Late-stage oxidative C(sp³)–H methylation

Summary Paragraph:

Frequently referred to as the “magic methyl effect”, installation of methyl groups, especially α to heteroatoms, has been shown to drastically increase the potency of bioactive molecules^1–3. Current methylation methods display limited scope and have not been demonstrated in complex settings¹. Herein we report a regio- and chemoselective oxidative C(sp³)–H methylation method compatible with late-stage functionalization. A key to affecting this new chemistry was the combination of a highly site- and chemoselective C–H hydroxylation with a mild, functional group tolerant methylation. Using a small molecule manganese catalyst Mn(CF₃PDP) at low loading (substrate/catalyst = 200) afforded targeted C–H hydroxylation on heterocyclic cores while preserving electron neutral and rich aryls. Fluorine or Lewis acid assisted formation of reactive iminium or oxonium intermediates enabled the use of a modestly nucleophilic organoaluminum methylating reagent that preserves other electrophilic functionalities on the substrate. The late-stage C(sp³)–H methylation is demonstrated on forty-one substrates housing sixteen different medicinally important cores incorporating electron-rich aryls, heterocycles, carbonyls, and amines. Eighteen pharmacologically relevant molecules with competing sites, including drugs (for example tedizolid) and natural products, are methylated site-selectively at the most electron rich, least sterically hindered position. Syntheses of two magic methyl substrates, an RORc inverse agonist and an S1P₁ antagonist, are demonstrated for the first time via late-stage methylation from the drug or its advanced precursor. Additionally, an unprecedented remote methylation of the B-ring carbocycle of an abiraterone analog is shown. The ability to methylate such complex molecules at late stages will reduce synthetic effort and thereby expedite broader exploration of the magic methyl effect in pursuit of novel small molecule therapeutics and chemical probes.

The introduction of methyl groups has the potential to drastically improve the biological activities of a drug candidate by altering its binding affinity, solubility, and metabolism^1–8. Such changes have been demonstrated to increase the potency of lead compounds up to more than 2000 folds and to enable interrogation of biological processes (Fig. 1a)^6–8. Although methyl groups are ubiquitous in small-molecule drugs¹, no general method is available to incorporate them in complex molecules at late stages. Accordingly, de novo synthesis, a rate-limiting step in drug discovery that impairs its overall atom-economy, is required^9,10. A practical synthetic method that allows selective installation of methyl groups from C–H bonds at sites adjacent to heteroatoms, where the magic methyl effect is often most significant, would streamline diversification of drug leads and encourage more comprehensive investigations of this effect. Over the past decade, considerable progress has been made in developing C(sp³)–H alkylation methods where the N-or O-heterocycle acts as a nucleophilic coupling partner^11–17. Such cross-couplings have shown broad scope with respect to alkyl electrophiles but limited scope of the metalated heterocyclic intermediates generated via substrate-controlled deprotonation or single-electron transfer (SET). Cases demonstrated with methyl electrophiles have focused on simple azacycles ^{11,13,15–17}. Expanding the heterocyclic scope to include dissymmetric substrates, epimerizable stereocenters, electrophilic functionalities (e.g., carbonyl, nitrile), remote basic amines, heteroaromatics, and halogenated aromatics remains a major challenge to be overcome for widespread use in late-stage diversification. Additionally, while direct C–C bond forming reactions may be desirable for installing larger and/or functionalized alkyl groups, direct methylation often results in inseparable mixtures with the starting material due to the methyl group’s small size and electron neutrality.

Fig. 1 |. C(sp³)–H methylation.

a, The magic methyl effect boosts potency of drugs and furnishes biological probes. b, This oxidative methylation proceeds through an electrophilic intermediate. Challenges included over-oxidation, unselective oxidation, elimination and unselective methylation pathways. c, Late stage oxidative methylation of antibiotic tedizolid. d, Oxidative C-H methylation is demonstrated on 16 different pharmaceutically relevant cores. Using only 1 equivalent of substrate, methylation proceeds site-selectively and with functional group tolerance to afford preparative yields in 41 examples (including 18 complex bioactive molecules).

