Catalytic Atroposelective Dynamic Kinetic Resolutions and Kinetic Resolutions towards 3-arylquinolines via S_NAr

Abstract

Herein we report the catalytic atroposelective syntheses of pharmaceutically relevant 3-arylquinolines via the nucleophilic aromatic substitution (S_NAr) of thiophenols into 3-aryl-2-fluoroquinolines mediated by catalytic amounts of Cinchona alkaloid-derived ureas. These reactions displayed a spectrum of dynamic kinetic resolution (DKR) and kinetic resolution (KR) characters depending upon the stereochemical stability of the starting material. Low barrier substrates proceeded via DKR while higher barrier substrates proceeded via KR. On the other hand, substrates with intermediate stabilities displayed hallmarks of both DKR and KRs. Finally, we also show that we can functionalize the atropisomerically enriched quinolines into pharmaceutically privileged scaffolds with minimal observed racemization.

Introduction

Atropisomerism is a type of conformational chirality that arises from hindered rotation about a σ-bond. It is different than point chirality as bond rotation can lead to racemization, and as such atropisomers can span the gamut of stereochemical stability. Medicinal chemists have wrestled with this issue for some time,^1–3 leading to LaPlante⁴ classifying atropisomers into three classes based on their stereochemical stabilities (Scheme 1), with Class-1 and Class-2 atropisomers racemizing on the second to month time scale at room temperature. Class-3 atropisomers possess barriers to rotation (ΔG_rac) greater than 29 kcal/mol and are considered to be sufficiently stereochemically stable for drug development.

Scheme 1.

(A) Examples of pharmaceutically relevant N-heterocycles. Atropisomeric axes designated with a curly arrow. (B) Atroposelective S_NAr towards 3-arylquinolines.

In addition to being a key attribute in myriad privileged chiral catalysts^5,6 and natural products,⁷ atropisomerism is becoming increasingly leveraged as a design element in medicinal chemistry.^8,9 A recent example is the mutant KRAS inhibitor AMG 510, a Class-3 atropisomer,¹⁰ as well as work from our group on improving the selectivity of kinase inhibitors by leveraging atropisomerism (Scheme 1A).^11,12

Due to the increasing importance of atropisomerism, there has been a huge increase in published atroposelective methodologies over the past decade, with most of the interest being placed on biaryl and binaphthyl scaffolds.^13,14 While there have been examples of the atroposelective synthesis of heterobiaryl scaffolds,^15–17 they have been significantly less studied than other classes of atropisomer. This represents a shortcoming in the field, as heterobiaryl scaffolds can perhaps be considered the ‘lifeblood’ of modern drug discovery.¹⁸ There are very few direct synthetic methodologies that provide access to atropisomerically pure aryl-substituted quinolines with the majority of precedence focusing on 1-aryl isoquinolines.^19–23 Recent examples towards atroposelective access towards other quinoline substitution patterns include a kinetic resolution (KR) via transfer hydrogenation of atropisomeric quinolines that allows for access to enantioenriched 5- or 8- aryl quinolines,²⁴ and an atroposelective Friedländer quinoline heteroannulation towards 4-aryl quinolines.²⁵

3-Arylquinolines and related scaffolds are well-represented in drug discovery. A search of the PDB reveals myriad examples of 3-arylquinolines that possess an axis of chirality (Scheme 1A), the majority existing as Class-1 atropisomers. To the best of our knowledge, there are currently no atroposelective routes towards 3-arylquinolines. We hypothesized that we could access enantioenriched atropisomerically stable 3-arylquinolines via an atroposelective nucleophilic aromatic substitution (S_NAr) at the 2-position of the quinoline. This strategy was inspired by the cation-directed desymmetrization of pro-atropisomeric pyrimidines from Smith.²⁶ Our group has applied this approach in the context of KRs to obtain pyrrolopyrimidine-based (PPYs) kinase inhibiting scaffolds.²⁷

3-Arylquinolines represent a different challenge than symmetrical pyrimidines since the starting materials are already atropisomeric, thus as with PPYs, stereoinduction occurs via a KR process if the starting materials are stereochemically stable. If, however, the starting material can racemize faster than the reaction occurs and the product is stereochemically stable, then a dynamic kinetic resolution (DKR) is possible, which would allow for up to near quantitative yields of enantioenriched products. In the previous examples of atroposelective S_NAr, a chlorine was replaced with an S-aryl group, and as chlorine and sulphur are similar in atomic radii, there is little difference in stereochemical stability from the starting material and product. In contrast, if the leaving group is a fluorine, then the increase in steric bulk imparted to the axis can be estimated to be 5 kcal/mol;²⁸ which could allow for DKR. For example, if the starting material is a Class-1 atropisomer that has a ΔG_rac of roughly 24 kcal/mol, the resulting product will be a Class-3 atropisomer with a ΔG_rac of about 29 kcal/mol (Scheme 1B).

