Miguel A.
Valle-Amores
a,
Claudia
Feberero
ab,
Ana
Martin-Somer
c,
Sergio
Díaz-Tendero
def,
Andrew D.
Smith
g,
Alberto
Fraile
*ae and
José
Alemán
*ae
aOrganic Chemistry Department (Módulo 1), Facultad de Ciencias, Universidad Autónoma de Madrid, 28049 Madrid, Spain. E-mail: alberto.fraile@uam.es; jose.aleman@uam.es; Web: https://www.uam.es/jose.aleman
bOrganic Chemistry Area, Chemistry Department, Universidad de Burgos, Burgos 09001, Spain
cApplied Physical Chemistry Department (Módulo 14), Facultad de Ciencias, Universidad Autónoma de Madrid, 28049 Madrid, Spain
dChemistry Department (Módulo 13), Facultad de Ciencias, Universidad Autónoma de Madrid, 28049 Madrid, Spain
eInstitute for Advanced Research in Chemical Sciences (IAdChem), Universidad Autónoma de Madrid, 28049 Madrid, Spain
fCondensed Matter Physics Center (IFIMAC), Universidad Autónoma de Madrid, 28049 Madrid, Spain
gEaStCHEM, School of Chemistry, University of St Andrews, North Haugh, Fife, Scotland, KY16 9ST, UK
First published on 11th October 2023
Herein, a formal highly enantioselective organocatalyzed [3 + 2] cycloaddition of furanone derivatives and azomethine ylides is presented. The success of this reaction resides in intramolecular hydrogen bond activation through an o-hydroxy group at the aromatic ring of the imine, allowing the formation of highly multifunctional bicyclic adducts with five stereogenic centers in a stereocontrolled manner. Furthermore, the reaction is paired with a highly efficient kinetic resolution of butenolides, achieving selectivity factors above 200. Using this methodology, furan-2(5H)-ones and furo[3,4-c]pyrrolidinones were obtained with high enantioselectivities. Quantum chemistry calculations reveal the crucial role of the hydrogen bond formed between the catalyst donor-units and the two reactants, which modifies their arrangement and promotes effective facial discrimination resulting in a highly selective kinetic resolution. In addition, further applicability of the kinetic resolution process is shown.
Regarding the different processes of kinetic resolution, we can distinguish between the use of chiral auxiliaries1c,2 and catalysts.1c,3 The principle of KR relies on the reaction of a chiral reagent or catalyst with each enantiomer of the racemic mixture that proceeds through the generation of two diastereomeric transition states. The difference in the free energy between these transition states (ΔΔG‡) dictates the difference in rate constants (k) for the reaction of each enantiomer, allowing their discrimination and determination of the efficiency of the KR using the selectivity factor (s) values.4 The catalytic KR can be divided into enzymatic,5 which has long been a popular strategy, and non-enzymatic processes,6 which include both metal catalysis7 and organocatalysis.8
Within the plethora of reactions studied for the catalytic kinetic resolution of racemates, [3 + 2] cycloadditions have been scarcely studied. The first example was described by Fu's group in 2005, who carried out the kinetic resolution of azomethine imines with activated alkynes catalysed by a chiral copper complex, achieving high enantioselectivities of the recovered dipole and with selectivity factors from 15 to 96 (Scheme 1a).9 Subsequently, other authors have published metal-catalyzed [3 + 2] kinetic resolution processes of different racemic dipolarophiles, reaching high ee and s values.10 Regarding organocatalyzed [3 + 2] cycloadditions, to date, there are only two precedents. The first was reported by Xie's group in 2010, who carried out the [3 + 2] cycloaddition of azomethine ylides to nitroolefins catalyzed by Takemoto's organocatalyst. While this procedure provided the KR of racemic 2-nitro-2H-chromene derivatives, the enantioselectivities obtained were from low to moderate (top, Scheme 1b).11 The second example relates to the three-component kinetic resolution catalysed by chiral bisphosphoric acids between racemic 2,3-allenoates and in situ formed azomethine ylides. In this case, 3-methylenepyrrolidine derivatives were obtained with high enantioselectivities (up to 94% ee) with the (R)-2,3-allenoates recovered in excellent enantioselectivities (up to 99% ee) (bottom, Scheme 1b).12 It should be noted that in both precedents, the authors must make use of azomethine ylide precursors that bear two electron-withdrawing (two esters are present) groups in the methylene carbon of imine which, inevitably, limits the structure of the final products.
