Optimized design and synthesis of chemical dimerizer substrates for detection of glycosynthase activity via chemical complementation

Article abstract of DOI:10.1016/j.bmc.2006.06.034

Glycosynthases catalyze the formation of a glycosidic bond between a glycosyl fluoride donor substrate and a glycosyl acceptor substrate with high yield, thus providing a valuable approach for the synthesis of carbohydrates and glycoconjugates. Chemical c

Full text of DOI:10.1016/j.bmc.2006.06.034

Bioorganic & Medicinal Chemistry 14 (2006) 6940–6953
Optimized design and synthesis of chemical dimerizer substrates
for detection of glycosynthase activity via chemical
complementation
Haiyan Tao, Pamela Peralta-Yahya, Hening Lin and Virginia W. Cornish*
Department of Chemistry, Columbia University, 3000 Broadway, New York, NY 10027, USA
Received 12 January 2006; revised 13 June 2006; accepted 19 June 2006
Available online 25 July 2006
Abstract—Glycosynthases catalyze the formation of a glycosidic bond between a glycosyl fluoride donor substrate and a glycosyl
acceptor substrate with high yield, thus providing a valuable approach for the synthesis of carbohydrates and glycoconjugates.
Chemical complementation can be used to link glycosynthase activity to the transcription of a reporter gene in vivo, providing a
selection for the directed evolution of glycosynthase enzymes with improved properties. In this approach, glycosynthase activity
is detected as covalent coupling between a small molecule disaccharide acceptor substrate and a small molecule disaccharide a-fluoro
donor substrate. Here we report the optimized design and synthesis of these small molecule substrates. These optimized substrates
are shown to give a robust, glycosynthase-dependent transcriptional read-out in the chemical complementation assay. The full syn-
thesis and characterization of these substrates are reported for the first time. These optimized chemical dimerizer substrates should
allow the potential of chemical complementation for the directed evolution of glycosynthases with diverse substrate specificities and
improved properties to be fully realized.
ꢀ 2006 Elsevier Ltd. All rights reserved.
1. Introduction
and to expand their substrate specificity.1,5,6 However,
the application of directed evolution to this class of en-
zymes has been limited because the reaction product is
not inherently screenable or selectable. The Withers
group developed an on-plate endo-cellulase coupled
screening assay for the directed evolution of the Agro-
bacterium sp. Abg:E358G glycosynthase. After combin-
ing mutations from two rounds of random mutagenesis,
a variant was identified with a 27-fold improvement in
activity and expanded substrate specificity.5,6 This
screen, however, depends on the identification of a cou-
pling enzyme that will cleave only the product of the
glycosynthase catalyzed reaction to release a chromo-
genic product, and hence may be difficult to extend to
glycosynthases with different substrate specificities.5,6
In addition, screens are limited to smaller library sizes
than selections.
Recently we demonstrated that chemical complementa-
tion can provide a high-throughput assay for the direct-
ed evolution of glycosynthase enzymes.1 Glycosynthases
are artificial enzymes derived from retaining glycosi-
dases, which have the ability to catalyze the formation
of a glycosidic linkage using a glycosyl fluoride donor.2
Compared to glycosyltransferases or natural glycosi-
dases, there are several advantages to the use of glyco-
synthases for carbohydrate synthesis.3,4 The glycosyl
fluoride donors are straightforward to synthesize com-
pared to the nucleotide diphosphate donors required
for natural glycosyltransferases. Glycosynthases gener-
ally give the product in higher yield than simple use of
the glycosidase. Glycosynthases can tolerate a broader
range of substrates than natural glycosyltransferases.3,4
Directed evolution provides an obvious route to
improve the activity of these glycosynthase variants
The chemical complementation assay developed by our
laboratory can be used to provide a general selection
strategy for evolving endo-glycosynthases.1,7 Chemical
complementation detects enzyme catalysis of bond
formation or cleavage reactions based on covalent cou-
pling of two small molecule ligands in vivo (Fig. 1).
The heterodimeric small molecule reconstitutes a tran-
scriptional activator, turning on the transcription of a
Keywords: Chemical complementation; Glycosynthase; Substrate
synthesis; High-throughput assay.
*
Corresponding author. Tel.: +1 212 854 5209; fax: +1 212 932
1289; e-mail: vc114@columbia.edu
Present address: Department of Biological Chemistry and Molecular
Pharmacology, Harvard Medical School, Boston, MA 02115, USA.
0968-0896/$ - see front matter ꢀ 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmc.2006.06.034

H. Tao et al. / Bioorg. Med. Chem. 14 (2006) 6940–6953
6941
Figure 1. Chemical complementation provides a high-throughput assay for glycosynthase activity. Chemical complementation detects enzyme
catalysis of bond formation or cleavage reactions based on covalent coupling of two small molecule ligands. The heterodimeric small molecule
reconstitutes a transcriptional activator, turning on the transcription of a downstream reporter gene. Here, a dexamethasone (Dex)–methotrexate
(Mtx) yeast three-hybrid system is used. Glycosynthase activity is detected as formation of a glycosidic linkage between a Mtx a-fluoride donor and a
Dex acceptor and activation of a LEU2 reporter gene.
downstream reporter gene. Bond formation is detected
as activation of an essential reporter gene; bond cleav-
age, repression of a toxic reporter gene. To extend the
chemical complementation assay to detect glycosynth-
ase activity, a methotrexate-disaccharide-fluoride donor
(Mtx-Lac-F) donor substrates, 4 and 5, and the first full
synthesis and characterization of both the Dex-Cel and
Mtx-Lac-F substrates. This improved design and full
characterization of the Dex-Cel and Mtx-Lac-F sub-
strates should facilitate further applications of chemical
complementation for the directed evolution of glyco-
synthase enzymes.
and
a
dexamethasone-disaccharide acceptor were
synthesized, and bond formation was detected in a
dexamethasone–methotrexate (Dex–Mtx) yeast three-
hybrid system1 (Fig. 1). In this system, a heterodimeric
Dex–Mtx small molecule is synthesized by the glyco-
synthase, thus dimerizing the hormone-binding domain
of the glucocorticoid receptor (GR), which binds to
Dex, and dihydrofolate reductase (DHFR), which
binds to Mtx. DHFR is fused to a DNA-binding
domain (DBD), and GR is fused to a transcription
activation domain (AD), such that Dex–Mtx effectively
reconstitutes the transcriptional activator (DBD-AD)
and increases transcription of a downstream reporter
gene. The reporter gene is LEU2, which allows for a
growth selection in the absence of leucine. Using the
LEU2 selection, we demonstrated that chemical com-
plementation can be used to read-out glycosynthase
activity. In addition, a Cel7B:E197S variant with a
five-fold increase in glycosynthase activity was isolated
from a Glu197 saturation library.1
1.1. Design and synthesis of the glycosynthase substrates
The glycosynthase substrates must incorporate the Dex
and Mtx ligands for use in the chemical complementa-
tion assay, yet still be efficient substrates for the
Cel7B:E197A glycosynthase variant.1,7 The structures
of the substrates designed for use in this study are shown
in Scheme 1. The design was based on the known
substrate specificities of the H. insolens Cel7B glycosi-
dase and Cel7B:E197A glycosynthase.8,9 Both kinetic
characterization and high-resolution structures of this
and related endo-glycosidases suggest that there are five
subsites in the active site that accommodate five glucose
units, and four of the five subsites (ꢀ2, ꢀ1, +1, and +2)
contribute most of the binding energy.10,11 In addition,
previous studies demonstrated that with lactose fluoride
as the donor substrate the acceptor subsites of
Cel7B:E197A may accommodate both mono- and disac-
charide acceptors and that disaccharides are better
acceptors than monosaccharides for the transglycosyla-
tion reaction.8 We therefore decided to use two
disaccharide compounds as the substrates for the
Cel7B:E197A glycosynthase. Dex–Mtx has proven to
be an efficient CID for reconstitution of a transcription-
al activator in the corresponding yeast three-hybrid sys-
tem.12 In addition, it has been shown that as long as the
linker is sufficiently long, the linker has little effect on the
activity of the Dex–Mtx CID in the yeast three-hybrid
assay.13 Therefore, we used cellobiose with Dex attached
at the anomeric position as the glycosyl acceptor sub-
strate, shown as 1; and lactosyl fluoride with Mtx
attached at the 60 position of the galactose unit as the
glycosyl donor substrates, shown as 4 and 5. Lactose
was chosen to avoid the self-polymerization reaction
of the cellobiose donor. Two Mtx-Lac-F donors, 4 and
5, were tested. The Mtx-Lac-F donor 5 has a shorter
As shown in Scheme 1, we designed the dexamethasone–
cellobiose (Dex–Cel) glycosyl acceptor 1 and methotrex-
ate-lactose-fluoride (Mtx-Lac-F) glycosyl donor 4 as the
glycosynthase substrates. However, as reported in our
first publication, instead of the desired substrate Mtx-
Lac-F 4, we obtained methotrexate-cellobiose-fluoride
(Mtx-Cel-F) 2. The activity of a known glycosynthase,
Humicola insolens Cel7B:E197A, was detected and a
directed evolution experiment was carried out using
the Dex-Cel 1 and Mtx-Cel-F 2 substrates.1 Although
substrates Dex-Cel 1 and Mtx-Cel-F 2 worked in the
chemical complementation assay, the synthesis of Mtx-
Cel-F 2 suffered from poor yields. A side elimination
reaction occurred at the final deprotection step in the
synthesis of 2, resulting in the by-product 3 and signifi-
cantly affecting the overall yield in which 2 could be
obtained. Thus, here we report an improved design
and synthesis of two methotrexate-lactose-fluoride

