Structure, stability and function of 5-chlorouracil modified A:U and G:U base pairs

Article abstract of DOI:10.1093/nar/gks1316

The thymine analog 5-chlorouridine, first reported in the 1950s as anti-tumor agent, is known as an effective mutagen, clastogen and toxicant as well as an effective inducer of sister-chromatid exchange. Recently, the first microorganism with a chemically different genome was reported; the selected Escherichia coli strain relies on the four building blocks 5-chloro-2′- deoxyuridine (ClU), A, C and G instead of the standard T, A, C, G alphabet [Marlie re, P., Patrouix, J., Doering, V., Herdewijn, P., Tricot, S., Cruveiller, S., Bouzon, M. and Mutzel, R. (2011) Chemical evolution of a bacterium's genome. Angew. Chem. Int. Ed., 50, 7109-7114]. The residual fraction of T in the DNA of adapted bacteria was <2% and the switch from T to ClU was accompanied by a massive number of mutations, including >1500 A to G or G to A transitions in a culture. The former is most likely due to wobble base pairing between ClU and G, which may be more common for ClU than T. To identify potential changes in the geometries of base pairs and duplexes as a result of replacement of T by ClU, we determined four crystal structures of a B-form DNA dodecamer duplex containing ClU:A or ClU:G base pairs. The structures reveal nearly identical geometries of these pairs compared with T:A or T:G, respectively, and no consequences for stability and cleavage by an endonuclease (EcoRI). The lack of significant changes in the geometry of ClU:A and ClU:G base pairs relative to the corresponding native pairs is consistent with the sustained unlimited self-reproduction of E. coli strains with virtually complete T→ClU genome substitution.

Full text of DOI:10.1093/nar/gks1316

Published online 28 December 2012
Nucleic Acids Research, 2013, Vol. 41, No. 4 2689–2697
doi:10.1093/nar/gks1316
Structure, stability and function of 5-chlorouracil
modified A:U and G:U base pairs
Amritraj Patra1, Joel Harp1, Pradeep S. Pallan1, Linlin Zhao1, Mikhail Abramov2,
Piet Herdewijn2,3 and Martin Egli1,*
2
1Department of Biochemistry, School of Medicine, Vanderbilt University, Nashville, TN 37232, USA, Laboratory
of Medicinal Chemistry, Rega Institute for Medical Research, KU Leuven, Minderbroedersstraat 10, 3000
3
Leuven, Belgium and Institute of Systems and Synthetic Biology, University of Evry-Val-d’Essonne, 5 Rue
´
´
Henri Desbruere, 91030 Evry Cedex, France
`
Received October 31, 2012; Revised November 16, 2012; Accepted November 19, 2012
INTRODUCTION
ABSTRACT
Uracil analogs with halogen substitution at the 5-position
represent an important class of compounds with regard to
their mutagenic activity (1). Such analogs were first
synthesized in the 1950s as potential anti-tumor agents
(2,3). The 5-fluorouracil (FU) analog is a well-known
anti-cancer drug for treatment of human malignancies
(4). 5-Chlorouracil and 5-bromouracil (ClU and BrU,
respectively) are associated with inflammation and are
considered to be carcinogenic (5). 5-Iodouracil (IU) was
shown to have lethal and mutagenic effects on bacterio-
phage T4 (6).
The thymine analog 5-chlorouridine, first reported in
the 1950s as anti-tumor agent, is known as an
effective mutagen, clastogen and toxicant as well
as an effective inducer of sister-chromatid
exchange. Recently, the first microorganism with a
chemically different genome was reported; the
selected Escherichia coli strain relies on the four
building blocks 5-chloro-20-deoxyuridine (ClU), A, C
and G instead of the standard T, A, C, G alphabet
`
¨
[Marliere,P., Patrouix,J., Doring,V., Herdewijn,P.,
Halogenated uracil residues are expected to exhibit base
pairing properties in double-stranded nucleic acids that
are closely related to those of thymine, thus involving
complementary pairs with A or wobble pairs with G
that are stabilized via Watson–Crick (W–C) hydrogen
bonds. However, substitution at the 5-position of uracil
can substantially alter the physical (electronic) and
chemical properties of the nucleobase, as evidenced by
changes in the UV spectra and the values of the pKa (7).
The effects of incorporation of FU into DNA on the struc-
ture and dynamics of the latter have been studied quite
extensively, arguably as a result of the interesting pharma-
cological properties of this U analog as an anti-cancer
agent (8–10). Interest in other halouracil analogs was pri-
marily focused on their use as radiosensitizing agents in
human cancers. BrU and IU were shown to be more ef-
fective in killing tumor cells using ionizing radiation
(11,12). However, ClU, though an effective mutagen,
clastogen and toxicant, as well as an effective inducer of
sister-chromatid exchange, is not as sensitive to ionizing
radiation as other thymine analogs (3). The relative lack of
interest in ClU may explain the fact that a thorough in-
vestigation of the impact of the replacement of T with ClU
on the base pairing geometry and duplex conformation
using structural tools is presently lacking.
Tricot,S., Cruveiller,S., Bouzon,M. and Mutzel,R.
(2011) Chemical evolution of
a
bacterium’s
genome. Angew. Chem. Int. Ed., 50, 7109–7114].
The residual fraction of T in the DNA of adapted
bacteria was <2% and the switch from T to ClU
was accompanied by a massive number of muta-
tions, including >1500 A to G or G to A transitions
in a culture. The former is most likely due to wobble
base pairing between ClU and G, which may be
more common for ClU than T. To identify potential
changes in the geometries of base pairs and
duplexes as a result of replacement of T by ClU,
we determined four crystal structures of a B-form
DNA dodecamer duplex containing ClU:A or ClU:G
base pairs. The structures reveal nearly identical
geometries of these pairs compared with T:A or
T:G, respectively, and no consequences for stability
and cleavage by an endonuclease (EcoRI). The lack
of significant changes in the geometry of ClU:A and
ClU:G base pairs relative to the corresponding
native pairs is consistent with the sustained unlim-
ited self-reproduction of E. coli strains with virtually
complete T!ClU genome substitution.
