Preparation of an amino acid intermediate for the dipeptidyl Peptidase IV inhibitor, saxagliptin, using a modified phenylalanine dehydrogenase

Article abstract of DOI:10.1002/adsc.200700013

The non-proteinogenic amino acid 2-(3-hydroxy-1-adamantyl)-(2S)- aminoethanoic acid [2, (S)-3-hydroxyadamantylglycine], is a key intermediate required for the synthesis of Saxagliptin, a dipeptidyl peptidase IV inhibitor under development for treatment of type 2 diabetes mellitus. Keto acid 2-(3-hydroxy-1-adamantyl)-2-oxoethanoic acid (1) was converted to (S)-3-hydroxyadamantylglycine by reductive amination using a phenylalanine dehydrogenase from Thermoactinomyces intermedius expressed in a modified form in Pichia pastoris or Escherichia coli. NAD (nicotinamide adenine dinucleotide) produced during the reaction was recycled to NADH (reduced form of nicotinamide adenine dinucleotide) using formate dehydrogenase. Pichia pastoris produces an endogenous formate dehydrogenase when grown on methanol, and the corresponding gene was cloned and expressed in E. coli. The modified phenylalanine dehydrogenase contains two amino acid changes at the C-terminus and a 12-amino acid extension of the C-terminus. The modified enzyme is more effective with keto acid 1 than the wild-type enzyme, but less effective with the natural substrate, phenylpyruvate. Production of multi-kg batches was originally carried out with extracts of Pichia pastoris expressing the modified phenylalanine dehydrogenase from Thermoactinomyces intermedius and endogenous formate dehydrogenase, and further scaled up using a preparation of the two enzymes expressed in E. coli.

Full text of DOI:10.1002/adsc.200700013

FULL PAPERS
DOI: 10.1002/adsc.200700013
Preparation of an Amino Acid Intermediate for the Dipeptidyl
Peptidase IV Inhibitor,Saxagliptin,using a Modified
Phenylalanine Dehydrogenase
Ronald L. Hanson,a,* Steven L. Goldberg,a David B. Brzozowski,a
Thomas P. Tully,a Dana Cazzulino,a William L. Parker,a Olav K. Lyngberg,a
Truc C. Vu,a Michael K. Wong,a and Ramesh N. Patela
a
Process Research and Development, Bristol-Myers Squibb, One Squibb Drive, New Brunswick, NJ 08903, USA
Fax : (+1)-732-227-3994; e-mail: ronald.hanson@bms.com
Received: January8, 2007
Abstract: The non-proteinogenic amino acid 2-(3-hy- and expressed in E. coli. The modified phenylalanine
droxy-1-adamantyl)-(2S)-aminoethanoic acid [2, (S)- dehydrogenase contains two amino acid changes at
3-hydroxyadamantylglycine], is a key intermediate the C-terminus and a 12-amino acid extension of the
required for the synthesis of Saxagliptin, a dipeptidyl C-terminus. The modified enzyme is more effective
peptidase IV inhibitor under development for treat- with keto acid 1 than the wild-type enzyme, but less
ment of type 2 diabetes mellitus. Keto acid 2-(3-hy- effective with the natural substrate, phenylpyruvate.
droxy-1-adamantyl)-2-oxoethanoic acid (1) was con- Production of multi-kg batches was originallycarried
verted to (S)-3-hydroxyadamantylglycine by reduc- out with extracts of Pichia pastoris expressing the
tive amination using a phenylalanine dehydrogenase modified phenylalanine dehydrogenase from Ther-
from Thermoactinomyces intermedius expressed in a moactinomyces intermedius and endogenous formate
modified form in Pichia pastoris or Escherichia coli. dehydrogenase, and further scaled up using a prepa-
NAD (nicotinamide adenine dinucleotide) produced ration of the two enzymes expressed in E. coli.
during the reaction was recycled to NADH (reduced
form of nicotinamide adenine dinucleotide) using
formate dehydrogenase. Pichia pastoris produces an Keywords: amino acids; enzyme catalysis; phenylala-
endogenous formate dehydrogenase when grown on nine dehydrogenase; reductive amination; Saxaglip-
methanol, and the corresponding gene was cloned tin
Introduction
an intermediate. This compound was originallypre-
pared using an asymmetric Strecker amino acid syn-
Glucagon-like peptide-1 (GLP-1) analogues and di- thesis.[4] We have previouslyprepared several unnatu-
peptidyl peptidase IV (DPP-IV) inhibitors are two ral amino acids[5–7] as intermediates for synthetic
promising new approaches currentlybeing explored
routes byreductive amination of the corresponding
for treatment of type 2 diabetes.[1–3] GLP-1, a peptide keto acids using l-amino acid dehydrogenases and
secreted bythe gut in response to feeding, has the
beneficial effects of increasing glucose-stimulated in-
sulin secretion, decreasing glucagon secretion, delay-
ing gastric emptying, and increasing the b-cell mass of
pancreatic islets.[1–2] GLP-1 (7–36) amide is rapidlyin-
activated byconversion to GLP-1 (9–36) amide by
DPP-IV.[1–3] To alleviate the inactivation of GLP-1,
more stable analogues of GLP-1 as well as inhibitors
of DPP-IV are approaches to provide improved con-
trol of blood glucose for diabetics.
have investigated this approach to afford an improved
Saxagliptin[4] (Figure 1), a DPP-IV inhibitor under
development byBristol-Meyrs Squibb, requires ( S)-
N-BOC-3-hydroxyadamantylglycine (3) (Scheme 1) as Figure 1. Structure of Saxagliptin.
Adv. Synth. Catal. 2007, 349, 1369 – 1378
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1369

Ronald L. Hanson et al.
FULL PAPERS
Scheme 1. Enzymatic reaction used for preparation of (S)-3-hydroxyadamantylglycine.
synthesis of intermediate 3. This report describes the idation of formate to CO2 using formate dehydrogen-
preparation of (S)-3-hydroxyadamantylglycine (2) ase (FDH) which is produced by Pichia pastoris
using a modified form of a recombinant phenylala- during growth on methanol. Wet cells, heat-dried cells
nine dehydrogenase cloned from Thermoactinomyces and cell extracts were compared. Pichia pastoris ex-
intermedius and expressed in Pichia pastoris or Es- tract was more effective than wet cells or heat-dried
cherichia coli. The amino acid 2 can be directlypro-
cells. Wet cells gave 64% yield, heat-dried cells gave
tected as its BOC derivative without isolation of 2 to 80% yield, whereas an extract gave close to 100%
afford intermediate 3. This biocatalysis procedure and yield. The best conditions using Pichia extract were
subsequent BOC protection has been scaled up to 0.25M keto acid, 0.5M ammonium formate, 1 mM
multi-kg scale to support the development of Saxa- NAD, 1 mM dithiothreitol, pH 8, and 12.5% w/v
gliptin.
