THE JOURNAL OF BIOLOGICAL CHEMISTRY
6185–6191, 1996
© 1996 by The American Society for Biochemistry and Molecular Biology, Inc.
Different Conformations and Site Selectivity of
HO22 -Co(III)-Bleomycin A2 and Co(III)-Bleomycin A2
Bound to DNA Oligomers*
(Received for publication, September 21, 1995, and in revised form, December 13, 1995)
Qunkai Mao, Patricia Fulmer, Wenbao Li, Eugene F. DeRose, and David H. Petering‡
From the Department of Chemistry, University of Wisconsin, Milwaukee, Wisconsin 53201
Conformational properties of HO2 2 -Co(III)-bleomycin
A2 (Form I) and Co(III)-bleomycin (Form II) bound to
DNA oligomers offering either principal cleavage site
for the drug, d(GGAAGCTTCC)2 or d(AAACGTTT)2, have
been studied by NMR methods. Form I binds in slow
exchange to these oligomers. It retains most of its solu-
tion nuclear Overhauser effects (NOEs) upon binding to
either oligomer. Pyrimidinyl methyl protons from the
metal domain of the drug make an NOE connection with
a G5 2-amino proton on DNA. The bithiazole intercalates
between base pairs involving either C6 and T7 or T6 and
T7 of the two DNA molecules, according to NOE connec-
tions between the bithiazole protons and protons from
these bases and changes in the positions of their chem-
ical shifts. Form II also retains most of its solution NOEs
upon association with the first oligomer. However, in
contrast to Form I it binds to DNA in fast exchange on
the NMR time scale over the temperature range of
5–35 °C and does not break the degeneracy of the DNA
proton chemical shifts. No intermolecular NOEs be-
tween Form II and the 10-mer have been detected. Like-
wise, the major perturbation in chemical shift of the
histidine H2 and guanine G5 protons seen in Form I-
DNA adducts is absent in Form II-DNA. The association
constant of Form II with d(GGAAGCTTCC)2 in 20 mM
HEPES buffer at pH 7.4 and 25 °C is 1.7 3 105 M21, and 1.0
mol of Form II bind per mol of 10-mer.
Bleomycin (Blm)1 is an antitumor agent used to treat human
cancer. It is comprised of metal- and DNA-binding domains
linked by a peptide and a disaccharide unit (Fig. 1). Fe(II)-Blm
reacts with O2 and reductants to generate HO2 2 -Fe(III)-Blm,
which initiates single and double strand DNA cleavage and
base release (Burger et al., 1981; Petering et al., 1990; Sam et
al., 1994; Stubbe and Kozarich, 1987). Efficient reaction is due,
in part, to the binding of the drug to DNA through the bithia-
zole and positively charged R group (Chien et al., 1977). There
is definite site selectivity for the reactions of Fe-Blm with DNA
to produce strand cleavage or base release. The drug prefers
initial attack on the deoxyribose unit of the sequences 59-
GpT-39 and 59-GpC-39 (McLean et al., 1989; Steighner and
Povirk, 1990).
The redox chemistry of Fe-Blm in solution or bound to DNA
* This work was supported by National Institutes of Health Grant
22184. The costs of publication of this article were defrayed in part by
the payment of page charges. This article must therefore be hereby
marked “advertisement” in accordance with 18 U.S.C. Section 1734
solely to indicate this fact.
‡ To whom correspondence should be addressed: Dept. of Chemistry,
University of Wisconsin-Milwaukee, P.O. Box 413, Milwaukee, WI
53201.
1
The abbreviations used are: Blm, bleomycin; NOE, nuclear Over-
hauser effect; NOESY, NOE spectroscopy.
6185
Vol. 271, No. 11, Issue of March 15, pp.
Printed in U.S.A.
can be modeled closely by Co-Blm (Xu et al., 1992a, 1992b).
Both form dioxygenated complexes, which undergo dimeriza-
tion to reach a peroxyl species, HO2 2 -Co(III)-Blm (Form I) and
Co(III)-Blm (Form II) or the corresponding iron species. How-
ever, in contrast to HO2 2 -Fe(III)-Blm, its cobalt analog is stable
in the presence of DNA.
The three-dimensional conformations of the two Co-Blm
structures have been determined by NMR spectroscopy (Xu et
al., 1994). Both forms display extensively folded structures in
which the peptide linker is close-packed against the metal
Downloaded from www.jbc.org by guest, on June 14, 2010
domain. In addition, in Form I the bithiazole moiety is folded
back over the metal domain. These structures also bind tightly
to DNA, offering the opportunity to examine them complexed
with DNA as models for the related Fe-Blms (Xu et al., 1992b).
