Part 6 – Multi-step enzymatic reactions

Introduction

Most chemists are inherently lazy and are always looking for short-cuts and ‘quick and dirty’ by doing as much as possible in one pot or by skipping product isolation after each step. This mindset is now part of what is called “process intensification”, a push toward higher efficiency in chemical processes.  I found a nice definition/explanation on the web[1]:

The objective of process intensification is to enable a process to achieve its maximal chemical or catalytic performance asymptote free of extraneous transport resistances and dispersive statistical phenomena, using minimal physical and financial resources. Crucial aspects of process intensification are well-defined operating conditions and its integrative approach, i.e. considering the overall process objectives rather than the isolated performance of individual unit operations. Due to tight operating windows, loss of degrees of freedom and nonlinear behaviour, integrated processes pose challenging control problems.

This blog issue will deal with sequential, multistep, reactions that require ‘action’, i.e. they do not happen spontaneously. This opposed to ‘cascade reactions’ or ‘domino reactions’ that require an (enzyme-catalyzed) initiating step that sets off a sequence of events. I once heard Nobel laureate Kyriacos Costa “KC” Nicolaou speak and I remember the most fantastic reaction schemes that I’m sure will impress you too and I am happy to include a reference to a review on cascades by his hand that also has some of his work.[2] Since most people use ‘cascades’ rather loosely, I will stick to that terminology as well.

Microorganisms are really good at cascades. Metabolic pathways tend to have high efficiency due to feedback controls that maintain a low concentration of irrelevant intermediates, all under relatively constant reaction conditions[3]. Also, if reactions are not compatible, the organism can perform them in different compartments (organelles). All in all pretty neat and worth trying to mimic in the lab!

I hope the following examples will give some  background and maybe some good ideas.

One enzyme, multiple substrates

Deoxyribose-phosphate aldolase (DERA)

This enzyme stitches together three molecules of aldehyde: chloroacetaldehyde to acetaldehyde to acetaldehyde. This cascade is the basis for DSM’s Atorvastatin-intermediate synthesis.[4]

Figure 1: Very chemo- and enantioselective coupling of several aldehydes by DERA. 

Polyketide synthases (PKS) and fatty acid biosynthesis (FAS)

The enzyme or enzyme complex stitches together a number of acetyl-CoA and malonyl-CoA molecules (for polyketides, [5]) (for fatty acids, [6]) to achieve the desired chain length.

Figure 2: The polyketide synthase enzyme complex couples an exact number of C2 fragments.

Haloalcohol dehalogenase (HHDH)

This enzyme can catalyze the reversible ring closure of a haloalcohol to give an epoxide. It can also perform ring-opening reactions with non-halogen nucleophiles. This is the basis for Codexis’ route to an Atorvastatin-intermediate.[7]

Figure 3: Ring closure and opening, catalyzed by HHDH. The epoxide is not isolated.

Cytochrome P450 oxidations

Cytochrome P450 enzymes can oxidize substrates leading to hydroxylation and other metabolism. In the case of multiple oxidations, it may happen that the product has not left the active site in-between reactions although in most cases it is assumed that it did. Measuring the time course of dihydroxylated product will show whether it follows the formation of mono-hydroxylated product or is simultaneous. In the metabolism of buspirone, changing the reaction conditions led to less di-hydroxylation so that would indicate a release of the product and a true two-step process.[8]

Figure 4: Cytochrome P450 oxidation of buspirone. Dihydroxylation can follow two pathways, each catalyzed by the same enzyme.

Multiple enzymes, one substrate

If you have a molecule with two functional groups that you both want to convert with enzymes, why not do all the enzymes in one pot and convert both groups? Although in principle this seems like a rather obvious thing to try, I’ll explain that this is deceptive. I think most people got it since I could not really find any good examples for this in half an hour. There is one report of reduction and hydrolysis of keto-esters but it does not specify whether ester hydrolysis is enzymatically catalyzed[9]. The resulting molecules are all reduced, leading me to believe that the sequence of events does not make a difference.

Figure 5: Two functional groups, two enzymes. What can be simpler?

