research papers
Acta Cryst. (2005). D61, 407±415 doi:10.1107/S090744490500082X 407
Acta Crystallographica Section D
Biological
Crystallography
ISSN 0907-4449
Crystal and solution structures of
7-amino-actinomycin D complexes with
d(TTAGBrUT), d(TTAGTT) and d(TTTAGTTT)
Eftichia Alexopoulos,
a
Elizabeth A. Jares-Erijman,
b
Thomas M. Jovin,
c
Reinhard
Klement,
c
Reinhard Machinek,
d
George M. Sheldrick
a
and Isabel
Uso
Â
n
e
*
a
Department of Structural Chemistry,
University of Go
È
ttingen, Tammannstrasse 4,
37077 Go
È
ttingen, Germany,
b
Departamento de
Quõ
Â
mica Orga
Â
nica, Facultad de Ciencias Exactas
y Naturales, Universidad de Buenos Aires,
1428 Buenos Aires, Argentina,
c
Department of
Molecular Biology, Max-Planck-Institute for
Biophysical Chemistry, 37077 Go
È
ttingen,
Germany,
d
Institute of Organic and
Biomolecular Chemistry, University of
Go
È
ttingen, Tammannstrasse 2, 37077 Go
È
ttingen,
Germany, and
e
ICREA at Instituto de Biologõ
Â
a
Molecular de Barcelona (IBMB±CSIC), Jordi
Girona 18-26, 08034 Barcelona, Spain
Correspondence e-mail: uson@ibmb.csic.es
# 2005 International Union of Crystallography
Printed in Denmark ± all rights reserved
The formation of the complex of 7-amino-actinomycin D with
potentially single-stranded DNA has been studied by X-ray
crystallography in the solid state, by NMR in solution and by
molecular modelling. The crystal structures of the complex
with 5
0
-TTAG[Br
5
U]T-3
0
provide interesting examples of
MAD phasing in which the dispersive component of the
MAD signal was almost certainly enhanced by radiation
damage. The trigonal and orthorhombic crystal modi®cations
both contain antibiotic molecules and DNA strands in the
form of a 2:4 complex: in the orthorhombic form there is one
such complex in the asymmetric unit, while in the trigonal
structure there are four. In both structures the phenoxazone
ring of the ®rst drug intercalates between a BrU±G (analogous
to T±G) wobble pair and a G±T pair where the T is part of a
symmetry-related molecule. The chromophore of the second
actinomycin intercalates between the BrU±G and G±BrU
wobble pairs of the partially paired third and fourth strands.
The base stacking also involves (A*T)*T triplets and Watson±
Crick A±T pairs and leads to similar complex three-
dimensional networks in both structures, with looping-out of
unpaired bases. Although the available NOE constraints of a
solution containing the antibiotic and d(TTTAGTTT) strands
in the ratio 1:1 are insuf®cient to determine the structure of
the complex from the NMR data alone, they are consistent
with the intercalation geometry observed in the crystal
structure. Molecular-dynamics (MD) trajectories starting from
the 1:2 complexes observed in the crystal showed that
although the thymines ¯anking the d(AGT) core are rather
¯exible and the G±T pairing is not permanently preserved,
both strands remain bound to the actinomycin by strong
interactions between it and the guanines between which it is
sandwiched. Similar strong binding (hemi-intercalation) of the
actinomycin to a single guanine was observed in the MD
trajectories of a 1:1 complex. The dominant interaction is
between the antibiotic and guanine, but the complexes are
stabilized further by promiscuous base-pairing.
Received 14 October 2004
Accepted 10 January 2005
PDB References: 7-amino-
actinomycin D±d(TTAGBrUT)
complex, orthorhombic,
1unm, r1umnsf; trigonal,
1unj, r1unjsf.
1. Introduction
Actinomycin D (AMD) is an anticancer antibiotic frequently
used as a component of the clinical VAC therapy (vincristine,
actinomycin, cyclophosphamide). It consists of two cyclo-
peptide rings connected to each other via a phenoxazone
moiety. The pharmacological action of AMD has been ratio-
nalized by its interaction with the double-stranded DNA helix:
the chromophore intercalates between the base pairs of DNA
and the cyclopeptides interact with the surface of the minor
groove by hydrogen bonding (Waring, 1981). As a result,
DNA transcription and replication is inhibited by blocking
polymerase translocation along the helix.
