University of Groningen

Novel halohydrin dehalogenases by protein engineering and database mining Schallmey, Marcus

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: 2015

Link to publication in University of Groningen/UMCG research database

Citation for published version (APA): Schallmey, M. (2015). Novel halohydrin dehalogenases by protein engineering and database mining. University of 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).

The publication may also be distributed here under the terms of Article 25fa of the Dutch Copyright Act, indicated by the “Taverne” license. More information can be found on the University of Groningen website: https://www.rug.nl/library/open-access/self-archiving-pure/taverne- amendment.

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.

Download date: 07-10-2021

   
  


  

 

 

 
 


!


Biotransformation and Biocatalysis, Groningen Biomolecular Sciences and Biotechnology Institute, University of Groningen, Nijenborgh 4, 9747 AG Groningen, The Netherlands.

  
"

Schallmey M, Floor RJ, Szymanski W, Janssen DB: 7.8 Hydrolysis and reverse hydrolysis: Halohydrin dehalogenases. In Comprehensive Chirality . Carreira EM, Yamamoto H (eds). Amsterdam: Elsevier; 2012:143–155. © 2012 Elsevier





# $


&
Halohydrin dehalogenases are members of a distinct class of enzymes that catalyze the release of halides from vicinal halohydrins via the formation of the corresponding epoxides. In the reverse direction, these enzymes also catalyze the nucleophilic epoxide-ring opening with alternative anions such as cyanide, azide, and nitrite, yielding important β-substituted alcohols. Some of these reactions proceed with excellent regio- and enantioselectivity, making halohydrin dehalogenases interesting biocatalysts for synthetic chemistry applications. This chapter gives an overview of mechanistic and engineering studies of these enzymes. Further, examples are presented which show the usefulness and versatility of these biocatalysts in the preparation of intermediates for enantiopure fine chemicals.



%
'  

'  
Halohydrin dehalogenases (also called haloalcohol dehalogenases, haloalcohol/ halohydrin epoxidases, or hydrogen-halide lyases; EC 4.5.1.-) (HHDHs) belong to a distinct group of enzymes, which catalyze the removal of a halide ion and a proton from a vicinal haloalcohol with formation of an epoxide ring [1, 2]. These reactions are different from the conversions catalyzed by other well-known dehalogenases such as haloalkane dehalogenases and haloacid dehalogenases, which displace a halogen by a hydroxyl group in a net hydrolytic reaction. For more information on reactions catalyzed by other dehalogenases, the reader is referred to other reviews [3–8]. HHDH enzymes are members of the short-chain dehydrogenase/reductase (SDR) enzyme superfamily but differ distinctively in structure and mechanism from the more than 160,000 reported SDR sequences in UniProt [9, 10]. HHDHs have only been discovered in a few bacterial strains, where they are involved in the biodegradation of xenobiotics such as epichlorohydrin ( Figure 1 ), 1,3-dichloro-2- propanol, or 1,2-dibromoethane. Mineralization of these compounds proceeds via the elimination of halide from haloalcohol intermediate to form an epoxide by intramolecular ring- closure. The epoxides are subsequently converted further, which may involve different reaction types (e.g. by epoxide hydrolases [11, 12]).

Figure 1. Degradation pathway of epichlorohydrin by Agrobacterium radiobacter AD1. Epichlorohydrin is first hydrolyzed by an epoxide hydrolase (EchA) to give 3-chloro-1,2- propanediol. Then, this haloalcohol is substrate for a HHDH (HheC) producing glycidol which is further metabolized [9]. The occurrence of this enzymatic dehalogenation reaction was first described in 1968 by Castro and Bartnicki, who studied a Flavobacterium strain that degraded 2,3-dibromo-1- propanol [13]. The partially purified enzyme was able to remove bromide or chloride anions from vicinal halohydrins, forming the corresponding epoxides, as well as to open the epoxide ring by terminal attack of bromide and chloride nucleophiles. Furthermore, it was shown that the ring-closure reaction proceeds via a stereospecific trans elimination of the proton and halide ion [14]. Later, it was found that also iodide can act as a nucleophile in the epoxide- ring opening reaction [15]. Over time, several studies reported the isolation of new bacterial strains able to utilize different haloalcohols or epichlorohydrin as sole carbon sources with proposed HHDH activities [16–23]. The interest in this novel type of dehalogenases rose upon the discovery that these enzymes can be employed for stereoselective biotransformations to produce optically active halohydrins from prochiral or racemic substrates [18–21, 24–29]. Of particular interest was the finding that HHDHs can also catalyze the cyanide-mediated epoxide-ring opening with the formation of novel carbon-carbon bonds [30], and that these reactions proceed with high enantioselectivity [31]. Subsequently, several HHDH genes were cloned and recombinantly expressed in Escherichia coli [9, 32, 33], allowing detailed kinetic and engineering studies [34–37]. The crystal structures of two wild-type HHDHs were solved by Dijkstra and coworkers [38, 39]. The availability of high quality structures allowed the engineering of the



(

# $

enzymes by site-directed mutagenesis and provided an explanation for observed enantioselectivities [35, 40, 41]. Further studies demonstrated that several non-natural nucleophiles such as azide or nitrite also could be used for epoxide-ring opening, showing the catalytic promiscuity of HHDHs [42–45]. The diversity of reactions catalyzed by these dehalogenases allowed their application in diverse processes, such as the removal of halogen compounds from polymer building blocks [46] and the large-scale preparation of an important intermediate in the atorvastatin side chain synthesis [47, 48].

