT calculations support a step-wise mechanism of this process and provide insight into the origin of both stereoselectivity and the preference for C–H versus O–H insertion.

Graphical Abstract

Introduction

Aryl/aryl carbenes, sometimes referred to as donor/donor carbenes,1 have emerged as attractive reagents due to their selectivity and functional group tolerance.^2,3 Aryl/aryl carbenes resist attack from heteroatom nucleophiles, expanding their range of utility compared to more electrophilic carbenes. Moreover, accessing aryl/aryl carbenes does not require high dilution, cryogenic temperatures, or the exclusion of ambient oxygen or water. Our group has demonstrated that aryl/aryl carbenes are phenomenal reagents for stereoselective intramolecular C–H insertion reactions. Previous work harnessing aryl/aryl carbenes to form 5- and 6-membered heterocycles proved high-yielding and stereoselective. While these methodologies are useful for accessing heterocycles, asymmetric carbocycle formation remains limited.^4–7

Substituted benzene-fused carbocycles, including benzocyclobutenes, indanes, and tetralins, are privileged scaffolds found in many compounds of biological interest (Figure 1).^8–12 The synthesis of smaller carbocycles, especially cyclobutanes and benzocyclobutenes, is challenging due to increased ring strain. The preparation of benzocyclobutenes historically has involved the use of strong acids and bases or UV light, often proceeding through highly reactive benzyne intermediates.^13–21 These harsh conditions can severely limit functional group tolerance and require strict air and moisture free techniques. Metal carbenes have been introduced as an alternative method of accessing this motif, but current methods still utilize highly reactive metal carbenes and thus lead to competitive side reactions.^22–23 While indane and tetralin synthetic methods are more plentiful and often milder, diastereo- and enantioselective reactions of these motifs remain scarce. A methodology utilizing aryl/aryl carbenes capable of accessing carbocycles in a stereoselective manner would find numerous applications in total synthesis and drug discovery.

Figure 1.

Examples of drugs and natural products containing substituted indanes and tetralins.

In previous work we have shown that the regioselectivity of aryl/aryl carbene C–H insertion reactions can be controlled through the placement of an “activating group” substituent (Figure 2), typically an electron-donating heteroatom (O, N, S). This group stabilizes the carbocation intermediate formed after hydride transfer to the metal-carbene center (the first step of C–H insertion), activating adjacent C–H bonds. If the activating group could be reliably excluded from the ring, it would open the door for novel stereoselective routes to new carbocyclic natural product families using aryl/aryl carbenes. The exocyclic activating group could also be used as a functional handle, increasing its utility in targeted synthesis. To achieve these transformations, we envisioned several strategic placements of the activating group, depending on the desired ring size. The activating group should block unwanted C–H insertion sites which lead to heterocyclic products, while leaving the preferred C–H bonds available to react.

Figure 2.

C–H insertion from aryl/aryl carbenes.

Our methodology was effective in furnishing stereoselective 4-, 5-, and 6-membered carbocycles under mild conditions, leading to a diverse array of insertion products in moderate to high yields. Notably, insertions proceeded smoothly in the presence of free hydroxyl groups, which to our knowledge, is completely unprecedented. Additionally, insertions with basic amines were achieved for the first time; traditionally, nitrogen-containing substrates require resonance stabilization through amide or carbamate groups to prevent catalyst inhibition. These results show an unprecedented level of functional group tolerance, not only for C–H insertion chemistry broadly, but aryl/aryl carbenes as well. Finally, we demonstrated the utility of our methodology for the asymmetric construction of biologically relevant scaffolds by synthesizing the 6,6,5, tricyclic core found in ergot alkaloid natural products such as agroclavine (Figure 1).

Methods

Full details of the methods and materials used in this study are reported in the Supporting Information.

General Procedure for one-pot hydrazone oxidation and C-H Insertion.

To a flame-dried vial were added hydrazone and CH₂Cl₂ (0.0125 M). MnO₂ (8.0 equiv) was then added and the mixture was reacted for 1–2 h. Upon full conversion of the hydrazone, a color change from clear to magenta was observed and Rh₂L₄ (0.01 equiv) was added. After stirring for another 1–16 h, the crude reaction mixture was filtered over celite to remove MnO₂, concentrated under reduced pressure and the crude material was purified by flash column chromatography on silica gel to obtain the desired product. Density functional theory (DFT) calculations were performed at the SMD(DCM)-PBE0-D3(BJ)/def2TZVPP//PBE0-D3(BJ)/def2-SV(P) level of theory using Gaussian 16 C.0.1.^24–28 Comprehensive conformational searches on Rh₂(MesCO₂)₄ were carried out using xTBCREST.²⁹ The TBSO group was simplified to TMSO throughout mechanistic modeling. Structural drawings were created with CYLview.³⁰ See SI for details on NMR calculations.