Theoretically, this concept can be applied to any molecule with different functional groups for the respective enzymes to work on. However, there is a high risk of channeling the reactions towards a slow-reacting intermediate and this concept only works well when all four reactions (in the example above) proceed at the same rate. It is unlikely that the ester will have the same k_cat and K_m as the acid, for instance. Therefore there is a good chance of having a severe bottleneck and a two-step approach is almost always better.

Let’s explain this by a simplified example. Assume we have two pathways from S --> P, with intermediates A and B. Further, assume that the rates are as listed in the figure, e.g. S --> A proceeds at 100/hr and the conversion of A --> P is much slower, at 5/hr. The upper route proceeds at an overall rate of 5/hr and thus intermediate A accumulates. Similarly, the lower route is at 25/hr and because the second step is so much faster, B has only a very low transient concentration.

When we measure the formation of P through both routes, it occurs at 30/hr. In the beginning of the process everything seems alright: the combination of enzymes gives a higher rate than one could get with the individual enzyme (ignoring inhibition of A or P on the conversion of S --> B in this simplification). Now, let us assume we have 500 molecules of S. In 4 hours, all the S is gone and we have sent 400 molecules through the upper route to A and 100 to B. The B is immediately converted to P as soon as it is formed, but A --> P needs 76 more hrs to complete (380/5).

Now consider using one enzyme first. S --> B can be done in 20 hours, and B --> P takes another 10 hrs, giving an overall process duration of 30 hrs. This is almost three times faster! As an added bonus, it now makes sense to use even more enzyme to convert B --> P, decreasing overall process time even further.

Convinced that there is no efficiency gain when using two enzymes simultaneously?

Figure 6: Simplified numerical example for two pathways. In practice, the rates are not constant and will slow down once the concentration approaches the K_m. Also, depending on the k_cat/K_m for A and S, the rate will not be 25+5, but lower than that.

Multiple enzymes, reaction sequence/pathway

For this concept to be elegant, I would limit this to cases where all the enzymes are present from the start and do not interfere with each other. Just a few examples here and many more in literature if you look for the work of Wolfgang Kroutil and Kurt Faber, both at Graz University, who have been at this for a while.

Nitrilase --> Penicillin acylase

My good friends from Delft have demonstrated the use of a nitrilase and penicillin acylase in antibiotic synthesis[10].

Figure 7: Some nitrilases are stuck in the amide stage and act more like a nitrile hydratase. This property is used in this strategy.

Phytase --> GPO/Catalase --> Aldolase --> Phytase

Conversion of glycerol and an aldehyde (R-CHO) to non-natural carbohydrates, also from Delft [11].

Figure 8: The phytase in step 1 is the same as in step 4 but requires a pH-shift.

KRED --> FDH, or KRED --> GHD, or KRED --> KRED

Technically, all the redox cofactor recycle systems would fall in this category. There are de-symmetrization oxidation-reduction cascades (cycles) that are very useful, as reviewed by folks from the University of Graz. [12]

Figure 9: KRED-FDH redox cycle.

KRED --> HNL --> NR

Jörg Schrittwieser is currently working on developing this sequence that includes a hydroxynitrile lyase and a new nitrile reductase enzyme while in Delft.[13] For the moment, chemical aerobic oxidation is considered but an alcohol oxidase (flavin dependent, with O₂) or a ketoreductase (NAD or NADP dependent) could be used too.

Figure 10: Oxidation – Reduction should use the same cofactor, in this case NADP⁺. There will be an imbalance, though, since the nitrile reduction requires 2 equivalents of hydride.

The challenge

Below is a partly hypothetical sequence that involves nine reaction steps that should not interfere with each other. Parts of this have been proven as shown above and in literature[14] and some of this idea was already posed a while back [15]. I challenge you to email me (devries.erikjan@gmail.com) a longer sequence (but don’t come with “repeat steps 6-9”) and I’ll compile them and will do a follow-up posting with your ideas. Let’s not worry about economic feasibility or technicalities like enzyme selectivity or optimal reaction conditions etcetera since we can change that easily by mutagenesis.

