Organocatalyzed enantio- and diastereoselective isomerization of prochiral 1,3-cyclohexanediones into nonalactones bearing distant stereocenters

Abstract

The lactonization of 2-(2-nitrophenyl)-1,3-cyclohexanediones containing an alcohol side chain and up to three distant prochiral elements is reported by isomerization under the mediation of simple organocatalysts such as quinidine. Through a process of ring expansion, strained nonalactones and decalactone are produced with up to three stereocenters in high er and dr (up to 99 : 1). Distant groups, including alkyl, aryl, carboxylate and carboxamide moieties, were examined.

Nine- and ten-membered lactones with up to three non-vicinal stereocenters (up to 99 : 1 er and dr) were prepared by ring expansion of prochiral alcohols (21 examples) upon exposure to quinidine and derivatives.

Introduction

Desymmetrization of prochiral reagents is an ambitious but rewarding strategy, as the chiral products feature one or more stereocenters, after a single-step reaction in which the catalyst binds in the vicinity of the enantiotopic groups for an effective transfer of the chiral information.¹ Applied to cyclic 2,2-disubstituted-1,3-diketones with enantiotopic carbonyl groups, this strategy was illustrated by Hajos and Parrish with the organocatalyzed enantioselective isomerization reaction of triketone 1 into ketol 2 (Scheme 1).² Even though the desymmetrization of 2,2-disubstituted-1,3-diones has been richly investigated since then,³ few reports actually showcase an enantioselective isomerization of these substrates. For example, Scheidt reported the conversion of enal 3, tethered to 2-methyl-1,3-cyclohexanedione scaffold, to β-lactone 4 in the presence of chiral N-heterocyclic carbene (NHC).⁴ Through noncovalent interactions, enone 5 was isomerized to bicyclic diketones 6 by exposure to chiral phosphoric acid (CPA), as described by Lam.⁵ Later, Dong demonstrated the enantioselective transformation of 2-acetaldehyde-1,3-cyclohexanedione 7 to bicyclic γ-lactone 8 by Rh-catalyzed ketone hydroacylation reactions.⁶ Note that the enantioselective isomerization of cyclic 1,3-diones by a process effecting a ring enlargement is unknown.

Scheme 1. Enantioselective isomerization of 2,2-disusbtituted-1,3-cyclohexanediones.

Nine-membered ring compounds have important applications in medicinal chemistry,⁷ but are challenging to synthesise due to ring strain,⁸ and their tendency to undergo transannular reactions to form bicyclic products.⁹

Amidst the enantioselective ones leading to medium-sized lactones from racemic or prochiral materials, a general strategy presented itself based on the in situ generation of electrophiles, binding to transition metal-chiral ligands or organocatalysts, reacting with enolates or olefins.¹⁰ To our knowledge, though, a single example of enantioselective synthesis of nonalactones from prochiral reagents was reported,^10a and it is hitherto unknown with distant stereocenters. Herein is reported the distal enantio- and diastereoselective ring expansion of 2-aryl-1,3-cyclohexanediones, into nonalactones with up to three distal stereocenters through an organocatalyzed isomerization.

Results and discussion

Our interest in the topic stems from our previous report of a domino sequence starting from aldehyde 9 (R = H), containing the motif 2-(2-nitrophenyl)-1,3-cyclohexanedione, which was converted into nonalactones when treated with carbanions (Scheme 1).¹¹ The resulting alcoholate initiated a series of transformations, eventually affording racemic nonalactones.

In a further step, the reactivity of the alcoholate holds promise for a enantioselective access to nonalactones 12 from alcohol 10 bearing two prochiral groups (R ≠ H). In the presence of a chiral bidentate organocatalyst, we envisaged that the hydroxyl and the carbonyl moieties of 10 would be engaged in a chiral environment allowing the enantioselective formation of the fused bicyclic lactol 11 preceding the ring expansion by retro-Claisen condensation into 12.¹² Notably, our blueprint implies a spatial discrimination of prochiral elements through noncovalent interactions.

