Reaction development and optimizationOur initial forays into the development of diversifiable amination reagents centered around the use of diazirines, which are historically known for their utility as carbene precursors in chemical biology22,23. Instead, we recently demonstrated their use in transition metal-catalyzed and photochemical decarboxylative aminations. While many carboxylic acids and their derivatives are commercial and widely available, this is far less true of the tertiary acids that would enable access to the alkyl amines required for the pharmaceuticals displayed in Fig. 1A. Furthermore, synthetic access into these tertiary acids, especially when remote from other functional groups, is not straightforward. To overcome this hurdle, we turned our attention to the use of olefins in concert with diazirine 1. Substituted olefins are abundantly available, in both commercial building blocks and natural products, and are simple to prepare in the laboratory via robust chemical processes (e.g. alkylation, cross- or ring-closing metathesis, Wittig, etc.).Carreira’s elegant hydrohydrazination work served as inspiration for the immediate starting point due to the mild cobalt-catalyzed conditions and good regioselectivity that were reported8. Indeed, upon initial reaction of diazirine 1 with 4-phenylbut-1-ene, low yields of the corresponding diaziridine were observed. While this served as confirmation that the C–N bond formed, extensive optimization of solvent, catalyst, and hydride source proved ineffective for raising the yield. However, inspired by Nojima, di-t-butyl peroxide was added to facilitate the formation of the cobalt hydride24. To our delight, this proved the key to unlocking the reaction (see Supplementary Fig. 9 for proposed mechanism and catalytic cycle). After a switch to olefin 2 as our model substrate (lower volatility compared to 4-phenylbut-1-ene), a re-optimization of catalyst, solvent, and hydride source afforded diaziridine 3a in 99% isolated yield (Table 1).Table 1 Optimization of the reactionOther cobalt catalysts such as Co(dpm)3 (entry 1), and Co(TPP)Cl (entry 2) gave trace amounts of the desired product. Mn(dpm)3 (entry 3), failed to show any appreciable regioselectivity (ca. 1.5:1), however, the products were obtained in 86% isolated yield after only 2 h at 0 °C. Ultimately, this turned out to be an excellent protocol for symmetrical olefins and will be discussed further (see below). Other counterions for the salen catalyst (e.g. OAc in cat-2, Cl in cat-3, entries 4 and 5), led to diminished yields as did Carreira’s catalyst (cat-4, entry 6). While a mixed solvent system (DCE:IPA 4:1) was found to be optimal for most substrates, IPA alone (entry 7) gave 3a in a slightly lower yield and longer reaction time (30 h). The IPA only conditions proved to be useful with tri-substituted alkenes as the substrate scope was further evaluated. Although the olefin was consumed within 3 h, a longer reaction time was required to achieve completion. At 3 h the isolated yield was 66% (entry 8), whereas prolonging the reaction time to 16 h increased the isolated yield to 93% (entry 9) and to 20 h, 99% (optimized conditions). As mentioned above, di-t-butyl peroxide was a crucial additive; in its absence, the yield dropped to 38% (entry 10). When using t-butyl peroxide instead, the yield decreased to 36% (entry 11), and the Mukaiyama hydration product was isolated instead as the major product. Other silanes, such as TES and PHMS, gave only trace amounts of 3a (entries 12 and 13). Higher temperatures were deleterious to the yield (60 °C, 56% yield, entry 14) as was running the reaction with no precautions (open to air and light, 54% yield, entry 15).Substrate scopeWith the optimized conditions in hand, the scope of building blocks containing unactivated olefins was broadly evaluated. Monosubstituted olefins (Fig. 2A), cyclic or acyclic disubstituted olefins (Fig. 2B), cyclic or acyclic trisubstituted olefins (Fig. 2C), and cis-cyclic olefins (Fig. 2D) all performed well, affording the desired diaziridine products as single regioisomers in good to excellent yields. Nitroarenes (3b) and classical aryl and aliphatic amine protecting groups were tested (e.g. Boc (3c, 3n, 3o, 3aa and 5a/b), Ac (3d), Cbz (3e) and Ts (3p, 3w and 