A Horseradish Peroxidase–Mediator System for Benzylic C–H Activation

Abstract

Enzyme–mediator systems generate radical intermediates that abstract hydrogen atoms under mild conditions. These systems have been employed extensively for alcohol oxidation, primarily in biomass degradation, but they are underexplored for direct activation of C(sp³)–H bonds in alkyl groups. Here, we combine horseradish peroxidase (HRP), H₂O₂, and redox mediator N-hydroxyphthalimide (NHPI) for C(sp³)–H functionalization of alkylbenzene-type substrates. The HRP–NHPI system is >10-fold more active than existing enzyme–mediator systems in converting alkylbenzenes to ketones and aldehydes under air, and it operates from 0–50 °C and in numerous aqueous-organic solvent mixtures. The benzylic substrate radical can be trapped through a reaction with NHPI, demonstrating the formation of benzylic products beyond ketones. Furthermore, we demonstrate a one-pot, two-step enzymatic cascade for converting alkylbenzenes to benzylic amines. Overall, the HRP–NHPI system enables the selective benzylic C–H functionalization of diverse substrates under mild conditions using a straightforward procedure.

Graphical Abstract

The selective functionalization of C(sp³)–H bonds in alkylbenzenes is of great synthetic interest because of the prevalence of alkylbenzene moieties in pharmaceutically relevant molecules.^1–3 One versatile method for benzylic functionalization is to use redox mediators (Figure 1a). In these approaches, an oxidant converts the mediator—often an N-hydroxy type compound—into the corresponding N-oxyl radical, which in turn performs hydrogen atom transfer (HAT) to generate a substrate radical that can be capped by diverse functional groups, including halogens,^4,5 cyanide,⁶ nitrate,⁷ and alkyl groups.⁸ Benzylic substrate radicals can also react with O₂,^9,10 leading to formation of benzylic ketones or aldehydes that can be subsequently converted to an array of different functional groups, including amines.¹¹ Redox mediator-based oxidation methods using transition metal catalysts typically require high temperatures and O₂ pressures (Figure 1b).^9,12–14 More recently, photocatalytic^15,16 and electrochemical^17–21 methods have enabled the oxidation of redox mediators at room temperature under air for functionalization of diverse substrates (Figure 1c).

Figure 1.

Current benzylic C–H oxidation methods. a) Overview of redox–mediator systems for benzylic functionalization. b) Transition metal catalysis. c) Electrochemistry. d) Laccase–mediator systems. e) This work, combining HRP and the redox mediator NHPI for benzylic C–H activation.

Using enzymes to generate mediator radicals is a promising alternative approach that avoids the need for electrochemical equipment or photo-irradiation and enables uniform distribution of the catalyst throughout a homogeneous reaction solution. In nature, white-rot fungi employ a variety of enzymes and natural redox mediators to oxidatively degrade lignin.²² Synthetic enzyme–mediator systems have been developed for several commercial applications, including lignin degradation and textile bleaching.²³ Particularly prominent are laccase–mediator systems, featuring the combination of laccase enzymes with mediators such as 2,2,6,6-tetramethylpiperidin-1-yl)oxyl (TEMPO), hydroxybenzotriazole (HOBT), and N-hydroxyphthalimide (NHPI). Laccase–mediator systems have been used extensively for benzylic alcohol oxidation.^24–28

Despite progress using enzyme-mediator systems for alcohol oxidation, their use for functionalization of alkylbenzenes is less explored. In one study, Trametes versicolor laccase (TvL) was reported with NHPI to oxidize a small panel of alkylbenzene derivatives to the corresponding ketone or aldehyde, although a pure O₂ atmosphere and a 24-hour reaction time were required (Figure 1d).²⁹ A similar system was also used to oxidize benzylic ethers to the corresponding esters.³⁰ Even though TvL has been reported to convert NHPI to the phthalimide N-oxyl (PINO) radical, laccases have at most a redox potential of ~0.8 V,³¹ while NHPI has a redox potential of 1.05 V,³² indicating that the generation of the PINO radical is thermodynamically uphill. This uphill reaction can in principle proceed, due to the PINO radical being continually depleted from the reaction (through reactions with substrates, for example), but we wondered whether enzymes with higher redox potentials could generate PINO and functionalize alkylbenzenes more efficiently.

