Jeremy
Nugent‡
a,
Adrián
López-Francés‡
b,
Alistair J.
Sterling‡
a,
Min Yi
Tay
a,
Nils
Frank
a,
James J.
Mousseau
c,
Fernanda
Duarte
*a and
Edward A.
Anderson
*a
aDepartment of Chemistry, Chemistry Research Laboratory, University of Oxford, 12 Mansfield Road, Oxford, OX1 3TA, UK. E-mail: fernandaduartegonzalez@chem.ox.ac.uk; edward.anderson@chem.ox.ac.uk
bDepartment of Organic Chemistry I, Faculty of Pharmacy and Lascaray Research Center, University of the Basque Country, UPV/EHU, Paseo de la Universidad 7, 01006, Vitoria-Gasteiz, Spain
cPfizer Worldwide Research and Development, Eastern Point Road, Groton, Connecticut 06340, USA
First published on 4th June 2024
Bridged bicycloalkanes such as bicyclo[1.1.1]pentanes (BCPs) and bicyclo[3.1.1]heptanes (BCHeps) are important motifs in contemporary drug design due to their potential to act as bioisosteres of disubstituted benzene rings, often resulting in compounds with improved physicochemical and pharmacokinetic properties. Access to such motifs with proximal nitrogen atoms (i.e. α-amino/amido bicycloalkanes) is highly desirable for drug discovery applications, but their synthesis is challenging. Here we report an approach to α-amino BCPs and BCHeps through the visible-light enabled addition of α-amino radicals to the interbridgehead C–C bonds of [1.1.1] and [3.1.1]propellane respectively. The reaction proceeds under exceptionally mild conditions and displays broad substrate scope, providing access to an array of medicinally-relevant BCP and BCHep products. Experimental and computational mechanistic studies provide evidence for a radical chain pathway which depends critically on the stability of the α-amino radical, as well as effective catalyst turnover.
These important rigid scaffolds are typically derived from [n.1.1]propellanes, which are convenient building blocks due to the diversity of functionality that can be introduced during ring-opening of the central C–C bond, especially using radicals19–32 and, for [1.1.1]propellane, anions.33–38 In the case of mono-substituted BCPs, synthetic approaches are most commonly anionic in nature (Fig. 1b); examples include the addition to 1 of aryl Grignard reagents,12 turbo amides,34,35 enolates,39 dithiane36 and azaallyl37,38 anions. While these methods provide ready access to valuable BCP building blocks, they are moisture-and/or air-sensitive and thus display limited functional group tolerance. Single electron strategies also enable the synthesis of monosubstituted BCPs, but have generally been limited to electron-deficient or thiyl (and related) radicals.24–28
α-Amino BCPs are highly desirable in medicinal chemistry as analogues of benzylamines – motifs found in many pharmaceuticals.40 The synthesis of these potentially valuable compounds has been mostly overlooked, with the few reported examples requiring lengthy reaction sequences, pyrophoric/strongly basic reagents, or being limited to 1° amines.37,41–44 As such, the synthesis of α-amino BCPs or BCHeps directly from [1.1.1]propellane 1 and [3.1.1]propellane 2 respectively represents an attractive yet unexplored route – especially in the latter case, as anionic additions to 2 are unfeasible.37 While these propellanes are well-established to react efficiently with electrophilic radicals,13,25,45,46 the addition of nucleophilic radicals is less studied.47 We questioned whether the direct addition of nucleophilic α-amino radicals (generated via photoredox-catalysed oxidation of simple N,N-dialkylanilines)48–58 to propellanes 1 or 2 could generate these useful α-amino bicycloalkanes in a single step. Here we report the successful development of this methodology, which represents the first examples of the ring-opening of [1.1.1] and [3.1.1]propellanes using α-amino radicals. We complement the development of this chemistry with a detailed mechanistic study that investigates the role of each reaction component, including the source of the BCP/BCHep bridgehead hydrogen atom.