The desymmetrization of meso cyclic acid anhydrides well illustrates the concept of activation of carbonyl groups for the addition of a nucleophile, effecting thereby the discrimination of vicinal enantiotopic groups. To that end, Oda¹³ and Aitken¹⁴ independently identified Cinchona alkaloids as efficient organocatalysts. If it was thus conceivable to promote the enantioselective isomerization of 10 with these alkaloids, whether the process of desymmetrization could establish distant stereocenters was uncertain. For that matter, different strategies were developed requiring catalysts/ligands with large structures.¹⁵

To test our hypothesis, the alcohols trans- and cis-10a were prepared from the corresponding isomers of the olefins 15a after a sequence of reactions, without purification, including ozonolysis into aldehydes 16a followed by a careful reduction (Scheme 2A). Starting from 5-methyl-1,3-cyclohexanedione 14a, the construction of 15a began with one-pot C-arylation and O-allylation steps, preceding a Claisen rearrangement for which the stereochemical outcome, the ratio trans/cis-15a (60/40), is rectifiable under Tsuji-Trost conditions.¹⁶ This is an important aspect as the configuration of alcohol 10a had an interesting bearing on the selectivity of the process (see the ESI for the optimization†).¹⁷

Scheme 2. (A and B) Preparation and enantioselective isomerization of trans-10a and cis-10a. (a) dr was measured by ¹H NMR spectroscopy of the crude reaction mixture and HPLC analysis, er was determined by HPLC analysis.

When exposed to quinidine (QD (1 equiv.), CHCl₃, 17 h, −40 to 0 °C), cis-10a was converted into trans-12a (79 : 21 diastereoisomers ratio (dr)), the relative configuration being determined by NOESY experiments, with enantioselectivity (96 : 4 enantiomers ratio (er), Scheme 2B) and with a yield of 50% yield over 3-step amounting to a theoretical 85% yield for each step. Note that operating instead with 0.5 equivalent of QD gave trans-12a with close selectivity (86 : 14 dr, 94 : 6 er) after a longer reaction time (40 h, 38% yield). On the other hand, trans-10a was lactonized into 12a (56% yield, 58 : 42 dr, trans/cis) upon exposure to QD (0.8 equiv.) in CHCl₃ at room temperature (16 h) and while trans-12a was formed without enantioselectivity, high enantiopurity (98 : 2 er) was measured for cis-12a.

Moreover, the treatment of cis-10a with quinine QN (1 equiv., −40 to 0 °C)—a natural pseudo enantiomer of QD—led to trans-12a (67% yield, 73 : 27 dr) having the attendant opposite enantioselectivity (3 : 97 er). Similarly, exposure of trans-10a to QN (0.5 equiv., rt) led to 12a (70% yield, 55 : 45 dr) and, while the formation of trans-12a occurred without enantioselectivity, the enantiopurity of cis-12a was excellent (3 : 97 er).

To sum up, both alcohols trans-10a and cis-10a were simply lactonized into 12a by an enantioselective ring expansion process upon exposure to readily available organocatalysts, that are easily recovered.

We then sought to identify the catalyst functions influencing the selectivity and, among them, the secondary benzylic alcohol appeared crucial for the enantioselectivity (Fig. 1). Bearing instead an alkyl ether or the acidic 2-nitroaniline, the corresponding catalysts induced the formation of rac-12a or were inactive.¹⁸

Fig. 1. Tuning the catalyst. (a) Reaction conditions: cis-10a, QD or 17a–17e in CHCl₃, 40 h, −40 °C, then 0 °C over 6 h; (b) dr was measured by ¹H NMR spectroscopy of the crude reaction mixture and HPLC analysis, er was determined by HPLC analysis.

While no improvement was noted with cupreidine 17a (trans-12a: 75 : 25 er), increasing the steric hindrance of the quinolin-6-ol scaffold was rewarding. Simply obtained by O-alkylation of 17a with i-butyl bromide (47% yield), 17b isomerized cis-10a to trans-12a (70 h, 51% yield) with better selectivity (90 : 10 dr, 96 : 4 er). Prepared by O-alkylation of 17a with bromocyclohexane (15% yield), 17c induced the formation of trans-12a (120 h, 63% yield) with high values of 95 : 5 dr and 96 : 4 er. For a slight modification of the ether appendage, 17d was synthesized by O-alkylation of 17a with (bromomethyl)cyclohexane in excellent yield (92%). Gratifyingly, the lactonization of cis-10a catalyzed by 17d was completed within a shorter timeframe (96 h, 60% yield) and with excellent selectivity (93 : 7 dr, 96 : 4 er). As noted with 17e, a bulkier group at this position was detrimental to trans-12a enantiopurity (90 : 10 er).