Figure 11: Now, this starts to look like a real pathway.

References

[1] http://www.bci.tu-dortmund.de/en/research/process-intensification

[2] KC Nicolaou, DJ Edmonds, and PG Bulger. Cascade reactions in total synthesis. Angew. Chem. Int. Ed. 2006, 45, pp 7134-7186. (http://medchem.rutgers.edu/mc504/pdfs/cascade_reactions.pdf) and for a review on enzyme-triggered cascades: SF Mayer, W Kroutil, and K Faber. Enzyme-initiated domino (cascade) reactions. Chem. Soc. Rev., 2001, 30, pp 332-339.

 [3] This is probably the best use of enzyme product inhibition that you can imagine. Other feedback controls are regulated cofactor regeneration, or energy/ATP use. No big pH swings are allowed, obviously all reactions occur at the same temperature. http://en.wikipedia.org/wiki/Biochemical_cascade

[4] S Jennewein, M Schürmann, M Wolberg, I Hilker, R Luiten, M Wubbolts, D Mink. Directed evolution of an industrial biocatalyst: 2-deoxy-D-ribose 5-phosphate aldolase. Biotechnol. J. 2006, 1 (5), pp 537-548. WA Greenberg, A Varvak, SR Hanson, K Wong, H Huang, P Chen, and MJ Burk. Development of an efficient, scalable, aldolase-catalyzed process for enantioselective synthesis of statin intermediates. PNAS 2004, 101 (16), pp 5788-5793. WO2007137816 Process for the preparation of epoxide intermediates for pharmaceutical compounds such as statins. D Mink, JH lutje Spelberg, EJ de Vries, 2007. WO2007128469 Process for the preparation of enantiomerically enriched nitriles. D Mink, JH lutje Spelberg, EJ de Vries, 2007.

And while I was researching for this blog post, I stumbled across the nice thesis by Angela Kinnell from Leeds University: Development of an Organo- and Enzyme-Catalysed Cascade Reaction, 2010. (http://etheses.whiterose.ac.uk/1132/1/Angela_Kinnell_Thesis_Final.pdf).

[5] Although technically this is sometimes one enzyme and sometimes a rather a complex multi-enzyme system that is forming a cluster: http://en.wikipedia.org/wiki/Polyketide_synthase

[6] The FAS principle is very similar to PKS: http://en.wikipedia.org/wiki/Fatty_acid_synthesis

[7] SK Ma, J Gruber, SC Davis, L Newman, D Gray, A Wang, J Grate,  GW Huisman  and RA Sheldon. A green-by-design biocatalytic process for atorvastatin intermediate. Green Chem., 2010, 12, pp 81-86.

[8] M Zhu, W Zhao, H Jimenez, D Zhang, S Yeola, R Dai, N Vachharajani, and J Mitroka. Cytochrome P450 3A- mediated metabolism of buspirone in human liver microsomes. Drug Metabolism and Disposition 2005, 33 (4), pp 500-507. (http://dmd.aspetjournals.org/content/33/4/500.full)

[9] NA Salvi, S Chattopadhyay. Asymmetric reduction of 3-aryl-3-keto esters using Rhizopus species. Bioorganic & Medicinal Chemistry 2006, 14 (14), pp 4918–4922.

[10] MA Wegman, LM van Langen, F van Rantwijk, RA Sheldon. A two-step, one-pot enzymatic synthesis of cephalexin from D-phenylglycine nitrile. Biotechnology and Bioengineering 2002, 79 (3), pp 356–361.

[11] R Schoevaart, F van Rantwijk, RA Sheldon. A four-step enzymatic cascade for the one-pot synthesis of non-natural carbohydrates from glycerol. J Org Chem. 2000, 65 (21), pp 6940-3.

