A general, enantioselective synthesis of β- and γ-fluoroamines

Abstract

In this Letter, we describe a short, high yielding protocol for the enantioselective (87–96% ee) and general synthesis of β-fluoroamines and previously difficult to access γ-fluoroamines from commercial aldehydes via organocatalysis.

Small molecule probes and drug candidates that contain one or more chiral fluorine atoms are commonplace and represent a critical component of the medicinal chemists’ toolbox.^1,2 In the context of amine-containing ligands, introduction of a fluorine atom either β- or γ- to the amine often maintains activity at the desired target while providing a 1–2 log reduction in pK_a. That shift in electronic structure often results in diminished ancillary pharmacology at cardiac ion channels, improved metabolic stability, enhanced pharmacokinetics, and increased CNS penetration.^1–3 In some cases, introduction of a chiral β-fluoroamine, due to the almost 2-log impact on pK_a, can diminish activity at the desired target; however, moving the fluorine atom to the γ-position has been shown to maintain the desired bioactivity, while still affording the benefits of the β-fluoroamine congeners.^1–6

While we and others have made considerable advances in the synthesis of chiral β-fluoroamines,^7–13 the synthesis of γ-congeners still relies on classical DAST approaches which are plagued with rearranged and dehydrated products.^1–6,14 Therefore, new synthetic strategies to access these elusive, chiral γ-fluoroamines were warranted.

Previous work from our lab (Fig. 1) employed a one-pot protocol to access chiral β-fluoroamines via organocatalysis in high yield and % ee (Eq. 1); however, the configurational instability of the incipient α-fluoroaldehyde required its immediate conversion to the β-fluoroamine.^7,8 Later efforts utilized a similar strategy, but employed the analogous α-chloroaldehydes, to access chiral N-terminal aziridines¹⁵ and chiral morpholines and piperazines;¹⁶ however, the configurational instability of the α-chloroaldehyde also necessitated immediate, one-pot use. To provide additional flexibility, we developed an approach (Fig. 1, Eq. 2) that improved yields and % ee for the synthesis of chiral morpholines and piperazines by reducing the α-chloroaldehyde to an alcohol, converting the hydroxyl to a leaving group, displacing the leaving group with an amino alcohol or diamine followed by base-induced cyclization.^16,17

Figure 1.

First generation organocatalytic approach to chiral β-fluoroamines and refined approach to chiral morpholines and piperazines.

These results led us to reflect on the β- and γ-fluoroamine problem, and apply this strategy to their asymmetric construction (Fig. 2). Here, we envisioned standard organocatalytic α-fluorination followed by reduction to the β-fluoroalcohol to provide a bench stable, chiral intermediate. Conversion of the hydroxyl to a leaving group followed by S_N2 displacement with an amine would provide chiral β-fluoroamines 2, whereas a cyanide displacement, followed by nitrile reduction, would afford rapid access to chiral γ-fluoroamines 3.

Figure 2.

Envisioned route to access both chiral β- and γ-fluoroamines via organocatalysis.

To validate this new approach, we prepared a number of β-fluoroalcohol substrates 4a–g under standard conditions in good yields (65–77%) and high enantioselectivity (87–96% ee) as expected from literature precedent (Scheme 1).^7,8,18–20 Conversion to the corresponding triflate, and displacement with benzylamine delivered the desired chiral β-fluoroamines 2a–c (Scheme 2) in high yields (84–96%) and in excellent enantioselectivity (90–94% ee). In essence, this new strategy sets the key stereocenter in a conformationally stabile way, affording the β-fluoroamines with reproducibly high % ee, and higher yields than the first generation chiral β-fluoroamine route.⁷

Scheme 1.

Organocatalytic, enantioselective synthesis of β-fluoroalcohols 4a–g.

Scheme 2.

Synthesis of chiral β-fluoroamines 2a–c.

In some instances, a β,β-difluoroamines have been shown to be an important pharmacophore,^1–7 and we wanted to determine if this new approach would grant access to this moiety as well. Here (Scheme 3), organocatalytic α-fluorination with excess NFSI leads to α,α-difluorination, and reduction affords the β,β-difluoroalcohol. Conversion to the triflate and displacement with benzylamine provides the desired β,β-difluoroamine 5 from the difluoroalcohol in 81% yield, representing another improvement over our first generation methodology.⁷

Scheme 3.

Synthesis of a β,β-difluoroamine 5.

While the ability to access β-fluoroamines was gratifying, our main objective with this approach was to gain access to chiral γ-fluoroamines, a chemotype that is very challenging to prepare in high enantioselectivity.^1–8,14 We began our attempts with a racemic 4a congener and surveyed a variety of leaving groups for cyanide displacement en route to γ-fluoroamines. Interestingly, when tosylate 6 was employed, all attempts (independent of cyanide source, stoichiometry, solvent, and temperature) afforded a 1,2-bis-cyano adduct 7 (Scheme 4). As more forcing conditions were required for tosylate displacement, a competing displacement of the fluoride by cyanide occurred. When triflate 8 was used, excess KCN in refluxing DCM provided the desired β-fluoronitrile 9, but in only 50% conversion accompanied by considerable decomposition. Further surveying of reaction conditions and refinement led to the discovery of optimal conditions (10 equiv KCN in the presence of 20 mol % 18-crown-6 at room temperature in MeCN for 16 h) to fully convert triflate 8 to the desired fluoronitrile 9, without any evidence of cyanide displacement of the fluoride.