Results and Discussion

We began our investigation by attempting to design an optimal substrate to access benzocyclobutenes. By using an activating group to block 1,5- insertion products, we are left with competition between 1,4- and 1,6- C–H insertion. To differentiate the sites electronically, we increased the substitution of the 1,4- position to better stabilize the carbocation intermediate that results from hydride transfer. Our first experiments employed methyl ether 7a, which can, ostensibly, form either an isochroman (8a’) or a benzocyclobutene (8a) (Table 1). Given the reduced reactivity of methyl ethers toward C–H insertion, we expected competition between these two pathways, ideally favoring the smaller ring. Treatment of 7a with MnO₂ and Rh₂(MesCO₂)₄ sequentially resulted in the exclusive formation of benzocyclobutene 8a in a 67% yield with >95:5 dr (the structure of which was confirmed by X-ray crystallography, Table 2). Upon further optimization we found that the yield increased when the reaction time was reduced to 1h, although they did not improve when the reaction was conducted in solvents other than CH₂Cl₂ (see SI). A screen of various rhodium tetracarboxylate catalysts (Figure 3.) revealed that Rh₂(MesCO₂)₄ provided the highest yield and diastereoselectivity. When a series of chiral catalysts were examined, no enantioselectivity was observed.^31–34

Table 1.

Catalyst Screening of 1,4 Insertion^a

Entry	Substrate	[Rh] Cat.	Yieldb	drc	erd
1	rac-7a	Rh2(MesCO2)4	75%	>95:5	-
2	rac-7a	Rh2(esp)2	57%	>95:5	-
3	rac-7a	Rh2(OAc)4	-e	-	-
4	rac-7a	Rh2(PTCC)4	60%	>95:5	50:50
5	(R)-7a	Rh2(PTCC)4	66%	>95:5	52:48
6	(R)-7a	Rh2(R-PTAD)4	65%	>95:5	46:54
7	(R)-7a	Rh2(S-PTAD)4	60%	>95:5	47:53

^aReaction Conditions: 7a (0.1 mmol), MnO₂ (0.8 mmol), solvent (6 mL), [Rh] (1 mol%).

^bIsolated yield.

^cdr was determined by ¹H NMR of unpurified reaction mixture.

^der was determined by chiral HPLC of purified product.

^eMultiple products in ¹H NMR spectrum.

Table 2.

Benzocyclobutene Substrate Scope^a

^aReaction Conditions: 7 (0.1 mmol), MnO₂ (0.8 mmol), CH₂Cl₂ (8 mL), Rh₂(MesCO₂)₄ (1 mol%); isolated yields; dr was determined by ¹H NMR of unpurified reaction mixture.

^bRh₂(RPTAD)₄ was used as the catalyst.

^cReaction was conducted at −20 °C to rt.

Figure 3.

Structures of rhodium tetracarboxylate catalysts.

Enantioenriched hydrazone (7a) was subjected to C–H insertion conditions, and surprisingly the resulting product was nearly racemic. We believe this could be the result of a reversible hydride transfer that enables bond rotation of the zwitterionic intermediate and hence, low enantioselectivity.

Examining the substrate scope of the reaction (Table 2), we see that benzocyclobutene products with various aliphatic substituents could be obtained in good yields and diastereoselectivities (8a-c). X-ray crystallography of 8a revealed that the aryl and alkoxy groups are trans. No carbocation rearrangement product was observed for cyclopropyl-substituted substrate (8c), despite the tertiary carbon adjacent to the site of carbocation formation. Replacing the methyl group with a phenyl group at the 1,4 position led to decreased diastereoselectivity (8d and 8g). We attribute this lower selectivity to the increased steric clash between cis-phenyl substituents. We then tested whether swapping out the methoxy substituent for an ethoxy group would still favor 1,4 C–H insertion over 1,6 C–H insertion. Substrate 8e yielded a minority of the isochroman and 8g gave exclusively 1,4 C–H insertion. This result suggests that the superior stabilization of the in-situ generated positive charge by the tertiary carbon is more determinant than the increased steric demand of making a 4-membered ring over a 6membered ring. A phenoxy-substituted product was also isolated in moderate yield with high diastereoselectivity (8f), despite being a worse activating group, due to resonance with the oxygen lone pair. Altering the electronic nature of the carbene center gave the desired electron-rich (8i) and electron-poor (8j) products in comparable yields and high diastereoselectivity. 8h, featuring an acetal group, could not be obtained under standard conditions. We have previously demonstrated that the substrate that we expected to form 8h actually forms o-QDM directly by an elimination mechanism after hydride transfer.³⁵