The binding of AMD to double-stranded DNA (dsDNA)
has been investigated extensively by X-ray crystallography
(Kamitori & Takusagawa, 1994; Robinson et al., 2001; Hou et
al., 2002), NMR (Brown et al., 1994; Chen et al., 1996; Lian et
al., 1996; Chou et al., 2002; Chin et al., 2003) and other spec-
troscopic techniques and is the subject of numerous thermo-
dynamic studies (Wilson et al., 1986; Chen, 1988; Chen & Sha,
2002; Chen et al., 2003; Qu et al., 2003). In general, the major
binding site involves 5
0
-GpC-3
0
sequences. The crystal struc-
tures of AMD or AMD derivatives complexed with dsDNA
show the phenoxazone ring intercalating into a GpC step, with
the cyclic pentapeptides anchored on the helix. Four hydrogen
bonds between threonines in AMD and guanine from the
DNA and another from the chromophore to the backbone
stabilize the complex. Additionally, hydrophobic contacts
strengthen the drug±DNA interaction (Jain & Sobell, 1972).
As the drug spans four base pairs, the ¯anking residues to
the guanines also in¯uence complex formation (Chen, 1988).
Binding and kinetic studies with oligomers containing
self-complementary and non-self-complementary ±XGCY±
sequences indicate differences in binding af®nities depending
on the nature of the adjacent bases. However, the DNA does
not always adopt a helical conformation, e.g. the AMD±
d(CGATCGATCG) complex consists of a slipped duplex with
the ApT dinucleotides looped out (Robinson et al., 2001). The
motif of bases looping out of the helix was also described for
5
0
-GXC/CYG-3
0
sequences (where X/Y are G/C or T/A)
interacting with AMD with the X/Y bases positioned
perpendicular to the stacked bases/chromophore (Chou et al.,
2002).
Three NMR (Lian et al., 1996; Chou et al., 2002; Chin et al.,
2003) and one X-ray (Hou et al., 2002) crystal structures of
non-complementary sequences bound to AMD have been
reported so far. In all cases base pairing between the
nucleotides seems to be forced. This results in mismatched
base pairs, which can be accompanied by strand slippage or
formation of hairpin loops. In the case of the sequence
d(GATGCTTC)
2
(Lian et al., 1996) investigated by NMR
methods, the non-complementarity was overcome by forma-
tion of T±T mismatches, resulting in a 2:1 DNA strand±drug
complex, with the chromophore intercalating between the
central GpC/CpG base pairs. The sequences used in two more
recent NMR studies (Chou et al., 2002; Chin et al., 2003)
formed hairpin loops leading to pairing between the bases of
the same strand and an overall strand±drug stochiometry of
1:1. In the only X-ray structure reported in the literature (Hou
et al., 2002) the DNA, which contains a CTG triplet sequence,
formed slipped duplexes involving mismatched T±T base
pairs.
AMD has also been shown to bind tightly and speci®cally to
single-stranded DNA sequences (Wadkins & Jovin, 1991;
Wadkins et al., 1996, 2000; Yoo & Rill, 2001; Chen et al., 2003,
2004). Fluorescence studies of 7-amino-actinomycin D
(7-AAMD; a ¯uorescent derivative of AMD shown to bind
dsDNA by Graves & Wadkins, 1989) with ssDNA indicate a
high sequence-dependence of the binding. In particular, there
seems to be an absolute requirement for guanine residues,
although not every guanine constitutes a potential binding
site. Thus, sequences containing a 5
0
-AGT-3
0
block were shown
to possess a high af®nity for the drug. AMD binding to ssDNA
has been shown to inhibit HIV reverse transcriptase and other
DNA polymerases (Rill & Hecker, 1996; Davis et al., 1998;
Imamichi et al., 2003), highlighting the pharmacological and
biochemical relevance of these investigations.
In a previous study, we proposed a model for the recogni-
tion of ssDNA sequences by actinomycin featuring the hemi-
intercalation of the chromophore between the two purine
residues (Wadkins et al., 1996). These and subsequent not
entirely consistent studies prompted the present investigation
of the interactions of 7-AAMD with non-complementary
5
0
-T
n
TAGTT
n
-3
0
(n = 2,3) by X-ray crystallography,
augmented by NMR data and molecular modelling. The
crystallographic studies were performed on the 7-AAMD±
5
0
-TTAG[Br
5
U]T-3
0
system.