6942
H. Tao et al. / Bioorg. Med. Chem. 14 (2006) 6940–6953
Scheme 1. Structure of the glycosynthase substrates used for the chemical complementation assay.
linker between the Mtx and Lac portions of the sub-
strate and provides a more efficient synthesis. However,
based on previous studies on the effect of linker length,
we had some concern that the shorter linker of 5 would
impair its activity in the yeast three-hybrid assay.13
26% yield for two steps. Several standard activators
were tested in an effort to improve the yield for this reac-
tion, but neither silver triflate nor silver zeolite gave the
desired product.14 The direct coupling of Dex to the
glycosyl bromide using either mercury or silver salts as
the activator also did not yield any desired product.
Therefore, Dex was installed to 6 using the Dex-thiol
derivative 713 as the nucleophile and gave 8 in 69% yield.
Finally, global deprotection with sodium methoxide
removed all acetyl protecting groups, and the Dex-Cel
acceptor 1 was prepared in four steps in 18% overall
yield.
Dex-Cel 1 is essentially as originally reported.1 Here we
report the full synthesis and characterization of this sub-
strate. As shown in Scheme 2, the synthesis of Dex-Cel
acceptor 1 was started from the commercially available
cellobiose octa-acetate. The treatment of cellobiose
octa-acetate with 30% hydrogen bromide in acetic acid
for 1 h afforded cellobiosyl-bromide. Then the glycosi-
dation reaction between cellobiosyl-bromide and 6-bro-
mo-1-hexanol was carried out using both mercury
cyanide and mercury bromide as activators to give 6 in
Although we intended Mtx-Lac-F 4 as the donor sub-
strate for the chemical complementation assay, in our
original publication the chemical complementation

H. Tao et al. / Bioorg. Med. Chem. 14 (2006) 6940–6953
6943
Scheme 2. Synthesis of Dex–Cel 1. Reagents and condition: (a) 30% HBr in HOAc, 1 h; (b) HO(CH2)6Br, Hg(CN)2, HgBr2, CH2Cl2, 26% for two
steps; (c) Dex-SH 7, MeOH, DIEA, 69%; (d) MeONa, MeOH, 100%.
assay was carried out with Mtx-Cel-F donor 2 instead.1
The synthesis of Mtx-Cel-F 2 suffered both from a side
reaction in the last deprotection step and poor yield,
thus the synthesis of the donor substrate had to be mod-
ified. As shown in Scheme 3, in our first effort to synthe-
size Mtx-Lac-F 4, a disulfide diamine linker was
installed to the c-carboxylic acid group of the Mtx,
and then its reduction product, the Mtx-thiol compound
13, was attempted for coupling with lactose bromo-
derivative 11 or 12 via a substitution reaction. However,
no product was observed using either N,N-diisopropyl-
ethylamine (DIEA) or sodium methoxide as the base.
In contrast, Mtx-thiol 13 reacted efficiently with dib-
romo-substituted lactose derivative10 when sodium
methoxide was used as the base. The difference in reac-
tivity between compounds 11 and 12 compared with 10
can be explained by nonbonding interactions with the
40 position in the transition state of the SN2 reaction.15
The nonbonding interaction from the axial substituent
of 11 or 12 is greater than that from the equatorial bro-
mine of 10. Since the product 14 led to the final product
Mtx-Cel-F 2, which is also a suitable substrate for glyco-
synthase Cel7B:E197A, we continued the synthesis from
compound 14. However, the compound Mtx-Cel-F 2
was difficult to obtain in pure form as an elimination
side-reaction occurred during the last deprotection step.
During the final deprotection step, a major by-product,
which we tentatively assigned as 3 (Scheme 1), arose.
Not surprisingly given their similarity in overall
chemical structure, it proved difficult to purify Mtx-
Cel-F 2 from by-product 3. Using reverse-phase high
performance liquid chromatography (HPLC), Mtx-
Cel-F 2 could be purified. The total yield for the synthe-
sis of compound 2, however, was significantly lowered
by this purification step (<10% for the last step). Mtx-
Cel-F 2 was synthesized from two components in eight
liner steps in 0.6% overall yield (counted from glutamic
acid a-methyl-c-tert-butyl ester).
Thus, as shown in Scheme 4, we came back to our
original design to synthesize Mtx-Lac-F 4 as the donor
substrate. To carry out the desired substitution reaction,
a better leaving group, p-toluenesulfonate, was intro-
duced to the 60 position of lactose derivative 9, and a
N-tert-butoxycarbonyl-2-aminoethanethiol was used to
attack the 60 position. In addition, a bulky base, potas-
sium tert-butoxide, was used as the base to catalyze the
substitution reaction. As the result, the lactose deriva-
tive 18 was prepared from lactose in 8.9% yield. First,
the 40 and 60 hydroxyl groups of lactose were first selec-
tively protected as the 4-methoxybenzylidene acetal.
Then the remaining hydroxyl groups of lactose were
protected using acetic anhydride and pyridine. Hydroly-
sis of the 4-methoxybenzylidene acetal using 50% triflu-
oroacetic acid (TFA) in dichloromethane gave 9 in 39%
yield. Compound 9 was a mixture of the a- and b-ano-
mers, and the ratio of the a to b anomer product varied.
Recrystallization of 9 from ethyl acetate and ether gave

6944
H. Tao et al. / Bioorg. Med. Chem. 14 (2006) 6940–6953
Scheme 3. Synthesis route toward Mtx-Cel-F 2. Reagents: (a) 4 equiv PPh3, 2 equiv CBr4, pyridine; (b) 1 equiv PPh3, 1 equiv CBr4, pyridine;
(c) 1—NaOMe, DMF, 2—Ac2O, TEA, DMF; (d) pyridine, Ac2O.
Scheme 4. Synthesis of lactose derivative 18. Reagents and conditions: (a) 1-(dimethoxymethyl)-4-methoxy-bezene, p-toluenesulfonic acid,
DMF, 50 ꢁC; (b) pyridine, Ac2O; (c) TFA, CH2Cl2, 39% for three steps; (d) p-tosyl chloride, DABCO, CH2Cl2, 65%; (e) pyridine, Ac2O, 100%;
(f) 1—potassium tert-butoxide, 2-(Boc-amino) ethanethiol, THF, 2—pyridine, Ac2O; (g) TFA, CH2Cl2, 35% for (f) and (g).
the pure a anomer and was used for characterization.
The mixture, however, could be used for the synthesis
of the final product. The primary 60 hydroxyl group
of 9 was then selectively activated using one equivalent
of p-toluenesulfonyl chloride and 1,4-diazabicy-
clo[2.2.2]octane (DABCO) to give 15, followed by pro-
tection of the 40 hydroxyl group of 15 using acetic
anhydride and pyridine to give 16. The toluenesulfonate
group of 16 was then substituted by N-tert-butoxycar-
bonyl-2-aminoethanethiol to give 17. Deprotection of
the Boc group of 17 using TFA gave the lactose deriva-
tive 18.
As shown in Scheme 5, a-methyl protected Mtx 21
was prepared essentially as original reported,12,16–18
from commercially available glutamic acid a-methyl-
c-tert-butyl ester and 4-(methylamino) benzoic acid.
N,N0-dicyclohexylcarbodiimide (DCC) was used as

H. Tao et al. / Bioorg. Med. Chem. 14 (2006) 6940–6953
6945
Scheme 5. Synthesis of Mtx-Lac-F 4. Reagents: (a) DCC, DIEA, CH2Cl2, 53%; (b) bromomethyl 2,4-pteridinediamine HBr–isopropanol complex,
DMAc, 75%; (c) TFA, CH2Cl2, 76%; (d) PyBOP, tert-butyl 6-aminohexanoate, DIEA, CH2Cl2, DMF, 65%; (e) TFA, CH2Cl2, 93%; (f) PyBOP,
CH2Cl2, DMF, DIEA, compound 18, 60%; (g) HF in pyridine, 62%; (h) LiOH, MeOH, H2O, 55%.
the coupling reagent in the synthesis of 19 to avoid
the coupling reaction between the carboxylic acid
group of glutamic acid a-methyl-c-tert-butyl ester with
the secondary amine of 4-methylamino benzoic acid.16
Then a-methyl-c-tert-butyl protected Mtx 20 was
prepared by a substitution reaction of 19 to 6-bro-

