University of Groningen
Omega transaminases: discovery, characterization and engineering
Palacio, Cyntia Marcela
IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite from it. Please check the document version below.
Document Version
Publisher's PDF, also known as Version of record
Publication date: 2019
Link to publication in University of Groningen/UMCG research database
Citation for published version (APA):
Palacio, C. M. (2019). Omega transaminases: discovery, characterization and engineering. Rijksuniversiteit Groningen.
Copyright
Other than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons).
Take-down policy
If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim.
Downloaded from the University of Groningen/UMCG research database (Pure): http://www.rug.nl/research/portal. For technical reasons the number of authors shown on this cover page is limited to 10 maximum.
Summary and perspectives
SUMMARY
Aminotransferases, or transaminases, are enzymes that catalyze the reversible transfer of an amino group from an amine donor to a ketone or aldehyde acceptor. These enzymes play an important role in the natural metabolism of amino acids and amines, but also have many realized and potential uses in applied biocatalysis. The asymmetric conversion of a ketone to a chiral amine is of particular interest, as the products appear as components of many bioactive compounds. One clearly successful example is the production of sitagliptin through a combined chemocatalytic and biocatalytic pathway, including a transaminase-mediated step that introduces an amine functionality and creates a chiral center (1). The substitution of chemical amination reactions by enzymatic transamination in multistep synthetic pathways is also attractive since it may give a more pure product, lower the amount of waste that is formed, reduce energy consumption, and simplify the whole production processes (2). Although some known natural transaminases are outstanding biocatalysts, several challenges remain open to make transaminases more widely applicable for efficient industrial processes. The enzyme stability required for high reaction temperatures and harsh conditions such as intensive stirring may be insufficient. In other cases, a broader substrate scope may be needed, and strategies to overcome equilibrium problems are required to transform transaminases investigated in the laboratory to widely used industrial biocatalysts. The work described in this thesis is aimed at exploring the properties of known and new transaminases and establishing the possibilities and limitations of their use in biocatalysis. Issues that are examined are enzyme stability, effect of coupled reactions on reaction equilibrium, and structural and catalytic properties of a novel class III transaminase.
In Chapters 2 and 3 we describe the discovery of a novel ω-transaminase termed PjTA in a strain of the Gram-negative bacterium Pseudomonas jessenii. This strain was isolated on basis of its capability to use caprolactam as a growth substrate. The PjTA enzyme belongs to the fold-type I group of PLP enzymes, in particular to the class III transaminases. The enzyme catalyzes the deamination of 6-aminohexanoic acid to produce 6-oxohexanoic acid, and was identified by proteomics profiling of genes induced by growth on caprolactam, employing the partial genome sequence to establish the putative identification and function of such proteins.
Using genome sequence information, gene clusters involved in caprolactam degradation in P. jessenii were identified. BLAST searches showed the presence of genes with high similarity to these P. jessenii caprolactam degradation genes in a strain of
Pseudomonas mosselii that was isolated as a caprolactam degrader from nylon industry
waste water (3). This suggests that the same caprolactam catabolic pathway is present in both strains. Additionally, searches against the non-redundant protein database revealed the presence of sequences with high similarity to PjTA in other Pseudomonas strains, indicating indicating that the ability to hydrolyze lactams or cyclic peptides may not be unusual in Pseudomonas.
During this study on bacterial caprolactam degradation by Ps. jessenii, we found that an ATP-dependent lactamase is involved in the conversion of caprolactam to 6-aminohexanoic acid (Chapter 2). This product can then be deaminated by the PjTA transaminase. The sequence analysis indicated that the ATP-dependent caprolactamase belongs to the ATP-dependent carboxylases/lactamase superfamily (4), and that it is further is similar to proteins annotated as 5-oxoprolinase or as hydantoinase. Other genes presumed to act in the caprolactam catabolic pathway were also discovered, with functions downstream of 6-aminohexanoic acid deamination. Proteomic studies, sequence comparison and (in the case of the lactamase and the transaminase) functional expression were used to identify the enzymes.