We hypothesized that horseradish peroxidase (HRP) would be more promising than TvL for redox mediator-based alkylbenzene functionalization, owing to its higher redox potential. HRP is a heme peroxidase that reacts with H₂O₂ to generate a highly oxidizing iron-oxo intermediate called Compound I (redox potentials ranging from 0.88 to 0.97 V have been reported at pH values ranging from 6 to 7).^33–35 Compound I oxidizes a variety of phenol and aniline substrates to the corresponding radicals, which diffuse away from the active site. The enzymatic intermediate is concurrently converted into Compound II, which is also highly oxidizing (redox potentials ranging from 0.87 to 0.99 V have been reported).^33–35 Compound II oxidizes another equivalent of the phenol or aniline substrate, restoring the enzyme to the resting Fe³⁺ state. HRP has previously been reported to oxidize the mediator HOBT, which forms a radical that is ~3 kcal/mol less reactive than PINO,²⁵ for pollutant breakdown and oxidative degradation of polybutadiene.^36–40 HRP is commercially available, has been previously used in a number of green synthetic applications,⁴¹ and has been shown to tolerate organic cosolvents.⁴²

We began by assessing the ability of TvL or HRP to react with a panel of redox mediators (TEMPO, HOBT, and NHPI) for the conversion of 1-(4-methoxyphenyl)ethan-1-ol (1), a benzylic alcohol, and 1-ethyl-4-methoxybenzene (2), an alkylbenzene, to the corresponding benzylic ketone (3) (Figure 2). Previously, TvL–mediator systems were used to catalyze the oxidation of these substrates.^24,29 To increase the stringency of the comparison and to assess synthetic utility, we used stoichiometric quantities of mediator and an air atmosphere instead of pure O₂. For the HRP reactions, H₂O₂ was included because it is necessary for the activity of HRP.

Figure 2.

Comparison of TvL–mediator and HRP–mediator systems. a) Structures of tested mediators. b) Oxidation of 1-(4-methoxyphenyl)ethanol (1). c) Oxidation of 1-ethyl-4-methoxybenzene (2).

The HRP–NHPI catalyst system effectively oxidized both the alcohol and the alkylbenzene substrates to the corresponding ketone, producing 94% and 61% yields, respectively. The HRP–HOBT system led to substantially lower yields of 20% and 6% for alcohol and alkylbenzene oxidation, respectively. These lower yields may be due to the HOBT radical being less reactive than the PINO radical; the bond dissociation energy (BDE) of the O–H bond in HOBT is 85 kcal/mol, while the BDE of the O–H bond in NHPI is 88 kcal/mol.²⁵ We observed oxidation of benzyl alcohol (16% yield) using the TvL–TEMPO system, but we detected no ketone product when the alkylbenzene substrate was used. This is likely because TEMPO cannot perform HAT on is alkylbenzene substrates.⁴³ The TvL–NHPI system gave low yield (6%) for benzyl alcohol oxidation and no detectable alkylbenzene oxidation, suggesting that HRP was much more effective than TvL at oxidizing NHPI under these reaction conditions.

Having identified the HRP–NHPI system as promising, we proceeded to explore and optimize reaction conditions for converting ethylbenzene derivatives to ketones, using 1-ethyl-4-methoxybenzene (2) as the initial substrate (Figure 3). Omission of HRP, NHPI, or H₂O₂ completely abolished product formation, indicating that all three components are necessary. HRP retained activity in a variety of cosolvents, including DMSO, acetone, acetonitrile, and DMF. Although a higher yield was obtained with acetone, we chose to move forward with DMSO as the cosolvent in order to avoid complications with solvent evaporation. We suspect poor substrate solubility is responsible for the lower yields at 10–30% DMSO, while the drop in yield at 70–90% DMSO is likely due to diminished HRP activity in this solvent system.

Figure 3.

[a] Yields determined by HPLC. [b] Standard conditions: 2 (1 mM), HRP (1 µM), NHPI (1 mM), H₂O₂ (5 mM, added slowly), PBS pH 5.5, 50% DMSO (v/v). 4 mL reaction volume with stirring. 2 h, 25°C. [c] H₂O₂ added at reaction start.

