Iron-Catalyzed Oxyfunctionalization of Aliphatic Amines at Remote Benzylic C–H Sites

Abstract

We report the development of an iron-catalyzed method for the selective oxyfunctionalization of C(sp³)–H bonds in aliphatic amine substrates. This transformation is highly selective for benzylic C–H bonds that are remote (i.e., at least 3-carbons) from the amine functional group. High site selectivity is achieved by in situ protonation of the amine with trifluoroacetic acid, which deactivates more traditionally reactive C–H sites that are α-to nitrogen. The scope and synthetic utility of this method are demonstrated via the synthesis and derivatization of a variety of amine-containing biologically active molecules.

Graphical Abstract

C–H bond functionalization reactions serve as powerful methods for streamlining the synthesis and late-stage modification of complex organic molecules. Aliphatic amines are particularly common functional groups in natural products and pharmaceuticals, comprising more than 33% of the top 100 selling pharmaceuticals in 2013.¹ As such, aliphatic amine-containing substrates represent important targets for selective C–H functionalization methods.² The transition metal-catalyzed C(sp³)–H oxygenation of aliphatic amines has been particularly well-studied, as these transformations mimic cytochrome P450 metabolism.³ The electron rich C–H bonds that are α-to nitrogen are typically the most reactive sites in these molecules. Thus, the treatment of unprotected aliphatic amines (e.g., 1 in Figure 1) with metal catalysts/oxidants most commonly results in α-oxidation to afford mixtures of amide, imine, and enamine products (e.g., A, B and D).⁴ These can then undergo further transformations such as hydrolysis and/or oxidation to generate secondary products such as C, E, and F.⁵

Figure 1.

Treatment of amines such as 1 with catalysts/oxidants leads to complex mixtures of C–H oxidation products.

Due to the high reactivity of the α-C–H bonds of aliphatic amines, it can be challenging to selectively functionalize at C–H sites that are remote to nitrogen (e.g., H^B in 1).⁶ Traditionally, remote C–H oxidation has required either: (i) the use of a directing group to deliver the catalyst to a remote C–H site⁷ or (ii) the use of a protecting group on nitrogen to attenuate the reactivity of the α-C–H bonds.⁸ We hypothesized that the strong preference for α-oxidation could be overridden using a much simpler approach, namely in situ protonation of the amine nitrogen.⁹ Protonation converts the electron-donating amine into an inductively electron-withdrawing ammonium salt.¹⁰ This should deactivate proximal C–H sites, thereby enabling selective oxidation remote to nitrogen.^11,12

Early precedent for this approach was disclosed by Asensio, who demonstrated the remote C(sp³)–H oxyfunctionalization of a small set of protonated aliphatic amines with TFDO.¹³ More recently, we and others have leveraged this protonation strategy to achieve transition-metal catalyzed remote oxyfunctionalization of a more diverse variety of nitrogen-containing substrates, using K₂PtCl₆,⁹ Fe(PDP),¹⁴ and Mn/Ru-based catalysts. ¹⁵ However, all of these reactions exhibit limitations with respect to substrate scope and/or selectivity. Furthermore, most require catalysts derived from noble transition metals and/or relatively expensive supporting ligands. Herein, we demonstrate the use of simple Fe salts in combination with picolinic acid to catalyze remote C(sp³)–H oxidation of protonated amines. This method is particularly effective for selective oxidation of remote benzylic C–H bonds in aliphatic amines, a substrate class that has not been addressed with previous catalysts/methods. This Letter describes the development, optimization, scope, and applications of this method in a variety of aliphatic amine-containing substrates.

Our test substrate for this transformation (1) contains both 1° and 2°-C–H bonds α-to nitrogen (H^A1 and H^A2) as well as benzylic C–H bonds remote to nitrogen (H^B). Furthermore, it is a 3° amine, and thus is not amenable to many of the protecting group/directing group strategies for remote oxidation that have been reported in the literature.^7,7 We first explored Gif-type^16,17 C–H oxidation conditions, using FeCl₃/picolinic acid as the catalyst and ^tBuOOH as the oxidant.¹⁸ When the unprotonated amine 1 was treated with 10 mol % of FeCl₃, 25 mol % of picolinic acid and 3 equiv of ^tBuOOH (TBHP) in a solvent mixture of pyridine/acetonitrile, a complex mixture of oxidation products was obtained (Figure 2a). These include multiple products of oxidation α-to nitrogen (e.g., B, F, and di-oxo-products related to A and B) along with traces of the desired benzylic oxidation product 1a. However, the protonation of 1 with CF₃CO₂H prior to treatment under otherwise analogous conditions resulted in the clean formation of 1a as the sole detectable oxidation product in 24% yield (Figure 2b). The high selectivity of this transformation is exemplified by the GCMS trace of the crude reaction mixture (compare Figure 2a to 2b).

