Biocatalytic synthesis of valaciclovir using commercial enzymes

Article abstract of DOI:10.1016/j.tetlet.2010.10.161

Proof-of-concept has been demonstrated for the biocatalytic transformation of aciclovir into valaciclovir, attaining high conversions from a solid-to-solid biotransformation in l-valine methyl ester using various formulations of Subtilisin Carlsberg activated for use in organic solvent.

Full text of DOI:10.1016/j.tetlet.2010.10.161

Tetrahedron Letters 52 (2011) 215–218
Contents lists available at ScienceDirect
Tetrahedron Letters
journal homepage: www.elsevier.com/locate/tetlet
Biocatalytic synthesis of valaciclovir using commercial enzymes
Kathleen McClean a, Christopher Preston b, David Spence b, Peter W. Sutton a, , John Whittall a
⇑
a StylaCats Ltd, The Heath Business & Technology Park, Runcorn, Cheshire WA7 4QX, UK
b Technical Development Department, GlaxoSmithKline, North Lonsdale Road, Ulverston, Cumbria LA12 9DR, UK
a r t i c l e i n f o
a b s t r a c t
Article history:
Proof-of-concept has been demonstrated for the biocatalytic transformation of aciclovir into valaciclovir,
Received 3 September 2010
Revised 13 October 2010
Accepted 29 October 2010
Available online 4 November 2010
attaining high conversions from a solid-to-solid biotransformation in
formulations of Subtilisin Carlsberg activated for use in organic solvent.
L-valine methyl ester using various
Ó 2010 Elsevier Ltd. All rights reserved.
Valaciclovir 1, the active ingredient of ValtrexÒ, is a water solu-
-valine ester prodrug of the HIV reverse transcriptase inhibitor
aciclovir 2. It can be hydrolysed to 2 and -valine 3 in vivo by the
action of enzymes, one of which is the ‘so-called’ human biphenyl
hydrolase-related protein (hBph-rp) that has been cloned by
Puente and López-Otin and expressed in E. coli (Scheme 1).1
Valaciclovir is currently produced commercially by a two-step
process using 1,3-dicyclohexylcarbodiimide (DCC) coupling of N-
decided to screen for valaciclovir hydrolysis activity as this can
be performed under mild aqueous conditions. Although the pre-
ferred enzyme for hydrolysis will not necessarily be preferred in
the reverse reaction, particularly in the case of lipases where the
water nucleophile appears to be introduced through a separate
tunnel,4 this screening approach reduces the potential to miss a
hit through the use of inappropriate conditions in the more diffi-
cult synthesis reaction.
ble
L
L
benzyloxycarbonyl-
L
-valine (CBz-
L
-valine) and aciclovir followed
Using a hydrolysis procedure adapted from the work of
by hydrogenolysis with a palladium catalyst to remove the protect-
ing group (Scheme 2). Alternative chemical methods for preparing
valaciclovir have been reported but, like the original method of
manufacture, all require the use of an amino acid moiety contain-
ing a nitrogen protecting group which has to be removed in a sub-
sequent step.2
The work of Puente and López-Otin led us to consider the pos-
sibility of synthesising valaciclovir by reverse hydrolysis using
hydrolase enzymes (Scheme 2, dashed arrow).1 An atom-efficient,
chemoselective method of preparing amino acid esters that does
not require protection of the amino acid nitrogen atom might pro-
vide significant production cost benefits for manufacture through
the use of fewer chemical transformations and reducing the costs
of starting materials.
The use of mammalian enzymes in active pharmaceutical ingre-
dient (API) production, particularly in the final chemistry stage, is
undesirable due to concerns over the risk of infection by transmis-
sible spongiform encephalopathies (TSEs).3 Therefore, we decided
to screen commercial hydrolases and a selection of non-commer-
cial hydrolases prepared from microbial sources, for activity using
the human enzyme hBph-rp as a positive control.
Kim et al.,5 62 immobilised and free commercial enzyme prepara-
tions were screened and 19 were competent catalysts for the
hydrolysis of valaciclovir to aciclovir. When the same enzyme from
different suppliers was tested, no qualitative difference was ob-
served. Of the microbial strains that were tested based on our
selection criteria,6 5/6 yeast strains, 7/14 Gram-negative bacterial
strains, 0/5 Gram-positive bacterial strains and 1/6 filamentous
fungi strains gave extracts displaying valaciclovir hydrolysis
activity.