Recent communications have begun to reveal how metallo-
bleomycins bind to DNA. NMR data on the interaction of Form
I with DNA 10-mer, d(CCAGGCCTGG)2 showed that the com-
plex binds in slow exchange with DNA and that the bithiazole
moiety partially intercalated into the base pair structure be-
tween C6 and C7 directly adjacent to the GpC cleavage site (Wu
et al., 1994). Based on provisional NOE assignments, it was
proposed that the pyrimidine methyl group of Form I was in the
proximity of a 2-amino proton of G5 at the site. Another study
on the interaction of Zn-Blm with DNA concluded that Blm
binds in the minor groove in rapid exchange with oligomeric
DNA (Manderville et al., 1994; Manderville et al., 1995).
The finding that Form I and Zn-Blm bind differently to DNA
oligomers despite containing the same linker and DNA binding
domains suggests that metal domain properties play a key role
in the binding process. From NMR studies and associated mo-
lecular dynamic calculations, it appears that the metal do-
mains of Forms I and II Co-Blm bind Co(III) with opposite
chiralities (Xu et al., 1994). Therefore, it was hypothesized that
Forms I and II should associate differently with DNA. The
present study compares the interaction of Forms I and II with
a DNA 10-mer containing a GpC site to test this hypothesis. It
also examines the binding of Form I to another DNA oligomer
bearing a GpT site to determine whether the features of bind-
ing of Form I to DNA are common to different preferential
binding sites.
EXPERIMENTAL PROCEDURES
Preparation of DNA Oligomers—Syntheses of d(GGAAGCTTCC) and
d(AAACGTTT) were accomplished using a Millipore Cyclone Plus DNA
synthesizer and nucleotide amidites from Millipore as starting materi-
als. The method followed directions provided by the manufacturer ex-
cept as modified below (Millipore, 1989). After synthesis of 30 mmol of
either oligomer, each was cleaved from the support and fully depro-
tected by incubation for 48 h in 30% ammonium hydroxide. Purification
of the oligomers was done with a Millipore C8 reverse phase column
(250 3 4.6 mm). Solvent A contained 0.1 M sodium acetate, pH 6.5, and
5% acetonitrile. Solvent B was a mixture of 0.1 M sodium acetate and
65% acetonitrile. The gradient began with 100% solvent A and reached
100% solvent B in 45 min with a flow rate of 3.0 ml/min. Finally, to
6186 DNA-bound Cobalt(III)-Bleomycin
FIG. 1. Structure of bleomycin. Only the deprotonated amide ni-
trogen contributes to the ligand charge among the metal binding sites
(dotted). It is also assumed that the peroxyl group is protonated.
remove trityl groups, 6 ml of 80% acetic acid was incubated for 1 h with
oligomer. To separate the dimethoxytrityl groups from the oligomer, a
Sephadex G-15 desalting column was used with 20 mM ammonium
bicarbonate, pH 7.4, as the eluate buffer.
Preparation of Form I and Form II and Their DNA Adducts—Blenox-
ane, the clinical mixture primarily of Blm A2 and B2, was a gift of
Bristol Myers Co. Bleomycin A2 was isolated from Blenoxane as de-
scribed previously (Xu et al., 1994). Then, a one-to-one mixture of Form
I and Form II was made by reacting Blm A2 with Co(II)Cl2 with stirring
under aerobic conditions. The mixture was separated by high pressure
liquid chromatography using a 250 3 4.6-mm C18 reverse phase semi-
preparative column (Bio-Rad) with the following gradient. Solvent A
contained water and 0.2% (v/v) trifluoroacetic acid. Solvent B was
methanol. The gradient changed linearly from 70% solvent A and 30%
solvent B to 63% solvent A and 37% solvent B within 10 min at a flow
rate of 2.0 ml/min. When 200-ml fractions were collected, two peaks
appeared in the chromatogram representing Form II and Form I with
characteristic retention times of 7.8 and 8.9 min, respectively. In order
to remove trifluoroacetic acid and methanol from the isolated Form I or
Form II mixtures, the samples were lyophilized several times in water.
The Form I or Form II DNA adducts were made by mixing isolated
Co-Blm species with DNAa or DNAb.
NMR Spectroscopy—Proton NMR spectra were acquired at 500 MHz
on a GE GN500 NMR spectrometer. Phase-sensitive two-dimensional
spectra were acquired using TPPI-States phase cycling (Marion et al.,
1989). NOESY spectra were acquired in D2O with a mixing time of 250
ms at 15 and 25 °C (Jeener, et al., 1979; Macura et al., 1981). TOCSY
spectra were acquired in D2O at 15 °C with a mixing time of 80 ms using
a WALTZ-17y mixing scheme (Bax, 1989; Braunschweiler and Ernst,
1983). The proton sweep widths in the D2O spectra were set to 6024 Hz,
centered on the water resonance, with 2048 complex points acquired in
t2, and 256 complex t1 increments collected, corresponding to acquisi-
tion times of 340 and 42 ms in the two dimensions. The residual HOD
peak was suppressed with an on resonance DANTE pulse sequence
applied during the second half of the 1-s recovery delay between scans.