6946
H. Tao et al. / Bioorg. Med. Chem. 14 (2006) 6940–6953
momethyl-2,4-pteridinediamine. Deprotection of the
tert-butyl protecting group of 20 with 50% TFA in
dichloromethane gave 21 in 30% yield for three steps.
The coupling reaction of compound 21 and t-butyl 6-
aminohexanoate using (benzotriazol-1-yloxy)tripyrroli-
dinophosphonium hexafluorophosphate (PyBOP) as
the coupling reagent, and the following deprotection
reaction using TFA resulted in compound 23 in 60%
yield for two steps. The compound 24 was then syn-
thesized from 18 and 23 in 60% yield using PyBOP
as the coupling reagent. The anomeric O-acetyl of
24 was converted to the a-fluoride compound 25 in
62% yield using a hydrogen fluoride–pyridine complex.
Global deprotection of 25 using lithium hydroxide
gave the desired Mtx-Lac-F donor substrate 4. Mtx-
Lac-F 4 was synthesized from two components in 10
linear steps in 1.8% overall yield (counted from
lactose).
Similar to the synthesis route of Mtx-Lac-F 4, the
a-methyl protected Mtx 21 was coupled to the lactose
derivative 18 in 49% yield using PyBOP as the coupling re-
agent (Scheme 6). Then the anomeric O-acetyl of 26 was
converted to the a-fluoride using a hydrogen fluoride–
pyridine complex, yielding compound 27 in 52% yield.
Global deprotection of 27 using lithium hydroxide gave
the desired Mtx-Lac-F donor substrate 5. Mtx-Lac-F 5
was synthesized from two components in 10 linear steps
in 0.9% overall yield (counted from lactose).
1.2. Chemical dimerizer substrate activities in the chem-
ical complementation assay
Having synthesized the chemical dimerizer substrates
Dex-Cel 1 and Mtx-Lac-F 4 and 5, we next determined
whether they were active in the chemical complementa-
tion assay. If active, the glycosynthase enzyme should
Scheme 6. Synthesis of Mtx-Lac-F 5. Reagents: (a) PyBOP, CH2Cl2, DMF, DIEA, compound 18, 49%; (b) HF in pyridine, 52%; (c) LiOH, MeOH,
H2O, 40%.

H. Tao et al. / Bioorg. Med. Chem. 14 (2006) 6940–6953
6947
be able to couple the two substrates, synthesizing
Dex-Cel-Lac-Mtx and thus reconstituting the transcrip-
tional activator from the DBD-GR and AD-DHFR fu-
sion proteins, here activating transcription of a LEU2
reporter gene (Fig. 1), providing a growth advantage
to yeast three-hybrid cells in the absence of leucine.1
The efficacy of the different chemical dimerizer sub-
strates is determined by carrying out a LEU2 mock
selection in the yeast selection strain V1019Y,19 in which
LexA is used as the DBD, and B42 as the AD. Both the
DBD-DHFR and AD-GR fusion proteins are expressed
from a GAL1 promoter. The DBD-DHFR gene is inte-
grated into the chromosome at the ade4 locus, and the
AD-GR gene is on a 2l plasmid. The LEU2 gene is inte-
grated into the chromosome under the control of six
tandem LexA operators.1,20 In the LEU2 mock selection
assay, a mixture of plasmids encoding a 1:10 ratio of the
Cel7B:E197A glycosynthase to the Cel7B glycosidase
(inactive control) variant was transformed en masse into
the yeast three-hybrid strain V1019Y. The resulting
transformants were incubated in selective media lacking
the appropriate auxotrophs and leucine, and containing
the donor and/or acceptor substrates (Fig. 2) and grown
at 30 ꢁC. After 7 days of growth, the cells were trans-
ferred to non-selective media and harvested the next
day. Then the plasmids were extracted from the cells,
and the genes encoding Cel7B and Cel7B:E197A were
analyzed on a DNA gel following restriction digestion.
The Cel7B:E197A glycosynthase gene was engineered
to obtain a unique NcoI site so that it could be readily
distinguished from the Cel7B glycosidase gene. Since
total plasmid DNA is analyzed in the mock selection,
effectively a very large number of samples are analyzed
in this assay. Further, because small differences in
activity are amplified during the selection process, this
assay is very sensitive.
and compound 5 conferred a 5-fold enrichment of glyco-
synthase. In addition, all three donor substrates had
background activity, conferring glycosynthase enrich-
ment even when used without Dex-Cel 1. Mtx-Cel-F 2
gave highest background with 5-fold of enrichment;
Mtx-Lac-F 4 gave a moderate background with 3-fold
of enrichment; and Mtx-Lac-F 5 gave the lowest back-
ground with 1.5-fold of enrichment. As we speculated
in our original publication, our hypothesis is that this
background results from covalent modification of
Mtx-Lac-F with a molecule endogenous to the cell that
either acts as a transcriptional activator on its own or
recruits other molecules with this function. By compar-
ison, Dex-Cel 1 alone did not show glycosynthase
enrichment. One reasonable explanation for the higher
activity of Mtx-Cel-F 2 compared to the other two
donor substrates is that it is a closer structural mimic
of the natural cellulose substrate, with an equatorial
bromine at the 40 position as opposed to the axial
hydroxyl group at the 40 position of lactose, and hence
a more efficient substrate for the Cel7B:E197A glyco-
synthase. The mock selection assay data establish unam-
biguously that compounds 1, 4, and 5 are suitable
substrates for detecting glycosynthase activity in the
chemical complementation selection system. If the back-
ground fold-enrichment with donor substrate alone were
subtracted from the fold-enrichment with donor and
acceptor substrates, donors 4 and 5 would be similar
in activity. However, these experiments cannot be used
to estimate the amount of fold-enrichment in the pres-
ence of 1 + 4 or 1 + 5 that is due to 4 or 5, respectively,
alone. Thus, we can only conclude that 4 is as or more
active than 5. The relative ease of synthesis and biolog-
ical activity of these substrates set the stage for now
using chemical complementation for the directed evolu-
tion of glycosynthase catalysts with a range of substrate
specificities.
As shown in Figure 2, the plasmids encoding the glyco-
synthase gene were enriched when the acceptor substrate
Dex-Cel 1 was incubated with any of the three donor
substrates 2, 4, or 5. Specifically, when combined with
the acceptor substrate Dex-Cel 1, compound 2 conferred
the highest (17-fold) enrichment of glycosynthase; com-
pound 4 conferred a 7-fold enrichment of glycosynthase;
2. Conclusion
In conclusion, we report the full design, synthesis, and
activity of optimized chemical dimerizer substrates for
the directed evolution of glycosynthase enzymes via
Figure 2. Dex–Cel glycosyl acceptor 1 and Mtx-Lac-F glycosyl donors 2, 4, and 5 are suitable substrates for detecting glycosynthase activity in the
chemical complementation assay. A mixture of plasmids encoding a 1:10 ratio of the Cel7B:E197A glycosynthase to the Cel7B glycosidase (inactive
control) was transformed en masse into the yeast three-hybrid strain V1019Y. The resulting transformants were incubated in selective media lacking
the appropriate auxotrophs and leucine, and containing the donor and/or acceptor substrates as indicated and grown at 30 ꢁC. After 7 days of
growth, the cells were transferred to non-selective media and harvested the next day. Then the plasmids were extracted from the cells, and the genes
encoding Cel7B and Cel7B:E197A were analyzed on a DNA gel following restriction digestion. The Cel7B:E197A glycosynthase gene was engineered
to obtain a unique NcoI site so that it could be readily distinguished from the Cel7B glycosidase gene. The picture was taken using a Canon SD400
digital camera, and the intensity of DNA bands was quantified using ImageQuantTM (molecule dynamics) without calibration.