On the other hand, furanones, a five-membered ring containing an oxygen atom, are a class of heterocyclic compounds of widespread interest in the organic, pharmacological and biological fields, containing diverse biological properties such as analgesic, anti-inflammatory, and anticancer properties among others.13
Some of the most used synthons that present the butenolide structure are the 5-alkoxyfuran-2(5H)-ones which allow achieving new approaches to acyclic and heterocyclic products.14,15 Feringa14c,16 and others15,17,18 have used butenolide derivatives with chiral auxiliaries such as menthol or sulfinyl groups to achieve diastereoselective asymmetric processes. Starting from these chiral precursors, many interesting structures with high diastereoselectivities have been synthesized. While demonstrative, green chemistry principles consider the use of chiral auxiliaries as not efficient or atom economical, emphasizing the need to develop effective catalytic processes.19 Therefore, the development of new catalytic procedures to carry out asymmetric reactions with 5-alkoxyfuran-2(5H)-ones would be highly desirable. Taking into account the scarce number of examples of organocatalyzed [3 + 2] kinetic resolution and the high synthetic value of furan-2(5H)-ones, we hypothesized that an efficient kinetic resolution could be achieved by following a match/mismatch20 process using a bifunctional organocatalyst and intramolecular hydrogen bond activation (Scheme 1c).21
![]() | ||
Scheme 2 Proof of concept for the intramolecular H-bond activation: OH role. The reactions were run with 0.1 mmol of imine 1 and 0.1 mmol of (±)-2a in 0.3 mL of p-xylene ([0.33] M). |
Entry | Cat (mol%) | Solvent | Conv.b (%) | ee 4bc (%) | ee (−)-2ac (%) | s (c (%)) |
---|---|---|---|---|---|---|
a The reaction was run with 0.1 mmol of imine 1b and 0.1 mmol of (±)-2a in 0.3 mL of indicated solvent ([0.33] M). b Conversion determined by 1H NMR. c Determined by chiral SFC. d Calculated conversion (c) = eeSM/(eeSM + eePR), selectivity factor (s) = ln[(1 − c)(1 − eeSM)]/ln[(1 − c)(1 + eeSM)]. e [0.16] M instead of [0.33] M. f The reaction was scaled up to 1.0 mmol of imine 1b. g The reaction was carried out with 0.05 mmol of imine 1b. | ||||||
1 | 3a (20) | p-Xylene | 48 | 97 | 87 | 190 (47) |
2 | — | p-Xylene | <5 | — | — | — |
3 | 3b (20) | p-Xylene | 40 | 96 | 65 | 100 (40) |
4 | 3c (20) | p-Xylene | 10 | — | — | — |
5 | 3d (20) | p-Xylene | 20 | — | — | — |
6 | 3e (20) | p-Xylene | 18 | — | — | — |
7 | 3f (20) | p-Xylene | 19 | — | — | — |
8 | 3g (20) | p-Xylene | <5 | — | — | — |
9 | 3a (20) | CH2Cl2 | 34 | 86 | 50 | 22 (37) |
10 | 3a (20) | Et2O | 47 | 95 | 91 | 120 (49) |
11 | 3a (20) | MTBE | 45 | 91 | 82 | 50 (47) |
12 | 3a (20) | Toluene | 48 | 92 | 98 | 110 (51) |
13e | 3a (20) | p-Xylene | 45 | 95 | 81 | 100 (46) |
14 | 3a (15) | p-Xylene | 42 | 97 | 76 | 150 (44) |
15 | 3a (10) | p-Xylene | 38 | 97 | 55 | 110 (36) |
16f | 3a (20) | p-Xylene | 48 | 97 | 92 | >200 (49) |
17g | 3a (20) | p-Xylene | 40 | 95 | 68 | 80 (41) |
Having determined that using imine 1b affords high selectivity, we continued looking for the optimal reaction conditions. Thus, different bifunctional organocatalysts (20 mol%) were studied, with Takemoto's thiourea catalyst 3a giving the highest selectivity for this cycloaddition reaction (compare entry 1 with entries 3–8, Table 1). Notably, a racemic background reaction was not operative, as no conversion to the product was obtained in the absence of catalyst 3a (entry 2). The use of squaramide-based catalysts led to a dramatic loss in conversion (entries 4 and 8). Interestingly, the pseudo-enantiomer catalyst 3e gave very low conversion in comparison with the organocatalyst 3b. Having identified the most promising catalyst, different solvents were screened (entries 9–12). The use of dichloromethane and tert-butylmethylether (entries 9 and 11, respectively) afforded reduced product conversions and enantioselectivities, and hence lower selectivity factors (s) than with p-xylene (entry 1). However, the reaction in diethylether and toluene (entries 10 and 12) gave rise to good conversion and enantioselectivity, but reduced s. Considering the data obtained, p-xylene was selected as the best solvent to continue the reaction screening. Subsequently, reaction concentration was also tested, with the highest selectivity at [0.33] M (entry 1) instead of [0.16] M (entry 13) (see the ESI‡). Finally, the catalyst loading was studied, with reduced product conversions and selectivity factors (s) observed when using 15 and 10 mol% catalyst (entries 14 and 15). The reaction was also scaled up to 1.0 mmol, with no detrimental effect on the stereoselectivity and obtaining the best results in terms of both enantiomeric excess and selectivity factor (4b: 47% and 2a: 46% yield, entry 16). Additionally, to determine if the use of a lower amount of imine 1b could provide similar results to that shown in entry 1, we carried out the reaction starting from 0.05 mmol of 1b (entry 17). However, the results obtained were worse, achieving only a 68% enantiomeric excess for furanone (−)-2a and a lower selectivity factor for the kinetic resolution.
With the optimized conditions in hand (c = 47%, s = 190, Table 1, entry 1), the scope of the dipolar cycloaddition and the efficiency of the KR for (±)-2a were evaluated (Scheme 3). For this purpose, a large assortment of imines 1 bearing an ortho-hydroxyl group at the aromatic ring and with different substituents were tested. When electron-donating groups are present at the aromatic ring of the imine such as the methyl (1c and 1d) or methoxy (1e and 1f) group, cycloaddition products 4c–f were obtained with high conversions and excellent enantioselectivities (94–99% ee). Moreover, the (R)-5-methoxyfuran-2(5H)-one (−)-2a could be recovered with high yields and good enantioselectivities (86–90% ee), achieving excellent selectivity factor values (s > 200) for 1c–e, while for 1f a reduced but still significant value was obtained (s = 90). However, the presence of a diethylamino group (1g) maintained the enantioselectivity of the cycloaddition (98% ee, s = 170) but led to decreased conversion (c = 25%). Furthermore, halogens at para- (1h–i) and meta-positions (1j) were employed. Cycloaddition products 4h–j were obtained with good conversions and high enantioselectivities (up to 95% ee for 4i). Regrettably, enantioselectivities of the recovered starting material (−)-2a were from moderate to good (up to 87% ee), affording reasonable selectivity factors (s = 28–80) for the resolution of (±)-2a. When a nitro group is present at the meta position (1k), the cycloadduct 4k and the furanone (−)-2a were obtained with excellent enantiomeric excess (95 and 90% ee, respectively) and, more importantly, the conversion (c = 49%) and the selectivity factor (s = 120) were very high. An ester group at the meta position (1l) afforded the cycloaddition product 4l with good enantioselectivity (91% ee), but with a decreased conversion and selectivity factor (c = 38%, s = 37). Two iminoesters with a naphthyl group (1m and 1n) were reacted at room temperature with the dipolarophile (±)-2a under the standard reaction conditions without obtaining the corresponding cycloadducts. To our delight, this limitation could be overcome by increasing the temperature to 50 °C, achieving thus 4m and 4n with excellent enantioselectivity values (93 and 97% ee, respectively) and with a high selectivity factor for both reactions (s = 90). Notably, alpha-substituted imines 1o and 1p, with a methyl or a phenyl group, respectively, worked well and allowed the asymmetric synthesis of 4-substituted pyrrolidines 4o and 4p, bearing a quaternary stereocentre, with good enantioselectivities (87 and 71% ee, respectively), high or complete diastereoselectivity (for 4p: 5:
1 dr) and increased conversion (up to 51%). Nevertheless, regarding the selectivity, the resolution was more effective with the α-methyl imine 1o (s = 45) than with the α-phenyl ring 1p (s = 11). Different electron-withdrawing groups at the imine moiety, such as N,N-dimethyl amide 1q, ethyl ester 1r, and Weinreb's amide 1s, also led to the corresponding bicyclic adducts 4q–s in good to high enantioselectivities (84–93% ee) and high conversions. The best results in terms of selectivity factor were achieved from Weinreb's amide 1s (s = 100), which allowed recovery of the (R)-5-methoxyfuran-2(5H)-one (−)-2a with excellent enantioselectivity (99% ee). The absolute configuration of the stereogenic centres of the hydroxylated cycloadducts 4 was assigned by X-ray crystallographic analysis of a monocrystal of 4j (3S,3aR,4S,6R,6aS)23 (middle, Scheme 3) and assuming the same stereochemical outcome for the rest of the products.