NOESY spectra in 90% H2O, 10% D2O were acquired with mixing times
of 200 ms at 5 and 15 °C. The water resonance was suppressed using a
“jump and return” read pulse, with an 80-ms delay between pulses,
corresponding to excitation maxima of 6 3125 Hz from the carrier
frequency centered on the water resonance (Plateau and Gueron, 1982).
A 45° phase shift was added to the normal mixing pulse-phase cycling
scheme to minimize radiation damping (Driscoll et al., 1989). The pro-
ton sweep widths were set to 12,048 Hz, with 2048 complex points
acquired in t2, and 256 complex t1 increments collected, corresponding
to acquisition times of 170 and 21 ms in the two dimensions. The delay
between scans was 1 s. Typically 96 scans were obtained per free
induction decay in all two-dimensional experiments. The NMR data
were processed on a Silicon Graphics INDY workstation using Felix 2.3
software (Biosym Technologies, Inc.). Sine-bell squared window func-
tions shifted by 60° followed by 2-Hz exponential multiplication were
applied in both dimensions, and the data were zero-filled to 2048 points
in t1 prior to Fourier transformation. Polynomial base-line correction
was applied to the F2 and F1 dimensions to obtain flat base lines.
One-dimensional spectra of samples in 90% H2O, 10% D2O were also
acquired using jump and return water suppression.
FIG. 2. Form I titration. One-dimensional proton NMR titration of
2 mM DNAa with Form I in 20 mM phosphate buffer and 0.1 M NaCl, pH
7.4, and 15 °C.
Downloaded from www.jbc.org by guest, on June 14, 2010
Association Constant of Co(III)-Blm A2 with DNA oligomers—Fluo-
rescence spectra of Co(III)-Blm A2 were recorded with an SLM AMINO
4800C spectrofluorimeter. Excitation at 300 nm produces an intense
emission from the bithiazole moiety at 350 nm, which is quenched upon
interaction with DNA. Titration of Form I, the DNA 10-mer, with DNAa
was carried out in 20 mM HEPES buffer at pH 7.4 and 25 °C. Plots of the
reduction in fluorescence intensity of the drug molecule as a function of
added oligomer were analyzed as described previously using the as-
sumption that Co(III)-Blm A2 binds to equivalent, noninteracting sites
on DNA (Chien et al., 1977). From the analysis one can determine K, the
association constant, and n, the number of base pairs/binding site,
using the following equation,
1/@Form II z DNAa# 5 1/nK @DNAo#@Form II# 1 1/n @DNAo# (Eq. 1)
in which [DNAo] represents the total concentration of DNA base pairs.
RESULTS
NMR Features of the Interaction of HO2 2 -Co(III)-Blm A2
(Form I) with d(GGAAGCTTCC)2 (DNAa)—A combination of
two-dimensional NMR techniques has been used to examine
the complexes of Form I and II bound to two DNA oligomers
incorporating either the GpC or GpT sites of cleavage (Petering
et al., 1990). The formation of 1:1 complexes of Form I and DNA
was monitored by following the disappearance of DNA imino
proton resonances. Both drug-DNA oligomer complexes were in
slow exchange on the NMR time scale as illustrated by the
observation of distinct resonances for free and bound DNA in
the imino region of the spectrum at intermediate drug-to-DNA
concentrations for the titration of d(GGAAGCTTCC) with Form
I (Fig. 2).
The binding of Form I lifts the 2-fold symmetry of the DNA
and thereby splits the degeneracy of its 1H NMR spectra. Fig. 3
illustrates this for the fingerprint region in the NOESY spec-
trum of the Form I-DNAa complex, showing a sequential NOE
pathway for each DNA strand. The sequential cross-peak be-
tween C6-H19 and T7-H6 protons is missing in this region,
suggesting that the drug has substantially perturbed this part
of the structure. In addition, the sequential cross-peaks be-
tween the G15 and T7 imino protons are missing in the H2O
NOESY spectrum, and these protons are significantly shifted
upfield. These results are consistent with the intercalation of
the drug between C6-G15 and T7-A14 (Wüthrich, 1986). A list
of intermolecular NOEs involving the bithiazole B5 and B59
protons in Table I confirms this interpretation by showing that
these protons interact with several protons in the C6-G15 and
T7-A14 base pairs. Together with the upfield shifts in position
 DNA-bound Cobalt(III)-Bleomycin 6187

FIG. 3. Fingerprint region of the 1H 250-ms NOESY spectrum of the 2 mM 1:1 Form I-DNAa complex in D2O, 20 mM phosphate (pH

Downloaded from www.jbc.org by guest, on June 14, 2010

7.4), and 0.1 M NaCl at 25 °C. a and b trace the sequential NOE pathways for each strand from 59 end to 39. The dashed circle indicates the missing
sequential NOE (C6-H19 to T7-H6).