6948
H. Tao et al. / Bioorg. Med. Chem. 14 (2006) 6940–6953
chemical complementation. The new synthesis route of
the glycosyl donor substrates Mtx-Lac-F 4 and 5 solved
a side elimination reaction that occurred in the final step
of the synthesis of glycosyl donor substrate Mtx-Cel-F
2, thus allowing us to prepare both substrates on a large
scale for directed evolution experiments. The establish-
ment of an efficient synthetic route and the full charac-
terization of the Dex-Cel 1 and Mtx-Lac-F 4 and 5
chemical dimerizer substrates provide not only efficient
substrates for the directed evolution of improved glyco-
synthase variants with cellobiose substrate specificity,
but also set the stage for the directed evolution of glyco-
synthase enzymes with a range of substrate specificities.
used for the mock selection assay contains ca. 20%
by-product 3 (estimate based on H NMR).
1
3.2. Synthesis of 6
To D-cellobiose octaacetate (2.00 g, 5.14 mmol) in a
50 mL round-bottomed flask was added 30% HBr in
HOAc (10 mL). The reaction was protected from light
and stirred at rt for 1 h. The reaction mixture was dilut-
ed with CH2Cl2 (60 mL) and poured into ice/water
(60 mL) in a separation funnel. The organic phase was
collected and washed with water (1· 50 mL) and satd
aq NaHCO3 (2· 50 mL). After drying over anhydrous
Na2SO4, the organic phase was concentrated to give a
white solid. Rf = 0.4 in 2:1 EtOAc:hexanes. The product
was lyophilized and was used directly in the next reac-
tion without further purification. The product from bro-
mination reaction (630 mg, 0.90 mmol), mercury
cyanide (230 mg, 0.90 mmol), and mercury bromide
(32 mg, 0.090 mmol) were dissolved in CH2Cl2 (3 mL)
and 6-bromo-1-hexanol (130 mg, 1.10 mmol) was added
to the reaction mixture. The reaction was protected
from light and stirred at rt overnight. The reaction mix-
ture was filtered through celite. The filter cake was
washed with CH2Cl2 (3· 5 mL), and the filtrate was con-
centrated and applied to a silica gel column. The prod-
uct (310 mg) was eluted using 2:1 hexanes/EtOAc in
26% yield for two steps: Rf = 0.4 in 1:1 hexanes/EtOAc;
1H NMR (400 MHz, CDCl3) d 5.15 (m, 2H), 5.07 (t,
J = 9.7 Hz, 1H), 4.90 (m, 2H), 4.50 (m, 2H), 4.44 (d,
J = 8.0 Hz, 1H), 4.37 (dd, J = 4.4, 12.5 Hz, 1H), 4.08
(dd, J = 5.0, 12.0 Hz, 1H), 4.03 (dd, J = 2.2, 12.5 Hz,
1H), 3.83 (m, 1H), 3.76 (t, J = 9.6 Hz, 1H), 3.65 (m,
1H), 3.57 (m, 1H), 3.35–3.49 (m, 3H), 1.98–2.12 (m,
21H), 1.84 (m, 2H), 1.57 (m, 2H), 1.25–1.48 (m, 4H);
13C NMR (100 MHz, CDCl3) d 170.7, 170.5, 170.4,
170.0, 169.7, 169.4, 169.2, 101.2, 101.0, 77.0, 73.4,
73.1, 73.0, 72.4, 72.0, 70.4, 68.2, 62.4, 62.0, 34.4, 33.2,
29.8, 28.4, 25.6, 21.6, 21.4; MS (FAB+) m/z 799.2
(M+H)+.
3. Experimental
3.1. General methods
Unless otherwise noted reagents were purchased from
Aldrich and used without further purification.
Glutamic acid a-methyl ester c-tert-butyl ester hydro-
chloric acid was purchased from Advanced Chemtech.
The dexamethasone derivative 7 and the 1:1 2-propa-
nol complex with 6-bromomethyl 2,4-pteridinediamine
monohydrobromide were synthesized as described.12,13
Mtx-Cel-F 2, Mtx-Lac-F 4 and 5 were purified on a
Waters delta 600 HPLC equipped with a Vydac C18
reverse-phase column using acetonitrile gradients.
Nuclear magnetic resonance (NMR) spectra were
recorded on a Bruker 300MHz, 400MHz or 500MHz
Fourier transform NMR spectrometer. High resolu-
tion mass spectra (HRMS) were recorded on a
JMS-HX110A mass spectrometer. Low resolution
MS were recorded on a JMS-LC mate mass spectrom-
eter. Standard protocols for molecular biology and
yeast genetics were used. Taq polymerase was pur-
chased from Promega (Madison, WI). The dNTPs
used in the polymerase chain reaction (PCR) were
purchased from Amersham Biosciences. Oligonucleo-
tides were purchased from Invitrogen. Restriction
enzyme NcoI was purchased from New England Bio-
labs (Beverly, MA). The transformation of yeast was
carried out by electroporation using a Bio-Rad gene-
pulserꢂ Xcell. Yeast plasmids were prepared using
EZNA kit (Omega Bio-Tek). Ultraviolet–visible
measurements were taken using a Molecular Devices
Spectramax 384. The picture of DNA gel was taken
3.3. Synthesis of 8
Compound 6 (218 mg, 0.27 mmol), Dex-SH (132 mg,
0.30 mmol) and DIEA (232 lL, 1.36 mmol) were dis-
solved in degassed MeOH (1.0 mL). The reaction mix-
ture was stirred at rt overnight. The reaction mixture
was concentrated, and the product was purified by silica
gel column chromatography using 40:1 CH2Cl2/MeOH.
The product (218 mg) was obtained as a white solid in
using
a Canon SD400 digital camera and was
analyzed using ImageQuantTM software (Molecule
Dynamics). The yeast strains and the plasmids used
in this study are listed in Table 1. The compound 2
1
69% yield: Rf = 0.5 in 10:1 CH2Cl2/MeOH; H NMR
(400 MHz, CDCl3) d 7.22 (d, J = 10.1 Hz, 1H), 6.89 (t,
Table 1. Strains and plasmids used in this study
Name
Description
Source/reference
Strains
V1019Y
Genotype
MATa trp1 ura3 6lexAop-LEU2 ade4::Pgal1
-
1,19
LexA-eDHFR(HIS3) GAL+ pBC398 pMW112
Plasmids
pHL1262
pHL1263
Details
P
P
met25-Cel7B 2 l URA3 pRS ori ampR
1
1
met25-Cel7B:E197A 2 l URA3 pRS ori ampR