The properties of the PjTA transaminase are further explored in Chapter 3, using enzyme that was recombinantly expressed in E. coli. Structural studies and sequence alignment indicate that the novel ω-transaminase indeed belonged to subgroup AT-II, class III aminotransferases of the fold type I superfamily of PLP enzymes. Furthermore, comparison of the crystal structures of PjTA to the well-studied ω-transaminases of
Chromobacterium. violeaceum (CvTA) and Vibrio fluvialis (Vf TA) and the recently elucidated
structure of Ochrobactrum anthropi (OaTA) showed that the overall structures and active sites are very well conserved.
Activity studies with the caprolactam degradation pathway intermediates 6-aminohexanoic acid (6-ACA) and 6-oxohexanoate (6-OHA) showed that PjTA is a better biocatalyst for these compounds when compared to the homologous ω-transaminases of CvTA and VfTA. Kinetic studies showed that specificity constants (kcat/KM) with 6-ACA and 6-OHA were 2-20 fold and 25 better for the PjTA enzyme, respectively. This is an indication that the PjTA enzyme may have evolved further for efficient deamination of the 6-aminohexanoic acid intermediate. Comparing the structures of the three enzymes does not indicate large or global differences that immediate make clear why PjTA shows better activity for 6-ACA. An interesting and potentially important difference is visible in the tunnel like active site. Here the 6-ACA carboxylate group of the external aldimine intermediate makes a salt bridge with a conserved arginine residue, Arg417. This arginine is known to undergo conformational changes in homologous enzymes, which allow it to interact either hydrophobically with alkyl or aromatic groups of amine donors or via electrostatic interactions with the carboxylate of an acceptor such as pyruvate. The interaction between the arginine and the 6-aminohexanoate carboxylate requires a more outward position of the arginine iminium than it can easily adopt in CvTA and Vf TA because a bulky Phe needs to move away in the latter two enzymes. In PjTA the Phe is replaced by a Ser, facilitating the repositioning of the carboxylate binding arginine.
Kinetic assays revealed that at high concentrations 6-OHA may be inhibitory for the
PjTA transaminase. Such substrate inhibition is suspected to occur due to binding of
the aldehyde acceptor to the PLP form of the enzyme in competition with the amino donor. Inhibition by amino donor is also conceivable, but in this case by binding of the
6
substrate to the already aminated (PMP-containing) form of the enzyme. Whereas it may be problematic for cell-free applications of a transaminase in biocatalysis, we expect that when incorporated in an in vivo pathway, the enzyme would work at low substrate concentrations, where expression levels, stability, and selectivity will be the main factors that determine performance.
The substrate scope study included in Chapter 3 also shows that PjTA has a preference towards aromatic substrates, as was reported earlier for CvTA and Vf TA (5). When linear substrates were tested, PjTA revealed higher catalytic efficiencies towards 4-aminobutanoic acid than towards the linear substrate 1-aminopentane. The cause of the improved activity might be the recognition of the α-carboxylate of 4-aminobutanoic acid by the aforementioned flexible Arg417 in the binding site of the PjTA (6). Differently,
CvTA and Vf TA displayed the best activity with 1-aminopentane. This is probably related
to the different molecular environment around Arg417 in these transaminases. Future protein engineering studies aimed at controlling PjTA activity may include replacing residues surrounding this arginine. Another protein engineering target is the stability of the enzyme. We observed rapid loss of activity under various conditions, such as storage at -20˚C in 50 mM potassium phosphate, pH 8, containing 10% glycerol and during storage in the same buffer at 4˚C.