We sought to approach C(sp³)–H methylation in N- and O-containing heterocycles in an oxidative fashion through a hydroxylated intermediate, with subsequent iminium or oxonium ion formation and methylation (Fig. 1b). Catalyst control could be leveraged to influence the site- and chemoselectivity of C–H hydroxylation in a broad range of heterocycles (Fig. 1c,d). Reports of alkylations of N-acyliminium ions are of limited scope^18–20. Although C(sp³–H oxidation α to heteroatoms has been well-demonstrated, substrate-controlled selectivities can afford poor site- and chemoselectivity thereby limiting examples in complex settings. Additionally, the strong hyperconjugative activation of hemiaminals and hemiacetals typically promotes overoxidation to the corresponding carbonyl, calling for reduction prior to or after methylation (Fig. 1b)^21–26.

The Mn(CF₃PDP)(MeCN)₂(SbF₆)₂ 1 catalyst was reported to uniquely control site- and chemoselectivity in hydroxylating strong methylene C(sp³)–H bonds while tolerating halogenated arenes, although the tolerance for electron neutral or rich aromatic and some heteroaromatic rings remained a challenge (Fig. 1b)²⁷. We questioned if sterically hindered catalyst 1 could result in faster C–H hydroxylation than alcohol oxidation for hyperconjugatively activated C–H bonds, and whether under milder oxidation conditions such a rate difference would increase chemoselectivity and yield for the hydroxylated product. Under the previously reported forcing conditions (10 mol% 1, 5 equiv. H₂O₂)²⁷, oxidation of arylated γ-lactam 2 afforded a significant amount of over-oxidation to the corresponding imide (Fig. 2a, 4b, 41%). Lowering the catalyst and hydrogen peroxide loadings (0.5 mol% 1, 2.0 equiv. H₂O₂) enabled the C–H bond α to nitrogen (α-N) to be hydroxylated to hemiaminal intermediates (hemiaminal, and hemiaminal acetate from AcOH) in an excellent 82% yield (4a). Consistent with slower alcohol oxidation, exposure of alcohol 4a to identical oxidation conditions afforded 84% hemiaminals with only 10% imide 4b (Fig. 2b). In contrast, exposure of alcohol 4a to the forcing conditions afforded predominantly imide 4b (54%) with only 17% hemiaminals. Under mild oxidation conditions, we additionally observed enhanced chemoselectivities for electron-rich and neutral aromatics and heteroaromatics, likely due to attenuation of similarly higher-energy overoxidation pathways (vide infra). Notably, this constitutes among the highest S/C ratio (substrate/catalyst = 200) reported to date for a preparative C(sp³)–H hydroxylation reaction in a complex setting. The ability to separate the hydroxylated intermediate prior to methylation, while not necessary for alkylation, avoids formation of an inseparable mixture of the methylated product and starting material often observed in direct methylation¹⁷. Expectedly, Fe(PDP) and Fe(CF₃PDP), previously employed for oxidative α-arylation of aliphatic peptides^28–30 gave a complex mixture of aromatic oxidation products (Extended Data). Mn(PDP), shown to hydroxylate simple linear amides³¹, was not reactive enough to promote preparative hydroxylation of 2, but can be uniquely effective for some sterically hindered substrates (vide infra).

