Monosaccharide-linked inhibitors of O6-methylguanine-DNA methyltransferase (MGMT): Synthesis, molecular modeling, and structure-activity relationships

Article abstract of DOI:10.1021/jm010006e

A series of potential inhibitors of the human DNA repair protein O⁶-methylguanine-DNA methyltransferase (MGMT) were synthesized, characterized in detail by NMR, and tested for their ability to deplete MGMT activity in vitro. The new compounds, ω-[O⁶-R-guan-9-yl]-(CH₂)_n-β-d- glucosides with R = benzyl or 4-bromothenyl and ω = n = 2, 4, ... 12, were compared with the established inhibitors O⁶-benzylguanine (O⁶-BG), 8-aza-O⁶-benzylguanine (8-aza-BG), and O⁶-(4-bromothenyl)guanine (4-BTG), which exhibit in an in vitro assay IC₅₀ values of 0.62, 0.038, and 0.009 μM, respectively. Potential advantages of the glucosides are improved water solubility and selective uptake in tumor cells. The 4-BTG glucosides with n = 2, 4, 6 show moderate inhibition with an IC₅₀ of ca. 0.5 μM, while glucosides derived from BG and 8-aza-BG showed significantly poorer inhibition compared to the parent compounds. The 4-BTG glucosides with n = 8, 10, 12 were effective inhibitors with IC₅₀ values of ca. 0.03 μM. To understand this behavior, extensive molecular modeling studies were performed using the published crystal structure of MGMT (PDB entry: 1QNT). The inhibitor molecules were docked into the BG binding pocket, and molecular dynamics simulations with explicit water molecules were carried out. Stabilization energies for the interactions of specific regions of the inhibitor and individual amino acid residues were calculated. The alkyl spacer is located in a cleft along helix 6 of MGMT. With increasing spacer length there is increasing interaction with several amino acid residues which play an important role in the proposed nucleotide flipping mechanism required for DNA repair.

Full text of DOI:10.1021/jm010006e

4050
J . Med. Chem. 2001, 44, 4050-4061
Mon osa cch a r id e-Lin k ed In h ibitor s of O6-Meth ylgu a n in e-DNA Meth yltr a n sfer a se
(MGMT): Syn th esis, Molecu la r Mod elin g, a n d Str u ctu r e-Activity Rela tion sh ip s
J ost Reinhard,† William E. Hull,‡ Claus-Wilhelm von der Lieth,‡ Uta Eichhorn,§ Hans-Christian Kliem,†
Bernd Kaina,‡ and Manfred Wiessler*,†
Division of Molecular Toxicology and Central Spectroscopy Department, German Cancer Research Center, Postfach 101949,
D-69009 Heidelberg, Germany, and Division of Applied Toxicology, Institute of Toxicology, University of Mainz,
Obere Zahlbacher Strasse 67, D-55131 Mainz, Germany
Received J anuary 5, 2001
A series of potential inhibitors of the human DNA repair protein O6-methylguanine-DNA
methyltransferase (MGMT) were synthesized, characterized in detail by NMR, and tested for
their ability to deplete MGMT activity in vitro. The new compounds, ω-[O6-R-guan-9-yl]-(CH2)n-
â-d-glucosides with R ) benzyl or 4-bromothenyl and ω ) n ) 2, 4, ... 12, were compared with
the established inhibitors O6-benzylguanine (O6-BG), 8-aza-O6-benzylguanine (8-aza-BG), and
O6-(4-bromothenyl)guanine (4-BTG), which exhibit in an in vitro assay IC50 values of 0.62,
0.038, and 0.009 µM, respectively. Potential advantages of the glucosides are improved water
solubility and selective uptake in tumor cells. The 4-BTG glucosides with n ) 2, 4, 6 show
moderate inhibition with an IC50 of ca. 0.5 µM, while glucosides derived from BG and 8-aza-
BG showed significantly poorer inhibition compared to the parent compounds. The 4-BTG
glucosides with n ) 8, 10, 12 were effective inhibitors with IC50 values of ca. 0.03 µM. To
understand this behavior, extensive molecular modeling studies were performed using the
published crystal structure of MGMT (PDB entry: 1QNT). The inhibitor molecules were docked
into the BG binding pocket, and molecular dynamics simulations with explicit water molecules
were carried out. Stabilization energies for the interactions of specific regions of the inhibitor
and individual amino acid residues were calculated. The alkyl spacer is located in a cleft along
helix 6 of MGMT. With increasing spacer length there is increasing interaction with several
amino acid residues which play an important role in the proposed nucleotide flipping mechanism
required for DNA repair.
In tr od u ction
as pseudosubstrates with MGMT and can completely
abolish its activity.
The human DNA repair protein O6-methylguanine-
DNA methyltransferase (MGMT) plays a critical role
in cancer therapy with alkylating reagents such as alkyl
nitrosoureas and alkyl triazenes. Among the resulting
DNA adducts, O6-alkylguanines are considered to be of
major importance for the induction of cancer, mutation,
and cell death. These adducts direct the incorporation
of either thymine or cytosine without blocking DNA
replication, resulting in GC to AT transition mutations.
MGMT provides protection against such lesions by
removing the O6-alkyl adduct and transferring it to a
cysteine acceptor group within the enzymes’s active
site.1-4 The resulting MGMT-thioether is not reacti-
vated; instead, the protein is ubiquitinated and de-
graded. Thus, MGMT is considered to be a suicide
enzyme since the transfer is not reversible and activity
can be recovered only by de novo protein synthesis.
Consequently, tumor cells and neoplastic tissue with
elevated MGMT levels can be resistant to alkylating
therapeutics but can be resensitized by depletion of
MGMT.5-7 Derivatives of the free base guanine react
Although potent inhibitors of MGMT have been found
and introduced into adjuvant therapy,4,8,9 they all have
major disadvantages. In all cases no selectivity for
neoplastic tissue has been achieved, leading to reduced
repair capacity in normal tissue and loss of protection
against the systemic side effects of chemotherapy. O6-
Benzylguanine (O6-BG), a well-established inhibitor
tested in phase I clinical trials,10,11 exhibits poor water
solubility, which limits its applications. Another very
promising compound, 4-(benzyloxy)-2,6-diamino-5-ni-
trosopyrimidine,9,12 has the disadvantage of a short half-
life in vivo. The compounds 8-aza-O6-benzylguanine and
O6-(4-bromothenyl)-guanine have IC50 values for MGMT
inhibition which are, respectively, 3 times and 75 times
lower than for O6-BG.8,9 However, these more potent
inhibitors did not show tumor selectivity in animal
testing.13-15 To overcome these disadvantages, ap-
propriate modifications in the pharmacophore are nec-
essary.
Previously published results6,8,9,16-20 indicate that
basically only three positions in the purine core are
suitable for structural variations, namely O,6 C8, and
N9. For example, specific modifications to substituents
at O6 or C8 may alter the electronic properties of the
purine ring system and enhance the leaving ability of
the alkyl group at O6.9,17 For modifications at N9, it is
* To whom correspondence should be addressed. Tel: +49-6221-
423311. Fax: +49-6221-423375. E-mail: m.wiessler@dkfz.de.
† Division of Molecular Toxicology, German Cancer Research Center.
‡ Central Spectroscopy Department, German Cancer Research
Center.
§ University of Mainz.
10.1021/jm010006e CCC: $20.00 © 2001 American Chemical Society
Published on Web 10/31/2001

O6-Methylguanine-DNA Methyltransferase
J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 24 4051
known that highly polar groups lead to a dramatic
decrease in MGMT depletion activity, whereas bulky
apolar groups do not hinder the ability to inhibit
MGMT.17,21 Not only substituent effects but also MGMT-
DNA interactions can influence ligand affinity since
DNA binding leads to steric hindrance of the competitive
interaction with small-molecule inhibitors.21
is enhanced. Glucose conjugates with antitumor activity,
so-called “trojan horses”, are accepted by SAAT1 and,
therefore, accumulate preferentially in neoplastic tissue.
Thus, appropriate glucosyl derivatives of MGMT inhibi-
tors should provide targeted depletion of MGMT activity
in neoplastic tissue and the desired reduction in MGMT-
based therapy resistance while maintaining MGMT-
based protection of normal tissue and minimizing
toxicity.
A better understanding of the molecular mechanism
of MGMT and structure-activity relationships for in-
hibitors can be obtained by computational approaches
based on the available X-ray crystal structures of
MGMT and biological data derived from in vitro studies.
The first alkyltransferase (ATase) of the MGMT family
to be crystallized was the 19 kDa C-terminal domain of
the ada gene product from Escherichia coli.22 Wibley et
al.23 reported the first homology model of the human
protein based on the crystal structure of the E. coli
ATase. A distortion of the DNA substrate was discussed
and later specified in terms of the proposed flipping
mechanism for the alkylated guanine,24 based on de-
scriptions of the analogous phenomenon for other
methyltransferases.25-28 Recently, the X-ray crystal
structures of the human enzyme29 and its benzylated
and methylated products30 have been determined. Al-
though a homology model suggested a deeply buried
active site, the crystal structure revealed that in the
human protein the active site is reasonably accessible
for pseudosubstrates. Site-directed mutagenesis studies
indicated that several amino acid residues are involved
in the MGMT reaction mechanism. The residues Arg128,
Tyr114, and Lys165 were found to be necessary for DNA
binding,31-33 while Pro140, Gly156, Tyr158, and Gly160
proved to be essential for reaction with either O6-BG or
oligonucleotides containing O6-methylguanine. Tyr114,
conserved in all known sequences, was shown to play a
key role in substrate specificity, DNA binding, and
protection against mutations induced by methylating
agents.34 Cells expressing the K165A mutant of MGMT
were still protected by the enzyme from the mutagenic
effects of O6-methylating agents and could not be
sensitized by O6-BG.33 Similar results were obtained
for various Gly156 mutants as well as for mutations
at Cys150, Asn157, Tyr158, Ser159, Leu168, and
Leu169.31,35 At codon 160 mutations can lead to either
increased sensitivity (G160A, G160W) or resistance
toward O6-BG (G160R and many others).36 The mutants
P140K and PVP(138-140)MLK showed a 10 000-fold
decrease in sensitivity against O6-BG.
To achieve this goal, it is essential to gain insight into
the molecular structure requirements of MGMT inhibi-
tors and the characteristics which lead to enhanced
MGMT depletion. Since polar groups at purine N9 are
unfavorable, we have introduced various [-O(CH2)n-]
alkyl spacers between C1 of â-d-glucose and N9 of
guanines bearing various O6 substituents. Here we
present the synthesis, purification, and spectroscopic
characterization of these novel MGMT inhibitors. The
results of initial in vitro inhibition assays are compared
with interaction energies obtained from detailed mo-
lecular modeling studies of MGMT-inhibitor complexes.
A detailed report concerning MGMT inhibition in tumor
cell cultures is presented elsewhere.39
Resu lts a n d Discu ssion
Syn th esis. A series of protected alkyl-â-d-glucopy-
ranosides (see Chart 1) was prepared according to step
A in Chart 2. The alkyl group was activated for
subsequent reaction by a bromine substituent in the ω
position. Alkyl groups (CH2)n were used as spacers with
n ) 2, 4, 6, 8, 10, 12 for compounds 1-6. For compound
7 a 2-butynyl moiety was employed. It was necessary
to protect the glucose hydroxyl groups with either acetyl
or benzoyl substituents. Since these are lost under basic
conditions, it was crucial to work in a water-free
environment.
The guanine derivative 8 (O6-benzyl) was synthesized
from 6-chloroguanine using a standard method for O6
substitution by nucleophilic replacement of the halo-
gen.40 Derivative 10 (O6-(4-bromothenyl)) was prepared
from 6-chloroguanine by an improved method intro-
duced by McElhinney et al.,8 whereby the chlorine is
first replaced by a trimethylamino group to give a
trimethylammonium salt. Subsequent displacement of
this group by an alcoholate gives higher overall yields
and can be performed at room temperature. The 8-aza
analogue 9 (O6-benzyl, 8-aza) was obtained by ring
closure of 6-benzyloxy-2,3,5-triaminopyrimidine using
nitrite under reductive conditions.41,42
Although data for purine-based inhibitors with sub-
stituents at N9 have been reported, the details of how
the substituents influence activity have not been elu-
cidated. Therefore, we have synthesized a set of such
compounds to determine the required structural fea-
tures of N9 substituents that are compatible with
MGMT depletion. Glucosidation represents a general
strategy for improving both drug solubility and target-
ing, two weaknesses of the inhibitors discussed above.
For example, â-d-glucosylisophosphoramide mustard
(Glu-IPM) exhibits in animal models the same thera-
peutic potency as isophosphoramide mustard but less
systemic toxicity.37 Glu-IPM is transported by SAAT1,
an active, membrane-bound, sodium-dependent glucose
cotransporter.38 In tumors the expression of transport
proteins for essential energy donors such as saccharides
The family of protected, halogen-activated alkyl-â-d-
glucosides was then used to alkylate position N9 of the
guanine derivatives (step B in Chart 2) to give the odd-
numbered protected compounds 11-27. Deprotection
(step C in Chart 2) then led to the even-numbered
unprotected glucosides 12-28. A problem with this
alkylation strategy is that N9 and N7 of the purines
are attacked equally well by suitable electrophiles;
however, N7-substituted MGMT inhibitors are inactive.
To maximize the N9/N7 substitution ratio, lithium
hydride was employed to activate the imidazole ring,
as demonstrated by Kjellberg and Liljenberg.43
All isolated products were characterized in detail by
spectroscopic methods as described in the Materials and