1
H NMR measurements were
made on the closely related 5
0
-TTTAGTTT-3
0
sequence, which
unfortunately did not yield crystals. Both the 2:1 and 1:1
strand±drug complexes were investigated by molecular-
dynamics simulations.
2. Materials and methods
2.1. Crystallization
The DNA oligomers 5
0
-TTAGTT-3
0
and 5
0
-TTAG[Br
5
U]T-3
0
were obtained from Biotez. Stock solutions were prepared in
water without further puri®cation. 7-Amino-actinomycin D
was purchased from Sigma±Aldrich Chemie GmbH (Stein-
heim, Germany) and dissolved in water as a stock solution.
The concentrations of both DNA and 7-AAMD were deter-
mined by UV-absorption measurements.
Red hexagonally shaped crystals were obtained via the
hanging-drop method at 293 K from solutions containing
2.3 mM 7-AAMD, 2.3 mM 5
0
-TTAGTT-3
0
,2.3mM gadolinium
nitrate, 2.7 M ammonium sulfate, 0.05 M sodium/potassium
tartrate and 0.1 M sodium citrate buffer pH 5.6. Their
diffraction pattern at 100 K showed streaks in one direction,
so the crystals were unsuitable for data collection, but a unit
cell could be determined [a = b = 71.5 (2), c = 109.3 (3) A
Ê
,
= = 90, = 120
]. Seeds of these crystals were transferred
via cat whiskers to drops containing 2.5 mM 7-AAMD,
2.5 mM 5
0
-TTAG[Br
5
U]T-3
0
,2.5mM gadolinium nitrate, 2.7 M
ammonium sulfate, 0.05 M sodium/potassium tartrate and
0.1 M sodium citrate buffer pH 5.6, which were equilibrated at
293 K. Crystals appeared within a period of 2±3 d and could be
frozen without breaking. Long rectangular plates of the
orthorhombic form were obtained when seeding experiments
were performed under the same conditions but without
gadolinium nitrate in either solution, but it is possible that the
seeding was incidental to their formation. It is interesting to
note that no gadolinium sites were determined in the crystal
structure of the hexagonal form once solved. This may be a
consequence of the high sulfate concentration in the crystal-
lization buffer.
research papers
408 Alexopoulos et al.
7-amino-actinomycin D±DNA complexes Acta Cryst. (2005). D61, 407±415
Since a preliminary in-house room-temperature data set
(not reported here) of the original trigonal 5
0
-(TTAGTT)-3
0
complex was consistent with the ®nal re®ned structure of the
5
0
-(TTAG[Br
5
U]T)-3
0
complex, it is reasonable to assume that
the replacement of thymine by bromouracil did not signi®-
cantly change the structure.
2.2. Data collection and processing
Diffraction data were collected at 100 K for both crystal
forms of the 7-AAMD complex with brominated DNA
(Table 1). Four synchrotron data sets were collected at EMBL,
c/o DESY, Hamburg at different wavelengths from the same
trigonal crystal (dimensions 0.3 0.3 0.01 mm) using
MAR345 image-plate (BW7B) and MAR165 CCD (BW7A)
detectors and processed with DENZO and SCALEPACK
(Otwinowski & Minor, 1997). The space group was assigned as
P3
1
21 or P3
2
21 with dif®culty because the re¯ections with l
odd were systematically weak. A ¯uorescence spectrum was
recorded before the MAD measurements in order to locate
the bromine K edge accurately. As can be seen from the
statistics in Table 1, the R
merge
values for the 2.5±2.4 A
Ê
shell
collected and processed in two of the data sets are dramati-
cally higher than those for the 2.6±2.5 A
Ê
shell and were
eventually excluded. The very same crystal had also diffracted
to 2.5 A
Ê
on our in-house source, so that the stronger source
improved the signal-to-noise ratio but not the resolution of
diffraction from this crystal.