Pichia extract, at 408C. The total turnover number
for the NAD cofactor was 237 with this keto acid
input and NAD concentration. With these conditions,
a total of 1.6 kg 2 was prepared in lab batches. Yields
before isolation were close to 100% with no (R)-en-
antiomer detected. The progress curves for a 30-g re-
action are shown in Figure 2.
Results and Discussion
Screening of Amino Acid Dehydrogenases and
Transaminases
The product from earlybatches was isolated as the
Of six amino acid dehydrogenases screened for reduc- free amino acid 2. The procedure for isolating the
tive amination of 1 to 2, onlytwo phenylalanine dehy- amino acid consisted of boiling the enzymatic reaction
drogenases were found to effect the conversion. Phe- mixture to drive off ammonia, adjusting to pH 3 with
nylalanine dehydrogenase (PDH) from Sporosarcina formic acid, filtration to remove precipitated protein,
[7]
species gave onlya 3% yield and heat-dried cells
of adsorption of the amino acid on Dowex 50 (H+)
Pichia pastoris containing phenylalanine dehydrogen- resin, elution with 1.5M ammonia and concentration
ase cloned from Thermoactinomyces intermedius gave of the rich effluent to give 2 as a crystalline solid. The
a 12% yield. Phenylalanine dehydrogenase from Rho- last run using this isolation procedure (787 g keto acid
dococcus, leucine dehydrogenase from Bacillus input) gave 804 g of 2 with a purityof 94.3 weight%
sphaericus, glutamate dehydrogenase from beef liver, and a yield of 96.0% from 1. All of the batches of 2
and alanine dehydrogenase from Bacillus subtilis isolated bythis procedure performed well in subse-
were not active. Four transaminases were also quent reactions for the synthesis of Saxagliptin. Sever-
screened, and onlybranched chain transaminase cata-
al additional batches were conducted to prepare the
lyzed the conversion of 1 to 2 (56-mg scale) using glu- N-BOC amino acid 3 without the resin purification
tamate as amino donor with a 49% yield.
Preparation of 2 with Phenylalanine Dehydrogenase
from Thermoactinomyces intermedius Expressed in
Pichia pastoris
Using fresh cells, the reaction was evaluated under
various conditions with phenylalanine dehydrogenase
from Thermoactinomyces intermedius expressed in
Pichia pastoris SMD 1168 (pPDH9K/10)
[SC16176].[7]
G
The enzymatic reductive amination reaction requires
ammonia and NADH (Scheme 1). NAD produced
during the reaction was recycled to NADH by the ox- acid 1; , amino acid 2.
Figure 2. Time course of reaction for conversion of 30 g keto
&
acid 1 to amino acid 2 by Pichia pastoris extract. , keto
^
1370
asc.wiley-vch.de
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2007, 349, 1369 – 1378

Amino Acid Intermediate for the Dipeptidyl Peptidase IV Inhibitor Saxagliptin
FULL PAPERS
step. To facilitate the BOC protection, protein was
first removed bymicrofiltration and ultrafiltration,
then ammonia was removed byboiling at pH 9.6. The
Use tests of E. coli SC16487 PDH/FDH extracts
and another Pichia construct with constitutive Ther-
moactinomyces intermedius PDH and formate-in-
input keto acid 1 could also be used without isolation. duced FDH gave only53% and 8% yields, respec-
The input keto acid 1 for one of the batches was a so- tively. The E. coli PDH/FDH enzyme activities were
lution containing about 1M NaCl and 0.25M higher than the methanol-induced Pichia PDH/FDH
NaHCO3, prepared without isolation byethyl acetate
SC16176 and the formate-induced Pichia had PDH/
extraction. The reaction was satisfactory, but proceed- FDH activities equal to the methanol-induced Pichia.
ed a little slower than that with the purified keto acid Dialysis of the extracts to remove possible inhibitors
1. Pilot plant batches of 4, 7, 10 and 37 kg gave close did not improve performance in the use test. After
to 100% conversion and were directlyconverted to 3 32 h of use test reaction most of the PDH activityre-
before isolation.
mained, although FDH was decreased. Addition of
more FDH, NAD or both after 32 h had little effect
on yield. In spectrophotometric enzyme assays, the
methanol-grown Pichia extract had measurable activi-
tywith 1, but the formate-grown Pichia extract and
Modified PDH Found in Pichia pastoris
Production of PDH and FDH in Pichia pastoris re- the E. coli extract did not. However, the latter two
quires growth of the cells on methanol to induce the extracts had higher activity with phenylpyruvate (nat-
enzyme activities.[7] The PDH enzyme expressed in ural substrate for PDH) than the the methanol-grown
Escherichia coli, inducible byisopropylthio- b-d-galac- Pichia extract. An extract of Pichia cells removed
topyranoside (IPTG), was also evaluated. Escherichia from a fermentor just before methanol induction was
coli BL21
(DE3) (pPDH155K) SC16144[7] extract con- inactive for reductive amination of 1, suggesting that
G
taining PDH cloned from Thermoactinomyces inter- the methanol-induced PDH was responsible for the
medius (but not FDH) combined with commercially reductive amination reaction, not an endogenous
available FDH from Candida boidinii was less effec- Pichia enzyme.
tive (45% conversion) than Pichia extract despite
Phenylalanine dehydrogenase from Thermoactino-
having higher PDH activity. E. coli heat-dried cells or myces intermedius was first cloned and sequenced by
wet cells gave much less conversion than extracts. An- Takeda et al.,[8] and this sequence information was
other strain of E. coli , SC16487, was constructed to used for PCR cloning and expression of the gene in
express Thermoactinomyces intermedius PDH and Pichia and E. coli.[7] The cloned pdh gene was re-iso-
Pichia pastoris FDH, both inducible byIPTG, to
lated from Pichia SC16176, and the DNA sequence
obtain expression levels of both enzymes higher than was compared with the original published sequence of
in Pichia pastoris and avoid the use of methanol in the pdh gene. The sequence of the 3’-end of the pdh
the pilot plant. Despite high PDH activitywith phe-
gene was found to be changed in Pichia SC16176
nylpyruvate, the PDH enzyme expressed in E. coli from the original Thermoactinomyces pdh gene. The
gave a low yield for conversion of keto acid 1 to modified PDH contains two amino acid changes at
amino acid 2 and a low activitywhen measured in a
the C-terminus and a 12 amino acid extension of the
spectrophotometric assaywith keto acid 1 in place of C-terminus (Figure 3). This fortuitous alteration in
phenylpyruvate (Table 1). This surprising result was the gene was probablydue to lack of a stop codon in
investigated further.
the primer used to amplifythe pdh gene for cloning
into the expression vector. The amino acid changes
Table 1. Effect of PDHmod on conversion of keto acid 1 to amino acid 2.
