The molecular basis for pyrimidine-selective DNA binding: Analysis of calicheamicin oligosaccharide derivatives by capillary electrophoresis

Article abstract of DOI:10.1021/ja000519v

Synthetic derivatives of the calicheamicin oligosaccharide were designed to probe the molecular basis for pyrimidine recognition. Binding affinities of the oligosaccharide derivatives for a range of DNA sequences were determined using capillary electropho

Full text of DOI:10.1021/ja000519v

J. Am. Chem. Soc. 2000, 122, 8413-8420
8413
The Molecular Basis for Pyrimidine-Selective DNA Binding:
Analysis of Calicheamicin Oligosaccharide Derivatives by Capillary
Electrophoresis
Kaustav Biswas,† Santona Pal,† Jeffrey D. Carbeck,‡ and Daniel Kahne*,†
Contribution from the Departments of Chemistry and Chemical Engineering, Princeton UniVersity,
Princeton, New Jersey 08544
ReceiVed February 11, 2000
Abstract: Synthetic derivatives of the calicheamicin oligosaccharide were designed to probe the molecular
basis for pyrimidine recognition. Binding affinities of the oligosaccharide derivatives for a range of DNA
sequences were determined using capillary electrophoresis. The results show that having an iodo-substituted
C-ring is neither necessary nor sufficient for pyrimidine-selective binding. Pyrimidine-selective binding depends
on maintaining the overall shape and rigidity of the calicheamicin oligosaccharide. These experiments support
the proposal that the novel pyrimidine selectivity of calicheamicin is the result of a shape-dependent induced-
fit interaction with DNA sequences.
Introduction
Calicheamicin γ1I (1, Figure 1) is an enediyne antibiotic that
is currently in advanced clinical trials for the treatment of
different kinds of cancer.1 It is believed to kill cells by effecting
double stranded cleavage of DNA. Calicheamicin has been
shown to bind in the minor groove of duplex DNA at pyrimidine
tracts including TCCT, TTTT, and CTCT. Following DNA
binding, the enediyne moiety rearranges in the presence of
reducing agents to form a diradical that abstracts hydrogen atoms
from both strands of the DNA backbone, leading to double
strand scission.2
Calicheamicin has attracted a great deal of attention over the
years for several reasons. In addition to having an unusual
structure and mechanism of action, it is the first known example
of a DNA binder that utilizes a carbohydrate moiety to recognize
DNA.3 It is also the first example of a minor groove binder
that is selective for pyrimidine sequences. Most other natural
small molecules that bind in the minor groove are selective either
for AT-rich regions of DNA or GC-rich regions of DNA.4 The
topology of the minor groove at AT-rich and GC-rich sites is
quite different, with AT sites containing a deep minor groove
that forms a cleft for flat, aromatic molecules, and GC sites
containing a wider and shallower groove suited to larger
† Department of Chemistry.
‡ Department of Chemical Engineering.
I
Figure 1. Structures of calicheamicin γ1 and synthetic oligosaccha-
rides.
(1) (a) Acute Myeloid Leukemia: Sievers, E. L.; Appelbaum, F. R.;
Speilberger, R. T.; Forman, S. J.; Flowers, D.; Smith, F. O.; Shannon-
Dorcy, K.; Berger, M. S. Blood 1999, 93, 3678-3684. (b) Ovarian
Cancer: Gillespie, A. M.; Broadhead, T. J.; Chan, S. Y.; Owen, J.;
Farnsworth, A.; Sopwith, M.; Coleman, R. E.; Online Abstracts of the 1998
Annual Meeting of the American Society of Clinical Oncology; see: http://
asco.infostreet.com/prof/me/html/98abstracts/ibt/m_1686.htm.
(2) (a) Zein, N.; Sinha, A. M.; McGahren, W. J.; Ellestad, G. A. Science
1988, 240, 1198-1201. (b) Zein, N.; Poncin, M.; Nilakantan, R.; Ellestad,
G. A. Science 1989, 244, 697-699. (c) Lee, M. D.; Ellestad, G. A.; Borders,
D. B. Acc. Chem. Res. 1991, 24, 235-243. (d) Nicolaou, K. C.; Dai, W.-
M. Angew. Chem., Int. Ed. Engl. 1991, 30, 1387-1416.
(3) (a) Kahne, D. Chem. Biol. 1995, 2, 7-12. (b) Kahne, D.; Silva, D.;
Walker, S. In Bioorganic Chemistry: Carbohydrates; Hecht, S. M., Ed.;
Oxford University Press: New York, 1999; pp 174-197.
(4) Geierstanger, B. H.; Wemmer, D. E. Annu. ReV. Biophys. Biomol.
Struct. 1995, 24, 463-493.
molecules that can hydrogen bond to the exocylic amino groups
of guanine that protrude into the floor of the groove.5 Cali-
cheamicin’s intriguing sequence selectivity has been the focus
of many investigations.
Initial work on the binding selectivity of calicheamicin
focused on the role of the iodine atom on the aromatic ring in
the recognition of TCCT sites. Schreiber and co-workers
suggested that favorable interactions between this iodine atom
and the exocyclic guanine amino groups in the floor of the minor
(5) Dickerson, R. E. Methods Enzymol. 1992, 211, 67-111.
10.1021/ja000519v CCC: $19.00 © 2000 American Chemical Society
Published on Web 08/10/2000

8414 J. Am. Chem. Soc., Vol. 122, No. 35, 2000
Biswas et al.
groove are critical for sequence recognition.6 Removal of the
iodine caused a dramatic decrease in binding to TCCT sites,
supporting Schreiber’s hypothesis.7 NMR studies showed,
however, that the iodine is in close proximity only to the 5′
guanine amino group, implying that only one of the guanine
amino groups can interact with the iodine.8,9 Nucleotide replace-
ment studies confirmed that only the 5′ guanine amino group
is essential for binding.7,10,11 While these experiments implicated
the aromatic iodine in the recognition of one G-C base pair,
they did not shed any light on the underlying basis for the
pyrimidine selectivity.
In fact, the molecular basis for pyrimidine recognition is still
unclear. Only a few studies have addressed how calicheamicin
can bind to recognition sites as different as TCCT and TTTT
with similar affinities. In the preceeding contribution NMR
studies on several different calicheamicin-DNA complexes
have shown that calicheamicin binds in the same orientation
with respect to all pyrimidine recognition sites.8,12 These results
imply that pyrimidine tracts share some common structural
features that calicheamicin is able to sense. We have proposed
that pyrimidine sequences may be able to distort at a lower
energetic cost to accommodate the drug.8 In independent
experiments with nucleosomes and A-tract DNA, Dedon and
Townsend have suggested that the drug recognizes some
“structural perturbations” associated with 3′ junctions of oli-
gopurine tracts.13 According to both hypotheses, the shape of
the drug is the key determinant of the pyrimidine selectivity.14
If so, changes to the drug that do not alter the shape significantly
should not destroy pyrimidine selectivity. Changes that alter
the shape or increase the flexibility substantially should cause
a degradation in selectivity.
substituents on the aromatic ring in pyrimidine-selective binding.
The two analogues were designed (2 and 4, Figure 1) to test
the individual contributions of overall shape and electronic
interactions to the overall binding affinities of the oligosaccha-
rides. Analogue 2 contains the iodine functionality but lacks
other substituents on the aromatic ring. This change alters the
steric surface that the oligosaccharide presents to the minor
groove. It also increases the flexibility of the molecule in the
vicinity of the C-D rings compared to the natural oligosac-
charide. This analogue enables a direct estimate of the impor-
tance of the iodine substituent in binding selectivity in com-
parison to maintaining the shape of the molecule. We also
wanted to determine whether the iodine per se is crucial, or
whether it can be replaced by a substituent of comparable size.
Therefore, we synthesized an oligosaccharide analogue (4) in
which the iodine was replaced with a methylthio substituent.
Analogue 4 presents the same steric surface16 and maintains
the rigidity of the natural system but has different electronic
properties along one edge of the aromatic ring. If pyrimidine
selectivity depends on maintaining the overall shape and rigidity
of the oligosaccharide, then one would predict that 4 should be
a pyrimidine selective binder that binds preferred recognition
sites with similar affinity as the natural calicheamicin oligosac-
charide. Conversely, if attractive interactions involving the
iodine are critical, then the binding affinity and pyrimidine
selectivity should decrease substantially.17
Synthesis of Calicheamicin Oligosaccharide Analogues.
Analogue 2 was synthesized following a convergent route
previously developed for the natural oligosaccharide 3 (Figure
1),18 using the known compound 3-iodo-4-hydroxy-benzalde-
hyde19 as the modified C-ring.