[12] JH Schrittwieser, J Sattler, V Resch, FG Mutti, and W Kroutil. Recent biocatalytic oxidation–reduction cascades. Curr. Opin. Chem. Biol., 2011, 15 (2), pp 249–256. (http://www.sciencedirect.com/science/article/pii/S1367593110001869)

[13] JH Schrittwieser. Asymmetric Cascade Reactions Incorporating Hydroxynitrile-Lyase Catalysed C–C Bond Formation. Go to http://www.tnw.tudelft.nl/en/ and search for ‘cascade schrittwieser’.

[14] JH Schrittwieser, I Lavandera, B Seisser, B Mautner, and W Kroutil. Biocatalytic Cascade for the Synthesis of Enantiopure β-Azidoalcohols and β-Hydroxynitriles. Eur. J. Org. Chem., 2009, 14, pp 2293–2298. And: B Seisser, I Lavandera, K Faber, JH lutje Spelberg, and W Kroutil. Stereo-Complementary Two-Step Cascades Using a Two-Enzyme System Leading to Enantiopure Epoxides. Adv. Synth. & Catal. 2007, 349 (8-9), pp 1399–1404.

 [15] JH lutje Spelberg, and EJ de Vries. Multi-step biocatalytic synthesis of epoxides & derivatives.  Specialty chemicals online News Article ID: 11764, 23 October 2006. I can email you a copy since it is not available online anymore.

Part 5 - Chirality, enantiomers, and enzyme enantioselectivity

Dear readers,

In the first month I had more than 500 page views from all over the World. I thank you all for your interest and hope you'll come back every now and then to enjoy some more.

Introduction

People seem to like my car analogies, so here comes another one. Enantiomers are mirror images of each other. In Figure 1A, you can see two cars. One is the left-hand drive (LHD) version and the other one is the opposite, right-hand drive (RHD). So, as long as we can tell which is the front of the car, by looking at where the steering wheel is placed we can say whether it is the English version (RHD) or the continental (LHD). Aerodynamic design conventions make it rather easy to determine what the front end of a car is, except for weird and failed exercises like in Figure 1B. This Zündapp model aptly names Janus shown here is an (almost) symmetrical car and consumers could not make heads or tails out of it. 

Figure 1: Cars as enantiomers[1].

How an enzyme treats enantiomers is like how a Brit would drive a continental rental car: the stick shift is on the wrong side and if feels uncomfortable enough to have to drive slower than your regular mainland European citizen. Of course in this analogy, the Brit is the non-preferred substrate, the stick is the catalytically active amino acid and the speed is the catalytic rate. Changing the enantioselectivity of an enzyme by protein engineering is like converting a car from LHD to RHD.

Are there gradations in chirality; is one molecule more chiral than another?[2]

A carbon atom with four different substituents is (at least somewhat) tetrahedral[3] and chiral since the groups can be arranged on the central atom in two mirror-image fashions.[4] In Figure 2A, I show three different chiral molecules; they all have four different substituents and I think most people will be inclined to say that molecule 1 is the ‘most chiral’ since the substituents differ the most, close to the central C-atom. In Figure 2B, an example where the discriminator (chlorine) is far away, yet the carbon atom has four different groups and the molecule is chiral. Valiant attempts at developing mathematical tools to quantify chirality have been made. The jury is still out as far as I’m concerned and I have listed two references to get you started [2]. Two thoughts occur to me: how do you make them and how do you analyze them? This is beyond the scope of this story, though.

Figure 2: A1, chloroethylbenzene; A2, deuterium (²H) and tritium (³H) labeled groups leading to chirality; A3, tetra-alkyl methane; B, (S)-4-chloro-diphenylmethanol.

Enzymes and enantioselectivity

Traditionally, an enzyme has been regarded as a lock to which only one of the enantiomeric keys would fit. Later, this model was modified to incorporate enzyme flexibility and became a glove that only fits one hand. I will use a few examples to show that enantioselectivity is even more subtle than that.

Example 1: Ketoreductase (KRED) reduction of ketothiolane (Figure 3A) and 4-chloro-benzophenone (Figure 3B).