Entry | PC | Solvent | Amine eq. | Light source | Time (h) | Yielda (%) (4a:5a) |
---|---|---|---|---|---|---|
a Yield determined by 1H NMR spectroscopy using trimethoxybenzene as internal standard. b 1 mol% of catalyst. c 2.5 mol% of catalyst. d 10 equiv. of water added. e Under air. PC = photocatalyst. | ||||||
1 | [Ir]1b | MeCN | 5 | 18 W 455 nm | 48 | 25 (4:1) |
2 | [Ir]1b | DMF | 5 | 18 W 455 nm | 48 | 42 (4:1) |
3 | [Ir]1b | DCE | 5 | 18 W 455 nm | 48 | 43 (4:1) |
4 | [Ir]1b | DMA | 5 | 18 W 455 nm | 48 | 36 (4:1) |
5 | [Ir]2b | DMA | 5 | 18 W 455 nm | 48 | 15 (4:1) |
6 | [Ir]3b | DMA | 5 | 18 W 455 nm | 48 | 35 (4:1) |
7 | 4CzIPNc | DMA | 5 | 18 W 455 nm | 48 | 45 (4:1) |
8 | 4CzIPNc | DMA | 10 | 18 W 455 nm | 48 | 60 (6:1) |
9 | 4CzIPNc | DMA | 10 | 30 W 440 nm | 24 | 65 (6:1) |
10 | 4CzIPNc , | DMA | 10 | 30 W 440 nm | 24 | 70 (6:1) |
11 | 4CzIPNc,d,e | DMA | 10 | 30 W 440 nm | 24 | 62 (6:1) |
12 | 4CzIPNc,d | DMA | 10 | None | 24 | <5 |
13 | Noned | DMA | 10 | 30 W 440 nm | 24 | 20 (6:1) |
With optimised conditions in hand, the scope of the aniline coupling partner was investigated, focussing first on variation of the N-arene substituent. We found this method of BCP installation to be successful with diversely functionalised (hetero)arylpyrrolidine substrates, with the α-amino BCP products generally obtained in good yields (Scheme 1). Electron-neutral and electron-poor para-substituted aniline substrates are well-suited to this reaction and gave good-to-excellent yields of the desired α-amino BCP products (4a–f, 52–80%). Substitution at the meta- and ortho-positions was also tolerated (4g–h, 35–46%), as were trisubstituted anilines (4i, 60%). The synthesis of BCPs substituted with heteroaryl dialkylanilines would be of high interest in a pharmaceutical context; pleasingly, we found that 2-, 3-, and 4-pyrrolidinopyridines were excellent substrates for this reaction, affording BCP products in high yields (4j–l, 63–70%); similarly a pyrrolidine–quinoline derivatives gave the BCP product 4m in good yield (61%).
We next investigated substrates in which the dialkylamine was varied. Pleasingly, piperidine (4n, 52%), azepane (4o–p, 43–47%) and morpholine (4q, 40%) substituted (hetero)arene BCPs were isolated in good yields, albeit with a slight increase in the amount of staffane side-product. Substrates bearing substituted pyrrolidines proved more challenging: while hexahydrocyclopenta[c]pyrrole 4q was obtained in excellent yield (71%), the use of non-symmetric substrates resulted in mixtures of product regioisomers, although high yields were still obtained (e.g.4r and 4s, 61%). For reasons that are unclear, acyclic dialkylamine systems generally resulted in low yields, with significant amounts of staffane formation (4u, 23%).
Recent studies suggest that nicotine may exert neuroprotective effects inducing defence mechanisms against pathologies associated with Alzheimer's or Parkinson's disease.19–28 Pleasingly, use of nicotine (6, Scheme 2) as a substrate for this α-aminobicyclopentylation reaction led to the corresponding BCP–nicotine derivative 7 (37%), highlighting the applicability of the chemistry to drug molecules. C–H abstraction occurs preferentially at the 2° position adjacent to the nitrogen atom due to the stability of the resulting α-amino radical.61 While a 3° radical at the opposing α-position should be significantly more stable than any of the 1° or 2° radicals that could be formed, we suggest that steric repulsion encountered during the C–H abstraction process prevents the formation of this radical, rationalising the observed regioselectivity of the reaction.