The synthesis of analogue 4 is described in Scheme 1. The
protected phenolic rhamnoside 518 was converted to 7 by the
exchange of protecting groups, followed by the substitution of
the iodine with a methylthio functionality by treatment with
methyllithium at -78 °C and subsequent quenching with
dimethyl disulfide. Protecting-group exchange followed by
deprotection of the benzylic alcohol with buffered pyridinium
hydrofluoride20 and Swern oxidation21 revealed the aromatic
aldehyde 10. Oxidation of the aldehyde functionality to the
To evaluate the relative importance of specific interactions
versus shape-selective recognition of DNA by the calicheamicin
oligosaccharide, we synthesized two synthetic analogues of the
oligosaccharide designed to probe these issues. Below we
present results demonstrating that shape complementarity plays
a more important role than specific attractive interactions in
determining pyrimidine-selective binding.
Results
(15) (a) Drak, J.; Iwasawa, N.; Danishefsky, S.; Crothers, D. M. Proc.
Natl. Acad. Sci. U.S.A. 1991, 88, 7464-7468. (b) Aiyar, J.; Danishefsky,
S. J.; Crothers, D. M. J. Am. Chem. Soc. 1992, 114, 7552-7554. (c)
Nicolaou, K. C.; Tsay, S.-C.; Suzuki, T.; Joyce, G. J. Am. Chem. Soc. 1992,
114, 7555-7557. (d) Sugiura, Y.; Uesawa, Y.; Takahashi, Y.; Kuwahara,
J.; Golik, J.; Doyle, T. W. Proc. Natl. Acad. Sci. U.S.A. 1989, 86, 7672-
7676.
(16) Iodine and the ethyl group have similar values in Taft’s steric scale
(see: March, J. AdVanced Organic Chemistry; John Wiley and Sons: New
York, 1992; p 285, and references therein). Sulfur is isosteric with
methylene. For example, see Mislow’s determination of the absolute
configuration of R-lipoic acid (Mislow, K. and Meluch, W. C. J. Am. Chem.
Soc. 1956, 78, 5920-5923) and the formation of similar polymorphic crystal
forms of D,L-methionine and D,L-norleucine (see Dalhus, B. and Gorbitz,
C. H. Acta Crystallogr., Sect. C (Cs. Str. Comm.) 1999, 55, 1105-1112,
and references therein). Therefore, the replacement of iodine by the
methylthio functionality is expected to result in the presentation of a
comparable steric surface on the aromatic ring.
Design. The aryl rhamnoside moiety of calicheamicin (C and
D rings, 1, Figure 1) plays an important role in oligopyrimidine
recognition.15 We synthesized analogues of the calicheamicin
oligosaccharide (3, Figure 1) in order to test the role of
(6) Hawley, R. C.; Kiessling, L. L.; Schreiber, S. L. Proc. Natl. Acad.
Sci. U.S.A. 1989, 86, 1105-1109.
(7) Li, T.; Zeng, Z.; Estevez, V. A.; Baldenius, K. U.; Nicolaou, K. C.;
Joyce, G. F. J. Am. Chem. Soc. 1994, 116, 3709-3715.
(8) (a) Walker, S.; Murnick, J.; Kahne, D. J. Am. Chem. Soc. 1993, 115,
7954-7961. (b) Walker, S. L.; Andreotti, A. H.; Kahne, D. E. Tetrahedron
1994, 50, 1351-1360.
(9) (a) Paloma, L. G.; Smith, J. A.; Chazin, W. J.; Nicolaou, K. C. J.
Am. Chem. Soc. 1994, 116, 3697-3708. (b) Ikemoto, N.; Kumar, R. A.;
Ling, T. T.; Ellestad, G. A.; Danishefsky, S. J.; Patel, D. J. Proc. Natl.
Acad. Sci. U.S.A. 1995, 92, 10506-10510. (c) Kumar, R. A.; Ikemoto, N.;
Patel, D. J. J. Mol. Biol. 1997, 265, 187-201.
(10) Bailly, C.; Waring, M. J. J. Am. Chem. Soc. 1995, 117, 7311-
7316.
(11) Chatterjee, M.; Mah, S. C.; Tullius, T. D.; Townsend, C. A. J. Am.
Chem. Soc. 1995, 117, 8074-8082.
(12) Kalben, A.; Pal, S.; Andreotti, A. H.; Walker, S.; Gange, D.; Biswas,
K.; Kahne, D. J. Am. Chem. Soc. 2000, 122, 8403-8412.
(13) (a) Mah, S. C.; Price, M. A.; Townsend, C. A.; Tullius, T. D.
Tetrahedron 1994, 50, 1361-1378. (b) Kuduvalli, P. N.; Townsend, C.
A.; Tullius, T. D. Biochemistry 1995, 34, 3899-3906. (c) Yu, L.; Salzberg,
A. A.; Dedon, P. C. Bioorg. Med. Chem. 1995, 3, 729-741.
(14) Recent work by Danishefsky and Crothers also support this model.
See: Sissi, C.; Aiyar, J.; Boyer, S.; Depew, K.; Danishefsky, S.; Crothers,
D. M. Proc. Natl. Acad. Sci. U.S.A. 1999, 96, 10643-10648.
(17) The dipolar interaction between the exocyclic nitrogen of guanine
and the iodine should be significantly stronger than that between the external
methyl group of the methylthio functionality and the guanine amino group.
The internal sulfur atom is not expected to contact the DNA due to steric
hindrance from the external methyl group and the two ortho-functionalities
on the aromatic ring (see Figure 1).
(18) Kim, S.-H.; Augeri, D.; Yang, D.; Kahne, D. J. Am. Chem. Soc.
1994, 116, 1766-1775.
(19) Barnes, J. H.; Borrows, E. T.; Elks, J.; Hems, B. A.; Long, A. G.
J. Chem. Soc. 1950, 2824-2833.
(20) Trost, B. M.; Caldwell, C. G.; Murayama, E.; Heissler, D. J. Org.
Chem. 1983, 48, 3252-3265.
(21) Omura, K.; Swern, D. Tetrahedron 1978, 34, 1651-1660.

Basis for Pyrimidine-SelectiVe DNA Binding
J. Am. Chem. Soc., Vol. 122, No. 35, 2000 8415
Scheme 1. Synthesis of the Calicheamicin Oligosaccharide Analogue 4a
a Reagents and conditions: (a) DIBAL-H, Et2O, -78 °C; TESOTf, Et3N, CH2Cl2 (70%, 2 steps); (b) MeLi, Et2O, -78 °C, then MeSSMe, -78
°C to -40 °C, then Et3N (74%); (c) AcOH, THF, H2O (3:1:1); Ac2O, pyr (82%, 2 steps); (d) HF‚pyr (88%); (e) DMSO, (COCl)2, CH2Cl2, then
Et3N, -78 °C (88%); (f) NBS, AIBN, CCl4, 80 °C (74%); (g) K2CO3, MeOH (74%); (h) ClPO(OEt)2, Et3N, THF (62%); (i) n-Bu3P, THF; then
DMAP, 13 (33%).
carboxylic acid in the presence of the thioether moiety proved
to be troublesome. After model studies with 4-(methylthio)-
benzaldehyde and various oxidizing agents, the desired trans-
formation was achieved by treatment of the aldehyde with
N-bromosuccinimide in the presence of a catalytic amount of
the radical initiator AIBN (2,2′-azobisisobutyronitrile). This led
to the in situ formation of an acid bromide,22 which was
hydrolyzed to the carboxylic acid 11 on workup. 11 was
subsequently deprotected and converted to 4, as described
previously.18
Estimation of Pyrimidine Selectivity. Analogue 2 was tested
for its ability to inhibit the cleavage of a radiolabeled DNA
duplex containing one calicheamicin binding site.7 The 20-mer
DNA was incubated with intact calicheamicin and various
concentrations of either analogue 2 or natural oligosaccharide
3 at 37 °C. Enediyne-induced cleavage was triggered by the
addition of â-mercaptoethanol, and the resulting products were
resolved by gel electrophoresis. Analogue 2 was found to inhibit
calicheamicin-induced DNA cleavage at much higher concentra-
tions (approximately 2 orders of magnitude) than the natural
oligosaccharide 3. Similar results were obtained with both the
TCCT (Figure 2) and the TTTT sites (data not shown).23 This
synthetic modification of the C ring thus leads to a significant
loss in affinity for pyrimidine sequences. We concluded that
selective DNA recognition requires, in addition to the iodine,
at least one or more of the three other substituents on the
aromatic ring.