To test this hypothesis, we synthesized quinoline 1a (Scheme 1b, R¹=F, R²=Me) and found that it existed as a Class-1 atropisomer and was amenable to S_NAr with thiophenol to give product 2a, which had a ΔG_rac = 28 kcal/mol. We next evaluated a series of bases and quinine-derived catalysts. The quaternary ammonium quinine based catalysts reported in Smith’s and our previous work^26,27 proved ineffective and unselective. Significant catalyst optimization studies led to the free amine - C-9 epi-aminoquinine derived N-iso-propylurea C4, which effected the S_NAr in 75% yield at 10:90 e.r. when carried out in the presence of K₂HPO₄ and thiophenol (see SI, Section F for full details on optimization). We next evaluated other reaction parameters, observing that the enantioselectivities remained largely unchanged, however, yields were improved in some cases. From these studies, which are described in the SI, our optimal reaction conditions were determined to be 20 mol% C4 with 20 equivalents of K₂HPO₄ in 30:70 n-hexanes:m-xylene at 0.1 M.

We then evaluated the scope of this DKR as seen in Scheme 2. In some cases, oxidation of the product sulphides (e.g., 2h) to sulphones (e.g., 3h) using mCPBA was needed to separate products from starting material and assess enantioselectivities. We first examined the effects of S_NAr using different thiophenols, finding that ortho-cresol (to give o-3b after oxidation), para-cresol (to give p-2b) and ortho-chlorothiophenol (to give 2c) resulted in decreases in enantioselectivity and yield compared to that of 2a.

Scheme 2.

[a] S_NAr reactions were performed with 1.0 equiv starting material, 20 mol% C4, 10.0 equiv HSPh, 20.0 equiv K₂HPO₄, and 70:30 m-Xylene/n-Hexanes (0.1 M) at r.t. for up to 80 h. [b] Oxidation to 3 used 1.0 equiv sulphide, 2.1 equiv mCPBA in 0.1 M EtOAc at r.t. for 18 h. [c] E.r.s are average of >2 trials. [d] Overall yields are average of >2 trials. [e] ΔG_rac are an average of >2 trials. [f] Trituration from 80:20 n-Hexanes/DCE, yields are from 1. [g] Trituration from 100% n-hexanes, yields are from 1.

We next evaluated a series of 3-aryl-2-fluoroquinolines using our optimal conditions. We observed that mild electron-donating groups on the 3-aryl (1d), or electron-withdrawing groups (substrates 1e-1g) gave enantioselectivities and yields comparable to 2a (e.g., 2d in 71% yield and 14:86 e.r., and 2e in 63% yield and 13:87 e.r.). We were able to obtain X-ray crystal structures of 2e, finding that the major enantiomer was in the (R_a)-configuration. Changing the R²-group on the quinoline scaffold to a chlorine (1h to give 3h in 79% yield, 12:88 e.r. after oxidation; 1i to give 3i in 90% yield, 17:83 e.r.), or phenyl (1j to give 2j in 86% yield, 16:84 e.r.) gave comparable enantioselectivities to 2a. Finally, we evaluated thiophene substrate 1k as the orientation of ‘5–6’ biaryl axes orientates the substitution of the 5-membered ring away from the axis, resulting in 1k possessing a lower ΔG_rac. Indeed, 1k was amenable to DKR, yielding 2k in 14:86 e.r. and 64% yield while existing as a stereochemically Class-3 atropisomer (ΔG_rac = 29 kcal/mol). Notably, the e.r. of 2k could be enriched to 3:97 e.r. in 47% yield from 1k via trituration. The sulphide products could be oxidized to the sulphone with no loss in enantioselectivity, and many of the sulphone products were crystalline solids, allowing us to enrich enantiopurity via trituration; for example, 3a could be isolated in 6:94 e.r. in 40% yield from 1a, and 3i could be isolated in6:94 e.r. in 42% yield from 1i.