Improving the stability of the PjTA enzyme was investigated in the work described in Chapter 4. We especially wanted to explore the use of computational design and in silico screening to search for a thermostability-enhanced version of PjTA without extensive laboratory screening. To this end, the FRESCO protocol was used, which should reduce the amount of laboratory resources and time needed to find thermostable variants. Additionally, high-temperature molecular dynamics (MD) simulations were examined as a tool to identify relatively flexible surface regions in PjTA. Such regions might be involved in early unfolding and thereby trigger exposure of hydrophobic patches that lead to aggregation and irreversible enzyme inactivation. Since MD was expected to discover only surface flexibility, buried residues and interface residues were omitted from the analysis.
Following this approach, a library of 96 single surface mutants was designed and constructed by oligonucleotide-assisted mutagenesis. Mutations comprised different classes of flexibility. Favored, high-priority mutations were those located at positions that were on bases of the MD results expected with highest probability to be stabilizing. Non-favored mutations were those not located in regions predicted to have high flexibility by high temperature MD. From the 96 variants that were selected for construction and expression, 64 were produced and obtained in soluble form. This included 51 mutants distributed over 29 positions in favored areas, and 13 mutations at 9 positions in non-favored areas. Stabilities were measured with a thermal shift fluorescence assay, and the results showed that stability was moderately improved in only a few variants. Most mutations were neutral or detrimental and only five mutants showed a thermostability
SUMMARY AND PERSPECTIVES
increase of more than 1˚C in TM,app. The five improved mutants were all located in the regions displaying relatively high flexibility. The success rate of 11% (5/46, 46 of the 51 mutations were in high priority areas designated by the FRESCO protocol, 5 were from alternative (stricter) energy criteria) is lower than what was reported previously for FRESCO by Wijma et al. 2014 (7). Four of the five stabilized mutants showed modest reduction of the catalytic activity, which is an inverse relationship that has been more often observed between stability and function (8, 9). By inspecting the structure, possible explanations for this reduced activity were found. For example, the stabilizing mutation at position M419L is located close to the tunnel-like active site of the enzyme and this may influence substrate entry or binding, causing some reduction of enzyme activity. One mutation, A393R, displayed a 30% increase in catalytic activity, for which we have no good explanation. The complexity of the catalytic cycle, with two half reactions each encompassing several chemicals steps, makes it difficult to correlate effects on rates with the nature of a mutation.
Twelve non-stabilizing mutations were discovered in high priority locations and four of them appeared significantly destabilizing, which might be due to the introduction of hydrophobic groups on the protein surface. This type of substitutions is more often predicted by FRESCO, mostly due to Rosetta energy calculations, and can be easily omitted from the laboratory testing step.
The results showed that the computational single-mutation strategy together with high temperature MD simulations have the potential to identify regions where stabilizing mutations should be introduced, although the results with PjTA are not spectacular. Recent results in our laboratory (10) and observations with a tetrameric halohydrin dehalogenase (11) indicate that mutations that stabilize the subunit interface may also be computationally designed by the FRESCO strategy. Work with an aminotransferase obtained from a metagenomic library (12) indicated that mutations improving cofactor binding may have a large positive effect on stability. The benefits of using high temperature molecular dynamics simulations to identify priority regions for introducing stabilizing mutations may be more pronounced in case of monomeric enzymes.
Finally, in Chapter 5 we describe a three-enzyme biocatalytic network for the conversion of the ether alcohols butyldiglycol (a primary ether alcohol) and 1-butoxy-2-propanol (a secondary ether alcohol) to the corresponding amines. This enzymatic network for alcohol to amine conversion was built by combining an alcohol dehydrogenase from Geobacillus stearothermophilus, the transaminase from
C. violaceum also used in Chapter 3 (CvTA), and an alanine dehydrogenase from Vibrio proteolyticus. The three enzymes were overexpressed in E. coli and used in purified
form. Three interconnected reactions, i.e. alcohol oxidation, L-alanine-driven carbonyl transamination, and reductive pyruvate amination with ammonia, were carried out in a one-pot system, allowing production of the desired amine and simultaneous recycling of the alcohol dehydrogenase cofactor NAD+ and the amine donor L-alanine. Thus, the
6
net reaction is alcohol plus ammonia gives organic amine plus water, which indeed is a highly attractive overall conversion because there is no use of reducing cofactors or an organic amine donor. Various concentrations of alcohol, ammonium, and alanine were tested to improve the yield of amine, which reached 40%. The optimization tests showed that none of the substrates could be added at high initial concentrations due to their inhibitory effect (alanine and the alcohol substrate) or a negative effect on the stability of the enzymes (ammonium).