4052 J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 24
Ch a r t 1. Synthesized Compounds
Reinhard et al.
Ch a r t 2. Reaction Schemes
Methods. For the unprotected glucosides NMR studies
were carried out at 11.7 T (500 MHz for 1H) so that
nearly complete and unambiguous assignments for all
1H and 13C resonances could be obtained for the com-
plete family of inhibitors (provided as Supporting
Information). NMR provided the only means for con-
firming the isomeric and stereochemical purity of the
isolated compounds (â anomer, N9 vs N7 substitution).
The parallel synthesis of a family of glucosides al-
lowed us to prepare the present set of guanine deriva-
tives under a standardized set of conditions. Thus, this
method is suitable for generating even larger libraries
of analogous structures. Regarding purification by
column and flash chromatography, we found that for
the compounds with long spacers, such as 24, there was
no advantage in performing a purification of the pro-

O6-Methylguanine-DNA Methyltransferase
J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 24 4053
Ta ble 1. MGMT Depletion Assay Using Cell-Free Extracts from HeLa S3 Cellsa
compound
IC50 [µM]
O6-benzylguanine (8)
0.62
25.0
0.038
2.0
0.009
0.680
0.450
0.450
0.032
0.030
0.030
0.250
2-[O6-benzylguan-9-yl]-ethyl-â-d-glucoside (12)
8-aza-O6-benzylguanine (9)
2-[8-aza-O6-benzylguan-9-yl]-ethyl-â-d-glucoside (28)
O6-(4-bromothenyl)-guanine (10)
2-[O6-(4-bromothenyl)-guan-9-yl]-ethyl-â-d-glucoside (14)
4-[O6-(4-bromothenyl)-guan-9-yl]-butyl-â-d-glucoside (16)
6-[O6-(4-bromothenyl)-guan-9-yl]-hexyl-â-d-glucoside (18)
8-[O6-(4-bromothenyl)-guan-9-yl]-octyl-â-d-glucoside (20)
10-[O6-(4-bromothenyl)-guan-9-yl]-decyl-â-d-glucoside (22)
12-[O6-(4-bromothenyl)-guan-9-yl]-dodecyl-â-d-glucoside (24)
4-[O6-(4-bromothenyl)-guan-9-yl]-but-2-ynyl-â-d-glucoside (26)
a
Detailed inhibition studies are presented in ref 39.
tected glucoside; good separations were obtained with
the deprotected compounds. However, for the shorter
spacers (ethyl, butyl), it proved to be very difficult to
separate the N9- and N7-derivatives (distinguishable
by NMR) in the deprotected glucoside form; in these
cases purification of the protected species was necessary
and successful.
MGMT In h ibition . In a first round of synthesis the
glucosides 12, 14, and 28 were prepared and compared
with their corresponding parent guanines 8, 10, and 9
in terms of their ability to inactivate MGMT (Table 1).
As expected from previous results, N9 alkylation of the
guanine leads to an increase in IC50 (factor 40 for 12 vs
8; factor 53 for 28 vs 9; factor 76 for 14 vs 10). The initial
results clearly showed that the 4-bromothenyl (4-BT)
derivative 10 was the most active inhibitor (IC50 ) 0.009
µM) and that its glucoside 14 (IC50 ) 2.0 µM) was nearly
as potent as the parent benzyl derivative 8 (O6-BG).
Therefore, in a second round of synthesis only the 4-BT
guanine derivative was used for coupling to glucosides
with various spacers, leading finally to the deprotected
compounds 16, 18, 20, 22, 24, and 26.
F igu r e 1. Superposition of the modeled structures of [S-
benzyl Cys145]MGMT (from the crystal structure data, PDB
entry: 1EH8, peptide CR backbone ) white/gray ribbon,
S-benzyl group ) brown ball-and stick model, arrow points to
Câ) and the MGMT:14 complex (magenta ribbon and red stick
model). The complex represents a snapshot from a molecular
dynamics simulation which included explicit water molecules
and used the crystal structure of native MGMT (PDB entry:
1QNT) as a template.
octyl-â-d-glucoside (20). During this work the crystal
structures of the methyl and benzyl Cys145-thioether
adducts of MGMT were published (PDB entries: 1EH7
(resolution 2.0 Å, R ) 0.199) and 1EH8 (resolution 2.5
Å, R ) 0.194), respectively).30 To test our simulation
results for the MGMT complex with 14, we superim-
posed the backbones of the 1000 structures (snapshots)
obtained during the MD run onto the crystal structure
of S-benzyl MGMT. The RMS deviation for all backbone
atoms was 0.79 ( 0.30 Å (mean ( SD), and the mean
distance between the center of the benzyl ring in the
covalently modified protein and the center of the 4-BT
thiophene ring in our complexes was only 1.02 ( 0.33
Å (range: 0.29-1.71 Å). This correspondence between
the modeled complex with 14 and the Cys145-benzy-
lated enzyme is illustrated in Figure 1 for one of the
MD-derived structures. Thus, the O6 substituent of the
tested ligand adopts a position within the binding pocket
of MGMT which corresponds closely (but not exactly)
to the position of the benzyl group attached covalently
to Cys145 via reaction with O6-BG. This finding was
also confirmd by an MD simulation for the MGMT
complex with ligand 20 (octyl spacer).
The experimentally determined IC50 values for these
additional compounds are also listed in Table 1. With
the butyl or hexyl spacer (16 or 18), the IC50 is 0.45 µM.
For longer spacers with n ) 8, 10, 12 (compounds 20,
22, and 24) there is a significant improvement in
inhibition with IC50 ) ca. 0.03 µM in each case. Finally,
for the more rigid medium-sized but-2-ynyl spacer an
intermediate IC50 of 0.25 µM was observed. Thus,
essentially equal and maximum activities for glucosides
were achieved with a flexible spacer of 8-12 methylene
groups. While the IC50 values for compounds 20, 22, and
24 are still about a factor of 3 higher than for the parent
guanine derivative 10, the glucosides are expected to
offer significant advantages in tumor selectivity, as
described in the Introduction.
Molecu la r Mod elin g. To gain more insight into the
experimentally observed structure-activity relation-
ships, we constructed a 3D model of the MGMT-ligand
complex using molecular dynamics (MD) simulations
(including solvent water molecules) and docking experi-
ments as described in the Methods. When we began
these modeling experiments, only the coordinates of the
crystal structure for O6-alkylguanine-DNA methyl-
transferase were available (Protein Databank (PDB)
entry 1QNT, resolution 1.90 Å, R value 0.198). These
data served as a basis for MD simulations of MGMT
complexes with 2-[O6-(4-bromothenyl)-guan-9-yl]-ethyl-
â-d-glucoside (14) and 8-[O6-(4-bromothenyl)-guan-9-yl]-
The crystal structure of the S-benzylated MGMT
product (PDB entry: 1EH8) reveals that Tyr158 and
Asn137 flank the S-benzylcysteine side chain so closely
that free rotation around either the cysteine or benzylic
CH2-S bond is not possible. A detailed interaction
analysis for the MGMT:20 complex was performed (see

4054 J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 24
Reinhard et al.
F igu r e 2. Definition of the structural subunits of MGMT
inhibitors (example: compound 24) for which enzyme:inhibitor
interaction analysis was performed by molecular modeling.
The subunits Sp1-Sp6 represent successive C2 of the spacer,
beginning at the guanyl N9 position.
F igu r e 3. Snapshot taken from a molecular dynamics simu-
lation of the MGMT:20 complex (octyl spacer). The crystal
structure of native MGMT (PDB entry: 1QNT) was used as a
template. The inhibitor is represented as a stick model
(green: C, white: H, red: O, blue: N, brown: Br, yellow: S).
The water-accessible surface of the protein is shown using
different colors for each amino acid residue. This view looks
into the “tunnel” region, which interacts with the guanine
moiety, and shows the O6 substituent emerging in the fore-
ground (lower right).
Ta ble 2. Local Interaction Energy Analysis for the Guanyl
Fragment of Compound 20 in Its Complex with MGMTa
guanyl regionb
residue
LTHO
LLNK
LPU1
LPU2
sum
Tyr 114
Gly 131
Met 134
Ar g 135
Asn 137
P r o 140
Cys 145
Va l 148
Asn 157
Tyr 158
Ser 159
Gly 160
su m
0.2
0.7
0.5
1.2
0.4
0.1
0.9
0.6
0.6
0.8
2.6
3.9
1.1
1.1
1.2
0.5
2.2
6.0
2.9
0.7
23.7
0.5
1.2
1.1
1.1
0.2
0.8
0.9
0.3
0.7
0.5
0.5
1.5
0.5
1.7
1.6
0.5
1.5
1.0
0.7
7.2
1.4
1.0
4.4
6.3
5.8
a
Mean values for the local interaction energies (nonbonded van
der Waals and Coulomb energies in kcal/mol) are negative and
represent contributions to the stabilization of the MGMT:20
complex, as determined from 1000 structures in the molecular
dynamics archive. Results are rounded to one decimal place and
b
values < 0.04 are not shown. As defined in Figure 2.
F igu r e 4. Snapshot taken from a molecular dynamics simu-
lation of the MGMT:24 complex (dodecyl spacer). The color
coding is the same as in Figure 3, but here the complex has
been rotated about 90° around a vertical axis to give a better
view of the cleft into which the spacer region of the inhibitor
fits. The glucose and guanine regions are, respectively, at the
left and right ends of the inhibitor molecule.
Methods), whereby the local interaction energies be-
tween specific portions of the ligand molecule (as defined
in Figure 2) and specific amino acid residues were
calculated. The complete results of this study for seven
4-BT ligands are provided in the form of Supporting
Information. Selected results for four regions of the
guanyl portion of ligand 20 are summarized in Table 2,
with negative nonbonded interaction energies repre-
senting a stabilization of the enzyme-ligand complex.
The interaction of the O6-substituent 4-BT (region
LTHO) with the enzyme (-7.2 kcal/mol) represented
about 30% of the total enzyme-guanyl stabilization
energy (-23.7 kcal/mol). The strongest interactions
involved Tyr158, Arg135, and Asn137 (-1.5, -1.2, -1.1
kcal/mol), consistent with the geometry of S-benzyl
MGMT discussed above. Weaker interactions of the O6
substituent of 20 were observed with Pro140, Ser159,
and Gly160 (-1.1, -1.0, -0.7 kcal/mol), whose impor-
tance in the binding of O6-BG (8) has been discussed in
several publications.31,33,36,44 The weaker interactions of
20 can be rationalized on the basis that in the complex
the mean distances between the 4-BT group (center of
ring) and the CR of Pro140, Ser159, and Gly160 range
from 5.7 to 5.9 Å and are 0.5-1.4 Å larger than the
corresponding distances in S-benzyl MGMT.
19% of the total interaction energy and exhibited
significant contacts with four of the residues that
interacted with the thiophene itself, namely Met134,
Arg135, Tyr158, and Ser159. The purine system (LPU1
+ LPU2) generated about 51% of the interaction with
MGMT, with about equal contributions coming from the
five- and six-membered rings. The major interactions
involved Met134, Arg135, Asn157, Tyr158, and Ser159
with minor contributions from Tyr114, Gly131, Asn137,
Pro140, Cys145, Val148, and Gly160 (Table 2). As can
be seen in Figures 3 and 4 several of these residues form
a tunnel-like pocket in the active site of MGMT which
can accommodate the guanyl portion of the ligand. At
the end of the tunnel are the residues Asn137, Pro140,
and Gly160 which interact only with the O6 substituent.
We find that the orientation of the guanyl portion of
the ligand is very similar to that modeled by Wibley et
al.23,29 for the ligand O6-BG. Therefore, we believe that
our computational approach provides realistic models
for MGMT complexes with our new ligands, such as 20
(Figure 3).
The region of the ether linkage (LLNK) in 20 between
the thiophene and the guanine moieties supplied about