H. Tao et al. / Bioorg. Med. Chem. 14 (2006) 6940–6953
6949
J = 5.8 Hz, 1H), 6.33 (dd, J = 1.8, 10.1 Hz, 1H), 6.11 (s,
1H), 5.16 (q, 2H), 5.06 (t, J = 9.6 Hz, 1H), 4.90 (m, 2H),
4.50 (m, 2H), 4.44 (d, J = 7.9, 1H), 4.37 (m, 2H), 4.02–
4.15 (m, 2H), 3.82 (m, 1H), 3.77 (t, J = 9.5 Hz, 1H),
3.67 (m, 1H), 3.35–3.60 (m, 4H), 3.18 (m, 1H), 2.66 (t,
J = 6.4 Hz, 2H), 2.60 (dd, J = 4.9, 12.7 Hz, 1H), 2.52
(t, J = 7.3 Hz, 2H), 2.29–2.45 (m, 4H), 1.98–2.24 (m,
22H), 1.75–1.91 (m, 2H), 1.47–1.63 (m, 9H), 1.21–1.37
(m, 4H), 1.12 (s, 3H), 0.94 (d, J = 7.2 Hz, 3H); 13C
NMR (100 MHz, CDCl3) d 186.8, 173.0, 170.7, 170.6,
170.4, 170.1, 169.9, 169.5, 169.3, 167.1, 153.0, 129.9,
125.2, 102.0, 101.1, 101.0, 100.3, 87.4, 77.0, 73.4, 73.1,
73.0, 72.8, 72.4, 72.1, 70.5, 68.2, 62.4, 62.0, 49.1, 48.9,
48.6, 44.4, 39.0, 36.8, 35.8, 35.1, 34.9, 32.9, 32.5, 32.1,
31.7, 30.1, 29.8, 29.0, 28.0, 26.0, 23.6, 23.6, 21.6, 21.4,
21.3, 21.2, 17.9, 15.2; HRMS (FAB+) m/z 1156.4835
((M+H)+, C55H79FNO22S+ requires 1156.4799).
The organic layer was washed with 1:1 brine:1 N HCl
(2· 100 mL), brine (3· 100 mL), dried over MgSO4,
and concentrated. To the residue, CH2Cl2 (8 mL) and
TFA (8 mL) were added. After stirring at rt for 1 h,
CH2Cl2and TFA were removed in vacuo. The residue
was dissolved in CH2Cl2 and applied to a silica gel col-
umn. 7.24 g of compound 9 was eluted with 25:1
CH2Cl2/MeOH in 39% yield as white solid: Rf = 0.2 in
25:1 CH2Cl2/MeOH; 1H NMR (400 MHz, CDCl3) d
6.23 (d, J = 3.7 Hz, 0.64H), 5.68 (d, J = 8.2 Hz,
0.37H), 5.40 (t, J = 9.8 Hz, 0.63H), 5.21 (m, 1.36H),
5.02 (m, 1H), 4.85 (m, 1H), 4.45 (m, 2H), 4.10 (m,
2H), 4.02 (m, 0.70H), 3.85 (m, 3.3H), 3.60 (m, 1H),
3.32 (br, 2H), 1.90–2.20 (m, 18H); 13C NMR
(100 MHz, CDCl3) d 170.4, 170.1, 170.0, 169.6, 169.3,
168.6, 168.5, 101.0, 100.9, 91.3, 88.7, 75.8, 74.4, 73.6,
73.5, 73.3, 72.8, 70.6, 70.3, 69.9, 69.6, 69.5, 69.1, 67.1,
61.9, 61.6, 61.4, 21.0, 20.9, 20.9, 20.8, 20.7, 20.6; recrys-
tallization of compound 9 in 1:2 ethyl acetate and ether
gave pure a-anomer as white needle: 1H NMR
(400 MHz, CDCl3) d 6.25 (d, J = 3.6 Hz, 1H), 5.45 (t,
J = 9.6 Hz, 1H), 5.22 (dd, J = 8.0, 10.0 Hz, 1H), 5.04
(dd, J = 3.7, 10.2 Hz, 1H), 4.90 (dd, J = 3.1, 10.2 Hz,
1H), 4.51 (d, J = 7.9 Hz, 1H), 4.47 (d, J = 12.0 Hz,
1H), 4.09 (m, 3H), 3.89 (m, 3H), 3.59 (t, J = 5.03 Hz,
1H), 2.93 (d, J = 4.3 Hz, 1H), 2.53 (dd, J = 4.4, 8.0 Hz,
1H), 2.02–2.18 (m, 18H), The 2.93 and 3.53 peak were
removed by treatment of CD3OD; 13C NMR
(100 MHz, CDCl3) d 170.1, 169.8, 169.6, 169.2, 168.9,
101.0, 88.9, 75.7, 74.4, 73.4, 70.6, 70.1, 69.6, 69.3, 67.9,
62.2, 61.7, 21.1, 21.0, 21.0, 20.9, 20.7. MS (FAB+) m/z
595.1862 ((M+H)+, C24H35O17 requires 595.1874).
3.4. Synthesis of 1
Compound 7 (20 mg, 0.017 mmol) was dissolved in
MeOH (0.9 mL) and NaOMe in MeOH (4.4 M,
5.0 lL, 0.022 mmol) was added. The reaction mixture
was stirred at rt for 40 min. Then Dowex-200 acidic res-
in (200 mg) was added, and the stirring was continued
for 5 more minutes. The resin was removed by filtration
and washed with MeOH. The filtrate was concentrated
to give a thin film on the side of the flask. The film
was washed with plenty of CH2Cl2 and EtOAc to
remove any soluble impurities and then dried in vacuo.
The product 1 (15 mg) was obtained as a white solid in
quantitative yield: 1H NMR (400 MHz, CD3OD) d 7.45
(d, J = 10.1 Hz, 1H), 6.32 (dd, J = 1.9, 10.1 Hz, 1H),
6.10 (s, 1H), 4.43 (d, J = 7.8 Hz, 1H), 4.30 (d,
J = 7.8 Hz, 1H), 4.26 (d, J = 11.2 Hz, 1H), 3.90 (m,
4H), 3.68 (dd, J = 5.4, 11.8 Hz, 1H), 3.63–3.20 (m,
14H), 3.14 (m, 1H), 2.74 (m, 1H), 2.66 (m, 2H), 2.60
(t, J = 7.3 Hz, 2H), 2.36–2.57 (m, 2H), 2.22 (m, 2H),
1.90 (m, 1H), 1.77 (q, J = 11.5 Hz, 1H), 1.40–1.70 (m,
13H), 1.23 (m, 1H), 1.10 (s, 3H), 0.91 (d, J = 7.4 Hz,
3H); 13C NMR (100 MHz, CD3OD) d 188.6, 175.2,
170.9, 155.8, 129.5, 124.8, 104.4, 104.0, 88.0, 80.6,
80.0, 77.7, 76.3, 76.3, 74.8, 74.3, 73.2, 72.8, 71.3, 70.7,
62.4, 61.8, 50.5, 50.2, 44.8, 40.0, 36.8, 36.3, 35.9, 35.7,
33.5, 32.5, 32.3, 30.8, 30.7, 29.7, 28.9, 26.8, 23.7, 17.9,
15.3; HRMS (FAB+) m/z 862.4033 ((M+H)+,
C41H64FNO15S+ requires 862.4059).
3.6. Synthesis of 15, 16
Compound 9 (2.22 g, 3.73 mmol), p-tosyl chloride
(439 mg,
(745 mg,
3.91 mmol),
and
DABCO
3.91 mmol) were added to a round-bottomed flask.
At 0 ꢁC, anhydrous CH2Cl2 (40 mL) was added to
the bottle. The reaction mixture was stirred at 0 ꢁC
for 4 h and white precipitate formed, then MeOH
(10 mL) was added to the reaction, and the mixture
was concentrated under vacuum. The residue was puri-
fied using a silica gel column (1:1 hexanes/EtOAc) to
give 1.81 g of 15 in 65% yield as white solid: Rf = 0.4
1
in 1:2 hexanes/EtOAc; H NMR (400 MHz, CDCl3) d
7.81 (d, J = 8.3 Hz, 2H), 7.40 (d, J = 8.3 Hz, 2H),
6.26 (d, J = 3.7 Hz, 1H), 5.41 (t, J = 9.8 Hz, 1H), 5.17
(dd, J = 8.0, 10.2 Hz, 1H), 5.00 (dd, J = 3.7, 10.3 Hz,
1H), 4.88 (dd, J = 3.3, 10.2 Hz, 1H), 4.46 (m, 2H),
4.27 (m, 1H), 4.00–4.09 (m, 4H), 3.82 (m, 2H), 2.87
(d, J = 5.2 Hz, 1H), 2.48 (s, 3H), 2.02–2.15 (m, 18H),
the 2.87 peak was removed by treatment of CD3OD;
13C NMR (100 MHz, CDCl3) d 170.1, 169.8, 169.6,
169.5, 169.0, 168.6, 145.2, 132.2, 129.8, 127.7, 100.7,
88.8, 75.4, 73.1, 72.2, 70.6, 69.4, 69.3, 69.2, 67.1,
66.3, 61.4, 21.8, 21.0, 21.0, 20.9, 21.8, 20.6; MS
(FAB+) m/z 749.1966 ((M+H)+, C31H41O19S requires
749.1963). To Compound 15 (300 mg, 0.40 mmol),
Ac2O (10 mL) and pyridine (10 mL) were added, and
the reaction mixture was stirred at rt overnight. The
mixture was concentrated and dried in vacuo to give
compound 16 in 100% yield as white solid: Rf = 0.4
3.5. Synthesis of 9
Lactose (10.8 g, 31.4 mmol), 4-methoxybenzaldehyde
dimethyl acetal (10.7 mL, 11.5 g, 62.9 mmol), and tolu-
ene-p-sulfonic acid monohydrate (597 mg, 3.14 mmol)
were dissolved in anhydrous DMF (60 mL). The reac-
tion vessel was placed in a 50 ꢁC water bath and con-
nected to an aspirator via a drying tower. The reaction
mixture was stirred at 50 ꢁC until all the solids were dis-
solved. Then the water bath was removed, 300 mL
CH2Cl2 was added to the mixture, and white precipitate
was formed and filtered. To the precipitate, acetic
anhydride (120 mL) and pyridine (100 mL) were added.
After stirred at rt overnight, solvents were removed in
vacuo. To the residue was added CH2Cl2 (150 mL).