The thermodynamic equilibrium of the three-enzyme amination network was another main factor determining the maximum degree of conversion, in accordance with previous studies (13). Therefore, equilibrium constants, according to Goldberg (14) notation, were calculated from experimental data. The results showed that for the first reaction, from alcohol to ketone or aldehyde, the equilibrium is in the unwanted direction of the alcohol. Similarly, the equilibrium constant of the transamination reaction is unfavorable for production of the amine, with the equilibrium strongly lying on the side of the substrates ketone/aldehyde and L-alanine. On the other hand, the alanine dehydrogenase-catalyzed recycling reaction, which recycles the amino donor alanine and the electron acceptor NAD+, displays a favorable equilibrium constant. Thus, the latter reaction appears to drive the enzymatic network to the amine production, and it will do so better if the ammonium concentration is higher.
Subsequently, we theoretically examined the overall reaction equilibrium as a function of substrate concentrations. Results showed that in order to achieve a good conversion (>90%), a high concentration (≥ 0.7M) of the amino donor ammonium should be used. Therefore, a range of ammonia concentrations (80 mM to 1M) was tested. As a result, conversion was increased from 40% to 60% when a total concentration of 280 mM of ammonium was used. For this, a step-wise addition of ammonia was required, as the enzymes precipitated or aggregated when a high concentration of ammonium was added in one step. The results again suggested that enhancing enzyme stability, now in the presence of high ammonia and salt, should be an important future goal.
PERSPECTIVES
Transaminases hold powerful characteristics for use as as catalysts, including a broad substrate scope and high product enantioselectivity in asymmetric conversions. Nevertheless, limitations due to unfavorable equilibrium, low thermal- and process stability, and narrow substrate scope should be solved to enable wider application in sustainable production of amines. Besides discovery of new natural transaminase variants, which will allow a broader range of conversions, some specific engineering aims or strategies emerge from the work reported here.
In Chapter 4, we attempted to increase the thermostability of the PjTA transaminase
SUMMARY AND PERSPECTIVES
through the FRESCO computational strategy, especially with the aim to examine if MD simulations can steer the choice of target regions. However, targeting mutations to the protein surface, where MD suggested priority regions, yielded only moderate improvements. For obtaining robust transaminase variants, it is advisable to shift attention from mutating surface residues to improving subunit interface interactions. Mutating residues on the subunit interface of the dimeric aminotransferases might strengthen the interactions between the monomers, increasing the thermostability of the enzyme if denaturation commences by dimer dissociation. Hayashi et al. (15) showed that a dimeric D-amino acid transaminase reached higher thermostability and catalytic efficiency when a tryptophan located at the interface of the subunits was substituted by a phenylalanine. Preliminary experiments in our lab indicate that computational design of mutations at the PjTA subunit interface indeed is very effective (10). As mentioned above, an option suggested by the work of Börner et al. (16) is to improve cofactor binding, which probably can also be achieved by computational methods.
In Chapter 5 we confirmed that both stability and equilibrium of an amination reaction are issues to overcome when designing a transaminase catalyzed process based on a multi-enzyme network. Although not tested in this thesis, one of the most studied strategies to improve the stability of enzymes under process conditions is immobilization. This strategy is widely used because immobilization can lead to markedly improved operational performance of biocatalysts. For instance, Koszelewski et al. (17) showed that the apparent optimum temperature of transaminase TA-117 increased from 30˚C to 40-50˚C when the enzyme was immobilized in sol-gel/celite. Immobilization also allowed the re-use of the biocatalyst through different cycles, even when used at elevated temperatures. Neto et al. (18) demonstrated that an ω-transaminase immobilized on solid carriers allowed the reuse of the enzyme for 8 cycles of 24 h, retaining more than 50% of the original activity. This strategy might be beneficial to increase the stability of the enzymes and allow the re-usability of complete multi-enzymatic reactions.