Strain
Enzymes
Substrate
U/g wet cells[a]
Yield [%] 2[b]
Pichia pastoris SC16176
PDHmod/FDH
formate
36.4
21.5
5.6
85.6
869.1
0.9
132.4
36.9
11.6
phenylpyruvate
keto acid 1
formate
phenylpyruvate
keto acid 1
formate
95
E. coli SC16487
E. coli SC16496
PDH/FDH
53
PDHmod/FDH
phenylpyruvate
keto acid 1
100
[a]
Activitymeasured in 25% w/v extracts as described in Experimental Section.
[b]
Percent conversion of 50 mgmLÀ1 keto acid 1 to amino acid 2 by12.5% w/v extract.
Adv. Synth. Catal. 2007, 349, 1369 – 1378
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
1371

Ronald L. Hanson et al.
FULL PAPERS
Figure 3. Comparison of 3’-ends of native pdh gene with modified gene reisolated from Pichia pastoris SC16176.
and C-terminal extension should result in a 1353 along with the FDH gene from Pichia pastoris to give
dalton increase in the subunit molecular weight vs. strain SC16496. Comparison of the protein subunits in
unmodified PDH. The modified PDH protein extracts of E. coli SC16487, Pichia pastoris SC16176,
(PDHmod) is apparentlybetter able to accept keto
and E. coli SC 16496 as shown in Figure 4 is consis-
acid 1 as substrate than is the unmodified PDH ex- tent with a slightlyincreased molecular weight for
pressed in E. coli SC16487. After this finding, the PDHmod. The PDH subunit molecular weights in E.
PDHmod gene was subsequentlycloned into E. coli coli SC16487, Pichia SC16176, and E. coli SC16496
were 37.9 kD, 38.6 kD and 39.3 kD, respectively
(mean of 3 lanes for each extract). The FDH subunit
in all strains had a molecular weight of about 42.9 kD.
The higher level of expression of PDH and FDH in
the new E. coli strain (SC16496) compared to the old
E. coli strain (SC16487) and to Pichia SC16176 is also
apparent in Figure 4. PDH from E. coli SC16487 and
PDHmod from Pichia SC16176 extracts were partially
purified bygel filtration chromatographyon an FPLC
Superdex 200 column to compare molecular weights
of the native enzymes. The activity peak for each
enzyme was found in the same fraction number indi-
cating a similar native molecular weight.
Preparation of 2 with the Modified Phenylalanine
Dehydrogenase from Thermoactinomyces
intermedius Expressed in Escherichia coli
Reductive amination was conducted using cell ex-
tracts from E.coli strain SC16496 expressing
PDHmod and cloned FDH from Pichia pastoris. Cells
from a 15-L tank had 133 U/g FDH, 65 U/g PDH
(phenylpyruvate assay), and 12.7 U/g PDH (assayed
with keto acid 1). The extract was used for conversion
of 30 g 1 to 2 in close to 100% yield, and this materi-
al, after filtration for protein removal, was converted
to 3 byBOC protection. Further experiments showed
that the E. coli extract could be used at 2.5% w/v
concentration instead of the 12.5% concentration
used for batches with Pichia pastoris extract. The pH
Figure 4. Comparison of PDH/FDH in E. coli SC16487 (ec
old) with PDHmod/FDH in Pichia pastoris SC16176 and in
E. coli SC 16496 (ec mod). 25% w/v extracts were diluted
10-fold with water and analyzed with an Agilent 2100 Bio-
analyzer using a Protein 200 Plus Labchip.
optimum for the reductive amination of
1 by
PDHmod was 8.4–9.0. The apparent Km for keto acid
1 with the E. coli extract at pH 8.75 using assayC
1372
asc.wiley-vch.de
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2007, 349, 1369 – 1378

Amino Acid Intermediate for the Dipeptidyl Peptidase IV Inhibitor Saxagliptin
FULL PAPERS
from the Experimental Section with varied keto acid 608C, whereas PDH retains most of its activityafter
was 23Æ5 mM (std. error). Because of the lower pH 1 h at 608C.
optimum (6.5–7.5[9] or 7.5[10] with more than 80% of
maximal activityat pH 8 [9–10]) for FDH, reactions
were carried out at pH 8.
Conclusions
After some further modifications to the procedure,
PDHmod and FDH expressed in E. coli have now Pichia extracts containing PDH from Thermoactino-
been used to prepare several hundred kg of BOC-pro- myces intermedius and endogenous FDH were used
tected amino acid 3 to support the development of for pilot plant batches (up to a 37-kg scale) giving
Saxagliptin.
close to 100% conversion of keto acid 1 to (S)-amino
acid 2 with no detectable (R)-enantiomer. PDH
cloned from Thermoactinomyces intermedius and ex-
pressed in Pichia pastoris was a modified version
(PDHmod) of the original enzyme. The modified
Heat Stability of PDHmod
Phenylalanine dehydrogenase from the thermophilic PDH contains two amino acid changes at the C-termi-
actinomycete Thermoactinomyces intermedius is a nus and a 12 amino acid extension of the C-terminus.
thermostable enzyme.[8] To compare the heat stability PDHmod is more effective with 1 and less effective
of PDHmod with PDH, extracts of E. coli SC16496 with phenylpyruvate than the original PDH.
(FDH/PDHmod) and SC16487 (FDH/PDH) were in- PDHmod is less heat-stable than PDH. E. coli ex-
cubated for 1 h at 0, 40, 50, 60 or 708C, then assayed tracts expressing both PDHmod and Pichia pastoris
for activity with phenylpyruvate or keto acid 1 as sub- FDH catalyze the conversion effectively and have
strate. Samples of the untreated centrifuged extracts been used for further scale-up to convert several hun-
were also run on an Agilent 2100 bioanalyzer protein dred kg of keto acid 1 to BOC-protected (S)-amino
200 labchip to separate and quantitate the PDH subu- acid 3.
nit bands. Specific activities are expressed as units per
mg PDH or PDHmod protein in the subunit band.