Figure 2. Cleavage inhibition reactions with calicheamicin oligosac-
charide 3 and analogue 2. Lane 1: 5′-end labeled d(GCTGAGTGTC-
CTGCACTGCC). Lane 2: cleavage of TCCT duplex with 3 µM
I
calicheamicin γ1 (1). Lanes 3, 5, 7, 9, 11: cleavage of TCCT duplex
with 3 µM calicheamicin γ1I in the presence of 10-6 M, 10-5 M, 10-4
M, 10-3 M, 10-2 M native oligosaccharide 3. Lanes 4, 6, 8, 10, 12:
cleavage of TCCT duplex with 3 µM calicheamicin γ1I in the presence
of 10-6 M, 10-5 M, 10-4 M, 10-3 M, 10-2 M analogue 2.
possible to measure accurately the binding affinity of similar
ligands to closely related DNA sequences using inhibition of
calicheamicin-induced DNA cleavage. Calicheamicin binds to
and cuts the different recognition sequences at different rates.11
Therefore, the use of cleavage inhibition to estimate and
compare differential binding affinities will introduce errors in
the measurement. Alternative methods that have been used to
measure DNA binding affinities of calicheamicin include
quantitative hydroxyl radical footprinting, circular dichroism,
and microcalorimetry.11,24 Of these, microcalorimetry is the most
direct method for evaluating binding constants, but it requires
large amounts of material and is not practical for evaluating
analogues produced through multistep syntheses. We wanted a
Capillary Electrophoresis. To compare the binding of the
native oligosaccharide 3 and the methylthio analogue 4 to
various DNA sequences, we needed a better binding assay than
we had used for 2. The cleavage-based assay allowed us to
determine that 2 binds poorly to DNA. However, it is not
(22) Cheung, Y.-F. Tetrahedron Lett. 1979, 3809-3810.
(23) Pal, S. Ph.D. Thesis, Princeton University, 1997.
(24) Krishnamurthy, G.; Ding, W.-D.; O’Brien, L.; Ellestad, G. A.
Tetrahedron 1994, 50, 1341-1349.

8416 J. Am. Chem. Soc., Vol. 122, No. 35, 2000
Biswas et al.
rapid, direct method that did not require invasive labeling of
the oligosaccharide, that was suitable for small amounts of
material, and that could be used under a wide range of
conditions. Capillary electrophoresis (CE) seemed appropriate
for our needs because it requires nanogram quantities of material
and the measurements are direct and fast.
Capillary electrophoresis has developed in the 1990s as a
highly sensitive technique for the separation of charged analytes.
Affinity capillary electrophoresis has been used extensively for
measuring protein-ligand affinities.25 Because both DNA itself
and most DNA ligands are charged, CE would seem to be ideal
for studying the differential DNA binding affinities of synthetic
ligands. Although preliminary investigations have validated CE
as a technique for measuring DNA-ligand affinities,26 it has
not yet been applied to address specific questions. Below, we
describe a CE-based assay for the quantitative measurement of
DNA binding affinities of synthetic oligosaccharides. This
method should be applicable to the study of other ligand-DNA
complexes.
Figure 3. Electropherogram sections from CE experiments involving
calicheamicin oligosaccharide (3) and the TCCT 15-mer duplex,
showing only the free ligand peak in the presence and absence of DNA
(UV absorbance is plotted against -1/time). Various concentrations
of ligand were titrated against a fixed concentration of DNA(25 µM)
in a 20mM NaCl, 30 mM Tris buffer at pH 7.5 (with 1% DMSO). The
difference between the ligand peak in the absence and presence of DNA
(as shown by the arrows) represents the estimate of the bound ligand
concentration for each data point. The concentrations of the free and
bound ligands were used to obtain the association constant by Scatchard
analysis.
CE is based on the mobility of charged and neutral species
through a buffer-filled capillary upon the application of voltage.
The different analytes are separated due to differences in their
charge-to-mass ratio, and the concentrations are measured by
online UV detection. Affinity capillary electrophoresis involves
measuring the change in mobility of a ligand (or receptor) in
the presence of different amounts of a receptor (or ligand) in
the buffer. The mobility change can be used to estimate binding
constants.25a We tried to validate affinity capillary electrophore-
sis as a method for measuring ligand-DNA interactions using
oligosaccharide 3 as the test case because a dissociation constant
for the binding of this oligosaccharide to DNA had been
reported.7 We chose to use 15-mer duplexes containing a single
binding site because 15-mers are long enough to be stable and
yet short enough to minimize undesired interactions with the
flanking sequences. Because the component in the buffer is
needed in larger quantities, we first examined the mobility of
ligand 3 in a capillary containing DNA in the buffer. The
presence of DNA in the buffer led to problems with detector
response, leading us to investigate alternative methods. Li and
Martin have reported the utility of capillary zone electrophoresis
for measuring the binding of small peptides to DNA.27 In
capillary zone electrophoresis, a pre-equilibrated mixture of
ligand and receptor is injected into the buffer-filled capillary.
Since most DNA ligands are positively charged, the separation
of the free ligand from the negatively charged free DNA and
the ligand-DNA complex is very efficient, leading to an
accurate estimation of the free ligand concentration in the
equilibrium mixture. Comparison of this value with the total
ligand concentration (obtained from a measurement in the
absence of DNA) affords an estimate of the bound ligand
concentration for each data point. The affinity constant can be
obtained from a Scatchard analysis of the above data (Figure
3).28
the TCCT duplex (Figure 3). There is one report regarding the
binding affinity of 3 to a TCCT site. Using a cleavage inhibition
assay, Nicolaou and co-workers estimated a dissociation constant
of 4.1 µM.7 With the CE assay, we found the association
constant (Ka) for the binding of 3 to TCCT to be 4.74 × 104
M-1 (Kd ) 21.07 µM, Figure 4a). This 5-fold discrepancy could
be due to different experimental conditions or to different
flanking sequences. However, the cleavage inhibition assay may
be less accurate because it is indirect and rests on the validity
of a number of assumptions. Townsend and Tullius noted a
similar discrepancy between Kd values obtained by cleavage
inhibition and other methods (i.e., footprinting and microcalo-
rimetry).11 To further calibrate our technique, we examined the
binding of the polyamide antibiotic netropsin to an AT-rich
duplex by capillary electrophoresis, and compared the affinity
constant to a reported literature value measured with DNase
footprinting.29 We obtained a dissociation constant of 1.66 µM,
in close agreement with the literature value of 1.0 µM.30
After validating our technique, we measured the interaction
of the calicheamicin oligosaccharide 3 with the other duplexes,
and repeated the experiments with the synthetic analogue 4 as
the ligand. Association constants were obtained from the CE
data (Figure 4). We compared the affinities of the methylthio
analogue 4 to the TCCT and TTTT sequences, and to the
corresponding affinities exhibited by the native iodo-oligosac-
charide 3 (Table 1). The substitution of methylthio for iodine
resulted in a 3-fold decrease in affinity for TCCT, and had no
effect on the affinity for TTTT. Thus, analogue 4 has a similar
affinity for these known calicheamicin binding sequences as 3.
Using the above conditions, we measured the affinity of 3 to
(25) (a) Colton, I. J.; Carbeck, J. D.; Rao, J.; Whitesides, G. M.
Electrophoresis 1998, 19, 367-382. (b) Heegaard, N. H. H.; Nilsson, S.;
Guzman, N. A. J. Chromatogr. B 1998, 715, 29-54.
(26) (a) Hamdan, I. I.; Skellern, G. G.; Waigh, R. D. Nucleic Acids Res.
1998, 26, 3053-3058. (b) Battersby, T. R.; Ang, D. N.; Burgstaller, P.;
Jurczyk, S. C.; Bowser, M. T.; Buchanan, D. D.; Kennedy, R. T.; Benner,
S. A. J. Am. Chem. Soc. 1999, 121, 9781-9789.
A comparison of the binding affinities to TCCT and TTTT
sites does not, by itself, permit us to draw any conclusions about
pyrimidine selectivity. There was the formal possibility that we
had just designed a better non-specific binder to DNA.
(27) Li, C.; Martin, L. M. Anal. Biochem. 1998, 263, 72-78.
(28) Scatchard, G. Ann. N. Y. Acad. Sci. 1949, 51, 600-604. The relative
linearity of the data from the binding isotherm, representing the formation
of a 1:1 complex, validates the use of a simple Scatchard analysis, rather
than a neighbor exclusion model (McGhee, J. D.; von Hippel, P. H. J. Mol.
Biol. 1974, 86, 469-489). We thank a reviewer for this observation.
(29) Rehfuss, R.; Goodisman, J.; Dabrowiak, J. C. Biochemistry 1990,
29, 777-781.

