Lipase enantioselectivity models

A strategy was developed that not only works well in the present situation involving the lipase-catalyzed model reaction 9 — 10 [9,10,42], but which may prove to be useful in other cases regarding the evolution of enantioselective enzymes as well. Accord-... 

Fig. 11.14. Further improvements in enantioselectivity of the lipase-catalyzed model reaction ofester 9 (S = serine F = phenylalanine C = glycine L= leucine V = valine). Details are outlined in [42].

There are several models for lipase enantioselectivity that can be used to predict which enantiomer of a chiral substrate will react fastest. Most models are substrate-based. This means that the lipase structure was not taken into account when the prediction model was set up. Instead, only common features of the fast-reacting enantiomer were considered. These models have great value when synthetic routes are planned but are less valuable in the understanding of the enantioselective process and design of more selective systems. 

If the slopes of the absorption/time curves differ considerably, a positive hit is indicated (i.e., an enantioselective lipase-variant has been identified) (16). Figure 5 shows two typical experimental plots, illustrating the presence of a non-selective lipase (top) and a hit (bottom) (16). As a consequence of the crudeness of the test, quantitative evaluation is not possible. Therefore, the hits need to be investigated separately in laboratory-scale reactions and evaluated quantitatively by conventional chiral GC. About 800 plots of this kind can easily be recorded per day. A total of 40 000 lipase-variants were generated by epPCR, saturation mutagenesis, cassette mutagenesis, and DNA shuffling and screened in the model reaction. 

Genetic engineering. The X-ray structures are known for many hydrolases, allowing for modeling of the substrate in the active site as well as structurally based, random or rational protein mutation to magnify or invert enantioselectivity. An example of the latter is provided by the rational design of a mutant of Candida antarctica lipase (CALB), which, instead of the wild-type R-selectivity, displayed... 

Although quite reliable empirical rules exist for the enantioselectivity of hydrolases for secondary alcohols (see Section 4.2.1.2), such rules are not as developed for primary alcohols, partly because many hydrolases often show low enantioselectivity. With some exceptions, lipases from Pseudomonas sp. and porcine pancreas lipase (PPL) often display sufficient selectivity for practical use. The model described in Figure 4.3 has been developed for Pseudomonas cepacia lipase (reclassified as Burkholderia cepacia), and, provided that no oxygen is attached to the stereogenic center, it works well for this lipase in many cases [41]. However, as soon as primary alcohols are resolved by enzyme catalysis, independent proof of configuration for a previously unknown product is recommended. 

As mentioned earlier (Section 4.2.1.1), empirical rules for the enantioselectivity of hydrolases have been developed. It is important to keep in mind that these rules do not work for all substrates. Most rules are based on pockets, which indicate how the steric bulk of the substituents in the substrate fit into the environment of the active site. Thus, such rules have been suggested for pig liver esterase(PLE) [66], the protease subtilisin [66-68], and certain lipases [69-71]. For secondary alcohols, most lipases follow the simple rule of Kazlauskas, which was developed for Pseudomonas cepacia, and which is depicted in Figure 4.4 [72]. This model implies that the fast-reacting enantiomers binds to the active site as described in Figure 4.4, whereas the slowly reacting one is not able to achieve a comfortable fit, because it will require the large substituent L to fit into the smaller pocket. In contrast to lipases, subtilisin displays opposite enantioselectivity toward secondary alcohols [68]. 

Conventional gas chromatography (GC) based on the use of chiral stationary phases can handle only a few dozen ee determinations per day. In some instances GC can be modified so that, in optimal situations, about 700 exact ee and E determinations are possible per day [29]. Such meclium-throughputmay suffice in certain applications. The example concerns the lipase-catalyzed kinetic resolution of the chiral alcohol (R)- and (S)-18 with formation of the acylated forms (R)- and (S )-19. Thousands of mutants of the lipase from Pseudomonas aeruginosa were created by error-prone PCR for use as catalysts in the model reaction and were then screened for enantioselectivity [29]. 

The kinetic results for the lipase-catalysed enantioselective hydrolysis of the esters (236)-(240) can be interpreted in terms of frontier orbital localization.213 The porcine pancreatic lipase (PPL)-mediated optical resolution of 18 racemic esters can be explained by a mechanistic model involving a W-shaped active conformation of the substrate lying in a diastereo-discriminating plane.214... 

Studies of the ability of the lipase B from Candida antarctica (CAL-B) to catalyse the enantioselective aminolysis of esters by cis- and firms-2-phenylcycloalkanamines (54 n = 1, 3, 4) have been followed up by molecular modelling approaches in order to probe the lipase-catalysed aminolysis mechanism. CAL-B possesses a typical serine-dependent triad, so it was possible, with access to an X-ray crystal structure of CAL-B, to model a series of phosphonamidates (55 n = 1, 3, 4) as analogues of the tetrahedral intermediate (TI) resulting from attack of the amine on the carbonyl of the acyl-enzyme. The results suggested as the most plausible intermediate for the CAL-B-catalysed aminolysis a zwitterionic TI resulting from the direct His-assisted attack of the amine on to a C=0 group of the acyl-enzyme.80... 

Enzymatic enantioselectivity in organic solvents can be markedly enhanced by temporarily enlarging the substrate via salt formation (Ke, 1999). In addition to its size, the stereochemistry of the counterion can greatly affect the enantioselectivity enhancement (Shin, 2000). In the Pseudomonas cepacia lipase-catalyzed propanolysis of phenylalanine methyl ester (Phe-OMe) in anhydrous acetonitrile, the E value of 5.8 doubled when the Phe-OMe/(S)-mandelate salt was used as a substrate instead of the free ester, and rose sevenfold with (K)-maridelic acid as a Briansted-Lewis acid. Similar effects were observed with other bulky, but not with petite, counterions. The greatest enhancement was afforded by 10-camphorsulfonic acid the E value increased to 18 2 for a salt with its K-enanliomer and jumped to 53 4 for the S. These effects, also observed in other solvents, were explained by means of structure-based molecular modeling of the lipase-bound transition states of the substrate enantiomers and their diastereomeric salts. 