A KRED enzyme uses so-called ‘three point interaction’ to orient the substrate in the active site. In this case, the ketone oxygen forms a hydrogen bond to a group in the enzyme that always locks it into place. Then, there are two sub-sites in the active site that need to accommodate the two groups flanking the ketone. Once the substrate molecule is bound in this preferred way (Figure 3A/B), the enzyme makes sure that the reducing hydride always comes from, say, above the plane of the paper/screen. In this way, only one enantiomer of the alcohol product is formed.

There are two 'flipped' alternative binding modes that can cause imperfect enantioselectivity[5] and they are shown in Figure 3C. Since the reduction of ketothiolane is rather temperature dependent, it is likely that enzyme flexibility is a factor that allows the substrate to occupy the alternative binding mode. At higher temperatures, the enzyme cannot distinguish what is ‘left’ and ‘right’. In the case of 4-chloro-benzophenone, it is more likely that the hydrogen bond is broken and that the ketone group exposes the other side (or ‘face’), leading to the opposite enantiomer.

Should one want to improve the selectivity of the KRED enzyme, in the case of ketothiolane one would want to focus on improving rigidity of the enzyme active site, whereas in the benzophenone case it seems more rewarding to strengthen the hydrogen bond to keep the ketone properly oriented. Thus, a detailed understanding of the root cause of lack of selectivity can guide a mutagenesis approach.

            By the way, there are KREDs that achieve >99% e.e. of these alcohols so it can be done! [6]

Figure 3: Some very basic active site models (A and B) and two alternative binding modes that lead to the opposite product-enantiomer (C).

Example 2: Epoxide hydrolysis by epoxide hydrolases.

Enantioselective hydrolysis of styrene oxide to phenylethane diol and its catalytic mechanism has been described by several groups. [7] A crude active site model is shown in Figure 4. A simplistic explanation for enantioselectivity in this case can be that with the non-preferred enantiomer, the catalytically active aspartate ‘cannot reach’ to the epoxide ring (indicated by the dotted line in Figure 4B) and no reaction takes place.

Figure 4: Side view (A) and top view (B). Note that only the enzyme-substrate alkylation step is shown because it is this step that determines the stereochemical outcome. In the following step, the covalent enzyme-substrate intermediate is simply hydrolyzed to retrieve the enzyme and release the diol.

This model can also explain a rather nice phenomenon that is often observed in epoxide hydrolysis: enantioconvergence. This happens when one substrate-enantiomer reacts on the terminal β-position with retention of configuration and the other substrate-enantiomer moves out of the binding pocket just a bit to allow reaction at the benzylic α-position with inversion of configuration. The result is going from a racemic epoxide to 100% enantiomeric excess (e.e.) of the diol in 100% yield.

Figure 5: Enantioconvergence rationalized.

X-Ray crystallography and computational chemistry can certainly aid in finding favorable binding modes. An example that I am familiar with: crystal structure elucidation of haloalcohol dehalogenases enzymes with substrate enantiomers in the active site showed how non-productive binding of the ‘wrong enantiomer’ occurred in the crystal. [8] I would also like to point to the impressive body of quantum chemical computational work done by Kathrin Hopmann on epoxide hydrolases and haloalcohol dehalogenases that present compelling evidence for the origins of the regioselectivity. [9]

Now, these simple cartoon models are exactly that: simple. They ignore kinetic parameters (rates and affinity constants) so the next two examples will show how kinetics can determine enantioselectivity.

Example 3: Epoxide hydrolysis by epoxide hydrolases (yes, again).

In Figure 6, two reactions catalyzed by EH from Agrobacterium radiobacter are shown. [10] In panel A, we see a ‘normal’ kinetic behavior: the most reactive enantiomer (R)-Cl-SO's concentration decreases rapidly. After 20 minutes, the remaining (S)-enantiomer is hydrolyzed at a slower pace. In panel B, we see that a slight modification to the substrate structure gives a completely different picture. In this case, the enantiomer that binds best is hydrolyzed preferentially but at a slow pace. After 50 minutes, when (R)-SO is gone, the second ‘gets its chance to bind’ and is hydrolyzed very rapidly and follows suit in just 10 minutes more. Well, you all know how I like comics and cartoons, but here a cartoon model is clearly inadequate at explaining this phenomenon.