 

*  
At the beginning of this study, only six HHDH gene sequences had been reported which originated from bacterial isolates, with similar dehalogenating activities, from at least three different geographic regions (England, The Netherlands, and Japan), and revealed only three different enzyme types. Based on sequence similarities and activity profiles, three phylogenetic groups, types A, B, and C, were proposed for the classification of HHDHs. Every group was represented by two cloned HHDH genes, i.e. hheA from Corynebacterium sp. strain N-1074 [32], and hheA2 from Arthrobacter sp. strain AD2 [9]; hheB from Corynebacterium sp. strain N-1074 [32], hheB2 from Mycobacterium sp. strain GP1 [42], and two identical hheC sequences from Agrobacterium radiobacter AD1 [9] and Rhizobium sp. strain NHG3 [33]. Although sequence identities within types A, B, and C exceeded 97%, identities between the groups were below 33%. Then, only enzyme sequences of type A and C identified each other as closest homologues in sequence databases whereas B-type enzymes showed closer similarity with other non-dehalogenase SDR superfamily members. Recently, many other HHDH enzyme sequences have been reported [49–53]. Phylogenetic analysis of these enzymes and their large evolutionary distance to other non-dehalogenase SDR enzymes suggests that these enzymes are not examples of post-industrial evolution of enzymatic activity to degrade halogenated xenobiotics but have evolved a long time ago from SDR-like ancestors [49, 54]. Although also several sequences with significant similarity can be found in metagenomic sequences data [54], they are much less abundant than SDR enzymes with more than 160,000 annotated sequences [10]. Thus, HHDHs are rare enzymes, which is in agreement with the fact that all known enzymes of this class have been obtained from microbial enrichment cultures.



 

HHDHs belong to the SDR protein superfamily and share various structural, sequence, and mechanistic features with this diverse enzyme family [9, 38, 55]. Just like SDR enzymes, which are mostly found to occur as homodimers or -tetramers, HHDHs are homotetramers formed by a pair of dimers ( Figure 2 ). HHDHs also exhibit the typical Rossmann fold structure of parallel β-sheets flanked by α-helices, but they lack the typical N-terminal Gly- rich motif T-G-x(3)-[AG]-x-G that shapes the NAD(P)+ cofactor binding site of SDR enzymes [56].



)
'  

Figure 2. Highly similar structures of A) gluconate 5-dehydrogenase from Streptococcus suis (PDB 3CXT) [57] and B) HheC from Agrobacterium radiobacter AD1 (PDB 1ZMT) [40]. The structure of the SDR enzyme is shown with D-glucurate (orange) and NADP + (blue) bound in one of its active sites. The HHDH structure is displayed with one active site-bound molecule of (R)- p-nitrostyrene oxide (red). HHDHs possess a highly conserved active site catalytic triad that is similar to the one in SDR enzymes [38, 56]. In SDR enzymes, the Tyr of the Ser-Tyr-Lys triad is responsible for the deprotonation of the substrate’s hydroxyl group, and the enzyme-bound NAD(P) + cofactor is responsible for hydride abstraction ( Figure 3A ). In HHDHs, the catalytic triad is composed of Ser-Tyr-Arg ( Figure 3B ). Instead of a cofactor-binding site, residues mapping to the Gly- rich T-G-x(3)-[AG]-x-G loop of SDR enzymes as well as residues from other loops form a spacious anion binding pocket [38, 58]. This pocket stabilizes anions via hydrogen bonds from backbone amides, slightly positively-charged edges of aromatic residues, as well as via a bound water molecule. Although distinct hydrogen bond donors as found in hydrolytic dehalogenases are missing, aromatic residues are present which have been shown to facilitate carbon-halogen bond cleavage [59]. Furthermore, HHDHs possess a proton relay system, transporting protons from the active site Arg toward the bulk solvent via the side-chain of a partly surface-exposed Asp. This is in contrast to SDR enzymes which employ the main-chain carbonyl of a highly conserved Asn and a hydroxyl group of the ribose moiety for interactions with the leaving proton [38]. Further understanding of the HHDH reaction mechanism was obtained through kinetic [35], mutational [9, 38], structural [38], and theoretical studies [60, 61] using HheC from Agrobacterium radiobacter AD1. During catalysis, the Ser forms a hydrogen bond with the substrate's hydroxyl group, the Arg lowers the pK a of the proton-abstracting Tyr, and an Asp that is located close to the surface of the enzyme is involved in proton transfer to the solvent via a hydrogen-bonded network of side chains and water molecules. The developing negative charge of the oxyanion is stabilized by Ser and leads to a nucleophilic attack at the neighboring carbon atom. In consequence, the halide-carbon bond is cleaved and the released halide anion is stabilized in the anion-binding site. The validity of this proposed reaction scheme was



+

# $

confirmed by quantum mechanical calculations [62].

Figure 3. Differences in reaction mechanism for the catalytic triad of A) Ser-Tyr-Lys of gluconate 5-dehydrogenase from Streptococcus suis (PDB 3CXT) [57] and B) Ser-Tyr-Arg of HheC from Agrobacterium radiobacter AD1 (PDB 1PX0) [38]. Both enzymes catalyze the abstraction or donation of a proton. However, SDR enzymes oxidize or reduce an alcohol or aldehyde employing a NAD(P) + or NAD(P)H cofactor for hydride transfer, respectively. HHDHs, on the other hand, cleave or form a halogen-carbon bond by intramolecular displacement or nucleophilic attack of a halide anion, facilitated via residues of the anion-binding pocket, respectively. In the epoxide-ring opening reactions, the mechanism is essentially the reverse. The nucleophilic attack of halide combined with oxygen protonation leads to the formation of a carbon-halogen bond and a hydroxyl group. Tyr and Ser contribute to a tetrahedral arrangement of the epoxide oxygen by donating hydrogen bonds towards its lone electron pairs. The Tyr now acts as an acid catalyst through activation by Arg, and Ser stabilizes the hydroxyl group, whereas Asp allows proton import from the solvent. Apparently, the anion- binding site can accommodate different nucleophiles for attack on the terminal position of the epoxide ring, including azide and cyanide, which cannot be detected in the ring-closure direction because the equilibrium is strongly in the direction of the ring-opened β-substituted alcohol. The charge of the surface-located Asp and its influence on the pK a of the active site groups may be responsible for the observed differences in pH optima of the epoxide-ring formation and epoxide-ring opening reaction (pH 4-5 and pH 8-9, respectively) [40].