3x); all were found to be compatible with the reaction conditions and, importantly, offer the opportunity for downstream synthetic manipulations of the orthogonally protected diamine motifs. In a similar vein, a free amine was tolerated (3f, 54% yield), and the presence of a nitrile group (3j) was also supported. Free hydroxy groups (3q, 4a) were compatible, as were many of their common protecting groups including acetyl (3r), MOM (3s), THP (3t), Cbz (3u), Bn (3v) and TBS (4e). Phenolic (3a-f, 3g) and thiophenolic ethers (3l), ketones (4b) and esters (4f and 4g) could be incorporated affording the desired products in good to excellent yields.Fig. 2: Scope of the hydroamination – building blocks and pharmaceutical core structures.A Monosubstituted alkene examples. B Disubstituted alkene examples. C Trisubstituted alkene examples. D Cyclic alkene examples. Reactions conducted with alkene (0.1 mmol), 1 (0.15 mmol, 1.5 eq), cat-1 (0.005 mmol, 0.05 eq), t-BuOOt-Bu (0.1 mmol, 1 eq), PhSiH3 (0.1 mmol, 1 eq), DCE (400 µL) and IPA (100 µL) at 40 °C for 20 h under argon atmosphere with protection from light. Isolated yields are reported. aReaction run at 0.1 mmol scale with perfluorinated diazirine 1. bReaction run at 1 g scale. cReaction run with IPA (100 µL) as the only solvent for 30 h. dReaction conducted with alkene (0.1 mmol), 1 (0.15 mmol, 1.5 eq), Mn(dpm)3 (0.005 mmol, 0.05 eq), PhSiH3 (0.1 mmol, 1 eq), DCE (400 µL) and IPA (100 µL) at 0 °C for 2 h under argon atmosphere with protection from light. Mn(dpm)3 = Tris(2,2,6,6-tetramethyl-3,5-heptanedionato)manganese(III). DCE = dichloroethane, IPA = isopropyl alcohol.Various simple hydrocarbons were successfully employed: linear alkenes (3h, 3i), methylenecyclohexane derivatives (3ab, 3ac) phenyl derivatives (3k, 3m, 3y), an adamantyl derivative (3z), methyl cyclohexenes (4c, 4d), norbornene (5c/d), and cyclooctene (5f/g). Heterocyclic substrates included Boc- or Ts-protected piperidines (3o, 3p, 3w, 3x, 3aa), an azepine (5a/b), and an 8-amino quinoline (3n). This last example is notable since both the starting material and diaziridine product may act as a ligand with the catalyst, yet 3n is still obtained in 54% isolated yield. Cis-olefins were generally hydroaminated in good yields (5c, 5f), however with azepine derivative 5a, a lower isolated yield was obtained (44%) due to loss of the Boc protecting group. This appears unique to this example since this phenomenon was not observed with other Boc-protected substrates (e.g. 3o, 3n, 3aa). While the cobalt-catalyzed protocol worked well in these examples, for symmetrical alkenes (Fig. 2D) the manganese-catalyzed version proved superior, giving the desired products in shorter reaction times with better yields (5b, 5d, 5g). To our delight, [1.1.1]propellane engaged well with the alternative manganese conditions, delivering 5e in 72% yield, which affords a different pathway into amine, hydrazine, and heterocyclic bicyclopentyl building blocks25.The reaction was conducted on gram-scale where 3h was isolated in 73% yield. Previously we demonstrated the use of a perfluorinated diazirine reagent (with C8F17) that allowed for both the initial diazirine reaction and subsequent diaziridine cleavage to be incorporated into fluorous phase workflows20. This eliminates the need for column chromatography and expedites the library synthesis of nitrogen-containing compounds for discovery scientists. Monosubstituted olefins (3g), disubstituted olefins (3x), and trisubstituted olefins (4d) all gave the expected hydroamination products, though 4d suffered a loss in yield. Many of the olefins displayed in Fig. 2 can be readily crafted into pharmaceuticals through known transformations including clobenzorex (from 3k), labetalol (from 3m), neramexane (from 3ac), quinocide (from 3n) and azelastine (from 5a/b).Having evaluated the initial scope and functional group compatibility of the hydroamination we turned our attention to the late-stage functionalization of more complex natural products with an emphasis on terpenoids (Fig. 3). Terpenes and terpenoids, many of which are commercially available, are not only common in the food and fragrance industries, but also serve as effective building blocks in medicinal chemistry and natural product synthesis. While these scaffolds