Given the large number of commercial enzymes that were
found to display valaciclovir hydrolysis activity, subsequent syn-
thetic studies focused on these, and in particular Subtilisin A (Sub-
tilisin protease from Bacillus licheniformis, also known as Subtilisin
Carlsberg) and ChiroCLEC-BL (a cross-linked form of the same en-
zyme), which displayed the highest activity of all the enzymes
tested including hBph-rp.
We anticipated that the synthesis of valaciclovir by reverse
hydrolysis would be highly challenging to realise biocatalytically
because of the poor solubility of
ganic and aqueous solvents. This is further hampered by the zwit-
terionic nature of -valine which reduces the concentration of the
L-valine and aciclovir in both or-
L
To identify non-mammalian commercial and non-commercial
enzymes that might be capable of synthesising valaciclovir, we
protonated acid form necessary for acyl enzyme formation at pH
values that are considered to be compatible with enzyme activity.
The possibility that the favoured amino acid form might be gener-
ated within the enzyme active site, along with precedent for the
enzymatic esterification of amino acids with sugars,7 led us to pur-
⇑
Corresponding author. Address: GlaxoSmithKline, Synthetic Chemistry, Gunnels
Wood Road, Stevenage, Hertfordshire SG1 2NY, United Kingdom. Tel.: +44 1438 76
8674; fax: +44 1438 76 4869.
sue this option. However, all attempts to esterify
L-valine with aci-
clovir failed to yield the desired product with the enzymes tested.8
E-mail address: peter.w.sutton@gsk.com (P.W. Sutton).
0040-4039/$ - see front matter Ó 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.tetlet.2010.10.161

216
K. McClean et al. / Tetrahedron Letters 52 (2011) 215–218
O
N
N
O
N
OH
H2N
N
O
N
N
2
O
N
+
O
N
O
H2N
O
NH3Cl
HO
1
NH3Cl
3
Scheme 1. Reagents: hBph-rp, pH 7.4 phosphate buffer.
O
O
N
N
O
N
i
N
O
N
O
N
O
H2N
N
OH
iii
H2N
N
H
H
NHCbz
4
2
ii
O
N
O
N
O
N
O
H2N
N
H
NH3Cl
1
-valine, DCC, DMAP, DMF; (ii) H2, 5% Pd/C, DMF, 0.5 M HClaq; (iii) desired enzyme catalysed esterification, L-valine.
Scheme 2. Reagents: (i) CBz-
L
Given the high polarity of the starting materials and product,
reaction as a potential biocatalytic route to valaciclovir (Scheme
3). Irreversible acyl donors such as p-nitrophenyl or vinyl esters
that have been used to selectively prepare amino acid esters of
other nucleoside analogues such as nelarabine,10 ribavarin11 and
lobucavir12 were not considered as they require preparation from
expensive N-protected amino acids. We instead chose the hydro-
we postulated that in an organic solvent they might bind tightly
to the protein resulting in inhibition. However, this appeared unli-
kely to be product related given that Ke and Klibanov have demon-
strated the propanolysis of racemic methyl phenylalaninate in
organic solvent in the presence of CLEC enzymes.9 In fact, we found
that the transesterification of valaciclovir with methanol could also
be effected under these conditions by a variety of enzymes of
which Subtilisin A and ChiroCLEC-BL were again preferred, both
chloride salt of L-valine methyl ester 5 as an acyl donor for this
non-selective transformation as it can be easily produced in a sin-
gle step from the amino acid, would give direct access to product as
the desired hydrochloride salt and provide transient salt protection
of the nitrogen atom.13
giving very similar activity. If L-valine were inhibiting the esterifi-
cation reaction, then spiking experiments should inhibit the
transesterification, but we found this not to be the case, and so pre-
sumably the esterification reaction is limited by the rate of acyl en-
zyme formation.
Following the successful transesterification of valaciclovir with
methanol, we instead turned our attention to the reverse of this
Disappointingly, attempts to transesterify L-valine methyl ester
hydrochloride with aciclovir in the presence of 4 Å molecular
sieves and various enzymes failed to afford any valaciclovir. Test-
ing the free amine of 5 instead, enantiopure valaciclovir was finally
produced in the presence of ChiroCLEC-BL with best rates observed
O
N
N
O
N
H2N
N
H
OH
O
2
N
O
N
+
O
N
O
O
H2N
N
H
NH3Cl
MeO
1
NH3Cl
5
Scheme 3. Desired enzyme catalysed transesterification.