' 
- 
 $

The structures of HheA2 [39] and HheC [38] were solved by X-ray crystallography and provide a structural basis for comparing both enzymes. Despite the fact that both enzymes exhibit only a modest sequence identity of 34% and 51% similarity, they are currently the



,
'  

most similar wild-type HHDH structures to each other with a RMSD of 1.3 Å over 234 C
atoms [39]. They employ the same loop which is part of the cofactor binding site in SDR proteins as part of the anion-binding site and have organized their catalytic triad in a similar way. An important difference lies in a leucine that lines up the active site of HheA2 whereas HheC has a bulky tryptophan at a similar position. Another tryptophan residue from a C- terminal extension of the opposing monomer protrudes into the active site of HheC, while this C-terminal extension is missing in HheA2. Altogether, this results in a more open active site for HheA2 which provides an explanation for the observed difference in enantioselectivity between both enzymes. For example, while HheC exhibits high ( R)-enantioselectivity in the epoxide-ring formation reaction from p-nitrophenyl-2-bromoethanol (E = 150) [36] or during azidolysis of the corresponding epoxide (E > 200) [42], no preference for either enantiomer could be observed using HheA2 with E values smaller than 10 [42, 43]. No halide-bound structure of HheA2 is available, but the location of a water molecule gives information on the putative halide binding mode. HheA2 uses different hydrogen bonding interactions to bind the halide as HheC. Instead of hydrogen bonds with backbone amides and a water molecule, HheA2 establishes hydrogen bonding of the anion with only one backbone amide and an Asn side chain [39]. Thus, A- and C type enzymes have different ways of anion binding but these differences do not change the position where the anion is bound [38].

# 
* "


/ 
The epoxide-ring formation reaction corresponds to the natural function of the enzyme in the bacteria from which they have been isolated [13]. Later, several studies revealed further substrates of halohydrin dehalogenases and showed that many different β-substituted haloalcohols can be converted with high enantioselectivity. Initially, it was found that phenyl- substituted haloalcohols and derivatives thereof, such as the chromogenic substrate 2-bromo- p-nitrophenylethanol [43], can be converted to the corresponding epoxides with enantioselectivities as high as E = 150 [35]. Also the 2,2-disubstituted haloalcohol 1-chloro- 2-methyl-2-propanol is converted with excellent rates [34]. Recently, also 3-alkenyl and heteroaryl chloroalcohols [63] as well as methyl [64] or ethyl esters [47, 48, 51, 52, 65] of 4- chloro-3-hydroxybutyric acid were reported to be halohydrin dehalogenase substrates. Especially the latter group of haloalcohols are important since they allow biocatalytic production of the atorvastatin side chain intermediate [66–70]. In the epoxide-ring opening reaction, a range of anionic nucleophiles is accepted. This includes cyanide, nitrite, azide, cyanate, thiocyanate, and formate, and in all cases almost exclusively β-substituted alcohols are produced ( Figure 4 ) [45]. The reactions result in different bond formation: carbon-nitrogen bonds in the case of azide and nitrite, carbon-carbon bonds with cyanide, and carbon-oxygen bonds with nitrite and formate. Also carbon-sulfur bonds can be formed, i.e. when thiocyanate is used as nucleophile whereas nucleophilic attack of cyanate produces oxazolidinones.



.

# $


Figure 4. Catalytic scope of halohydrin dehalogenase-catalyzed epoxide ring-opening of epoxides and various nucleophiles (modified from [45]). An overview of reactions catalyzed by different halohydrin dehalogenases, including the range of accepted halohydrin substrates and epoxides for conversions with other nucleophiles, is shown in Table 1 . The ability of halohydrin dehalogenases to break or form different chemical bonds, allowing a multitude of chemical reactions and products, is an example of remarkable catalytic promiscuity, which is an area of intensive research [71–75]. Whereas with most enzymes the stray reactions are very slow, the nucleophilic ring opening catalyzed -1 by halohydrin dehalogenases occurs at decent rates (k cat ) such as 40 s , or about two orders of magnitude higher than that observed with natural nucleophiles such as bromide or chloride [45]. Epoxide-ring opening rates for halides are in the range of 0.2 to 0.8 s -1 and cyanolysis of epoxides occurs at similar rates. Rates observed for nitrite and azide are 10-fold, even 100- fold higher, respectively.



0
'  

a Table1. , Highest , ratereported; Azidolysis 1,3-dibromo-2-propanol ,-pxbtn HheC nitrite Othernucleophiles HheC 1,2-epoxybutane HheC p HheC 1,2-epoxy-2-methylbutane 1,2-epoxy-2-methyl-3-butene 1,2-epoxy-2-cyclohexyl 1,3-dichloro-2-propanol Dehalogenation thiocyanate product( cyanate 12.1 HheB HheA2 3,3-dimethyl-1,2-epoxybutane epichlorohydrin HheC 1,2-epoxybutane HheC Cyanolysis 109 ( 1-chloro-pent-3-en-2-ol HheC 2-chloro-1-fur-2-yl-ethanol k ( Enzyme 1-chloro-2-methyl-2-propanol Substrate E R ntosyeeoie HheC -nitrostyrene oxide --hoo4pey-u--n2o HheC )-1-chloro-4-phenyl-but-3-en-2-ol --hoohnltao He 1. 12.1 HheC )-2-chlorophenylethanol b HheC b Catalytic activitiesCatalytic and substrate scopehalohyof HheC b HheC b , , 1,2-epoxybutaneWith substrate. as hA 2. 27.7 HheA HheA2 9.7 cloned not 63.8 HheC 117 HheB-type 4.0 HheB HheA2 39.7 HheC HheB













