Transaminase mimics

The class of enzymes called transaminases use the coenzyme pyridoxamine phosphate 23 to convert a substrate a -keto acid to an o -amino acid, while the pyridoxamine itself is converted to pyridoxal phosphate 24 (Fig. 1.12). Later the pyridoxal phosphate coenzyme is converted back to the pyridoxamine form by transaminations in the reverse direction from a sacrificial amino acid. We set out to mimic this interesting process by attaching pyridoxamine to a cyclodextrin, so there would be a preference for transaminating keto adds that could bind to the cyclo dextrin in water. 

In our first example (Fig. 1.13), the first in which a coenzyme was linked to a cyclodextrin, we synthesized compound 25, with a pyridoxamine covalently linked to a C-6 of /3-cyclodextrin. We saw transaminations of pyruvic acid, of phenylpyruvic acid, and of indolepyruvic acid to form alanine, phenylalanine, and tryptophan, respectively, and with high selectivity for the hydrophobic phenyl and indole derivatives relative to simple pyruvic acid. Relative to its reaction with simple pyridoxamine in solution, without an attached or unattached cyclodextrin, the indolepyruvate reacted 50 times faster with compound 25. Also, we saw a 5 1 preference for the formation of the L-phenylalanine relative to the d enantiomer in transamination by compoimd 25.  

In compound 25 we had attached the pyridoxamine to the primary C-6 position of the cyclodextrin, but in an earlier study we had seen that there were sometimes, but not always, advantages to attaching catalytic groups to the secondary face of a cyclodextrin. Thus we attached pyridoxamine to the secondary face of jS-cyclodextrin, and saw again a preference for the transamination of indolepyruvic acid and of phenylpyruvic add, but the preferences were only approximately half as large as those with compound 25. Thus here the original attachment of the cofactor to the primary face of the cyclodextrin was actually the best. 

Another comparison substituted the cyclodextrin-binding group by a synthetic macrocycle that also strongly binds hydrophobic substrates in water solution. We synthesized 

We saw that the endo pair of isomers, with the pyridoxamine directed over the cyclodextrin cavity (e.g. 27), was particularly effective in transaminating a substrate, p-t-butylphenylpyruvic acid, that is oriented along the cyclodextrin axis. By contrast, the exo compounds (e.g. 28) were more effective with w-t-butylphenylpyruvic acid, which holds 

In our first study we attached a pyridoxamine unit to a primary carbon of jS-cyclodextrin (structure 12). We saw that pyridoxamine alone is able to transaminate pyruvic acid to form alanine, phenylpyruvic acid to form phenylalanine, and indolepyruvic acid to form tryptophan, all with equal reactivity by competition experiments. However, when the cyclodextrin was attached to the pyridoxamine there was a 200-fold preference for the indolepyruvate over pyruvate in one-to-one competition, forming greater than 98% of tryptophan, and in the competition with phenylpynivate and pyruvate the phenyManine was formed in greater than 98% as well. Thus the ability of the substrates to bind into the cyclodextrin cavity led to striking selectivities. In addition there was some chiral induction in these processes, since )3-cyclodextrin is itself chiral, but the magnitudes of the induction were quite modest. 

We devised a method to perform the selective tosylation of the C-2 hydroxyl group in )3-cyclodextrin, and used it to attach the pyridoxamine to this secondary side of the cyclodextrin. Again we saw some preference for transamination of the aromatic ketoac-ids, but by less than we had observed with the pyridoxamine attached to C-6. The tryptophan synthesis was only 25 times as fast as that for alanine, while the phenylalanine formation was 18 times as fast as alanine in competitive reactions. Modest enantioselec-tivities were observed as well with this C-2 linked pyridoxamine, and they differed in detail from those produced with the C-6 isomer that we had reported earlier. 