Scheme 2 Bicyclopentylation of nicotine (6, 10 equiv.); 7 was isolated as a 4:1 mixture with the corresponding BCP-staffane. |
Considering the excellent results obtained in the bicyclopentylation reactions of 1 to form α-amino BCPs, and the high relevance of recently discovered BCHeps as potential bioisosteres of meta-substituted arenes,13 we questioned whether [3.1.1]propellane 2 would also perform well in this nucleophilic radical addition chemistry. We first studied the stability of 2 under blue LED irradiation (440 nm), which confirmed that 2 is stable for several hours.61 Under the optimised conditions developed for the bicyclopentylation, we explored the reaction of 2 with aniline 3a. The use of 1 equiv. of amine 3a under the optimized reaction conditions afforded the desired product 8a in low yield (18%, Table 2, entry 1), as a 4:1 mixture with the corresponding BCHep staffane 9a (which notably constitutes the first example of this type of ‘dimer’ formation for [3.1.1]propellane).13 An increase in yield but a similar product-staffane ratio was observed using 5 equiv. of amine 3a (47% (4:1), entry 2), while the use of 10 equiv. of amine resulted in enhancement of both the yield and product:staffane ratio (82% (5:1), entry 3).
These conditions were applied to a range of amine substrates (Scheme 3). We observed that anilines featuring neutral and electron-withdrawing N-aryl groups were well-tolerated, affording the corresponding α-amino BCHeps in good to excellent yields (8a–c, 52–82%), while more electron-deficient N-aryls (para-fluorine or meta-cyano substitution) led to low yields of BCHep product (8d–e, 20–21%). Notably, the replacement of the N-aryl and pyrrolidine rings with pyridine or morpholine motifs respectively was successful, generating BCHeps that feature multiple ‘drug-like’ functionalities (8f–g, 43–56%). Interestingly, only phenyl, para-fluorophenyl and pyridine corresponding BCHep staffanes (9a, 9d, 9f); staffane formation was not observed for other substrates.
Scheme 3 Substrate scope for α-amino bicyclo[3.1.1]heptane reaction; isolated yields shown. a Isolated as a 5:1 mixture with 9a (from 8a) or 9d (from 8d). b Isolated as a 10:1 mixture with 9f. |
Fig. 2 (a) Proposed mechanistic cycle for the formation of α-amino BCPs from amines and [1.1.1]propellane 1 (n = 1) or [3.1.1]propellane 2 (n = 3). (b) Stern–Volmer quenching plot for amine 3a and [1.1.1]propellane 1 with 4CzIPN. (c) Calculated E° values (V vs. SCE) for 3a, 12 and 14. (d) Calculated H-atom transfer barriers using amines 3a, 3u, DMA, and Et2O.64 Free energies were calculated at 298.15 K, and the standard concentration of each species was adjusted for the experimental molar ratios (3a/3u: 10.0 equiv., DMA: 21.6 equiv. Et2O: 12.0 equiv.). |
The preference for HAT transfer from 3a, rather than the solvent or H2O, was further explored using deuterium-labelling studies (Table 3). We first confirmed that under the standard conditions, no deuterium incorporation was observed in the presence of D2O, ruling out the reduction of the BCP radical as a catalyst turnover step (entry 2). Use of d7-DMF (as a surrogate for DMA) also led to no product deuteration (entry 3). However, 80% D-incorporation was observed using d4-N-phenylpyrrolidine (entry 4, d4-3a), albeit this reaction proceeded in very low yield. A significantly greater amount of staffane was observed, which is consistent with the slower rate of deuterium atom transfer compared to HAT with h4-3a (4a:5a = 1.3:1 vs. 6.4:1). Only 34% D-incorporation was observed when using d4-N-phenylpyrollidine in combination with a DMA/Et2O solvent mixture (entry 5), suggesting that these solvents may also act as H-atom sources in the presence of deuterated substrate.
