Synthesis of Biaryl Atropisomers via Site-Selective C–H Functionalization

Abstract

We describe an approach to form conformationally rigid atropisomers with a variety of nucleophiles not commonly applicable to transition-metal-catalyzed methods. The use of organic photoredox catalysis renders this method operationally simple, as direct substrate oxidation followed by nucleophilic attack may furnish the products via site-selective C–H functionalization in moderate to quantitative yields. Density functional theory (DFT) computations estimated the rotational barriers and half-lives of the products.

Graphical Abstract

INTRODUCTION

Atropisomerism, a type of conformational chirality that arises from hindered rotation about a σ bond, has in the past decades become increasingly prevalent in both industry and academia (Figure 1).^1,2 These biaryl scaffolds have provided chemists the ability to access enantioselective transformations across a range of reaction classes and have found widespread industrial applications.³ Approximately 15% of all U.S. Food and Drug Administration (FDA)-approved small molecules contain at least one atropisomeric axis, and from 2010 to 2018, that number rose to ~30% (Figure 1).^1,4 The inclusion of atropisomers in active pharmaceutical ingredients (APIs) primarily arises from an improved understanding that freely rotating atropisomers do not exist in “flatland” but rather occupy significant chemical space, as they can sample 360° of rotational freedom.⁵ However, the motifs along the biaryl framework can only bind to the given target in a subset of these conformations.⁶ As a result, limiting rotation by generating more sterically rigid axes has been leveraged to lock the substituents in the desired conformations.⁷

Figure 1.

Atropisomeric molecules in chiral catalysts and pharmaceuticals.

Given the increasing importance of atropisomers, strategies to access this biaryl motif have garnered significant interest in the past few decades.^8,9 However, introduction of a C–H functionalization-based strategy would be ideal, as it makes use of unfunctionalized, easily accessed starting materials (Figure 2A). Despite this appeal, such transformations commonly feature at least one of the following: directing/pre-activating groups, non-earth-abundant transition-metal-based catalysts, and a limited coupling partner scope.¹⁰ The final limitation is particularly evident considering that the vast majority of reported methodologies are incapable of C–H amination, particularly on unactivated arenes.¹¹ For methods that begin with a configurationally stable axis, ortho-hydroxy functionalization remains a common approach, which limits the product to bisphenol-based directing groups.¹² With respect to desymmetrization approaches that generate a stable axis, a broader directing group scope (aldehydes, pyridines, amines, etc.) may be achieved with Pd-, Rh-, and Ir-based catalysts; however, the coupling partner scope remains relatively limited.^13,14 Given these shortcomings, a desymmetrization-based methodology capable of accessing rigid atropisomers with multiple axial stereocenters from a broad array of unactivated coupling partners may expedite access to diverse pharmaceutically relevant biaryl compounds.

Figure 2.

Arene and biaryl C–H functionalization approaches.

Since 2012, our group has demonstrated the synthetic utility of acridinium salts as potent single-electron oxidation catalysts for a range of synthetic tranformations.¹⁵ To this end, in 2015, our laboratory disclosed the use of acridinium-based photoredox catalysts for the C–H amination of arenes with azoles and ammonia (Figure 2B).¹⁶ The well-established mechanism of this transformation proceeds via the intermediacy of aromatic cation radicals, generated by excited state acridinium, that undergo nucleophilic addition. This general mechanistic paradigm allowed us to achieve cyanation, amination (primary amines), alkylation, and (radio)halogenation of (hetero)-aromatics.^17-20 The result of this body of work is a general platform in which a range of nucleophiles may be used to accomplish a site-selective C–H substitution without the need for pre-functionalization of the arene substrate, directing groups, or a strong inherent steric/electronic bias within the substrate.^21-25 We hypothesized that this mode of reactivity may be useful to access biaryl atropisomeric compounds via desymmetrization and site-selective C–H functionalization with a broad array of nucleophilic coupling partners (Figure 2C). Herein, we disclose C–H functionalization reactions of configurationally unstable biaryls in the ortho position to generate rigid atropisomers.