Selecting catalysts 17c and 17d, the scope of the process was first examined with prochiral 5-alkyl-2-(2-nitrophenyl)-1,3-cyclohexanediones (Scheme 3). Substituted with i-propyl group, trans-12b was isolated (62% yield—over three steps as every other given yields) with 92 : 8 dr and 99 : 1 er after exposure of cis-10b to 17d (results obtained with 17c are provided in the ESI†). Substituted with the n-propyl group, cis-10c was lactonized with excellent selectivity (trans-12c: 94 : 6 dr, 98 : 2 er). The i-butyl derivative cis-10d was transformed into trans-12d with 94 : 6 dr and 91 : 9 er, values obtained at −10 °C due to a solubility issue.

Scheme 3. Enantioselective lactonization of prochiral 2-(2-hydroxyethyl)-2-(2-nitrophenyl)-1,3-cyclohexanediones. (a) Reaction conditions: 17c, 17d or QD (1 equiv.) in CHCl₃ at −40 °C then 0 °C over 6 h, C = 0.12 M, 96 h. (b) dr was measured by both ¹H NMR spectroscopy of the crude reaction mixture and HPLC analysis, er was determined by HPLC analysis. (c) −10 °C then 0 °C over 6 h. (d) −20 °C then 0 °C over 6 h. (e) CH₂Cl₂ as co-solvant (1 : 1, v/v). (f) −40 °C to 0 °C.

We then investigated the desymmetrization of 5-aryl-2-(2-nitrophenyl)-1,3-cyclohexanediones. Starting with the lactonization of phenyl derivative cis-10e, trans-12e (71% yield, 93 : 7 dr) was isolated with moderate enantiopurity (86 : 14 er). A suitable crystal for X-ray crystallography was obtained with enhanced values (99 : 1 dr, 97 : 3 er) allowing the determination of the absolute (S, S)-configuration of the major isomer. The isomerization of the 4-methoxyphenyl derivative cis-10f led to trans-12f (49% yield) with high selectivity (95 : 5 dr, 90 : 10 er). Lactonization of the 4-fluorophenyl derivative cis-10g into 12g (64% yield, 40 h) occurred with an intriguing switch of diastereoselectivity. While the cis-lactone was so far formed in traces, cis-12g became slightly predominant (38 : 62 dr, trans/cis), the relative configuration being determined by NOESY experiments, and although the enantiopurity of cis-12g was low (67 : 33 er), trans-12g was enantioenriched (98 : 2 er).

With derivatives of 5-alkylcarboxylate-2-(2-nitrophenyl)-1,3-cyclohexanedione such as 10h, the selectivity of the process was worth examining. Unlike the previous pattern, cis-12h was isolated enantiopure (20% yield, 99 : 1 er) while trans-12h was obtained with lower enantiopurity (57% yield, 61 : 39 er), the relative configuration being determined by NOESY experiments. As noted at the beginning of the study, QD induced the lactonization of cis-10a with high enantioselectivity but with moderate diastereoselectivity. This proved to be advantageous, though, in this context as the treatment of cis-10h with QD led to an equal amount of trans- and cis-12h, the latter being isolated with an exquisite enantiopurity (43% yield, 98.5 : 1.5 er), values that remain consistent on a larger scale experiment (2.2 mmol, 41% yield, 98.5 : 1.5 er).¹⁹

Decreasing the steric hindrance of the ester, as with cis-10i, was slightly beneficial since 17d catalyzed the formation of cis-12i in 26% yield and 99 : 1 er, trans-12i (40% yield) being isolated with 65 : 35 er values. The treatment of the hindered 10j (unknown relative configuration), bearing a spiro γ-lactone moiety, afforded 12j (34% yield) with modest selectivity (77 : 23 dr, 60 : 40 er).

After examining hydrogen bond acceptors, the isomerization of scaffolds bearing also hydrogen bond donors was evaluated with the secondary benzylcarboxamide derivative cis-10k. In this setting, 17c performed best the lactonization into trans-12k (120 h, 61% yield) and, in contrast with the carboxylate esters, the trans-isomer was prevalent (94 : 6 dr) and enantioenriched (88 : 12 er). On the other hand, 17d enabled the formation of trans-12k in shorter reaction time (72 h, 60% yield, 93 : 7 dr) but with lower selectivity (78 : 22 er). Embedding a more acidic N–H bond, the anilinecarboxamide cis-10l was isomerized into trans-12l (120 h, 66% yield, 95 : 5 dr) by 17c without enhancement of the enantiopurity (85 : 15 er). But this structural modification had more consequences when exposed to 17d, as the reaction of cis-10l (72 h, trans-12l: 67% yield, 90 : 10 dr) proceeded with higher enantioselectivity (84 : 16 er) than with cis-10k (78 : 22 er). When comparing the influence of alkyl carboxylate and secondary carboxamide functions on the selectivity, the latter steers the reaction pathway toward the enantioselective production of the trans-isomer.