Basis for Pyrimidine-SelectiVe DNA Binding
J. Am. Chem. Soc., Vol. 122, No. 35, 2000 8417
Table 1. Association Constants for the Binding of the
Calicheamicin Oligosaccharide (3) and the Synthetic Analogue (4)
to 15-mer DNA Duplexes
Ka (104 M-1
TTTT TATA TAAT TGGT
4.74 ( 0.25 2.36 ( 0.44
4 (SCH3) 1.73 ( 0.20 2.31 ( 0.41
)
ligand
3 (I)
TCCT
a
a
a
a
a
a
a For low affinity sites, association constants could not be determined
showed that neither the native oligosaccharide 3 nor the synthetic
analogue 4 bound to these DNA sequences (the free ligand peak
height remained the same in the presence and absence of the
duplexes). The association constants could not be determined
due to the absence of any measurable binding (the binding
affinities of analogue 2 to DNA could not be quantitated by
capillary electrophoresis due to a similar reason). Thus, 4
maintains the oligopyrimidine selectivity exhibited by the native
oligosaccharide.
Discussion
Role of the Iodine in Pyridine Selectivity. Previous studies
have attributed the role of a specific contact to the aromatic
iodine. NMR studies have shown that the iodine is in close
proximity to a guanine amino group present in some of the
sites.8,9 Studies with synthetic derivatives and nucleotide
replacements have also shown that the iodine plays a critical
role in the recognition process.7,10,11 In an effort to quantitate
the electronic contribution due to the iodine, the interaction with
the guanine amino group has been measured by different
techniques. Using their cleavage inhibition assay, Nicolaou and
co-workers have estimated that the iodine-amino group interac-
tion is worth ∼2.3 kcal/mol. This estimate was based on a
comparison of the affinities of the native oligosaccharide 3 and
the corresponding des-iodo analogue for a TCCT site.7 Townsend
and co-workers concluded that the iodine-amino group interac-
tion is only worth ∼1 kcal/mol based on comparing the extent
of cleavage of DNA sequences containing guanine or inosine
by calicheamicin.11
Capillary electrophoresis gave us an opportunity to evaluate
the non-covalent iodine-amino group interaction by a direct
biophysical technique. By comparing the binding of 3 and
analogue 4 to both TCCT and TTTT sites, we have concluded
that the iodine-amino group interaction is worth about 0.6 kcal/
mol.31 Furthermore, both 3 and 4 bind only to oligopyrimidine
sequences, demonstrating that the role of the iodine in pyrimi-
dine selectivity is largely steric and that it can be successfully
substituted by a functionality of similar size. The large decrease
in binding affinity to the TCCT site when the iodine is replaced
by hydrogen may be due to at least three factors. First,
substitution of the iodine by hydrogen creates a gap along one
edge of the C ring that would be a void in a complex with DNA.
Second, the change also reduces the rigidity of the molecule
because rotation around the C-D linkage is less hindered.
Increased flexibility could well reduce specificity for TCCT
sites. Finally, the loss of the iodine-nitrogen interaction also
decreases the affinity, albeit modestly.
Figure 4. Scatchard plots of the binding of (a) the calicheamicin
oligosaccharide 3 and (b) the methylthio analogue 4 to DNA 15-mer
duplexes with TCCT or TTTT as the binding site. The data were fit to
eq 1 using Kaleidagraph (see Experimental Section).
Therefore, we also evaluated the binding of oligosaccharides 3
and 4 to 15-mer duplexes containing TATA, TAAT, and TGGT,
respectively, as the binding sites. These DNA sequences have
similar base pair compositions as the TTTT and TCCT sites,
but they do not contain sequential pyrimidines, the common
feature of all calicheamicin binding sites. CE experiments
(30) The experiments with netropsin and a TATA-containing 15-mer
duplex were performed in DNase I buffer (10 mM Tris‚Cl, 8 mM MgCl2,
2 mM CaCl2) in order to replicate the conditions used in the DNase titration
experiment. The DNA sequence was 5′-CGTCGGTATACCACG-3′, in-
corporating the 8-mer duplex used in the footprinting experiments in the
center. Netropsin concentrations were determined spectroscopically from
the published extinction coefficient (ꢀ296 ) 20200 M-1 cm-1). Histamine
was used as a positive marker in all the separations to monitor changes in
EOF. The DNA and the ligand were initially equilibrated at 25 °C in the
running buffer for 15 min before injection into the capillary. The experiments
were performed in a procedure similar to the experiments with the synthetic
oligosaccharides. Different concentrations of netropsin were titrated with
10 µM DNA to obtain the affinity constant by Scatchard analysis.
Role of Molecular Shape in Pyrimidine Selectivity. We
have suggested that the pyrimidine-selective binding exhibited
by calicheamicin is due to an induced-fit interaction between
the oligosaccharide and the DNA duplex.8a This hypothesis was
based on two observations. First, the pyrimidine selectivity
cannot be explained on the basis of specific directional contacts
because the functional group arrays at different pyrimidine sites
We obtained an affinity constant of (6.0 ( 0.23) × 105 M-1 (Kd ) 1.66
µM), in close agreement to the value obtained by footprinting, 106 M-1
(Kd ) 1.0 µM).
(31) The ∆∆G was calculated from the measured Ka’s at 298 K.

8418 J. Am. Chem. Soc., Vol. 122, No. 35, 2000
Biswas et al.
differ markedly. Second, the conformation of the DNA changes
in a consistent way in all the specific complexes we have
examined.
ence (2-fold) that the oligosaccharide displays for TCCT over
TTTT sequences is due to the iodine, but this functionality plays
no role in the general selectivity for pyrimidine sites.
Induced-fit recognition is one way that proteins achieve
selectivity in DNA binding. For example, TATA binding protein
binds to a range of AT sites in the regulatory regions of genes.32
The recognition sites share an ability to adapt to the shape of
the protein. Von Hippel and Berg have noted that induced-fit
recognition of DNA requires a relatively rigid protein.33 The
calicheamicin oligosaccharide contains a number of unusual
structural features that enforce a particular shape.34-36 NMR
studies on calicheamicin have shown that the oligosaccharide
has a well-defined conformation and that this conformation is
maintained on binding.8,12 Hence, the conformational changes
in the DNA and the lack thereof in calicheamicin are consistent
with the hypothesis that recognition involves an induced-fit
process in which the DNA adapts to the ligand.
Experimental Section
General Methods. Unless otherwise stated, all chemicals were
purchased from Aldrich or Sigma and used without further purification.
Pyridinium hydrofluoride (70% HF) was purchased from Fluka.
Methylene chloride, toluene, pyridine, and triethylamine were distilled
from calcium hydride under dry argon. Diethyl ether and tetrahydrofuran
were distilled from potassium benzophenone ketyl under dry argon.
All reactions were carried out under argon atmosphere with freshly
distilled solvents under anhydrous conditions unless otherwise noted.
Analytical thin-layer chromatography (TLC) was performed using
silica gel 60 F254 precoated plates (0.25 mm thickness) with a
fluorescent indicator. The developed plates were examined under short-
wave UV light and stained with anisaldehyde. Flash column chroma-
tography was performed using silica gel 60 (230-400 mesh) from EM
Science.
In this work we have examined the relative importance of
specific functional groups versus overall shape in the selective
recognition of pyrimidine sequences. As shown with the
oligosaccharide analogue 2, which contains an iodine attached
to the aromatic ring but lacks the other three substituents, there
is a loss in binding affinity to DNA. These substituents may
contribute to the energetics of binding by making favorable
contacts to the DNA, but none of them have been singled out
as making particularly strong contacts. It seems more likely that
the substituents on the aromatic ring play a more subtlesbut
essentialsrole in binding. Ortho substituents restrict rotation
around bonds to aromatic rings.37 The presence of ortho
substituents on the aromatic ring likely enforces a particular
conformation of the calicheamicin oligosaccharide. Removing
all three substituents would not only leave a gap along one edge
of the C ring, but it would permit the C and D rings to adopt a
number of other conformations. If the binding selectivity is
largely determined by the oligosaccharide inducing an appropri-
ate fit in the DNA duplex, increased flexibility would be
expected to destroy binding selectivity. For entropic reasons,
the flexibility could also lead to a decrease in binding affinity.
In contrast, analogue 4, in which the iodine is replaced by a
substitutent of comparable size that differs in its electronic
properties, maintains pyrimidine selectivity. Apparently, main-
taining the overall shape and rigidity of the oligosaccharide is
more important in determining pyrimidine selectivity than any
specific contacts between the oligosaccharide and the DNA.