O6-Methylguanine-DNA Methyltransferase
J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 24 4055
Ta ble 3. Interaction Energy Analysis for Individual Amino Acid Residues in MGMT:Inhibitor Complexesa
compound spacer
14 ethyl
16 butyl
18 hexyl
20 octyl
22 decyl
24 dodecyl
26 but-2-ynyl
region
A
B
A
B
A
B
A
B
A
B
A
B
A
B
Ser 113
Tyr 114
Gln 115
Lys 125
Ala 127
Ar g 128
Gly 131
Gly 132
Met 134
Ar g 135
Gly 136
Asn 137
P r o 138
Va l 139
P r o 140
Ile 141
Cys 145
His 146
Va l 148
Va l 149
Cys 150
Ser 151
Ser 152
Ala 154
Va l 155
Asn 157
Tyr 158
Ser 159
Gly 160
Gly 161
Va l 164
Lys 165
Leu 168
su m
0.5
4.3
2.8
0.5
3.3
2.8
0.1
5.0
1.5
0.1
3.9
1.5
0.6
0.3
3.6
0.7
0.1
1.3
1.6
1.8
0.1
2.5
3.0
0.1
0.8
2.8
0.7
0.1
1.3
1.6
0.1
2.6
0.6
2.0
0.6
3.8
1.1
0.5
1.6
3.8
1.2
3.2
1.1
0.5
1.6
3.8
0.4
1.0
0.3
1.4
3.0
0.5
1.4
3.0
0.4
0.7
1.8
1.4
0.1
2.2
2.1
1.5
1.5
0.7
1.8
0.7
0.5
0.9
1.1
0.5
0.9
0.1
0.1
0.9
1.1
0.1
2.1
2.2
0.1
0.9
0.4
1.4
0.6
0.2
0.4
4.7
0.1
0.3
1.4
0.5
0.2
2.7
2.3
2.6
3.9
1.9
2.8
2.6
0.1
0.6
0.1
0.6
0.2
1.1
0.8
0.2
0.1
0.8
0.4
0.3
0.7
0.2
0.9
0.6
0.2
0.3
1.3
0.9
0.6
2.2
1.4
0.4
1.1
1.2
0.5
0.3
1.0
0.4
1.1
0.2
1.6
1.2
0.9
0.1
0.4
0.1
0.5
0.1
0.5
1.0
0.9
1.3
0.2
1.7
1.0
5.6
3.1
1.5
1.0
0.9
1.3
0.2
1.7
0.3
0.3
0.6
0.4
0.3
0.6
0.4
0.3
1.2
0.9
6.8
3.1
1.0
0.6
0.5
0.1
0.1
33.1
1.1
0.2
0.5
1.0
5.4
2.3
0.7
0.5
0.4
0.1
0.3
0.4
2.1
6.4
2.7
0.5
0.1
1.0
0.4
2.9
6.0
2.9
0.7
0.7
2.0
5.7
2.4
1.2
0.7
1.4
3.9
0.7
0.1
0.4
0.2
0.3
21.5
5.8
30.0
8.5
9.7
34.9
11.3
37.2
15.8
29.7
9.1
19.1
2.5
a
Mean values for interaction energies (nonbonded van der Waals and Coulomb energies in kcal/mol) are negative and represent
contributions to the stabilization of the MGMT:20 complex, as determined from 1000 structures in the molecular dynamics archive; A )
total interaction energy for complete ligand; B ) interaction energy for glucosyl + spacer region only (N9 substituent). Only residues
with total interactions (A) > 0.1 are shown; results are rounded to one decimal place and values <0.04 are not shown.
The favored orientation of the purine ring system
determined by modeling directs the alkyl spacer at N9
along a cleft on the protein surface near helix 6. This is
illustrated in Figure 4 for compound 24. For the
complete series of 4-BT ligands with various spacers,
Table 3 summarizes the interaction energies between
individual amino acid residues and the entire ligand
(column A) or just the glucosyl+spacer region (N9
substituent) (column B). When the entries in columns
(A) and (B) in Table 3 are nearly equal, then the
corresponding amino acid residue interacts primarily
with the N9 substituent rather than the guanyl portion
of the ligand. This is the case for Tyr114, Gln115,
Ala127, Arg128, Gly131, and Ser152, which have been
identified to play an important role in the proposed
nucleotide flipping mechanism required for DNA re-
pair,30 and also for Lys125, Cys150, Ser 151, and
Val155. Thus, our model of the MGMT:inhibitor complex
and the data of Table 3 reveal that our ligands can
interact strongly with and therefore block residues
normally involved in DNA binding.
general, the interactions of the alkyl spacer and the
glucose moiety with the face of the binding cleft are
significantly weaker than the interactions of the “active”
guanyl portion of the ligand in the binding “tunnel”. For
the inhibitor with the shortest ethyl spacer (14) or the
more rigid butynyl spacer (26), the interaction energies
for both the spacer and the glucose moiety are low. For
these ligands the bulky glucose fits poorly into the
binding cleft and prohibits an optimal fit of the guanyl
portion, whose interaction energy is reduced to ca. 16
kcal/mol. With longer, more flexible spacers (butyl to
dodecyl) the guanyl moiety can achieve its maximal
stabilization in the binding pocket (21-23 kcal/mol),
while the spacer itself can achieve up to 9 kcal/mol
stabilization in the hydrophobic cleft. Maximum inter-
action with the N9 substituent is achieved with the
decyl spacer (22). With the dodecyl spacer (24) the
glucose moiety extends outside of the binding cleft and
contributes little to the total binding interaction (Figure
4).
Figure 5 presents a correlation diagram for the
calculated interaction energies (total, spacer+glucose,
spacer alone) and the experimental IC50 values of the
4-BT glucoside inhibitors studied. While the data for the
family of six inhibitors with (CH2)n spacer show a crude
negative correlation between the energies and IC50, the
data for the unique inhibitor 26 with butynyl spacer
The MD-derived interaction energies for MGMT:
inhibitor complexes can also be viewed in terms of the
local contributions made by each of the 15 structural
elements of the ligand (Figure 2), summed across all
amino acid residues involved, as shown in Table 4 for
the 4-BTG glucosides 14, 16, 18, 20, 22, 24, and 26. In

4056 J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 24
Reinhard et al.
Ta ble 4. Total Interaction Energies for Specific Inhibitor Regions in MGMT:Inhibitor Complexesa
compound (spacer)
row
region
LTHO
LLNK
LPU1
LPU2
gu a n yl
Sp1
Sp2
Sp3
Sp4
Sp5
Sp6
sp a cer
Glc1
Glc2
Glc3
14 (ethyl)
16 (butyl)
18 (hexyl)
20 (octyl)
22 (decyl)
24 (dodecyl)
26 (but-2-ynyl))
1
2
3
4
5.5
2.3
4.2
3.8
15.7
2.0
6.5
3.9
5.9
5.1
21.5
1.7
8.6
3.6
6.2
5.0
23.4
2.1
7.2
4.4
6.3
5.8
23.7
1.5
1.3
2.0
1.6
6.7
3.9
6.0
4.8
21.3
2.1
1.9
2.3
1.9
0.9
7.5
3.7
4.9
4.5
20.6
1.2
1.5
1.6
1.1
1.1
1.1
7.4
0.6
0.7
0.0
0.0
0.4
1.6
29.7
6.8
2.5
3.8
3.5
16.6
0.7
∑ 1-4
5
6
7
8
9
10
2.3
1.6
1.9
0.9
∑ 5-10
2.0
0.8
0.2
0.3
0.4
2.1
3.8
21.5
4.0
0.8
1.1
0.3
0.3
2.0
4.5
30.0
5.6
1.1
0.9
0.4
0.3
1.5
4.1
33.1
6.4
2.0
0.4
0.2
0.3
2.0
4.9
34.9
9.0
2.5
2.1
1.1
0.3
0.8
6.8
37.2
1.6
0.3
0.2
0.0
0.0
0.5
1.0
19.1
11
12
13
14
Glc4
15
Glc56
glu cosyl
w h ole in h ibitor
∑ 11-15
∑ 1-15
a
Energies are negative and in kcal/mol (see footnote in Table 3); structural regions are defined in Figure 2.
and for the total energy and may be due to the complex
influence of spacer length on the interactions between
protein and the glucosyl and guanyl subunits. However,
due to the crudeness of the methods used for determin-
ing interaction energies and IC50 values and the rela-
tively small differences observed, a more detailed quan-
titative analysis does not seem warranted at this time
and would require the measurement of true binding
constants and the calculation of free energies of binding
(∆G).
In summary, the modeling studies provide us with a
simple rationale for interpreting the experimental MGMT
inhibition assay results, at least on a qualitative basis.
Glucosides 12, 14, and 28 with C2 spacers are signifi-
cantly poorer ligands compared to their parent guanines
8, 9, and 10. The modeling experiments with 14 show
that this can be explained by the poor fit of the glucosyl
unit into the MGMT binding groove and the resulting
nonoptimal binding geometry for the guanyl unit (lower
interaction energy). With C4 and C6 spacers (16, 18) a
substantial improvement in interaction energy but only
a modest improvement in inhibition was observed. The
best inhibitors (20, 22, and 24) with spacers of 8-12
carbons exhibited the strongest protein-spacer interac-
tions, and with the C10 spacer a maximum in protein-
spacer and protein-glucose interaction was achieved.
Elongation of the spacer from 10 to 12 carbons provided
no improvement in inhibition and nearly eliminated
binding stabilization via the glucosyl moiety. In addi-
tion, longer alkyl spacers, which are undesirable on the
basis of solubility, can show diminishing returns due
to the increased entropic penalty paid to form a complex
with reduced degrees of rotational freedom. Thus, for
more detailed quantitative structure-activity relation-
ships it will also be important to consider the reduction
in enthalpy required to orient the spacer and glucose
unit in the appropriate conformation which provides a
good “fit” to the surface of the protein and a maximum
in stabilization energy. Finally, the MGMT:inhibitor
model allows us to investigate local protein-spacer
interactions in more detail and to design and test more
structured spacers which may provide specific enhanced
binding interactions.
F igu r e 5. Correlation plot for MGMT:inhibitor interaction
energies and IC50. The vertical axis represents the calculated
interaction energies (dH in kcal/mol) for the total ligand, the
spacer+glucose subunit (sp+GLC), and the spacer alone (sp);
the horizontal axis represents the IC50 values estimated from
the in vitro assay (Table 1). Data points are shown for
compounds 14, 16, 18, 20, 22, and 24 and were analyzed by
linear regression (solid line: spacer alone, r2 ) 0.82). Data
points for the unique ligand (26) with butynyl spacer are
marked with an arrow and were omitted from the analysis.
deviate significantly from the trend. Although the
interaction energy for 26 is 11 kcal/mol less than for 16
(butyl spacer), the IC50 for 26 is actually about a factor
of 2 lower than for 16. This apparent conflict suggests
that the modeling of the MGMT:26 complex may be
inadequate and deserves further study. For the inhibi-
tors with (CH2)n spacer a simple linear regression
analysis indicates that the best correlation is obtained
between IC50 and the interaction energy for the alkyl
spacer alone (Figure 5, r2 ) 0.82). A poorer correlation
(r2 ) 0.54) is observed for the spacer+glucose energy