The formation of a trisubstituted lactone was next studied from a meso reagent. Advantageously synthesized as a single isomer, cis-10m was converted by 17c into the trisubstituted lactone 12m (40% yield), isolated as a single diastereoisomer (99 : 1 dr) and with appreciable enantioselectivity (85 : 15 er). In contrast, exposure of cis-10m to 17d gave 12m (99 : 1 dr, 55 : 45 er) with low enantioselectivity, demonstrating thus the superior affinity of 17c for the topography of meso compounds. A step further, we sought to prepare the persubstituted lactone 12n from cis/trans-10n (50 : 50 dr). In this complex scenario however, 12n (53% yield, 58 : 32 : 10 dr) was obtained without enantioselectivity (56 : 44 er).

Subsequent to the isomerization of substrates with at least two prochiral elements, the strategy was applied to 10o bearing a single prochiral group (Scheme 4). Note that it was not part of our initial plan devised in Scheme 1 since it was unclear whether the chiral information of lactol 11o would be preserved during the ring expansion into enolate (12o)⁻.

Scheme 4. Enantioselective formation of 12o–qvia the corresponding enolate. Reaction conditions: (a) 17d or QD (1 equiv.) in CHCl₃ at −40 °C, 40 h then 0 °C over 6 h. The yields are given for three steps.

We were therefore surprised to measure the high enantioselectivity of 12o (75% yield, 94 : 6 er) obtained by treatment of 10o with 17d, which dropped to 88 : 12 er when QD was instead employed. Indicative that a quaternary carbon within the 1,3-cyclohexanedione ring remains moderately compatible with the chemistry, 10p was converted into 12p (33% yield) with 72 : 28 er value. Pleasingly, bearing a longer alcohol appendage, 10q was converted into decalactone 12q with significant enantioselectivity (60% yield, 88 : 12 er).²⁰

Amidst all the structural modulations examined within the study, the aromatic substitution of the prochiral scaffold was left to explore. To that end, alcohols cis-10r,s were synthesized, each one embedding a different halo-substituted 2-nitrophenyl ring (Fig. 2).

Fig. 2. Enantioselective isomerization of cis-10r,s,v into nonalactones trans-12r,s,v. Reaction conditions: (a) 17d or QD (1 equiv.) in CHCl₃ at −40 °C, then 0 °C over 6 h. (b) dr was measured by ¹H NMR spectroscopy of the crude reaction mixture and HPLC analysis, er was determined by HPLC analysis. (c) −10 °C. The yields are given for three steps.

Compared to the case of cis-10a, 4-bromo-2-nitrophenyl derivative cis-10r was isomerized with less diastereoselectivity (12r: 78% yield, 75 : 25 dr) but enhanced enantioselectivity (98.5 : 1.5 er) when exposed to 17d. With a different pattern of substitution, 5-chloro-2-nitrophenyl derivative cis-10s was converted by 17d into 12s (65% yield, 72 : 28 dr) with excellent enantioselectivity (97.5 : 2.5 er).

To modulate the position of the nitro group, we set about preparing 2-(3-nitrophenyl)-1,3-cyclohexanedione derivative 10t. But the instability of the precursor aldehyde 9t thwarted our plan and a similar outcome was noted with the 4-nitrophenyl analogue 9u. We surmised that a steric effect, induced by the ortho substitution of the aromatic ring, could prevent the degradation of the aldehyde. Lending credence to this hypothesis, 2-(2-bromo-4-nitrophenyl) derivative 9v was successfully obtained and converted into alcohol cis-10v. In the presence of 17d, the formation of trans-12v occurred slowly at −10 °C (120 h, 60% yield) with modest selectivity (69 : 31 dr, 68 : 32 er). A slight enhancement was noted upon the catalysis of QD (40 h, −10 °C, 38% yield), trans-12v being formed with higher enantiopurity (78 : 22 er) but at the expense of the diastereoselectivity (47 : 53 dr). The nitro group in the ortho-position of the aromatic ring seems therefore favorable to achieve high stereoselectivity.