6950
H. Tao et al. / Bioorg. Med. Chem. 14 (2006) 6940–6953
1
in 1:1 hexanes:EtOAc; H NMR (400 MHz, CDCl3) d
7.77 (d, J = 8.3 Hz, 2H), 7.38 (d, J = 8.0 Hz, 2H),
6.26 (d, J = 3.7 Hz, 1H), 5.45 (t, J = 9.7 Hz, 1H), 5.34
(d, J = 2.7 Hz, 1H), 5.09 (dd, J = 7.9, 10.4 Hz, 1H),
5.00 (dd, J = 3.7, 10.3 Hz, 1H), 4.95 (d, J = 3.4 Hz,
1H), 4.92 (dd, J = 3.2, 10.4 Hz, 1H), 4.48 (d,
J = 7.9 Hz, 1H), 4.43 (dd, J = 2.0, 12.0 Hz, 1H), 4.12
(d, J = 4.2 Hz, 1H), 4.09 (d, J = 3.8 Hz, 1H), 4.00 (m,
2H), 3.93 (d, J = 6.3 Hz, 1H), 3.83 (t, J = 9.7 Hz,
1H), 2.47 (s, 3H), 1.95–2.18 (m, 21H); MS (EI+) m/z
789.1962 ((MꢀH)+, C33H41O20S+ requires 789.1912).
hexanes:EtOAc; 1H NMR (400 MHz, CDCl3) d 7.69
(d, J = 8.8 Hz, 2H), 6.87 (d, J = 7.4 Hz, 1H), 6.57 (d,
J = 8.8 Hz, 2H), 4.76 (m, 1H), 3.76 (s, 3H), 2.87 (s,
3H), 2.38 (m, 2H), 2.22 (m, 1H), 2.07 (m, 1H), 1.42 (s,
9H); 13C NMR (100 MHz, CDCl3) d 172.6, 172.3,
166.7, 151.8, 128.7, 121.3, 111.1, 80.8, 52.5, 52.3, 31.8,
30.3, 28.2, 27.3.
3.9. Synthesis of 20
Compound 19 (1.04 g, 2.96 mmol) and 6-bromomethyl
2,4-pteridinediamine monohydrobromide (1:1 com-
pound with 2-propanol) (0.94 g, 2.37 mmol) were dis-
solved in dimethyl acetamide (DMAc) (8 mL). The
reaction mixture was stirred at 55 ꢁC for 6 h. The filtrate
was concentrated and then purified using a silica gel col-
umn 20:1 CH2Cl2/MeOH to give 0.93 g of 20 in 75%
yield as orange solid: Rf = 0.2 in 10:1 CH2Cl2/MeOH;
1H NMR (300 MHz, CD3OD) d 8.97 (s, 1H), 7.75 (d,
J = 8.7 Hz, 2H), 6.91 (d, J = 8.8 Hz, 2H), 4.96 (s, 2H),
4.59 (dd, J = 5.2, 9.3 Hz, 1H), 3.73 (s, 3H), 3.29 (s,
3H), 2.38 (t, J = 7.2 Hz, 2H), 2.19 (m, 1H), 2.05 (m,
1H), 1.42 (s, 9H); 13C NMR (75 MHz, CD3OD) d
173.1, 172.9, 169.0, 163.8, 156.6, 152.1, 151.4, 149.6,
149.5, 145.8, 129.4, 129.3, 122.4, 112.4, 80.9, 55.9,
52.7, 39.3, 31.8, 28.1, 26.5.
3.7. Synthesis of 17, 18
At ꢀ78 ꢁC, a 1.0 M solution of potassium tert-butoxide
solution in THF (2.87 mL, 2.87 mmol) was diluted with
THF (10 mL) and degassed using argon. A solution of
2-(Boc-amino) ethanethiol (580 lL, 609 mg, 3.44 mmol)
in THF (3 mL) was added successively, degassed, and
stirred for 30 min. A solution of compound 16
(794 mg, 1.00 mmol) in THF (10 mL) was added and
degassed, and the mixture was gradually warmed to
rt and stirred for 2 h. A mixture of acetic anhydride
(12 mL) and pyridine (10 mL) was added to the reac-
tion and the reaction mixture was stirred at rt over-
night. The reaction mixture was quenched with water
(20 mL) and extracted with ethyl acetate (2· 40 mL).
The organic layer was washed with brine (2· 40 mL),
dried over Na2SO4, and then purified using a silica
gel column (3:1 CH2Cl2/EtOAc) to give 369 mg prod-
uct as white solid: Rf = 0.4 (3:1 CH2Cl2/EtOAc); MS
(FAB+) m/z 796.5 (M+H)+. The product is a mixture
of compound 17 and an unknown impurity. To the
bottle of the product in CH2Cl2 (15 mL) was added
TFA (5.0 mL). After stirring at rt for 1 h, CH2Cl2
and TFA were removed under vacuum. The residue
was dissolved in CH2Cl2 and purified using a silica
gel column (from 40:1 CH2Cl2/MeOH to 10:1
CH2Cl2/MeOH with 0.5% TEA) to give 244 mg com-
pound 18 in 35% yield for two steps as white solid:
Rf = 0.1 in CH2Cl2/MeOH; 1H NMR (500 MHz,
CDCl3) d 7.91 (br, 3H), 6.23 (d, J = 3.5 Hz, 0.36H),
5.69 (d, J = 8.1 Hz, 0.64H), 5.44 (m, 1.15H), 5.25 (t,
J = 9.0 Hz, 0.54H), 5.03 (m, 2.72H), 4.50 (d,
J = 7.7 Hz, 1H), 4.43 (d, J = 11.1 Hz, 1H), 3.80–4.13
(m, 4H), 3.19 (m, 4H), 2.90 (m, 2H), 2.73 (m, 1H),
2.58 (m, 1H), 1.96–2.18 (m, 21H); HRMS (FAB+)
m/z 696.2162 ((M+H)+, C28H42NO17S+ requires
696.2173).
3.10. Synthesis of 21
To compound 20 (500 mg, 0.955 mmol) in CH2Cl2
(2.0 mL) was added TFA (2.0 mL). After stirring at
rt for 1 h, CH2Cl2 and TFA were removed in vacuo.
The residue was dissolved in CH2Cl2 and purified
using a silica gel column (from 20:1 CH2Cl2/MeOH
to 5:1 CH2Cl2/MeOH) to give 339 mg 21 in 76% yield
as orange solid: Rf = 0.2 in 5:1 CH2Cl2/MeOH; 1H
NMR (400 MHz, CD3OD) d 8.58 (s, 1H), 7.75 (d,
J = 8.9 Hz, 2H), 6.83 (d, J = 8.9 Hz, 2H), 4.86 (s,
2H), 4.59 (dd, J = 5.1, 9.2 Hz, 1H), 3.74 (s, 3H),
3.24 (s, 3H), 2.45 (t, J = 7.2 Hz, 2H), 2.24 (m, 1H),
2.10 (m, 1H); 13C NMR (100 MHz, CD3OD) d
174.5, 171.8, 167.7, 162.3, 156.4, 150.6, 150.5, 147.8,
145.9, 128.0, 121.1, 120.1, 110.4, 54.6, 51.9, 50.9,
37.9, 29.7, 25.5.
3.11. Synthesis of 22
To a solution of Compound 21 (25 mg, 0.053 mmol) and
PyBOP (28 mg, 0.054 mmol) in DMF (1 mL) was added
a solution of tert-butyl 6-aminohexanoate (27 mg,
0.144 mmol) in anhydrous CH2Cl2 (2 mL) and DIEA
(40 lL, 30 mg, 0.23 mmol). The mixture was stirred at
rt overnight and then concentrated under vacuum. The
residue was purified using a silica gel column (20:1
CH2Cl2/MeOH to 10:1 CH2Cl2/MeOH) to give 22 mg
of 23 in 65% yield as orange solid: Rf = 0.4 in 10:1
3.8. Synthesis of 19
Glutamic acid a-methyl ester c-tert-butyl ester hydro-
chloric acid (1.00 g, 3.94 mmol), 4-methylamino benzoic
acid (0.835 g, 5.50 mmol), and DCC (1.140 g,
5.50 mmol) were dissolved in anhyd CH2Cl2 (10 mL).
DIEA (2.40 mL, 13.80 mmol) was added slowly. The
reaction mixture was stirred at rt overnight, and then
diluted with CH2Cl2 (20 mL), filtered through Celite,
and washed with CH2Cl2. The filtrate was concentrated
and then purified using a silica gel column (from 4:1 hex-
anes/ethyl acetate to 2:1 hexanes:ethyl acetate) to give
0.73 g of 19 in 53% yield as white solid: Rf = 0.2 in 1:1
1
CH2Cl2/MeOH; H NMR (400 MHz, CD3OD) d 8.55
(s, 1H), 7.74 (d, J = 9.0 Hz, 2H), 6.84 (d, J = 9.0 Hz,
2H), 4.82 (s, 2H), 4.54 (q, 1H), 3.72 (s, 3H), 3.23 (s,
3H), 3.11 (t, J = 7.0 Hz, 2H), 2.32 (m, 2H), 2.16 (m,
4H), 1.51 (m, 2H), 1.41 (m, 11H), 1.27 (m, 2H); HRMS
+
(FAB+) m/z 638.3438 ((M+H)+, C31H44N9O6 requires
638.3415).