*To whom correspondence should be addressed. Tel: +1 615 343 8070; Fax: +1 615 322 7122; Email: martin.egli@vanderbilt.edu
ß The Author(s) 2012. Published by Oxford University Press.
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by-nc/3.0/), which
permits non-commercial reuse, distribution, and reproduction in any medium, provided the original work is properly cited. For commercial re-use, please contact
journals.permissions@oup.com.

2690 Nucleic Acids Research, 2013, Vol. 41, No. 4
Marliere et al. (13) recently evolved genomic DNA
composed of the three canonical bases A, C and G and
the artificial base ClU in an Escherichia coli
strain lacking thymidylate synthase and requiring exogen-
ous T. Selection over 25 weeks in a specially developed
cultivation device yielded descendants that grew essen-
tially with only ClU instead of T. The DNA of adapted
bacteria contained 90% ClU and 10% T and this residual
fraction was forced to <2% by disrupting the trmA gene
for tRNA U54 methyltransferase.
The above pioneering study prompted us to analyze the
base pair geometries of ClU with A and G in more detail.
To date, no crystallographic study of a DNA duplex
comprising ClU:A or ClU:G base pairs has been
reported. A nuclear magnetic resonance (NMR) investiga-
tion by Theruvathu et al. (14) did not bring to light any
substantial difference in the geometries of ClU:A and T:A
base pairs. Also, the geometries of ClU:G and T:G wobble
pairs exhibited similar geometries (15). In another study,
the stacking patterns of a halogenated (F, Cl or Br)
uridine overhang at the 30-terminus of an octamer
RNA:DNA hybrid duplex were analyzed in presence of
rhodium or iridium hexamine salt (16). The crystal struc-
tures with ClU and BrU were similar in that the dangling
ends were located atop the terminal base pair, whereas FU
was ejected from the helical stack.
MATERIALS AND METHODS
Protein expression and purification
Bacillus halodurans genomic DNA was purchased from
American Type Culture Collection (ATCC, Manassas,
VA, USA). The Asp132!Asn mutant of BhRNase H
(Met58 to Lys196) was expressed in E. coli and purified
as described previously (18). The protein solution was
concentrated to 25 mg/ml.
Synthesis of 50-O-dimethoxytrityl-5-chloro-20-
deoxyuridine, 30-[(2-cyanoethyl)-(N,N-diisopropyl)]-
phosphoramidite and incorporation of ClU into
oligonucleotides
To a colorless solution of 5-chloro-20-deoxyuridine (19)
(600 mg, 2.28 mmol) in pyridine (20 ml), 4,40-
dimethoxytrityl chloride (930 mg, 2.75 mmol) was added
in one portion at room temperature (RT). The reaction
mixture was stirred for 12 h and turned to a yellow–orange
color. After the starting material had disappeared, the
reaction mixture was cooled in an ice bath, methanol
(1 ml) was added and the reaction mixture was
concentrated and co-evaporated twice with toluene. The
residue was dissolved in dichloromethane, washed with
H2O, dried over Na2SO4 and purified by column chroma-
tography on silica gel to yield 50-O-dimethoxytrityl-5-
chloro-20-deoxyuridine (1.08 g, 84%). This compound
(1.08 g, 1.91 mmol) was dissolved in dichloromethane (10
ml) and cooled in an ice bath. N,N-Diisopropylethylamine
(1.5 ml, 8.76 mmol) and 2-cyanoethyl N,N-diisopropyl-
chlorophosphoramidite (0.58 ml, 2.6 mmol) were added.
The reaction solution was stirred for 30 min at RT.
Upon completion, the reaction mixture was concentrated
and co-evaporated twice with toluene. The crude material
was purified by column chromatography on silica to
yield 50-O-dimethoxytrityl-5-chloro-20-deoxyuridine, 30-
[(2-cyanoethyl)-(N,N-diisopropyl)]-phosphoramidite (1.0g,
To compare the geometries of ClU:A and ClU:G base
pairs with those of the corresponding T:A and T:G pairs,
respectively, and to examine possible effects of these
artificial pairs on the conformation of duplex DNA, we
determined crystal structures of four Dickerson–Drew
Dodecamer (DDD) B-form duplexes containing
two ClU:A pairs ([d(CGCGAA{ClU}TCGCG)]2,
referred to as ClU7 here, and [d(CGCGAAT
{ClU}CGCG)]2; ClU8), four ClU:A pairs ([d(CGCGAA
{ClUClU}CGCG)]2; ClU7/8), or two ClU:G base pairs
([d(CGCGAATT{ClU}GCG)]2, ClU9) in complex with
Bacillus halodurans RNase H (BhRNase H) at resolutions
31
68%). P NMR (CDCl3, 25ꢀC): d = 149.08, 148.7. High-
˚
between 1.5 and 1.7 A. Crystals grown for the duplexes
resolution mass spectrometry (HRMS) calculated for
C39H46ClN4O8P, [MH+] 765.2794, found 765.2820.
All four ClU-modified, high-pressure liquid chromatog-
raphy-purified DDD oligonucleotides were purchased
from Trilink (San Diego, CA, USA). The DNAs were
annealed and mixed with the protein at 1.2:1 molar ratio
in the presence of 5 mM MgCl2.
alone were not of diffraction-quality and we therefore
resorted to using RNase H as a scaffold. DNA duplexes
act as inhibitors of RNase H (17) and the enzyme binds
double-stranded DNA non-specifically and without
perturbing the structure of the central T:A region of the
duplex (18), thus not compromising our geometric
analysis of ClU:A or ClU:G pairs (the latter being
adjacent to the four central T:A pairs).