Entry | Substrates | Solvent | Additive | % D | Yield (%) (4a:5a)/(8a:9a) |
---|---|---|---|---|---|
a d4-3a = 98% D. | |||||
1 | 1 + h4-3a | DMA/Et2O | H2O | 0 | 70 (6.4:1) |
2 | 1 + h4-3a | DMA/Et2O | D2O | 0 | 70 (6.4:1) |
3 | 1 + h4-3a | d7-DMF/pentane | D2O | 0 | 70 (6.4:1) |
4a | 1 + d4-3a | d7-DMF/pentane | D2O | 80 | 7 (1.3:1) |
5a | 1 + d4-3a | DMA/Et2O | D2O | 34 | 10 (1.6:1) |
6a | 1 + d4-3a | d7-DMF/Et2O | D2O | 63 | 10 (1.5:1) |
7 | 2 + h4-3a | DMA/n-Bu2O | H2O | 0 | 82 (5:1) |
8a | 2 + d4-3a | DMA/n-Bu2O | H2O | 38 | 19 (3.5:1) |
9 | 2 + h4-3a | DMA/n-Bu2O | D2O | 21 | 79 (5:1) |
10a | 2 + d4-3a | DMA/n-Bu2O | D2O | 49 | 10 (3.5:1) |
11a | 2 + d4-3a | d7-DMF/n-Bu2O | D2O | 74 | 8 (2:1) |
Equivalent deuteration studies were next performed using [31.1]propellane 2 as acceptor, which confirmed that the substrate 3a is again a capable H atom donor, with 38% D-incorporation using d4-N-phenylpyrrolidine, DMA and H2O (entries 7 and 8). However, additional experiments revealed that in the case of 2, use of D2O resulted in a surprising 21% D-incorporation and a much superior yield (entry 9). Furthermore, 49% deuteration was observed using a combination of d4-3a and D2O (entry 10), and the introduction of d7-DMF further increased the extent of deuteration to 74%, confirming the participation of multiple H-atom sources, including the solvent (entry 11). The BCHep/staffane ratios gradually decreased from 5:1 (entries 7 and 8), progressing to 3.5:1 (entries 9 and 10), and finally reaching 2:1 (entry 11), showing that the bicycloheptylation reaction also features a fine balance between HAT and staffane formation.
Additional evidence for our mechanistic proposal was obtained using the kinetic isotope effect (KIE) observed for the HAT step in the reactions of 1. Since staffane formation is independent of the deuteration state of the amine, the product:staffane ratio for h4- and d4-N-phenylpyrollidine (Table 3, entries 1 vs. 4) should approximate the HAT KIE, i.e. kH/kD ≃ (4a:5a)/(d-4a:d-5a) (see ESI† for further discussion). Using this approach, a KIE of 4.9 ± 0.5 was obtained, which is in reasonable agreement with the value obtained from computation (kH/kD = 6.1).
While a chain process is evidently possible, the low quantum yield of 0.84 obtained for this reaction suggests catalyst turnover is important.61 Since reduction of the BCP radical is not viable to achieve turnover (at least for [1.1.1]propellane, as evidenced by the lack of deuteration of 4a in the presence D2O for this propellane), it may be that a fast HAT process between BCP radical 12 and low-concentration radical cation 10 could occur that would lead to the formation of iminium ion 14. This iminium ion could then be reduced by 4CzIPN− to reform the neutral organophotoredox catalyst, and an α-amino radical 11 (ΔEcalc = +0.88 V).
Finally, the successful use of different propellanes in this aminobicycloalkylation offers the opportunity to compare the relative propensity of the two to undergo ring-opening. A competition experiment was therefore undertaken in which the reaction was conducted using an equimolar mixture of 1 and 2; this experiment revealed that the formation of BCP derivative 4e is ∼3.9 times faster than BCHep product 8d as judged by NMR monitoring of the proportions of products formed during the reaction.61 This reveals an enhanced reactivity of [1.1.1]propellane 1 compared to [3.1.1]propellane 2, at least in this particular setting of nucleophilic radical addition chemistry (Scheme 4).
Footnotes |
† Electronic supplementary information (ESI) available: Experimental procedures, computational details, copies of 1H and 13C NMR spectra. See DOI: https://doi.org/10.1039/d4sc01368a |
‡ These authors contributed equally. |
This journal is © The Royal Society of Chemistry 2024 |