Catalytic Asymmetric Synthesis of trans-Configured β-Lactones: Cooperation of Lewis Acid and Ion Pair Catalysis



The development of the first trans-selective catalytic asymmetric [2+2] cyclocondensation of acyl halides with aliphatic aldehydes furnishing 3,4-disubstituted β-lactones is described. This work made use of a new strategy within the context of asymmetric dual activation catalysis: it combines the concepts of Lewis acid and organic aprotic ion pair catalysis in a single catalyst system. The methodology could also be applied to aromatic aldehydes and offers broad applicability (29 examples). The utility was further demonstrated by nucleophilic ring-opening reactions that provide highly enantiomerically enriched anti-aldol products.


Although β-lactones are fairly reactive four-membered heterocyclic ring systems, they were already described in 1883.1, 2 β-Lactones offer rich potential as synthetic intermediates3 and, most importantly, they can be viewed as masked and activated aldol products because they readily undergo nucleophilic ring-opening reactions as a result of their intrinsic ring strain, which is similar to that of epoxides (β-lactones: ca. 23 kcal mol1; epoxides: ca. 27 kcal mol).4 Various hard nucleophiles, such as metal alkoxides, amines, or C-nucleophiles can regioselectively cleave the acyl[BOND]oxygen bond providing the corresponding aldol adducts in a divergent manner.3, 5 Accordingly, the development of catalytic, asymmetric [2+2] cycloadditions of ketenes6 and aldehydes79 offers the possibility of replacing catalytic asymmetric aldol reactions, which often require the preformation and isolation of enolate equivalents such as silyl ketene acetals.10

Various bioactive natural and synthetic products possessing a β-lactone structural motif can act as specific enzyme inhibitors.11, 12 Among the derivatives of natural products, tetrahydrolipstatin (orlistat, Xenical) has been identified as an anti-obesity agent and is available as over-the-counter weight-loss medication in a number of countries. Of late, trans-configured β-lactones such as tetrahydrolipstatin have generated renewed interest due to the recent findings that they can serve as specific inhibitors of fatty acid synthase (FAS-TE), which is an approved drug target for cancer treatment.13 Owing to the significant activity of trans-configured β-lactones, a number of approaches have been reported for their synthesis.12b, 14 Unfortunately, the [2+2] cycloaddition approach using ketenes and aldehydes provides, in general, cis isomers as the major products. To the best of our knowledge, there is only one [2+2] cycloaddition available for the catalytic enantioselective formation of trans-configured β-lactones,9b which appears to be limited to the use of aromatic aldehydes, whereas many bioactive systems, such as tetrahydrolipstatin, carry an aliphatic chain at the 4-position of the 3,4-disubstituted 2-oxetanone.1518

The aim of the present work was to develop a trans-selective catalytic asymmetric [2+2] cyclocondensation of acyl halides 1 and aliphatic or aromatic aldehydes 2.19 A solution to this problem would provide a protocol for the challenging catalytic enantioselective anti-aldol addition reactions.20 The development of such a methodology was based upon the idea that if an enolate such as 5, and not a ketene such as 4, represented the reactive intermediate, the trans-configured product would be expected to be formed in preference by activation of the aldehyde with a mono-coordinating Lewis acid (Scheme 1). The choice of a Lewis acid with only one available coordination site was expected to be essential to avoid a cyclic transition state leading to the cis-configured product through a Zimmerman–Traxler-type transition state. The catalyst could thus combine the cooperative action of a Lewis acid and a nucleophilic moiety to generate and direct the enolate to the aldehyde through the open transition state 6. In this state, a staggered conformation around the generated C[BOND]C bond9b, 15 could be adopted in which the nucleophilic function should be oriented gauche to the aldehyde H for steric reasons. The initial aldol adduct could then cyclize to form the heterocyclic product.

Scheme 1.

Concept for the asymmetric formation of trans-configured β-lactones 3 (L.A.=Lewis acid with only one available coordination site).

Results and Discussion

N,N′-Bis(salicylidene)ethylenediamine (salen) ligands were chosen because of their modular nature and their ready availability.21 Kozlowski and co-workers first described catalyst systems with the general structure 7 that incorporate basic amino moieties into the structurally well-defined salen complex.22

Thumbnail image of

To evaluate the role of the amino functionalities, ligands lacking the nucleophilic or basic groups were studied as control systems. With the standard salen complex 8 a (R=tBu), the desired product was not formed in the simple model reaction of acetyl bromide (1 A) and dihydrocinnamaldehyde (2 a) in CH2Cl2 at −20 °C (Table 1, entry 1). With R=H or iBu, good conversion into the nearly racemic cycloaddition product was found (Table 1, entries 2 and 3). In contrast, with complex 8 d, which carries dimethylaminomethylene donors, a promising enantiomeric excess of 50 % was attained.