NMR spectra were recorded on a JEOL GSX 270 or a Varian Inova
500 Fourier transform NMR spectrometer. Chemical shifts (δ) are
reported in parts per million (ppm) downfield from tetramethylsilane
(TMS). Coupling constants (J) are reported in hertz (Hz). Multiplicities
are abbreviated as follows: singlet (s), doublet (d), triplet (t), quartet
(q), multiplet (m), doublet of doublets (dd), and broad singlet (bs). High-
resolution mass spectra were obtained on a VG ZAB under conditions
of fast atom bombardment by Dr. Ron New at the University of
California at Riverside Mass Spectrometry Facility.
Oligosaccharide 2. Methyl 4,6-Dideoxy-4-(((2,6-dideoxy-4-S-(4-
((6-deoxy-3-O-methyl-r-L-manno-pyranosyl)oxy)-3-iodobenzoyl)-4-
thio-â-D-ribo-hexopyranosyl)oxy)amino)-2-O-(2,4-dideoxy-4-(N-eth-
ylamino)-3-O-methyl-r-L-xylopyranosyl)-â-D-glucopyranoside (2).
Rf 0.2 (10% MeOH/CHCl3 with 1% NH3); 1H NMR (CD3OD, 270
MHz) δ 8.36 (d, J ) 2.0 Hz, 1 H, ArH), 8.01 (dd, J ) 8.6, 2.0 Hz, 1
H, ArH), 7.26 (d, J ) 8.6 Hz, 1 H, ArH), 5.69 (d, J ) 2.0 Hz, 1 H,
D-1), 5.37 (m, 1 H, E-1), 5.08 (dd, J ) 10.3, 1.9 Hz, 1 H, B-1), 4.40
(t, J ) 2.0 Hz, 1 H, D-2), 4.21 (m, 3 H, D-5, B-3 and A-1), 4.06 (dq,
J ) 10.6, 6.3 Hz, 1 H, B-5), 3.91 (t, J ) 9.8 Hz, 1 H, A-3), 3.86 (t, J
) 10.9 Hz, 1 H, E-5axial), 3.72-3.61 (m, 4 H, A-5, B-4, D-3 and
E-5equatorial), 3.58 (t, J ) 9.2 Hz, 1 H, D-4), 3.54 (m, 4 H, E-3 and
OCH3), 2.76-2.66 (m, 3 H, E-4 and NCH2), 2.38 (ddd, J ) 12.7, 2.9,
1.5 Hz, 1 H, E-2equatorial), 2.24 (t, J ) 9.8 Hz, 1 H, A-4), 1.95 (dt, J )
12.4, 2.9 Hz, 1 H, B-2equatorial), 1.74 (ddd, J ) 12.7, 10.3, 2.9 Hz, 1 H,
B-2axial), 1.44 (m, 1 H, E-2axial), 1.36 (d, J ) 5.9 Hz, 3 H, B-6), 1.27
(d, J ) 5.3 Hz, 3 H, A-6), 1.21 (d, J ) 5.3 Hz, 3 H, D-6), 1.15 (t, J
) 7.2 Hz, 3 H, NCH2CH3); HRFABMS calcd for C35H55IN2O15S (M)+
902.2369, found 902.2380.
Synthesis of Oligosaccharide 4. 2-((tert-Butyldiphenylsiloxy)-
methyl)-5-((6-deoxy-3-O-methyl-2,4-bis-O-(triethylsilyl)-r-L-man-
nopyranosyl)oxy)-3,4-dimethoxy-6-iodotoluene (6). DIBAL-H (5.85
mL of a 1.0 M solution in toluene, 5.85 mmol, 5.0 equiv) was added
to a solution of 5 (1.0 g, 1.17 mmol) in 45 mL of Et2O at -78 °C. The
reaction was warmed to -20 °C over 30 min, quenched with 5.0 mL
methanol, and poured into a solution of 1 N HCl (50 mL). The product
was extracted with Et2O (4 × 50 mL), dried over Na2SO4, concentrated,
and purified by flash chromatography (50% EtOAc/petroleum ether)
to give 700 mg (72%) of the rhamnoside diol as a white solid: Rf 0.2
Conclusions
In this work, we report the use of capillary electrophoresis
to address specific questions about small molecule-DNA
molecular recognition. We show that the calicheamicin oli-
gosaccharide binds to TCCT and TTTT sequences with similar
affinity. Furthermore, we show that synthetic changes to the
oligosaccharide that alter the shape of the molecule cause a
degradation in selectivity, while analogues that maintain the
overall shape, even while removing specific contacts, demon-
strate a similar oligopyrimidine selectivity. The modest prefer-
1
(50% EtOAc/petroleum ether); H NMR (CDCl3, 270 MHz) δ 7.70-
7.67 (m, 4 H, ArH), 7.42-7.35 (m, 6 H, ArH), 5.64 (d, J ) 1.3 Hz, 1
H, H-1), 4.76 (s, 2 H, benzylic CH2), 4.51 (dd, J ) 3.3, 1.6 Hz, 1 H,
H-2), 4.32 (dq, J ) 9.8, 6.0 Hz, 1 H, H-5), 3.84 (dd, J ) 9.3, 3.3 Hz,
1 H, H-3), 3.76 (s, 3 H, OCH3), 3.64 (t, J ) 9.7 Hz, 1 H, H-4), 3.64
(s, 3 H, OCH3), 3.57 (s, 3 H, OCH3), 2.50 (s, 3 H, ArCH3), 1.32 (d, J
) 6 Hz, 3 H, H-6), 1.04 (s, 9 H, C(CH3)3); 13C NMR (CDCl3, 67.5
MHz) δ 152.8, 149.7, 142.9, 137.9, 135.9, 133.4, 129.5, 128.7, 127.5,
102.7, 93.8, 80.9, 71.7, 71.1, 70.2, 67.0, 61.2, 60.6, 58.6, 57.1, 57.0,
26.8, 25.5, 19.3, 17.5, 17.5; HRFABMS calcd for C33H43O8SiI (MNa)+
745.1669, found 745.1650.
(32) Kim, J. L.; Burley, S. K. Struct. Biol. 1994, 1, 638-653.
(33) von Hippel, P. H.; Berg, O. G. Protein-Nucleic Acid Interactions;
Saenger, W., Heinemann, U., Eds.; CRC Press: Boca Raton, Fl, 1989; pp
1-18.
(34) Walker, S.; Valentine, K. G.; Kahne, D. J. Am. Chem. Soc. 1990,
112, 6428-6429.
(35) Walker, S.; Yang, D.; Kahne, D.; Gange, D. J. Am. Chem. Soc.
1991, 113, 4716-4717.
(36) Walker, S.; Gange, D.; Gupta, V.; Kahne, D. J. Am. Chem. Soc.
1994, 116, 3197-3206.
To a solution of the diol (254.2 mg, 0.35 mmol) in 10 mL CH2Cl2
at 0 °C was added Et3N (233 µL, 1.75 mmol, 5.0 equiv), followed by
(37) Oki, M. Top. Stereochem. 1983, 14, 1-81.

Basis for Pyrimidine-SelectiVe DNA Binding
J. Am. Chem. Soc., Vol. 122, No. 35, 2000 8419
the addition of triethylsilyl trifluoromethanesulfonate (240 µL, 1.05
mmol, 3.0 equiv). The reaction was maintained at 0 °C for 20 min,
quenched with 1.0 mL methanol, diluted with 10 mL Et2O, and washed
with water (20 mL), saturated aqueous NaHCO3 (20 mL), and brine
(20 mL). The organic layer was dried over Na2SO4, concentrated, and
purified by flash chromatography (2% tert-butyl methyl ether/petroleum
ether) to give 314.3 mg (94%) of 6 as a colorless oil: Rf 0.45 (5%
tert-butyl methyl ether/petroleum ether); 1H NMR (CDCl3, 270 MHz)
δ 7.70-7.67 (m, 4 H, ArH), 7.45-7.33 (m, 6 H, ArH), 5.31 (d, J )
1.3 Hz, 1 H, H-1), 4.79 (d, J ) 10.9 Hz, 1 H, benzylic H), 4.72 (d, J
) 10.9 Hz, 1 H, benzylic H), 4.47 (dd, J ) 2.6, 1.3 Hz, 1 H, H-2),
4.21 (dq, J ) 9.8, 6 Hz, 1 H, H-5), 3.74 (t, J ) 9.8 Hz, 1 H, H-4), 3.73
(s, 3 H, OCH3), 3.63 (s, 3 H, OCH3), 3.56 (dd, J ) 9.8, 2.6 Hz, 1 H,
H-3), 3.42 (s, 3 H, OCH3), 2.50 (s, 3 H, ArCH3), 1.25 (d, J ) 6 Hz, 3
H, H-6), 1.04 (s, 9 H, C(CH3)3), 1.01-0.89 (m, 18 H, Si(CH2CH3)3),
0.73-0.54 (m, 12 H, Si(CH2CH3)3); 13C NMR (CDCl3, 67.5 MHz) δ
152.8, 150.7, 143.2, 137.8, 135.7, 133.5, 129.5, 128.5, 127.5, 104.8,
94.3, 81.4, 72.3, 72.2, 68.6, 61.2, 60.6, 58.6, 57.1, 26.8, 25.6, 19.3,
18.0, 7.0, 6.8, 5.2, 4.9; HRFABMS calcd for C45H71O8Si3I (MNa)+
973.3399, found 973.3435.