Our success with benzocyclobutenes encouraged us to apply this method to form larger carbocycles. While larger rings feature less ring strain, kinetic selectivity is a liability. 1,6- C–H insertion could be especially difficult due to the strong kinetic preference for 1,5- C–H insertion. Accessing 6- membered carbocycles would be a battle between C–H activation and kinetic favorability.

The first substrate tested underwent a 1,6 C–H insertion to access a tetralin (11a), in high yield and diastereoselectivity (Table 3). We then screened the reaction with several dirhodium catalysts (see SI), wherein high reactivity and diastereoselectivity could be observed for most catalysts. The best enantioselectivity was obtained using Rh₂(R-PTAD)₄³⁶ and Rh₂(S-TCPTTL)₄³⁷ with a 97:3 and 6:94 er, respectively. The former was chosen to further explore substrate scope.

Table 3.

Substrate Scope for the Formation of Oxygen-Substituted Carbocycles^a

^aReaction Conditions: 9 (0.1 mmol), MnO₂ (0.8 mmol), CH₂Cl₂ (8 mL), Rh₂(R-PTAD)₄ (1 mol%); isolated yields; dr was determined by ¹H NMR of the unpurified reaction mixture; er was determined by chiral HPLC.

^bRh₂(S-TCPTTL)₄ was used as the catalyst.

^cRh₂(SBTPCP)₄ was used as the catalyst.

^dThe structure and configuration were further determined by X-ray structure.

By shortening the carbon chain of our hydrazone precursor, indane (10a) was synthesized in a 96% yield, albeit with slightly lower diastereo- and enantioselectivity than its tetralin counterpart. Substrates with altered alkoxy groups were synthesized to evaluate the effect of varying size and rigidity of the activating group on the reaction. Moderate to excellent yields and high stereoselectivity were observed for all substrates (10b-10d and 11b-11d), with no discernable trends related to size or rigidity. Acetals afforded the desired C–H insertion products 10e and 11e in excellent yields and enantioselectivity. Secondary ethers were suitable; however, the enantioenriched hydrazone was required to achieve high enantioselectivity for tetralin 11f. These results are in agreement with a previous study from our group demonstrating substrate control for chiral substrates.³⁸ The structure of 11f was confirmed by X-ray crystallography and revealed that in contrast to what was observed for benzocyclobutenes, the phenyl group is syn to the methoxy group.

Next the efficacy of non-alkoxy activating groups was explored. Silyl ethers 10g and 11g were prepared in good yields, affording products with a functional group handle for further diversification. Higher diastereo- and enantioselectivity was observed for indane (10g) over tetralin (11g). This opposes trends observed for the methyl ethers, possibly due to the increased bulk from the TBS group. To test the chemoselectivity of the reaction, we designed substrates containing a hydroxyl group that enables a competing O–H insertion reaction pathway (10h, 11h). We observed exclusive C–H insertion products in good yields with moderate to excellent stereoselectivity. These results highlight the unprecedented functional group tolerance of aryl/aryl carbenes in contrast to their more reactive donor/acceptor counterparts.^39–42

We then varied the electronic character of the carbene by including alterations to R¹ (Table 3). A slight decrease in yield was observed when R¹ is an electron-rich phenyl ring (10i and 11i). Minimal changes were seen for tetralins when R¹ was electron-poor (11j) while indanes exhibit an increase in stereoselectivity (11j). We were also able to confirm the synconfiguration of 11j by X-ray crystallography. Replacing the phenyl group with a methyl or a CF₃ resulted in a significant decrease in stereoselectivity (11k and 11l), and the isolated yields were much lower due to the formation of azine and styrenyl byproducts. Electron-rich heterocycles were also tolerated. Nicotinyl 11n and thiophene 11m were isolated in reduced yields but featured high stereoselectivities. Synthesis of several 7- and 8-membered rings were also attempted. Benzyl- ether protected substrates (10o and 11o) offer two sites with activated C–H bonds, leading to either smaller 5- and 6membered carbocycles, or larger 7- and 8-membered heterocycles, respectively. While the larger rings benefit from activation by two sources, the smaller-kinetically favored carbocycles were afforded in both cases. Attempts at 7 membered carbocycles were made by further extending the carbon chain of the hydrazone precursor. The reaction of methyl-ether 12a yielded exclusively the 5-membered ring, revealing that the kinetic preference for 1,5 insertion dictates the reaction outcome rather than the activating group in this case. Reaction of 12g leads to a complex reaction mixture without any insertion observed. 12h was also isolated as the 5membered ring albeit in a reduced yield. Results from quantum chemical calculations for the C–H insertion of 12h (below) reveal a larger activation barrier for 1,7 than 1,5 C–H insertion (vide infra).