appear sporadically throughout the hydrofunctionalization literature, to date there is no general hydroamination that has been demonstrated to be effective across numerous members of the class. Classical methods reported for C–N bond formation on terpenes include the Ritter reaction26, mercuration with anilines and azides27,28, and others. More recent hydrofunctionalization methods include Boger’s hydroazidation of citronellol29, Carreira’s hydrohydrazination of camphene8, Glorius’ iminative bis-functionalization of perillol, limonene, and nootkatone30, Lin and Xu’s independent approaches to the electrochemical diazidation of limonene oxide and nootkatone31,32, and Engle’s aminoarylation of sclareol33. Monosubstituted olefins that successfully underwent the diazirine-based hydroamination include eugenol (6a) and sclareol (6i). Reaction on the disubstituted isopropenyl group of nookatone (6b), cannabidiol (6c), perillol (6d), limonene oxide (6e), isopulegol (6f), valencene (6g), and limonene (6k) all afforded the desired products in moderate to good yields. Bisabolol (6h), citronellic acid (6l), citronellal (6m), and citronellol (6n) serve as examples of suitable trisubstituted prenyl groups that were able to undergo the hydroamination. Finally, camphene was converted to 6j selectively as the exo-product in 75% yield.Fig. 3: Scope of the hydroamination – terpene natural products.Reactions conducted with alkene (0.1 mmol), 1 (0.15 mmol, 1.5 eq), cat-1 (0.005 mmol, 0.05 eq), t-BuOOt-Bu (0.1 mmol, 1 eq), PhSiH3 (0.1 mmol, 1 eq), DCE (400 µL) and IPA (100 µL) at 40 °C for 20 h under argon atmosphere with protection from light. Isolated yields are reported. aReaction run with IPA (500 µL) as only solvent for 30 h. DCE dichloroethane, IPA isopropyl alcohol.Importantly, each of the above reactions were performed directly on the natural products, without the need for protecting groups. As a result, the functional group tolerance was further illuminated: α,β-unsaturated ketones (6b), free phenols (6a, 6c), allylic alcohols (6d), aliphatic alcohols (6f, 6h, 6i, 6n), epoxides (6e), carboxylic acids (6l) and aldehydes (6m) were all tolerated. Notably, kinetically-driven chemoselectivity was observed whereby disubstituted olefins reacted significantly faster than trisubstituted olefins. Thus, for cannabidiol (6c), perillol (6d), valencene (6g), bisabolol (6h), and limonene (6k) the observed products were a result of mono-hydroamination only on the isopropenyl group, leaving the trisubstituted olefins untouched. However, when the reaction time was increased to 30 h in the presence of a twofold amount of diazirine, catalyst, peroxide, and silane, the bis-hydroamination products appeared and slowly became more dominant. Presumably this is a steric effect of the initial binding of the metal and migratory insertion to the olefin. However, in the mono-aminated substrates displayed in Figs. 2 and 3, the steric environment around (but not directly attached to) the reactive olefin appears to have significantly less influence as shown in 3o, 6c, and 6j.Synthetic applicationsWith the substrate scope in hand, we endeavored to highlight the utility of the diaziridine intermediates through their conversion to amines, hydrazines, and various heterocyclic species. As shown in Fig. 4, this was achieved with a variety of target-oriented syntheses, diversity-oriented approaches, and late-stage functionalization, all from diazirine reagent 1. Primaquine, a member of the 8-aminoquinoline class of drugs, is used for the prevention and treatment of malaria and is found on the World Health Organization’s List of Essential Medicines. Quinocide is the constitutional isomer of primaquine, and the major contaminant formed during its synthesis34. Access to quinocide (8) is critical for quality control and impurity assessment of primaquine; however, its availability is somewhat limited and at an exorbitant cost. The treatment of quinoline 7 under the cobalt-catalyzed conditions gave diaziridine 3n in 54% yield, which was then exposed to TMSCl and LiCl in DMF to reveal the amine with concomitant cleavage of the Boc group to produce quinocide•HCl (8) in 80% yield (Fig. 4A). This compares favorably to the previous route where the amine was installed via a six-step sequence