26.8 40.0 0.1 1.0 0.5 1.9 18 0.1 1.9 0.1 22.3 21.9 c a t 2.5 2.2 [s a a a a a a a a a a E = E 28 product( E = E 24 = E 24 product( product( product(S)-3-hydroxy-3-methyl-4-pentenenitrile, E product( > E 200, configurationof remaining chloroalcohol: a a a a - 1 product( E > E 200, configurationof remaining chloroalcohol: > E 200, configurationof remaining chloroalcohol: Comments ] drindehalogenases. R S R R R R --ylhxl3hdoyrpoirl,E=16 [79] )-3-cyclohexyl-3-hydroxypropionitrile,E= 106 --zd--yrxbtn,E=2 [45] )-1-azido-2-hydroxybutane,E28= --zd-4ntohnlehnl 0 [42] [79])-2-azido-(4-nitrophenyl)ethanol, > E 200 )-3-hydroxy-3-methylpentanenitrile,E200> [79 )-3-hydroxy-4,4-dimethylpentanenitrile,E109= --hoo2hdoyuyoirl,e 52 [31] )-1-chloro-2-hydroxybutyronitrile,ee95.2% ( ( ( =11 [79] 141= S S S [63] ) [63] [63] ) ) [16] [36] [28] [25] [16] Reference [45] [45] [45] [45] [30] [24] [34] [36] ]



1

# $


Kinetic studies using product inhibition measurements and stopped-flow fluorescence experiments that follow intrinsic protein fluorescence indicated that epoxide-ring formation (dehalogenation) proceeds via an ordered uni-bi mechanism, in which bromide release is the rate-limiting step [35]. According to this scheme (Figure 5 ), binding of substrate to form the enzyme•halohydrin complex (k 1) is followed by a fast unimolecular reaction producing the ternary enzyme•halide•epoxide complex (k 2). The ternary complex is resolved by slow release of halide (k 3) followed by epoxide release as the last step of the catalytic cycle (k 4).

Figure 5. Ordered uni-bi mechanism of epoxide-ring formation accompanied by halide release from haloalcohol substrate (based on [35]). The exact pathway of halide release is not clear in structural terms since there is no obvious tunnel connecting the halide binding site to solvent, except through the substrate binding site, which however is closed off by the product. Fluorescence experiments indicate that significant conformational changes may accompany halide binding and release, but to what extend this involves the anion-binding loop is unclear [37]. Epoxide formation from haloalcohols and epoxide-ring opening by halide are equilibrium reactions governed by thermodynamics. The equilibrium depends on the nature of the halide, and changes from epoxide towards halohydrin in the order of chloride, bromide, to iodide [1]. Shifting of conversions to the product side can only be accomplished by adding one of the reactants in excess, by changing the reaction conditions, or, in the direction of epoxide formation, by removing one of the products from the reaction mixture. Product removal can, for example, be achieved by removal of epoxide with an epoxide hydrolase, leading to diol formation and more complete conversion of chlorohydrins such as 2,3-dichloro-1-propanol or 2-chloro-1-phenylethanol [29]. Also sequential two-step-single-enzyme approaches employing the subsequent nucleophilic epoxide-ring opening by other nucleophiles such as azide [76] or cyanide [64] were successful methods to obtain higher conversion yields. In situ product removal is another possibility to shift the equilibrium to higher product yields. For example, this principle was successfully applied in HHDH-catalyzed reactions by using OH --loaded anion exchange resins to remove HX produced during the dehalogenation of chlorohydrins such as 2-chloro-1-phenylethanol [77] or the adsorption of epichlorohydrin to macroporous resins produced from the dehalogenation of 1,3-dichloro-2-propanol [78].

 3

  *
Biochemical studies on different halohydrin dehalogenases showed that epoxide-ring formation proceeds in a highly stereoselective manner for several substrates [18–21, 24, 27, 29]. Epoxide-ring opening reactions occur with high regioselectivity, nucleophilic attack taking place on the terminal carbon atom, and, dependent on the substrate, also with high stereoselectivity [30, 42, 44, 79, 80]. Kinetic studies showed that the observed ( R)-preference of HheC [29] is due to preferential binding of the (R)-configured haloalcohols over the ( S)- enantiomer [35]. Later, Dijkstra and coworkers revealed an unproductive binding mode for



2
'  
the non-favored ( S)-epoxide [40]. Examining structures obtained after soaking crystals of HheC with the ( R)- and ( S)-enantiomers of p-nitrostyrene oxide, it was found that the aromatic rings of both enantiomers occupied similar positions in the active site, but that the orientation of the epoxide rings was different ( Figure 6 ). For the preferred ( R)-epoxide, the epoxide oxygen orients towards the active site Tyr and Ser with the expected hydrogen bonding interaction that are involved in catalysis, whereas the oxygen of the ( S)-epoxide orients away from the catalytic residues without proper hydrogen bonding pattern. The epoxide ring thus has an inverted orientation for the non-preferred enantiomer, with the oxygen instead of the terminal carbon pointing towards the anion-binding site. The higher relative activity of HheC with haloalcohols having a terminal halide substituent as compared to haloalcohols with a terminal hydroxyl group was observed when comparing 1,3-dichloro-2-propanol and 2,3-dichloro-1-propanol and is an example of the enzyme’s high regioselectivity. It can be explained as follows [9]. The anion-binding pocket defines a space for optimal stabilization of the halide that is released during carbon-halogen bond cleavage and serves as a site for activation of the nucleophile in epoxide-ring opening reactions. The hydrogen bonding capabilities of the general acid-base Tyr and of Ser define a subsite within the active site where protonation and deprotonation can occur and where the substrate's oxygen must bind. Because there is room for the R-group only in a narrow cavity pointing towards the surface of the protein, epoxide formation reactions ideally require binding of a haloalcohol with a terminal halogen in the anion-binding site and a hydroxyl on the flanking secondary carbon allowing an anti-displacement reaction. This mode of action is in agreement with the highly regiospecific nucleophilic attack on the β-carbon of the epoxide ring observed with different nucleophiles [31, 42–45, 79, 81]. The regioselectivity was also explained by quantum mechanical calculations [60].