The HRP–NHPI system gave similar yields using temperatures ranging from 0 to 50 °C (Figure 3). Lower reaction temperatures should prove useful for minimizing evaporation of volatile substrates. Strikingly, the yield decreased only slightly with a 10-minute reaction time instead of 2 hours, indicating fast reaction kinetics. Because slow addition of hydrogen peroxide was previously reported to increase the yield of oxidation reactions using peroxygenase enzymes,⁴⁴ and heme peroxidases such as HRP can be inhibited by concentrated H₂O₂ solutions,^45,46 we added H₂O₂ dropwise over the 2-hour window for the reactions in Figure 3. However, similar yields were obtained when all of the H₂O₂ was added at the start of the reaction time.

Sub-stoichiometric loadings of NHPI led to considerable decreases in yield compared to stoichiometric loading under these reaction conditions. We suspect this is due to HRP having a high K_M for NHPI, as has been reported for Trametes villosa laccase.⁴⁷ To assess the K_M, we performed a series of reactions using the fluorescent substrate (6-methoxynaphthalen-2-yl)methanol (Figure S2a).^48,49 By varying the concentration of NHPI while keeping other conditions constant, we can estimate the K_M value from a pseudo-Michaelis Menten plot. Figure S2c indicates the K_M for NHPI is likely in the low millimolar range, hence why a stoichiometric amount of NHPI is necessary to achieve reasonable yields with 1 mM substrate. As shown below in Figure 4b, when higher alkylbenzene concentrations are used, NHPI can play a catalytic role in this reaction, consistent with previous reports on non-enzymatic systems.⁵⁰

Figure 4.

a) Substrate scope for analytical scale aromatic HRP–NHPI oxidation reactions. Yields determined by HPLC. b) Selected scaled-up substrates and isolated yields. See SI for detailed reaction conditions. c) Formation of NHPI adduct as the major product with adjusted reaction conditions. Yields determined by HPLC.

We found that using super-stoichiometric quantities of NHPI also decreased the yield of 3. In this case, we attribute the decreased yield to an increase in the formation of side products, in particular an NHPI-adduct species (see Table S1). We found that 1 mM NHPI was effective in forming the desired product (3) while minimizing the formation of this side product.

We next explored the range of substrates that could be oxidized by HRP–NHPI in analytical-scale reactions (Figure 4a). A variety of substitutions were tolerated at the para position of ethylbenzene derivatives, although yields were modest when the substituents were electron-withdrawing groups. We did not detect any oxidation products corresponding to HAT adjacent to ether groups, such as in substrate 5. The BDEs of C–H bonds adjacent to ether groups tend to be higher than benzylic C–H BDEs, making HAT by the PINO radical less favorable.⁵¹ Substitutions were also tolerated on the aromatic ring at the ortho and meta positions and on the alkyl chain. Specifically, benzylic ethers (18) were converted to esters, as was previously demonstrated with laccases.³⁰ The catalyst system was effective on primary benzylic carbons (9 and 10), affording predominantly the aldehyde product and only small amounts (<6 %) of the carboxylic acid overoxidation product for some substrates (see HPLC chromatograms in supporting information). This finding is consistent with previous reports that the PINO radical can oxidize a variety of primary benzyl alcohols to the corresponding aldehydes.⁴³ Alkyl groups adjacent to pyrrole (15) and pyridine (14) heterocycles were also oxidized. To explore the utility of this method on a pharmaceutically relevant substrate, we evaluated whether the HRP–NHPI system could oxidize ibuprofen and found that the corresponding ketone was the major product.

Substrates corresponding to compounds 3, 11, and 12 were selected for preparative-scale reactions (Figure 4b). Each of the products was isolated in high purity on 100–400 mg scale after filtration and column chromatography (see experimental procedures and NMR spectra in supporting information). The substrate concentration was increased to 10 mM for these preparative-scale reactions to decrease the amount of solvent required, and 0.1 equivalents of NHPI were included. Up to 5 product molecules were obtained per NHPI molecule added to the reaction, demonstrating that NHPI can play a catalytic role in this system.