originating from nitroethane35. Mecamylamine is an antagonist of nicotinic acetylcholine receptors that is used for smoking cessation and hypertension36. Following the selective hydroamination of camphene (9) as previously discussed, diaziridine 6j was converted to the free amine with HI (82% yield) followed by reductive amination with paraformaldehyde to afford mecamylamine•HCl (10) in 62% yield. This provides a more tractable preparatory scale route for discovery, which avoids the use of hydrogen cyanide that is common in the industrial process (Fig. 4A). Neramexane is an NMDA antagonist that is being clinically investigated for a number of indications including Alzheimer’s disease and tinnitus37,38. The reaction of diazirine 1 with 1,1,3,3-tetramethyl-5-methylenecyclohexane (11) produced diaziridine 3ac in 59% yield, which was hydrolyzed in the presence of HI to furnish neramexane•HCl (12) in 57% yield (Fig. 4A).Fig. 4: Synthetic applications: Target- and diversity-oriented synthesis of pharmaceuticals and related compounds via hydroamination.A Target-oriented synthesis of mecamylamine, neramexane, and quinocide. B Diversity-oriented synthesis of four piperidine-based fragments of a muscarinic M1 receptor candidate. C Target-oriented and late-stage functionalization approaches to a splicing modulator candidate.Synthetic medicinal chemistry workflows are best expedited through common intermediates that can be readily diversified for structure-activity relationship (SAR) studies. Azaspirocycle 19 is a muscarinic M1 receptor agonist from Heptares Therapeutics with potential applications in various neurological disorders39. Piperidine building blocks 15, 16, and 18 were targeted by the researchers via three distinct routes and three different sets of intermediates using classical chemistry: reductive aminations, SN2 displacements, and nucleophilic additions. Piperidine 17, an obvious derivative to be examined for SAR in this subset of fragments, could not be prepared through these routes and thus was not evaluated. In a more streamlined approach, piperidines 13 and 14 were subjected to the hydroamination conditions with diazirine 1 to afford diaziridines 3p and 3w in 87% and 67% yield, respectively. Treatment with either pyrazole-forming conditions (Method A) or cleavage to the free amine (Method B) rapidly delivered all four building blocks (15–18) in 55-77% yield (Fig. 4B). Taken together with our previously reported diversification reactions, it is easily envisioned how a small number of strategically chosen intermediates can be multiplied into a large library of medicinally relevant scaffolds covering a significant amount of chemical space.Phthalazinone 25 is a splicing modulator, developed by Remix Therapeutics, that may be useful in treating disease through targeting RNA40. In the original route, it was prepared from 6-bromophthalazin-1(2H)-one over three steps in ca. 3% overall yield. The lowest yielding steps are the Mitsunobu reaction to install the azepine, followed by deprotection (Fig. 4C, inset). By leveraging the manganese-catalyzed version of our hydroamination, symmetrical cyclic olefin 20 was smoothly converted into diaziridine 5b in 85% yield. Condensation of 5b with ester 21 effected a late-stage functionalization to afford splicing modulator 25 in 64% yield. This approach can be used not only to rapidly prepare 25 in high yield (54% yield over two steps from olefin 20), but also interrogate the SAR around the azepine. Alternatively, a three-step approach can be conducted that would efficiently enable the late-stage variation of heterocycle 25. Here, diaziridine 5b was first condensed with 5-bromo-3-hydroxyisobenzofuran-1(3H)-one 22 in 89% yield, followed by Suzuki coupling with 24 to afford the desired product in 77% yield (58% yield over three steps from olefin 20) (Fig. 4C).