Crystallization and structure determination
Our structures reveal nearly identical geometries of the
ClU:A and ClU:G base pairs compared with T:A and
T:G, respectively. In line with the structural similarities,
UV melting experiments of DNA duplexes with ClU pairs
uncovered only very minor consequences of the replace-
ment of T by ClU opposite A or G relative to the parent
duplexes with T:A or T:G pairs, respectively. These obser-
vations regarding structure and stability were mirrored at
the level of function. Thus, the restriction endonuclease
EcoRI did not display a preference in its ability to recog-
nize and cleave the natural recognition sequence GjAATT
C (‘j’ marks the cleavage site) in the native DDD
compared with duplexes featuring the modified recogni-
tion sequences GjAA(ClU)TC or GjAAT(ClU)C.
Crystallization experiments were performed by the sitting
drop vapor diffusion technique at 4ꢀC using a sparse
matrix screen (Hampton Research, Aliso Viejo, CA,
USA) (20). A quantity of 1 ml complex solution was
mixed with 1 ml of reservoir solution and equilibrated
against 40 ml reservoir wells. Crystals appeared in
droplets containing 0.2 M magnesium acetate, 0.1 M
sodium cacodylate (pH 6.5) and 20% (w/v) PEG 8000
within 2–3 days. Crystals were mounted in nylon loops,
cryo-protected in reservoir solution containing 20%
glycerol and frozen in liquid nitrogen. Diffraction data
were collected using either a Mar225 or Mar300 CCD
detector on the 21-ID F/D beam lines of the Life

Nucleic Acids Research, 2013, Vol. 41, No. 4 2691
Sciences Collaborative Access Team (LS-CAT) at the
Advanced Photon Source, Argonne National Laboratory
(Argonne, IL, USA). Data were integrated and scaled with
the program HKL2000 (21). The structures were
determined by the Molecular Replacement technique
using the program MOLREP (22,23) and the BhRNase
H structure with PDB ID code 3D0P (protein alone) as
the search model. Initial refinement was carried out with
the program REFMAC (24) and DNA duplexes were then
gradually built into the electron density, 2–3 bp at a time,
followed by further refinement. Manual rebuilding was
performed with the program COOT (25). Water molecules
and metal ions were added gradually and isotropic/
Translation Libration Screw-motion (TLS) refinement
was continued with the program PHENIX (26). A
summary of crystallographic parameters is provided in
Table 1. Helical parameters were calculated with the
program CURVES (27). Illustrations were generated
with the program UCSF Chimera (28).
Cleavage assays
Cleavage assays with EcoRI were conducted following a
previous protocol (29) with modifications. Briefly, 5 mM
32P-labeled DNA was incubated at 37ꢀC with EcoRI
buffer and 20 U of EcoRI (high fidelity; New England
Biolabs, Ipswich, MA, USA) in a volume of 10 ml, with
final concentrations of 50 mM potassium acetate, 20 mM
Tris–acetate, pH 7.5, 10 mM magnesium acetate and
1 mM dithiothreitol. Reactions were started by adding
EcoRI and were stopped by removing an aliquot of 1 ml
reaction mixture at various times and adding it to 9 ml of
20 mM EDTA (pH 9.0) in 95% (v/v) formamide, followed
by 20% PAGE analysis.
Coordinates
Final coordinates and structure factors for all four
dodecamers have been deposited in the Protein Data
Bank (http://www.rcsb.org.) The PDB ID codes are
4HUF (ClU7), 4HTU (ClU8), 4HUG (ClU7/8) and
4HUE (ClU9).
UV thermal melting
Melting of each oligonucleotide (ꢁ7.5 mM duplex concen-
tration) was performed in 10 mM sodium phosphate, pH
7.4 and 0.1 mM EDTA in the presence of 150 mM NaCl.
Absorbance versus temperature profiles were measured
with a Varian Cary 300 spectrophotometer at 260 nm
wavelength and 1-cm path length at heating or cooling
rates of 0.5ꢀC/min. Melting temperatures were determined
with Varian Cary UV Tm Analysis software and are
averages of the maxima of the first derivative of the
95-point smoothed curves from heating and cooling
experiments. The final Tm values are based on six inde-
pendent measurements. Hyperchromicities were determ-
ined by calculating the difference of absorbance between
high- and low-temperature baselines and dividing by the
absorbance of the low-temperature baseline.
RESULTS
Overall structures of the ClU-modified DDD:RNase
H complexes
To gain a better understanding of the potential structural
change due to the replacement of the thymine 5-methyl
group by chlorine, we selected the DDD B-form DNA
as a template and synthesized four modified DDDs with
ClU either located opposite A (ClU7, ClU8, ClU7/8) or
opposite G (ClU9). We used crystals of these modified
DDDs in complex with BhRNase H to study the
geometries of ClU:A and ClU:G pairs as the crystals of
the oligonucleotides alone were not of sufficient quality
for a detailed structural analysis. In the crystal of the
Table 1. Crystal data, data collection parameters and structure refinement statistics
Structure/duplex
ClU8
ClU9
ClU7
ClU7/8
Space group
P212121
Cell dimensions
˚
a (A)
64.08
64.76
116.47
64.36
64.81
116.62
64.19
64.75
116.47
63.83
64.64
116.29
˚
b (A)
˚
c (A)
Data collection
Wave length (A)
˚
0.97872
˚
Resolution (A)
35.88–1.49
1.52–1.49
33.28–1.56
1.59–1.56
33.22–1.69
1.72–1.69
43.23–1.64
1.67–1.64
˚
Outer shell (A)
Unique reflections (outer shell)
Completeness (outer shell) (%)
R-merge (outer shell)
77 514 (3752)
96.9 (95.1)
0.071 (0.824)
27.3 (2.5)
68 234 (3339)
97.3 (96.6)
0.082 (0.800)
19.8 (2.1)
54 372 (2620)
98.3 (97.3)
0.061 (0.675)
29.8 (2.9)
59 446 (2888)
99.7 (99.3)
0.072 (0.674)
26.7 (3.0)
I/s(I) (outer shell)
Refinement
Working set reflections
Test set reflections
R-work/R-free
74 662
3759
0.196/0.223
2244/896
476/2/4
30.18
68 154
3453
0.197/0.226
2203/897
523/2/4
33.43
54 301
2766
0.202/0.237
2192/833
375/2/4
27.80
59 370
2996
0.197/0.220
2215/833
341/2/4
27.96
No. of protein/DNA atoms
No. H2O/Mg2+/ligands
Average B-factors
˚
RMSD bonds (A)
0.009
1.4
0.009
1.3
0.008
1.3
0.008
1.4
RMSD angles (ꢀ)

2692 Nucleic Acids Research, 2013, Vol. 41, No. 4
Figure 1. Overall structure of the BhRNase H:ClU8-DDD (ClU8) complex. The asymmetric unit contains two independent complexes, both con-
sisting of a single RNase H molecule bound to a modified DDD duplex. Protein chains in all four structures are labeled a (complex 1) and b
(complex 2) and duplex strands are labeled c and d (complex 1) and e and f (complex 2), with nucleotides numbered 1–12 (c/e) and 13–20 (d/f).