2-((tert-Butyldiphenylsiloxy)methyl)-5-((6-deoxy-3-O-methyl-2,4-
bis-O-(triethylsilyl)-r-L-mannopyranosyl)oxy)-3,4-dimethoxy-6-me-
thylthio-toluene (7). MeLi (65 µL of a 1.4 M solution in Et2O, 0.085
mmol, 2.5 equiv) was added to a solution of 6 (32.5 mg, 0.034 mmol)
in 1 mL Et2O at -78 °C, followed by the addition of MeSSMe (20
µL, 0.17 mmol, 5.0 equiv) after 5 min. The reaction temperature was
maintained at -78 °C for 10 min, quenched with 200 µL Et3N, and
poured into 5 mL of saturated aqueous NaHCO3. The aqueous layer
was extracted with Et2O (3 × 5 mL), the organic layers were combined,
dried over Na2SO4, concentrated, and purified by flash chromatography
(2% EtOAc/petroleum ether) to give 22 mg (74%) of the des-iodo
methylthio compound 7 as a colorless oil: Rf 0.4 (5% EtOAc/petroleum
ether); 1H NMR (CDCl3, 270 MHz) δ 7.71-7.68 (d, J ) 7.2 Hz, 4 H,
ArH), 7.42-7.34 (m, 6 H, ArH), 5.27 (d, J ) 2.0 Hz, 1 H, H-1), 4.76
(d, J ) 11.2 Hz, 1 H, benzylic H), 4.69 (d, J ) 11.2 Hz, 1 H, benzylic
H), 4.47 (t, J ) 2.0 Hz, 1 H, H-2), 4.25 (dq, J ) 9.2, 6.0 Hz, 1 H,
H-5), 3.75 (s, 3 H, OCH3), 2.72 (t, J ) 9.2 Hz, 1 H, H-4), 3.65 (s, 3
H, OCH3), 3.49 (dd, J ) 9.2, 2.6 Hz, 1 H, H-3), 3.41 (s, 3 H, OCH3),
2.54 (s, 3 H, ArCH3), 2.27 (s, 3 H, ArSCH3), 1.27 (d, J ) 6.5 Hz, 3 H,
H-6), 1.04 (s, 9 H, C(CH3)3), 1.02-0.91 (m, 18 H, Si(CH2CH3)3), 0.70-
0.59 (m, 12 H, Si(CH2CH3)3); 13C NMR (CDCl3, 67.5 MHz) δ 152.8,
152.5, 144.1, 138.6, 135.7, 133.6, 129.5, 128.8, 127.5, 125.6, 104.7,
81.7, 72.5, 71.8, 68.5, 61.1, 60.5, 58.2, 56.9, 26.9, 19.3, 18.9, 18.2,
17.3, 7.0, 6.8, 5.2, 4.9; HRFABMS calcd for C46H74O8SSi3 (MNa)+
893.4309, found 893.4316.
2.55 (s, 3 H, ArCH3), 2.29 (s, 3 H, ArSCH3), 2.16 (s, 3 H, OCOCH3),
2.13 (s, 3 H, OCOCH3), 1.21 (d, J ) 5.9 Hz, 3 H, H-6), 1.04 (s, 9 H,
C(CH3)3); 13C NMR (CDCl3, 67.5 MHz) δ 170.2, 170.1, 152.9, 151.2,
143.6, 138.9, 135.7, 133.6, 129.5, 129.2, 127.5, 125.0, 100.6, 72.3,
68.5, 67.9, 61.2, 60.6, 58.2, 57.6, 26.8, 21.0, 19.3, 19.0, 17.5, 17.3;
HRFABMS calcd for C38H50O10SSi (MNa)+ 749.2791, found 749.2819.
5-((6-Deoxy-2,4-di-O-acetyl-3-O-methyl-r-L-mannopyranosyl)-
oxy)-3,4-dimethoxy-4-hydroxymethyl-6-methylthio-toluene (9). To
a plastic vial containing 8 (50 mg, 0.06 mmol) was added 1.5 mL of
a freshly prepared solution of buffered pyridinium hydrofluoride (10
mL THF, 5.7 mL pyridine and 2.1 g pyridinium hydrofluoride).20 The
reaction mixture was quenched after 3 h with 5 mL of saturated aqueous
NaHCO3. The aqueous layer was extracted with CH2Cl2 (3 × 5 mL),
and the organic layers were combined, dried over Na2SO4, concentrated,
and purified by flash chromatography (45% EtOAc/petroleum ether)
to give 29.6 mg (88%) of benzylic alcohol 9 as a white solid: Rf 0.2
1
(45% EtOAc/petroleum ether); H NMR (CDCl3, 270 MHz) δ 5.77
(dd, J ) 3.3, 1.9 Hz, 1 H, H-2), 5.54 (d, J ) 1.3 Hz, 1 H, H-1), 5.07
(t, J ) 9.9 Hz, 1 H, H-4), 4.72 (s, 2 H, benzylic H), 4.46 (dq, J ) 9.9,
5.9 Hz, 1 H, H-5), 3.96 (dd, J ) 9.9, 3.3 Hz, 1 H, H-3), 3.91 (s, 3 H,
OCH3), 3.84 (s, 3 H, OCH3), 3.43 (s, 3 H, OCH3), 2.60 (s, 3 H, ArCH3),
2.29 (s, 3 H, ArSCH3), 2.16 (s, 3 H, OCOCH3), 2.13 (s, 3 H, OCOCH3),
1.20 (d, J ) 5.9 Hz, 3 H, H-6); 13C NMR (CDCl3, 67.5 MHz) δ 170.2,
170.1, 153.1, 151.5, 143.7, 137.7, 129.1, 125.4, 100.5, 72.2, 68.6, 67.9,
61.4, 60.7, 57.8, 57.6, 21.0, 18.9, 17.5, 17.1; HRFABMS calcd for
C22H32O10S (MNa)+ 511.1613, found 511.1636.
4-((6-Deoxy-2,4-di-O-acetyl-3-O-methyl-r-L-mannopyranosyl)-
oxy)-2,3-dimethoxy-6-methyl-5-methylthio-benzaldehyde (10). To a
solution of dimethyl sulfoxide (24 µL, 0.34 mmol, 2.2 equiv) in 6 mL
CH2Cl2 at -78 °C was added oxalyl chloride (85 µL of a 2.0M solution
in CH2Cl2, 0.17 mmol, 1.1 equiv). After 6 min, the reaction mixture
was treated with a solution of 9 (75.5 mg, 0.15 mmol) in 2 mL of
CH2Cl2. The temperature was allowed to rise to -60 °C over 15 min,
followed by the addition of Et3N (100 µL, 0.75 mmol, 5.0 equiv). The
temperature was further raised to -40 °C over 45 min and the reaction
quenched with 10 mL of saturated aqueous NaHCO3. The aqueous layer
was extracted with CH2Cl2 (3 × 10 mL), and the organic layers were
combined, dried over Na2SO4, concentrated, and purified by flash
chromatography (30% EtOAc/petroleum ether) to give 65.7 mg (88%)
of aromatic aldehyde 10 as a white solid: Rf 0.45 (30% EtOAc/
petroleum ether); 1H NMR (CDCl3, 270 MHz) δ 10.45 (s, 1 H, CHO),
5.76 (dd, J ) 3.3, 2.0 Hz, 1 H, H-2), 5.72 (d, J ) 1.9 Hz, 1 H, H-1),
5.09 (t, J ) 9.9 Hz, 1 H, H-4), 4.37 (dq, J ) 9.9, 6.0 Hz, 1 H, H-5),
3.98 (dd, J ) 9.9, 3.3 Hz, 1 H, H-3), 3.97 (s, 3 H, OCH3), 3.88 (s, 3
H, OCH3), 3.43 (s, 3 H, OCH3), 2.78 (s, 3 H, ArCH3), 2.29 (s, 3 H,
ArSCH3), 2.18 (s, 3 H, OCOCH3), 2.14 (s, 3 H, OCOCH3), 1.19 (d, J
) 5.9 Hz, 3 H, H-6); 13C NMR (CDCl3, 67.5 MHz) δ 191.3, 170.0,
158.6, 155.6, 143.4, 141.2, 126.9, 125.4, 100.3, 72.0, 68.9, 67.7, 62.1,
60.9, 60.8, 57.7, 21.0, 18.8, 18.1, 17.4; HRFABMS calcd for C22H30O10S
(MNa)+ 509.1457, found 509.1462.