TABLE I also signals that the metal domain interacts with the oligomer.
Bithiazoles and base intermolecular NOEs With the bithiazole intercalated between C6-G15 and T7-A14,
Complex NOE the metal domain-linker structure intact, and the metal do-
Form I-DNAa (10-mer) main in proximity to G5, it appears that Form I retains a
B59-C6 H5 6.97, 5.52 compact, folded structure when bound to DNA. Evidently,
B59-C6 NH2 6.97, 6.46 Form I can bind to a GpC site of cleavage without substantial
B59-T7 CH3 6.97, 1.70
reorganization of its solution structure, bringing the minor
B59-G15 NH 6.97, 12.3
B5-A14 H8 7.04, 7.52 groove cleavage site of the drug into apposition with the DNA
B5-A14 H19 7.04, 5.52 and metal domains.
B5-A14 H29, H20 7.04, 2.32/2.62 NMR Features of the Interaction of HO2 2 -Co(III)-Blm A2
B5-G15 H8 7.04, 7.44
B5-G15 H19 7.04, 5.58
(Form I) with d(AAACGTTT)2 (DNAb)—A similar picture is
Form I-DNAb (8-mer) emerging from an examination of the Form I-DNAb adduct.
B59-T6 NH 6.73, 13.25 The bithiazole B5 and B59 proton resonances are shifted to a
B59-T6 CH3 6.73, 1.36 higher field (8.11 and 7.77 ppm to 6.85 and 6.73 ppm, respec-
tively) upon binding of Form I to DNAb. Correspondingly, the
T6 and T7 imino protons are upfield shifted, consistent with
of the bithiazole B5 and B59 resonances from 8.11 and 7.77 ppm
the intercalation of the B5-B59 edge of the bithiazole between
to 7.04 and 6.97 ppm, respectively, these results confirm that
T6-A14 and T7-A13. Intermolecular NOEs between these parts
the sulfur-containing edge of the bithiazole intercalates be-
of the two molecules also support the binding of the bithiazole
tween C6-G15 and T7-A14. Similar data have been obtained
from an analysis of the Form I-DNAc adduct, leading to the to this region of DNAb (Table I). In addition, as with the Form
same interpretation (Wu et al., 1994). I-DNAa complex, the folded structure of the metal domain-
Form I itself retains almost all of the intramolecular NOEs linker remains in the DNA complex, but the B5-PCH3 NOE is
seen in the absence of DNAa at positions similar to those missing (Table II). Instead, the PCH3 protons make an NOE
previously reported (Table II) (Xu et al., 1994). Thus, in the connection with a G5 2-amino proton (Fig. 4B). Also, the imid-
DNAa adduct, extensive folding exists in the peptide linker azole H2 is highly shifted from 8.69 to 9.14 ppm as in Form
region, which generates a close-packed structure involving the I-DNAa. Clearly, Form I binds to both sites, GpC and GpT,
metal domain and linker similar to that in free Form I. How- with closely related conformations.
ever, the NOE between B5 and pyrimidinyl (P) methyl hydro- NMR Features of the Interaction of Form II, Co(III)-Blm A2
gens is missing. In the unbound form this NOE shows that the with d(GGAAGCTTCC)2—Form II binds to DNAa in faster
metal domain is folded over the DNA domain (Xu et al., 1994). exchange than Form I and does not break the 2-fold symmetry
Evidently, this interaction is lost upon intercalation of the of the DNAa NOESY spectrum (Figs. 5 and 6). In data not
bithiazole into DNA. shown, the fast exchange regions for the adduct existed over
The PCH3 protons at 2.55 ppm display NOE interactions the temperature ranges of 5–35 °C. Nevertheless, small pertur-
with an exchangeable base proton at 10.22 ppm. Although bations in chemical shifts of protons throughout both the drug
highly shifted, this proton is identified as a G5 2-amino proton and DNA structures indicate that a binding interaction does