H. Tao et al. / Bioorg. Med. Chem. 14 (2006) 6940–6953
6951
3.12. Synthesis of 23
3.15. Synthesis of 4
To compound 22 (13 mg, 0.020 mmol) in CH2Cl2
(2.0 mL) was added TFA (1.0 mL). After stirring at
rt for 3 h, CH2Cl2 and TFA were removed in vacuo.
The residue was dissolved in CH2Cl2 and purified
using a silica gel column (from 5:1 CH2Cl2/MeOH
to 2:1 CH2Cl2/MeOH) to give 11 mg 23 in 93% yield
as orange solid: Rf = 0.2 in 5:1 CH2Cl2/MeOH; 1H
Compound 25 (3.0 mg, 0.0025 mmol) was dissolved in
MeOH (0.5 mL) and cooled with ice. 60 lL of
1.67 mM lithium hydroxide hydrate aq (0.10 mmol)
was slowly added to the MeOH solution. The reaction
mixture was then allowed to warm to rt. After stirring
for 6 h, MeOH (10 mL), water (2 mL), and 40 lL of
10% AcOH aq solution were added to the reaction mix-
ture, and then the mixture was concentrated to dryness.
The mixture was purified by HPLC using a C18 reverse
phase column with a gradient from 6% CH3CN aq to
15% CH3CN aq in 50 min. 1.3 mg of compound 4 was
obtained in 55% yield as orange solid: 1H NMR
(500 MHz, CD3OD) d 8.61 (s, 1H), 7.77 (d, J = 9.0 Hz,
2H), 6.90 (d, J = 9.0 Hz, 2H), 5.55 (dd, J1 = 48.6 Hz,
1H), 4.58 (s, 3H), 4.49 (m, 1H), 4.41 (d, 1H), 3.84 (m,
5H), 3.68 (t, 2H), 3.47, (m, 5H), 3.47 (s, 3H), 3.27 (s,
3H), 3.19 (s, 2H), 3.10 (t, 2H), 2.83 (d, J = 7.0 Hz,
1H), 2.69 (m, 2H), 2.09–2.27 (m, 6H), 1.31–1.59 (m,
6H); MS (ESI+) m/z 953.5 (M+H)+.
NMR (400 MHz, 80% CD3OD + 20% CDCl3)
d
8.55 (s, 1H), 7.74 (d, J = 9.0 Hz, 2H), 6.83 (d,
J = 9.0 Hz, 2H), 4.82 (s, 2H), 4.54 (q, J = 4.6 Hz,
1H), 3.72 (s, 3H), 3.23 (s, 3H), 3.11 (t, J = 7.0 Hz,
2H), 2.32 (m, 2H), 2.21 (m, 3H), 2,09 (m, 1H),
1.56 (m, 2H), 1.42 (m, 2H), 1.29 (m, 2H); MS
+
(FAB+) m/z 582.3 ((M+H)+, C27H36N9O6 requires
582.3).
3.13. Synthesis of 24
To a solution of compound 23 (35 mg, 0.060 mmol) and
PyBOP (35 mg, 0.067 mmol) in DMF (1 mL) was added
a solution of compound 18 (36 mg, 0.052 mmol) in
anhydrous CH2Cl2 (2 mL) and DIEA (40 lL, 30 mg,
0.23 mmol). The mixture was stirred at rt overnight
and then concentrated under vacuum. The residue was
purified using a silica gel column (20:1 CH2Cl2/MeOH
to 10:1 CH2Cl2/MeOH) to give 39 mg of 24 in 60% yield
as orange solid: Rf = 0.1 in 15:1 CH2Cl2/MeOH; 1H
NMR (500 MHz, CD3OD) d 8.55 (d, 1H), 7.74 (d,
J = 8.8 Hz, 2H), 6.85 (d, J = 8.6 Hz, 2H), 6.22 (d,
J = 3.7 Hz, 0.53H), 5.74 (d, J = 8.3 Hz, 0.45H), 5.47 (t,
J = 3.5 Hz, 1H), 5.38 (t, 0.53H), 5.28 (t, 0.44H), 5.09
(m, 1H), 4.97 (m, 2H), 4.84 (s, 2H), 4.60 (dd, J = 2.8,
7.5 Hz, 1H), 4.54 (m, 1H), 4.46 (t, 1H), 4.14 (m, 1H),
4.07 (m, 0.52H), 3.92 (m, 2.45H), 3.72 (s, 3H), 3.24 (d,
3H), 3.11 (m, 2H), 2.77 (m, 1H), 2.65 (m, 3H), 1.91–
2.34 (m, 28H), 1.56 (t, J = 7.1 Hz, 2H), 1.44 (t,
J = 7.2 Hz, 11H), 1.27 (m, 2H); MS (FAB+) m/z
1259.3 (M+H)+.
3.16. Synthesis of 26
To a solution of compound 21 (25 mg, 0.053 mmol) and
PyBOP (28 mg, 0.054 mmol) in DMF (1 mL) was added
a solution of compound 18 (27 mg, 0.039 mmol) in
anhydrous CH2Cl2 (2 mL) and DIEA (40 lL, 30 mg,
0.23 mmol). The mixture was stirred at rt overnight
and then concentrated under vacuum. The residue was
purified using a silica gel column (20:1 CH2Cl2/MeOH
to 10:1 CH2Cl2/MeOH) to give 22 mg of 26 in 49% yield
as orange solid: Rf = 0.8 in 5.7:1 CH2Cl2/MeOH; 1H
NMR (500 MHz, CD3OD) d 8.57 (s, 1H), 7.74 (d,
J = 8.8 Hz, 2H), 6.86 (t, 2H), 6.22 (d, J = 3.7 Hz,
0.46H), 5.74 (d, J = 8.3 Hz, 0.52H), 5.45 (d,
J = 3.3 Hz, 1H), 5.38 (t, J = 9.8 Hz, 0.45H), 5.28 (t,
J = 9.0 Hz, 0.56H), 5.09 (m, 1H), 4.97 (m, 2H), 4.84 (s,
2H), 4.66 (m, 1H), 4.57 (m, 1H), 4.45 (t, J = 8.4 Hz,
1H), 4.14 (m, 1H), 4.05 (m, 0.53H), 3.92 (m, 2.54H),
3.73 (s, 3H), 3.25 (s, 3H), 2.74 (m, 1H), 2.28 (m, 3H),
1.89–2.35 (m, 26H); MS (FAB+) m/z 1145.7 (M+H)+.
3.14. Synthesis of 25
Compound 24 (5 mg, 0.0040 mmol) was placed in a
1.5 mL microcentrifuge tube. 70% HF-pyridine
(100 lL) was added and the reaction mixture was
shaken on an orbital shaker at rt for 2.0 h. The reac-
tion mixture was then diluted with CH2Cl2 (30 mL)
and poured into iced brine (30 mL). The organic layer
was separated from the aqueous layer, washed with
iced brine (1· 30 mL), cold satd NaHCO3 (30 mL),
and then dried and concentrated. The residue was
purified by silica gel column with 20:1 to 15:1
CH2Cl2/MeOH to give 3.0 mg compound 25 in 62%
yield as orange solid: Rf = 0.4 in 10:1 CH2Cl2/MeOH;
1H NMR (500 MHz, CD3OD) d 8.57 (s, 1H), 7.75 (d,
J = 9.0 Hz, 2H), 6.86 (d, J = 9.0 Hz, 2H), 5.72 (dd,
1H), 5.47 (d, 1H), 5.36 (t, 1H), 5.09 (m, 1H), 4.85
(m, 57H), 4.66 (d, J = 7.9 Hz, 1H), 4.55 (m, 2H),
4.14 (m, 4H), 3.72 (s, 3H), 3.25 (s, 3H), 3.12 (m,
2H), 2.77 (m, 1H), 2.65 (m, 3H), 1.91–2.34 (m,
25H), 1.28–1.58 (m, 6H); MS (FAB+) m/z 1219.5
(M+H)+.
3.17. Synthesis of 27
Compound 26 (10 mg, 0.0087 mmol) was placed
in a 1.5 mL microcentrifuge tube. 70% HF-pyridine
(100 lL) was added and the reaction mixture was shaken
on an orbital shaker at rt for 2.0 h. The reaction mixture
was then diluted with CH2Cl2 (30 mL) and poured into
iced brine (30 mL). The organic layer was separated from
the aqueous layer, washed with iced brine (1· 30 mL),
cold satd NaHCO3 (30 mL), and then dried and concen-
trated. The residue was purified by silica gel column with
20:1 to 15:1 CH2Cl2:MeOH to give 5 mg compound 27 in
52% yield as orange solid: Rf = 0.1 in 15:1 CH2Cl2/
1
MeOH; H NMR (500 MHz, CD3OD) d 8.56 (s, 1H),
7.75 (d, J = 8.8 Hz, 2H), 6.86 (d, J = 8.8 Hz, 2H), 5.69
(dd, J = 2.7, 53.3 Hz, 1H), 5.45 (d, J = 3.3 Hz, 1H),
5.35 (t, J = 9.75 Hz, 1H), 5.09 (dd, J = 3.5, 10.4 Hz,
1H), 4.96 (m, 2H), 4.67 (d, J = 7.8 Hz, 1H), 4.58 (m,
1H), 4.50 (d, J = 11.2 Hz, 1H), 4.17 (dd, J = 5.3,
12.2 Hz), 4.07 (m, 1H), 3.95 (m, 2H), 3.73 (s, 3H), 3.25