Figure 6. Productive and non-productive binding modes of A) ( R)- and B) ( S)- p-nitrostyrene oxide (pNSO), respectively, in the active site of HheC as molecular reasoning for its stereo- and regioselectivity. The orientation of the catalytic residues Ser132, Tyr145, and Arg149 as well as of the epoxide in A) and B) was taken from published crystal structures 1ZMT and 1ZO8, respectively [40]. In the pNSO-complexed structures, the position of the anion (green) is occupied by a water molecule but several other HheC structures confirm the location of bromide or chloride in the pocket (PDB 1PWX or 1PWZ, respectively) [38].





# $


!  

* 
$$  
Removal of halogenated contaminants Several examples show the applicability of halohydrin dehalogenases in the removal of organohalogen compounds from contaminated waste sites as well as their application in food industry. In most of the cases, whole cells have been applied in processes for the removal of toxic halogenated compounds. For example highly toxic and xenobiotic 1,2,3-trihalopropanes could be degraded by a genetically engineered Agrobacterium radiobacter strain able to utilize 2,3-dichloropropanol via its own halohydrin dehalogenase and epoxide hydrolase [82]. Whole cells with halohydrin dehalogenase activity can be applied for the decontamination of sites polluted with 1,3-dichloro-2-propanol [22, 83–85] or 2,3-dichloro-1-propanol [17, 23, 33, 84]. Recently, the removal of halogenated contaminants such as epichlorohydrin, 1,3-dichloro-2- propanol, and 3-chloro-1,2-propanediol from cationic polyamino-polyamide-epichlorohydrin (PAE) resins used as paper coatings to increase its wet strength properties in food contact applications has been implemented at multi-plant scale [46]. In food applications such as acid hydrolysis of proteins and during the deodorization of vegetable oils, toxic mono- and dichlorinated propanols and corresponding fatty acid esters can be formed, causing food safety issues. Processes have been described to prepare non-toxic glycerol suitable for food industry applications by employing halohydrin dehalogenases as whole cell lysate in combination with different lipases [86] or as purified enzymes in a cascade reaction in combination with Candida antarctica lipase A and an epoxide hydrolase from Agrobacterium radiobacter AD1 [87]. Kinetic resolution of halohydrins Halohydrin dehalogenase activity of whole cells or partially purified enzyme fractions has been employed for the production of optically active C 3 and C 4 fine chemical precursors [88–92]. Recently, processes employing halohydrin dehalogenases have received lots of attention due to their application in the synthesis of highly valuable, chiral precursors for pharmaceutical chemistry applications. As one of the first examples, optically active ( R)-3-chloro-1,2-propanediol has been produced from prochiral 1,3-dichloro-2-propanol [18]. Here, resting cells of Corynebacterium sp. strain N-1074 harbor an ( R)-specific halohydrin dehalogenase producing ( R)- epichlorohydrin. Subsequent epoxide hydrolysis yielded 93.7% final product ( R)-3-chloro- 1,2-propanediol with 83.8% enantiomeric excess (ee) [93]. Also purified HheB has been used to prepare ( R)-epichlorohydrin from prochiral 1,3-dichloro-2-propanol with an ee of 90% in the first reaction minutes [25]. A second example employed another B-type enzyme to produce even higher ee of >95% and yielding 10.7% ( R)-epichlorohydrin by addition of bromide to the reaction. After optimizing the pH of the reaction, yields of ( R)-epichlorohydrin could be increased to 63% with 89.5% ee. Furthermore, it is possible to obtain the opposite (S)-enantiomer from racemic substrates and additional halide via the transhalogenation activity of halohydrin dehalogenases. With (rac )-epibromohydrin and chloride as well as ( rac )-epichlorohydrin and bromide, ( S)- epichlorohydrin with 97.5% ee, 30% yield as well as >95% ee, 14.5% yield could be obtained [27]. In another approach, ( S)-epichlorohydrin with 99.5% ee was obtained directly from prochiral 1,3-dichloropropanol using whole cells carrying an ( S)-selective enzyme [20]. Either (R)- or ( S)-3-chloro-l,2-propanediol could be produced from racemic 3-chroro-1,2-