Fig. 2 |. Reaction development.

a, Optimization of the oxidative methylation reaction. For achiral substrates, (R,R)- and (S,S)-1 can be used interchangeably. b, Exposure of hemiaminal to mild C–H hydroxylation developed here (0.5 mol% 1, 2 equiv. H₂O₂) gives little overoxidation to the imide. The previous forcing condition (10 mol% 1, 5 equiv. H₂O₂) results in imide as the major product. ^aNo methylation. ^bMixture of hemiaminal (64%-71%) and hemiaminal acetate from AcOH (13%-18%). ^c1 equiv. ^d2 equiv. ^eTFAA (1 equiv.), TMSOTf (1 equiv.). ^fMsCl (1 equiv.), NEt₃ (1 equiv.), NaHCO₃ wash; AlMe₃ (3 equiv.), −78 °C, 2 h; rt 1 h. ^gMeMgBr (3 equiv.), −78°C, 3 h.

A further challenge with an oxidative methylation approach was to identify a way to activate the hemiaminal/hemiacetal towards attack by a nucleophilic methyl source without resulting in either undesirable elimination to the enamine, or attack at other electrophilic moieties in complex substrates (Fig. 1b). The modestly nucleophilic and Lewis acidic nature of organoaluminum reagents suggested they could achieve such selective reactivity. Their high affinity for fluorine coupled to their tolerance of Lewis acids afforded a means of generating reactive iminium or oxonium species from transient C–F or from C–OH bond ionization^32,33. High functional group tolerance was also expected given organoaluminum reagents’ ability to methylate oxoniums at late stages in the presence of other electrophilic functionalities³⁴.

After significant experimentation abbreviated in Fig. 2a, we arrived at a scalable general procedure employing either diethylaminosulfur trifluoride (DAST) or boron trifluoride diethyl etherate (BF₃•OEt₂) as hydroxyl activators for iminium formation with AlMe₃ as an inexpensive, commercial methylating reagent. Thermally stable bis(2-methoxyethyl)aminosulfur trifluoromethanesulfonate (Deoxo-Fluor) may also be used. In general, the fluorine-assisted methylation strategy should be used in substrates containing Lewis basic or electrophilic functional groups, while ones lacking such functionality can be methylated via the BF₃ activation strategy (Extended Data). When unreacted hemiaminal acetate is observed, esterification of the hemiaminal by trifluoroacetic anhydride (TFAA) and subsequent activation of both esters with trimethylsilyl triflate (TMSOTf) to furnish the iminium can be employed³⁴. The capacity to vary the ionization method with AlMe₃ is critical to the broad scope of this methylation, providing a facile handle to tune reactivity and/or selectivity for a given substrate in cases where unreacted hemiaminal intermediates or enamine byproducts are observed (vide infra).

Alternative activation modes with AlMe₃ and methylating reagents with DAST were examined (Extended Data). In hemiaminals, base-mediated formation of an activated C–O bond (i.e., mesylation) gave predominantly elimination (Fig. 2a). Grignard reagents, even at cryogenic temperatures, afford diminished yields relative to AlMe₃, likely due to poor functional group tolerance. Although ineffective for methylation of heterocyclic substrates, these reagents can be used to effect methylation in challenging linear secondary amine and carbocyclic substrates (vide infra, 51, 53).

We explored oxidative methylation for its capacity to methylate a collection of twenty-two compounds comprising ten different heterocyclic cores commonly found in pharmaceuticals (Fig. 3). A gram-scale methylation of 2 was successfully done in 71% yield via DAST activation; an ethyl group could also be installed using commercial triethylaluminum (5, 51%). Methylated δ-lactam 6 was isolated in 58% yield; analogous to the γ-lactam, under more forcing oxidation conditions δ-lactam gave predominantly imide (60%). For amide 2, methylated 3 was observed in preparative yields with both DAST and BF₃ ionization (Fig. 2a). However, in oxazolidinones, housing more labile carbonyls, fluorination with DAST furnished significantly higher yields (7, 55% versus 10% with BF₃, Extended Data; 8, 63%). Pyrrolidine, the fifth most common nitrogen heterocycle in drugs^35,36, undergoes hydroxylation with no significant over-oxidation to pyrrolidinone, followed by BF₃-promoted methylation to afford mono-methylated product 9 in 54% yield. Critical for late-stage derivatization and orthogonal to most radical processes, high site-selectivity for methylation at the less sterically hindered methylene site was observed in substrates bearing more activated tertiary (3°) aliphatic, 3° benzylic, and 3° α-carbonyl C–H bonds to afford products in preparative yields (10-13). Full stereoretention was measured with chiral substrates leading to 12 and 13, indicating that the high regioselectivity is attributed to catalyst control of Mn(CF₃PDP) 1 in the C–H cleavage step. Methyl ester, ketone, acetate, and nitrile, not well tolerated with strongly nucleophilic methylation reagents, were maintained using DAST activation/AlMe₃ methylation (12-15). Methylation on a 3-phenylpyrrolidine derivative proceeded regioselectively at the methylene site distal from the phenyl group, furnishing the 5-methylated product 16 in useful yield. Such chemoselectivity for electron neutral aromatics has not been previously demonstrated: at higher catalyst loadings, 1 afforded poor yields and chemoselectivities²⁷.