on the basis of its NOE connections with G5 1-imino and G5 occur. The maintenance of the degenerate, uninterrupted NOE
2-amino hydrogens, which are its nearest neighbor protons pathway in Fig. 4 shows that Form II does not intercalate into
(Fig. 4A). Thus, the metal domain is closely associated with the the DNA base structure like Form I. Still, the B5 and B59
guanine base at the specific cleavage site and within a few protons are shifted upon interaction with DNAa. As with Form
angstroms of the binding site for the DNA domain. A large shift I, almost all of the NOEs seen in solution remain unperturbed
of imidazole H2 upon binding to DNA from 8.61 to 9.05 ppm in the DNA-bound structure (Table I). There is no NOE con-
6188 DNA-bound Cobalt(III)-Bleomycin
TABLE II
Drug-DNA complexes and CoBlm Intramolecular NOEs
Form I-DNAb Form I-DNAa Form II-DNAa
Complex NOEa Complex NOEa Complex NOEa
H2-VaCH3 9.14, 0.60 H2-TCH3 9.05, 1.16 H2-TCH3 8.79, 1.12
H2-Aa 9.14, 3.62 H2-VaCH3 9.05, 0.57 H2-VaCH3 8.79, 0.69
H2-Ab9 9.14, 3.05 H2-Aa 9.05, 3.65 H2-Aa 8.79, 3.31
H2-TCH3 9.14, 1.20 H2-Ab9 9.05, 3.20 H2-Hb9 8.79, 3.15
H4-Hb 7.57, 5.45 H2-ANH 8.79, 6.43
H4-Va 7.57, 1.17
H4-VaCH3 7.57, 0.60 H4-Va 7.52, 1.12 H4-Va 7.60, 1.13
Hb-G1 5.45, 5.32 H4-Hb 7.52, 5.42 H4-Hb 7.60, 5.51
H4-VaCH3 7.52, 0.57 H4-VaCH3 7.60, 0.69
Hb-G2 5.45, 4.27
Hb-G1 5.42, 5.36 Hb-G1 5.51, 5.23
Hb-G2 5.42, 4.10 Hb-G2 5.51, 4.08
VaCH3-Vb 0.60, 3.47 VaCH3-Vb 0.57, 3.41 VaCH3-Vb 0.69, 3.36
VaCH3-Vg 0.57, 3.66 VaCH3-Vg 0.69, 3.47
VaCH3-Vg 0.60, 3.70 VgCH3-Vb 0.85, 3.36
VgCH3-Vb 0.91, 3.47 VgCH3-Vb 0.87, 3.41 VNH-VgCH3 9.89, 0.85
VNH-VgCH3 8.68, 0.87 VNH-Vb 9.89, 3.36
TCH3-Ta 1.20, 4.08 VNH-Vb 8.68, 3.41
Ta-TCH3 4.11, 1.12
Ta-TCH3 4.02, 1.16 TNH-VaCH3 8.31, 0.69
TNH-VaCH3 9.23, 0.60 TNH-VaCH3 9.33, 0.57 TNH-TCH3 8.31, 1.12
TNH-Va 9.23, 1.17 TNH-Va 9.33, 1.12 TNH-Vb 8.31, 3.36

Downloaded from www.jbc.org by guest, on June 14, 2010

TNH-Vb 9.23, 3.47 TNH-Vb 9.33, 3.41 TNH-Va 8.31, 1.13
BNH-TCH3 8.51, 1.20 BNH-Ta 8.57, 4.02 BNH-Ta 8.04, 4.11
BNH-Ta 8.51, 4.08 BNH-TCH3 8.57, 1.16 BNH-TCH3 8.04, 1.12
BNH-Tb 8.57, 4.43
VNH-VgCH3 8.72, 0.91
VNH-Vb 8.72, 3.47
a
Ordered pair of chemical shifts in parts per million for the protons involving the NOE.

FIG. 5. Form II titration. One-dimensional proton NMR titration of

FIG. 4. Portions of the jump-return NOESY spectra (tm 5 200
ms) of the 2 mM Form I-DNA (1:1). Spectra were recorded in H2O/
D2O (90/10), 20 mM phosphate (pH 7.4), and 0.1 M NaCl. a, Form
I-DNAa, 15 °C; b, Form I-DNAb, 5 °C.
nection between PCH3 and the DNA minor groove as is present
in the Form I-DNA structures. Nor have any other intermolec-
ular NOEs been assigned. Furthermore, the highly shifted G5
2-amino proton and H2 proton resonances detected in Form
I-DNA adducts are not seen with Form II; instead, they are
only modestly perturbed from their solution values (8.58 – 8.79
and 6.75– 6.85 ppm, respectively for H2 and G5 2-amino). In-
terestingly, the DNA protons in the fingerprint region which
are most perturbed by the presence of Form II are those in the
GC base pairs at the ends of the 10-mer. Thus, the H6, H5, and
H19 protons of C9 are shifted downfield 0.05, 0.05, and 0.06
ppm, respectively, and, correspondingly, those of C10 are
shifted, 0.26, 0.29, and 0.09 ppm. Evidently, Form II-DNAa
2 mM of DNAa with Form II in D2O and 20 mM phosphate buffer, pH 7.4,
15 °C.
adopts a different conformation than Form I-DNAa-c, while
maintaining much, if not all, of the solution conformation of
this form of the drug.