4-((6-Deoxy-2,4-di-O-acetyl-3-O-methyl-r-L-mannopyranosyl)-
oxy)-2,3-dimethoxy-6-methyl-5-methylthio-benzoic acid (11). To a
solution of aromatic aldehyde 10 (56.5 mg, 0.11 mmol,) in 10 mL CCl4
was added N-bromosuccinimide (45 mg, 0.24 mmol, 2.1 equiv) and
2,2′-azobisisobutyronitrile (2 mg, 0.01 mmol, 0.1 equiv). The reaction
mixture was heated to reflux for 90 min, cooled to room temperature,
and washed with brine (15 mL). The aqueous layer was extracted with
CH2Cl2 (3 × 10 mL), the organic layers combined, dried over Na2-
SO4, concentrated, and purified by flash chromatography (5% MeOH/
CHCl3 with 1% AcOH) to give 43 mg (74%) of aromatic acid 11 as a
white solid: Rf 0.2 (10% MeOH/CHCl3); 1H NMR (CDCl3, 270 MHz)
δ 5.78 (dd, J ) 3.3, 2.0 Hz, 1 H, H-2), 5.54 (d, J ) 1.9 Hz, 1 H, H-1),
4.99 (t, J ) 9.9 Hz, 1 H, H-4), 4.48 (dq, J ) 9.9, 5.9 Hz, 1 H, H-5),
3.99 (dd, J ) 9.9, 3.3 Hz, 1 H, H-3), 3.89 (s, 3 H, OCH3), 3.86 (s, 3
H, OCH3), 3.41 (s, 3 H, OCH3), 2.47 (s, 3 H, ArCH3), 2.31 (s, 3 H,
ArSCH3), 2.14 (s, 3 H, OCOCH3), 2.11 (s, 3 H, OCOCH3), 1.14 (d, J
) 5.9 Hz, 3 H, H-6); 13C NMR (CDCl3, 67.5 MHz) δ 171.3, 170.8,
152.8, 151.2, 144.6, 134.7, 126.2, 101.2, 77.7, 72.9, 69.3, 68.5, 61.3,
60.7, 57.4, 20.1, 20.0, 18.3, 17.4, 17.1; HRFABMS calcd for C22H30O11S
(MNa)+ 525.1406, found 525.1418.
2-((tert-Butyldiphenylsiloxy)methyl)-5-((6-deoxy-2,4-di-O-acetyl-
3-O-methyl-r-L-mannopyranosyl)oxy)-3,4-dimethoxy-6-methylthio-
toluene (8). 7 (284.2 mg, 0.32 mmol) was dissolved 30 mL of AcOH:
THF:H2O (3:1:1), and the reaction mixture was maintained at 25 °C
for 12 h. After dilution with 30 mL ice-cold water, the reaction was
quenched with the careful addition of solid NaHCO3 until the solution
was shown to be basic by pH paper. The aqueous layer was extracted
with EtOAc (3 × 40 mL), and the organic layers were combined, dried
over Na2SO4, and concentrated. The crude diol was azeotroped with
toluene (3 × 20 mL) and taken to the next step without further
purification.
To a solution of the diol in 20 mL of pyridine was added DMAP (8
mg, 0.06 mmol, 0.2 equiv), followed by acetic anhydride (155 µL, 1.63
mmol, 5.0 equiv). The reaction mixture was maintained at 25 °C for 3
h, followed by the removal of pyridine in vacuo. The residue was
dissolved in 30 mL of CH2Cl2 and washed with 1 N HCl (40 mL). The
aqueous layer was extracted with CH2Cl2 (3 × 30 mL), and the organic
layers were combined, dried over Na2SO4, concentrated, and purified
by flash chromatography (25% EtOAc/petroleum ether) to give 194
mg (82%, 2 steps) of diacetate 8 as a colorless oil: Rf 0.6 (30% EtOAc/
1
petroleum ether); H NMR (CDCl3, 270 MHz) δ 7.71-7.68 (m, 4 H,
ArH), 7.42-7.34 (m, 6 H, ArH), 5.80 (dd, J ) 3.3, 1.9 Hz, 1 H, H-2),
5.51 (d, J ) 1.9 Hz, 1 H, H-1), 5.08 (t, J ) 9.9 Hz, 1 H, H-4), 4.75 (d,
J ) 10.5 Hz, 1 H, benzylic H), 4.70 (d, J ) 10.5 Hz, 1 H, benzylic H),
4.49 (dq, J ) 9.9, 5.9 hz, 1 H, H-5), 3.96 (dd, J ) 9.9, 3.3 Hz, 1 H,
H-3), 3.78 (s, 3 H, OCH3), 3.65 (s, 3 H, OCH3), 3.43 (s, 3 H, OCH3),
4-((6-Deoxy-3-O-methyl-r-L-mannopyranosyl)oxy)-2,3-dimethoxy-

8420 J. Am. Chem. Soc., Vol. 122, No. 35, 2000
Biswas et al.
6-methyl-5-methylthio-benzoic Acid (12). To a solution of diacetate
11 (32.8 mg, 0.06 mmol) in 1.5 mL of MeOH:THF (2:1) was added
K2CO3 (27 mg, 0.19 mmol, 3.0 equiv). The reaction mixture was
maintained at 25 °C for 6 h and quenched by the careful addition of
acetic acid until the solution was shown to be acidic by pH paper. All
solvents were removed in vacuo, and the residual acetic acid was
removed by means of a toluene azeotrope (3 × 2 mL). The crude
product was purified by flash chromatography (5% MeOH/CHCl3 with
1% AcOH) to give 20.2 mg (74%) of the diol 12 as a white solid: Rf
0.3 (5% MeOH/CHCl3 with 1% AcOH); 1H NMR (CD3OD, 500 MHz)
δ 5.46 (d, J ) 1.5 Hz, 1 H, H-1), 4.47 (dd, J ) 3, 1.5 Hz, 1 H, H-2),
4.32 (dq, J ) 9.5, 6.0 Hz, 1 H, H-5), 3.92 (s, 3 H, OCH3), 3.85 (s, 3
H, OCH3), 3.69 (dd, J ) 9.5, 3 Hz, 1 H, H-3), 3.56 (t, J ) 9.5 Hz, 1
H, H-4), 3.54 (s, 3 H, OCH3), 2.51 (s, 3 H, ArCH3), 2.28 (s, 3 H,
ArSCH3), 1.98 (bs, 2 H, OH), 1.26 (d, J ) 6.0 Hz, 3 H, H-6); 13C
NMR (CD3OD, 125 MHz) δ 175.4, 151.9, 150.1, 144.9, 133.4, 125.8,
104.3, 81.4, 71.8, 71.2, 67.5, 61.0, 60.6, 56.6, 18.3, 17.7, 17.4;
HRFABMS calcd for C18H26O9S (MNa)+ 441.1195, found 441.1194.
Methyl 4,6-Dideoxy-4-(((2,6-dideoxy-4-S-(4-((6-deoxy-3-O-meth-
yl-r-L-mannopyranosyl)oxy)-2,3-dimethoxy-6-methyl-5-methylthio-
benzoyl)-4-thio-â-D-ribo-hexo-pyranosyl)oxy)amino)-2-O-(2,4-dideoxy-
4-(N-ethylamino)-3-O-methyl-r-L-xylo-pyranosyl)-â-D-glucopyrano-
side (4). To a solution of aromatic acid 12 (20 mg, 0.047 mmol) in 1.5
mL of THF were added Et3N (9 µL, 0.061 mmol, 1.3 equiv) and ClPO-
(OEt)2 (9 µL, 0.059 mmol, 1.25 equiv). The reaction was maintained
at 25 °C for 20 min and filtered through Celite to remove insoluble
Et3NuˆHCl. The solvent was removed in vacuo, and the residue was
purified by filtration through a short silica column (5% MeOH/EtOAc)
to give 16.0 mg (62%) of 13 which was used immediately for the next
coupling step.