O6-Methylguanine-DNA Methyltransferase
J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 24 4057
[Ph.D. Thesis: Conformational Analysis of Oligosaccha-
rides in free and bound states; University of Heidelberg,
(http://www.ub.uni-heidelberg.de/archiv/605/)]. The program
PAVDO was used to extract relevant details from the large
DISCOVER output files and to generate tables and lists which
were used as input for the CAT software (conformational
analysis tools; http://www.dkfz-heidelberg.de/spec/martin/
cat2000). CAT was used to convert the raw data into the
desired interaction matrices and to calculate the desired
averages. Each MD simulation required ca. 7 days of computa-
tion with four parallel processors on an IBM SP2 computer.
Ma ter ia ls a n d Meth od s
MGMT Assa y. The MGMT assay has been described
previously.45 Briefly, 200 µg of extracted cellular protein from
HeLaS3 cells (ATCC, Rockville, MD) in 200 µL of 70 mM
HEPES buffer (with 1 mM dithiothreitol (DTT), 5 mM EDTA,
pH 7.8) was incubated at 37 °C with a defined concentration
of MGMT inhibitor (added as a DMSO solution). After 30 min
an excess of [3H]-methylated DNA (120 000 cpm) was added,
and the incubation was continued for an additional 90 min.
The reaction was stopped by the addition of 400 µL TCA (13%),
and the DNA was hydrolyzed by heating the reaction mixture
for 30 min at 98 °C. The precipitated protein was washed three
times with 400-µL portions of 5% TCA, solubilized in 0.1 N
NaOH, and analyzed by liquid scintillation counting using the
cocktail Rotiszint eco plus (Roth, Karlsruhe) and a TRI-CARB
2100 TR liquid scintillation analyzer (Canberra-Packard,
Dreieich). Enzyme activity was expressed as fmol of [3H]methyl
transferred to TCA-insoluble protein material per mg of total
cellular protein. Percent inhibition was calculated relative to
untreated control samples. Each assay was repeated three
times, and IC50 values were determined graphically from plots
of percent inhibition vs inhibitor concentration.
Molecu la r Mod elin g. The crystal structure of the alkyl-
transferase MGMT (PDB entry: 1QNT, resolution 1.90 Å, R
value 0.198) was used to derive a model for ligand binding in
the active site. To reduce the required computational time for
molecular dynamics, the amino acid residues 1-82 of the
N-terminal region were deleted since these residues do not
participate in the active site around Cys145 which is com-
pletely located within residues 83-176 of the C-terminal
domain. Hydrogen atoms were added to the crystal structure
assuming a pH of 7. The atom types and partial charges
defined for the CFF91 force field were assigned using the
automatic procedures of the modeling software INSIGHT II
(Molecular Simulations, Inc., San Diego, CA). The ligand
molecules were constructed manually using the BUILDER
option of INSIGHT II. The guanine portion of the ligand
molecules was initially docked by hand into the active site
according to the orientation found for O6-benzylguanine in
previous modeling experiments by Wibley et al.23,29
An a lytica l Meth od s. HPLC separations were carried out
on a J asco HPLC system, consisting of a PU980 pump, a
DG980-50 ternary gradient unit, a LG980-02 three-line de-
gaser, and a UV975 intelligent UV-vis detector. Solvents were
of the highest purity available; water was degassed under
vacuum, other solvents by ultrasound. A variety of HPLC
methods and solvent gradients were employed and are desig-
nated here by roman numerals (see Table B, Supporting
Information). For methods I, III, IV, VI, VII, VIII, X, and XII
a Lichrospher column (100-RP18-(E)-5 µm; 250 × 4 mm) and
methanol/water at a flow rate of 1 mL/min were used with
UV detection at 254 nm. Methods II and XI utilized a
Lichrospher column (100-RP18-(E)-5 µm; 250 × 4 mm), aceto-
nitrile/water at a flow rate of 1 mL/min, and UV detection at
254 nm. For methods V and IX the setup was as follows:
Purospher column (Si80-5 µm; 250 × 4.6 mm), chloroform/
methanol at 1 mL/min, UV detection at 254 nm. TLC analysis
was performed with solvents of the highest purity available,
and mixtures are reported in %(vol/vol). Precoated plastic
sheets were from Macherey-Nagel, Type Polygram SIL G/UV254
.
Column and flash chromatography were carried out with silica
gel from Macherey-Nagel, Type Silica Gel 60, either 70-230
mesh ASTM (column chromatography) or 230-400 mesh
ASTM (flash chromatography).
Mass spectrometry was performed with a Finnigan MAT
TSQ-7000 using electrospray ionization (ESI) or with a J EOL
J MS 700 in high-resolution mode using fast-atom bombard-
ment (HR-FAB). Uncorrected melting points were determined
with a Bu¨chi apparatus. UV spectra were recorded with a
Hitachi Uvikon 931 spectrophotometer.
The protein-ligand complex was placed in the center of a
box (50 × 40 × 50 Å) which was filled with 2720 water
molecules. The program DISCOVER (Molecular Simulations,
Inc., San Diego, CA) was used to perform a molecular dynamics
(MD) simulation with explicit solvent molecules under a
standardized protocol for all complexes investigated. To ensure
that the tertiary structure of the protein was maintained
during the simulation, all backbone atoms of MGMT(83-176)
1H- and 13C NMR spectra were recorded using Bruker NMR
spectrometers: AC-250 (5.87 T, 250 MHz 1H frequency) and
1
AM-500 (11.74 T, 500 MHz H frequency), with 5-mm sample
tubes and standard Fourier transform NMR techniques. In
general, the protected glucosides were dissolved in CDCl3 and
measured at 30 °C at 5.87 T (data reported in text below),
while the deprotected glucosides (final products) were mea-
sured at 37 °C or 40 °C in DMSO-d6 at 11.74 T (detailed data
are reported in Tables D-G, Supporting Information). For all
compounds one-dimensional 1H and 13C (1H-decoupled and
DEPT) spectra were obtained. In general, a complete first-
order 1H spin system analysis was performed with Bruker’s
WIN NMR software using zero filling and resolution enhance-
ment (Lorentz-Gauss line shape transformation). For com-
pounds 10, 14, 19, and 20 additional experiments were
performed at 11.74 T: 2D HH-COSY, 2D CH correlation via
were tethered to their original coordinates with
a force
constant of 100 kcal/Å, while side chains were unrestrained.
The simulation protocol was defined as follows: integration
step size ) 1 fs, dielectric ) 1.0, cutoff for nonbonded
interactions ) 10 Å, T ) 300 K, periodic boundary conditions.
In the initial phase an energy minimization of the complex
was performed (1000 iterations). This was followed by a
temperature equilibration period of 50 ps (velocity rescaling
for ∆T > 10 K) and a production period of 1 µs with constant
bath temperature maintained by the standard DISCOVER
algorithm. During the production period, snapshots of the
molecular ensemble were stored every 1000 steps (1 ps). This
resulted in an archive of 1000 conformations which was used
to produce trajectories for various structural parameters and
to carry out the interaction analyses described in the Results.
The ENCLOSE option of DISCOVER was used to calculate
the nonbonded interactions (van der Waals and Coulombic)
for specified structural elements of the ligand (Figure 2) with
individual residues of MGMT. For each ligand fragment the
calculation was performed for all residues for which at least
one atom was within a radius of 5 Å measured from any
atom of the ligand fragment. Water molecules were not
included in the interaction calculation. All reported results
represent average values for n ) 1000 structures. These
detailed analyses were carried out using software developed
in the Central Spectroscopy Department by Martin Frank
n
1J CH or via J CH (COLOC, compound 19), 1D 13C with gated
and selective 1H decoupling to observe and assign J CH for
nonprotonated carbons. Detailed measurements were also
made for 10 and 20 in CDCl3/CD3OD at 30 °C and for the butyl
spacer analogue (not used in this study) of the 8-aza compound
28 in DMSO. Thus, for these compounds complete and
unequivocal signal assignments could be made a priori. These
results, other literature data, and chemical shift predictions
using SpecTool software (version 2.1, Chemical Concepts,
Weinheim, FRG) and WIN-SPECEDIT (version 3.0, Bruker)
served as a basis for all other assignments reported here. The
few remaining ambiguities are duly noted in the results. 1H
chemical shifts are reported in ppm relative to the following
internal references: tetramethylsilane (TMS) in CDCl3 ) 0.0,
residual DMSO-d5 ) 2.49 ppm, residual CHD2OD ) 3.30; 13
C
chemical shifts are relative to CDCl3 ) 77.0, DMSO ) 39.5,
or CD3OD ) 49.0 ppm. Spin-spin J couplings are reported in