Figure 2.

GCMS trace of the oxidation of 1 conducted: (a) without an acid additive and (b) after protonation with CF₃CO₂H.

We next optimized this reaction with respect to Fe salt, solvent, acid additive, catalyst loading, oxidant loading, and oxidant addition procedure. Full details of this optimization can be found in Table S1, and only highlights of these studies are discussed below. The reaction proceeds in similar yield with a variety of simple Fe salts [e.g., FeCl₃, FeCl₂, Fe(OTf)₃, Fe(OAc)₂, Fe(BF₄)₂]. However, without picolinic acid, minimal conversion is observed. The MeCN solvent can be substituted with H₂O, but the pyridine additive is crucial for reactivity. A variety of strong acids can be used for protonation of the amine (e.g., CF₃CO₂H, HBF₄, HCl, H₂SO₄), and the choice of acid has a relatively minor impact on the reaction yield. Finally, the slow addition of ^tBuOOH improves conversion significantly. Overall, our best conditions involve use of the amine substrate as the limiting reagent in conjunction with 1.1 equiv of CF₃CO₂H, 5 mol % of FeCl₃, 12.5 mol % of picolinic acid in a pyridine/MeCN (1 : 5) solvent mixture with slow addition of ^tBuOOH (70 wt % in H₂O, 18 equiv), resulting in 65% yield of 1a and <5% of other oxidation products, along with 14% remaining starting material (see Figure S1 for a ¹H NMR spectrum of the crude reaction mixture). Product 1a was obtained in 60% isolated yield under these conditions. Notably, attempts to apply the recently reported Fe(PDP)/H₂O₂ system¹⁴ to substrate 1 resulted mainly in decomposition, and none of product 1a was detected (Figure S2). This result highlights the complementarity of the current transformation to existing methods.

We next examined the impact of chain length on reactivity and selectivity (Table 1). For substrates bearing 3-, 4-, or 5-carbon chains between the protonated amine and the aromatic ring (3, 1, and 2, respectively), the benzylic C–H oxidation product was obtained in isolated yields ranging from 20–60%. The lowest yield was observed with product 3a. In this case, the major side product was phenyl vinyl ketone (14% yield), which derives from a retro-Michael reaction of the initial product 3a.¹⁹ In contrast, amine substrates 4 and 5, which contain shorter 2- and 1-carbon chains between the protonated amine and the aromatic ring, were unreactive. This is likely because the benzylic sites in these systems are highly deactivated due to their proximity to the electron-withdrawing protonated amine. We also examined analogues of 4 and 5 that contain benzylic sites at the remote 4-position of the aromatic ring (substrates 6 and 7). In both cases, C–H oxidation occurs selectively at the benzylic site remote to the nitrogen to afford 6a and 7a in 63% and 60% yield, respectively.

Table 1.

Catalytic oxidation of alkylphenylamines.

substrate	product	isolated yielda
(1)	(1a)	60%
(2)	(2a)	56%
(3)	(3a)	20 (38%)b
(4)	(4a)	no reaction
(5)	(5a)	no reaction
(6)	(6a)	63%c
(7)	(7a)	60%c

^aReactions run under standard conditions [CF₃CO₂H (1.1 equiv), FeCl₃ (5 mol %), picolinic acid (12.5 mol %), ^tBuOOH (70 wt % in H₂O (18 equiv) in pyridine/MeCN (1:5) at rt, 48 h].

^bYield in parentheses determined by ¹H NMR spectroscopic analysis of crude reaction mixture.

^cRun with 12 equiv of ^tBuOOH.