Association Constant of Co(III)-Blm A2 with DNAa—The
strikingly different properties of binding of Form I and Form II
to DNAa prompted the measurement of the association con-
stant of Form II with DNAa. Following a method used previ-
ously, Form II was titrated with DNAa and the extent of their
interaction determined by the degree of quenching of the bi-
thiazole fluorescence upon binding to the oligomer (Chien et al.,
1977). Fig. 7a shows fluorescence emission spectra of Form II in
the absence and presence of DNAa. The quenching that was
observed is qualitatively similar to that observed when metal-
free Blm interacts with DNA (Chien et al., 1977). In Fig. 7, b
and c, a summary of the fluorescence titrations is presented,
 DNA-bound Cobalt(III)-Bleomycin 6189

FIG. 6. Fingerprint region of the 250-ms NOESY spectrum of 2

mM Form II-DNAa adduct (1:1) in D2O, 20 mM phosphate (pH 7.4)
at 15 °C.

Downloaded from www.jbc.org by guest, on June 14, 2010

together with the a sample plot of data from one titration based
on Equation 1. Its linearity validates the simple binding model
described in Equation 1. From four titrations the average and
standard deviation for the association constant and number of
base pairs per binding site were 1.6 6 0.2 3 105 M21 and 10.2
6 0.6, respectively. The latter figure indicates that a maximum
of one drug molecule associated with each 10-mer. Considering
the magnitude of this equilibrium constant and the concentra-
tions used in the NMR experiments under similar conditions,
virtually all of the Form II added to DNAa during the NMR
titration study became bound to the oligomer.
DISCUSSION
Detailed structural information is beginning to appear about
metallobleomycins bound to DNA. The structure of a Zn-Blm
DNA oligomer has been published (Manderville et al., 1995).
NMR information about a Form I adduct with 10-mer DNA has
also been described (Wu et al., 1994). Although a structure for
this complex is not yet in hand, comparative NMR studies offer
the opportunity to survey how variations in drug structure and
DNA sequences affect adduct formation.
Previous studies have suggested that the DNA and metal
domains of ON-Fe(II)-Blm, OC-Fe(II)-Blm, O2-Fe(II)- and O2-
Co(II)-Blm-DNA, and Fe(III)-Blm interact with DNA (Albertini
and Garnier-Suillerot, 1984; Antholine and Petering, 1979; FIG. 7. Fluorescence titration of DNAa by Form II. a, emission
Antholine et al., 1981; Chikira et al., 1989; Fulmer and Peter- spectra of Form II and Form II in the presence of a ratio of 1.25:1.00
ing, 1994). Features of the Form I-oligonucleotide structure DNAa to Form II. b, summary of four plots of the decrease in fluores-
cence emission of Form II as a function of added DNAa. Error bars
described above confirm the intimate interaction of both drug represent ranges of values. Conditions were as follows: 20 mM HEPES,
domains with DNA. Indeed, in each of the three Form I-DNA pH 7.4, and 25 °C. c, double reciprocal plot based on Equation 1 to
structures that have been examined, it is an element of the determine K and n.
metal domain that associates with the G5 part of the preferred
GpC or GpT site of cleavage. The DNA domain also intercalates bithiazole intercalates into the base pair structure of all three
between several stacks of base pairs, involving C6-T7, T6-T7, Form I-DNA structures (Table I and Wu et al. (1994)). The
and C6-C7, so the site of the primary interaction of bithiazole is finding that the PCH3 group in the metal domain is in close
not unique but does include a member of the cleavage site on proximity to G5 in each structure supports the close association
the 39 side (Wu et al., 1994). of the metal domain with DNA (Fig. 4). In the Form I complex
Knowledge of Form I-DNA conformation is derived from with d(CCAGGCCTGG)2, the assignment of this NOE to an
several types of NMR information. The intramolecular NOEs interaction between a G4 or G5 2-amino proton and PCH3
in Table II show that Form I retains its solution conformation protons could not be distinguished (Wu et al., 1994). The pres-
for the linker and metal domain when bound to DNAa and ence of the same NOE in all three Form I-DNA complexes now
DNAb. The intermolecular NOEs as well as the upfield chem- supports the identification of this NOE as a G5 2-amino proton-
ical shifts of the bithiazole protons and protons of the base PCH3 proton interaction in each complex.
pairs involving bases at positions 6 and 7 indicate that the A synthesis of these structural determinants shows that
6190 DNA-bound Cobalt(III)-Bleomycin
Form I exists in each DNA complex as a structure in which the
metal domain and linker are folded similarly to those found in
the absence of DNA. Using the unbound structure as a model,
the metal domain is in close association with G5, the peroxide
coordinated to Co occupies a position between G5 and T6 or C6,
and the bithiazole interacts between the base pairs at positions
6 and 7. Evidently, the metal domain as well as the DNA
domain interacts with DNA as previously hypothesized and
participates in site selectivity at the guanine residue of the
GpC and GpT pairs (Xu et al., 1994). This conclusion is but-
tressed by the finding that the chemical shift of the histidine
H2 proton, part of the metal domain, is markedly perturbed in
all of the DNA adducts. A role for the metal domain in site
selectivity has been previously hypothesized (Carter et al.,
1990; Kane and Hecht, 1994).
