Recently the US EPA gave Codexis a Presidential Green Chemistry Award for a process designed for making a statin intermediate. The product, HN, a cyano alcohol, is prepared in a few cascading biocatalytic steps from the halo ketone using three different enzymes.
Codexis' multi-enzyme process towards the statin intermediate HN
The wild-type enzymes all performed below commercialisation limits and needed to be improved through directed evolution. The final process combined reactions catalysed by an optimised alcohol dehydrogenase (ADH), an optimised cofactor-regenerating glucose dehydrogenase (GDH) and an optimised halohydrin dehalogenase (HHDH).
As is the general rule in synthetic chemistry, the introduction of the chiral centre is best done as early as possible in the synthetic route to prevent the accrual of waste products which, at best, do not interfere with the process but which in most cases make downstream processing complicated.
In the HN process described above, the chirality is introduced in the first reaction step and the following steps do not change the enantiomeric purity of the product. The HN process is a prime example of an industrial biocatalytic reaction cascade that dramatically decreases the overall cost of the manufacturing process.
Biocatalytic potential of ADH & HHDH
HHDHs and short-chain ADHs are structurally closely related. The sequence identity between them is generally about 20-25%. These similarities and the fact that the cofactor-independent HHDHs have a conserved topology for a cofactor-binding domain, indicate that they have a common evolutionary origin.
ADHs catalyse the reduction of a wide range of ketones, yielding optically active alcohols. Similarly, the reduction of an a-chloro ketone yields an optically active haloalcohol.
Because ketones are pro-chiral substrates, the yield of enantiomerically pure alcohol can theoretically approach 100%. The ee of the resulting haloalcohol depends on the enantioselectivity of the enzyme, which is in some cases very good but is more often only moderate.
Recently, Julich Chiral Solutions achieved the production of both enantiomers of 2-chloro-1-phenylethanol on a multi-kg scale. (R)-2a was prepared optically pure using an ADH obtained from Rhodococcus sp. (ADH RS1), combined with a glucose dehydrogenase from Bacillus subtilus (GDH BS) for cofactor regeneration.
The other enantiomer of the haloalcohol, (S)-2a, was obtained using an ADH obtained from Lactobacillus brevis (ADH LB). The cofactor is regenerated through the oxidation of isopropanol by the same enzyme.
Various enzymes are known that catalyse the enantioselective ring-opening of epoxides such as glutathione transferases, epoxide hydrolases and HHDHs. Currently three types of HHDHs are known, designated HheA, HheB and HheC.
These cofactor-independent and metal-free enzymes can catalyse two different reactions: ring-closure of a haloalcohol to give the epoxide and ring-opening of an epoxide with anionic nucleophiles to give a range of 2-substituted alcohols. The recombinant HHDH obtained from Agrobacterium radiobacter AD1 (HheC) is the most studied.
Generally, aromatic haloalcohols are converted with a higher enantioselectivity than aliphatic haloalcohols. Recently it was shown that, by a single point mutation (W249F), an HheC variant could be obtained that displayed a more than seven-fold increase in initial activity and a six-fold increase in enantioselectivity, making this a superior catalyst for enantioselective ring-closure reactions.
In a kinetic resolution, one enantiomer of a racemic mixture is converted at a higher rate, leaving the other behind. The advantage is that the remaining enantiomer of the substrate can always be obtained in an optically pure form.
The discrimination of an enzyme towards both enantiomers is called enantioselectivity and can be expressed as the E-value. A kinetic resolution with a very high enantioselectivity or E-value (>200) yields the remaining enantiomer optically pure (>99%) with a yield close to the maximum of 50%.
Even if the E-value of the resolution is moderate (e.g. E=10) the residual enantiomer of the racemic substrate can still be obtained optically pure, though with a yield of less than 30%.
It is not only halides that can be used as a nucleophile in the ring-opening ofepoxides. HHDHs catalyse the ring-opening of various substituted epoxides with cyanide. This reaction gives optically active b-hydroxy nitriles, which can easily be converted to the corresponding 1,3 functionalised derivatives, such as 1,3 aminoalcohols.
The highest enantioselectivity was observed inthe ring-opening of aromatics, but simple aliphatic epoxides such as epoxybutane are ring-openedwith reasonable enantioselectivity by HheC (E=29) but not by HheA (E=4), nor HheB (E=1).