Finally, to demonstrate the versatility of our [3 + 2] cycloaddition kinetic resolution, we studied the influence of the incorporation of a variety of substituents at C(3) and C(5) within the furan-2(5H)-one as well as the use of 2-(5H)-pyrrolones (Scheme 4). Varying the substituent at C(5) from the methoxy to ethoxy group gave excellent conversion to the product, giving an s factor >200 for the kinetic resolution, and provided resolved (R)-5-ethoxyfuran-2(5H)-one (−)-2b in excellent enantiomeric excess.
Introducing a thiolate group resulted in the successful formation of the cycloadduct 4u with a high degree of enantioselectivity (92% ee). However, the furanone 2c was obtained with no enantioselectivity. This observation can be attributed to the high acidity of the thioacetal proton (H-5) in furanone 2c, which can lead to complete epimerization in the presence of organocatalyst 3a, resulting in a racemic mixture, followed by auto-selfcondensation as it was previously described in the literature.24 The reaction with a more reactive pseudo ester such as the 3-bromo-5-methoxyfuran-2(5H)-one (±)-2d was studied under the optimized conditions, achieving the cycloadduct 4v with good enantioselectivity (87% ee) and moderate conversion (c = 35%) and selectivity. The reactions with the aromatic derivative (±)-2e worked well, giving rise to the cycloadduct 4w with good selectivity, while the heteroaromatic species (±)-2f gave the cycloadduct 4x in excellent enantioselectivity (s = 120), but with low conversion, accounting for the recovery of the pyrrole derivative (−)-2f in low enantiomeric excess. With the incorporation of an alkenyl group the cycloaddition reaction was facile, leading to the formation of the corresponding adduct 4y and the pseudoester (−)-2g with excellent enantioselectivities and yields. Notably, in these recent examples, it was possible to generate a quaternary carbon stereocenter in a precisely controlled manner. In contrast, we also studied the reactivity of 5-methoxypyrrol-2(5H)-ones containing an acetyl substituent at the nitrogen atom. Regrettably, under the specified conditions the acetyl derivative (±)-2h reacted to give the lactam 4z and 2-(5H)-pyrrolone 2h in high yield, but as a racemic mixture. Furthermore, attempts were made to conduct [3 + 2] cycloaddition reactions using 5-methoxyfuran-2(5H)-ones containing a methyl substituent at the alpha position. However, no conversion was observed in these reactions.
Considering the excellent results obtained in the kinetic resolution of 5-methoxyfuran-2(5H)-one (±)-2a and to showcase the utility of the developed process, we envisioned the use of the enantioenriched isomer (−)-2a as the starting material for new asymmetric reactions (Scheme 5). Thus, the crude reaction product obtained during the [3 + 2] cycloaddition was submitted to a one-pot procedure. Therefore, several Michael additions14g and a [3 + 2] cycloaddition18b were carried out to demonstrate the applicability of our methodology (Proc. A, Scheme 5). Thus, the addition of (R)-1-phenylethan-1-amine or thiophenol to the crude reaction mixture from the cycloaddition (equations a and b) afforded the corresponding aza-Michael or thio-Michael products (5 and 6, respectively) with excellent enantio- and diastereo-selectivity and in high yields (only 50% of the final product can be obtained). Furthermore, the addition of 11H-dibenzo[b,e]azepine 5-oxide led to the cycloadduct 7 with excellent results (equation c).
Finally, to corroborate the configurational stability of furanone, the same reactions showed before, starting from the cycloaddition reaction crude products, were conducted from the (R)-5-methoxyfuran-2(5H)-one (−)-2a obtained by kinetic resolution and the subsequent purification by flash chromatography on silica gel (Proc. B, Scheme 5). The enantiomeric excesses achieved in these transformations brought to light that the 5-methoxyfuran-2(5H)-one (−)-2a obtained by kinetic resolution is configurationally stable and can be isolated in the enantioenriched form without any racemization.