However, as illustrated in Scheme 5, this structural requirement advantageously paves the way to the synthesis of enantioriched alkaloids and derivatives. For instance, the indole 18a was prepared from 12a by reduction of the nitro group (Zn, AcOH, >95% yield) and the regioselective halogenation occurred in 6-position (19, 72% yield), offering a synthetic handle for the decoration of the aromatic ring. Underscoring the strain of the indolononalactone, the spontaneous air oxidation of the aromatic ring of 18a into hydroxylindolenine was noted (∼20% after two months of storage at rt, see the ESI for the details†). For that matter, we purposely performed the consecutive ring-expansion reaction of 18a by oxidative cleavage (NaIO₄)^21a delivering the 12-membered ring 20 (56% yield). Not only this demonstrated a short route to enantioenriched 12-membered ring from 6-membered ring 10avia12a,^21b but 20 provides a platform for further ring expansion, as illustrated by Unsworth, to access macrocycles.^21c

Scheme 5. Synthetic manipulations.

Palladium-catalyzed alkylative dearomatization of the indole 18a led to trans-allylic indolenine 21 (53% yield, non-optimized) after the distal diastereoselective formation of a quaternary carbon (97 : 3 dr).²² Bearing an additional reactive function, the aldehyde 22 was smoothly obtained after oxidative cleavage of the allylic appendage (>95% yield). Prepared from 12h by hydrogenation (>95% yield), the indolic ethyl ester 18h was treated with NaOH (1 equiv., 1,4-dioxane, H₂O) to give the carboxylic acid lactone 18ha (56% yield, 68% brsm), demonstrating thus the robustness of the strained but sterically hindered ring.²³ To access new scaffolds, 12a (80 : 20 dr) was deconstructed with LiAlH₄ into triol 23 (42% yield, 75 : 25 dr) and next converted into δ-lactone 24 (PhI(OAc)₂/TEMPO, 59% yield, relative configuration determined by NOESY experiments).²⁴

Conclusions

While they are among the most difficult cyclic scaffolds to prepare for kinetic and thermodynamic reasons,^8b nonalactones and decalactone with up to three non-vicinal stereocenters were prepared by ring expansion of prochiral alcohols (21 examples). Modulation of the configuration and substitution patterns of the prochiral material allowed the exploration of various steric and electronic scenarios (alkyl, aryl, carboxylate and carboxamide), picturing the perimeter and potential of the strategy with high values of dr and er (up to 99 : 1) owing to the 2-nitrophenyl function. Whether other electron-withdrawing groups may enable the enantioselective desymmetrization of the corresponding alcohol remains to be determined. However, this current limitation is counterbalanced by the enantio- and diastereoselective access to indoles, indolenines and derivatives thus provided, structural motifs notably encountered in many alkaloids (natural products or pharmaceuticals). From this investigation emerges a tool box in which readily available organocatalysts—QD or QN, 17c, 17d—effectively synthesized isomers of lactones 12 from cis-10 or trans-10. Bearing a quinolin-6-ol substituted with a methylenecyclohexyl ether, which increased the volume of the appendage, 17d offered a good balance between reactivity and stereoselectivity in most cases. Quinidine and quinine being commercially available, the two other derivatives are prepared in one step from cupreidine and are easily recovered after reaction, offsetting the catalytic load used to perform the energy-demanding ring expansion step. For that matter, the process of ring expansion of trans-10a into cis-12a required higher temperature to occur, without inducing lower enantioselectivity.

Futures studies will be focused on some features and implications of this complex and yet simple to implement process, such as the inversion of selectivity observed with fluorophenyl and carboxylate ester derivatives, or the origins of the stereoselectivity induced by the alkyl ether of the quinolin-6-ol appendage of the catalyst.

Data availability

All experimental and characterization data including HPLC traces and NMR spectra are available in the ESI.† Crystallographic data for compound 12e has been deposited in the Cambridge Crystallographic Data Centre under accession number CCDC 2162117.

Author contributions

A. A. and O. G. conducted the investigation and prepared the ESI.† C. F. conducted the DFT calculations. L. B. and E. P. contributed to the formal analysis of the results. J. M. reviewed the manuscript. M. D. P. conceptualized and supervised the research, wrote the manuscript.

Conflicts of interest

There are no conflicts to declare.