RESULTS AND DISCUSSION

Previous reports from our group demonstrated that 1,3-dimethoxybenzene substitutes primarily at the 4 position. Computations showed that this position features the greatest difference in natural population analysis (NPA) values between the cation radical and neutral species, making this position the most electrophilic and, therefore, probable site of functionalization as the cation radical. We presumed this selectivity may be leveraged with biaryl derivatives of the substrate to accomplish an overall desymmetrization reaction and generate configurationally stable atropisomers.²⁵ Acridinium photocatalyst A proved competent with pyrazole as the nucleophilic coupling partner, forming compound 1 in 78% yield with complete positional selectivity, despite five distinct sites for C–H functionalization (Figure 3). Compound 1 was additionally synthesized on a preparative scale using both batch and flow protocols. To determine the configurational stability of the biaryl axis, density functional theory (DFT) computations were conducted to estimate the rotational barrier of enantiomerization (ΔG^‡); the corresponding half-life (t_1/2) for racemization at 23 °C was then calculated from this value (see the Supporting Information for details). Through this, we estimated compound 1 to be a rigid atropisomer with a rotational barrier of 31.3 kcal/mol and t_1/2 for a racemization of 220 years.

Figure 3.

C–H functionalization scope for biaryl atropisomer formation. The B3LYP-6-31G′(d,p) level of theory was used to estimate rotational barriers (ΔG^‡) and corresponding half-lives (t_1/2) at 23 °C. ^aIsolated yield of the gram-scale flow reaction. ^bIsolated yield of the gram-scale batch reaction. ^cPhotocatalyst A (5 mol %), TMSCN (4.5 equiv), and 10:1 MeCN/phosphate buffer at pH 7 (0.1 M). ^dNo (2,2,6,6-tetramethylpiperidin-1-yl)oxyl (TEMPO). ^eSynLEDs (465–470 nm). ^f–OTBS was substituted with ─Me for computational purposes. ^gMeCN (0.1 M).

Further exploration of the reaction revealed that a range of azoles proved to be capable nucleophiles (2–4), affording configurationally stable atropisomers (ΔG^‡ > 29.8 kcal/mol) in moderate to good yields. Cyanide could also be employed as a nucleophile, although the calculated barrier of rotation for the product biaryl (ΔG^‡ = 27.3 kcal/mol) rendered this stereocenter considerably more labile (t_1/2 = 91 days) likely due to the small steric nature of the cyano group. The final class of nucleophiles explored were primary amines, with several examples proving suitable for the transformation, despite concerns of unselective C–H amination through aminium cation radical formation. Notably, amino-acid-derived valine methyl ester hydrochloride furnished compound 8 as a 3:2 mixture of inseparable diastereomers in 40% yield. While biaryl 8 features a rotational barrier exceeding 31 kcal/mol, smaller nucleophiles delivered products (5–7) calculated to have t_1/2 values on the order of days due to a lack of steric bulk associated with the groups on the ortho positions of the biaryl axis. For this reason, a derivative of the model substrate bearing a ─C(Me)₂OTBS motif was synthesized as computations suggested that it may be sterically bulky enough to generate higher rotational barriers. Biaryl 9 was formed in good but still slightly lowered yields compared to compound 1; however, it featured a rotational barrier predicted to be greater than 37 kcal/mol, corresponding to a half-life of over 7000 millennia. Atropisomer 10 also featured an exceptionally high barrier. Compound 11 was the most notable, as it was formed via nucleophilic addition of ammonia, one of the smallest nucleophiles demonstrated in the scope. Despite its size, the product was still predicted to feature a barrier of nearly 32 kcal/mol.

Given our success in using 1,3-dimethoxyphenyl-substituted biaryls to generate stable atropisomers with a variety of nucleophiles, we sought to explore further variation of the upper ring substitution pattern. Initial attempts at using an electron-rich upper arene resulted in a poor yield and off-target C–H functionalization (see the Supporting Information). This was presumably due to increased delocalization of the charge density on the cation radical, lowering the overall reactivity of the intermediate and allowing for reactivity at undesired positions.²⁵ For this reason, we explored electron-neutral and electron-poor upper ring substitution patterns. Compound 12 was formed in excellent yield; however, it remained freely rotating as the nitrogen lone pair of isoquinoline proved too small to generate a stable axis of chirality.²⁶ Increasing substitution in an 8-bromonapthalene derivative restores conformational stability (ΔG^‡ = 27.1 kcal/mol) in product 13 while demonstrating tolerance for a potential cross-coupling handle. ortho-Substituted esters (14 and 15) and protected amines (16) were also well-tolerated, with the latter proving notable, as aniline-derived aromatics commonly inhibit C–H substitution due to stabilization of the cation radical charge density, rendering them less reactive.²⁷ Biaryls 15 and 16 demonstrated C4 addition being the preferred regioisomer, which was relatively surprising, as this site of functionalization had only been observed in a trace yield throughout our substrate scope. While NPA would still predict C2 as the preferred site of addition, we hypothesize that the inclusion of large substituents at both ortho positions of the upper ring (not present in compounds 1–14) prevents nucleophilic addition to the most electrophilic site. This results in addition at C4, the second most electrophilic position of the cation radical. Despite this, the regioisomers are still separable by column chromatography, allowing access to the desired substitution pattern.