Figure 5 shows that the PDHmod enzyme has lower Experimental Section
specific activity with phenylpyruvate but higher specic
activitywith keto acid 1 than the original PDH Preparation of Keto Acid 1
enzyme. PDHmod also has lower heat stability than
PDH, with PDHmod losing all activityafter 1 h at
The preparation of keto acid 1 has been described.[11] De-
tailed procedures for the synthetic route to Saxagliptin are
provided in a published patent application.[12]
Construction and Growth of Recombinant Strain P.
pastoris SMD1168 (pPDH9K/10) (SC16176,ATCC
74408)
Construction of the final recombinant vector (pPDH9K),
transformation into P. pastoris strain SMD1168, growth of
the cells and expression of the enzyme activities were de-
[7]
scribed previously. PDH expression in this system is under
the control of the AOX1 promoter with induction of expres-
sion bymethanol. Expression of the endogenous FDH in
this strain is also inducible with methanol.
Sequencing of PDH and PDHmod
Plasmids pBMS2000-PDH and pBMS2000-PDHmod were
sent to the Bristol-Myers Squibb Core Sequencing Facility,
and sequencing was performed using the BigDye terminator
kit and a model 377 DNA sequencing unit (Applied Biosys-
tems, Foster City, CA) using a series of external and internal
primers.
Construction of E. coli Strain Containing PDHmod
and FDH
Figure 5. Heat Stabilityand substrate specificityof
PDHmod and PDH expressed in E. coli, Upper panel
&
PDHmod, lower panel PDH; , specific activitywith keto
A
two-step construction of the expression vector
^
acid 1; specific activitywith
phenylpyruvate.
pBMS2000-PPFDH-PDHmod was employed. The P. pastoris
Adv. Synth. Catal. 2007, 349, 1369 – 1378
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
1373

Ronald L. Hanson et al.
FULL PAPERS
Scheme 2. Oligonucleotides used for amplification of P. pastoris FDH gene. Restriction sites used for subsequent ligation to
pBMS2000 are indicated.
FDH gene was subcloned into expression vector
pBMS2000[13] using oligonucleotide primers containing the
5’ and 3’ end of the P. pastoris FDH gene along with com-
patible restriction endonuclease cleavage sites (see
Scheme 2).
High-fidelityPCR amplification of the P. pastoris FDH
gene was carried out in four 100-mL aliquots, each contain-
ing 1 X TaqPlus reaction buffer (Stratagene, LaJolla, CA),
0.2 mM each deoxynucleotide triphosphate (dATP, dCTP,
dGTP, and dTTP), 0.4 nM each oligonucleotide, 2.5 U Taq-
Plus DNA polymerase (Stratagene), and 10 pg plasmid
DNA containing the cloned P. pastoris FDH gene. The am-
plification conditions included incubation at 948C for 4 min,
followed by25 cycles of incubation at 94 8C for 1 min; 508C
for 1 min; and 728C for 1.5 min, using a Perkin–Elmer
Model 480 thermocycler with autoextension.
The PCR reaction mixture was extracted with an equal
volume of 1:1 phenol:chloroform (GibcoBRL, Gaithersburg,
MD), and centrifuged at 13,000”g for 5 min. The upper
aqueous phase was removed and placed in a new microcen-
trifuge tube. DNA was precipitated byaddition of 0.1
volume 3M sodium acetate and 2 volumes ice-cold ethanol.
After centrifugation at 13,000”g for 5 min, liquid was aspi-
rated from the tube, and the pellet washed with 0.5 mL ice-
cold 70% ethanol. Liquid was aspirated again, and the
pellet was allowed to air dryfor 30 min at room tempera-
ture.
in a shaker for 1 h at 378C and 225 rpm. Colonies with plas-
mid DNA were selected on LB agar plates containing 50 mg/
mL kanamycin sulfate (Sigma Chemicals, St. Louis, MO).
Plasmids with the desired insert were identified bycolony
PCR in capillarytubes using the RapidCycler (Idaho Tech-
nology, Salt Lake City, UT). Each reaction mixture con-
tained 50 mM Tris-HCl (pH 8.3), 4 mM MgCl2, 0.25 mg/mL
bovine serum albumin, 2% sucrose 400, 0.1 mM cresol red,
0.4 nM each primer (Scheme 2), and 2.5 U Taq DNA poly-
merase (Promega Corp., Madison, WI). The reaction mix-
ture was divided into 10-mL aliquots, and pipetted into the
wells of a round-bottom microtiter plate. A kanamycin-re-
sistant colonywas picked using a disposable plastic inocula-
tion needle, swirled into the reaction mixture, and transfer-
red to LB-kanamycin agar. Each reaction mixture aliquot
was drawn into a 30 mL capillarytube, and the tube was
flame-sealed at both ends. Cells were lysed and DNA dena-
tured byincubation at 94 8C for 30 seconds; amplification
was performed using 30 cycles of incubation at 948C for 0
seconds; 408C for 0 seconds, and 728C for 60 seconds using
a RapidCycler Thermocycler (Idaho Techonolgies, Salt Lake
City, UT). Samples were electrophoresed on a 1.0% TAE
agarose gel for 2 h at 100 V. Seven samples out of 17 tested
showed a strong band at 1100 base pairs. One colonycon-
taining this plasmid (referred to herein as pBMS2000-
PPFDH) was chosen for the next step in the plasmid con-
struction.
Amplified DNA was digested with 20 units each of BspHI
and BamHI for 3 h at 378C in a total volume of 50 mL. In
parallel, the pBMS2000 vector (2 mg) was digested with
BspHI and BamHI. The digested samples were electrophor-
esed on a 1.0% TAE agarose gel for 2 h at 100 V. The
bands corresponding to the FDH gene (1100-base pair frag-
ment) and linearized vector (4700-base pair fragment) were
separatelyexcised from the gel and purified using the QIA-
quick Gel Extraction Kit (Qiagen, Chatsworth, CA). The
concentrations of the isolated fragments were estimated by
electrophoresis against the low molecular weight mass
ladder (Invitrogen Corp., Carlsbad, CA) and ligated with
1 U T4 DNA ligase (Invitrogen) in a 5:1 (insert:vector)
molar ratio in a total volume of 10 mL at 228C for 2 h. DNA
was precipitated byaddition of 15 mL H2O and 250 mL 1-bu-
tanol, and pelleted at 13,000”g in a microcentrifuge for
5 min. Liquid was removed byaspiration, and the DNA was
dried in a SpeedVac (Savant Instruments, Farmingdale, NY)
for 5 min under low heat. The pellet was resuspended in
5 mL H2O.
