Direct functionalization of amino groups in complex organic molecules is one of the most important key technologies in modern organic synthesis, especially in the synthesis of bio-active chemicals and pharmaceuticals. Whereas numerous chemical reactions of amines have been developed to date, a selective, practical method for functionalizing complex amines is still highly demanded. Here we report the first late-stage N-alkylation of pharmaceutically relevant amines with alcohols at ambient temperature. This reaction was achieved by devising a mixed heterogeneous photocatalyst in situ prepared from Cu/TiO2 and Au/TiO2. The mixed photocatalytic system enabled the rapid N-alkylation of pharmaceutically relevant molecules, the selective mono- and di-alkylation of primary amines, and the non-symmetrical dialkylation of primary amines to hetero-substituted tertiary amines.
New synthetic strategies for producing complex organic molecules (e.g. bio-active natural products and pharmaceuticals) have been continuously demanded because only a limited number of chemical reactions have been available for selectively converting molecules bearing various functional groups into desired compounds1. Alkylamines represent an important class of functionality in valuable yet complex molecules2. Representative examples are shown in Fig. 1a. The alkylamino functionality is essential for drug design, as they often improve the oil-water partition coefficient (log P), reduce their toxicity, and increase their bioavailability (prodrugs)3.
The N-alkylation of primary amines (1) with alcohols (ROH) is a highly efficient method for preparing secondary alkylamines (2) and tertiary alkylamines (3, Fig. 1b)4,5,6,7,8,9,10. The most frequently used catalysts for this reaction include transition metal complexes (e.g. Ru11,12,13,14,15, Ir16,17,18,19,20, Fe21,22,23,24,25, Co26,27, and Mn28,29), and heterogeneous catalysts30,31,32,33,34,35. We have recently contributed to the development of molecular iron and phosphazene catalysts21,36. These methods allow green access to various amines without producing stoichiometric waste other than water. However, because they invariably require high reaction temperatures (typically >80 °C), their application to the late-stage functionalization37 of thermally unstable amines remains to be explored.
The photocatalytic N-alkylation of amines using alcohols is potentially a powerful method for functionalizing complex amines because this reaction generally proceeds at room temperature38,39,40,41,42,43. We have recently demonstrated that Ag/TiO2 promotes the chemoselective N-methylation of amines44. This photocatalyst enabled, for the first time, the gram-scale synthesis of tertiary amines by means of photocatalytic N-methylation at room temperature. However, the functionalization of complex amines by means of photocatalytic N-alkylation with alcohols remains unexplored simply because of the lack of a suitable method: the above-mentioned photocatalytic methods suffer from a limited scope in terms both of amine substrates and of alcohol reagents. Moreover, previously reported systems require the use of a large excess (>140 equiv) of alcohols and long reaction times (>4 h for completion of 0.2-mmol scale reactions). Thus, the benefits of developing an effective method for the room-temperature functionalization of complex amines are undeniable and warrant a thorough exploration of metal-loaded TiO2 systems. Herein we report the first late-stage N-alkylation of pharmaceutically relevant amines with alcohols at ambient temperature (Fig. 1c). The strategy of mixing two metal-loaded photocatalysts resulted in a synergistic increase in reaction rate for the N-alkylation of pharmaceutically relevant molecules. The selectivity with respect to the mono- and di-alkylation of primary amines was solvent-controlled. Facile synthesis of rivastigmine and alverine as well as venlafaxine-d6 and imipramine-d3 was demonstrated using methanol, ethanol, or deuterated methanol as alkylating reagent.
Development of Cu–Au mixed photocatalytic system
We first examined the N,N-dimethylation of primary amine 1a to give pharmaceutically relevant target (rac)-rivastigmine (3aa; a chiral (S)-form being used for Alzheimer’s and Parkinson’s diseases) using methanol with metal-loaded TiO2 photocatalysts under light irradiation (Xe lamp, λ = 300–470 nm) at 25 °C (Fig. 2). Among those tested, Cu/TiO2 was the most effective photocatalyst, giving the tertiary amine 3aa in 89% isolated yield without involving the cleavage of the benzylic C–N bond or carbamate linkages (entry 1). While Ag/TiO2 and Pd/TiO2 also afforded 3aa, neither Au/TiO2, Pt/TiO2 nor TiO2 were effective (entries 2–6). Conversions of 1a in entries 1–5 were >96%, but quantitative analysis of intermediates failed due to their instability. Inspired by the recent advent of synergistic photocatalysis45,46 and bimetallic heterogeneous catalysis31, we next investigated a mixed photocatalytic system consisting of Cu/TiO2 and Au/TiO2 in order to further establish a more efficient reaction system. Au/TiO2 was selected because it represents the most reactive titania-based photocatalyst for the dehydrogenation of primary alcohols to aldehydes47. Indeed, mixing these two photocatalysts resulted in a synergistic acceleration of the dimethylation of 1a, and gave 3aa in 70% yield in 2 h (entry 8). This result proved to be better than those obtained using Cu in combination with other metals (entries 9–11) or the sole use of Cu (entry 12). The Cu–Au promoted dimethylation of 1a (1.0 mmol) completed in just 4 h at 25 °C (entry 13). Furthermore, small-scale reaction proceeded more quickly at 50 °C to afford 3aa in 97% yield within 12 min (entry 14). Such a rapid dimethylation of amines by methanol has hitherto been unachievable. Reaction could be similarly promoted by irradiation with a UV-LED (λ0 = 365 nm), and hardly proceeded in the dark (entries 15 and 16).