Piperidine derivatives are substrates of particular interest, since piperidine is the most common nitrogen heterocycle in drug molecules (Figure 3).²⁰ Based on the results in Table 1, we anticipated that C–H sites on the piperidine ring would be relatively unreactive towards C–H oxygenation due to their proximity to the electron-withdrawing protonated amine. Indeed, minimal reactivity was observed when 1-methylpiperidine or 1,4-dimethylpiperidine were subjected to our standard conditions (≤11% yield of the corresponding C–H oxygenation products).²¹ However, piperidine substrates bearing remote benzylic sites underwent clean oxidation to form 8a–15a in modest to good isolated yield and high site selectivity (Figure 3). In the series of substrates with substituted aromatic rings, a higher yield was obtained with an electron donating substituent (p-OMe, 9a) than the relatively electron neutral (p-H, 8a) and (p-F, 10a) substituents. Increasing the chain length between the piperidine ring and the benzylic site furnished 11a in 64% yield. A variety of substituents were tolerated on the piperidine nitrogen, including N-benzyl (12a), N-homobenzyl (13a), and N-allyl (14a). In all three of these examples, oxidation occurred selectively at the remote benzylic site rather than at benzylic or allylic C–H bonds proximal to nitrogen. An N-butyronitrile-containing substituent was also compatible and reacted to form 15a in 47% isolated yield. No evidence for nitrile hydrolysis was observed under these conditions.

Figure 3.

Substrate scope for catalytic oxidation of benzylic C-H bonds. Reactions were all run under standard conditions.

We surveyed a series of other alicyclic amine substrates that are common motifs in pharmaceuticals and/or natural products. The nitrogen of each was alkylated with a side chain containing a potential site for remote benzylic oxidation. As shown in Figure 3, C–H oxygenation proceeded in moderate to good yield for substrates containing piperidine (16a, 62%), pyrrolidine (17a, 68%), 3-azabicyclo[3.1.0]hexane (18a, 45%), cis-octahydrocyclo-penta[c]pyrrole (19a, 48%), 3-azabicyclo[3.2.1]octane (20a, 40%), and 2,3,4,5-tetrahydro-1H-benzo[d]azepine (21a, 47%). This latter result motivated us to examine a derivative of the weight loss drug lorcaserin,²² which contains both 2° and 3° benzylic C–H bonds in the nitrogen heterocycle. However, here again, the remote benzylic oxidation product 22a was obtained selectively in 31% isolated yield. The majority of the mass balance in this system was unreacted starting material.

A final set of studies focused on the application of this transformation to the synthesis and derivatization of amine-containing bioactive molecules (Figure 4). The oxidation of readily available precursor 23 afforded the piperidine alkaloid (±)-sedaminone (23a) in 48% isolated yield. A related transformation was used to prepare several pharmaceuticals from the butyrophenone family, a class of antipsychotic drugs for the treatment of schizophrenia. For example, double benzylic oxidation of 24 delivered lenperone (Elanone V) (24a), in modest 26% yield.²³ Melperone (Buronil) (25a) was obtained in 52% yield from precursor 25. Similarly, deoxy-haloperidol (26a) was formed in 41% yield from precursor 26.²⁴ The late-stage of oxidation of protonated L-687384 (27), a sigma receptor agonist,²⁵ afforded ‘oxo-L-687384’ (27a) in 62% yield. Finally, protonated dehydroabietylamine (28), a cannabinoid receptor agonist,²⁶ underwent remote benzylic oxidation to generate 28a in 27% isolated yield.²⁷ Notably, oxygenation at this particular benzylic position has been shown to result in bioactive dehydroabietylamine derivatives.²⁸

Figure 4.

Applications toward synthesis and derivatization of bioactive molecules. Reactions were run under standard conditions unless otherwise indicated. Reaction run with ^a10 mol% of FeCl₃ and 54 equiv of ^tBuOOH over 5 d; ^b6 equiv of ^tBuOOH over 24 h; ^c3 equiv of ^tBuOOH over 24 h. ^dIsolated as the HCl salt.

In summary, this paper describes the development of an Fe-catalyzed method for selective oxidation of remote benzylic C–H bonds in aliphatic amine substrates. Our approach takes advantage of amine protonation, a strategy that electronically deactivates proximal C–H bonds, and thereby renders remote oxidation feasible and selective. Notably, even C–H bonds that are doubly activated (i.e., those that are benzylic/allylic and also α-to the amine nitrogen) are sufficiently deactivated by amine protonation such that oxidation occurs selectively at remote benzylic sites. Applications to the synthesis and late-stage modification of amine-containing natural products and pharmaceuticals were successfully demonstrated. These results suggest that the method could prove useful in the late-stage derivatization of bioactive amine-containing molecules.