Figure 6: Hydrolysis of 4-chlorostyrene oxide (A, Cl-SO) and styrene oxide (B, SO).

Example 4: Switch of enantioselectivity of Subtilisin Carlsberg catalyzed transesterifications in organic solvent due to additives.

The effect of co-solvents and additives on the reaction rates of (L) and (D) phenylalanine derivatives in cyclohexane was studied (Figures 7 and 8, [11]). It turned out that the effect on the reaction rates of the (L)-enantiomer was different than on the (D)-enantiomer. This spurred a few years of debate between Alex Klibanov and Jaap Broos about water activity, enzyme flexibility and conformational change. Anyway, the sum of the effects is that in some cases the enzyme is S-selective and in some cases R-selective for the same reaction, without mutagenesis. I remember a Friday afternoon discussion I had with Jaap about the rate limiting step in this reaction. It could be that with some additives the acylation is rate limiting whereas in the presence of other additives de-acylation is rate limiting. If these two reaction steps proceed with different enantioselectivity, this could give an opposite stereochemical outcome. A detailed study into the kinetics of each individual reaction step should give the answer but I don’t think that was ever done.

Figure 7: Subtilisin catalyzed transesterification.

Figure 8: Effect of additives on the reaction rates (V₀) of the individual enantiomers (V₀(L) and V₀(D)). Additives used were entry 1: ethanol; 2: acetonitrile; 3: 1-propanol; 4: propionitrile; 5: butyronitrile; 6: no additive; 7: tert-butanol; 8: 2-methyl-2-pentanol; 9: 3-methyl-3-pentanol; 10: 2-methyl-2-butanol.

Conclusions

Successfully predicting whether an enzyme will be R-selective or S-selective in a certain reaction is very hard to do. Even the best crystal structure with the highest resolution, into which substrate or products are soaked or computer docked will only give an indication. Granted, computational chemistry has come a long way from the days of “you tell me the answer and I’ll calculate it for you”[12] but I challenge this readership to show me a case where a computational prediction of enantioselectivity was accurate (sign and magnitude). I think I have shown that even a rather simple extrapolation of results from one molecule to an analog or from one reaction condition to another is tricky.

Fortunately, it is not difficult to just measure the enantioselectivity. (Semi-) random mutagenesis is very powerful and rather agnostic of kinetic parameters and has been used quite often to improve or even really switch enantioselectivity.

References

[1] A: The air intake on the rear fender indicates that this is a rear-engine car (think Porsche or VW bugs) so the placement of the engine is not indicative of where the front is. Likewise with which wheels are driven by the transmission (FWD vs. RWD). The best indicator is looking at which wheels move when you turn the steering wheel (unless you’re looking at a fork lift) and where the mirrors are positioned. B: The Janus’s fenders have a cut-out to allow wheels to turn corners, so the right hand side of the car in the picture is the front. And while we’re at it, the Janus has an internal mirror plane and could be called a ‘micro car’ as well as a ‘meso car’.           http://en.wikipedia.org/wiki/Zündapp_Janus

[2] PW Fowler. Quantification of chirality: attempting the impossible. Symmetry: Culture and Science 2005, 16(4), pp 321-334 (http://symmetry.hu/content/fowler-05-4.pdf). And then have a look at this nice presentation from N Duncan-Gould. Recent advances in quantifying chirality. 2008. http://www.scs.illinois.edu/denmark/presentations/2008/gm-2008-12_9.pdf

[3] Winner of the first-ever Nobel prize for Chemistry, oddly enough not for the tetrahedral carbon configuration. http://en.wikipedia.org/wiki/Jacobus_Henricus_van_'t_Hoff and look at http://www.nobelprize.org/nobel_prizes/chemistry/laureates/1901/

And look this paper full of definitions: K Mislow, and J Siegel. Stereoisomerism and Local Chirality. J. Am. Chem. Soc. 1984, 106 (11) pp 3319-3328 (https://www.uzh.ch/oci/efiles/OCVII/ja00323a043.pdf)

[4] I’d like to take this opportunity to highlight the work of Dr. Wolter ten Hoeve who in his thesis with the best title ever (“The long and winding road to planar carbon”) describes the synthesis of molecule 3 from Figure 2A in 80% enantiomeric excess, as well as efforts at making a carbon atom with four substituents that is mostly flat. Check out the (Dutch) summary: http://dissertations.ub.rug.nl/FILES/faculties/science/1979/w.ten.hove/Planar_carbon0001.PDF.