Another way to increase the stability of transaminases is the usage of whole cells, which also was not examined in this thesis. Whole cell biocatalysis is a promising strategy because stability of the enzymes is improved and addition of cofactors is unnecessary, making network reactions as used in Chapter 5 more economical. Klatte et al. (19, 20) showed a redox self-sufficient amination of alcohols through a network reaction similar to that studied here in Chapter 5, using plasmid-born overproduction in E. coli of the three enzymes: transaminase, alcohol dehydrogenase, and alanine dehydrogenase.
One major challenge to overcome when designing reactions with ω-transaminases is the unfavorable thermodynamic equilibrium of the reaction (21), as described in Chapter 5. A widely used strategy is to shift the equilibrium by the addition of an excess of one of the substrates. Truppo et al. (22) showed that using a large excess of the amino donor isopropylamine with acetophenone as amino acceptor resulted in 95% conversion. Although this strategy might seem simple to implement, an important consideration
6
is that enzyme activity may be lowered when a large access of substrate is added due to substrate inhibition or enzyme unfolding. The latter may affect all enzymes used in a multi-enzymatic system, and even salting-out effects may occur when using high substrate concentrations. The choice of amine donor will also influence the position of the thermodynamic equilibrium.
Finally, the removal of a product or co-product is another uncomplicated method to shift the equilibrium of transaminase mediated reactions. This can be accomplished through, for example, the continuous physical separation of the product from the reaction mixture, such as through volatilization of acetone when isopropylamine is used as amino donor (23), or the design of cascade reactions (24), such as the recycling of the amino acceptor pyruvate back to alanine (see Chapter 5), its conversion to lactate by a dehydrogenase, or its enzymatic decarboxylation.
SUMMARY AND PERSPECTIVES
REFERENCES
1. Savile CK, Janey JM, Mundorff EC, Moore JC, Tam S, Jarvis WR, Colbeck JC, Krebber A, Fleitz FJ, Brands J, Devine
PN, Huisman GW, Hughes GJ. 2010. Biocatalytic asymmetric synthesis of sitagliptin manufacture. Science 329:305–
310.
2. Malik MS, Park E, Shin J. 2012. Features and technical applications of ω-transaminases. Appl Microb Biotechnol
94:1163–1171.
3. Hong S-J, Park G-S, Khan AR, Jung BK, Shin J-H. 2016. Draft genome sequence of a caprolactam degrader bacterium:
Pseudomonas taiwanensis strain SJ9. Braz J Microbiol 69:2015–2016.
4. Weidenweber S, Schühle K, Demmer U, Warkentin E, Ermler U, Heider J. 2017. Structure of the acetophenone carboxylase core complex: prototype of a new class of ATP-dependent carboxylases / hydrolases. Sci reports 7:1–10. 5. Kaulmann U, Smithies K, Smith MEB, Hailes HC, Ward JM. 2007. Substrate spectrum of ω-transaminase from
Chromobacterium violaceum DSM30191 and its potential for biocatalysis. Enzyme Microb Technol 41:628–637.
6. Steffen-Munsberg F, Vickers C, Thontowi A, Schätzle S, Meinhardt T, Svedendahl Humble M, Land H, Berglund
P, Bornscheuer UT, Höhne M. 2013. Revealing the structural basis of promiscuous amine transaminase activity.
ChemCatChem 5:154–157.
7. Wijma HJ, Floor RJ, Jekel PA, Baker D, Marrink SJ, Janssen DB. 2014. Computationally designed libraries for rapid enzyme stabilization. Protein Eng Des Sel 27:49–58.