(A) The duplex in the first complex (colored green) is fully resolved in the electron density map and is contacted by three RNase H molecules
(colored in white, gray or dark gray, whereby symmetry mates are marked with hash symbol). (B) The duplex in the second complex (colored in
beige) is contacted by two RNase H molecules (colored in pink or magenta) and four C:G pairs at one end of the duplex are not visible in the
electron density map. (C) The superimposition of the two independent complexes illustrates that the ClU8-DDDs exhibit similar conformations and
virtually identical orientations of the original RNase H molecule relative to the DNA duplex. ClU residues are labeled and chlorine atoms are
highlighted as light green spheres, selected water molecules are shown as cyan spheres, RNase H amino acids interacting with the DNA are labeled
and the so-called phosphate (P-) binding site is marked by an arrow.
native DDD bound to RNaseH studied earlier (18),
protein–DNA contacts in the complex were limited to
the backbone of the C:G tracts (underlined) at both
ends: CGCGAATTCGCG. Thus, we expected base
pairing in the central GAATTC region where we
replaced T opposite either A or G by ClU not to be
affected by binding of RNase H (Figure 1).
DNA is only partially resolved and either three or four
base pairs at one end of the duplex are missing in the
individual complex crystals (Figure 1B and C). Similarly,
two (ClU7 crystal structure) and three N-terminal amino
acids (ClU7/8 crystal structure) in the RNase H molecules
from the second complex could not be resolved in the
electron density maps. Examples of the quality of the
final electron density are shown in Figure 2 and
Supplementary Figures S1–S3.
Inspection of the lattice interactions in the structures of
the complexes reveals the origin of the diminished order of
four base pairs in the second duplex. Three RNase H
molecules interact with the duplex from the first
complex, whereby two proteins cradle one end and a
loop entailing residues T90, G91 and E92 of a third is
stacked against the terminal base pair at the other end
(Figure 1A). In contrast, the second duplex only exhibits
interactions with RNase molecules at one end, with the
other not being stabilized by protein contacts and jutting
out into the solvent (Figure 1B). The comparison between
the orientations of RNase H molecules around the two
independent duplexes reveals similar positions of the
˚
Crystals diffracted to resolutions between 1.5 and 1.7 A
and belong to space group P212121 (Table 1). The struc-
tures were solved by the molecular replacement technique
using the protein portion of the complex between RNase
H and the native DDD (18) as the search model. All four
crystals of the complexes with ClU-modified DDDs are
isomorphous (Table 1). However, unlike in the structure
of the complex between RNase H and the native DDD
where the duplex sits on a dyad and the asymmetric unit
contained two protein molecules and two DNA single
strands, crystals of the complexes with ClU-modified
DDDs feature two independent complexes consisting of
RNase H bound to a duplex (Figure 1A and B). The
first of these displays clear electron density around all nu-
cleotides, whereas in the second, the density around the

Nucleic Acids Research, 2013, Vol. 41, No. 4 2693
RNase H and A or ClU/T nucleotides from the central
tetramer.
Geometry of ClU:A and ClU:G base pairs
A central question that we wanted to address with our
structural studies is whether there are any changes in the
pairing modes and/or geometries of ClU:A and ClU:G
base pairs relative to the native T:A and T:G pairs,
respectively. Analysis of the duplexes containing either
two or four ClU:A pairs reveals that they are of the
standard W–C type with formation of two hydrogen
bonds. Comparison between the geometries of ClU:A
and T:A pairs by superimposing the ClU7 and ClU8
duplexes (Figure 4A) or the ClU7/8 and native DDD
(31) duplexes (Figure 4C) demonstrates that replacing
the methyl group of T by chlorine is of little consequence.
Except for the methyl group carbon and chlorine, whose
positions deviate somewhat as a result of the longer C-Cl
bond relative to C-CH3, the ClU:A and T:A pairs neatly
overlap. Similarly, like T:G the ClU:G pair adopts the
familiar wobble geometry, with G and ClU being shifted
into the minor and major grooves, respectively, under for-
mation of two hydrogen bonds (Figure 4B). As with the
ClU:A and T:A pairs, the superimposition of the ClU9
duplex and a DDD featuring T:G pairs (32) illustrates
the nearly identical geometries of the two pairs.
Moreover, calculated geometric parameters for these
duplexes (27), including rise, twist, x- and y-displacement,
etc. (see the Supplementary Material) confirm that
chlorine attached at the 5-position of thymine in place
of a methyl group does not trigger a substantial difference
in either pairing behavior or base pair geometry.
