Enantioselective β-C–H Arylation of α-Fluoroalkyl Cyclopropane- and Cyclobutane- Carboxylic Acids

Abstract

Fluoroalkyl-substituted small-membered rings are highly valuable scaffolds in drug discovery; nevertheless, there existed no general methods to prepare them enantioselectively. Herein, we report a highly enantioselective β-C–H arylation of simple and readily available α-fluoroalkyl cyclopropane- and cyclobutane- carboxylic acids via C–H activation. This transformation enabled by chiral monoprotected-N-amino sulfonamide (MPASA) ligand is compatible with a broad range of aryl iodides, affording an efficient route to diverse small-membered rings with all-carbon quaternary centers bearing fluoroalkyls, which have been difficult to access to date. The synthetic utility of arylation product was demonstrated by converting the carboxylic acid into alcohol and amine which may be used to prepare new analogues of bioactive compounds. Preliminary mechanistic studies were also conducted, giving initial insights into catalyst speciation and stability of this reaction.

Graphical Abstract

Modern medicinal chemistry has witnessed the essential role of fluorine in the developments of numerous pharmaceutical compounds.¹ Due to their unique electronic and structural properties, fluorine and fluoroalkyls, when being strategically placed on biological molecules, often offer a significant enhancement in physiochemical properties such as metabolic stability, lipophilicity and membrane permeability. Trifluoromethyl cyclopropane, for instance, has been used as the bioisostere of t-Bu group to improve metabolic stability by suppressing undesired oxidation of C–H bonds.² Several works have been applying this strategy, leading to the discovery of more potent analogues of finasteride, FLT3 inhibitor and BRS-3 angonist (MK-5046) (Figure 1A).^2–4 Difluoromethyl (CF₂H) group, moreover, can act as lipophilic hydrogen-bond donor due to the presence of two strongly electron-withdrawing fluorine atoms.^5,6 Due to this special property, CF₂H group has been exploited as a bioisostere of alcohols and thiols. An important example of utilizing difluoromethyl group as a lipophilic hydrogen-bond donor was the discovery of HCV N3 protease inhibitor in which a stereospecific hydrogen-bonding between CF₂H group and carbonyl group of Leu amino acid residue proved to be a key feature of the potent inhibitor (Figure 1A).⁷ The utility of this strategy was also showcased in the development of glecaprevir and voxilaprevir, which both have a difluoromethyl-substituted cyclopropane ring embedded in their scaffold and are approved by FDA for treatment of HCV infection.¹ Besides trifluoromethyl- and difluoromethyl-substituted cyclopropane rings, fluoroalkyl-substituted cyclobutane rings have also gained more attention from the community of medicinal chemistry recently.^8,9

Figure 1.

Biological values and enantioselective syntheses of small rings bearing fluoroalkyls

Despite therapeutics potentials of chiral small rings bearing fluoroalkyl substituents in drug discovery^3–11, there exists no general enantioselective method to synthesize such motifs to date. Even though tremendous effort has been made in both asymmetric transition metal catalysis and biocatalysis which both employ carbene intermediates, these methods usually required specific type of substrates or cumbersome preparation of catalysts and are not applicable towards construction of quaternary stereocenters bearing fluoroalkyls in broad sense (Figure 1B).^12–20 It is important to note that by employing a different approach, migratory difluorination of styrene with simple chiral aryl iodide catalyst, Jacobsen lab elegantly demonstrated a catalytic enantioselective method towards constructing quaternary stereocenters with difluoromethyl group, albeit only applicable to acyclic systems.²¹

We wondered if enantioselective C–H activation could be employed as an alternative yet distinct approach to prepare these chiral scaffolds (Figure 1C). Cyclocarboxylic acids bearing α-fluoroalkyls are readily available. In addition, the diversity of aryl iodide is also superior to the styrene used in the carbene chemistry. Furthermore, carbene chemistry is limited to the construction of cyclopropane scaffold. Herein, we described the discovery and development of ligand-enabled desymmetrization of α-fluoroalkyl cyclopropane- and cyclobutane- carboxylic acids via arylation reaction, which led to the preparation of four different chiral scaffolds featuring all-carbon quaternary stereocenters bearing fluoroalkyl groups (CF₃ and CF₂H), which have been challenging to synthesize to date.