Ten Hoeve worked in the group of stereochemistry pioneer Prof Hans Wynberg, an extraordinary man that I had the pleasure to meet frequently. You have to check out these links to get a glimmer: http://en.wikipedia.org/wiki/Frederick_Mayer_(spy)#Hans_Wynberg and the obituary that does him justice: http://www.grinnell.edu/offices/communications/magazine/extras/hans-wynberg

[5] I know the term enantioselectivity is wrong, but most people call it like this.

[6] J Liang, E Mundorff, R Voladri, S Jenne, L Gilson, A Conway, A Krebber , J Wong, G Huisman, S Truesdell, and J Lalonde. Highly Enantioselective Reduction of a Small Heterocyclic Ketone: Biocatalytic Reduction of Tetrahydrothiophene-3-one to the Corresponding (R)-Alcohol. Org. Process Res. Dev., 2010, 14 (1), pp 188–192; MD Truppo, D Pollard, and P Devine. Enzyme-Catalyzed Enantioselective Diaryl Ketone Reductions. Org. Lett., 2007, 9 (2), pp 335–338.

[7] A few come to mind from the EPOX project: Dick Janssen, Roland Furstoss, Michael (Ernie) Arand, Manfred Reetz, but there are many, many more. For the main catalytic mechanisms, just Google “catalytic mechanism epoxide hydrolase”.

[8] RM de Jong (LI: nl.linkedin.com/pub/rené-de-jong/29/297/3b2), JJW Tiesinga, A Villa, LX Tang, DB Janssen, and BW Dijkstra. Structural Basis for the Enantioselectivity of an Epoxide Ring Opening Reaction Catalyzed by Halo Alcohol Dehalogenase HheC. J. Am. Chem. Soc., 2005, 127 (38), pp 13338-13343. http://gbb.eldoc.ub.rug.nl/FILES/root/2005/JAmChemSocdeJong/2005JAmChemSocdeJong.pdf

[9] KH Hopmann (LI: www.linkedin.com/pub/kathrin-hopmann/9/966/a12). Nitrile Hydratases and Epoxide-Transforming Enzymes: Quantum Chemical Modeling of Reaction Mechanism and Selectivities. KTH, Stockholm, Doctoral thesis, 2008. 

http://kth.diva-portal.org/smash/record.jsf?pid=diva2:13342

[10] JH lutje Spelberg, R Rink, RM Kellogg, and DB Janssen. Enantioselectivity of a recombinant epoxide hydrolase from Agrobacterium radiobacter. Tetrahedron: Asymmetry 1998, 9 (3), pp 459–466

[11] J Broos, JFJ Engbersen, IK Sakodinskaya, W Verboom, and DN Reinhoudt. Activity and enantioselectivity of serine proteases in transesterification reactions in organic media. J. Chem. Soc., Perkin Trans. 1, 1995, pp 2899-2905. (http://pac.iupac.org/publications/pac/pdf/1996/pdf/6811x2171.pdf ) and some of the follow-up discussions are summarized here: J Broos. Impact of the Enzyme Flexibility on the Enzyme Enantioselectivity in Organic Media Towards Specific and Non-specific Substrates. Biocatalysis and Biotransformation, 2002, 20 (4), pp. 291–295. (http://gbb.eldoc.ub.rug.nl/FILES/root/2002/BiocatBiotrBroos/2002BiocatalBiotransfBroos.pdf)

[12] I had to follow a course on thermodynamics given by Herman Berendsen while he and his group were writing the GROMOS software and I seemed to hear that joke a lot from them.