%
'  
propanediol using Pseudomonas sp. or Alcaligenes sp. with ee of 99.5% and 99.4%, respectively [19]. Epoxide ring-opening with cyanide After the initial observation that HheA catalyzes the non-enantioselective cyanolysis of epichlorohydrin, 1,2-epoxypropane and 1,2-epoxybutane to produce the corresponding β- hydroxynitriles [30], a later paper showed that employing this enzyme for the desymmetrization of prochiral 1,3-dichloro-2-propanol yields 65.3% ( R)- γ-chloro- β- hydroxybutyronitrile with 95.2% ee [31]. The formation of β-hydroxynitriles with high ee was subsequently found in cyanolysis reactions with various aliphatic epoxides by HheC [43, 45, 79]. Also derivatives of 2-phenyl- or 2-phenoxy-substituted epoxides were transformed to the corresponding β-hydroxynitriles [79, 94]. The presence of the additional substituent causes a very high enantioselectivity (E > 200), which can be used for producing enantiopure tertiary alcohols (see below). Statin side chain Atorvastatin belongs to the molecule class of statins which are inhibitors of 3-hydroxy- 3-methylglutaryl-coenzyme A reductase. This enzyme catalyzes an early rate-limiting step in cholesterol biosynthesis. Inhibiting this key enzyme leads to reduced low density lipoprotein (LDL) cholesterol levels. LDL levels are directly correlated to mortality rates in coronary heart diseases [67] making Lipitor (atorvastatin calcium) to one of the best selling drugs in the world with annual sales of more than US$10 billion in 2010 [48]. Atorvastatin has two chiral centers and its application as a drug requires >99.5% ee and >99% diasteromeric excess [95]. During its synthesis, chirality of the second stereo center is controlled by the chirality of the side chain precursor, ethyl ( R)-4-cyano-3-hydroxybutyrate, by diastereomeric induction [48]. Since chiral purity of atorvastatin depends solely on the chiral purity of the side chain precursor, extensive efforts were directed to develop biocatalytic processes for the production of this important chiral precursor also employing halohydrin dehalogenases [67]. Starting from racemic methyl 4-chloro-3-hydroxybutyrate in the presence of cyanide, mutant HheC-W249F has been employed to produce enantiomerically pure methyl ( S)-4- cyano-3-hydroxybutyrate and unreacted methyl ( S)-4-chloro-3-hydroxybutyrate with 96.8% and 95.2% ee, respectively [64]. Although this route does not lead to the valuable ( R)- β- hydroxynitrile precursor for atorvastatin synthesis, it provides an alternative route for the preparation of other optically pure chiral building blocks. Especially the biocatalytic route developed at Codexis Inc. (Redwood City, CA, USA) [96], has drawn lots of attention due to its application in a green-by-design process for the production of ethyl ( R)-4-cyano-3-hydroxybutyrate with a yearly demand exceeding 10 5 kg as pharmaceutical precursor for the atorvastatin side chain [48] ( Figure 7 ).



(

# $


 

    !                        

            
  
    
                                Figure 7. Biocatalytic production of ethyl ( R)-4-cyano-3-hydroxybutyrate from ethyl ( S)-4- chloro-3-hydroxybutyrate is achieved by using a HheC variant. The haloalcohol is produced from ethyl 4-chloro-3-ketobutyrate by reduction using a ketoreductase and replenishing of NADH through the oxidation of glucose. This process employs an engineered HheC variant which is able to produce ethyl ( R)-4- cyano-3-hydroxybutyrate with an ee higher than 99.9% [47]. In detail, an alcohol dehydrogenase variant from Candida magnoliae reduces prochiral 4-chloro-3-ketobutyrate ester to produce ( S)-4-chloro-3-hydroxybutyrate ester. In the following two-step-one-enzyme reaction, a highly evolved HheC variant converts this haloalcohol to the corresponding ( S)- epoxide which is subsequently transformed in the presence of cyanide to ( R)-4-cyano-3- hydroxybutyrate ester. Besides being commercially competitive with established chemical processes, this process also is an environmentally friendly alternative to established chemical routes. Epoxide-ring opening with azide The production of optically pure β-azidoalcohols through epoxide-ring opening with azide enables a chemoenzymatic route towards optically pure β-amino alcohols. These compounds are used in asymmetric synthesis as chiral ligands and auxiliaries [97] and in medicinal chemistry as β-blockers [98]. High enantioselectivities have been reported for various arylic- [42, 80, 99, 100] or aliphatic epoxides [76, 80]. Similar to the cyanolysis of 2,2-disubstituted epoxides, also the biocatalytic azidolysis of geminally methyl- or ethyl- substituted epoxides shows excellent enantioselectivities when using HheC [80, 100]. Furthermore, optically enriched tertiary β-aziridines have been obtained from chiral β- azidoalcohols produced via halohydrin dehalogenase-catalyzed epoxide azidolysis [100]. Nitrite-mediated epoxide hydrolysis Halohydrin dehalogenases also catalyze the epoxide-ring opening with the ambident nitrite nucleophile via the attack of nitrogen or oxygen atom [44]. For styrene oxide derivatives, the majority of the reaction occurs via oxygen-centered attack on the epoxide ring, yielding unstable nitrite ester that ultimately hydrolyzes to the corresponding diols. The minor, nitrogen-centered attack yields the corresponding β-nitroalcohols. Since here formation of nitroalcohols is negligible, the nitrite-mediated epoxide-ring opening by halohydrin



)
'  
dehalogenases can be seen as an alternative to epoxide hydrolysis [44]. Using HheC and mutants thereof, high enantioselectivities exceeding the ones obtained with other available epoxide hydrolases [91] have been observed for the conversion of arylic epoxides to the corresponding ( R)-configured diols [44]. Nevertheless, the preference for vicinal diol or β- nitroalcohol formation seems to be dependent on the reacting epoxide since conversions of epoxybutane and nitrite yielded roughly equal amounts of 1-nitro-2-butanol and 1,2- butanediol [45]. Recently, the nitrate-mediated hydrolysis of ( rac )-2-benzyloxymethyl-2-methyloxirane has been described producing enantiomerically pure (S)-2-benzyloxymethyl-2-methyloxirane (Figure 8 ). Further synthetic steps lead to ( S)-chromanemethanol, an intermediate for the total synthesis of enantiopure vitamin D [101].