The Form I-DNA conformation is not unique among metal-
lobleomycins bound to DNA; a different one has been reported
for Zn-Blm-DNA (Manderville et al., 1994, 1995). On the basis
of the present experiments, the binding mode of Form II to
DNAa is also distinct from that of Form I. Furthermore, Form
II associates with DNA in a conformation that includes exten-
sive folding in the linker region that has not been observed for
Zn-Blm either in solution or when bound to DNA (Akkerman et
al., 1988, Manderville et al., 1994, 1995).
Fluorescence titration experiments have shown that Form II
binds to DNAa with an association constant comparable with
ones measured for bleomycin and similar to or 10-fold larger
than values for Cu-Blm under conditions of low ionic strength
(Chien et al., 1977; Povirk et al., 1981; Roy et al., 1981). At 1.7
3 105 M21, it is sufficiently large to insure that virtually all of
the detectable drug binds to DNA with the concentrations and
ionic strength used in the NMR experiments (Fig. 7). It is also
large enough to question whether the fast exchange behavior of
Form II in the NMR experiments seen in Fig. 5 results from
rapid exchange of Form II either bound to a particular site on
the 10-mer or free in solution.
An attractive alternative is that Form II slides along the
10-mer such that drug exchanges rapidly between sites on the
oligomer while bound to it. Other examples of sliding by ligands
binding to DNA oligomers have been hypothesized based on
NMR studies in which association between ligand and DNA
does not break the symmetry of the DNA NMR spectrum
(Leupin et al., 1986; Pelton and Wemmer, 1990). In the present
study, the finding that DNA proton chemical shifts were most
perturbed in the GC base pairs at the ends of the molecule may
also be consistent with this interpretation (data not shown).
The lack of any assignable intermolecular NOEs suggests that
sliding is very fast, thereby reducing the intensity of NOEs
related to any particular drug-site interaction. Because the fast
exchange binding of Form II to DNA is qualitatively similar to
that of Zn-Blm, the latter complex may also be able to slide
along the DNA molecule (Manderville et al., 1994, 1995).
The possibility that Form II can slide quickly along DNA
stands in contrast to the apparent static binding of O2-Co(II)-
Blm to DNA (Xu et al., 1992b). The latter permits such mole-
cules in close proximity along the DNA structure to exist for
extended periods without undergoing bimolecular oxidation
reduction to produce Form I and Form II (Xu et al., 1992a). It
is hypothesized that O2-Co(II)-Blm is folded, possibly like Form
I, when bound to DNA (Chikira et al., 1989). The hypothesis
that Form II may move along the DNA structure while bound
to it suggests that sliding by extended metallobleomycin struc-
tures might play a role in the mechanism by which sites are
selected for DNA cleavage by the drug.
A major question arises from these findings: why does Form
II not bind to DNA like Form I? Each contains the same DNA
domain structure with its bithiazole and positively charged
tail, which provides electrostatic stabilization for the drug-
DNA adduct. The net charge on the metal domain of Form I is
11; the charge of the metal domain of Form II is either 1 or 21,
depending on the charge on the ligand occupying its sixth
coordination position, probably either water or hydroxide.
Therefore, the positively charged metal domains of both forms
should interact favorably with the negative charge of the DNA
polymer and not be an obvious source of discrimination be-
tween them. A possible answer to this question relates to the
fact that Forms I and II may adopt different structures in
solution; the first has the bithiazole folded over the metal
domain, and the second has the bithiazole tail region extended
from the folded metal domain-linker region (Xu et al., 1994).
Apparently, these conformations substantially remain in their
DNA adducts. According to molecular dynamics calculations on
Form I and Form II, the ligand structure in each wraps oppo-
sitely about Co(III) to yield two chiral metal domains (Xu et al.,
1994). One result of such differential folding is that the two
metal domains cannot identically interact with DNA (Xu et al.,
1994). This conclusion is consistent with the differential behav-
ior of the H2 and G5 2-amino protons in the two DNA adducts
and the lack of an NOE connection between the PCH3 protons
Downloaded from www.jbc.org by guest, on June 14, 2010
and a G5 2-amino proton in the Form II-DNA complex. A
related outcome is that the position of the bulky disaccharide is
different in the two conformations. A third consequence of the
different chiralities is that only in Form I can the bithiazole
fold back upon the metal domain in Form I, placing the perox-
ide between them. In contrast, in Form II it is only possible to
fold the DNA domain over the metal domain such that the sixth
ligand position is on an outer face of the metal domain, not
between them, making it impossible for those two structures to
bind similarly to DNA. Therefore, it is suggested that interca-
lation by the bithiazole requires a specific metal domain con-
figuration, which permits a particular folded conformational
relationship to exist between the metal and DNA domains. The
reason for this requirement needs further study.