15N isotopic labelingThe stable isotopic labeling of small molecules is an important analytical tool used across a variety of fields due to its low natural isotopic abundance (e.g. 2H, 13C, 15N), which delivers a high signal to noise ratio when observed in NMR, MS, and MRI. Recent advances in newer techniques such as hyperpolarization further increase the sensitivity41. The synthesis of 15N-labeled pharmaceuticals and agrochemicals allows for the in vivo study of both the metabolism and environmental fate of candidate molecules (Fig. 5A)42. 15N NMR has also been used to study protein packing and the conformation of protein-ligand complexes43. The incorporation of 15N monitoring by mass spectrometry enables the quantification of the metabolism of proteins and other biomolecules (e.g. metabolomics)44. The elucidation of organic and organometallic mechanisms is also enhanced through the monitoring of reaction intermediates and determination of kinetics45.Fig. 5: Late-stage 15N installation and its potential applications.A Applications of 15N-containing molecules. B Hydroamination with 15N-diazirine to insert 15N-atoms into target molecules. C Previous unlabeled decagram synthesis of diazirine 1. D Redesigned synthesis of 15N-diazirine (15N-1) from 15N-ammonia as the only nitrogen source. E Synthesis of 15N-containing drug candidate 15N-25. (Panel A Created with BioRender.com released under a Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International license (https://creativecommons.org/licenses/by-nc-nd/4.0/deed.en)).Despite the utility of 15N-containing molecules, their syntheses still predominantly rely on 15N-ammonium salts46, or a small number of simple building blocks such as 15N-ammonia47, 15N-hydroxylamine47, 15N-hydrazine monohydrate48, and 15N-urea49. In nearly all cases, this necessitates a time-consuming radical redesign of the synthetic route for the small molecule that needs to be labeled. Several efforts have been made toward the development of an electrophilic 15N-nitrogen source. Jones reported a 15N-diazonium reagent that afforded an azo species en route to the preparation of 15N-labeled deoxynucleosides50. Unkefer developed 1-chloro-1-[15N]nitrosocyclohexane for the generation of 15N-labeled amino acids51. While useful for their given targets, both approaches lack generality. Bis-15N labeled diazirine (1) can serve as a more universal 15N-labeled electrophilic nitrogen source when coupled not only with the hydroamination chemistry described above, but also with the previously reported decarboxylative aminations. The mild reactions conditions allow for the label(s) to be installed at a late-stage, minimizing cost and waste, and the diversification of the diaziridines affords immediate access to myriad 15N amines, hydrazines, and nitrogen-containing heterocycles. Most importantly, the tedious redesign of synthetic routes would no longer be required when unlabeled diazirine 1 is used for the initial synthesis (Fig. 5B).The current synthesis of diazirine 1 requires using hydroxylamine to install the first nitrogen atom and a large excess of liquid ammonia to install the second. This route is neither practical nor economically feasible for the synthesis of 15N-labeled diazirine 1 with ca. 100 eq of 15N-ammonia going to waste (Fig. 5C). While other methods exist for the conversion of ketones to diazirines, none are suitable for α,α,α-trifluoroacetophenone47,52. Instead, we endeavored to devise a new route to 15N-1 that used 15NH3 as the lone source of isotopically labeled nitrogen (Fig. 5D). Toward this end, a modified literature procedure was used to generate 15N-phosphorus-nitrogen ylide 3053, which required ca. 7 equivalents of 15NH3. This bench stable intermediate was treated with n-BuLi and ketone 26 to affect a Wittig-like reaction that furnished the corresponding 15N-labeled imine. Treatment of the imine with t-BuOCl followed by 15NH3 (ca. 5 equiv.) allowed the oxidation/cyclization sequence to proceed and delivered 15N-diaziridine 32. Oxidation with NaOCl and catalytic TBAI gave 15N-1 (>95% yield by NMR). To demonstrate the effectiveness of the late-stage isotopic incorporation, cyclic olefin 21 was treated with 15N-1 under the manganese-catalyzed hydroamination conditions, which furnished 15N-labeled diaziridine 33 in 72% yield. Condensation of 15N-diaziridine 33 with ester 21 delivered splicing modulator 15N-25 in 75% yield (Fig. 5E). The simple reagent synthesis, mild late-stage functionalization conditions, and ability to engage 15N-1 under any present or future diazirine protocol makes this an attractive option for stable isotope incorporation.
Regioselective hydroamination of unactivated olefins with diazirines as a diversifiable nitrogen source