4058 J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 24
Reinhard et al.
Hz. For signal assignments the numbering scheme shown in
Chart 1 is used, and the alkyl spacer is numbered beginning
at the glucoside -OCH2.
3.3 g (5 mmol) of 1-bromo-2,3,4,6-tetra-O-benzoyl-â-d-glucopy-
ranoside were reacted. Flash chromatography on SiO2 (4 ×
20 cm) with petroleum ether/ethyl acetate (4:1) gave 4 as a
colorless oil (1.92 g, 48%). C42H43BrO10 (787.70); TLC: Rf 0.26
(SiO2, petroleum ether/ethyl acetate, 4:1); ESI-MS: m/z (% int.)
579.1 (100) [M + H-Br(CH2)8OH]+, 787.1/789.2 (5) [M(79Br/
81Br) + H]+ (calc. 787.21/789.21), 809.1/811.1 (30) [M(79Br/81Br)
+ Na]+.
Syn t h eses. Syn t h esis of Act iva t ed Alk yl Glu cosid es
(1-7). The Ko¨nigs-Knorr method (Chart 2, step A) was used
to prepare the required ω-bromoalkyl-â-d-glucopyranosides, as
described below.46 Detailed NMR data with assignments are
presented below only for the representative compounds 1, 4,
and 7 (ethyl, octyl, butynyl spacer); the corresponding data
for compounds 2, 3, 5, and 6 are provided as Supporting
Information. The specific 13C NMR assignments given are
generally unambiguous with the exception of Glc2, Glc5
(assigned by analogy with 19) and the inner carbons of long
alkyl chains (Alk4 to Alk(ω-3), which were assigned by analogy
with 19 and by considering shift increment rules for alkanes.
2-Br om oeth yltetr a -O-a cetyl-â-d -glu cop yr a n osid e (1).
2-Bromoethanol (12.5 g, 100 mmol) and 9.85 g (20 mmol) of
1,2,3,4,6-penta-O-acetyl-â-d-glucopyranoside were dissolved in
50 mL of dry methylene chloride and stirred with 4-Å molec-
ular sieves for 1 h under argon. After addition of 5.4 mL (30
mmol) of trimethylsilyltriflate stirring was continued until no
glucose educt could be detected by TLC. The reaction was
stopped by addition of 3.3 g (30 mmol) of triethylamine, and
the pH was adjusted to 6-7 with solid sodium carbonate. The
mixture was filtered, and the filtrate was shaken with four
80-mL portions of saturated sodium carbonate solution. The
organic phase was dried over sodium sulfate. After evaporation
of the solvents the product was separated by flash chroma-
tography on SiO2 (7.5 × 25 cm) with ethyl acetate/petroleum
ether (1:2) to yield pure 1 (4.1 g, 36.1%). Recrystallization was
achieved by adding water to a methanol solution of the
product; cooling to 4 °C gave analytically pure white needles.
1H NMR δH (250 MHz, CDCl3): 1.1-1.30 (8H, Alk3,4,5,6);
1.48 (m, Alk2b); 1.56 (m, Alk2a); 1.719 (tt, Alk7); 3.313 (t,
Alk8); 3.544 (dt, Alk1b); 3.924 (dt, Alk1a); 4.186 (ddd, Glc5);
4.529 (dd, Glc6b); 4.661 (dd, Glc6a); 4.865 (d, Glc1); 5.551 (dd,
Glc2); 5.703 (dd, Glc4); 5.940 (dd, Glc3); 7.24-7.38 (m, 4 Bz3,5);
1
7.4-7.52 (m, 4 Bz4); 7.83-8.02 (m, 4 Bz2,6). H J couplings:
7.83 (Glc1,2); 9.72 (Glc2,3); 9.62 (Glc3,4); 9.76 (Glc4,5); 3.35
(Glc5,6a); 5.05 (Glc5,6b); -12.09 (Glc6a,6b); -9.69 (Alk1a,1b);
6.14 (Alk1a,2); 6.6 (Alk1b,2); 6.87 (Alk7,8).
13C NMR δC (62.9 MHz, CDCl3): 25.54 (Alk3); 27.83 (Alk6);
28.38 (Alk4); 28.82 (Alk5); 29.20 (Alk2); 32.60 (Alk7); 33.76
(Alk8); 63.13 (Glc6); 69.79 (Glc4); 70.08 (Alk1); 71.86 (Glc2);
72.05 (Glc5); 72.87 (Glc3); 101.18 (Glc1); 128.14, 128.21, 128.21,
128.25 (4 Bz3,5); 128.72, 128.73, 129.28, 129.52 (4 Bz1); 129.6
129.6 129.6 129.67 (4 Bt2,6); 132.96, 133.05, 133.06, 133.26
(4 Bz4); 164.90, 165.07; 165.68; 165.97 (4 CdO).
10-Br om odecyltetr a-O-ben zoyl-â-d-glu copyr an oside (5).
As described for 3, 1 g (4.3 mmol) of 1-bromo-10-decanol and
3.3 g (5 mmol) of 1-bromo-2,3,4,6-tetra-O-benzoyl-â-d-glucopy-
ranoside were reacted. Flash chromatography on SiO2 (4 ×
20 cm) with petroleum ether/ethyl acetate yielded pure 5 as
an colorless oil (1.62 g, 46%). C44H47BrO10 (815.76); TLC: Rf
0.34 (SiO2, petroleum ether/ethyl acetate 6:1, developed twice);
ESI-MS: m/z (% int.) 579.1 (100) [M + H - Br(CH2)10OH]+,
815.2/817.2 (5) [M(79Br/81Br) + H]+ (calc. 815.24/817.24), 837.2/
839.2 (92/100) [M(79Br/81Br) + Na]+.
C
16H23BrO10 (Mr ) 455.05); TLC: Rf 0.49 (SiO2, petroleum
ether/ethyl acetate 1:1); ESI-MS: m/z (% int.) 477.1/479.1 (98/
100) [M(79Br/81Br) + Na]+ (calc. 477.04/479.03).
1H NMR δH (250 MHz, CD3OD): 1.960; 1.999; 2.027; 2.049
(4 s, acetyl CH3); 3.474 (ddd, Alk2b), 3.542 (ddd, Alk2a); 3.855
(ddd, Alk1b); 3.872 (ddd, Glc5); 4.102 (ddd, Alk1a); 4.134 (dd,
Glc6b); 4.271 (dd, Glc6a); 4.731 (d, Glc1); 4.898 (dd, Glc2); 5.014
(dd, Glc4); 5.254 (dd, Glc3). 1H J couplings (Hz): 7.95 (Glc1,2);
9.50 (Glc2,3); 9.43 (Glc3,4); 10.0 (Glc4,5); 4.67 (Glc5,6a), 2.48
(Glc5,6b); -12.35 (Glc6a,6b); -11.54 (Alk1a,1b); 5.40 (Alk1a,-
2a); 5.55 (Alk1a,2b); 5.43 (Alk1b,2a); 7.30 (Alk1b,2b); -10.70
(Alk2a,2b).
13C NMR δC (62.9 MHz, CD3OD): 20.53; 20.55; 20.62; 20.74
(4, acetyl CH3); 31.17 (Alk2); 63.07 (Glc6); 69.83 (Alk1); 71.01
(Glc4); 72.68 (Glc2); 72.91 (Glc5); 74.17 (Glc3); 101.83 (Glc1);
171.15; 171.22; 171.58; 172.27 (4 CdO).
12-Br om od od e cylt e t r a -O-b e n zoyl-â-d -glu cop yr a n o-
sid e (6). As described for 3, 1 g (3.7 mmol) of 1-bromo-12-
dodecanol and 2.5 g (3.7 mmol) of 1-bromo-2,3,4,6-tetra-O-
benzoyl-â-d-glucopyranoside were reacted. Flash chroma-
tography on SiO2 (4 × 20 cm) with petroleum ether/ethyl
acetate (4:1) gave pure 6 as an colorless oil (1.95 g, 62%).
C
46H51BrO10 (843.81); TLC: Rf 0.34 (SiO2, petroleum ether/
ethyl acetate 6:1, developed twice); ESI-MS: m/z (% int.) 579.0
(95) [M + H - Br(CH2)12OH]+, 843.1/845.1 (3) [M(79Br/81Br) +
H]+ (calc. 843.27/845.27), 865.2/867.1 (84/100) [M(79Br/81Br) +
Na]+.
4-Br om ob u t -2-yn ylt e t r a -O-a ce t yl-â-d -glu cop yr a n o-
sid e (7). In 5 mL of dry 1,2-dichloroethane were dissolved and
stirred with 4-Å molecular sieves for 1 h under argon 745 mg
(5 mmol) of 4-bromobut-2-yn-1-ol and 780 mg (2 mmol) of
1,2,3,4,6-penta-O-acetyl-â-d-glucopyranoside. After addition of
352 µL (2 mmol) of trimethylsilyltriflate, stirring was contin-
ued until no glucose educt could be detected by TLC. The
reaction was stopped by addition of 218 mg (2 mmol) of
triethylamine, and the pH was adjusted to 6-7 with solid
sodium carbonate. The mixture was filtered, and 25 mL of
chloroform was added to the filtrate, which was then shaken
with four 25-mL portions of saturated sodium carbonate
solution. The organic phase was dried with sodium sulfate.
After evaporation of the solvents flash chromatography on SiO2
(5.5 × 20 cm) with petroleum ether/ethyl acetate (1:1) gave
the product 7 (160 mg, 16.7%). An analytically pure sample
was obtained by recrystallization from methanol at -30 °C.
4-Br om obu tyltetr a -O-a cetyl-â-d -glu cop yr a n osid e (2).
For this particular case only, the method of Helferich and
Zirner47 was used as published to produce 2 (1.2 g, 10%).
C
18H27BrO10 (483.31); TLC: Rf 0.39 (SiO2, petroleum ether/
ethyl acetate 1:1); ESI-MS: m/z (% int.) 505.0/507.0 (42)
[M(79Br/81Br) + Na]+ (calc. 505.07/507.07), 987.3/989.2 (45/100)
[2M(79Br/81Br) + Na]+.
6-Br om oh exyltetr a-O-ben zoyl-â-d-glu copyr an oside (3).
After 25 mL of nitromethane was dried for 20 h with 3-Å
molecular sieves under argon, 5.5 g (30 mmol) of 1-bromo-6-
hexanol and 3.3 g (5 mmol) of 1-bromo-2,3,4,6-tetra-O-benzoyl-
â-d-glucopyranoside were added, and the reaction mixture was
stirred for 60 min. After addition of 0.7 g of silver phosphate,
the reaction was monitored by TLC until no glucose educt
could be detected. After addition of 75 mL of ethyl acetate,
the mixture was filtered, and the solvent was evaporated.
Purification by flash chromatography on SiO2 (5.5 × 20 cm)
with petroleum ether/ethyl acetate (4:1) gave 3 as a colorless
oil (2.2 g, 58%). C40H39BrO10 (759.65); TLC: Rf 0.25 (SiO2,
petroleum ether/ethyl acetate, 4:1, developed twice); Rf 0.53
(SiO2, petroleum ether/ethyl acetate 2:1); ESI-MS: m/z (% int.)
579.1 (100) [M + H - Br(CH2)6OH]+, 759.2/761.2 (5) [M(79Br/
81Br) + H]+ (calc. 759.18/761.18), 781.1/783.1 (20) [M(79Br/81Br)
+ Na]+.
C
18H23BrO10 (479.28); TLC: Rf 0.37 (SiO2, petroleum ether/
ethyl acetate 1:1); Rf 0.11 (SiO2, petroleum ether/ethyl acetate
2:1); ESI-MS: m/z (% int.) 496.3/498.1 (30) [M(79Br/81Br) +
NH4]+, 501.1/503.2 (92/100) [M(79Br/81Br) + Na]+ (calc. 501.04/
503.03).
1H NMR δH (250 MHz, CDCl3): 2.01, 2.03, 2.07, 2.09 (4 s,
acetyl CH3); 3.73 (ddd, Glc5); 3.94 (t, 2H, Alk4), 4.16 (dd,
Glc6b); 4.27 (dd, G6a); 4.42 (t, 2H, Alk1); 4.73 (d, Glc1); 5.00
(dd, Glc2); 5.10 (dd, Glc4); 5.24 (dd, Glc3). 1H J couplings: 7.8
(Glc1,2); 9.3 (Glc2,3); 9.3 (Glc3,4); 9.4 (Glc4,5); 4.5 (Glc5,6a);
2.3 (Glc5,6b); -12.3 (Glc6a,6b); 1.9 (Alk1,4).
8-Br om ooctyltetr a -O-ben zoyl-â-d -glu cop yr a n osid e (4).
As described for 3, 1.22 g (5.8 mmol) of 1-bromo-8-octanol and