The similarities observed at the level of individual base
pairs extend to the overall conformations of the
dodecamer duplexes. As mentioned above, RNase H mol-
ecules in the crystal structures of the complexes with
ClU-modified DDDs contact the duplexes exclusively in
the CG portions, either at both ends (duplex 1, Figure 1A)
or at only one end (duplex 2, Figure 1B). As a result, a
hallmark of the DDD, the narrow A-tract minor groove
remains largely unaffected by the interactions with the
protein (Figure 5). Introduction of T:G mismatches at
the border of the A-tract results in a wider minor groove
in the ClU9 and G:T DDDs at the location of the
mismatch pairs. However, replacement of T by ClU has
no obvious effect on the change in groove width. The
depth of the minor groove along the entire dodecamer
exhibits much less variation compared with the width
and neither G:T mismatch pairs nor ClU:A/G pairs
cause any significant changes relative to the native DDD
(Figure 5).
Figure 2. Quality of the final electron density. Fourier (2Fo–Fc) sum
electron density drawn at the ꢁ1.0 s threshold (A) around the central
ApApTpClU tetramer in the ClU8 duplex, and (B) around base pair
A5:ClU20 (top) and base pair ClU8:A17 (bottom). Atoms are colored
beige, red, blue, orange and light green for carbon, oxygen, nitrogen,
phosphorus and chlorine, respectively.
protein that binds a phosphate group at its P-binding
site but a considerable shift between the second protein
molecules contacting the same end of the two duplexes
(Figure 1C).
Protein–DNA interactions in the ClU-DDD:RNase
H complex structures
The two complexes per asymmetric unit in the four crystal
structures exhibit similar interactions between RNase H
and the ClU-modified DDD (Figure 1A, white RNase H a
and green duplex; Figure 1B, pink RNase H b and beige
duplex). In both cases, the phosphate group of C3 is
lodged at the phosphate-binding pocket (Figure 3) that
harbors a phosphate of a DNA nucleotide separated by
2 bp from the scissile RNA phosphate in the structure of
an RNA:DNA hybrid bound to BhRNase H (30). In the
crystal structure of the complex between RNase H and
the native DDD, the phosphate of the terminal G12 sits
at the active site and thus mimics a phosphate from the
RNA strand (18). The phosphate of residue G4 from the
paired strand is anchored at the phosphate-binding
pocket. Compared with these interaction modes between
RNA:DNA hybrid and native DDD and RNase H,
ClU-modified duplexes thus exhibit a less intimate inter-
action with the protein as only phosphates from one
strand are contacted by amino acids (P3 by T104, S147
and T148 and P4 by W139; Figure 3). As a consequence,
only one Mg2+ ion (MgB) is observed at the protein active
site, whereby the ion coordination sphere comprises D71,
E109 and four water molecules (Figure 3). In addition to
the protein–DNA contacts involving phosphate groups,
four residues from
a symmetry-related RNase H
molecule interact with the terminal base pair of the first
duplex. R126 and P177 stack onto the nucleobases of G24
and C1, respectively, and I64 and L179 form hydrophobic
interactions with the base and sugar moieties, respectively,
of nucleotide C1 (Figures 1A and 3). In the case of the
second complex where the symmetry-related RNase H
Thermodynamic stability of ClU-modified duplexes
To examine whether replacement of T by ClU affects
duplex stability, we carried out UV melting studies with
the native DDD, a DDD with T:G mismatches and all
four ClU-modified DDDs (Table 2). The melting tempera-
tures for duplexes with two (ClU7, ClU8) or four ClU:A
pairs (ClU7/8) vary only slightly and are similar to the
Tm of the native DDD. Insertion of T:G mismatches is
molecule adopts
a somewhat different orientation
(Figure 1A, gray RNase H b# versus Figure 1B,
magenta RNase H a#), amino acids establishing contacts
with the terminal base pair include K143, N170 and T173
(Figure 1B). However, neither duplex 1 nor duplex 2 in the
four crystal structures exhibits interactions between

2694 Nucleic Acids Research, 2013, Vol. 41, No. 4
Figure 3. Interactions between RNase H and ClU modified DDDs. The original RNase H molecule (white ribbon) binds phosphates from one DNA
strand; the phosphate of C3 (P3) is lodged at the phosphate-binding pocket and W139 forms a hydrogen bond to the phosphate of G4. The active
site that normally harbors the scissile phosphate from the RNA strand (18) remains unoccupied and only one of the two Mg2+ ions (purple sphere) is
present, with some water molecules (cyan spheres) taking the place of phosphate oxygens. Amino acid side chains from a symmetry-related RNase H
molecule (I64, L179, P177 and R126) interact with the terminal C1:G24 base pair. Hydrogen bonds and Mg2+ coordination sphere are indicated with
thin solid lines and two glycerol molecules trapped at the protein–DNA interface are highlighted with carbon atoms colored in black.
Figure 4. Comparison between the geometries of ClU:A and ClU:G and those of T:A and T:G base pairs, respectively. (A) Superimposition of the
d(ClU7pT8pC9):d(G16pA17pA18) (light blue carbon atoms) and d(T7pClU8pC9):d(G16pA17pA18) (gray carbon atoms) trimer portions from the
ClU7 and ClU8 structures, respectively. (B) Superimposition of the d(T8pClU9pG10):d(C15pG16pA17) (gray carbon atoms) and
d(T7pT8pC9):d(G16pA17pA18) (light blue carbon atoms) trimer portions from ClU9 and a G:T mismatch-containing DDD structure [PDB ID
113D (32)], respectively. (C) Superimposition of the central hexamers from the ClU7/8 structure (gray carbon atoms) and the native DDD [PDB ID
436D (31)] (light blue carbon atoms). Chlorine atoms are highlighted in green, hydrogen bonds are indicated with thin solid lines (omitted in panel C
for clarity), and fluorine atoms of two 20-deoxy-20-fluoroarabino-Ts in the reference structure are highlighted in purple (panel C).
accompanied by a steep drop in the stability that is similar
EcoRI cleavage assays with ClU-modified DDDs
for the duplex featuring two ClU:G pairs (ClU9). Overall,
The EcoRI endonuclease recognizes the self-com-
plementary hexamer 50-GjAATTC-30:30-CTTAAjG-50
and cleaves between G and A (j) under formation of
these data argue against an either stabilizing or
destabilizing effect of the ClU nucleotide analog in
B-form DNA.