Table 1. Investigation of the influence of nonbasic and Lewis/Brönsted basic substituents R on salen–Al complexes 8 for the formation of β-lactone 3 Aa. inline image
Entry8RT [°C]t [h]Yield [%][a]ee [%][b] (isomer)
  1. [a] Yield determined by 1H NMR spectroscopy using acetophenone as internal standard. [b] Enantiomeric excess (ee) determined by HPLC. [c] Conversion of aldehyde determined by 1H NMR spectroscopic analysis.

18 atBu−20290
28 bH−204573[c]2 (R)
38 ciBu−50→−204895[c]6 (R)
48 dCH2NMe2−50→−20486550 (R)
58 e
original image
−50→−20484944 (R)
68 f
original image
−50→−20485444 (R)
78 g
original image
−50→−20485140 (R)
88 h
original image
−50→−20484732 (R)
98 i
original image
−50→−20486910 (R)
108 j
original image
−50→−20486210 (R)
118 kCH2SMe−5037946 (R)

Slightly lower enantioselectivities were observed with piperidine (44 % ee; Table 1, entry 5), morpholine (44 % ee; Table 1, entry 6), or N-phenyl piperazine donors (40 % ee; Table 1, entry 7). Catalysts containing α-branched amino functions were less selective (Table 1, entries 8–10). These results show that the efficiency of the basic group decreases with increasing steric hindrance. Catalyst 8 k (Table 1, entry 11), carrying a nucleophilic thioether moiety, was found to be more reactive but less enantioselective than 8 d.

For the reactivity of catalysts 8 dj, the ratio and addition order of acetyl bromide and the auxiliary base iPr2NEt is crucial. For the investigations summarized in Table 1, an excess of acetyl bromide was used relative to iPr2NEt (3 equiv of acetyl bromide, 2.5 equiv iPr2NEt), the latter being added as the last component to the reaction mixture. With an increased amount of base, less product formation was observed. An equal amount of base and acetyl bromide significantly impeded the reaction, whereas with an inverse order of addition, that is, first the base, then acetyl bromide, no reaction at all took place. Furthermore, when an excess of iPr2NEt was used, the catalyst was completely inhibited. Under these conditions the catalyst is apparently deactivated by coordination of the base. In contrast, under the conditions detailed in Table 1, the amino groups of 8 dj might be partly protonated, which could disturb the formation of a reactive enolate. On the other hand, the amino functions might be acylated as in the case of 8 l, but there would not be enough base remaining for efficient enolate formation.

Thumbnail image of

Nevertheless, with propionyl bromide (1 B) as methylketene source, trans-configured β-lactone 3 Ba was preferentially formed at −20 °C (trans/cis=71:29) with a promising ee value of 66 %, albeit in a low yield of 28 % (Scheme 2).

Scheme 2.

Formation of α,β-disubstituted β-lactone 3 Ba catalyzed by 8 d.

Different catalyst species possessing either acylated or protonated amino functionalities were expected to be present that might exhibit different selectivity profiles. The influence of structurally well-defined, positively charged aprotic ammonium moieties was subsequently investigated because the development of a trans-selective catalytic asymmetric [2+2] cyclocondensation might also be achieved if acyl halide enolates 9 represent the reactive intermediates (Scheme 3). A possible approach involved the use of a catalyst that included a Lewis acid moiety working in concert with a positively charged aprotic organic functionality (Q+) within the ligand system that could direct the enolate through formation of an ion pair. The enolate could then undergo an enantioselective aldol addition to the aldehyde, which is activated by the mono-coordinating Lewis acid; a key aspect being that the acyl halide enolate would have to react faster than the ketene intermediate. Because this preference is usually not observed,23 the unstable enolate24, 25 would have to be generated within the catalyst sphere by attack of the halide counterion on the ketene intermediate, and be stabilized by ion pair formation with the organic cationic functionality Q+.26 Compared with conventional bifunctional Lewis acid/Lewis base catalysts,27 this cooperative catalysis concept would have the principal advantage that the cationic aprotic organic functionality does not deactivate the Lewis acid by a self-quenching process.