O6-Methylguanine-DNA Methyltransferase
J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 24 4059
13C NMR (62.9 MHz): 13.70 (Alk4); 20.54, 20.56, 20.67,
20.67 (4 acetyl CH3); 56.19 (Alk1); 61.75 (Glc6); 68.30 (Glc4);
71.01 (Glc2); 71.96 (Glc5); 72.78 (Glc3); 81.19, 82.34 (Alk2,3
assignments not definitive); 98.31 (Glc1); 169.34, 169.37,
170.21, 170.59 (4 CdO).
Syn th esis of Gu a n in e Der iva tives (8-10). The required
O6-R guanine derivatives were prepared as described in the
Results and outlined below.
O6-Ben zylgu a n in e (8). In 20 mL of benzyl alcohol was
carefully dissolved 0.6 g (26 mmol) of sodium hydride, and the
solution was heated to 130 °C. In small portions 2 g (11.8
mmol) of 2-amino-6-chloropurine was added, and the reaction
mixture was stirred for 18 h, then cooled to room temperature,
and neutralized with acetic acid. The benzyl alcohol was
removed by distillation. Purification by column chromatogra-
phy (SiO2, dichloromethane/methanol, 20:1), yielded the prod-
uct 8 (0.91 g, 31%). C12H11N5O (241.25); TLC: Rf 0.31 (SiO2,
dichloromethane/methanol 10:1); ESI-MS: m/z (% int.) 241.8
(100) [M + H]+ (calc. 242.10), 263.8 (5) [M + Na]+, 483.2 (15)
[2M + H]+.
1H NMR δH (500 MHz, DMSO-d6): 5.627 (d, BT6); 6.247 (s,
br, 2-NH2); 7.308 (dt, BT3); 7.650 (d, BT5); 7.820 (s, H8); 12.4
1
(br, N9-H). H J couplings: 1.52 (BT3,BT5); 0.57 (BT3,BT6).
13C NMR δC (125 MHz, DMSO-d6): 60.47 (T, BT6); 107.88
(S, BT4); 112.75 (S,vbr, G5); 124.90 (D, BT5); 130.53 (D, BT3);
138.35 (D,br G8); 140.62 (S, BT2), 155.99 (s,vbr, G4); 158.73
(S,br, G6); 159.27 (S, G2). All guanine signals except G2 were
broad (br) or very broad (vbr) due to tautomerization between
N7H and N9H forms.
Syn th esis of P r otected Nu cleosid e An a logu es (11, 13,
15, 17, 19, 21, 23, 25, a n d 27). The bromoalkyl glucosides were
coupled with guanine derivatives via the following standard
procedure (Chart 2, step B). One equivalent of the purine and
1 equiv of the activated alkylglucoside were dried for 12 h at
1 × 10-3 mmHg. The guanine derivatives were dissolved in
very dry DMF (10 mL/mmol) containing 4-Å molecular sieves.
Subsequently 1.1 equiv of lithium hydride was added, and the
mixture was stirred for 1 h while increasing the temperature
to 80 °C. The glycoside was dissolved in dry DMF (5 mL/mmol)
and added dropwise to the solution of the guanine. Stirring
was continued until one of the educts could no longer be
detected by TLC. The reaction mixture was cooled to room
temperature, 0.1 M phosphate buffer (KH2PO4/Na2PO4, pH 7)
was added (10 mL/mmol), and the mixture was stirred for 15
min and then filtered. The filtrate was extracted with three
100-mL portions of chloroform. The organic phase was dried
over sodium sulfate and evaporated until no DMF was left in
the oily mixture. Purification of the product was achieved by
column or flash chromatography (see below). The HPLC
methods used are summarized in the footnote to Table B
(Supporting Information). With the exception of 23, where the
reaction mixture was used directly for step C, all compounds
1H NMR δH (250 MHz, CD3OD): 5.52 (s, 2H, Bn7); 7.30 (m,
1H, Bn4); 7.34 (m, 2H, Bn3,5); 7.48 (m, 2H, Bn2,6); 7.822 (s,
1H, G8). Aromatic protons were assigned by analogy with 12.
13C NMR δC (62.9 MHz, CD3OD): 68.97 (T, Bn7); 113.37
(br S, G5); 129.11 (S, Bn4); 129.37 (D, Bn3,5); 129.45 (D,
Bn2,6); 137.93 (S, Bn1); 140.04 (D, G8); 157.09 (br S, G4);
161.43 (S, G6); 161.70 (S, G2). Tautomerization between N9-H
and N7-H leads to significant linebroadening (ca. 20 Hz) at
G4,G5. Bn2,6 and Bn3,5 assigned by analogy with 12.
8-Aza -O6-ben zylgu a n in e (9). In a mixture of 200 mL of
acetone and 20 mL of acetic acid was suspended and stirred
for 2 h 4.9 g (20 mmol) of 2,4,5-triamino-6-benzyloxypyrimidine
and 1.38 g (20 mmol) sodium nitrite. The acetone was reduced
in volume, and the mixture was shaken with 100 mL of water
and 5 × 100 mL of chloroform to extract the product. Purifica-
tion was performed via flash chromatography (SiO2; chloroform/
methanol 15:1), and 9 was crystallized from ethanol/water 1:1
(3.05 g, 63%). C11H10N6O (242.24); TLC: Rf 0.40 (SiO2, dichlo-
romethane/methanol 10:1); ESI-MS: m/z (% int.) 243.0 (50)
[M+H]+ (calc. 243.09), 264.9 (20) [M + Na]+, 485.2 (80) [2M +
H]+, 507.1 (50) [2M + Na]+.
1
were characterized by H and 13C NMR at 5.87 T (11.74 T for
19). For the representative compounds 13, 19, 25, and 27, only
the key NMR data required for characterization are presented
below; the complete data for all compounds are provided as
Supporting Information.
2-(O6-Ben zylgu a n -9-yl)et h yl-â-d -t et r a -O-a cet ylglu co-
sid e (11). The coupling reaction was carried out with 241 mg
(1 mmol) of O6-benzylguanine and 455 mg (1 mmol) of
2-bromoethyl-tetra-O-acetyl-â-d-glucopyranoside. Column chro-
matography (3 × 15 cm) with ethyl acetate/petroleum ether
(2:1) gave pure 11 (73.6 mg, 17.7%). C28H33N5O11 (615.60);
TLC: Rf 0.5 (SiO2, methylene chloride/methanol 20:1); HPLC
(method II): Rt ) 13.8 min; ESI-MS: m/z (% int.) 616.2 (100)
[M + H]+ (calc. 616.22), 638.2 (55) [M + Na]+.
1H NMR δH (250 MHz, CD3OD): 5.6 (s, Bn7); 7.32 (m, Bn4);
7.37 (m, Bn3,5); 7.52 (m, 2H, Bn2,6).
13C NMR δC (62.9 MHz, CD3OD): 69.74 (Bn7); 121.30 (G5);
129.38 (Bn4); 129.56 (Bn2,6; Bn3,5); 137.26 (Bn1); 156.13 (G4);
162.62 (G6); 164.06 (G2).
2-[O6-(4-Br om oth en yl)gu an -9-yl]eth yl-â-d-tetr a-O-acetyl-
glu cosid e (13). The coupling reaction was carried out with
163 mg (0.5 mmol) of O6-(4-bromothenyl)guanine and 225 mg
(0.5 mmol) of 1-(2-bromoethyl)-2,3,4,6-tetra-O-acetyl-â-d-glu-
copyranoside. Column chromatography (1 × 20 cm) with
acetone/petroleum ether (1:1) gave pure 13 (30 mg, 8.5%).
4-Br om oth en yl Alcoh ol. In 400 mL of 2-propanol were
stirred for 1 h 25 g (130 mmol) of 4-bromothiophen-2-aldehyde
and 5.7 g (150 mmol) of sodium borohydride. The reaction was
stopped by addition of 10 mL of water. The crude product was
obtained by filtration and extracted with n-hexane (22.5 g,
89%). C5H5BrOS (193.06); TLC: Rf 0.38 (SiO2, toluol/methanol
4:1); ESI-MS: m/z (% int.) 190.7/192.7 (40) [M(79Br/81Br) - H]-
(calc. 190.92/192.92).
C
26H30BrN5O11S (700.52); TLC: Rf 0.17 (SiO2, acetone/petro-
leum ether 1:1); HPLC (method II): Rt ) 13.9 min; ESI-MS:
m/z (% int.) 700.1/702.1 (21/23) [M(79Br/81Br) + H]+ (calc.
700.09/702.09), 722.1/723.9 (21/23) [M(79Br/81Br) + Na]+, 1399.4/
1401.3 (40/100) [2M(79Br/81Br) + H]+.
1H NMR δH (250 MHz, CDCl3): 2.539 (t, BT6-OH); 4.731
(dd, BT6); 6.887 (dt, BT3); 7.155 (d, BT5); 3J 6,OH ) 5.8; 4J 3,5
)
4
1.44; J 3,6 ) 0.8.
1H NMR δH (250 MHz, CDCl3), selected data: 3.775 (ddd,
Alk1b); 4.17 (m, Alk2b); 4.19 (ddd, Alk1a); 4.319 (ddd, Alk2a);
4.431 (d, Glc1); 4.863 (br s, 2H, G2-NH2); 5.634 (d, BT6b); 5.663
13C NMR δC (62.9 MHz, CDCl3): 59.38 (T, BT6); 109.27 (S,
BT4); 122.31 (D, BT5); 127.43 (D, BT3); 145.88 (S, BT2).
O6-(4-Br om oth en yl)gu a n in e (10). In 12 mL of DMSO was
dissolved 13.89 g (72 mmol) of 4-bromothenyl alcohol, and the
mixture was carefully treated with 1.92 g [48 mmol] of sodium
hydride with stirring for 1 h. Then 5.49 g (24 mmol) of
2-aminopurin-6-yltrimethylammonium chloride was added in
small portions, and the mixture was stirred for 1 h. The
mixture was neutralized with 3.9 mL of acetic acid, followed
by addition of 240 mL of diethyl ether, and stirred for 1.5 h.
The solid material was dried and washed with water, and 10
was recrystallized from methanol at -15 °C (5.5 g, 70%). C10H8-
BrN5OS (326.17); TLC: Rf 0.4 (SiO2, dichloromethane/metha-
nol 10:1); Rf 0.22 (SiO2, toluol/methanol 4:1); mp 212-214 °C;
ESI-MS: m/z (% int.) 325.8/327.8 (98/100) [M(79Br/81Br) + H]+
(calc. 325.971/327.969), 347.8/349.8 (19/20) [M(79Br/81Br) +
Na]+; λmax (methanol) ) 239, 284 nm.
1
(d, BT6a); 7.110 (dt, BT3); 7.182 (d, BT5); 7.582 (s, G8). H J
couplings: 7.84 (Glc1,2); -10.5 (Alk1a,1b); -14.7 (Alk2a,2b);
1.47 (BT3,5); 0.68 (BT3,6); -12.7 (BT6a,b).
13 C NMR δC (62.9 MHz, CDCl3), selected data: 43.06 (Alk2);
61.74 (BT6); 67.45 (Alk1); 100.53 (Glc1); 109.13 (BT4); 115.32
(G5); 124.02 (BT5); 130.85 (BT3); 139.88 (BT2); 140.62 (G8);
154.19 (G4); 158.81 (G2); 160.20 (G6).
4-[O6-(4-Br om ot h e n yl)gu a n -9-yl]b u t yl-â-d -t e t r a -O-
a cetylglu cosid e (15). The reaction was carried out with 163
mg (0.5 mmol) of O6-(4-bromothenyl)guanine and 241 mg (0.5
mmol) of 4-bromobutyltetra-O-acetyl-â-d-glucopyranoside. Col-
umn cromatography (3 × 25 cm) with acetone/petroleum ether
(1:1) gave pure 15 (86 mg, 24.8%). C28H34BrN5O11S (728.58);
TLC: Rf 0.17 (SiO2, acetone/petroleum ether 1:1); HPLC