Acknowledgment—We appreciate the use of the spectrofluorimeter in
the laboratory of Dr. Sally Twining.
REFERENCES
Akkerman, M. A. J., Haasnoot, C. A. G., and Hilbers, C. W. (1988) Eur. J. Biochem.
173, 211–225
Albertini, J. P., and Garnier-Suillerot, A. (1984) Biochemistry 23, 47–53
Antholine W. E., and Petering, D. H. (1979) Biochem. Biophys. Res. Commun. 91,
528 –533
Antholine, W. E., Petering, D. H., Saryan, L. A., and Brown, C. E. (1981) Proc. Natl.
Acad. Sci. U. S. A. 78, 7517–7520
Braunschweiler, L., and Ernst, R. R. (1983) J. Magn. Reson. 53, 521–528
Bax, A. (1989) Methods Enzymol. 176, 151–168
Burger, R. M., Peisach, J., and Horwitz, S. B. (1981) J. Biol. Chem. 256,
11636 –11644
Carter, B. J., Murty, V. S., Reddy, K. S., Wang, S.-N., and Hecht, S. M. (1990)
J. Biol. Chem. 265, 4193– 4196
Chien, M., Grollman, A. P., and Horwitz, S. B. (1977) Biochemistry 16, 3641–3647
Chikira, M., Antholine, W. E., and Petering, D. H. (1989) J. Biol. Chem. 264,
21478 –21480.
Driscoll, P. C., Clore, M. G., Beress, L., and Gronenborn, A. M. (1989) Biochemistry
28, 2178 –2187
Fulmer, P., and Petering, D. H. (1994) Biochemistry 33, 5319 –5327
Jeener, J., Meier, B. H., Bachmann, P., and Ernst, R. R. (1979) J. Chem. Phys. 71,
474 – 492
Kane, S. A., and Hecht, S. M. (1994) Progress Nucleic Acid Res. Mol. Biol. 49,
313–352
Leupin, W., Chazin, W. J., Hyberts, S., Denny, W. A., and Wüthrich, K. (1986)
Biochemistry 25, 5902–5910
Macura, C., Huang, Y., Suter, D., and Ernst, R. R. (1981) J. Magn. Reson. 43,
259 –281
Manderville, R. A., Ellena, J. F., and Hecht, S. M. (1994) J. Am. Chem. Soc. 116,
10851–10852
Manderville, R. A., Ellena, J. F., and Hecht, S. M. (1995) J. Am. Chem. Soc. 117,
7891–7903
Mao, Q., Fulmer, P., Otvos, J., Antholine, W., DeRose, E., and Petering, D. (1995)
FASEB J. 9, A1349
Marion, D., Ikura, M., Tschudin, R., and Bax, A. (1989) J. Magn. Reson. 85,
393–399
 DNA-bound Cobalt(III)-Bleomycin
McLean, M. J., Dar, A., and Waring, M. M. (1989) J. Mol. Recognit. 1, 184 –192
Millipore Corp. (1989) Millipore Cyclone Plus DNA Synthesizer: Operator’s Man-
ual, 7–1-7–30
Pelton, J. G., and Wemmer, D. E. (1990) J. Am. Chem. Soc. 112, 1393–1399
Petering, D. H., Byrnes, R. W., and Antholine, W. E. (1990) Chem. Biol. Interact.
73, 133–182
Plateau, P., and Gueron, M. (1982) J. Am. Chem. Soc. 104, 7310 –7311
Povirk, L. F., Hogan, M., Dattagupta, N., and Buechner, M. (1981) Biochemistry
20, 665– 671
Roy, S. N., Orr, G. A., Brewer, C. F., and Horwitz, S. B. (1981) Cancer Res. 41,
4471– 4477
Sam, J. W., Tang, X-J., and Peisach, J. (1994) J. Am. Chem. Soc. 116,
5250 –5256
6191
Steighner, R. J., and Povirk, L. F. (1990) Proc. Natl. Acad. Sci. U. S. A. 87,
8350 – 8354
Stubbe, J., and Kozarich, J. (1987) Chem. Rev. 87, 1107–1136
Wu, W., Vanderwall, D. W., Stubbe, J., Kozarich, J. W., and Turner, C. J. (1994)
J. Am. Chem. Soc. 116, 10843–10844
Wüthrich, K. (1986) NMR of Proteins and Nucleic Acids, p. 269, John Wiley and
Sons, Inc., New York
Xu, R. X., Antholine, W. E., and Petering, D. H. (1992a) J. Biol. Chem. 267,
944 –949
Xu, R. X., Antholine, W. E., and Petering, D. H. (1992b) J. Biol. Chem. 267,
950 –955
Xu, R. X., Nettesheim, D., Otvos, J. D., and Petering, D. H. (1994) Biochemistry 33,
907–916
Downloaded from www.jbc.org by guest, on June 14, 2010