Fig. 3 |. Ten different heterocyclic cores, commonly found in pharmaceuticals, were explored in the Mn(CF₃PDP) 1-catalyzed C–H hydroxylation and methylation.

Twenty-two heterocycles including lactams, oxazolidinones, pyrrolidines, piperidines, azepane, azabicycloheptane, quinolines, and isochroman were oxidatively methylated in preparative overall yields (54% average) using limiting substrate. General oxidation: substrate, catalyst (0.5 mol%), AcOH in MeCN, −36 °C; H₂O₂ (2 or 5 equiv.) in MeCN syringe pump 1 h. Mixture passed through silica plug, EtOAc flush, concentrated prior to isolation or methylation. For insoluble substrates, CH₂Cl₂ added to MeCN and/or 0 °C. ^aDAST Activation: crude in CH₂Cl₂ (0.2 M), DAST (1 equiv.) added at −78 °C; room temperature (rt) for 1 h; cooled to −78 °C, AlMe₃ added, stirred 2 h; rt for 1 h. ^bBF₃ Activation: crude in CH₂Cl₂ (0.2 M), −78 °C, AlMe₃ (3 equiv.) and BF₃•OEt₂ (2 equiv.) sequentially added, stirred 1 h; rt for 3 h. ^cTriethylaluminum. ^d2 mol% (S,S)-1. ^eAlMe₃ −78 °C, 3 h. ^f1 mol% (S,S)-1. ^gFor facile purification, hemiaminal isolated before methylation. 10 mol% (S,S)-1, rt, starting material recycled 1x.

In piperidines, the most common nitrogen heterocycle in small molecule drugs^35,36, both DAST and BF₃ activation should be tried: enamine formation competes strongly with methylation and is highly dependent on both the substrate and the mode of hemiaminal activation (Fig. 3). For example, a pipecolic acid derivative gave 2% of the methylated product with 60% enamine byproduct under DAST activation, whereas the BF₃ activation furnished 17 in 47% overall yield. Alternatively, methylation of piperidinyl-2-methyl acetate using BF₃ did not fully convert the hemiaminal to methylated product, whereas DAST activation afforded 18 in 64% yield. Gamma-substituted piperidines are prevalent structures in drugs, such as in paliperidone and paroxetine. An N-nosyl intermediate in the synthesis of paliperidone was selectively methylated using the DAST protocol to give 19 in 37% yield with no protection of the benzisoxazole ring γ to nitrogen. However, methylation of γ-(4-bromophenyl)piperidine with DAST resulted predominantly in enamine formation, while the BF₃ activation strategy affording 39% yield of methylated 20. Notably, all piperidines furnished a single observed methylated diastereomer, likely as a result of the rigid half-chair conformation of the iminium intermediate³⁷.