6952
H. Tao et al. / Bioorg. Med. Chem. 14 (2006) 6940–6953
(s, 3H), 2.74 (m, 1H), 2.60 (m, 3H), 1.90–2.36 (m, 24H);
experiment was digested with NcoI for 3 h and loaded
onto an ethidium bromide stained 1% agarose gel. Since
only Cel7B:E197A gene can be digested, colonies encod-
ing Cel7B glycosidase can be easily distinguished from
those encoding Cel7B E197A.
MS (FAB+) m/z 1105.4 (M+H)+.
3.18. Synthesis of 5
Compound 27 (4.0 mg, 0.0036 mmol) was dissolved in
MeOH (0.5 mL) and cooled with ice. Lithium hydroxide
hydrate (1.5 mg, 0.036 mmol) was dissolved in water
(0.05 mL), and the solution was slowly added to the
MeOH solution. The reaction mixture was then allowed
to warm to rt. After stirring for 6 h, MeOH (10 mL),
water (2 mL), and 20 lL of 10% AcOH aq solution were
added to the reaction mixture, and then the mixture was
concentrated to dryness. The mixture was purified by
HPLC using a C18 column with a gradient from 6%
CH3CN aq to 15% CH3CN aq in 50 min. 1.2 mg of com-
Acknowledgments
The authors thank Dr. J. Cherry and Dr. C. Ward from
Novozymes for providing the Cel7B genes and the Cel7-
B:E197A glycosynthase. We are grateful for financial
support for this work from the National Science Foun-
dation (CHE-0350183) and National Institutes of
Health (GM62867).
1
pound 5 was obtained in 40% yield as orange solid: H
NMR (500 MHz, CD3OD) d 8.58 (s, 1H), 7.73 (d,
J = 9.0 Hz, 2H), 6.87 (d, J = 9.0 Hz, 2H), 5.50 (dd,
(J1 = 53.5 Hz, 1H), 4.55 (s, 3H), 4.48 (m, 1H), 4.38 (d,
J = 6.5 Hz, 1H), 3.82 (m, 6H), 3.64 (m, 3H), 3.44, (m,
7H), 3.27 (s, 3H), 3.19 (s, 1H), 2.77 (m, 2H), 2.58 (m,
2H), 2.02–2.40 (m, 4H); HRMS (ESI+) m/z 840.3054
((M+H)+, C34H47FN9O13S+ requires 840.2998).
Supplementary data
The H NMR, 13C-NMR, COSY and HSQC spectra of
1
compounds synthesized in this paper. Supplementary
data associated with this article including calculation
details and predicted activities can be found, in the
online version, at doi:10.1016/j.bmc.2006.06.034.
3.19. Mock selection assay
The plasmids for mock library selection were prepared
from Escherichia coli strains HL1262 (pHL1262 encod-
ing Cel7B glycosidase) and HL1263 (pHL1263 encoding
Cel7B:E197A glycosynthase)1 using QIAprepꢂ maxiprep
kits, and plasmid concentration was determined by UV
absorption at 260 nm. Then the plasmid mix was trans-
formed to yeast strain V1019Y using high efficiency yeast
transformation protocol.21 Fifty microliters of transfor-
mances was incubated in 50 mL SC media containing
2% glucose, lacking uracil, histidine, and tryptophan at
30 ꢁC for 3 days. The cell suspension was then
centrifuged at room temperature using a Sorvail RT7
Plus centrifuge at 2000 rpm for 5 min. The cell pellet
was re-suspended in 5.0 mL of 10% glycerol. To set up
mock selections, 20 lL of the cell suspension was added
into 13 mL Falcon tubes containing 2 mL SC media con-
taining 2% galactose, 2% raffinose, and indicated small
molecules, lacking uracil, histidine, tryptophan, methio-
nine, and leucine. The tubes were incubated in a rotary
shaker at 200 rpm and 30 ꢁC. On the 7th day, 300 lL
of cell culture was transferred into 3 mL non-selective
synthetic complete media containing 2% glucose, lacking
uracil, histidine, and tryptophan, and incubated for
another day. The resulted cells were collected for plasmid
preparation using EZNA kit (Omega Bio-Tek). The plas-
mid mixture from each selection experiment was ampli-
fied using Taq polymerase with primers VWC98450
(50GCA TAC GTC ACT AGT ATG GCT CGC GGT
ACC GCT CT30) and VWC98630, (50GCA TAC GTC
CCC GGG TTA ATG GTG ATG GTG ATG GTG
CTG AAC CTC CTG GTA GGT C30). The PCR pro-
gram used was as follows: step 1, 94 ꢁC, 5 min; step 2,
94 ꢁC, 0.5 min, 50 ꢁC, 0.5 min, 72 ꢁC, 3 min, 30 cycles;
step 3, 72 ꢁC, 10 min. The PCR product was then puri-
fied and quantified using UV absorption at 260 nm.
Equal amount of PCR product from each selection
References and notes
1. Lin, H.; Tao, H.; Cornish, V. W. J. Am. Chem. Soc. 2004,
126, 15051–15059.
2. Mackenzie, L. F.; Wang, Q.; Warren, R. A. J.; Withers, S.
G. J. Am. Chem. Soc. 1998, 120, 5583–5584.
3. Daines, A. M.; Maltman, B. A.; Flitsch, S. L. Curr. Opin.
Chem. Biol. 2004, 8, 106–113.
4. Sears, P.; Wong, C. H. Science 2001, 291, 2344–2350.
5. Kim, Y. W.; Lee, S. S.; Warren, R. A.; Withers, S. G.
J. Biol. Chem. 2004, 279, 42787–42793.
6. Mayer, C.; Jakeman, D. L.; Mah, M.; Karjala, G.; Gal, L.;
Warren, R. A.; Withers, S. G. Chem. Biol. 2001, 8, 437–443.
7. Baker, K.; Bleczinski, C.; Lin, H.; Salazar-Jimenez, G.;
Sengupta, D.; Krane, S.; Cornish, V. W. Proc. Natl. Acad.
Sci. U.S.A. 2002, 99, 16537–16542.
8. Fort, S.; Boyer, V.; Greffe, L.; Davies, G. M. O.;
Christiansen, L.; Schulein, M.; Cottaz, S.; Driguez, H.
J. Am. Chem. Soc. 2000, 122, 5429–5437.
9. Schou, C.; Rasmussen, G.; Kaltoft, M. B.; Henrissat, B.;
Schulein, M. Eur. J. Biochem. 1993, 217, 947–953.
10. Ducros, V. M.; Tarling, C. A.; Zechel, D. L.; Brzozowski,
A. M.; Frandsen, T. P.; von Ossowski, I.; Schulein, M.;
Withers, S. G.; Davies, G. J. Chem. Biol. 2003, 10, 619–628.
11. MacKenzie, L. F.; Sulzenbacher, G.; Divne, C.; Jones, T.
A.; Woldike, H. F.; Schulein, M.; Withers, S. G.; Davies,
G. J. Biochem. J. 1998, 335 Pt 2, 409–416.
12. Lin, H.; Abida, W. M.; Sauser, R. T.; Cornish, V. W.
J. Am. Chem. Soc. 2000, 122, 4247–4248.
13. Abida, W. M.; Carter, B. T.; Althoff, E. A.; Lin, H.;
Cornish, V. W. Chembiochem 2002, 3, 887–895.
14. Toshima, K.; Tatsuta, K. Chem. Rev. 1993, 93, 1503–1531.
15. Capon, B. Chem. Rev. 1969, 69, 407–498.
16. Hart, B. P.; Haile, W. H.; Licato, N. J.; Bolanowska, W. E.;
McGuire, J.J.;Coward,J. K. J. Med.Chem. 1996,39, 56–65.
17. Kokuryo, Y.; Kawata, K.; Nakatani, T.; Kugimiya, A.;
Tamura, Y.; Kawada, K.; Matsumoto, M.; Suzuki, R.;
Kuwabara, K.; Hori, Y.; Ohtani, M. J. Med. Chem. 1997,
40, 3280–3291.

H. Tao et al. / Bioorg. Med. Chem. 14 (2006) 6940–6953
6953
18. Piper, J. R.; Montgomery, J. A. J. Org. Chem. 1977, 42,
208–211.
20. Licitra, E. J.; Liu, J. O. Proc. Natl. Acad. Sci. U.S.A. 1996,
93, 12817–12821.
19. Baker, K.; Sengupta, D.; Salazar-Jimenez, G.; Cornish,
V. W. Anal. Biochem. 2003, 315, 134–137.
21. Colby, D. W.;Kellogg, B. A.;Graff,C.P.;Yeung,Y.A.;Swers,
J. S.; Wittrup, K. D. Methods Enzymol. 2004, 388, 348–358.