Two data sets were collected at 100 K from the same
orthorhombic crystal of dimensions 0.2 0.2 0.1 mm. The
®rst data set was collected with a Bruker rotating-anode
generator and SMART6000 CCD detector in-house to 2.4 A
Ê
resolution using Cu K radiation. These data were integrated
and scaled using the programs PROTEUM and SADABS
(Bruker AXS). The systematic absences indicated a C-centred
lattice and space group C222
1
. A further data set from the
same crystal was collected to 2.0 A
Ê
on the X11 beamline at
EMBL c/o DESY using a MAR165 CCD detector. However,
after processing with DENZO, SCALEPACK and XPREP
(Bruker AXS), the space group appeared to be P2
1
2
1
2
1
with
the same unit-cell parameters. The superlattice re¯ections
with h + k 6 2n were however very weak (about 1/5 of the
intensity of the re¯ections with h + k =2n). The C-centred cell
was used for the successful structure solution and re®nement
reported here.
2.3.
1
H NMR spectroscopy
1
H NMR spectra were acquired at 291 K on a Varian Inova
500 (500.17 MHz) spectrometer at the Department of Organic
Chemistry, University of Go
È
ttingen. AMD and
2
H
2
O
(99.996% deuterium) were purchased from Sigma.
The assignments of AMD in
2
H
2
O were based on literature
data (Angerman et al., 1972; Brown et al., 1994) and con®rmed
by DQF-COSY, TOCSY and NOESY (mixing time 350 ms).
Assignments of aromatic protons of T
3
AGT
3
were based on
comparison with literature data for mononucleotides, di-
nucleotides (Bovey, 1972) and a number of duplex DNA
molecules and were con®rmed by the DQF-COSY, TOCSY
and NOESY experiments. NOESY spectra were collected with
mixing times of 150, 250, 350, 400 and 500 ms.
3. Results and discussion
3.1. Structure determination of the orthorhombic form
The orthorhombic form was solved ®rst. After a number of
inconclusive attempts to solve the structure by SAD (single-
wavelength anomalous diffraction) in both P2
1
2
1
2
1
and C222
1
,
the structure was solved in C222
1
by treating the two wave-
lengths, although neither was close to the absorption edge, as a
pseudo-MAD (multiple-wavelength anomalous diffraction)
experiment using XPREP to generate the F
A
values.
SHELXD (Sheldrick et al., 2001; Schneider & Sheldrick, 2002)
research papers
Acta Cryst. (2005). D61, 407±415 Alexopoulos et al.
7-amino-actinomycin D±DNA complexes 409
Table 1
Data-collection statistics for both crystal forms.
Values in parentheses refer to the outer 0.1 A
Ê
resolution shell (except for the unit-cell parameters and occupancies, where they are standard uncertainties).
Orthorhombic form Trigonal form
Wavelength (A
Ê
) 1.5418 0.8110 0.8463 0.9196 0.9204 0.9050
X-ray source Rotating anode X11 BW7B BW7A BW7A BW7A
Resolution (A
Ê
) 2.4 2.0 2.5 2.4 2.4 2.5
Independent re¯ections 3039 5138 11493 12988 12964 11527
Space group C222
1
C222
1
P3
2
21 P3
2
21 P3
2
21 P3
2
21
Unit-cell parameters
a (A
Ê
) 51.56 (5) 51.59 (3) 71.48 (7) 71.22 (6) 71.17 (6) 71.22 (7)
b (A
Ê
) 70.89 (7) 70.96 (4) 71.48 (7) 71.22 (6) 71.17 (6) 71.22 (7)
c (A
Ê
) 39.66 (4) 39.45 (3) 109.32 (13) 108.41 (12) 108.35 (12) 108.40 (13)
Data statistics
hI/(I)i 53.8 (13.3) 19.9 (11.2) 26.4 (8.5) 16.8 (4.4) 16.5 (3.5) 21.3 (5.3)
R
merge
(%) 7.5 (32) 3.2 (12.3) 7.2 (28) 8.8 (73) 7.5 (70) 8.0 (48)
Completeness (%) 99.5 (99.9) 99.6 (98.8) 99.0 (94.3) 98.9 (92.1) 98.8 (92.1) 99.0 (93.3)
Redundancy 41.5 (29.3) 13.9 (13.5) 10.9 (9.4) 30.6 (3.1) 30.2 (2.9) 30.9 (2.8)
f
0
² (electrons) 0.77 1.14 1.50 6.14 8.50 3.33
f
00
² (electrons) 1.28 3.05 3.20 4.34 3.80 3.70
Br occupancy 0.85 (11) 0.57 (4) 0.69 (2) 0.59 (3) 0.50(3) 0.49 (3)
² From Sasaki (1989).