Figure 8. Preparation of ( S)-2-benzyloxymethyl-2-methyloxirane from racemic epoxide using HheC and nitrite. The enantiopure epoxide is converted further to ( S)-chromanemethanol which is a chiral building block enantiopure vitamin D [101]. Other nucleophiles Other anions such as cyanate, thiocyanate, and formate have been reported to be accepted as nucleophiles in the halohydrin dehalogenase-catalyzed production of terminal ( R)- configured nucleophile adducts. Similar to the aforementioned nitrite-mediated epoxide-ring opening, also the nucleophilic attack of formate leads to unstable formate esters resulting in the formation of vicinal ( R)-diols [45]. Despite the ambident character of cyanate, epoxide-ring opening by this anion produces only β-isocyanates which spontaneously cyclize to oxazolidinones [81]. Although oxygen attack on the epoxide-ring could also lead to organic cyanates, these compounds isomerize chemically to the corresponding β-isocyanates. Formation of enantiopure ( R)-5-substituted-2- oxazolidonones has been observed for cyanate-mediated conversions of geminally substituted epoxides such as 1,2-epoxy-2-methylbutane [81]. The enantioselective preparation of oxazolidinones attracts lots of interest from the synthetic community due to their antibiotic activity against multi-drug resistant Gram-positive bacteria [102]. Actually, this class of antimicrobial agents has been the first one to be approved for clinical trials since 1972 [103] with estimated US sales of over US$ 695 million in 2012 (source: http://www.drugs.com/stats/top100/2012/sales). In contrast, conversions with thiocyanate do not lead to the corresponding oxazolidine-2- thione. Here, not the nitrogen but the sulphur attacks the epoxide-ring. The produced β- thiocyanates rearrange slowly under formation of inorganic cyanate to the corresponding



+

# $

thiiranes useful for fragrance preparations [45]. Tertiary alcohols Especially processes for the outstanding enantioselectivity of halohydrin dehalogenases with 2,2-disubstituted epoxides (E > 200) have been exploited for the production of optically pure tertiary alcohols. These conversions provide access to optically pure tertiary - azidoalcohols and -hydroxynitriles [79, 80, 100] as well as optically pure diols via the nitrite- mediated hydrolysis [101]. These compounds are difficult to obtain by non-biocatalytic methods [104–106]. Cascade reactions The concept of enzymatic cascades comprising halohydrin dehalogenases has also been applied to the production of optically pure fine chemical precursors with enantiomeric excesses higher than 99%. Initially only the natural activity of halohydrin dehalogenases has been applied to produce chiral halohydrins, epoxides, and diols in combination with an epoxide hydrolase [29]. To improve yields inherent to kinetic resolutions, processes according to dynamic kinetic resolution principles have been developed by employing enzyme-catalyzed racemization of substrates [76] or by chemical racemization via an iridium-based catalyst [107]. Recently, enzymatic cascades have been applied to produce optically pure halohydrins and epoxides as well as -azidoalcohols and -hydroxynitriles starting from prochiral α- chloroketones. The first step in these processes employs a stereoselective alcohol dehydrogenase for the introduction of chirality via enantioselective reduction of α- chloroketones to the corresponding chiral halohydrins. Then, in a second step, halohydrin dehalogenases convert the halohydrin to the corresponding optically pure epoxide. Enantiocomplementary synthesis of the desired ( S)- or ( R)-product can be accomplished by choosing an appropriate ( S)- or ( R)-selective alcohol dehydrogenase, respectively. Kroutil and coworkers have been able to produce either epoxide enantiomer with >99% ee from aliphatic and arylic α-chloroketones [108]. Addition of an alternative nucleophile such as azide or cyanide enables the optically pure synthesis of -azidoalcohols by the same enzyme [94] and -hydroxynitriles [47, 48], respectively. Lately, this multicatalytic process has been further developed by the introduction of a chemically-catalyzed process into the same reaction vessel ( Figure 9 ). Haloketones are transformed into enantiopure β-azidoalcohols by engineered cells containing an alcohol dehydrogenase and haloalcohol dehalogenase. Then β-azidoalcohols react with terminal acetylenes in a Huisgen-type 1,3-dipolar cycloaddition to form β-hydroxytriazoles [109].

Figure 9. Combination of halohydrin-dehalogenase producing designer cells and ‘click’ reaction in the one-pot-four-step preparation of enantiopure β-hydroxytriazoles [109].



,
'  

4
 

Enhancing enzyme stability. Studies of Janssen and coworkers showed that the activity of homotetrameric HheC is lost under oxidizing conditions due to the formation of monomeric enzyme with intramolecular disulfide bonds [34]. Replacement of all four cysteine residues with an alanine or serine identified the mutant C153S with greatly enhanced stability rendering the enzyme more useful for biocatalytic applications. Modifying stereoselectivity. Replacement of all four HheC tryptophans with a phenylalanine residues identified W139 and W249 as key residues for altering the enzyme’s enantioselectivity [36]. Mutation W139F leads to a HheC variant with complete loss of enantioselectivity due to similar k cat /K M values for both enantiomers of p-nitrophenyl-2-bromoethanol. This loss of enantioselectivity is mostly caused by opposite effects on the K M for both enantiomers. The wild-type HheC enzyme shows preferred binding of the ( R)- over the ( S)-enantiomer. A similar effect, mostly affecting k cat values, has been described for mutation W249F leading to an improved discrimination for the ( R)-enantiomer by a factor of six by increasing the kcat for the ( R)- while reducing the k cat for the ( S)-enantiomer. The enhancement of enantioselectivity by mutation W249F has been applied to ring-opening reactions with halides [36, 64, 107], cyanide [64], and nitrite [44] using arylic [44] and aliphatic epoxides [64]. Enhancing the catalytic rate.

Whereas the W139F substitution shows at maximum a less than two-fold increase of k cat for p-nitrophenyl-2-bromoethanol, a site-saturation library at this position identified a W139C mutation resulting in an enzyme variant exhibiting five-fold elevated k cat values [110]. Modifications in the anion binding pocket have also been successful to improve catalytic properties of HheC. Mutation Y187F disrupts a possible hydrogen bond thus enabling higher anion exchange rates previously found to be the slowest step of the kinetic mechanism [37].