We next sought to probe the substitution patterns on the lower aromatic ring (Figure 4). We found that shifting the methoxy groups to the 2,6 positions allowed for a meta-selective desymmetrization (17 and 18) with moderate-to-high rotational barriers similar to those of the starting material. Targeting the same position, mesityl-derived compound 19 was formed in a quantitative yield; however, reversed stoichiometry was required, as diamination was competitive and consumed the desired product rapidly. Overaddition of the nucleophile was not unique to compound 19, as it was observed in other substrates but was either less prevalent or could be reduced with a shorter reaction time. We also sought to incorporate an electron-deficient pyrimidine-derived aromatic, as it deviates from a desymmetrization reaction and may be deprotected to a biologically relevant biaryl uracil derivative (20). The pyrimidine-containing biaryl starting material has a redox potential of E_p/2 = +2.38 V versus saturated calomel electrode (SCE), over 300 mV outside the range of the highly oxidizing acridinium photocatalyst (E*_red = +2.15 V versus SCE) used for the prior substrates.²⁸ To address this, we hypothesized that triphenyl pyrylium tetrafluoroborate (TPT) may serve as a compatible photoredox catalyst due to its sufficiently high excited state reduction potential (*E^1/2_red = +2.55 V versus SCE).²⁹ With a slight increase in catalyst loading and a weak external oxidant [tert-butyl hydrogen peroxide (TBHP)], we isolated and deprotected the anticipated C–H amination product to form the desired rigid biaryl uracil derivative 20 in 67% overall yield (ΔG^‡ = 32.0 kcal/mol).

Figure 4.

Biaryl substrate scope, lower ring. The B3LYP-6-31G′(d,p) level of theory was used to estimate rotational barriers (ΔG^‡) and corresponding half-lives (t_1/2) at 23 °C. ^aArene/nucleophile (3:1). ^bTBHP = tert-butyl hydrogen peroxide.

Finally, DFT computations suggested that an extraordinarily stable C–N axis of chirality [ΔG_C–N^‡ = 45.6 kcal/mol and t_1/2 (23 °C) = 8.0 × 10³ eons] could be formed via coupling of 2-methylbenzimidazole with the model substrate (Figure 5). The reaction itself proved successful, generating the desired product in 54% yield as a 1:1 mixture of separable diastereomers (±21a and ±21b). A crystal structure of compound ±21b was acquired, allowing us to confidently identify its relative stereochemical configuration as well as the orientation of compound ±21a by a process of elimination. No epimerization of compound ±21b to compound ±21a was observed after 4 months on the benchtop or resubmission to the reaction conditions, lending support for the rotational barrier and half-life computational estimates. With this result, we demonstrated that this method is amenable to forming both C–C and C–N atropisomers with high rotational barriers.

Figure 5.

Two axes of chirality, C–C and C-N. The B3LYP-6-31G′(d,p) level of theory was used to estimate rotational barriers (ΔG^‡) and corresponding half-lives (t_1/2) at 23 °C. X-ray structure details: 50% ellipsoid contour probability, CCDC 2417600. ^aPrepared under standard reaction conditions as a 1:1 mixture of separable diastereomers in 54% combined yield.

CONCLUSION

We have developed a method for the synthesis of rigid atropisomers featuring both C–C and C–N axial chirality capable of incorporating several types of nucleophiles without the need for transition metals or pre-installed directing/activating groups. The ortho and meta positions of starting biaryls were shown to undergo efficient C–H functionalization with a variety of nucleophiles previously demonstrated to be successful in past reports from our group. The resultant atropisomers and their computationally derived rotational barriers and half-lives were calculated for each compound and demonstrate this method’s ability to form rigid atropisomers featuring biologically relevant motifs. It is our hope that the synthetic utility of this transformation will afford ready access to more structurally diverse atropisomers. Investigations into rendering this transformation atroposelective are ongoing.

Data Availability Statement

The data underlying this study are available in a published article and online Supporting Information.