Other simple cyclic amines, such as azepane, azabicycloheptane, and decahydroquinoline, were selectively methylated using the BF₃ activation protocol α-N to afford 40%-46% overall yields of mono-methylated products 21-23. Tetrahydroisoquinoline, among the top 20 nitrogen heterocycles in FDA-approved drugs³⁶, was oxidatively methylated using BF₃ activation in good yields for both a brominated and an unsubstituted aromatic structure (24, 51%; 25, 50%), with lower yields observed using DAST activation. In contrast to the majority of radical based oxidation methods that oxidize isochromans to isochromanones, under Mn(CF₃PDP) 1 catalysis little over-oxidation is observed. 6,8-Dibromoisochroman was oxidatively methylated using DAST/AlMe₃ in 74% yield (26); BF₃ activation for these types of oxygen heterocycles afforded ring-opening products³⁸.

We explored the ability of highly site- and chemoselective Mn(CF₃PDP) 1 catalyzed C–H hydroxylation, coupled to a Lewis-acid/fluorine-promoted methylation, to provide a general method for installing methyl groups directly into the hydrocarbon cores of complex, bioactive molecules, thereby avoiding lengthy and costly de novo synthesis (Fig. 4)^1,9,10. Cromakalim acetate, a potassium channel activator housing a γ-lactam with tertiary and secondary hyperconjugatively activated α-N C(sp³)–H bonds, underwent oxidative methylation at the less activated but more sterically accessible secondary site in good yield (27, 51%). The acetate could be readily removed to furnish methylated cromakalim 28 in 85% yield. The methyl ester of indoprofen, an anti-inflammatory drug investigated for spinal muscular atrophy³⁹, was oxidatively methylated at its central isoindolinone core in synthetically useful yields (29, 33%). The enhanced chemoselectivity of oxidative methylation with 1 under reduced loadings is evident when comparing with results at higher loadings (10 mol%), where 29 was obtained in diminished yields (7%) due to poor chemoselectivity. Chloroindoprofen methyl ester, a derivative with decreased electron density on the aromatic ring, undergoes oxidative methylation in higher yields (30, 55%). Fenspiride, an antitussive drug, was oxidatively methylated in a useful overall yield (31, 24%) at a methylene site adjacent to the quaternary center of an unprotected spiro-oxazolidinone, using (S,S)-Mn(PDP)(SbF₆)₂ catalyst that is less sensitive to sterics²⁷. The basic piperidine nitrogen of fenspiride was protected with HBF₄ and rendered a strong electron-withdrawing group, deactivating a distal benzylic site and three α-N sites towards C–H oxidation⁴⁰. Notably, SET reactions proceeding via basic amine catalysis (e.g. quinuclidine) are not amenable to this kind of nitrogen protection strategy and therefore do not undergo remote C–H functionalizations^17,24. A derivative of pozanicline, a neuroprotective drug evaluated for ADHD treatment⁴¹, undergoes α-N oxidative methylation at the pyrrolidine in useful yield and diastereoselectivity (32, 34%, 6:1 dr). The HBF₄ protection deactivates the basic pyridine moiety and its proximal ethereal sites from oxidation with 1. While DAST activation produced similar yields, a higher diastereoselectivity was obtained with BF₃ activation (6:1 vs. 3:1), possibly due to different iminium counterions (Fig. 1b). The nosyl group on pyrrolidine, a convenient chromophoric protecting group for secondary amines, was readily removed using thiophenol and subsequently protected with tert-butyloxycarbonyl (Boc) to afford 33 in 57% overall yield. Underscoring the unique chemoselectivity of this method, a derivative of the antidepressant diclofensine was oxidatively methylated at its tetrahydroisoquinoline core to afford 34 in useful yield despite the presence of a very electron rich methoxyphenyl. Mild, reductive desulfonation followed by reductive amination furnished Me-diclofensine 35 in 82% yield. The antidepressant drug citalopram, upon HBF₄ protection of the tertiary amine, is oxidatively methylated at its dihydroisobenzofuran core to afford 36. DAST activation was used on the majority of these densely functionalized substrates whereas BF₃ activation was more effective on the tetrahydroisoquinoline core.