We initiated our effort towards developing an enantioselective arylation of α-fluoroalkyl carboxylic acid by selecting α-CF₃ cyclopropane carboxylic acid 1 as the model substrate. Even though enantioselective arylation of simple cyclopropane carboxylic acid has been realized with MPAAM and MPAA ligand with aryl iodides and aryl boronic esters respectively^22,23, those conditions did not give successful results with α-CF₃ substituted substrates (See Scheme S1, SI). The poor reactivity of these substrates can be partially attributed to the significantly weakened binding by the α-CF₃ group. It was also observed in our previous study that alcohol substrate with CF₃ substituent performed poorly compared to the ones with simple alkyl substituents.²⁴ We then turned to test other ligand scaffolds, including MPAA derived from β-aminoacid (L3), MPASA^24–26 (L4 and L5) and MPAThiol²⁷ (L6) that have been successfully developed in our group and other laboratories (See Table S1, SI for details). Gratifyingly, all of these ligands showed excellent stereoselectivity (89–98% ee) with MPASA ligand (L4) giving the highest reactivity (65% yield) when 10 mol% Pd catalyst was used. Encouraged by this promising result, we decided to extensively and strategically modified the MPASA ligand L4 in order to find more reactive ligands which could lower catalyst loading and reaction temperature simultaneously without reducing either yield or enantioselectivity. A library of MPASA ligands in which different key elements including amino acid side chain, protecting group of amino moiety and sulfonamide were then systematically altered and tested using low palladium catalyst loading (2 mol%) (Table 1). Varying the side chains of the amino acid backbone did not lead to any improvement in reactivity (L4-L11 and L13-L14). Interestingly, replacing a benzyl group (L11) by a phenyl group (L12) or introducing an oxygen atom into the substituents (L15 and L16) resulted in significant loss in reactivity. Changing acetyl-protecting group to benzoyl-protecting group (L17-L20) proved to be ineffective, giving extremely low yield. Finally, different sulfonamides were incorporated into the MPASA ligand (L21-L25), showing that electron-withdrawing groups such as CF₃ and NO₂, which may enhance binding affinity to carboxylate by increasing electrophilic nature of palladium catalyst, significantly improved the reactivity, up to 64% yield with L24, while preserving the excellent stereocontrol.

Table 1.

Ligand Evaluation in Enantioselective Arylation of α-CF₃ Cyclopropane Carboxylic Acid^a,b

^aConditions: 1 (0.1 mmol), ArI (0.2 mmol), Pd(OAc)₂ (2 mol%), ligand (L) (5 mol%), Na₂CO₃ (0.1 mmol), Ag₂CO₃ (0.1 mmol), HFIP (1.0 mL), 70°C, 24h.

^bYields were determined by ¹H NMR analysis. Enantiomeric excess (ee) determined by SFC.

With the optimal conditions in hand, we set to explore the scope of arylation reaction with respect to aryl iodides using L25 as the standard ligand.²⁸ A wide range of substituents at para- position on aryl iodides were compatible at the optimal reaction conditions, delivering desired products at excellent enantioselectivity (>97% ee in all cases) (Table 2). Aryl iodides bearing electron-donating groups such as methoxyl, alkyl, halides, tosylate and phenyl gave the coupling products in synthetically useful yield (41–60%) (2a-2g). Aryl iodides with electron-withdrawing and chemically reactive groups such as -CO₂Me, -NO₂ and -COMe were compatible, giving the arylated products in consistently high enantioselectivity albeit with slightly lower yield (38–55% yield) (2h-2j). Meta- and ortho-substituted aryl iodides were also tolerated in the C–H arylation reaction (2k-2q). Aryl iodides with meta-halide gave desired products (2k-2l) in moderate yield (39–59%) (99% ee). Electron-withdrawing groups including -CO₂Me, -NO₂ and -COMe were slightly more reactive at meta-substituted position than at para-substituted ones, delivering arylated product at 57–73% yield (2n-2p). The scope of this C–H arylation reaction could also be extended to multi-substituted aryl iodides without loss of stereoselectivity. Aryl iodides with both bromide and other functional groups including ester and methoxy were compatible, affording the coupling products at moderate yield (32–42%) and very high enantioselectivity (>98%ee in all cases) (2r-2t). Indeed, the arylated products with halides, ester, NO₂ and COMe groups could be functionalized further via cross-coupling and reduction, offering a potential pathway to access other chemical space.

Table 2.

Scope of β-Arylation of α-CF₃ Cyclopropane Carboxylic Acid^a,b

^aConditions: 1 (0.1 mmol), ArI (0.2 mmol), Pd(OAc)₂ (2 mol%), ligand (L25) (2 mol%), Na₂CO₃ (0.1 mmol), Ag₂CO₃ (0.1 mmol), HFIP (1.0 mL), 70°C, 24h.

^bIsolated yield. Enantiomeric excess (ee) determined by SFC.

Next, we examined if the arylation reaction was suitable for α-CF₂H cyclopropane carboxylic acid, which is a common precursor to valuable motifs in medicinal chemistry. It is worth noting that the CF₂H group has a significantly more acidic proton than methyl group due to strong inductive effect from two fluorine atoms. This proton might be deprotonated via CMD mechanism, then interrupting the catalytic reaction. To our delight, the arylation reaction was completely suitable for coupling this carboxylic acid substrate to a broad range of aryl iodides with a slightly higher palladium loading (5 mol%), giving products in moderate to excellent yield (41–82%) with exceptional regioselectivity – only one regioisomer was formed – and enantioselectivity (>93% ee in all cases) (4a-4q) (Table 3). Electron-donating groups such as halides and phenyl (4a-4c and 4n) were good coupling partners, delivering products in synthetically useful yield (59–65%) (See SI for details). Electron-withdrawing groups at para- and meta- position (4d-4h and 4i-4o) were all compatible. Heteroaryl iodides could also be coupled with α -CF₂H cyclopropane carboxylic acid, given that their 2-position was occupied to attenuate their binding to palladium (II) catalyst. Even though the arylated products were isolated at moderate yield (51–56%), high enantioselectivity was observed with more than 90% ee for both cases (4p-4q).