Nucleic Acids Research, 2013, Vol. 41, No. 4 2695
sticky ends with 50-overhangs. We carried out cleavage
assays with the restriction enzyme and ClU-modified
DDDs, as well as the native DDD and a DDD with
T:G mismatch pairs as reference duplexes (Figure 6). As
expected, T:G and ClU:G mismatches abrogated cleavage.
Conversely, the native DDD and all three modified
duplexes with ClU:A pairs were cleaved by EcoRI
without obvious effects of the ClU modification in terms
of the time course of cleavage.
opportunity to replace it in vivo by starvation and exogen-
ous introduction of unnatural alternatives (33).
Incorporation of 5-halogenopyrimidines into DNA was
established decades ago (34). Among the analogs with
fluoro-, bromo-, chloro- or iodo-substituents, ClU
exhibits the closest likeness to T (14), is readily converted
to the nucleoside triphosphate in the cell (35) and lacks the
photolabile behavior of BrU and IU (36). Starting from an
E. coli strain lacking thymidylate synthase and furnishing
exogenous ClU instead of T, 25 weeks of selection in a
cultivation device yielded a strain with A, G, C and 90%
ClU and 10% T in its genome (13). Through additional
disruption of the tRNA U54 methyltransferase gene, the
ClU content was further reduced to <2%. This T!ClU
transliteration was accompanied by more than 1500 A to
G and G to A transitions in a particular culture, whereby
the former was about twice as common. The frequency of
these transitions suggests that ClU is prone to mispairing
with G, although it is not clear whether ClU:G resembles
the T:G pair in the wobble configuration with two
hydrogen bonds or whether the mismatch pair exhibits a
different hydrogen bonding pattern as a consequence of
subtle changes in pKa, dipole moment and/or hydration of
the chlorouracil base compared with thymine (15).
DISCUSSION
Our investigation of the structure, stability and function
of DNA duplexes with ClU in place of T was motivated by
the recently demonstrated genome-wide transliteration of
T with ClU in E. coli by combining tight metabolic selec-
tion and long-term automated cultivation of bacterial
populations (13). Among nucleobases, only thymine is
unique to DNA and the fact that its metabolism is
separated from RNA biosynthesis provides an
We selected the DDD B-form DNA as a template to
analyze the base pairing behavior of ClU opposite either A
or G by X-ray crystallography. Because crystals of the
DNAs alone did not diffract to high resolution, we
decided to determine the structures of their complexes
with BhRNase H. We previously found that crystals of
complexes between the endonuclease and the native
DDD or chemically modified DDDs diffract X-rays to
˚
resolutions of around 1.5 A (18,29,37). Indeed, all
crystals of complexes with ClU-modified DDDs diffracted
˚
to better than 1.7 A resolution (Table 1). Two of the
Figure 5. Minor groove widths (solid lines) and depths (dashed lines)
in ClU-modified DDD duplexes in complex with RNase H compared
with native DDD d(CGCGAATTCGCG) [PDB ID 436D; (31)] and a
DDD d(CGCGAATTTGCG) with T:G mismatch pairs [underlined,
PDB ID 113D; (32)]. All parameters were calculated with the
program Curves (27).
complexes feature duplexes with two T:A pairs replaced
by ClU:A (ClU7 and ClU8) and a third contains a duplex
with all four T:A pairs in the central A-tract of the DDD
replaced by ClU:A (ClU7/8). A fourth complex is between
BhRNase H and a DDD in which two ClU:G pairs
Table 2. UV-melting temperatures of the native DDD, ClU-modified DDDs and a DDD with T:G mismatches
Duplex
DDD
ClU7
ClU8
ClU7/8
ClU9
T:G DDD
Tm (ꢀC)
Hyperchromicity (%)
62.6 0.9
15.5
61.8 1.2
14.3
64.0 0.5
15.7
63.8 0.6
13.5
35.4 0.7
15.6
36.5 0.5
16.3
Figure 6. PAGE assay of EcoRI cleavage experiments with ClU-modiEed DDDs, the native DDD d(CGCGAATTCGCG) and the DDD
d(CGCGAATTTGCG) with T:G mismatch pairs (underlined).

2696 Nucleic Acids Research, 2013, Vol. 41, No. 4
bracket the central T:A tract. The first three structures
illustrate that ClU:A and T:A pairs adopt nearly identical
configurations with two hydrogen bonds (Figures 2, 4A
and 4C), thus confirming the earlier observations based on
the structure of a B-form DNA duplex with ClU:A pairs
analyzed by solution NMR (14). Similarly, the geometry
of ClU:G mismatch pairs in the structure of the fourth
complex closely resembles that of the T:G pair in the
structure of the B-form duplex of the same sequence (32)
(Figure 4B). Crystals of the RNase H complexes were
grown at near neutral pH (6.5) and the wobble configur-
ation with two hydrogen bonds confirms previous reports
of the similarity of the ClU:G and T:G pairs as analyzed
by NMR in solution (15).
The similarities at the level of conformation indicate
that chlorine appears to closely mimic the methyl substitu-
ent of thymine. Indeed, when both bond length and van
der Waals radius are taken into account, the two substitu-
ents exhibit very similar sterics. The length of the C(5)-Cl
bond is clearly longer than that of the C(5)-CH3 bond
ignore the switch from a methyl group to chlorine in the
major groove, T ! ClU base substitution changes groove
geometry and electrostatics in a subtle fashion and may
affect nucleic acid recognition by proteins involved in
replication and transcription. Indeed, in vitro kinetic
assays of DNA polymerization using human polymerase
b and E. coli DNA polymerase I exo- Klenow fragment
with templates containing either ClU or T showed
facilitated incorporation of dGTP opposite ClU
compared with T, particularly at increased values of pH
(i.e. 9.0) (39). A high mutation rate is important for the
evolutionary process and the massive number of A ! G
mutations seen in the adaptation of E. coli to ClU in place
of T is consistent with facile formation of the ClU:G
wobble pair.