Fig. 4 |. Application of oxidative methylation for late stage functionalization.

a, Selective methylation of drugs, drug precursors, intermediates and natural products underscores the power of this method for late stage applications. Generally, 0.5 to 5 mol% (S,S)-1 and 2 or 5 equiv. H₂O₂ were used for oxidation. Higher catalyst and oxidant loadings were applied when conversions were low. b, Methylation of an RORc inverse agonist precursor rapidly furnishes the analogue with 13-fold potency boost. c, Methylation of antibiotic tedizolid acetate furnishes Me-tedizolid. d, Methylation of linear aniline in S1P₁ antagonist methyl ester occurs at a position where magic methyl effect was observed to contribute to a 2135-fold potency boost. e, Remote methylation of a carbocycle on an abiraterone analog. ^aDAST activation. ^bBF₃ activation. ^cTMSOTf activation: TFAA, rt, 1 h; cooled to −78 °C, AlMe₃ and TMSOTf sequentially added, 2 h; then rt, 1 h. ^dDeoxo-Fluor activation. ^eMesylation activation: MsCl and Et₃N added, rt, 1 h; NaHCO₃ wash, dried, condensed; redissolved in CH₂Cl₂, AlMe₃ added at −78 °C, stirred 2 h; then rt, 1 h. ^fOxidation intermediates isolated before methylation. ^g1 M NaOH/MeOH. ^hStarting material recycled 1x. ⁱFor insoluble substrates, CH₂Cl₂ added to MeCN and/or 0 °C. ^jHBF₄ protection, ref. 40. ^k10 mol% (S,S)-Mn(PDP)(SbF₆)₂. ^l10 mol% (S,S)-1. ^mPhSH, Cs₂CO₃; Boc₂O. ⁿMg, NH₄Cl; formaldehyde, formic acid. ^o10 mol% (RR)-1. ^p2 equiv. TMSOTf. ^qTMSOTf (1.2 equiv.), 0 °C, 1 h, then MeMgBr (3.0 equiv.) −78 °C, 4 h, repeated once.

A precursor to pyrroloisoquinoline, a prevalent structure in compounds with neurotransmitter uptake inhibitor properties⁴², undergoes selective oxidative methylation at the less sterically hindered methylene site, versus the more activated tertiary, benzylic site, to furnish 37 (44% yield, Fig. 4a). Oxidation of a carbamate precursor to antibiotic tedizolid furnished substantial amounts of hemiaminal acetate that could be methylated in a useful overall yield under the TFAA/TMSOTf-assisted methylation (38, 44%). This method is operationally facile and can be performed on gram scale with no loss in efficiency (45% yield). Fluorination afforded lower yields of methylated product 38 due to unconverted hemiaminal acetate. The core piperidine of a paroxetine precursor and metabolite⁴³ was oxidatively methylated in useful overall yields (39, 34%) preferentially at the less sterically hindered methylene site remote from the 3-acetoxymethyl group. Nosyl deprotection and subsequent Boc protection afforded 40 in 86% yield. A piperidine derivative of the anti-inflammatory drug celecoxib was mono-methylated to afford 41 in good overall yield in the presence of an oxidatively labile tolyl group and pyrazole, both tolerated during C–H oxidation with 1 and requiring no protection. In these piperidine substrates, BF₃ activation was effective in furnishing methylated products.

Methylation of proline-based di-, tri-, and tetrapeptides proceeded with good overall yields and mass balances (42, 43, 44) with 1 under fluorine-assisted oxidative methylation conditions (Fig. 4a). Deoxo-Fluor may be used in substrates like tetrapeptide 44, where isolation from the polar byproducts of DAST is challenging. Although effective in promoting arylation of peptides with electron rich aromatic nucleophiles, BF₃ activation in the AlMe₃ methylation of peptides afforded complex mixtures, likely arising from activation of the amide carbonyls³⁰. Ambroxide, a naturally occurring terpenoid, also undergoes selective oxidative methylation using DAST at a methylene site α to oxygen on its tetrahydrofuran ring to afford 45 in 32% yield (with 19% of sclareolide lactone). The use of BF₃ in this case promoted ring-opening. Significantly, Fe(PDP), ruthenium-mediated oxidation, and 1 under forcing conditions all afforded sclareolide lactone as the major product isolated (see SI)^10,21,28.