We next applied our optimal S_NAr conditions to ortho-nitro substrate 1l, which possessed a ΔG_rac of ~ 25 kcal/mol (Scheme 3), corresponding to racemization on the hour to day timescale at room temperature. As 1l racemizes on a similar time course as the S_NAr, the reaction could display both hallmarks of DKR and KR.^29,30 Indeed, we isolated sulphide product 2l in 15:85 e.r. in 80% yield, while the starting material 1l was recovered in 63:37 e.r. and 14% yield. To examine this, we used the Fiaud and Kagan equations²⁹ to determine the predicted conversion based on the observed enantiomeric excess (ee) of starting material and product in the context of an idealized kinetic resolution, finding a prediction of 27% conversion. As the observed isolated yield of 80% outperforms the calculated conversion, this reaction has a significant degree of DKR-character, with some KR character, which may have some conversion related effects on the observed enantiopurity. This is represented clearly when overlaid onto the plot of predicted e.r. versus conversion of a KR (See SI, Section J).^30,31 To examine this further, we evaluated other substrates (1m-1o) that we expected to possess similar ΔG_rac to 1l. Using the same analysis described above, we calculated the expected conversions based on the observed enantioselectivities for products 2m-2o and substrates 1m- 1o. While these substrates yielded attenuated enantioselectivities, we observed similar perturbations in isolated yields from the expected conversions for a purely kinetic resolution process.

Scheme 3.

DKR and KR hybrid. S_NAr reactions were performed under the same conditions as Scheme 2. [a] Each substrate was performed 3 times with similar results, however, the reported e.r.s are of one trial. Reported yields are obtained from isolated product and starting material, as an average of >3 trials. [b] To determine the theoretical conversion, the Fiaud equation was used: conv% = 100(ee₁/[ee₁ + ee₂]). See SI for more details.

Finally, we evaluated stereochemically stable substrates (Scheme 4, ΔG_rac > 26.9 kcal/mol), observing pure kinetic resolutions. For example, substrate 1p, which possesses an ortho-OCF₃ group yielded sulphide 2p in 45% isolated yield and 9:91 e.r. with 1p isolated in 83:17 e.r and 54% yield, correlating to an s-factor of 21. ortho-Phenyl substituted 1q yielded an s-factor of 16 in 44% conversion. Other ortho substitutions such as methyl (1r) or chloro substitution (1s, or 1t) yielded s-factors between 4.7 and 15. ortho-CF₃ substituted 1u yielded an s-factor of 27, allowing for the isolation of 2u in 5:95 e.r. albeit in 25% yield. Finally, 1v, which possesses a Ph at the quinoline C-4 position, was less reactive, but yielded a s-factor of 15.

Scheme 4.

Kinetic resolution. [a] Conversions and s-factors are reported as an average of >2 trials and were determined using the Fiaud and Kagan equations: conv% = 100(ee₁/[ee₁ + ee₂]); s-factor, s = ln[(1-conv)(1-ee₁)]/ln[(1-conv)(1+ee₁)]. [b] Reactions were performed on 50 mg substrate. [c] Exemplary e.r.s of starting material and product are included above for one trial. [c] Isolated yields are reported as an average of at least two trials. See SI for more details.

While diaryl sulphides are themselves of great interest in diverse fields,^32,33 we also sought to determine the synthetic applicability of these products (see SI, Section K). As sulphones are known to be good leaving groups for S_NAr,³⁴ we first subjected enantioenriched sulphone 3a (4:96 e.r.) to a two-step amination procedure,²⁷ isolating enantioenriched 2-aminoquinoline 4a, with minimal racemization, in 10:90 e.r. and 64% yield. We next leveraged a two-step strategy,³⁵ wherein the sulphone is subjected to methanol and tert-butoxide at room temperature, followed by a demethylation using BBr₃ to give 2-quinolinones. We observed that the resultant 2-quinolinones possessed lower stereochemical stabilities, resulting in significant racemization when 2a was used, however, compounds with higher ΔG_rac such as 3q proved amenable to this series, allowing us to obtain the 4q and 5q in good yields with no observed racemization from 3q.

Overall, this work provides access to enantioenriched 3-aryl quinolines in synthetically useful yields and enantiopurities, with the sulphide products transformable to pharmaceutically relevant pharmacophores with little racemization in most cases. As these and related heterocyclic motifs are ‘pharmaceutically privileged’, this work can have numerous applications towards biologically active atropisomeric heterocycles.

Scheme 5.

Enantioenriched products that are obtainable from sulphide products.