ACCESSION NUMBERS
4HUF (ClU7), 4HTU (ClU8), 4HUG (ClU7/8), and
4HUE (ClU9).
˚
˚
(1.73 A versus 1.5 A, respectively), but the van der Waals
˚
radius of a methyl group (2 A) exceeds that of chlorine
˚
(1.75 A) by roughly the same amount. Thus, the two
SUPPLEMENTARY DATA
substituents will be nearly indistinguishable for a protein
probing the major groove of a B-form duplex. Indeed, we
demonstrate here that EcoRI cuts the ClU7-, ClU8- and
ClU7/8 DDDs with two or more Ts in the recognition site
replaced by ClU (Figure 6). Conversely, replacing the C:G
pairs in the target sequence with either T:G or ClU:G
mismatches abolishes cleavage by EcoRI. Chlorine not
only mimics the native methyl substituent of thymine,
thus allowing ClU:A pairs to evade recognition by
enzymes, but replacement of T opposite A by ClU does
not lead to any obvious changes in the thermodynamic
stability of B-form DNA (Table 2). Also, T:G and
ClU:G mismatch pairs result in very similar losses of sta-
bility. The stability data support the notion that chlorine
and methyl at the 5-position of uracil are basically inter-
changeable. Although chlorine can, in principle, partici-
pate in halogen bonds (38), the observed distances
between chlorine and phosphate oxygens in the
ClU-modified DDD structures are all clearly above the
Supplementary Data are available at NAR Online:
Supplementary Tables 1–6 and Supplementary Figures
1–4.
ACKNOWLEDGEMENTS
P.H. is indebted to KU Leuven, FWO and ERC for finan-
cial support. We are grateful to Dr. Z. Wawrzak,
Northwestern University, for assistance with X-ray dif-
fraction data collection.
FUNDING
Funding for open access charge: US National Institutes of
Health [R01 GM55237 to M.E.].
Conflict of interest statement. None declared.
˚
sum of the van der Waals radii for Cl and O (3.27 A):
˚
˚
˚
REFERENCES
minimum 5.09 A, maximum 6.23 A and average 5.49 A.
Chlorine atoms are also poorly hydrated and the
shortest contact between a chlorine and water in our struc-
1. Lindahl,T. (1993) Instability and decay of the primary structure
of DNA. Nature, 362, 709–715.
˚
tures is 3.71 A and thus not indicative of a hydrogen bond.
2. Heidelberger,C., Chaudhuri,N.K., Danneberg,P., Mooren,D. and
Griesback,L. (1957) Fluorinated pyrimidines, a new class of
tumor-inhibitory compounds. Nature, 179, 663–666.
3. Morris,S.M. (1993) The genetic toxicology of 5-fluoropyrimidines
and 5-chlorouracil. Mutat. Res., 297, 39–51.
4. Longley,D.B., Harkin,D.P. and Johnston,P.G. (2003)
5-Fluorouracil: Mechanisms of action and clinical strategies. Nat.
Rev. Cancer, 3, 330–338.
5. Henderson,J.P., Byun,J., Takeshita,J. and Heinecke,J.W. (2003)
Phagocytes produce 5-chlorouracil and 5-bromouracil, two
mutagenic products of myeloperoxidase, in human inflammatory
tissue. J. Biol. Chem., 278, 23522–23528.
6. Byrd,D.M. and Prusoff,W.H. (1975) Multiplicity reactivation of 5
iodouracil substituted, nonviable bacteriophage T4td8. Antimicrob.
Agents Chemother., 8, 558–563.
Similar melting temperatures established for the native
DDD and the ClU-modified DDDs are also inconsistent
with differential polarizations of the thymine and
chlorouracil nucleobases as this should affect stacking
and thus stability.
The similarity between ClU and T in terms of pairing
properties, conformation and duplex stability helps
rationalize the successful outcome of the in vivo evolution
of E. coli strains that rely on a ClU, A, C, G alphabet
instead of T, A, C, G (13). It is unlikely that a similar
success could be achieved with the FU or BrU analogs
that exhibit more significant steric and stereoelectronic de-
viations from T compared with ClU. Although we found
in one instance (EcoRI) that an enzyme appeared to
7. Jang,Y.H., Sowers,L.C., Cagin,T. and Goddard,W.A. III (2001)
First principles calculation of pKa values for 5-substituted uracils.
J. Phys. Chem. A, 105, 274–280.

Nucleic Acids Research, 2013, Vol. 41, No. 4 2697
8. Coll,M., Saal,D., Frederick,C.A., Aymami,J., Rich,A. and
Wang,A.H.-J. (1989) Effects of 5-fluorouracil/guanine wobble base
pairs in Z-DNA: molecular and crystal structure of d(CGCGFG).
Nucleic Acids Res., 17, 911–923.
9. Lukin,M. and de los Santos,C. (2006) NMR structures of
damaged DNA. Chem. Rev., 106, 607–686.
10. Parker,J.B. and Stivers,J.T. (2011) Dynamics of uracil and
5-fluorouracil in DNA. Biochemistry, 50, 612–617.
11. Kinsella,T.J. (1996) An approach to the radiosensitization of
human tumors. Cancer J. Sci. Am., 2, 184–193.
maximum-likelihood method. Acta Crystallogr. D Biol.
Crystallogr., 53, 240–255.
25. Emsley,P. and Cowtan,K. (2004) Coot: Model-building tools for
molecular graphics. Acta Crystallogr. D Biol. Crystallogr., 60,
2126–2132.
26. Adams,P.D., Afonine,P.V., Bunkoczi,G., Chen,V.B., Davis,I.W.,
Echols,N., Headd,J.J., Hung,L.W., Kapral,G.J., Grosse-
Kunstleve,R.W. et al. (2010) PHENIX: a comprehensive
Python-based system for macromolecular structure solution. Acta
Crystallogr. D Biol. Crystallogr., 66, 213–221.