8. Poole LB, Loveys DA, Hale SP, Gerlt JA, Stanczyk SM, Bolton PH. 1991. Deletion of the omega-loop in the active site of staphylococcal nuclease. 1. Effect on catalysis and stability. Biochemistry 30:3621–7.
9. Gleason FK. 1992. Mutation of conserved residues in Escherichia coli thioredoxin: Effects on stability and function. Protein Sci 1:609–616.
10. Q. Meng, C. Palacio, H.J. Wijma EL. Robust ω-transaminases by computational stabilization of subunit interfaces. In preparation.
11. Arabnejad H, Lago MD, Jekel PA, Floor RJ, Thunnissen AWH, Scheltinga ACT Van, Wijma HJ, Janssen DB. 2017. A robust cosolvent-compatible halohydrin dehalogenase by computational library design. Protein Eng Des Sel 30:175– 189.
12. Börner T, Bartsch S, Vogel A, Adlercreutz P, Grey C. 2017. Three in one: temperature, solvent and catalytic stability by engineering the cofactor-binding element of amine transaminase. ChemBioChem 18:1482–1486.
13. Höhne M, Kühl S, Robins K, Bornscheuer UT. 2008. Efficient asymmetric synthesis of chiral amines by combining transaminase and pyruvate decarboxylase. ChemBioChem 9:363–365.
14. Goldberg PW, Roth A. 2014. Bounds for the query complexity of approximate equilibria. ACM 8:639–656.
15. Hayashi H, Inoue Y, Kuramitsu S, Morino Y, Kagamiyama H. 1990. Effects of replacement of tryptophan-140 by phenylalanine or glycine on the function of Escherichia coli aspartate aminotransferase. Biochem Biophys Res Commun
167:407–412.
16. Börner T, Rämisch S, Reddem ER, Bartsch S, Vogel A, Thunnissen A-MWH, Adlercreutz P, Grey C. 2017. Explaining operational instability of amine transaminases: Substrate-induced inactivation mechanism and influence of quaternary structure on enzyme–cofactor intermediate stability. ACS Catal 7:1259–1269.
17. Koszelewski D, Müller N, Schrittwieser JH, Faber K, Kroutil W. 2010. Immobilization of ω-transaminases by encapsulation in a sol-gel/celite matrix. J Mol Catal B Enzym 63:39–44.
18. Neto W, Schürmann M, Panella L, Vogel A, Woodley JM. 2015. Enzymatic immobilisation of ω-transaminase for industrial application: Screening and characterisation of commercial ready to use enzyme carriers. J Mol Catal B Enzym
117:54–61.
19. Klatte S, Wendisch VF. 2015. Role of L-alanine for redox self-sufficient amination of alcohols. Microb Cell Fact 14:1–10. 20. Klatte S, Wendisch VF. 2014. Redox self-sufficient whole cell biotransformation for amination of alcohols. Bioorg Med
Chem 22:5578–5585.
21. Höhne M, Brundiek H. 2016. Transaminases – A Biosynthetic Route for Chiral Amines, p. 4–451. In Lutz Hilterhaus, Andreas Liese, Ulrich Kettling, and GA (ed.), Applied Biocatalysis From From Fundamental Science to Industrial ApplicationsFirst. Wiley-VCH.
22. Truppo MD, Rozzell JD, Moore JC, Turner NJ. 2009. Rapid screening and scale-up of transaminase catalysed reactions. Org Biomol Chem 7:395–398.
23. Park E, Dong J, Shin J. 2013. ω-Transaminase-catalyzed asymmetric synthesis of unnatural amino acids using isopropylamine as an amino donor. RSC oublishing 11:6929–6933.
24. Simon RC, Richter N, Busto E, Kroutil W. 2014. Recent developments of cascade reactions involving ω-transaminases. ACS Catal 4:129–143.