Scheme 4.

Optimization of β-fluoronitrile 8 synthesis.

With racemic 9 in hand, we then evaluated a number of reducing agents to provide the γ-fluoroamine. Interestingly, the vast majority of common methods for nitrile reduction (LiAlH₄, DIBALH, H₂/Pd, Ni(0)/NaBH₄) failed to reduce the β-fluoronitrile without considerable decomposition or defluorination. Ultimately, good results were obtained with the milder conditions of InCl₃/NaBH₄,^21,22 delivering the γ-fluoroamines in yields up to 90%.

Having developed a route to racemic γ-fluoroamines, attention was now directed at accessing chiral γ-fluoroamines, an elusive and difficult to prepare pharmacophore. Here, the chiral β-fluoroalcohols 4a–g (Scheme 1) were converted into the corresponding triflates, which were then displaced under the optimized cyanide displacement conditions to deliver chiral β-fluoronitriles 9a–g in yields ranging from 71% to 93% (Table 1). Our optimized reduction protocol with InCl₃/NaBH₄ smoothly provided the chiral γ-fluoroamines 3a–g in excellent yields (73–90%) and enantioselectivity (87–96% ee). The overall yields starting from commercial aldehydes 1 range from 40–58%. Of particular utility of this approach is the commercial availability of both enantiomers of the organocatalyst, enabling either enantiomer of the chiral γ-fluoroamine to be prepared. We were now able to access chiral β-fluoro- and γ-fluoroamines, as well as β,β-difluoroamines, providing a range of finely tuned amine basicity (with pK_as of 10.7 (parent amine) to 9.0 (β-fluoro), to 7.3 (β,β-difluoro) to 9.7 (γ-fluoro)). Attempts to prepare a γ,γ-difluoro congener, to provide an amine substrate with a pK_a of ~8.7 failed, as we were unable to displace the β,β-difluorotriflate with cyanide in yield greater than 10%, despite surveying a broad spectrum of reaction conditions (cyanide source, temperature, solvent, additives).²³

Table 1.

Chrial γ-fluoroamines 10a–g via organocatalysis

Starting material	Yield of 9a–ga	Yield of 3a–gb	% eec
4a	9a, 71%	3a, 87%	94%
4b	9b, 92%	3b, 84%	87%
4c	9c, 73%	3c, 86%	92%
4d	9d, 93%	3d, 83%	96%
4e	9e, 77%	3e, 90%	96%
4f	9f, 82%	3f, 89%	95%
4g	9g, 84%	3g, 73%	96%

^aAll reactions were performed on a 1.0 mmol scale.

^bAll reactions were performed on a 0.5 mmol scale.

^CEnantiomeric excess determined by ¹⁹F NMR using the (R)-Mosher amide of the final amine.

By synthesizing β-fluoronitriles, we envisioned them not only as precursors to γ-fluoroamines, but also as intermediates poised to access a variety of fluorinated scaffolds using the nitrile as a handle. To illustrate this idea, we performed common reactions to exemplify the β-fluoronitrile as a lynchpin providing access to other, difficult to prepare, fluorinated moieties. As seen in Scheme 5, the β-fluoronitrile 10 was used in a [3+2] cycloaddition with sodium azide to provide β-fluorotetrazole 11, a common carboxylic acid bioisostere.²⁴ Additionally, hydrolysis of the nitrile with hydroxylamine provided amide oxime 12, a precursor for oxadiazole synthesis.²⁵ Overall, the chiral β-fluoronitrile linchpin offers rapid entry to a wide range of valuable fluorinated functional groups with subtle perturbations of pK_a and electronic properties.

Scheme 5.

Synthesis of a β-fluoro tetrazole and a β-fluoro amide oxime.

In summary, we have developed a powerful extension of our one-pot, chiral β-fluoroamine work, that overcomes issues related to variable % ee due to configurationally unstable α-fluoroaldehydes, by a two-pot protocol that affords β-fluoroamines in high yields and reproducibly high enantioselectivity (90–94% ee). A further extension allows access to highly elusive and previously difficult to prepare chiral γ-fluoroamines using a three-pot protocol in good overall yields (40–58%) and excellent enantioselectivities (87–96%) from commercial aldehydes. Importantly, outside of classical DAST chemistry, this work represents the only other approach for the enantioselective synthesis of γ-fluoroamines, an important pharmacophore in drug discovery and development. Moreover, the chiral β-fluoronitrile linchpin offers rapid access to a wide range of valuable fluorinated functional groups with subtle perturbations of pK_a and electronic properties. Additional refinements are under development and will be reported in due course.