Scheme 3.

Modified concept for the asymmetric formation of trans-configured β-lactones 3 (L.A.=Lewis acid with only one available coordination side, Q+=cationic aprotic organic functionality).

To establish proof of principle, a range of enantiopure Al–salen complexes 8, which differed in the substituents R at the 6-position of the phenol rings, were applied to the transformation of propionyl bromide (1 B) with dihydrocinnamaldehyde (2 a) as a model reaction (Table 2). To compare the effect of the substituents R, the reactions presented in Table 2 were all performed at −20 °C and stopped after 2 h, unless indicated otherwise. The most simple catalyst system 8 b (X=H) produced β-lactone 3 Ba with moderate cis-selectivity in almost racemic form (Table 2, entry 1). For the more bulky salen 8 c, carrying isobutyl substituents, both reduced reactivity and cis-selectivity were noted, whereas tert-butyl substituents in 8 a again completely prevented all product formation, presumably for steric reasons.

Table 2. Optimization of a cooperative organic ion pair/Lewis acid catalyst.[a]inline image
Entry8RT [°C]Yield [%][b]ee [%][c]trans/cis[d]
  1. [a] The catalyst was prepared in situ if not mentioned otherwise. [b] Yield determined by 1H NMR analysis using acetophenone as internal standard. [c] Enantiomeric excess of the trans-isomer determined by chiral column HPLC. [d] Ratio determined by 1H NMR spectroscopic analysis. [e] 21 % ee for the cis isomer. [f] 23 % ee for the cis isomer. [g] Reaction time 5 h. [h] Use of preformed catalyst. [i] Reaction time 24 h.

18 bH−204212[e]21:79
28 ciBu−20280[f]25:75
38 atBu−200
48 m
original image
58 n
original image
68 o
original image
78 p
original image
88 q
original image
98 p
original image
10[h]8 p
original image

In contrast, catalysts 8 mq possessing positively charged substituents at the phenol 6-position led to enhanced reactivity. Moreover, the enantioselectivity was significantly improved. Most strikingly, the cationic ammonium or heterocyclic functionalities provided high trans-selectivities (Table 2, entries 4–8). As a general trend, the enantioselectivity was slightly reduced with increased steric bulk of the cationic moieties. The planar pyridinium system 8 p was the most selective catalyst, which might be explained by a more efficient contact ion pair formation as a result of the planar sp2-nitrogen atom compared with the tetrahedral ammonium functionalities. Imidazolium derivative 8 q, in which the positive charge is delocalized over two nitrogen atoms, led to lower enantio- and diastereoselectivity.

To attain a higher level of enantioselectivity, the reaction temperature had to be further decreased and the catalyst was isolated prior to use; in previous experiments it was prepared in situ. Whereas the reaction catalyzed by the trimethylammonium system 8 m was extremely slow at −50 °C as a result of poor catalyst solubility, the pyridinium catalyst 8 p was significantly more reactive (Table 2, entry 9) and rendered high enantio- and trans-selectivity as well as providing a good yield at −70 °C (Table 2, entry 10). Compound 8 p was readily prepared in four steps from 2-hydroxy-5-tert-butylbenzaldehyde (11) with an overall yield of 87 % (Scheme 4).

Scheme 4.

Synthesis of catalyst 8 p in four steps from aldehyde 11.

The β-lactone formation triggered by 8 p generally provided high trans-selectivities with all the aliphatic aldehydes tested (Table 3). The enantioselectivities did not significantly depend on the aldehyde, and almost identical results were obtained with substrates possessing long aliphatic side chains (Table 3, entries 4–6), with or without a C[DOUBLE BOND]C double bond, β-branched isovaleraldehyde (Table 3, entry 10) or sterically undemanding aldehydes such as propanal, butanal, or pentanal (Table 3, entries 7–9). In particular, these last results are remarkable because high enantioselectivities have previously never been reported for these sterically undemanding aliphatic aldehydes in alternative catalytic asymmetric cycloadditions with acyl halides. Both yields and enantioselectivities were higher in the case of valeroyl bromide (1 C; Table 3, entries 13–18) compared with propionyl bromide (1 B). Again, the enantioselectivity was found to be almost independent of the aldehyde used.