PDHmod refers to a modified Thermoactinomycetes inter-
medius phenylalanine dehydrogenase that differs from the
published DNA sequence[8] bya change of the last two
amino acids and an additional 12 amino acids at the carbox-
yl terminus. This change was introduced into plasmid
pPDH9K/10 which was subsequentlytransformed into P.
pastoris SMD1168 (deposited as strain ATCC 74408).[14]
Oligonucleotide primers containing the 5’- and 3’-ends of
the PDHmod gene along with compatible restriction endo-
nuclease cleavage sites were prepared, as shown in
Scheme 3.
Reaction conditions for amplification and purification of
the PDHmod byPCR were identical to those used for the P.
pastoris FDH gene except chromosomal DNA prepared
from ATCC 74408 was included as template for the reac-
tion. The resulting fragment was digested with 20 units each
of NdeI and SmaI for 1 h at 258C, followed by2 h at 37 8C,
in a total volume of 50 mL. In parallel, a version of the
pBMS2000 vector with an NdeI site at the initiation codon
(2 mg) was digested with NdeI and SmaI using identical con-
ditions. The digested samples were separatelyelectrophor-
esed on a 1.0% TAE agarose gel for 2 h at 100 V. The
bands corresponding to the PDHmod gene (1200-base pair
fragment) and linearized vector (4700-base pair fragment)
were excised from the gel and purified using the QIAquick
Gel Extraction Kit (Qiagen). Ligation of the two fragments,
The resuspended DNA was transformed byelectropora-
tion into 0.04 mL E. coli DH10B competent cells (Invitro-
gen) at 25 mF and 250 W. SOC medium was immediately
added (0.96 mL; SOC=0.5% yeast extract, 2% tryptone,
10 mM NaCl, 2.5 mM KCl, 10 mM MgCl2, 10 mM MgSO4,
and 20 mM glucose per liter), and the cells were incubated
1374
asc.wiley-vch.de
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2007, 349, 1369 – 1378

Amino Acid Intermediate for the Dipeptidyl Peptidase IV Inhibitor Saxagliptin
FULL PAPERS
Scheme 3. Oligonucleotides used for amplification of T. intermedius PDHmod gene.
transformation of E. coli, and screening for colonies contain-
ing inserts with the PDHmod gene (forming pBMS2000-
PDHmod) were performed as described above for PPFDH.
For construction of pBMS2000-PPFDH-PDHmod,
pBMS2000-PDHmod (2 mg) was cleaved with 10 U each
HindIII and SmaI in a 50-mL reaction for 1 h at 258C, fol-
lowed by1 h at 37 8C. Ten units of T4 DNA polymerase (In-
vitrogen) and 2 mL of a 2.5 mM mixture of all four deoxyri-
bonucleoside triphosphates were added and the sample was
incubated at 118C for 20 min. The reaction was electrophor-
esed on a 1.0% TAE agarose gel for 2 h at 100 V. The 1800-
base pair fragment was excised and isolated using the QIA-
quick Gel Extraction Kit (Qiagen). This fragment contains,
in order, the tac promoter, groES gene, and the PDHmod
gene (as a transcriptional fusion). Next, pBMS2000-PPFDH
(2 mg) was digested with 10 units restriction endonuclease
Sma I in a 50 mL volume for 2 h at 258C, then treated with
0.4 U shrimp alkaline phosphatase (United States Biochemi-
cals, Cleveland, OH) for 1 h at 378C. Plasmid DNA was
electrophoresed for 2 h at 100 V on a 1.0% TAE agarose
gel, isolated, and extracted with the QIAquick kit. The two
fragments were ligated with 1 U T4 DNA ligase in a 6.5:1
(insert:vector) molar ratio at 168C for 4 h in a 10 mL final
volume. After 1-butanol extraction and centrifugation, the
DNA was transformed into electrocompetent DH10B cells.
Kanamycin-resistant colonies were screened for the pres-
ence of the PDHmod gene with the two PDHmod-specific
primers as previouslydescribed for FDH. A second round
of PCR screening was conducted byusing a DNA primer
homologous to the 5’ end of the PPFDH gene and an inter-
nal primer (antisense) homologous to the PDHmod gene.
Onlythose constructs able to support amplification of a
1400-base pair fragment possessed the two genes in the
same orientation. One such plasmid was found and the ori-
entation confirmed bydiagnostic restriction digestion with
KpnI, which gave the expected fragments of 5422 and 1826
base pairs. This plasmid was designated “pBMS2000-
PPFDH-PDHmod.”
Expression of both genes was induced byaddition of filter-
sterilized 1M isopropyl b-d-1-thiogalactopyranoside(IPTG)
ACHTREUNG
to a final concentration of 35 mM and the fermentation was
continued for 24–48 h. Cells were pelleted bycentrifugation
at 6,500”g for 5 min, washed once with an equal volume of
50 mM ammonium formate pH 7.0, and repelleted. Cells
were stored frozen at À208C or used immediately. The
pellet was resuspended in 50 mM ammonium phosphate,
pH 7.0 at 10 mL/g wet cell weight and sonicated 3”15 sec-
onds using a Fisher Scientific Model 50 Sonic Dismembrator
(Fisher Scientific, Pittsburgh, PA), power setting 15 with a
microtip. Debris was pelleted bycentrifugation at 13,000” g
for 5 min at room temperature.
Expression was examined by sodium dodecyl sulfate-poly-
acrylamide gel electrophoresis (SDS-PAGE). One mL of the
cell extract was mixed with 5 mL of 4X NuPAGEꢁ LDS
buffer (Invitrogen) and brought up to 19 mL with distilled
water. Samples were heated at 708C for 10 min. One mL of
a 1M dithiothreitol solution was added to the mixture and
10 mL applied to a 10% NuPAGEꢁ Bis-Tris polyacrylamide
mini-gel. Electrophoresis was carried out at 200 V for 50–
60 min and the gel stained in a solution consisting of 0.1%
(w/v) Coomassie Blue (Sigma), 40% (v/v) ethanol, and 10%
(v/v) acetic acid. The gel, immersed in the stain, was heated
in a microwave oven until boiling was evident, then shaken
at 40 rpm on an orbital shaker for 15 min. The gel was
washed thoroughlywith deionized water and covered with
destaining solution (GelClearꢁ; Invitrogen). The solution
was again heated just to the point of boiling and shaken
gentlyfor at least 2 h. Two prominent bands at M 43,000
r
and 40,000 were seen upon induction, corresponding to the
expected molecular weight of the subunits of FDH and
PDHmod. Samples were also found to possess both FDH
and PDH activities when tested for reductive amination as
described below. This recombinant E. coli strain was given
the internal designation of SC 16496.