The reaction was studied considering the endo approach for the two enantiomers of 2 (exo approaches were discarded since they are less favourable than the endo ones,15c,21b see the ESI‡).25 We considered a two-step process. The first step is the activation of the imine 1b by an H transfer from 1b to catalyst 3a. In the second step, new C–C bonds are formed, yielding the 5-membered ring (Fig. 1 and 2, respectively). For both enantiomers, the H transfer requires ∼18 kcal mol−1 from the initial pre-association complex, PAC (pre-H transfer: complex formed by the catalyst and the two reactants 1b and (±)-2a) (Fig. 1). However, the main difference was found in the second cycloaddition step (Fig. 2). While for the TS leading to the formed cycloadduct a low barrier of ∼3.5 kcal mol−1 is observed (black line), in the alternative case a transition state was located at a much higher energy, ∼9 kcal mol−1 (blue line), than the initial pre-association-complex (PAC, Fig. 2); i.e., even higher than the first TS for H transfer. Therefore, for the non-observed isomer, the much higher barrier for the cycloaddition prevents its formation, thus explaining the experimentally obtained adduct.
To obtain further molecular insights into the origin of the stereocontrol, we analysed the non-covalent interaction (NCI) at the second TS (TS CC, Fig. 3) by means of NCI plots, using the NCI code.31 NCI plots show the different interaction regions using a colour code to rank those interactions. Red is used for destabilizing interactions, blue for stabilizing interactions and green for delocalized weak interactions. The intensity of these colours is associated with the interaction strength.
For both TSs, NCI plots revealed weak delocalized interactions (green) between the three moieties interacting (Fig. 3). However, the orientations of 1b and both enantiomeric dipolarophiles 2a with respect to the catalyst are slightly different, leading to different non-covalent interactions in the C–C bond forming transition state. TS-A (observed reaction) presents a strong NH⋯O hydrogen bond (blue flat circular region) between the ammonium group, generated after the H transfer to the catalyst, and the hydroxy group of ylide 1b (top image in Fig. 3A). This leads to a stronger intramolecular hydrogen bond in 1b between the hydroxy group and the iminic nitrogen (actually it appears as H bonded to both atoms). While for TS-B, this hydrogen bond is formed with the iminic nitrogen of 1b (instead of the hydroxy group) and is weaker (only slightly blue) (top image in Fig. 3B). The significance of the hydroxyl group on the aromatic ring of the imine becomes apparent from these findings. It plays a crucial role in establishing a beneficial intermolecular interaction with the organocatalyst through hydrogen bonding. In contrast, the absence of this hydroxyl group in imine 1a prevents such an interaction, making it impossible to form a structured and organized complex between 1a and 3a. Consequently, this leads to low enantioselectivity.
Concerning the NCI between the catalyst and dipolarophile 2a, at TS-A there are two hydrogen bonds between the thiourea group in the catalyst and oxygen atoms 1 and 2 in (+)-2a, with 1 being the stronger one (bottom image in Fig. 3A). For TS-B (not-formed adduct), the different orientation of the methoxy group of (−)-2a allows the formation of a very weak hydrogen bond (green) between carbonyl oxygen atom 1 in the ylide 1b (bottom image in Fig. 3B) and the catalyst, displacing the furanone unit to the right.
This displacement leads to the formation of only one hydrogen bond between oxygen 2 in furanone (−)-2a (NH⋯O) and the catalyst, which is weaker than in TS-A (lighter blue color). In addition, for TS-B, there is a third CH⋯O bond (3), not present in TS-A, but it is very weak (green). The arrangement of the reactants at TS-B makes the region between the three moieties more crowded than in TS-A, hampering the approach of the two reactants to form the new C–C bond. This is reflected in a higher barrier for CC bond formation (TS-B energy is ∼10 kcal mol−1 higher than that of TS-A).
Footnotes |
† Dedicated to M. Rosario Martín Ramos for her great contributions to the 5-alkoxyfuran-2(5H)-one chemistry. |
‡ Electronic supplementary information (ESI) available. CCDC 2160961. For ESI and crystallographic data in CIF or other electronic format see DOI: https://doi.org/10.1039/d3qo01471a |
This journal is © the Partner Organisations 2024 |