4060 J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 24
Reinhard et al.
(method II): Rt ) 14.8 min; ESI-MS: m/z (% int.) 727.9/729.8
(90/100) [M(79Br/81Br) + H]+ (calc. 728.12/730.12), 749.9/751.9
(90/85) [M(79Br/81Br) +Na]+.
ethyl acetate/petroleum ether (2:1, 3:1, 4:1) gave pure 27 (61
mg, 19.8%). C27H32N6O11 (616.58); TLC: Rf 0.32 (SiO2, acetone/
petroleum ether 1:1); HPLC (method II): Rt ) 13.5 min; ESI-
MS: m/z (% int.) 617.2 (40) [M + H]+ (calc. 617.22), 639.1 (20)
[M + Na]+, 1233.7 (100) [2M + H]+, 1255.6 (20) [2M + Na]+.
1H NMR δH (250 MHz, CDCl3, selected data): 4.091 (ddd,
Alk1b); 4.288 (ddd, Alk1a); 4.471 (d, Glc1); 4.565 (ddd, Alk2b);
4.666 (ddd, Alk2a); 5.478 (br s, 2H, 2-NH2); 5.601 (d, Bn7b);
5.623 (d, Bn7a); 1H J couplings: 7.88 (Glc1,2); -10.5 (Alk1a,-
1b); 5.1 (Alk1a,2a); 4.8 (Alk1a,2b); 7.85 (Alk1b,2a); 4.77
(Alk1b,2b); -14.5 (Alk2a,2b); -12.2 (Bn7a,7b).
6-[O6-(4-Br om oth en yl)gu a n -9-yl]h exyl-â-d -tetr a -O-ben -
zoylglu cosid e (17). The reaction was carried out with 163
mg (0.5 mmol) of O6-(4-bromothenyl)guanine and 378 mg (0.5
mmol) of 6-bromohexyltetra-O-benzoyl-â-d-glucopyranoside.
Flash chromatography (3.5 × 30 cm) with petroleum ether/
acetone (1:1) gave pure 17 (150 mg, 29.8%). C50H46BrN5O11
S
(1004.91); TLC: Rf 0.31 (SiO2, acetone/petroleum ether 1:1);
Rf 0.23 (SiO2, ethyl acetate/petroleum ether 2:1); HPLC
(method II): Rt ) 20.8 min; ESI-MS: m/z (% int.) 1004.0/1006.3
(100/75) [M(79Br/81Br) + H]+ (calc. 1004.22/1006.22), 1025.9/
1027.9 (17/13) [M(79Br/81Br) + Na]+.
13C NMR δC (62.9 MHz, CDCl3, selected data): 45.68 (Alk2);
67.06 (Alk1); 68.79 (Bn7); 70.77 (Glc2); 100.48 (Glc1); 121.69
(G5); 128.30 (Bn4); 128.36 (Bn3,5); 128.45 (Bn2,6); 135.48
(Bn1); 153.56 (G4); 161.63 (G6); 161.76 (G2).
8-[O6-(4-Br om oth en yl)gu a n -9-yl]octyl-â-d -tetr a -O-ben -
zoylglu cosid e (19). The reaction was carried out with 0.82 g
(2.5 mmol) of O6-(4-bromothenyl)guanine and 1.82 g (2.5 mmol)
of 8-bromooctyltetra-O-benzoyl-â-d-glucopyranoside. Flash chro-
matography (4.5 × 25 cm) with petroleum ether/acetone (3:1)
gave pure 19 (1.00 g, 37.7%). C52H50O11BrN5S (1032.97);
TLC: Rf 0.50 (SiO2, acetone/petroleum ether 1:1); ESI-MS: m/z
(% int.) 1032.5/1034.5 (85/100) [M(79Br/81Br)+H]+ (calc. 1032.25/
1034.25), 1054.5/1056.4 (15/18) [M(79Br/81Br) + Na]+.
1H NMR (500 MHz, CDCl3, selected data): 3.523 (dt, Alk1b);
3.904 (dt, Alk1a); 3.971 (t, 2H, Alk8); 4.831 (d, Glc1); 4.964
(br, 2H, G2-NH2); 5.643 (d, 2H, BT6); 7.111 (dt, BT3); 7.178
(d, BT5); 7.582 (s, G8). 1H J couplings: 7.84 (Glc1,2); -9.7
(Alk1a,b); 7.3 (Alk7,8).
Dep r otection of th e Nu cleosid e An a logu es (12, 14, 16,
18, 20, 22, 24, 26, a n d 28). For removal of acetyl or benzoyl
groups (Chart 2, step C), 1 equiv of the nucleoside analogue
was dissolved in methanol (2 mL/mmol), and 0.1 equiv of a
freshly prepared 0.1 M sodium methanolate solution was
added. The reaction was monitored by TLC. The mixture was
neutralized with Dowex (H+ 50WX2) ion exchanger and
filtered. The solvent was evaporated from the filtrate, and the
final product was purified by flash chromatography and, if
possible, by recrystallization. Note: compound 24 was syn-
thesized without purification of the protected glucoside 23. The
reaction mixture from step B was directly dissolved and used
for step C since good separation of N9 and N7 derivatives was
possible in the deprotected form (see Results). Detailed
analytical results and complete NMR data are presented as
Supporting Information.
13C NMR δC (125 MHz, CDCl3, selected data): 25.64 (Alk3);
26.39 (Alk6); 28.76 (Alk5); 28.92 (Alk4); 29.30 (Alk2); 29.74
(Alk7); 43.60 (Alk8); 61.79 (BT6); 70.23 (Alk1); 101.34 (Glc1);
109.15 (BT4); 115.36 (G5); 124.06 (BT5); 130.84 (BT3); 139.72
(G8); 139.85 (BT2); 154.16 (G4); 158.78 (G2); 160.22 (G6).
Selected 13C-1H couplings for carbon: G8 (209.2 d, 4.0 t); BT3
(174.2 d); BT5 (190.9 d, 8.1 d).
Su p p or tin g In for m a tion Ava ila ble: Tables A1-A7,
detailed interaction energies between each amino acid residue
in the MGMT binding region and each structural segment of
the inhibitors shown in Tables 3 and 4; Table B, details of the
HPLC purifications, yields, and UV absorption data for 4-BTG
10-[O6-(4-Br om oth en yl)gu an -9-yl]decyl-â-d-tetr a-O-ben -
zoylglu cosid e (21). The reaction was carried out with 163
mg (0.5 mmol) of O6-(4-bromothenyl)guanine and 378 mg (0.5
mmol) of 10-bromodecyltetra-O-benzoyl-â-d-glucopyranoside.
Flash chromtatography (4.5 × 25 cm) with petroleum ether/
1
(10) and the inhibitors 12-28; additional detailed H and 13C
NMR data and assignments for the precursor glycosides 2, 3,
5, and 6 and for the protected nucleosides 11, 15, 17, 21, and
acetone (3:2) gave pure 21 (142 mg, 35.8%). C54H54BrN5O11
S
1
23 in Table C; complete H and 13C NMR chemical shift data
(1061.02); TLC: Rf 0.37 (SiO2, acetone/petroleum ether 1:1);
ESI-MS: m/z (% int.) 1060.2/1062.2 (80/100) [M(79Br/81Br) +
H]+ (calc. 1060.23/1062.23).
12-[O6-(4-Br om oth en yl)gu a n -9-yl]d od ecyl-â-d -tetr a -O-
ben zoylglu cosid e (23). The reaction was carried out with
163 mg (0.5 mmol) of O6-(4-bromothenyl)guanine and 421 mg
(0.5 mmol) of 12-bromododecyltetra-O-benzoyl-â-d-glucopyran-
oside. The raw product was used without purification for step
C.
with assignments for the inhibitors 12, 14, 16, 18, 20, 22, 24,
26, 28 in Tables D and E, respectively; and diagnostic J CH and
J HH of useful structural information in Tables F and G,
respectively.
Refer en ces
(1) D’Incalci, M.; Citti, L.; Taverna, P.; Catapano, C. V. Importance
of the DNA repair enzyme O6-alkyl guanine alkyltransferase
(AT) in cancer chemotherapy. Cancer Treat. Rev. 1988, 15, 279-
292.
4-[O6-(4-Br om ot h en yl)gu a n -9-yl]b u t -2-yn yl-â-d -t et r a -
O-a cetylglu cosid e (25). The reaction was carried out with
163 mg (0.5 mmol) of O6-(4-bromothenyl)guanine and 240 mg
(0.5 mmol) of 4-bromobut-2-ynyltetra-O-acetyl-â-d-glucopyran-
oside. Flash chromatography (3 × 20 cm) with ethyl acetate/
(2) Pegg, A. E.; Byers, T. L. Repair of DNA containing O6-alkylgua-
nine. FASEB J . 1992, 6, 2302-2310.
(3) Mitra, S.; Kaina, B. Regulation of repair of alkylation damage
in mammalian genomes. Prog. Nucleic Acid Res. Mol. Biol. 1993,
44, 109-142.
petroleum ether (2:1) gave pure 25 (135 mg, 37.3%). C29H30
-
BrN5O11S (724.54); TLC: Rf 0.53 (SiO2, acetone/petroleum
ether 1:1); HPLC (method II): Rt ) 14.8 min; ESI-MS: m/z
(% int.) 725.5/727.5 (88/100) [M(79Br/81Br) + H]+ (calc. 724.09/
726.09).
(4) Pegg, A. E.; Dolan, M. E.; Moschel, R. C. Structure, function,
and inhibition of O6-alkylguanine-DNA alkyltransferase. Prog.
Nucleic Acid Res. Mol. Biol. 1995, 51, 167-223.
(5) Morimoto, K.; Dolan, M. E.; Scicchitano, D.; Pegg, A. E. Repair
of O6-propylguanine and O6-butylguanine in DNA by O6-alkyl-
guanine-DNA alkyltransferases from rat liver and E. coli.
Carcinogenesis 1985, 6, 1027-1031.
(6) Dolan, M. E.; Pegg, A. E.; Dumenco, L. L.; Moschel, R. C.; Gerson,
S. L. Comparison of the inactivation of mammalian and bacterial
O6-alkylguanine-DNA alkyltransferases by O6-benzylguanine
and O6-methylguanine. Carcinogenesis 1991, 12, 2305-2309.
(7) Dolan, M. E.; Mitchell, R. B.; Mummert, C.; Moschel, R. C.; Pegg,
A. E. Effect of O6-benzylguanine analogues on sensitivity of
human tumor cells to the cytotoxic effects of alkylating agents.
Cancer Res. 1991, 51, 3367-3372.
1H NMR δH (250 MHz, CDCl3, selected data): 4.395 (t, 2H,
Alk1); 4.723 (d, Glc1); 4.876 (t, 2H, Alk4); 5.019 (br s, 2-NH2);
5.653 (d, 2H, BT6); 7.115 (dt, BT3), 7.190 (d, BT5), 7.756 (s,
1
G8); H J couplings: 7.88 (Glc1,2); 1.9 (Alk1,4).
13C NMR δC (62.9 MHz, CDCl3, selected data): 32.93 (Alk4);
55.94 (Alk1); 61.79 (BT6); 79.91 (Alk2); 80.67 (Alk3); 98.30
(Glc1); 109.12 (BT4); 115.29 (G5); 124.04 (BT5); 130.85 (BT3);
138.60 (G8); 139.67 (BT2); 153.83 (G4); 159.09 (G2); 160.28
(G6).
2-(8-Aza -O6-ben zyl-gu a n -9-yl)eth yl-â-d -tetr a -O-a cetyl-
glu cosid e (27). The reaction was carried out with 121 mg (0.5
mmol) of 8-aza-O6-benzylguanine and 225 mg (0.5 mmol) of
2-bromoethyltetra-O-acetyl-â-d-glucopyranoside. Column chro-
matography (3 × 20 cm) with a stepwise applied gradient of
(8) McElhinney, R. S.; Donnelly, D. J .; McCormick, J . E.; Kelly, J .;
Watson, A. J .; Rafferty, J . A.; Elder, R. H.; Middleton, M. R.;
Willington, M. A.; McMurry, T. B. H.; Margison, G. P. Inactiva-
tion of O6-alkylguanine-DNA alkyltransferase. 1. Novel O6-
(hetarylmethyl)guanines having basic rings in the side chain.
J . Med. Chem. 1998, 41, 5265-5271.