We also examined an interesting modified version in which the j8-cyclodextrin had a pyridoxamine attached to the primary C-6 carbon but the other six hydroxymethyl groups on the primary side were deoxygenated to become methyl groups. This leads to quite a different hydrophobic cavity, but it showed properties similar to those of the original compound that still had its primary hydroxyls unmodified. 

In a full paper we described this work in detail, including some work on pyridoxamine derivatives that did not involve cyclodextrins and other work in which a synthetic binding group was used instead of the cyclodextrin.  

We also set about to see whether we could get good chiral induction in the product amino acids, not by the simple accident of the chirality of the cyclodextrin but by basic groups that could direct the proton transfer involved in transamination so as to give a preference for one enantiomer of the product amino acid. We described some of this work, and in it also referred to work reported by Tabushi in which the same general principle was applied. In our work only relatively modest selectivities were seen the largest optical ratio (L/D) in the product was only 6.8. 

---

Liu L, Rozenman M, Breslow R. Hydrophobic effects on rates and substrate selectivities in polymeric transaminase mimics. J. Am. Chem. Soc. 2002 124 12660-12661. 

Liu, L, M Rozenman and R Breslow (2002). Hydrophobic effects on rates and substrates selectivity in polymeric transaminase mimics. Journal of the American Chemical Society,... 

Figure 37 (a) Cyclodextiin-based transaminase mimic (b) Polymeric PEI transaminase mimic. 

Figure 38 (a) Pyridoxalamine with unpolar side chains (b) Catalytic cycle of noncovalent transaminase mimic. 

In the described examples, the pyridoxamine was covalently attached to the polymer while in most real transaminase enzymes the pyridoxamine coenzyme forms a noncovalent active holoenzyme with the protein (apoenzyme). A new artificial transaminase mimic was developed, in which the pyridoxamine binds noncovalently and reversibly to the polymer. The pyridoxamine attached, for example, to a steroid side chain 99 or 100, together with modified PEI 101 (molecular weight of 60000 and 8.7% dodecyl chains) forms the artificial holoenzyme (Figure 38a). The transamination of pyruvic acid was accelerated 28000-fold with 99 + 101 compared to 10 000 with the covalent pyridoxamine-polymer 98 enzyme mimic. This was due to the fact that the noncovalent system 99 - -101 is more dynamic and therefore can adopt a more suitable geometry for the reaction. The artificial transaminase shows effective rate enhancements in converting the ketoacid into the amino acid, but also the pyridoxamine is converted to pyridoxal. The conversion to pyridoxamine is a necessary step in the catalytic cycle to achieve high turnovers however, this was still not possible with the noncovalent model system. It was observed that the reverse process is very slow and actually in all artificial models so far thermodynamically unfavorable. However, it was possible to use sacrificial amino acids at elevated temperatures (60 °C) that were decarboxy-lated to recycle the pyridoxal 102 to pyridoxamine 100 with modest turnover numbers of 81 (Figure 38b). " ... 

Scheme 22 Preparation of a-amino acids using the MIP transaminase mimic developed by Nicholls et al ...

A great deal of early research in the field established the foundations that now govern the concept of enzyme mimicry using MIPs, including the importance of structure-function relationships in determining binding and catalytic activity. Nicholls etal. prepared and evaluated an MIP transaminase mimic for the reaction of phenylpyruvic acid and pyridoxamine (Scheme 22). ... 

One of the earliest published attempts to create antibodies with catalytic activity had as its goal the generation of a transaminase. Raso and Stollar prepared V-(5-phosphopyridoxyl)-3 -amino-L-tyrosine 154 as a mimic of the Schiff s base intermediate that is formed during the pyridoxal-dependent transamination of tyrosine and showed that it was a site-directed inhibitor of the enzymes tyrosine transaminase and tyrosine decarboxylase.132 Partially purified polyclonal antibodies, elicited against y-globulin conjugates of the hapten, recognized both the... 

A particular example from our laboratory is observed in compound 5, in which a basic amino group held rigidly on a mimic of the coenzyme pyridoxamine phosphate is able to convert ketoacids to amino acids with high enantioselectivity (15) (Fig. 3). This reaction is modeled closely to the way in which transaminase enzymes achieve the same goal. 