Alkylation of amines by synergistic Cu–Au photocatalysis: substrate scope
With this optimized photocatalytic system, next, the substrate scope was checked. The results of photocatalytic N-alkylation of amines are summarized in Fig. 3. A chiral substrate (S)-1a was straightforwardly converted to rivastigmine [(S)-3aa] with retention of the absolute configuration at the benzylic position. The selectivity for mono- and dialkylation of amines 1b and 1c could be precisely controlled by tuning the reaction conditions. Irradiation of amine 1b and alcohols in hexane or cyclopentyl methyl ether (CPME) gave predominantly secondary amines 2bb–2bg. For the first time, photocatalytic methods have been successfully used with the presence of cyclopropyl, cyclobutyl, chloroalkyl, and oligomeric alkoxy groups in 2bd–2bg being tolerated. Moreover, the use of only 2–4 equiv of alcohol was sufficient for N-alkylation of 1b to 2bd–2bg. Exclusive monoalkylation of 1b with 2-propanol to 2bh proceeded under neat conditions. Selectivity for monoalkylation to 2bb–2bh over dialkylation was >97:3, as determined by GC/MS analysis. In contrast, dialkylation of 1b with ethanol and 1-propanol under neat conditions with longer irradiation time gave tertiary amines 3bb and 3bc in 89% and 83% yields, respectively. Similar results were also seen for amine 1c. Amines bearing core structures important to pharmaceuticals were also efficiently converted to the desired products 3da–3ga. In all these cases, the superior reactivity of the Cu–Au system with respect to either Cu/TiO2 or Au/TiO2 was confirmed (Table S6 in the Supplementary Information). Lysine derivative 1 h•HCl and protected glucosamine 1i were converted to tertiary amines 3 ha and 3ia in good yields, respectively. This method was also effective in the synthesis of alverine (5, a drug used for irritable bowel syndrome) and the functionalization of desloratadine (6, a drug used for treating allergies).
Sequential, non-symmetrical dialkylation
Having demonstrated that the photocatalytic system enabled the precise control of mono- vs di-alkylation, one-pot, sequential synthesis of non-symmetrical tertiary amines from primary amines and alcohols was demonstrated (Fig. 4a). Successive reaction of 1b with alcohols R1OH and R2OH yielded non-symmetrical amines 3bd (75%) and 3be (67%).
Application to deuterated drug synthesis
Regio-specifically deuterated drugs have recently begun receiving significant attention because of their improved metabolic stability with respect to their hydrogen analogues. This is making the development of new and efficient synthetic methods for their production an important emergent area in medicinal chemistry48. Here, deuterium atoms were precisely installed to pharmaceutical structures at the desired methyl groups by using Cu–Au mixed photocatalysis (Fig. 4b). Photocatalytic reaction of 1a with commercially available deuterated methanol (CD3OD) produced hexadeuterated (rac)-rivastigmine (3aa-d6) efficiently on a gram scale (1.0 g, 91% yield). This protocol also allowed us to rapidly access other hexa- and trideuterated drugs such as venlafaxine-d6 (3ja-d6) and imipramine-d3 (8-d3).
The origin of the superior reactivity of the mixed Cu–Au photocatalytic system is under investigation. Whereas photocatalytic activity of Au/TiO2 in methanol dehydrogenation is higher than that of Cu/TiO2 (Figure S1 in the Supplementary Information), reducing the reactivity of Au/TiO2 in producing 3aa is significantly lower than that of Cu/TiO2 (Fig. 2, entry 4 vs entry 1). These results imply a greater contribution by Au/TiO2 to alcohol dehydrogenation and by Cu/TiO2 to imine reduction, respectively (Fig. 4c).