Production of PDHmod/FDH by E. coli SC16496 in
a 15-L Fermentation
Expression of FDH and PDHmod
One frozen vial of SC16496 was thawed and the entire con-
tents (1.5–1.8 mL) were transferred to a 4-L flask containing
1 L of MT5 medium. The inoculum flask was incubated at
308C for ca. 24 h and 250 rpm. The optical densityof the re-
sulting inoculum (OD600) at this time was ca. 5.5–5.7 AU/cm.
Based on a desired starting OD600 of 0.35 AU/cm for the fer-
mentor stage, the appropriate quantityof broth was then
used to inoculate a 23-L tank containing 15 L of the produc-
tion medium, MT5M1. The modified MT5M1 medium was
composed of 4% glycerol, 1.85% Tastone 154 (Sensient Bi-
onutrients), 2% Hy-Pea (Quest International), 0.125%
(NH4)2SO4, 0.6% Na2HPO4, and 0.04% UCON LB625 anti-
foam (Dow Chemical). The pH of the medium was adjusted
to 7.4 with NaOH. Solutions of MgSO4 heptahydrate and
pBMS2000-PPFDH-PDHmod was transformed into Escheri-
chia coli JM110. In shake-flasks studies, JM110(pBMS2000-
PPFDH-PDHmod) was grown for 18 h at 288C, 250 rpm in
MT5 medium (2.0% Yeastamin, 4.0% glycerol, 0.6%
sodium phosphate [dibasic], 0.3% potassium phosphate
[monobasic], 0.125% ammonium sulfate, 0.0256% magnesi-
um sulfate [heptahydrate; added post-autoclaving from a
sterile 1M solution], and 50 mg/mL kanamycin sulfate
[added post-autoclaving from a filter-sterilized 50 mg/mL so-
lution]). The optical densityat 600 nm (OD 600) was recorded
and cells sufficient to give a starting OD600 of 0.35 AU/cm
were added to fresh MT5/kanamycin medium. Flasks were
shaken at 250 rpm, 288C until the OD600 was 0.8–1.0 AU/cm.
Adv. Synth. Catal. 2007, 349, 1369 – 1378
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
1375

Ronald L. Hanson et al.
FULL PAPERS
kanamycin, prepared separately and filter-sterilized, were
added to the tank medium prior to inoculation to yield final
A solution (7024 mL) containing ammonium formate
(442.5 g, 7.017 mol) and keto acid 1 (786.7 g, 3.510 mol) was
concentrations of 0.0246% and 0.005%, respectively. The brought to pH 8.0 with 276 mL concentrated ammonium hy-
fermentation process conditions for the tank containing 15 L
of medium were as follows: temperature 308C; pH control
at or above pH 7.0 with ammonium hydroxide only; agita-
tion 500 rpm (tip speed 130 m/min); aeration 15 Lpm; back
pressure 690 mbar; foam controlled bythe addition of
UCON LB625 on demand.
At the target OD600 of ca. 0.8 AU/cm, (1.5–3.0 h into the
run), a filter-sterilized IPTG solution was added to yield a
final concentration of 35 mM. The run was considered com-
pleted after glycerol was exhausted and growth ceased, and
an OD600 of 50–53 AU/cm was usuallyachieved by35 h. The
whole broth was cooled to 4–108C and cells were recovered
bycentrifugation followed bya wash with 50 mM ammoni-
um formate pH 7.0 buffer. A total of 560–600 g of cell paste
was then recovered and stored at À758C.
droxide. NAD (9.834 g, 14.82 mmol) and dithiothreitol
(2.163 g, 14.027 mmol) were added, then the solution was
added to the bioreactor containing the Pichia pastoris ex-
tract. The solution was maintained at 408C and stirred at
150 rpm. Portions of concentrated ammonium hydroxide
were added at 0 (44 mL), 3 (25 mL) and 18 (27 mL) h after
the start of the reaction to adjust the pH to 8.0. After 25 h,
the solution contained 818.9 g (3.637 mol, 100% yield) of
amino acid 2 as measured byHPLC analysis and no detecta-
ble keto acid 1 or (R)-enantiomer of the amino acid.
To reduce the level of ammonia, the reaction mixture,
14 L, was heated to boiling and sparged with steam, with n-
butanol (1.3 L) added to control foaming. When foaming
subsided, the mixture was cooled to 378C and adjusted to
pH 3.0 with 88% formic acid (579 g). Precipitated protein
was removed byfiltration. The filtrate was diluted to 80 L
with water, applied to a 25 i.d. ” 30 hcm column of Dowex
50W-X8 (H+), 50–100 mesh, and the column was washed
with water. No amino acid 2 was detected in the effluent
(TLC, silica gel, butanol-acetic acid-water, 4:1:1, ninhydrin,
Rf =0.33). The product was eluted from the resin with 1.5M
ammonium hydroxide, collecting 4 L fractions, and was de-
tected in fractions 13–20 (TLC). These were combined and
concentrated to a solid on a rotaryevaporator to give 804 g
of 2 with a potencyof 94.3% byHPLC. The isolated yield
from the keto acid was thus 96%.
Screening of Amino Acid Dehydrogenases
Reaction mixtures contained 0.15 M ammonium formate,
0.05 M keto acid 1, 1 mM NAD, 1 mM dithiothreitol, 1U
FDH (Boehringer Mannheim), and an amino acid dehydro-
genase in a final volume of 1 mL. The pH was adjusted to
8.5 with NH4OH. Beef liver glutamate dehydrogenase, Ba-
cillus subtilis alanine dehydrogenase, and S. ureae phenylala-
nine dehydrogenase were from Sigma. Rhodococcus phenyl-
alanine dehydrogenase was from Jülich. Leucine dehydro-
genase was partiallypurified from
Bacillus sphaericus
ATCC 4525. Phenylalanine dehydrogenase from Thermoac-
tinomyces intermedius expressed in Pichia pastoris was pre-
[7]
pared previouslyas dried cells.
Solutions were incubated
Pilot Plant Scale Reductive Amination
at 308C for 20 h, then analyzed by HPLC.