Based on the very different behaviors of lipases A (CAL-A) and B (CAL-B) from Candida antarctica towards polyfunctional compounds in non-aqueous media, Liljeblad et al82 reported a novel lipase-catalyzed method for the resolution of A -heterocyclic amino esters using methyl pipecolinate 69 as a model compound. For this purpose, the chemo-and enantioselective alcoholysis and transesterification reaction of 69 in the presence of CAL-B and the A-acylations using CAL-A were studied, (cf. fig. 37 and 38). 

Hundreds of impressive examples of enantioselective lipase-catalysed reactions are known, including industrial processes as in the case of the BASF method of chiral amine production (Collins et al. 1997 Breuer et al. 2004 Schmid and Verger 1998). However, the classical problem of substrate acceptance or lack of enantioselectivity (or both) persists. We were able to meet this challenge in model studies regarding the hydrolytic kinetic resolution of the ester rac-1 with formation of carboxylic acid 2, catalysed by the lipase from Pseudomonas aeruginosa. The wild-type (WT) lipase is only slightly (S )-selective, the selectivity factor amounting to a mere E = 1.1 (Scheme 1). 

Lipases are the most frequently used enzymes in organic chemistry, catalyzing the hydrolysis of carboxylic acid esters or the reverse reaction in organic solvents [3,5,34,70]. The first example of directed evolution of an enantioselective enzyme according to the principle outlined in Fig. 11.2 concerns the hydrolytic kinetic resolution of the chiral ester 9 catalyzed by the bacterial lipase from Pseudomonas aeruginosa [8], This enzyme is composed of 285 amino acids [32]. It is an active catalyst for the model reaction, but enantioselectivity is poor (ee 5 % in favor of the (S)-acid 10 at about 50 % conversion) (Fig. 11.10) [71]. The selectivity factor E, which reflects the relative rate of the reactions of the (S)- and (R)-substrates, is only 1.1. 

Although these lipase variants, which have only two amino acid substitutions, show notable degrees of enantioselectivity in the model reaction, they are not better than the best variants of previous experiments. However, the efforts are not in vain because they pave the way to a novel strategy for additional improvements in enantioselectivity as discussed below. 

A second example of the use of directed molecular evolution for natural product synthesis is the use of lipases by Reetz and colleagues. This work is based on the kinetic hydrolytic resolution of racemic mixtures, in which one enantiomer is preferentially hydrolyzed and the chiral product is thus enriched. Utilizing both random mutagenesis and directed techniques such as CAST,64 they have improved the stereoselectivity of a lipase from Pseudomonas aeruginosa (PAL) on a number of occasions with different substrates. One of the first examples utilized the model substrate 2-methyldecanoic acid /xnitrophenyl ester, for which the wild-type enzyme has an enantioselectivity of E= 1.1. As a consequence of five mutations accumulated through random mutagenesis, followed by saturation mutagenesis, the enantioselectivity was increased to 25.8.123 More... 

The enzyme in its mature form is composed of 285 amino acids [77]. Upon applying the algorithm described in Sect. 1, it can be calculated that the number of mutant lipases in which one amino acid per enzyme molecule is substituted by one of the remaining 19 amino acids is 5415 [ 15 ]. It was of great interest to see whether such small libraries (or even smaller ones) would contain lipases with significantly improved enantioselectivity in the model reaction [15,18,43,76]. 

It was also interesting to observe that the mutant lipase which shows an ee of 81% in the model reaction of the p-nitrophenyl ester displays similar performance if the corresponding ethyl ester is used (80 % ee) [76]. Nevertheless, it was not clear at this stage whether the degree of enantioselectivity would continue to climb in further mutagenesis experiments. 

Ke, T., Tidor, B., and Klibanov, A. M., Molecular-modeling calculations for enzymatic enantioselectivity taking hydration into account, Biotechnol. Bioeng., 57,741-745,1998. Haeffner, F., Norin, T., and Hull, K., Molecular modeling of the enantioselectivity in lipase-catalyzed transesterification reactions, Biophys. J., 74, 1251-1262, 1998. Bernstein, F. C., Koetzle, T. R, WiUiams, G. J. B., Meyer, E. F. J., Brice, M. D., Rodgers, J. R., Kennard, O., Shimanouchi, T., and Tasumi, M., The protein data bank a computer-based archival file for macromolecular structures, J. Mol. Biol., 112, 535-542, 1977. Parida, S. and Dordick, J. S., Tailoring lipase specificity by solvent and substrate chemistries, J. Org. Chem., 58, 3238-3244, 1993. 

A very straightforward approach in route C (Fig. 2) would have been the direct enzyme-catalyzed peptide formation (cf. Chen et al. [18]) by enantioselective aminolysis of diester 9 with (S)-tert-leucine methylamide 13 or even racemic 13. This would combine three synthetic objectives the resolution of (rac)-9, the resolution of (roc)-13 and the coupling step. In orientating experiments monoester 10 was tested as a model substrate. It was contacted with an equal amount of (S)-amine 13 in the presence and absence of an organic solvent. Solid or liquid subtilisin Carlsberg preparations (Alcalase 2.0 T or Alcalase 2.5 L, respectively) were used as the catalyst. Only with the liquid enzyme preparation was the formation of minor amounts of one of two possible diastereoisomeric peptides observed [19], whereas most of the ester was hydrolyzed to the acid. Likewise, a few selected lipases also provided negative results. 

---