O6-Methylguanine-DNA Methyltransferase
J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 24 4061
(9) Chae, M. Y.; Swenn, K.; Kanugula, S.; Dolan, M. E.; Pegg, A.
E.; Moschel, R. C. 8-Substituted O6-benzylguanine, substituted
6(4)-(benzyloxy)pyrimidine, and related derivatives as inacti-
vators of human O6-alkylguanine-DNA alkyltransferase. J . Med.
Chem. 1995, 38, 359-365.
(29) Wibley, J . E.; Pegg, A. E.; Moody, P. C. Crystal structure of the
human O6-alkylguanine-DNA alkyltransferase. Nucleic Acids
Res. 2000, 28, 393-401.
(30) Daniels, D. S.; Mol, C. D.; Arvai, A. S.; Kanugula, S.; Pegg, A.
E.; Tainer, J . A. Active and alkylated human AGT structures:
a novel zinc site, inhibitor and extrahelical base binding. EMBO
J . 2000, 19, 1719-1730.
(31) Xu Welliver, M.; Pegg, A. E. Point mutations at multiple sites
including highly conserved amino acids maintain activity but
render O6-alkylguanine-DNA alkyltransferase insensitive to O6-
benzylguanine Biochem. J . 2000, 347, 519-526.
(32) Kanugula, S.; Goodtzova, K.; Edara, S.; Pegg, A. E. Alteration
of arginine 128 to alanine abolishes the ability of human O6-
alkylguanine-DNA alkyltransferase to repair methylated DNA
but has no effect on its reaction with O6-benzylguanine. Bio-
chemistry 1995, 34, 7113-7119.
(33) Xu Welliver, M.; Kanugula, S.; Loktionova, N. A.; Crone, T. M.;
Pegg, A. E. Conserved residue lysine-165 is essential for the
ability of O6-alkylguanine-DNA alkyltransferase to react with
O6-benzylguanine. Biochem. J . 2000, 347, 527-534.
(34) Goodtzova, K.; Kanugula, S.; Edara, S.; Pegg, A. E. Investigation
of the role of tyrosine-114 in the activity of human O6-alkylgua-
nine-DNA alkyltransferase. Biochemistry 1998, 37, 12489-
12495.
(35) Xu, W. M.; Leitao, J .; Kanugula, S.; Pegg, A. E. Alteration of
the conserved residue tyrosine-158 to histidine renders human
O6-alkylguanine-DNA alkyltransferase insensitive to the inhibi-
tor O6-benzylguanine. Cancer Res. 1999, 59, 1514-1519.
(36) Xu, W. M.; Leitao, J .; Kanugula, S.; Meehan, W. J .; Pegg, A. E.
Role of codon 160 in the sensitivity of human O6-alkylguanine-
DNA alkyltransferase to O6-benzylguanine. Biochem. Pharma-
col. 1999, 58, 1279-1285.
(37) Pohl, J .; Bertram, B.; Hilgard, P.; Nowrousian, M. R.; Stuben,
J .; Wiessler, M. D-19575-a sugar-linked isophosphoramide
mustard derivative exploiting transmembrane glucose transport.
Cancer Chemother. Pharmacol. 1995, 35, 364-370.
(38) Veyhl, M.; Wagner, K.; Volk, C.; Gorboulev, V.; Baumgarten, K.;
Weber, W. M.; Schaper, M.; Bertram, B.; Wiessler, M.; Koepsell,
H. Transport of the new chemotherapeutic agent beta-d-gluco-
sylisophosphoramide mustard (D-19575) into tumor cells is
mediated by the Na+-d-glucose cotransporter SAAT1. Proc. Natl.
Acad. Sci. U.S.A. 1998, 95, 2914-2919.
(39) Reinhard, J .; Eichhorn, U.; Wiessler, M.; Kaina, B. Inactivation
of O6-methylguanine-DNA methyltransferase (MGMT) by glucose-
conjugated inhibitors. Int. J . Cancer 2001, in press.
(40) Balsinger, R. W.; Montgomery, J . A. Synthesis of potential
anticancer agents. XXV. Preparation of 6-alkoxy-2-aminopu-
rines. J . Org. Chem. 1960, 25, 1573-1575.
(41) Shealy, Y. F.; Clayton, J . D.; O’Dell, C. A.; Montgomery, J . A.
v-Triazolo[4, 5-d]pyrimidines. II. O-substituted derivatives of
8-aza-guanine and 8-aza-hypoxanthine. J . Org. Chem. 1962, 27,
4518-4523.
(42) O’Brien, D. E.; Cheng, C. C.; Pfleiderer, W. Pyrimidines. XVIII.
2,4-diamino-5-nitro-6-arylaminopyrimidines. Nitration study of
2,4-diamino-6-chloropyrimidine and a literature correction. J .
Med. Chem. 1966, 9, 573-575.
(43) Kjellberg, J .; Liljenberg, M. Regioselective alkylation of 6-(â-
methoxyethoxy)guanine to give the 9-alkylguanine derivative.
Tetrahedron Lett. 1986, 27, 877-880.
(44) Davis, B. M.; Roth, J . C.; Liu, L.; Xu, W. M.; Pegg, A. E.; Gerson,
S. L. Characterization of the P140K, PVP(138-140)MLK, and
G156A O6-methylguanine-DNA methyltransferase mutants: im-
plications for drug resistance gene therapy. Hum. Gene Ther.
1999, 10, 2769-2778.
(45) Preuss, I.; Haas, S.; Eichhorn, U.; Eberhagen, I.; Kaufmann, M.;
Beck, T.; Eibl, R. H.; Dall, P.; Bauknecht, T.; Hengstler, J .;
Wittig, B. M.; Dippold, W.; Kaina, B. Activity of the DNA repair
protein O6-methylguanine-DNA methyltransferase in human
tumor and corresponding normal tissue. Cancer Detect. Prev.
1996, 20, 130-136.
(10) Dolan, M. E.; Pegg, A. E. O6-benzylguanine and its role in
chemotherapy. Clin. Cancer Res. 1997, 3, 8837-847.
(11) Friedman, H. S.; Kokkinakis, D. M.; Pluda, J .; Friedman, A. H.;
Cokgor, I.; Haglund, M. M.; Ashley, D. M.; Rich, J .; Dolan, M.
E.; Pegg, A. E.; Moschel, R. C.; McLendon, R. E.; Kerby, T.;
Herndon, J . E.; Bigner, D. D.; Schold, S. C., J r. Phase I trial of
O6-benzylguanine for patients undergoing surgery for malignant
glioma. J . Clin. Oncol. 1998, 16, 3570-3575.
(12) Roy, S. K.; Moschel, R. C.; Dolan, M. E. Pharmacokinetics and
metabolism in rats of an inactivator of O6-alkylguanine-DNA
alkyltransferase. Drug Metab. Dispos. 1996, 24, 1205-1211.
(13) Middleton, M. R.; Kelly, J .; Thatcher, N.; Donnelly, D. J .;
McElhinney, R. S.; McMurry, T. B.; McCormick, J . E.; Margison,
G. P. O6-(4-bromothenyl)guanine improves the therapeutic index
of Temozolomide against A375M melanoma xenografts. Int. J .
Cancer 2000, 85, 248-252.
(14) Dolan, M. E.; Stine, L.; Mitchell, R. B.; Moschel, R. C.; Pegg, A.
E. Modulation of mammalian O6-alkylguanine-DNA alkyltrans-
ferase in vivo by O6-benzylguanine and its effect on the sensitiv-
ity of a human glioma tumor to 1-(2-chloroethyl)-3-(4-methyl-
cyclohexyl)-1-nitrosourea. Cancer Commun. 1990, 2, 371-377.
(15) Mitchell, R. B.; Moschel, R. C.; Dolan, M. E. Effect of O6-
benzylguanine on the sensitivity of human tumor xenografts to
1, 3-bis(2-chloroethyl)-1-nitrosourea and on DNA interstrand
cross-link formation. Cancer Res. 1992, 52, 1171-1175.
(16) Moschel, R. C.; McDougall, M. G.; Dolan, M. E.; Stine, L.; Pegg,
A. E. Structural features of substituted purine derivatives
compatible with depletion of human O6-alkylguanine-DNA alkyl-
transferase. J . Med. Chem. 1992, 35, 4486-4491.
(17) Chae, M. Y.; McDougall, M. G.; Dolan, M. E.; Swenn, K.; Pegg,
A. E.; Moschel, R. C. Substituted O6-benzylguanine derivatives
and their inactivation of human O6-alkylguanine-DNA alkyl-
transferase J . Med. Chem. 1994, 37, 342-347.
(18) Mineura, K.; Fukuchi, M.; Kowada, M.; Terashima, I.; Kohda,
K. Differential inactivation of O6-methylguanine-DNA methyl-
transferase activity by O6-arylmethylguanines. Int. J . Cancer
1995, 63, 148-151.
(19) Dolan, M. E.; Roy, S. K.; Garbiras, B. J .; Helft, P.; Paras, P.;
Chae, M. Y.; Moschel, R. C.; Pegg, A. E. O6-alkylguanine-DNA
alkyltransferase inactivation by ester prodrugs of O6-benzylgua-
nine derivatives and their rate of hydrolysis by cellular esterases.
Biochem. Pharmacol. 1998, 55, 1701-1709.
(20) Abril, N.; Luque, R. F.; Prieto, A. M.; Rafferty, J . A.; Margison,
G. P.; Pueyo, C. Bacterial and mammalian DNA alkyltrans-
ferases sensitize Escherichia coli to the lethal and mutagenic
effects of dibromoalkanes. Carcinogenesis 1997, 18, 1883-1888.
(21) Pegg, A. E.; Chung, L.; Moschel, R. C. Effect of DNA on the
inactivation of O6-alkylguanine-DNA alkyltransferase by 9-sub-
stituted O6-benzylguanine derivatives. Biochem. Pharmacol.
1997, 53, 1559-1564.
(22) Moore, M. H.; Gulbis, J . M.; Dodson, E. J .; Demple, B.; Moody,
P. C. Crystal structure of a suicidal DNA repair protein: the
Ada O6-methylguanine-DNA methyltransferase from E. coli.
EMBO J . 1994, 13, 1495-1501.
(23) Wibley, J . E.; McKie, J . H.; Embrey, K.; Marks, D. S.; Douglas,
K. T.; Moore, M. H.; Moody, P. C. A homology model of the three-
dimensional structure of human O6-alkylguanine-DNA alkyl-
transferase based on the crystal structure of the C-terminal
domain of the Ada protein from Escherichia coli. Anticancer Drug
Des. 1995, 10, 75-95. [Erratum: Anticancer Drug Des. 1995,
439].
(24) Vora, R. A.; Pegg, A. E.; Ealick, S. E. A new model for how O6-
methylguanine-DNA methyltransferase binds DNA. Proteins
1998, 32, 3-6.
(25) Brown, L. R.; Deng, J .; Clarke, N. D. Dominant sensitization
variants of human O6-methylguanine-DNA-methyltransferase
obtained by a mutational screen of surface residues. Mutat. Res.
DNA Repair 2000, 459, 81-87.
(26) Cheng, X. DNA modification by methyltransferases. Curr. Opin.
Struct. Biol. 1995, 5, 4-10.
(46) Paulsen H. Fortschritte bei der selektiven chemischen Synthese
verzweigter Oligosaccharide. Angew. Chem. 1982, 94, 184-201.
(47) Helferich, B.; Zirner, J . Notiz u¨ber Tetraacetyl-[4-brom-n-butyl]-
â-d-glucosid. Chem. Ber. 1962, 96, 374.
(27) Cheng, X.; Blumenthal, R. M. Finding a basis for flipping bases.
Structure 1996, 4, 639-645.
(28) Klimasauskas, S.; Kumar, S.; Roberts, R. J .; Cheng, X. HhaI
methyltransferase flips its target base out of the DNA helix. Cell
1994, 76, 357-369.
J M010006E