Recombinant Pichia pastoris frozen cells (114.7 kg) were
thawed and added to deionized water (340.4 kg) containing
ammonium formate (1.45 kg) in an 800-L vessel. After thaw-
ing, the cells were suspended bystirring, then adjusted to
pH 7 with concentrated NH4OH. Cells were disrupted by
passages through a microfluidizer at 14000 psig, and the cell
extract was clarified bycentrifugation with a Sharples cen-
trifuge.
Transamination using Branched Chain Transaminase
The solution contained 0.10 M sodium glutamate, 0. 05 M 1
(neutralized with 0.05 M NaOH), 0.1 mM pyridoxal phos-
phate, and 1 mg branched chain transaminase (Biocatalytics)
in a final volume of 1.0 mL in 50 mM potassium phosphate
buffer pH 8.0. The solution was incubated in a microfuge
tube at 378C for 68 h, then analyzed by HPLC. The solution
contained 5.53 mg/mL (49.2% yield) 2 and 7.05 mg/mL re-
maining 1.
Deionized water (196.1 kg), keto acid
1
(40 kg,
178.4 mol), and ammonium formate (22.3 kg, 353.6 mol)
were charged into an 800-L vessel. The pH was adjusted to
8 with concentrated NH4OH, then NAD (499.5 g, 0.753
mol), dithiothreitol (111 g, 0.720 mol), 30.4 kg deionized
water, and 407 kg clarified cell extract containing 3467 kU
formate dehydrogenase and 646 kU phenylalanine dehydro-
genase (assayC) were added. With slow stirring, the solu-
tion was maintained at 408C, and pH 8.0. After 49 h, the so-
lution contained 0.44 kg keto acid 1 and 36.95 kg of amino
acid 2 as measured byHPLC analysis (99% conversion). In-
soluble material was removed from the reaction mixture by
microfiltration (0.2 micron ceramic Membrane, 88 square
feet), followed bythree diafiltrations (133.3 kg water each)
for a total filtrate of 978 kg. Protein was then removed by
ultafiltration (5000 MW cut-off Millipore Prostak Mem-
brane, 50 square feet), followed bydiafiltration (twice with
20 kg deionized water). The filtrate (1013 kg) was adjusted
to pH 9.6 with 25% sodium hydroxide solution for concen-
tration and removal of ammonia. Water and ammonia were
removed bydistillation (jacket temperature 110–130 8C) to
Reductive Amination using an Extract from
Recombinant Pichia pastoris
Recombinant Pichia pastoris frozen cells (2.25 kg) were
added to deionized water (6.75 L) containing ammonium
formate (28.65 g, 0.454 mol). After thawing, the cells were
suspended using a Janke and Kunkel Ultra-turrax T25 ho-
mogenizer, then adjusted to pH 7 with concentrated
NH4OH and cooled with crushed ice to give a 25% w/v cell
suspension in 50 mM ammonium formate. Cells were dis-
rupted bytwo passages through
a microfluidizer at
12000 psi, and cell debris was removed bycentrifugation at
20,000”g and 48C. Supernatant (7024 mL) containing
230998 U (assayA) or 124641 U (assayB) phenylalanine de-
hydrogenase and 80080 U formate dehydrogenase was
added to a 16-L vessel of a New Brunswick Scientific Bio-
flow IV bioreactor.
1376
asc.wiley-vch.de
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2007, 349, 1369 – 1378

Amino Acid Intermediate for the Dipeptidyl Peptidase IV Inhibitor Saxagliptin
FULL PAPERS
give 407 kg solution containing amino acid 2 (~100 g/L)
which was converted to the BOC derivative 3.
CuSO4 in water), flow rate was 1 mLminÀ1, detection was at
240 nm, temperature was 408C, and injection volume was
10 mL. Retention times were (S)-enantiomer 3.2 min, (R)-en-
antiomer 11.2 min, keto acid 1 5.2 min.
Reductive Amination using an Extract from
Recombinant Escherichia coli SC16496
JM110[pBMS2000-PPFDH-PDHmod]
Enzyme Assays
Phenylalanine dehydrogenase assay A contained in 1.0 mL
at 408C: 0.4 mM NADH, 5 mM sodium phenylpyruvate,
0.75 M NH4OH adjusted to pH 8.75 with HCl. Absorbance
decrease was monitored at 340 nm. Phenylalanine dehydro-
genase assayB contained in 1.0 mL at 40 8C: 1 mM NAD,
10 mM l-phenylalanine, 0.1 MK2HPO4 adjusted to pH 10.0
with 1 N NaOH. Absorbance increase was monitored at
340 nm. Phenylalanine dehydrogenase assay C contained in
1.0 mL at 408C: 0.4 mM NADH, 50 mM keto acid 1 (dis-
solved in 1 equivalent NaOH solution), 0.75M NH4OH ad-
justed to pH 8.75 with HCl. Absorbance decrease was moni-
tored at 340 nm. The formate dehydrogenase assay con-
tained in 1.0 mL at 408C: 1 mM NAD, 100 mM ammonium
formate, 100 mM potassium phosphate buffer, pH 8.0. Ab-
sorbance increase was monitored at at 340 nm. Enzyme ac-
tivityunits were calculated as mmol/min based on the rates
of absorbance change.
Frozen cells of recombinant Escherichia coli 25% w/v in
50 mM ammonium formate pH 7 were suspended using a
Janke and Kunkel Ultra-turrax T8 homogenizer, then
cooled with crushed ice and disrupted bytwo passages
through a microfluidizer at 12000 psi. Cell debris was re-
moved bycentrifugation at 20,000” g and 48C. Supernatant
(266 mL) containing 2456 U (assayA) or 768 U (assayC)
phenylalanine dehydrogenase and 8801 U formate dehydro-
genase was added to a 1-L bottle.
A
solution (266 mL) containing ammonium formate
(16.74 g, 0.2654 mol) and keto acid 1 (29.76 g, 0.1327 mol)
was brought to pH 8.0 with 12.7 mL concentrated ammoni-
um hydroxide. NAD (372 mg, 0.561 mmol) and dithiothrei-
tol (81.8 mg, 0.530 mmol) were added, then the solution was
added to the bottle containing the Escherichia coli extract.
The bottle was maintained at 408C on a shaker at 40 rpm.
Concentrated ammonium hydroxide was added periodically
to maintain pH 8.0. After 38 h, the solution contained 31.5 g
(0.140 mol, 100% yield) of amino acid 2 as measured by
HPLC analysis and no (R)-enantiomer of the amino acid.