Table 3. Application of the dual activation catalyst 8 p to the reaction of aliphatic aldehydes. inline image
Entry3R1R28 p [mol %]Yield [%][a]ee [%][b]trans/cis[c]
  1. [a] Yield of isolated product. [b] Enantiomeric excess of the trans isomer determined by chiral column HPLC or GC (see the Supporting Information). [c] Ratio determined by 1H NMR analysis. [d] Determined by 1H NMR spectroscopic analysis. [e] Reaction performed at −40 °C.

13 BaMe(CH2)2Ph1082[d]8897:03
23 BaMe(CH2)2Ph2.5738597:03
33 AaH(CH2)2Ph107670
43 BbMenHept10778796:04
53 BcMe(CH2)3CH[DOUBLE BOND]CH21074[d]8896:04
63 BdMe(CH2)8CH[DOUBLE BOND]CH210628794:06
73 BeMeEt1076[d]8795:05
83 BfMenPr10679397:03
93 BgMenBu10648997:03
103 BhMeiBu1076[d]8794:06
113 BiMecHex1078[e]8087:13
123 AiHcHex108288
133 CanPr(CH2)2Ph10919498:02
143 CcnPr(CH2)3CH[DOUBLE BOND]CH210969598:02
153 CenPrEt10639497:03
163 CfnPrnPr10939598:02
173 CgnPrnBu10929396:04
183 ChnPriBu10769496:04

In contrast to the aldehydes described above, the α-branched cyclohexylcarboxaldehyde gave very little conversion at −70 °C. Therefore, the temperature had to be increased to −40 °C to provide the product with useful selectivity (80 % ee, trans/cis=87:13) and yield (78 %; Table 3, entry 11). With acetyl bromide (1 A) the reaction with cyclohexylcarboxaldehyde could again be performed at −70 °C, providing an enantiomeric excess of 88 % (Table 3, entry 12), whereas with acetyl bromide/dihydrocinnamaldehyde the ee dropped to 70 % (Table 3, entry 3). The method is apparently less general for the formation of 4-monosubstituted β-lactones. Although 10 mol % of catalyst 8 p was commonly used for the experiments in Table 3, a similar result was also achieved with just 2.5 mol % (entry 2).

trans-Configured β-aryl-substituted β-lactones are considerably less stable than their β-alkyl-substituted counterparts due to a marked sensitivity towards elimination, which is accelerated by electron donors on the aromatic ring.28 As a consequence, alternative workup conditions were necessary in the present study for product isolation. The transformation of propionyl bromide (1 B) with benzaldehyde (2 j) was investigated as a model reaction. All attempts at aqueous workup (acidic, basic, or neutral) led to either complete or partial decomposition of the product. However, by catalyst quenching with an excess of iPr2NEt at low temperature and filtration through a plug of deactivated silica gel (followed by column chromatography) it was possible to isolate the targeted β-lactone 3 Bj (Table 4, entries 1 and 2).

Table 4. Application of 8 p to the reaction of aromatic aldehydes. inline image
Entry3R1R2X 8 p [mol %]Yield [%][a]ee [%][b]trans/cis[c]
  1. [a] Yield determined by 1H NMR spectroscopic analysis using acetophenone as internal standard. [b] Enantiomeric excess of the trans-isomer determined by chiral column HPLC. [c] Ratio determined by 1H NMR spectroscopic analysis.

13 BjMeH2.5709396:04
23 BjMeH10809194:06
33 BkMeCF32.5847597:03
43 BlMeCl2.5757796:04
53 BmMeCN2.5877098:02
63 BnMeNO22.5907897:03
73 BoMeBr2.5817196:04
83 CjnPrH10469497:03
93 CknPrCF310938096:04
103 ClnPrCl10628395:05
113 CmnPrCN10908198:02
123 CnnPrNO210888097:03
133 ConPrBr10708898:02

For reactions of propionyl bromide with different aromatic aldehydes, 2.5 mol % of 8 p was usually sufficient (Table 4). With benzaldehyde, an ee value of 93 % was attained (Table 4, entry 1). In the case of electron-poor aryl groups, the products were formed with good yield but the enantioselectivity was decreased to 70–78 % ee although high trans selectivity was maintained (Table 4, entries 3–7). On the other hand, no product could be isolated with 4-methylbenzaldehyde or 4-methoxybenzaldehyde; in the first case due to decomposition of the product during workup and, in the second case, maybe because of low substrate reactivity.

Reactions between a range of aromatic aldehydes and n-valeroyl bromide required a catalyst loading of 10 mol % to achieve useful yields (Table 4, entries 8–13). In this case, again, higher enantioselectivities (80–94 % ee) were generally attained than with propionyl bromide.