The sultam ring in an advanced intermediate of an RORc inverse agonist 46 (Fig. 4b) was oxidatively methylated with 1 using TFAA/TMSOTf activation/AlMe₃ to afford 47 as the syn-diastereomer. Other activation modes, such as BF₃, resulted in deleterious elimination pathways. Notably, an oxidatively labile phenyl moiety and a doubly activated benzylic methylene site were tolerated. Previous installation of the syn-methyl group afforded a 13-fold increase in RORc SRC1 selectivity relative to the unmethylated version; however, it required a six-step de novo synthesis proceeding in 1.4% overall yield⁵. This analog and others are now accessible via cross-coupling with methylated intermediate 47.

Tedizolid, a commercial oxazolidinone antibiotic for acute bacterial skin infections, bears numerous oxidatively sensitive functional groups such as a pyridine, tetrazole, and an N-methyl (Fig. 4c). Mn(CF₃PDP) 1 oxidation (2 mol%) of tedizolid acetate 48 proceeded in ca. 53% yield of oxidized products (3:1 hemiaminal/acetate) with no protection of the dense nitrogen functionalities. The significant challenge was to identify a procedure to install the methyl group from the hemiaminal intermediates. Activation via fluorination furnished primarily eliminated products not observed on the simpler core structure (38, Fig. 4a), while BF₃ activation resulted in side reactions. However, under TFAA/TMSOTf activation, elimination was suppressed and the methylated product 49 was obtained in a remarkable 40% overall yield, 74% per step, comparable to that of the simpler precursor 38. Deprotection of the acetate in 92% yield afforded Me-tedizolid 50, an interesting candidate for future biological evaluation given that a 9-fold boost in potency has been reported for similar oxazolidinone cores with methylation at the same position (Fig. 1a)⁴⁴.

We questioned if the scope of this reaction could be extended beyond heterocycles, and found that oxidative C–H methylation is not restricted to substrates that can form iminium or oxonium intermediates: promising reactivity has been observed for both imines and remote alcohols generated via C–H hydroxylation with 1. An S1P₁ antagonist, whose benzylic and aromatic methylations afforded a 2135-fold potency increase⁶, was methylated in its methyl ester form (51, Fig. 4d). While oxidation of the antagonist was successful without need for protection of the aniline motif, the resulting imine was much less reactive than an iminium and required a stronger nucleophile than AlMe₃. In this case, we found that methylmagnesium bromide at cryogenic temperatures, with TMSOTf activation of the imine, produced the methylated product without eroding the amide and ester functional groups (52, 14%, 52% per step).

At higher catalyst loadings, Mn(CF₃PDP) 1 is an effective catalyst for methylene C–H bond hydroxylations²⁷. Abiraterone analog 53 was hydroxylated in ca. 32% yield (with 16% ketone) in one step, without recycling the starting material as required in Fe(CF₃PDP) catalysis (Fig. 4e)⁴⁰. In carbocyclic substrates, displacement of a C–F bond or ionization with a Lewis acid is difficult; however, mesylates of such aliphatic alcohols are stable and can be activated by AlMe₃ to undergo substitution⁴⁵. By replacing fluorination with mesylation, 53 was successfully methylated as a single observed diastereomer (54, 15% overall yield, 53% per step), likely through a carbocation intermediate. To the best of our knowledge, this is the first method that enables such remote methylation at unactivated C(sp³)–H bonds. The discovery of this reactivity underscores the importance of developing methylene oxidations that afford predominately alcohol products.