K. McClean et al. / Tetrahedron Letters 52 (2011) 215–218
217
20
15
10
5
80
60
40
20
0
ChiroCLEC-BL
Subtilisin A
Vacuum (250 mbar)
Atmospheric Pressure
0
0
50
100
150
0
5
10
15
20
Time/d
Time/h
Figure 1. Screening results for the transesterification of L-valine methyl ester (free
amine) with aciclovir using ChiroCLEC-BL or Subtilisin A at room temperature in the
presence of 4 Å molecular sieves.
Figure 2. Gram scale synthesis of valaciclovir in the presence of precipitated
Alcalase at 50 °C, with and without applied vacuum (250 mbar).
in neat methyl
L
-valinate or with tert-butanol cosolvent (Fig. 1).
and cross-linked catalyst with no observable racemisation.19 On a
gram scale, by replacing molecular sieves with reduced pressure
(250 mbar) as a driving force, preliminary experiments gave con-
versions of up to 70% from a 20 volume process (Fig. 2). However,
the mixtures were observed to become more viscous over the time
course of the reactions, even where a constant volume was main-
tained by periodic addition of fresh valine methyl ester. In fact, aci-
clovir/valaciclovir mixtures, isolated from laboratory scale reaction
by filtration, acidification and precipitation were found to contain a
significant quantity of polymeric valine.
Surprisingly, Subtilisin A was unable to produce valaciclovir under
these conditions even though it had displayed similar activity to
ChiroCLEC-BL in the reverse reaction. Given that all enzymes re-
quire a minimum quantity of water for activity,14 we suspected
that the discrepancy in activity between Subtilisin A and ChiroC-
LEC-BL in the forward and reverse transesterification reactions
might have arisen as a result of different catalyst hydration levels
in the different solvents used and that the CLEC enzyme might
have contained sufficient water before solvent addition.
Unfortunately ChiroCLEC-BL is no longer commercially avail-
able and so it was important to obtain an alternative formulation
of Subtilisin Carlsberg that gives a comparable activity before reac-
tion optimisation. The activation of lyophilised enzymes in organic
solvent has been studied extensively,15 but we found that by using
the simple procedure reported by Chen et al.,16 Subtilisin A, precip-
itated from aqueous solution with tert-butanol, gave comparable
wt/wt activity to the CLEC enzyme following filtration and an ace-
tone wash. In fact, the cheap, commercial formulation Alcalase
2.5 L, that is, available in bulk quantities, could be treated in a sim-
ilar manner to give a catalyst of comparable activity and both could
be cross-linked with glutaraldehyde in organic solvent to produce
an insoluble version of the enzyme without loss of activity. Alter-
natively the enzyme could be precipitated in situ by the addition
of Alcalase 2.5 L directly to the reaction mixture. However, Alcalase
2.5 L contains a significant quantity of 1,2-propanediol (30%) and
conversions were inferior using this technique, presumably due
to competition of the diol nucleophile with aciclovir.
With a suitable first generation catalyst in hand we next turned
our attention to improving the productivity of the transesterifica-
tion reaction. We found that increasing the reaction temperature
had a significant positive impact on reaction rate and was opti-
mum at 50 °C. Water content was critical to maintain activity with
optimum rates observed at 1–3% v/v of water. Further increases in
water content resulted in reduced product formation. Under iden-
tical conditions, untreated Subtilisin A gave no product, illustrating
that hydration of the dehydrated catalyst in organic solvent was
not sufficient for reactivation as previously observed by Halling
et al.17 Presumably this is because the protein is rigid in organic
solvent, preventing conformational change.18
Treatment of neat
lase (30 mg in 1 ml of
L
-valine methyl ester with precipitated Alca-
-valine methyl ester) at 50 °C gave an emul-
L
sion after three days whereas in a parallel experiment, in the
absence of catalyst, no solid was observed to form over the same
period.