When the photocatalyst recyclability was tested, the reused mixed Cu–Au photocatalysts were found to have comparable reactivity to the pristine mixture for 10 cycles (82–98% yield of 3aa, Table S7 in the Supplementary Information). The fresh and used photocatalysts were investigated by (scanning) transmission electron microscopy [(S)TEM] and powder X-ray diffraction (Figures S2–S20). TEM indicated that Cu/TiO2 comprised Cu nanoparticles of mean particle size 1.7 ± 0.3 nm, while Au/TiO2 comprised Au nanoparticles of mean particle size 7.75 ± 0.46 nm (Figures S2–S6). Whereas bright field imaging of the combined photocatalysts suggests the presence of both Cu and Au nanoparticles (Figure S7), STEM analysis suggests a more complicated picture (Figures S8–S19). Hence, a pristine sample of combined Cu/TiO2 and Au/TiO2 photocatalysts reveals an essentially uniform Cu background punctuated by discrete Au nanoparticles (Figure S19a). In contrast, after photocatalytic reaction (Table S7, run 1) the Cu signals have become more localized and coincident with the Au signals (Figure S19b). This is consistent with the formation of individually heterobimetallic nanoparticles at room temperature by light irradiation, in spite of the fact that high temperatures (ca. 160 °C) are normally needed for their formation49. The formation of similar Cu–Au heterobimetallic nanoparticles was seen after the irradiation of a mixture of Cu/TiO2 and Au/TiO2 in methanol in the absence of amine (Figures S20 and S21). This pre-irradiated Cu–Au photocatalyst also showed similar reactivity to the pristine analogue in the N,N-dimethylation of 1a (Table S8, entry 2). Nevertheless, the pristine mixture of Cu/TiO2 and Au/TiO2 showed slightly higher reactivity than the pre-irradiated mixed photocatalysts (Table S8, entry 1 vs entries 2 and 3), implying that the formation of heterobimetallic nanoparticles is not prerequisite for the high reactivity of the current photocatalytic system.
We have established a mixed Cu–Au photocatalytic system for the rapid N-alkylation of pharmaceutically relevant amines. The synthesis and functionalization of drugs, the controllable mono- and dialkylation of primary amines, and the non-symmetrical dialkylation of primary amines to hetero-substituted tertiary amines have been demonstrated by the mixed photocatalytic system. Studies for further improvement of the photocatalytic system, targeting the ultra-fast N-methylation of amines applicable to 11C-positron emission tomography (PET) using 11CH3OH are currently underway50.
A representative procedure for N-methylation of 1a to 3aa by Cu (5 wt%)/TiO2 and Au (5 wt%)/TiO2 (Fig. 2, entry 13) is as follows: Cu (5 wt%)/TiO2 (22 mg, 1.6 mol% Cu), Au (5 wt%)/TiO2 (22 mg, 0.51 mol% Au), anhydrous CH3OH (10 mL, 250 mmol), and 1a (222.1 mg, 1.00 mmol) were added successively to a cylindrical Pyrex glass reaction vessel (diameter: 50 mm, height: 130 mm with a top window made of Pyrex) connected to a rubber balloon. After the resulting mixture was sonicated for 30 sec and deaerated by Ar bubbling via cannula for 5 min, the vessel was immersed in a water bath (kept at 25 °C using a cooling circulator), and stirred for 4 h with irradiation [300 W Xe lamp (Ushio: BA-x300/ES1 Technology; CERMAX PE300BF) equipped with a UV cold mirror (λ = 300–470 nm)]. The presence of 3aa (95% yield) in the reaction mixture was indicated by GC/MS analysis using 2,2-dimethylpropan-1-ol as an internal standard. The reaction mixture was filtered through a 0.45 µm membrane filter and the photocatalyst was washed with CH3OH (10 mL). HCl (35–37%, 12 M aq, 0.5 mL, 6 mmol) was added to the solution (pH 1–2) and stirred at rt for 30 min. After methanol was evaporated, the residue was dissolved in H2O (20 mL) and washed with ethyl acetate (20 mL). To the aqueous layer, sodium carbonate (s) was added (pH 10) and extracted with ethyl acetate (2 × 30 mL). After washing with brine, the organic layer was dried over Na2SO4 and concentrated under reduced pressure to afford (rac)-rivastigmine (3aa) as a light-yellow oil (227.8 mg, 91% yield). All new compounds were fully characterized (see Supplementary Information).
Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
This work was supported by JSPS KAKENHI (Grant Number JP26410115, to H.N.), the Tobe Maki Foundation (to H.N.), the JGC-S Foundation (to H.N.), the Advanced Catalytic Transformation program for Carbon utilization (ACT-C, Grant #JPMJCR12YJ, JST, Japan, to S.S.), the Royal Society (International Exchange Scheme, to A.E.H.W. and H.N.), and UK EPSRC (EP/J500380/1, to K.J.). L.M.W. thanks the CSC program. The authors wish to thank Professors Ryoji Noyori (NU) and Akihiko Kudo (Tokyo University of Science) for fruitful discussions as well as R. Nagata and M. Shibata (NU) for their technical assistance.