Heat Stability of PDHmod
The heat stabilityand substrate specificityof PDH and
PDHmod are compared in Figure 5. For this experiment,
25% w/v extracts of E. coli SC16496 (FDH/PDHmod) and
SC16487 (FDH/PDH) in 50 mM ammonium formate pH 7
were microfluidized and centrifuged at 20000”g. One mL
samples of the centrifuged extract were incubated for 1 h at
0, 40, 50, 60 or 708C, then assayed for activity with phenyl-
pyruvate or keto acid 1 as substrate. Samples of the untreat-
ed centrifuged extracts were also diluted 1:10 with water
and run on an Agilent 2100 bioanalyzer protein 200 labchip
to separate and quantitate the PDH subunit bands. The
PDH concentration was 2.99 mgmLÀ1 and the PDHmod
concentration was 5.17 mgmLÀ1 in the centrifuged extracts
(mean of 4 samples each). Specific activities are expressed
as units per mg PDH or PDHmod protein in the subunit
band.
Preparation of (S)-N-BOC-3-
Hydroxyadamantylglycine (3)
The reaction mixture was filtered through a 0.1 micron filter
for clarification and subsequentlythrough a 5 kD ultrafilter
to remove soluble proteins. The filtrate was charged to a 1-
L reaction flask equipped with a distillation apparatus and a
receiving flask The pH was adjusted to 9.6 with 25 wt%
NaOH, then was heated to distill off excess water and
remove ammonia. The solution, containing 25 g of 2, was
concentrated to 200 mL and K3PO4 (70 g, 0.33 mol) was
added. The reaction mixture was cooled to room tempera-
ture then treated with BOC2O (26 g, 0.12 mol) in THF
(200 mL). The reaction was stirred at room temperature
until deemed complete byHPLC (2 h). The bottom layer
was discarded, and water (150 mL) was added to the top
layer, followed by isopropyl acetate (200 mL). This mixture
was adjusted to pH 2 with conc. HCl (15 mL). After mixing,
the aqueous layer was separated and the organic layer was
concentrated to 150 mL bydistillation at 1 atmosphere. The
organic layer was stirred, cooled to room temperature and
heptane (150 mL) was added over 30 min. The suspension
was stirred for another h, then the product was filtered out
and dried at 408C under vacuum to obtain 3 (32 g, 88%
yield).
References
[1] B. Gallwitz, Treat. Endocrinol. 2005, 4, 361–370.
[2] L. L. Nielsen, Drug Discovery Today 2005, 10, 703–
710.
[3] E. M Sinclair, D. J. Drucker, Curr. Opin. Endocrinol.
Diabet. 2005, 12, 146–151.
[4] D. J. Augeri; J. A. Robl, D. A. Betebenner, D. R.
Magnin, A. Khanna, J. G. Robertson, A. Wang, L. M.
Simpkins, P. Taunk, Q. Huang, S.-P. Han, B. Abboa-
Offei, M. Cap, L. Xin, L. Tao, E. Tozzo, G. E. Welzel,
D. M. Egan, J. Marcinkeviciene, S. Y. Chang, S. A.
Biller, M. S. Kirby, R. A. Parker, L. G. Hamann, J.
Med. Chem. 2005, 48, 5025–5037.
[5] R. L. Hanson, J. Singh, T. P. Kissick, R. N. Patel, L. J.
Szarka, R. H Mueller, Bioorg. Chemistry 1990, 18, 116–
130.
HPLC Analysis of Amino Acid 2
Samples of the reductive amination reaction mixture were
diluted with water to about 2 mg/mL and placed in a boiling
water bath for 1 min to stop the reaction and precipitate
proteins. After cooling, the samples were filtered through
0.2 micron nylon filters into HPLC vials. Samples were ana-
lyzed with a Regis Davankov Ligand Exchange 15”0.46 cm
column. The mobile phase was 25% methanol/75% (6 mM
Adv. Synth. Catal. 2007, 349, 1369 – 1378
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
1377

Ronald L. Hanson et al.
FULL PAPERS
[6] R. L. Hanson, M. D. Schwinden, A. Banerjee, D. B.
Brzozowski, B-C. Chen, B. P Patel, C. G. McNamee,
G. A. Kodersha, D. R. Kronenthal, R. N. Patel, L. J.
Szarka, Bioorg. Med. Chem. 1999, 7, 2247–2252.
[7] R. L. Hanson, J. M. Howell, T. L. LaPorte, M. J. Dono-
van, D. L. Cazzulino, V. Zannella, M. A. Montana,
V. B. Nanduri, S. R. Schwarz, R. F. Eiring, S. C.
Durand, J. M. Wasylyk, W. L. Parker, M. S Liu, F. J.
Okuniewicz, B-C. Chen, J. C. Harris, K. J. Natalie Jr.,
K. Ramig, S. Swaminathan, V. W. Rosso, S. K. Pack,
B. T. Lotz, P. J. Bernot, A. Rusowicz, D. A. Lust, K. S.
Tse, J. J. Venit, L. J. Szarka, R. N. Patel, Enzyme
Microb. Technol. 2000, 26, 348–358.
[10] J. J. Allais, A. Louktibi, J. Baratti, Agric. Biol. Chem.
1983, 47, 2547–2554.
[11] J. D. Godfrey, R. T. Fox, F. G. Buono, J. Z. Gougoutas,
M. F. Malley, J. Org. Chem. 2006, 71, 8647–8650.
[12] T. C. Vu, D. B. Brzozowski, R. Fox, J. D. Godfrey, R. L.
Hanson, S. V. Kolotuchin, J. A. Mazzullo Jr., R. N.
Patel, J. Wang, K. Wong, J. Yu, J. Zhu, D. R. Magnin,
D. J. Augeri, L. G. Hamann, (Bristol-Myers Squibb
Company), PCT Int. Appl. WO 2004/052850 A2, 2004.
[13] S. W. Liu, T. J. Franceschini, (Bristol-Myers Squibb
Company), US Patent 6,068,991, 2000.
[14] R. Hanson, M. J. Donovan, S. Goldberg, P. A. Jass, W-S
Li. , R. Patel, K. Ramig, L. J. Szarka, J. J. Venit, (Bris-
tol-Myers Squibb Company), US Patent 6,140,088,
2000.
[8] H. Takada, T. Yoshimura, T. Ohshima, E. Nobuyoshi,
K. Soda, J. Biochem. 1991, 109, 371–376.
[9] C. T. Hou, R. N. Patel, A. I. Laskin, N. Barnabe, Arch.
Biochem. Biophys. 1982, 216, 296–305.
1378
asc.wiley-vch.de
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2007, 349, 1369 – 1378