Nitrogen-containing compounds, especially those containing basic nitrogen atoms, have been challenging to obtain by C–H insertion with aryl/aryl carbenes.^4,5 Previous work to access insertions with nitrogen-containing activating groups required the use of amides and carbamates to be efficacious. We initiated our study with hydrazones containing a morpholine moiety, which inductively withdraws electron density from the nitrogen (Table 4). Insertion products 14a and 15a were observed in moderate yields and moderate to high stereoselectivity. Pyrrolidine scaffolds were also tolerated, affording tetralin 15b in moderate yields with high stereoselectivity. Compounds containing less-basic nitrogen atoms 15c, 15d and 15e were obtained in moderate-to-high yields and stereoselectivity. The structure of 15c was confirmed by X-ray crystallography and was consistent with prior results. Boc and tosyl protecting groups provided insertion products for methylamines (15d and 15e) respectively; these protecting groups can be easily cleaved for further functionalization.

Table 4.

Nitrogen-Containing Carbocycles Substrate Scope^a

^aReaction Conditions: 13 (0.1 mmol), MnO₂ (0.8 mmol), CH₂Cl₂ (8 mL), Rh₂(R-PTAD)₄ (1 mol%); isolated yields; dr was determined by ¹H NMR of the unpurified reaction mixture; er was determined by chiral HPLC.

^bTHF was used as the solvent.

^cRh₂(S-TCPTTL)₄ was used as the catalyst.

^der was determined using an NMR chiral shift reagent.

^eThe structure and configuration were further determined by X-ray structure.

Motivated by the positive results with benzene-fused carbocycles, we attempted to cyclize an indole-containing substrate to obtain the tricyclic core found in ergot alkaloids. We were pleased to see that our initial substrate (17a), afforded the desired tricyclic core in a 94% yield, 90:10 dr, and 97:3 er (Table 5). This reaction also exhibits the impressive chemoselectivity seen earlier, proceeding with an unprotected indole nitrogen (17b and 17c) and free hydroxyl group (17c). We sought to replicate our success with nitrogen-containing activating groups on an indole to afford products that further resemble ergot alkaloid natural products. The desired carbocycles (17d and 17e) were obtained in good yields. Although the diastereoselectivity is lower than that of the oxygen-containing substrates 17a and 17b, the major products 17d and 17e are trans diastereomers. Most C–H insertion reactions of aryl/aryl carbenes lead to cis products. While trans products have been reported for particular substrate/catalyst pairings, they are rare.⁴² It is likely this switch to trans selectivity is related to the fused-indole backbone’s rigidity, but it is currently unclear why this is exclusive to the nitrogen-containing indoles. In our previous multivariate linear regression (MLR) study, geometric parameters describing the products were correlated with stereochemical outcomes.⁴² This method enables analysis of substrate-catalyst interactions without direct calculation of the transition state energies (see SI for more details). The reduced steric demand of the oxygen substituents results in a preference for cis products 17a and 17b. Further studies are underway to understand the origin of this reversal and apply this method to the synthesis of ergot alkaloid natural products and analogs.

Table 5.

Indole-Containing Carbocycles Substrate Scope^a

^aReaction Conditions: 16 (0.1 mmol), MnO₂ (0.8 mmol), CH₂Cl₂ (8 mL), Rh₂(R-PTAD)₄ (1 mol%); isolated yields; dr was determined by ¹H NMR of the unpurified reaction mixture; er was determined by chiral HPLC.

^bRh₂(S-TCPTTL)₄ was used as the catalyst.

^cTHF was used as the solvent.

^dThe structure and configuration were further determined by X-ray structure.