There are some limitations to the breadth of substrates that can be oxidized using this method. Compounds bearing phenol or aniline functionalities tend to be oxidized by HRP directly, leading to a complex mixture of products, including substrate dimers or oligomers. Additionally, certain alkylbenzene substituents are prone to C–C bond cleavage, consistent with previous reports.²¹ Compounds with tertiary benzylic carbons, such as cumene, are also prone to C–C bond cleavage (Figure S3). However, careful control of reaction temperature and solvent identity can minimize C–C bond cleavage in PINO-based reactions.⁵²

We investigated whether other direct benzylic functionalization could be achieved beyond formation of ketones and aldehydes. Specifically, we wondered whether we could functionalize the benzylic position with NHPI itself through coupling of the PINO radical with a benzylic substrate radical. These types of NHPI adducts are of synthetic interest because they can be converted through nonenzymatic reactions to alcohols, hydroxylamines, amines, or alkenes.^53–55 By increasing the concentration of HRP and NHPI, as well as depleting the O₂ in the system by purging with argon, we generated the benzylic NHPI adduct as the major product (Figure 4c).

Finally, we also explored the conversion of an alkylbenzene to a benzylic amine by combining the HRP–NHPI system with a transaminase enzyme in a two-step, one-pot reaction cascade (Figure 5). We first used the HRP–NHPI system to convert 2-ethylnaphthalene to a benzylic ketone (see experimental procedures in supporting information). Next, we added an engineered sitagliptin-producing transaminase (SitaTA)¹¹ along with its required reaction components directly into the same reaction mixture. The addition of isopropylamine, which is necessary for the transaminase reaction, raises the pH and decreases the activity of the HRP–NHPI system (data not shown). The fact that the HRP–NHPI system is inhibited during the second reaction step is useful for preventing oxidation of the amine product. We obtained a 10% yield across the two-step process, consistent with yields previously reported for transaminase-based cascade reactions.⁵⁶ This two-step, one-pot enzymatic cascade demonstrates the advantage of using an enzyme to generate the PINO radical in the first step; the semi-aqueous HRP–NHPI reaction solution can be directly combined with the transaminase enzyme and associated reaction components without any purification between the enzymatic reaction steps.

Figure 5.

One-pot, two-step conversion of 2-ethylnaphthalene to (R)-1-(naphthalen-2-yl)ethan-1-amine using HRP–NHPI and SitaTA. Yield and ee determined by HPLC.

In summary, the HRP–NHPI catalyst system enables selective benzylic C–H activation for a variety of substrates under mild reaction conditions. Under matched conditions, we observed dramatically higher product formation compared to the HRP–HOBT (10-fold) and TvL–NHPI (no product detected) catalyst systems. The fact that the HRP–NHPI catalyst system consists of commercially available reagents (HRP, NHPI, and H₂O₂), functions in a conventional “dump-and-stir” reaction setup, and tolerates a range of temperatures and solvent systems make it easy to implement. An admitted disadvantage of the system relative to laccase–mediator systems is that HRP requires H₂O₂, which is more expensive than O₂, the oxidant used by laccases. Nonetheless, H₂O₂ is an easy-to-use commodity chemical, and its use as a liquid reagent allows its concentration in the reaction solution to be controlled more easily than the concentration of O₂. Our results complement recent studies using photocatalysts for converting alkylbenzenes to ketones,^57–60 as well as recent studies using cytochrome P450s, peroxygenase enzymes, and artificial metalloenzymes to convert alkylbenzenes to alcohols, ketones, or amines.^61–66 Relative to peroxygenase enzymes, the HRP–NHPI system offers the advantage of not requiring the substrate to fit within a confined active site, making it compatible for use with bulky substrates. Another practical advantage is the commercial availability of HRP, which avoids the need for recombinant enzyme expression and purification.

We anticipate that benzylic substrate radicals generated by the HRP–NHPI system will be possible to intercept with an array of different reagents, which should allow for the synthesis of halogens, azides, or nitriles, among others (Figure 1a).¹⁴ As an initial demonstration, we show here the formation of a benzylic NHPI adduct, an intermediate that can be converted into multiple different products. Additionally, we showcase that ketones formed using HRP–NHPI can be converted to benzylic amines by a transaminase, using a convenient one-pot two-step procedure that requires no workup or purification between reaction steps. Given the broad range of conditions compatible with the system, additional carbonyl-reactive enzymes should be possible to incorporate into two-step cascades. In the future, directed evolution of HRP will be a promising approach to access more potent mediator radicals, allowing for improved yields with electron-deficient substrates and enabling C–H activation beyond the benzylic position.