In consequence Y187F exhibits a four-fold higher k cat value for the epoxide-ring formation from ( R)- p-nitrophenyl-2-bromoethanol. Highly evolved halohydrin dehalogenases. As mentioned above, HheC has been subjected to a thorough directed evolution study to tailor the biocatalyst to process demands defined by the industrial scale production of the statin side chain precursor ethyl ( R)-4-cyano-3-hydroxybutyrate [47]. Various techniques were applied to introduce rational and random mutations into the enzyme. The directed evolution program addressed the increase of temperature stability, increase of reaction rates, tolerance to high substrate and product loads, and high chiral purity of the desired product. After 18 consecutive rounds of diversity generation and screening, protein sequence activity relationships (ProSAR) were derived from the screening of more than 585,000 clones. Finally, after analyzing the data for all beneficial mutations, a final variant with at least 35 mutations out of a pool of 43 substitutions ( Figure 10 ) produced ethyl ( R)-4-cyano-3-hydroxybutyrate with a 4,000-fold increase in volumetric productivity thus allowing an economically viable route for the biosynthetic production of this important statin side chain building block. All of the deposited sequences [111] contain the previously described C153S substitution rendering the enzyme more stable whereas none of the sequences carried the W249F mutation enhancing enantioselectivity. The latter finding is not surprising since this mutation does



.

# $

promote enantioselectivity opposite to the target molecule. The residues around the active site are heavily mutated. About 40% of the residues within 7.5 Å of the active site were assessed, including four out the nine non-catalytic substrate binding residues. A cluster of amino acids lining the substrate access channel seems to have been targeted extensively, suggesting that this region may be very important for the activity change. Prospected process variants also carry mutations of several more distant residues, of which the individual effect cannot be explained without assessing the biochemical properties of the enzyme variants. Based on these sequence data, one can estimate that HheC could be mutated at more than 100 positions and that the final variant is changed at 35 of its 254 residues [47]. This is a surprisingly high number of mutations for a directed evolution study and shows the extraordinary tolerance of HheC towards amino acid substitutions.

Figure 10. Location of 43 amino acid substitutions present in the final population of highly evolved HheC variants generated by Fox and co-workers [47]. The mutated residues are shown in red on the cartoon display of tetrameric HheC (green) with bound ( R)- p-nitrostyrene oxide (blue) in each of the four active sites (PDB 1ZMT).



0
'  

#   

 
Halohydrin dehalogenases are attractive biocatalysts since they can catalyze a range of chemically diverse reactions. Moreover, they are valuable tools for organic synthesis due to their ability to form various bonds such as carbon-halogen, carbon-nitrogen, carbon-carbon, carbon-sulfur and carbon-oxygen bonds with high enantio- and regioselectivities. Historically, this class of enzymes has been identified in organisms applied for the removal of organohalogens from contaminated waste streams. Consequently, halohydrin dehalogenase activities from these organisms have been exploited for the bioremediation of xenobiotics. Apart from those environmental applications, such enzymatic organohalogen removal has been described in the field of food applications [86, 87] or paper coatings for food applications [46]. Despite the high volumes of such food technology processes, the highest potential for biocatalytic processes lies in the preparation of fine chemicals or their precursors in pharma- and nutraceutical applications [112–114]. To date, several applications of this versatile enzyme class have been described to produce chiral C3 and C4 precursors for valuable pharmaceuticals such as the former blockbuster drug atorvastatin (Lipitor) [48] or the nutraceutical vitamin E [101]. Furthermore, halohydrin dehalogenases provide biocatalytic access to a wide range of -substituted alcohols enriching the synthetic chemist’s toolbox [45]. Only a limited number of halohydrin dehalogenases is readily available to scientific and industrial researchers at the moment. Our experience indicates that these enzymes can be easily produced in E. coli at a satisfactory level using standard recombinant expression systems. In the future, novel halohydrin dehalogenase enzymes have to be obtained to expand the knowledge and versatility of this interesting enzyme class. The limited number of available enzymes is partially due to the fact that this enzymatic activity is scarcely distributed in nature and has only been discovered in few bacterial strains. The natural diversity may be larger than currently known, and several enrichment cultures with typical HHDH activity have been described without enzyme sequence information reported. The observed activities in these cultures may represent enzymes with valuable properties such as higher activities [28, 84] or enantioselectivities opposite to the widely investigated HheC [19–21]. In view of the low abundance of halohydrin dehalogenase genes, which is also evident from blast sequence similarity searches of genomic databases, classical enrichment cultures or methods using sequence- or activity-based pre-enrichment before library construction may be more suitable than standard metagenomic approaches for the discovery of new halohydrin dehalogenases. The aforementioned limitations in terms of available sequence space and thus biocatalytic diversity and applicability can be addressed by directed evolution [115–117]. Methods such as rational design, error-prone PCR or gene shuffling have successfully provided halohydrin dehalogenase variants with improved target activities. The capacity of HheC to tolerate more than 13% amino acid substitutions on the sequence level [47] is remarkable and makes it a promising candidate for further biocatalyst evolution. Currently, Codexis Inc. provides halohydrin dehalogenase panels with different enzyme variants in a 96-well plate format. This gives access to a collection from which a mutant with the desired activity can be retrieved by screening. Halohydrin dehalogenases already possess broad promiscuous activity for the epoxide- ring opening with many nucleophiles. Enhancement of promiscuous catalytic properties can also be addressed by directed evolution of biocatalysts expanding the usefulness of enzymes



1

# $

for synthetic applications [118]. A further expansion of the list of accepted nucleophiles will be a scientifically challenging but rewarding task.



%2