Figure 3 A mimic of the enzyme transaminase achieves high stereoselectivity in the product amino acid because of geometric control by the attached basic chain.

Generally, their studies involved hydrolytic and fragmentation reactions. We took up the study of such systems as mimics for transaminase enzymes and showed with the polyaziridines that we achieve very large accelerations of the conversion of ke-toacids to amino acids (27). This result reflected several ways in which these polyamines mimic enzymes. 

To solve this problem, we used a mimic of a different enzyme, diaUcylglycine decarboxylase (30). In this enzyme, pyridoxal phosphate reacts with an alpha-disubstituted glycine to perform an irreversible decarboxylation (Fig. 6) while converting the pyridoxal species to a pyridoxamine. We imitated this with our model transaminations using pyridoxal species that carry hydrophobic chains, and we were able to achieve as many as 100 catalytic turnovers. Thus, we could imitate one enzyme—the ordinary transaminases—by also imitating another enzyme that solved the turnover problem. 

Figure 6 The mechanism used in the oxidative decarboxylation of alpha disubstituted glycines by an enzyme, which, in mimics, solved the problem of converting pyridoxal species to pyridoxamine species in biomimetic transaminase systems.

Serious untoward effects are rarely caused by erythromycin. Among the allergic reactions observed are fever, eosinophilia, and skin eruptions, which may occur alone or in combination each disappears shortly after therapy is stopped. Cholestatic hepatitis is the most striking side effect. It is caused primarily by erythromycin estolate and rarely by the ethylsuccinate or the stearate. The illness starts after about 10 to 20 days of treatment and is characterized initially by nausea, vomiting, and abdominal cramps. The pain often mimics that of acute cholecystitis. These symptoms are followed shortly thereafter by jaundice, which may be accompanied by fever, leukocytosis, eosinophilia, and elevated transaminases in plasma. Biopsy of the liver reveals cholestasis, periportal infiltration by neutrophils, lymphocytes, and eosinophils, and occasionally, necrosis of neighboring parenchymal cells. Findings usually resolve... 

Extensive studies on mimics of transaminase enzymes follow, and then mimics of enzymes that use thiamine pyrophosphate as their coenzyme. Studies on mimics of enzymes that perform aldol condensations are then described. After a brief section on mimics of enzymes that use coenzyme as a coenzyme, there is a description of mimics... 

Skouta, R, S Wei and R Breslow (2009). High rates and substrate selectivities in water by polyvinylimidazoles as transaminase enzyme mimics with hydrophobicaUy bound pyridoxamine derivatives as coenzyme mimics. Journal oftheAmerican Chemical... 

Breslow, R, AW Czarnik, M Lauer, R Leppkers, J Winkler and S Zimmerman (1986). Mimics of transaminase enzymes. Journal of the American Chemical Society, 108(8),... 

Further studies on artihcial transaminases were done by Breslow et al. by attaching the pyridoxamine to a PAMAM dendrimer (Figure 39). In contrast to the PEI polymer 101, which has not a well-defined structure, it is possible to build more structurally defined transaminase models with one pyridoxamine unit at the core of a PAMAM dendrimer. However, this dendrimeric transamination mimic 103 and even further recent dendrimeric transaminase enzyme mimics developed by Breslow et al. were not as potent as the polymeric ones described above. 

We have made a number of enzyme mimics in which a coenzyme group is attached to a cyclodextrin. For instance, much work has gone into mimics for transaminase, with pyridoxal or pyridoxamine units attached to P cyclodextrin [21]. In addition, as described above, a cyclodextrin with an attached thiazolium ring is a mimic of enzymes that use thiamine pyrophosphate as coenzyme. However, we thought that the attachment of a coenzyme B12 mimic to cyclodextrin would lead to a particularly interesting catalyst. 

---