Table 3.

Scope of β-Arylation of α-CF₂H Cyclopropane Carboxylic Acid^a,b

^aConditions: 1 (0.1 mmol), ArI (0.2 mmol), Pd(OAc)₂ (5 mol% for aryl or 10 mol% for heteroaryl idodides), ligand (L) (L25 (5 mol%) for aryl or L23 (10 mol%) for heteroaryl idodides) Na₂CO₃ (0.1 mmol), Ag₂CO₃ (0.1 mmol), HFIP (1.0 mL), 70°C for aryl or 100°C for heteroaryl idodides, 24h.

^bIsolated yield. Enantiomeric excess (ee) determined by SFC.

We next explored the feasibility of extending this enantioselective arylation to cyclobutane scaffolds, although α-alkyl cyclobutane carboxylic acids were incompatible with aryl iodides via Pd(II)/Pd(IV) catalytic cycle as shown previously.^22,23,29 Surprisingly, α-CF₃ and α-CF₂H cyclobutane carboxylic acids could be coupled with a wide variety of aryl iodides (Table 4). Simple iodobenzene and aryl iodides with electron-withdrawing and electron-donating at para-position were suitable coupling partners in the arylation reaction with α-CF₃ cyclobutane carboxylic acid, giving the corresponding products in moderate yield (36–54%) and good enantioselectivity (85–89% ee) (6a-6h). Meta-substituted aryl iodides provided the desired products at synthetically useful range (43–58%) and similar level of enantioselectivity (86–89% ee) (6i-6k). The reaction also worked with ortho-substituted and multisubstituted aryl iodides, affording the coupling products at high level of enantiocontrol (6l-6o). α-CF₂H cyclobutane carboxylic acid was showed to be a viable substrate, delivering the desired adducts at slightly lower yield (31–46% yield) and lower stereoselectivity (82–86% ee) (8a-8e).

Table 4.

Scope of β-Arylation of α-Fluoroalkyl Cyclobutane Carboxylic Acid^a,b

^aConditions: 1 (0.1 mmol), ArI (0.2 mmol), Pd(OAc)₂ (5 mol%), ligand (L25) (5 mol%), Na₂CO₃ (0.1 mmol), Ag₂CO₃ (0.1 mmol), HFIP (1.0 mL), 60–70°C, 24h (See SI for details).

^bIsolated yield. Enantiomeric excess (ee) determined by SFC.

Due to increasing frequency of small-membered rings bearing fluoroalkyls in drug discovery, we turned to explore synthetic utilities of arylation adducts. The product 2a could be conveniently converted to alcohol (9) or protected amine (10) by simple reduction with LiAlH₄ or Curtius rearrangement respectively, affording the corresponding products in good to excellent yield (Figure 2A). It is worth to note that the alcohol and protected amine may have potentials to prepare new analogues of pharmaceutical-related products such as VX-659 or BACE1 inhibitor.^30,31

Figure 2.

Derivatization of arylation product and preliminary mechanistic study

We also conducted rudimentary mechanistic studies using α-CF₃ cyclopropane carboxylic acid as the model substrate to gain preliminary insights about reaction mechanism. A non-linear effect (NLE) was not observed, suggesting that monosubstituted catalyst-ligand complex was the only catalyst species formed during the reaction at this standard condition (Figure 2B).³² A same excess experiment was also run, showing an overlap in reaction progress between two different initial concentrations, implying there was no catalyst deactivation in the arylation reaction and further demonstrating the robustness of this reaction with this substrate (See SI for details).³³ Based on preliminary results of mechanistic studies, a reaction mechanism was proposed (Figure 2C). In the presence of MPASA ligand, a monosubstituted catalyst-ligand complex was formed and coordinated by carboxylic acid 1 to form int-1, which then underwent C–H activation to generate five-membered palladacycle int-2. Oxidative addition of int-2 with aryl iodide and subsequent reductive elimination gave product-bound palladium complex int-4. Abstraction of iodide from this complex afforded the desired product and regenerated the active palladium catalyst.

In summary, we have successfully developed a general method towards synthesis of small-membered rings with all-carbon quaternary centers bearing fluoroalkyl groups (CF₃ and CF₂H) in one step directly from readily available starting materials. This method not only is complementary to existing methods employing carbene intermediates but also opens up a new and simple way to access new chemical space that has not been explored extensively to date. The success of this new catalytic system was hinged on the development of MPASA ligands whose reactivity could be tuned conveniently by both sulfonamide and amino acid moieties. This catalytic method featured consistently high level of enantiocontrol across a broad range of aryl iodides, including nitrogen-substituted heterocycles. The arylation products could be conveniently converted to other synthetically useful and important classes of organic compounds such as alcohol and amine. Further studies to uncover mechanistic details and to explore other transformations are ongoing in our laboratory.

Supplementary Material

The Supporting Information is available free of charge at Full experimental details, mechanistic studies and characterization of new compounds (PDF).