12. Saif,M.F., Cheng,T.-C., Berk,G. and Kinsella,T.J. (2007) IPdR: A
novel oral#radiosensitizer. Expert. Opin. Investig. Drugs, 16,
1415–1424.
27. Lavery,R. and Sklenar,H. (1989) Defining the structure of
irregular nucleic acids: conventions and principles. J. Biomol.
Struct. Dyn., 6, 655–667.
13. Marliere,P., Patrouix,J., Doring,V., Herdewijn,P., Tricot,S.,
Cruveiller,S., Bouzon,M. and Mutzel,R. (2011) Chemical
evolution of a bacterium’s genome. Angew. Chem. Int. Ed., 50,
7109–7114.
28. Pettersen,E.F., Goddard,T.D., Huang,C.C., Couch,G.S.,
Greenblatt,D.M., Meng,E.C. and Ferrin,T.E. (2004) UCSF
Chimera, a visualization system for exploratory research and
analysis. J. Comp. Chem., 25, 1605–1612.
14. Theruvathu,J.A., Kim,C.H., Rogstad,D.K., Neidigh,J.W. and
Sowers,L.C. (2009) Base-pairing configuration and stability of an
oligonucleotide duplex containing a 5-chlorouracil-adenine base
pair. Biochemistry, 48, 7539–7546.
15. Theruvathu,J.A., Kim,C.H., Darwanto,A., Neidigh,J.W. and
Sowers,L.C. (2009) pH-dependent configurations of a
5-chlorouracil-guanine base pair. Biochemistry, 48, 11312–11318.
29. Pallan,P.S., Prakash,T.P., Li,F., Eoff,R.L., Manoharan,M. and
Egli,M. (2009) A conformational transition in the structure of a
20-thiomethyl-modified DNA visualized at high resolution.
Chem. Comm., 2017–2019.
30. Nowotny,M., Gaidamakov,S.A., Crouch,R.J. and Yang,W. (2005)
Crystal structures of RNase H bound to an RNA/DNA hybrid:
substrate specificity and metal-dependent catalysis. Cell, 121,
1005–1016.
16. Cruse,W., Saludjian,P., Neuman,A. and Prange,T. (2001)
´
Destabilizing effect of a fluorouracil extra base in a hybrid RNA
duplex compared with bromo and chloro analogues. Acta
Crystallogr. D Biol. Crystallogr., 57, 1609–1613.
17. Lima,W.F. and Crooke,S.T. (1997) Binding affinity and specificity
of Escherichia coli RNase H1: impact on the kinetics of catalysis
of antisense oligonucleotide-RNA hybrids. Biochemistry, 36,
390–398.
18. Pallan,P.S. and Egli,M. (2008) Insights into RNA/DNA hybrid
recognition and processing by RNase H from the crystal structure
of a non-specific enzyme-dsDNA complex. Cell Cycle, 7,
2562–2569.
19. Rui,E.K. and Kim,J.N. (1989) The oxidative chlorination of
pyrimidine and purine bases and nucleosides using acyl
chloride-dimethyl-formamide-m-chloroperbenzoic acid system.
Nucleosides, Nucleotides, 8, 43–48.
20. Jancarik,J. and Kim,S.H. (1991) Sparse matrix sampling: a
screening method for crystallization of proteins J. Appl. Cryst.,
24, 409–411.
21. Otwinowski,Z. and Minor,W. (1997) Processing of X-ray
diffraction data collected in oscillation mode. Meth. Enzymol.,
276, 307–326.
22. Vagin,A. and Teplyakov,A. (1997) MOLREP: an automated
program for molecular replacement. J. Appl. Crystallogr., 30,
1022–1025.
23. Collaborative Computational Project, Number 4. (1994) The
CCP4 suite: Programs for protein crystallography. Acta
Crystallogr. D Biol. Crystallogr., 50, 760–763.
24. Murshudov,G.N., Vagin,A.A. and Dodson,E.J. (1997)
Refinement of macromolecular structures by the
31. Tereshko,V., Minasov,G. and Egli,M. (1999) The Dickerson-Drew
B-DNA dodecamer revisited - at atomic resolution. J. Am. Chem.
Soc., 121, 470–471.
32. Hunter,W.N., Brown,T., Kneale,G., Anand,N.N., Rabinovich,D.
and Kennard,O. (1987) The structure of guanosine-thymidine
˚
mismatches in B-DNA at 2.5-A resolution. J. Biol. Chem., 262,
9962–9970.
33. Kornberg,A. and Baker,T. (1992) DNA Replication, 3rd edn. W.
H. Freeman, New York.
34. Dunn,D.B. and Smith,J.D. (1957) Effects of 5-halogenated uracils
on the growth of Escherichia coli and their incorporation into
deoxyribonucleic acids. Biochem. J., 67, 494–506.
35. Munch-Petersen,A. (ed.), (1983) Metabolism of Nucleotides,
Nucleosides and Nucleobases in Microorganisms. Academic Press,
New York.
36. Willis,M.C., Hicke,B.J., Uhlenbeck,O.C., Cech,T.R. and
Koch,T.H. (1993) Photocrosslinking of 5-iodouracil-substituted
RNA and DNA to proteins. Science, 262, 1255–1257.
37. Pallan,P.S. and Egli,M. (2009) The pairing geometry of the
hydrophobic thymine analog 2,4-difluorotoluene in duplex DNA
as analyzed by X-ray crystallography. J. Am. Chem. Soc., 131,
12548–12549.
38. Auffinger,P., Hays,F.A., Westhof,E. and Ho,P.S. (2004) Halogen
bonds in biological molecules. Proc. Natl Acad. Sci. USA., 101,
16789–16794.
39. Kim,C.H., Darwanto,A., Theruvathu,J.A., Herring,J.L. and
Sowers,L.C. (2010) Polymerase incorporation and miscoding
properties of 5-chlorouracil. Chem. Res. Toxicol., 23, 740–748.