Extended Data

Extended Data Table 1 |.

Reaction Optimization.

Entry	Substr.	Catalyst	Loading (mol%)	Additive	[Nu]	4a (OH)/S2 (%)	4a (OAc)/S3 (%)	3/7 (%)	4b (%)	4c/S4 (%)	rsm (%)
Oxidation
1b	2	Fe(PDP)	3 x 5	-	-	<5k	0	-	<5k	-	0
2c	2	Fe(CF3PDP)	3 x 5	-	-	8k	0	-	6k	-	0
3d	2	Mn(PDP)(OTf)2	1	-	-	12	0	-	0	-	75
4	2	Mn(PDP)(SbF6)2	1	-	-	28	7	-	<5k	-	35
5e	2	Mn(CF3PDP) 1	10	-	-	13k	10	-	41	-	0
6	2	1	1	-	-	51	21	-	9	-	0
7	2	1	0.5	-	-	64	18	-	<5k	-	4
Methylation
8f	2	1	0.5	BF3•OEt2	AlMe3	<5k	0	63	<5k	0	11
9f	S1	1	0.5	BF3•OEt2	AlMe3	11	5	10	-	4	27
10g	S1	1	0.5	DAST	AlMe3	0	14k	55	-	0	16
11g	2	1	0.5	DAST	AlMe3	0	0	64	<5k	0	12
12g	2	1	0.5	Deoxo-Fluor	AlMe3	0	0	61	6	0	5
13h	2	1	0.5	TFAA/TMSOTf	AlMe3	0	0	51	<5k	14	9
14h	S1	1	0.5	TFAA/TMSOTf	AlMe3	0	0	46	-	20	13
15i	2	1	0.5	MsCl/Et3N	AlMe3	15	0	0	<5k	39	6
16g	2	1	0.5	DAST	ZnMe2	17	9	0	11	0	14
17g,j	2	1	0.5	DAST	MeMgBr	24	<5k	24	<5k	0	9

^aGeneral oxidation (unless otherwise noted): 2 (0.3 mmol), catalyst (x mol%, (R,R) and (S,S) used interchangeably), AcOH (15 equiv.), MeCN (0.5 M), −36 °C; H₂O₂ (2 equiv.) in MeCN (3.75 mL) syringe pump 1 h. Mixture passed through silica plug, EtOAc flush, concentrated prior to isolation or methylation. Isolated product yields.

^bProcedure ref. 28.

^cProcedure ref. 29.

^dProcedure ref. 31.

^e5 equiv. H₂O₂.

^fGeneral BF₃ alkylation: crude in CH₂Cl₂ (0.2 M), −78 °C, AlMe₃ (3 equiv.) and BF₃•OEt₂ (2 equiv.) sequentially added, stirred 1 h; room temperature (rt) for 3 h.

^gGeneral fluorine alkylation: crude in CH₂Cl₂ (0.2 M), fluorine additive (1 equiv.) added at −78 °C; rt for 1 h; cooled to −78 °C, nucleophile (3 equiv.) added, stirred 2 h; rt for 1 h.

^hGeneral TMSOTf alkylation: crude in CH₂Cl₂ (0.2 M), TFAA (1 equiv.) added, stirred 1 h; cooled to −78 °C, AlMe₃(3 equiv.) and TMSOTf (1 equiv.) sequentially added, stirred 2 h; rt for 1 h.

ⁱCrude in CH₂Cl₂ (0.2 M), MsCl (1 equiv.) and Et₃N (1 equiv.) added, stirred 1 h; washed NaHCO₃, dried, reduced; redissolved in CH₂Cl₂, AlMe₃ (3 equiv.) added at −78 °C, stirred 2 h; rt for 1 h.

^jMeMgBr (3 equiv.) added at −78°C, stirred 3 h.

^kYield by crude ¹H NMR.