The proposed mode of operation is depicted in Scheme 5. The aldehyde is assumed to bind with its sterically more accessible lone pair to the only free Al coordination site, to form an octahedral complex. Because the unstable, highly reactive acyl bromide enolate2325 is expected to be present in only minute concentrations and could therefore not be spectroscopically detected in our studies despite intensive attempts (NMR and IR spectroscopy, mass spectrometry), it is likely that the short-living reactive intermediate is generated directly in the catalyst sphere by reaction of the catalyst’s bromide counterion with the ketene.29 This attack is expected to occur selectively trans to the residue R1 to minimize repulsive interactions, thus selectively forming the E-configured enolate. This reactive species should nucleophilically attack the aldehyde through an open transition state that adopts a staggered conformation with the enolate C1 atom gauche to the aldehyde H atom. Our model is in agreement with the products’ configuration.30

Scheme 5.

Working model for the cooperative action of the ion pair and Lewis acid catalysis.

The synthetic utility of the trans-configured products was finally showcased by conducting nucleophilic ring-opening reactions with a primary amine and with Weinreb amine (Scheme 6), giving access to the anti-aldol products in good yield and with high enantiomeric excess.

Scheme 6.

Formation of enantioenriched anti-aldol products by nucleophilic ring-opening reactions.


This work has explored a new strategy within the context of asymmetric cooperative catalysis.31 It combines the concepts of Lewis acid and aprotic organic ion pair catalysis in a single catalyst system (Figure 1).32

Figure 1.

The concept of ion pair directed Lewis acid catalysis.

This mode of activation has allowed us to develop the first trans-selective catalytic asymmetric [2+2] cyclocondensation of acyl halides with aliphatic aldehydes, furnishing 3,4-disubstituted β-lactones. Despite the high sensitivity of the 4-aryl-substituted β-lactones towards elimination, the methodology could also be applied to aromatic aldehydes by making use of a different workup procedure. In general, this new methodology offers broad applicability and represents an alternative to asymmetric anti-aldol additions.

Experimental Section

General procedure for the formation of catalysts 8a–k and 8m–q in situ: A solution of Me3Al in toluene (2 m, 0.10 mL, 0.21 mmol, 1 equiv) was added to a solution of the corresponding salen ligand (0.21 mmol, 1 equiv) in CH2Cl2 (3.5 mL). The mixture was stirred for 3 h at ambient temperature.

General procedure for the formation oftrans-configured β-lactones 3 with aliphatic substituents at the β-position: Aldehyde 2 (0.75 mmol, 1 equiv), acyl bromide 1 (4.5 mmol, 6 equiv), and diisopropylethylamine (1.875 mmol, 2.5 equiv) were successively added at −70 °C to a mixture of complex 8 p (0.075 mmol, 0.1 equiv) in CH2Cl2 (3 mL). The resulting heterogeneous mixture was stirred at −70 °C for 24 h then poured into aqueous 1 m HCl (30 mL) and extracted with CH2Cl2 (2×20 mL). The combined organic phase was dried over MgSO4 and filtered through a short plug of silica gel. CH2Cl2 was subsequently removed in vacuo and the crude product mixture was purified by flash chromatography if not indicated otherwise.

General procedure for the formation oftrans-configured β-lactones 3 with an aryl substituent in the β-position: Aldehyde 2 (0.75 mmol, 1 equiv), acyl bromide 1 (4.5 mmol, 6 equiv), and diisopropylethylamine (1.875 mmol, 2.5 equiv) were successively added at −70 °C to a mixture of complex 8 p (0.0188 mmol, 0.025 equiv) in CH2Cl2 (3 mL). The resulting heterogeneous mixture was stirred at −70 °C for 24 h, then the reaction mixture was quenched with diisopropylethylamine (4 mL) and filtered through a short plug of Et3N-deactivated silica gel. CH2Cl2 was subsequently removed in vacuo and the crude product mixture of the trans-configured β-aryl-substituted β-lactones 3 was purified by rapid flash chromatography using Et3N-deactivated silica gel.


This work was financially supported by a TH research grant TH-30/04-2, the Swiss National Science Foundation (SNF; postdoctoral fellowship to J.C.) and F. Hoffmann-La Roche. We thank Reuter Chemische Apparatebau KG (RCA) for the generous donation of enantiomerically pure 1,2-diaminocyclohexane.