A major limitation of asymmetric ketone reduction is that only secondary alcohols are accessible and not sterically hindered tertiary alcohols, so that 2,2-disubstituted epoxides cannot be prepared. Fortunately, the ring-opening of epoxides bearing a substituent on the non-terminal position was well accepted by HheC. Conversion of 2-methyl-1,2-epoxybutane with cyanide occurred with very high enantioselectivity (E>200) yielding the corresponding b-hydroxynitrile in >99%.
Since nitrite (NO2-) is an ambident nucleophile, a ring-opening with this nucleophile can yield either the nitroalcohol (attack by the nitrogen atom) or the nitrite ester (attack by oxygen). The ring-opening of racemic 3c catalysed by HheC occurs with a very high enantioselectivity (E >200), yielding the remaining (S)-epoxide optically pure in 48% yield.
The product of this reaction is the nitrite ester with the nitro alcohol as minor side product. The regioselectivity, ranging from highly oxygen-selective to highly nitrogen-selective, depends on the structure of the substrate and the enzyme.
The ring-opening of epoxides by azide ions leads to the formation of azidoalcohols, which are precursors for various compounds, such as amino alcohols and aziridines. Various chemical catalysts have been described that catalyse the conversion of epoxides to azidoalcohols.
The regioselectivity of chemical ring-opening of the epoxide depends on electronic factors as well as steric factors. The ring-opening of (substituted) styrene oxides, for example, occurs predominantly at the more substituted carbon atom, since the phenyl group stabilises the formation of a positive charge in the transition state.
In the absence of an HHDH and with an excess of sodium azide, para-nitrostyrene oxide 3c was ring-opened, giving a regioisomeric mixture of 2-azido-2-(para-nitrophenyl)ethanol (63%) and 2-azido-1-(para-nitrophenyl)ethanol 4c (37%).
The regioselectivity of the HheC-catalysed ring-opening, however, is completely the opposite. It is solely directed at the terminal carbon atom resulting in the formation of 4c (98%) with only a trace of the other regioisomer.
In addition, the reaction is absolutely enantioselective (E>200). A kinetic resolution of 3c catalysed by HheC using 0.6 equivalents of NaN3 yielded the (S)-3c in >99% and (R)-4c in 96%.
Enzymatic ring-opening with azide has many advantages over chemical methods. The enzymatic reactions proceed at room temperature and neutral pH, using 0.5 equivalent of the cheap NaN3 (compared to, for example, the expensive TMSN3 that is used in similar enantioselective ring-opening reactions with salen catalysts),16 in the absence of heavy metal catalysts and yielding an optically pure product with no regioisomer present.
As mentioned, the ring-opening of aromatic epoxides generally occurs with a higher enantioselectivity than that of aliphatic epoxides. On the other hand, the enzymes have a very high activity for the azidolysis of aliphatic epoxides.
The activity towards epoxybutane or epichlorohydrin is 400 and >1,000 mmol/min/mg protein respectively. These ring-opening activities are more than ten times higher than the best ring-closure activity towards haloalcohols, which were considered to be their natural and best substrates.
Alternative routes
Combining an enzymatic step with a chemical reaction step would reduce the costs of a multi-step reaction. Recently, a one-pot biocatalytic conversion of 1-chloro-2-octanone to (R)-1,2-epoxyoctane was described. The reduction catalysed by whole cells of R. ruber DSM 44541 was performed at pH 13, yielding the optically pure epoxide as the sole product.
An HHDH-catalysed ring-closure of racemic para-nitro-2-bromo-1-phenylethanol 2c yielded the remaining haloalcohol (S)-2c optically pure in 47% yield and the epoxide (R)-3c in 52% yield, but with a moderate 90%.
After the separation of the substrate and product, recrystallisation of the epoxide (R)-3c yielded this compound in optically pure form and in 40% yield. Chemical ring-closure of the optically pure haloalcohol yielded the other enantiomer of the epoxide optically pure. Using this chemoenzymatic strategy, the kinetic resolution of 2c gives access to both optically pure epoxide enantiomers.
The HHDH-catalysed ring-opening of epoxides like para-nitrostyrene oxide 3c with nitrite as the nucleophile yields a nitrite ester as the major product.18 By performing the ring-opening under slightly acidic conditions the nitrite ester is hydrolysed in situ yielding the corresponding diol .
In this case, the HHDH acts as an epoxide hydrolase, an enzyme that catalyses the direct hydrolysis of epoxides to diols. Since the nitrite is released during the hydrolysis step, this ion can also be considered as a catalyst.
SpecChem online.com