found four heavy-atom sites with high correlation coef®cients
(CC all/weak = 51.7/31.8). After phasing and density modi®-
cation with SHELXE (Sheldrick, 2002) using a solvent
content of 45%, further density modi®cation was performed
using DM (Cowtan & Main, 1996) with a solvent content of
30%. The contrast and connectivity output by SHELXE were
a little better for the original hand (0.390 and 0.867, respec-
tively) than for the inverted structure (0.335 and 0.846). A 2:1
DNA±AMD complex was traced by hand in the DM map. A
posteriori the map correlation coef®cient against the ®nal
re®ned structure was a little higher after SHELXE (0.794)
than DM (0.775).
The model was subjected to least-squares re®nement
against F
2
of the synchrotron data set with SHELXL97
(Sheldrick & Schneider, 1997). 5% of the re¯ections were
selected in thin shells as an R
free
set for cross-validation
(Bru
È
nger, 1992). In the ®rst ten re®nement steps, the complex
was built by hand using XFIT (McRee, 1999) with a stepwise
increase of the resolution starting from 3 A
Ê
. Geometrical 1,2-
and 1,3-distance restraints were taken from Parkinson (1996)
and additional 1,2- and 1,3-distance restraints were generated
with SHELXPRO (Sheldrick & Schneider, 1997) from the
structures of actinomycin D (Scha
È
fer et al., 1998) and
bromouracil (Sternglanz & Bugg, 1975). Planarity, chiral
volume and antibumping restraints and a Babinet solvent
model (Moews & Kretsinger, 1975) were employed, giving an
initial R
work
of 0.29 (R
free
= 0.32). A few water molecules were
added manually, selecting from the highest difference density
peaks those that were approximately spherical and made
reasonable contacts, and 12-parameter overall anisotropic
scaling (Uso
Â
n et al., 1999) was applied. During the re®nement,
it was noticed that the B values of the Br atoms were higher
than those of the other atoms of the corresponding base, so a
common occupancy was re®ned for them, giving values of
0.85 (11) for the in-house data and 0.57 (4) for the synchrotron
data that were collected later. Although a little bromouracil
may have been substituted by thymine by the use of seed
crystals, this is clear evidence of bromine loss as a result of
radiation damage (Ennifar et al., 2002). The ®nal re®nement
statistics are summarized in Table 2.
3.2. Structure determination of the trigonal form
The trigonal form was ®rst solved by molecular replacement
by a rather indirect route that involved an intermediate
lowering of the space-group symmetry using a model from the
orthorhombic structure and the program COMO (Jogl et al.,
2001). Shortly afterwards, the four-wavelength bromine-MAD
analysis led to a very clear independent solution of the
structure. F
A
values were derived by XPREP with re®nement
of the f
0
and f
00
values. SHELXD found 16 heavy-atom sites
research papers
410 Alexopoulos et al.
7-amino-actinomycin D±DNA complexes Acta Cryst. (2005). D61, 407±415
Table 2
Final re®nement statistics for the orthorhombic and trigonal structures.
Orthorhombic Trigonal
Space group C222
1
P3
2
21
Resolution range (A
Ê
) 20±2.0 20±2.5
R
work
0.236 0.258
R
free
0.279 0.296
Water molecules 13 14
Data/restraints/parameters 5127/2493/2356 11503/17002/8996
R.m.s. deviations from idealized geometry
Bond lengths (A
Ê
) 0.006 0.005
1,3-distances (A
Ê
) 0.019 0.015
Non-zero chiral volumes (A
Ê
3
) 0.004 0.003
Distances from restrained planes (A
Ê
) 0.205 0.159
Mean B factors (A
Ê
2
)
DNA atoms 21.9 30.1
7-AAMD atoms 15.2 49.8
PDB code 1unm 1unj
Figure 1
View of (a) the upper part of the complex (unit 1), consisting of strands 1
and 3 interacting with actinomycin molecule 1, and (b) the lower part
(unit 2), consisting of strands 2 and 4 interacting with actinomycin
molecule 2. Colour coding: guanine, blue; thymine, yellow; adenine, pink;
bromouracil, orange; sugar, grey; phosphate, red.