Density functional theory (DFT) calculations were performed at the SMD(DCM)-PBE0-D3(BJ)/def2TZVPP//PBE0-D3(BJ)/def2-SV(P) level of theory using Gaussian 16 C.0.1 to elucidate the mechanism of carbocycleforming C–H insertions catalyzed by dirhodium complexes.^24–28,30 Comprehensive conformational searches on Rh₂(MesCO₂)₄ using xTB-CREST were conducted to ensure the lowest energy conformers were computed for all stationary points.²⁹ The results reveal a low-energy structure in which the catalyst does not maintain C₄ symmetry, indicating the flexibility of Rh₂(MesCO₂)₄ despite its rigid ligand groups (see SI for catalyst conformations). The TBSO group was simplified to TMSO for structure modeling, and the Rh-carbene intermediate was used as the reactant for computations to focus on the selective C–H insertion step. The computational results reveal that the hydride shift and subsequent ring closure may occur via a concerted or stepwise process. The stepwise mechanism with a shallow zwitterionic minimum on the potential energy surface (see IRCs in SI) is consistent with our previous work on S_E2 mechanisms for C– H insertions of aryl/aryl carbenes.³⁸

5-Membered ring formation, modeled with substrate 10h (similar results were found for OCH₃ and OTMS systems; see SI), shows a preference for forming the cis product with a barrier approximately 2 kcal/mol lower than that for formation of the trans product, consistent with experimental observations. The results also indicate a clear energetic preference for hydride transfer (C–H insertion) over nucleophilic attack of the oxygen lone pair (O–H insertion). The predicted barrier for 5-membered ring formation via O– H insertion is 20 kcal/mol, while the barrier for C–H insertion is only 4.3 kcal/mol, highlighting a significant difference in reactivity (see SI for details).

Increasing ring size leads to higher predicted barriers for C–H insertion. As shown in Figure 4, visualized using CYLView 1.0,³⁰ the favored cis C–H insertion free energy barrier increased from 4.3 kcal/mol for 10h to 6.8 kcal/mol for 11h to 12.3 kcal/mol for 12h. Despite the formation of 5membered 12h, formation of the 7-membered ring product 12h’ was also investigated, and a higher barrier was found (>14 kcal/mol; see energy profile in SI). The geometries of the transition structures for 11h and 12h reveal a beneficial O–H hydrogen bond between the hydroxyl groups of the substrate and the carboxylate ligands on the catalyst, with O•••H distances ranging from 1.59 Å to 1.89 Å (see SI for structures). Despite this stabilizing interaction, 7-membered ring (12h’) formation still has a higher barrier than 5-membered ring (12h) formation due to unfavorable strain associated with bringing the shifting hydride and carbene center into close proximity. We also observed a possible pathway for formation of side products for the 12h’ trans system. Formation of the intermediate structure can be facilitated by the carboxylate ligands on the [Rh] catalyst, as deprotonation of the OH group following the H-shift occurs along the reaction coordinate (Scheme 1). A similar scenario was reported for a combined C–H insertion/Cope rearrangement (CHCR) reaction, where this type of transition state leads to the formation of side products.⁴³ In the case of 12h’, completion of the cyclization reaction involves reverse proton transfer followed by C–C bond formation. Compound 12h’ was not observed in the experiment, and our computational results reveal both that its formation requires a higher barrier and opens up a pathway leading to the formation of side products.

Figure 4.

Transition structures for cis C–H insertion reactions of various alkyl chain lengths (n=1, 2, 3) with OH substituents. Critical bond lengths are labeled in Å, energies are in kcal/mol.

Scheme 1.

Mechanistic Scheme for the Formation of a Deprotonated Intermediate (IM) Along the Reaction Pathway of Substrate 12h. The Computed Relative Free Energies (in kcal/mol) for Each Stationary Point are Labeled

Conclusion

In summary, we have developed a method that enables the stereo-, regio-, and chemoselective synthesis of a variety of fused benzene carbocycles, including benzocyclobutenes, indanes, and tetralins. This method exhibits extremely high functional group tolerance, yielding C–H insertion products in the presence of basic amines, and even free hydroxyl groups. DFT calculations and QSSR analysis were performed to help rationalize features of the reaction outcomes such as C–H versus O–H insertion, competing ring sizes, and alternative reaction pathways. Construction of the 6,6,5 tricyclic core found in ergot alkaloids can be achieved with either high cis or trans selectivity, the latter of which is rare for aryl/aryl carbenes. Future work will utilize this methodology for the targeted synthesis of ergot alkaloid natural products as well as the efficient assembly of carbocyclic compounds for drug discovery.

Supplementary Material

The Supporting Information is available free of charge at “ACS Publications website.”

Experimental procedures, compound characterization and computational data (PDF)