Proof-of-concept has been demonstrated for the biocatalytic
transformation of aciclovir into valaciclovir, attaining high conver-
sions from a solid-to-solid biotransformation in L-valine methyl
ester using various forms of Subtilisin Carlsberg activated for use
in organic solvent. Subtilisin Carlsberg ultimately proved to be
unsuitable for scale-up, displaying insufficient activity towards
the hydrochloride salt of L-valine methyl ester that might have
allowed direct access to valaciclovir as the desired hydrochloride
salt and inadequate selectivity towards the more active free base,
resulting in polymer formation. In order to realise a practical
process for the biocatalytic production of valaciclovir it would,
therefore, be necessary to identify an alternative enzyme or
Subtilisin Carlsberg variant that accepts
hydrochloride as an acyl donor or that accepts the free amine of
-valine methyl ester as an acyl donor but not an acceptor. In
the latter case, the rate of valaciclovir synthesis might be expected
to improve in the absence of competing peptide formation. An
improved enzyme might also allow access to more efficient pro-
cesses for the preparation of a variety of amino acid ester prodrugs
of nucleosides20 and other drugs21 that have recently attracted
considerable interest.
L-valine methyl ester
L
We have also demonstrated that Subtilisin Carlsberg and other
hydrolases can transesterify
(valaciclovir). Transient protection of nitrogen should prevent
competing polypeptide formation and given that -valine repre-
L-valine ester hydrochloride salt
L
The reaction is a solid-to-solid transformation where both start-
ing material and product have similarly poor solubility (<1 mg/ml
in tert-butanol or L-valine methyl ester) and so the substrate solu-
tion concentration remains constant throughout the majority of
the reaction resulting in a linear rate until high conversions.
Attempts to improve the reaction rate by the use of a range of ionic
liquids to enhance substrate solubility were unsuccessful.
Under optimum conditions, yields of up to 89% were achieved at
20 mg/ml concentration using either precipitated or precipitated
sents a particularly hindered amino acid, it is likely that a range
of amino acid ester salts could be used as acylating agents for
the resolution of racemic alcohols. Amino acid ester based acylat-
ing agents would be cheap and produce a basic product, that is,
readily separable from the neutral starting alcohol, therefore,
avoiding the need for chromatographic separation. This would pro-
vide a complementary approach to the use of succinic anhydride
and a single step alternative to the use of the N-Boc protected ami-
no acid vinyl esters recently reported by Moody and coworkers.22

218
K. McClean et al. / Tetrahedron Letters 52 (2011) 215–218
12. Hanson, R. L.; Shi, Z.; Brzozowski, D. B.; Banerjee, A.; Kissick, T. P.; Singh, J.;
Acknowledgement
Pullockaran, A. J.; North, J. T.; Fan, J.; Howell, J.; Durand, S. C.; Montana, M. A.;
Kronenthal, D. R.; Mueller, R. H.; Patel, R. N. Bioorg. Med. Chem. 2000, 8, 2681–
2687.
We acknowledge Dr. X.S. Puente for kindly providing plasmid
pAR- Bph as a gift.
13. Gill, I.; Lopez-Fandino, R.; Vulfson, E. J. Am. Chem. Soc. 1995, 117, 6175–6181.
14. Zaks, A.; Klibanov, A. M. J. Biol. Chem. 1988, 263, 3194–3201.
15. Bordusa, F. Chem. Rev. 2002, 102, 4817–4867. and references therein..
16. Chen, S.-T.; Tu, C.-C.; Wang, K.-T. Bioorg. Med. Chem. Lett. 1993, 3, 539–542.
17. Partridge, J.; Hutcheon, G. A.; Moore, B. D.; Halling, P. J. J. Am. Chem. Soc. 1996,
118, 12873–12877.
References and notes
1. Puente, X. S.; López-Otin, C. J. Biol. Chem. 1995, 270, 12926–12932.
2. Beauchamp, L. M. U.S. Patent 4957,924, 1990; Chem. Abstr. 1990, 112, 112,048.
3. Minimising the Risk of Transmitting Animal Spongiform Encephalopathy Agents via
Human and Veterinary Medicinal Products, In European Pharmacopoeia 6.0,
2010, Chapter 5.2.8.
4. Wittrup Larsen, M.; Zielinska, D. F.; Martinelle, M.; Hidalgo, A.; Juhl Jensen, L.;
Bornscheuer, U. T.; Hult, K. ChemBioChem 2010, 11, 796–801.