To a solution of disulfide 14 (2.9 mg, 3 µmol) in 1 mL of
deoxygenated THF was added excess n-Bu3P (34 µL, 0.13 mmol, 45
equiv). Disulfide reduction was shown to be complete in 20 min by
TLC (3:1 Et2O:MeOH). The activated acid 13 (16.0 mg, 28 µmol, 9
equiv) in 0.5 mL THF and DMAP (5 mg, 0.036 mmol, 12 equiv) were
added, and the reaction mixture was maintained at 25 °C for 4 h. The
solvent was removed in vacuo, and the crude product was purified by
flash chromatography (6:1 Et2O:MeOH) to give 1.7 mg (32%, 2 steps)
of aryl tetrasaccharide 4 as a white solid: Rf 0.25 (3:1 Et2O:MeOH);
1H NMR (CD3OD, 500 MHz) δ 5.53 (d, J ) 1.5 Hz, 1 H, D-1), 5.38
(bs, 1 H, E-1), 5.07 (dd, J ) 10.3, 1.8 Hz, 1 H, B-1), 4.46 (t, J ) 2.0
Hz, 1 H, D-2), 4.28 (dq, J ) 9.5, 6.5 Hz, 1 H, D-5), 4.21 (m, 2 H, B-3
and A-1), 4.04 (dq, J ) 10.6, 6.5 Hz, 1 H, B-5), 3.92 (t, J ) 9.5 Hz,
1 H, A-3), 3.90 (s, 3 H, OCH3), 3.86 (s, 3 H, OCH3), 3.84 (t, J ) 10.0
Hz, 1 H, E-5axial), 3.70 (dd, J ) 9.5, 2.2 Hz, 1 H, D-3), 3.69-3.63 (m,
3 H, A-5, B-4 and E-5equatorial), 3.57 (t, J ) 9.5 Hz, 1 H, D-4), 3.54 (s,
3 H, OCH3), 3.51 (m, 4 H, E-3 and OCH3), 3.38 (s, 3 H, OCH3), 3.37
(dd, J ) 8.8, 7.5 Hz, 1 H, A-2), 2.74-2.65 (m, 3 H, E-4 and NCH2),
2.43 (s, 3 H, ArCH3), 2.40 (ddd, J ) 12.8, 2.9, 1.5 Hz, 1 H, E-2equatorial),
2.31 (s, 3 H, ArSCH3), 2.25 (t, J ) 9.5 Hz, 1 H, A-4), 1.97 (dt, J )
12.5, 2.9 Hz, 1 H, B-2equatorial), 1.77 (ddd, J ) 12.8, 10.3, 2.8 Hz, 1 H,
B-2axial), 1.46 (ddd, J ) 13.2, 11.0, 3.7 Hz, 1 H, E-2axial), 1.39 (d, J )
6.5 Hz, 3 H, B-6), 1.37 (d, J ) 6.0 Hz, 3 H, A-6), 1.26 (d, J ) 6.5 Hz,
3 H, D-6), 1.14 (t, J ) 7.0 Hz, 3 H, NCH2CH3); 13C NMR (CD3OD,
DEPT, 30 unique non-quaternary carbons) δ 104.4, 103.1, 100.7, 99.0,
81.3, 79.6, 76.2, 71.4, 70.5, 69.7, 68.6, 67.5, 61.6, 60.7, 56.7, 56.1,
55.4, 52.5, 42.0, 38.2, 34.3, 30.0, 27.4, 18.7, 18.3, 17.8, 17.4, 16.9,
13.7, 13.5; HRFABMS calcd for C39H64N2O17S (MH)+ 897.3724, found
897.3703.
buffer (10 mM MgCl2; 70 mM Tris‚HCl, pH 7.6; 5 mM DTT), and
5-10 units of T4 polynucleotide kinase, which were incubated at 37
°C for 1 h. The enzyme was denatured by heating the reaction mixture
at 90 °C for 5 min. The reaction mixture was mixed with 100 µL of
1× TE buffer (pH 7.2) and excess ATP was removed on a G25
sephadex spin column. The 5′-labeled oligomer (1 pmol) and 20 times
excess of the “cold” complementary strand were hybridized in a volume
of 100 µL containing 10 mM Tris‚HCl, pH 7.6, which was incubated
at 90 °C for 5 min and slow-cooled to 25 °C over 90 min. The DNA
sequence was d(GCTGAGTGNNNNGCACTGCC), where NNNN
represented a calicheamicin binding site (TCCT or TTTT).
Different concentrations of the inhibitor, native oligosaccharide 3
or the mono-iodo analogue 2 (10-6 M to 10-2 M) in DMSO were
incubated with 3 µΜ calicheamicin, also in DMSO, in a 20 mM NaCl,
30 mM Tris‚HCl buffer (pH 7.5), along with 1 fmol of labeled DNA
duplex. The final DMSO concentration was 10% (v/v). The reaction
mixture was incubated at 37 °C for 1 h. Calicheamicin-dependent DNA
cleavage was initiated by the addition of 1 µL of â-mercaptoethanol
(prewarmed to 37 °C) to attain a final concentration of 100 mM. The
cleavage reaction was allowed to proceed for 11 min at 37 °C, and
quenched by heating at 95 °C for 4 min. An equal volume of gel loading
buffer, containing 96% formamide, 7.5 mM EDTA, 0.05% (w/v)
bromophenol blue and 0.05% (w/v) xylene cyanol, was added to the
reaction mixture and the cleavage products separated on a 19%
denaturing polyacrylamide gel (containing 6.2 M urea).
Capillary Electrophoresis. CE experiments were conducted on a
Beckman P/ACE 5510 capillary electrophoresis system, with the anode
on the injection side and cathode on the detection side. Uncoated fused
silica capillaries (50 µm i.d.), obtained from Polymicro Technologies
Inc. (Phoenix, AZ), were used with a total length of 27 cm, the length
to the detector being 20 cm. The running buffer was 20 mM NaCl, 30
mM Tris‚Cl (pH 7.5). The capillary was initially rinsed with the
following reagents: 1 N HCl for 5 min, deionized water for 2 min, 0.1
N NaOH for 15 min, deionized water for 2 min, and finally, running
buffer for 5 min. The applied separation voltage was 10 kV, at 25 °C.
The capillary was rinsed with 0.1 N NaOH for 2 min, deionized water
for 2 min, and running buffer for 2 min between successive separations.
Fresh capillaries were prepared each day. The analytes were quantitated
by UV detection at 214 nm.
The nucleotide sequence of the 15-mer duplexes were 5′-CGT-
GCGNNNNGCACG-3′ where NNNN represents the various binding
sites (TCCT, TTTT, TATA, TGGT, and TAAT, respectively). The
sequence contains only one binding site per duplex. Equimolar amounts
of the two complementary strands were annealed for each duplex, by
heating to 95 °C for 5 min, followed by slow cooling to room
temperature over 3-5 h. The DNA and the ligand were initially
equilibrated at 25 °C in the running buffer (with 1% DMSO) for 15
min before injection into the capillary. To measure the affinity of the
synthetic oligosaccharides for the different DNA 15-mer duplexes,
successive samples were run with a fixed concentration of DNA and
varying concentrations of the ligands. Absorbances were plotted against
-1/time.39 For each ligand concentration, a run was performed in the
absence of the DNA, to calibrate the ligand concentration (peak heights
were found to be directly proportional to concentration). The difference
between total ligand concentration and the amount of free ligand in
the presence of DNA gave the concentration of the bound ligand. To
obtain the association constants by Scatchard analysis, the data were
fit to eq 1 using Kaleidagraph software (version 3.08d, Synergy
Software, Reading, PA).
Duplex DNA. Purified synthetic oligonucleotides were obtained from
the Princeton Synthesis Facility. Following dialysis to remove the TFA
salts, the strands were lyophilized and dissolved in deionized water.
Absorbances were measured, and the concentrations were determined
from the calculated extinction coefficients.38 The samples were stored
at 4 °C for further use.
[Ligand]bound/[Ligand]free ) Ka[DNA]total - Ka[Ligand]bound (1)
Acknowledgment. This work was supported by Grants GM
53066 and GM 42733 from the National Institutes of Health.
K.B. acknowledges a graduate fellowship from Bristol-Myers
Squibb. We thank Drs. George Ellestad and Wei-Dong Ding
of Lederle Laboratories for generously providing us with
calicheamicin.
Inhibition of Cleavage Assay.7 The DNA oligomers (20-mers) were
5′-end labeled in a 10 µL reaction mixture containing 1-5 pmol of
oligonucleotide, 5-7× excess of [γ-32P] ATP (3000 Ci/mmol), kinase
JA000519V
(38) Borer, P. N. CRC Handbook of Biochemistry and Molecular Biology;
Nucleic Acids, Vol. I; Fasman, G. D., Ed.; CRC Press: Cleveland, 1975; p
589.
(39) Mammen, M.; Colton, I. J.; Carbeck, J. D.; Bradley, R.; Whitesides,
G. W. Anal. Chem. 1997, 69, 2165-2170.