with high correlation coef®cients (all/weak = 59.2/49.3). 30
cycles of SHELXE density modi®cation assuming 30%
solvent were performed for both enantiomorphs (P3
1
21 and
P3
2
21). The contrast and the connectivity were lower for
P3
1
21 (0.321 and 0.843, respectively) than for P3
2
21 (0.382 and
0.885, respectively) con®rming the latter space group; the map
correlation coef®cient against the re®ned structure was 0.832
(after recalculation with an improved version of SHELXE).
Four copies of the core of the orthorhombic structure could be
®tted into the resulting map and further bases could be traced
by hand.
The model was re®ned in the same way as the orthorhombic
form, except that fourfold non-crystallographic symmetry
(NCS) restraints were also employed. Although the high-
energy remote synchrotron data set was used for the re®ne-
ment reported in Table 2, a common bromine occupancy was
re®ned for each of the four wavelengths. The occupancies,
shown in Table 1 in the order in which the data sets were
collected, again strongly indicate that the Br atoms were lost
from their original sites during data
collection as a result of radiation
damage. For both crystal forms, the data
sets at wavelengths for which f
0
of
bromine is least negative were
measured ®rst. Thus, the dispersive
MAD differences and the pseudo-SIR
differences caused by loss of bromine
during irradiation will have the same
phase and reinforce one another,
fortuitously strengthening the MAD
experiment and explaining the high-
quality maps obtained. This is especially
true for the orthorhombic structure,
where the very modest MAD dispersive
difference of 0.37 electrons is
enhanced by an average of 9.8 elec-
trons per site (calculated from the
change in re®ned occupancy) caused by
loss of bromine! For this reason, it is
strongly advisable to collect the wave-
length with the most negative f
0
(i.e. the
in¯ection point) last in a MAD experi-
ment.
3.3. Overall crystal structure
description
Both crystal modi®cations of the
7-AAMD±5
0
-TTAG[Br
5
U]T-3
0
complex
have very similar core structures, so the
orthorhombic form that was determined
to the highest resolution is discussed
here. The drug-to-strand stoichiometry
is 2:4. The four conformationally
different single strands are partly
paired, forming two antiparallel double
strands with extensive base stacking. In
each one of these double strands an actinomycin molecule
intercalates with the phenoxazone ring inserted between two
guanines (Fig. 1).
The ®rst double strand, shown in Fig. 1(a), is formed by the
single strands 1 and 3, which are connected by two stacked
base pairs, T106±A303 and BrU105±G304. The latter is an
unusual G±T wobble pair (Fig. 2a) in which N1 and O6 of the
guanine are hydrogen bonded to O3 and N3, respectively, of
the pyrimidine. The stacking continues with the phenoxazone
ring of the actinomycin and guanine G104, which pairs with
thymine T206 of a symmetry-related molecule, again in
wobble mode (Fig. 2b). In strand 1 the phosphate backbone
performs a sharp turn after G104 with a torsion angle
G104
of
137.7
. Adenine A103 and the thymines T102 and T101 do not
participate in the stacking; the angles of their planes to that of
the guanine G104 are 65.0, 27.8 and 60.7
, respectively. In this
way, they can interact with symmetry-related strands. T101
and T102 participate in (A*T)*T triplets (Fig. 2c), whereas
adenine A303 forms an A±T pair. The remaining two bases of
research papers
Acta Cryst. (2005). D61, 407±415 Alexopoulos et al.
7-amino-actinomycin D±DNA complexes 411
Figure 2
(a) The mismatched wobble base pair G304±BrU105 with hydrogen bonding between the N1 and
the O6 of the guanine with the O2 and the N3 of the bromouracil, respectively; (b) the mismatched
base pair G104±T206* showing similar interactions; (c) in a T*A*T triplet, A403 makes Watson±
Crick interactions with T102* and Hoogsten interactions with T101*; (d) -stacking of BrU305 to
the phenoxazone ring of actinomycin molecule 1 (* indicates symmetry equivalent).