5. Kim, I.; Song, X.; Vig, B. S.; Mittal, S.; Shin, H.-C.; Lorenzi, P. J.; Amidon, G. L. Mol.
Pharmacol. 2004, 1, 117–127.
6. Microbes were chosen which have documented hydrolase activity or have been
isolated from environments where hydrolase activities might be expected to be
used to degrade contaminants/xenobiotics. Additional criteria that were
imposed on strain selection were availability, pathogenicity, ease of culture
and intellectual property issues.
7. Bhandya, S. R.; Soundar, D. Indian J. Chem. 2007, 46B, 2026–2044; Kenchaiah, L.;
Soundar, D. J. Biotechnol. 2005, 117, 49–56.
8. Using the best enzyme hits from the hydrolysis screen, conditions that were
attempted included the use of neat organic solvent, solvent/aqueous buffer or
aqueous buffer alone at both low and high pH. Using conditions described in
18. Griebenow, K.; Klibanov, A. M. J. Am. Chem. Soc. 1996, 118, 11695–11699.
19. In a typical experiment, water (40
acyclovir (40 mg) in -valine methyl ester (2 ml). After brief shaking 4 Å
molecular sieves were added and the resultant mixture stirred at 50 °C. 25
samples of the reaction mixture were taken at regular intervals, diluted with
10% TFA in H2O (1 ml) and passed through a 4 m PTFE syringe filter. Analysis
was performed by HPLC, injecting 5 l of diluted sample onto a Spherisorb
phenyl (250 Â 4.6 mm, 5 m) column and eluting with 20:80:0.1 MeOH/H2O/
ll) was added to the catalyst (40 mg) and
L
ll
l
l
l
TFA at a flow rate of 1 ml/min, with the UV detector set at 254 nm. Valaciclovir
enantiopurity was determined by HPLC using a Crownpak CR (+) column and
eluting with H2O/MeOH/90% aqueous HClO4 (950:50:5) as the mobile phase at
a flow rate of 0.75 ml/min, with the UV detector set at 254 nm.
20. Gao, H.; Mitra, A. K. Tetrahedron Lett. 2000, 41, 1131–1136; Vig, B. S.; Lorenzi, P.
J.; Mittal, S.; Landowski, C. P.; Shin, H.-C.; Mosberg, H. I.; Hilfinger, J. M.;
Amidon, G. L. Pharm. Res. 2003, 20, 1381–1388; Aggarwal, S. K.; Gogu, S. R.;
Rangan, S. R. S.; Agrawal, K. C. J. Med. Chem. 1990, 33, 1505–1510; Kim, D.-K.;
Lee, N.; Kim, Y.-W.; Chang, K.; Im, G.-J.; Choi, W.-S.; Kim, K. H. Bioorg. Med.
Chem. 1999, 7, 419–424; Sriram, D.; Yogeeswari, P.; Srichakravarthy, N.; Ratan
Bal, T. Bioorg. Med. Chem. Lett. 2004, 14, 1085–1087.
the following references, treatment of
in the presence of Subtilisin A also failed to afford any product in the presence
of subtilisin A, whereas N-Boc- -phenylalanine could be esterified: Liu, C.-F.;
L-valine or L-phenylalanine with ethanol
21. Mahfouz, N. M.; Hassan, M. A. J. Pharm. Pharmacol. 2001, 53, 841–848; Mathew,
A. E.; Mejillano, M. R.; Nath, J. P.; Himes, R. H.; Stella, V. J. J. Med. Chem. 1992, 35,
145.
L
Tam, J. P. Org. Lett. 2001, 3, 4157–4159.
22. Brossat, M.; Moody, T. S.; de Nanteuil, F.; Taylor, S. J. C.; Vaughan, F. Org. Process
Res. Dev. 2009, 13, 706–709.
9. Ke, T.; Klibanov, A. M. J. Am. Chem. Soc. 1999, 121, 3334–3340.
10. Mahmoudian, M.; Eaddy, J.; Dawson, M. Biotechnol. Appl. Biochem. 1999, 29,
229–233.
11. Tamarez, M.; Morgan, B.; Wong, G. S. K.; Tong, W.; Bennett, F.; Lovey, R.;
McCormick, J. L.; Zaks, A. Org. Process Res. Dev. 2003, 7, 951–953.