Base pairs in the duplex exhibit slide 1.96 A˚ and intermediate values for X-displacement 3.23 A˚, as in ADNA, while their inclination to the helical axis is not prominent.. Importantly,
Trang 1Solution structure of 2¢,5¢ d(G4C4)
Relevance to topological restrictions and nature’s choice of phosphodiester links Bernard J Premraj1, Swaminathan Raja1, Neel S Bhavesh2, Ke Shi3, Ramakrishna V Hosur2,
Muttaiya Sundaralingam3and Narayanarao Yathindra1
1
Department of Crystallography and Biophysics, University of Madras, Guindy Campus, Chennai, India;2Department of Chemical Sciences, TIFR, Colaba, Mumbai, India;3Department of Chemistry, The Ohio State University, Columbus, OH, USA
The NMR structure of 2¢,5¢ d(GGGGCCCC) was
deter-mined to gain insights into the structural differences between
2¢,5¢- and 3¢,5¢-linked DNA duplexes that may be relevant
in elucidating nature’s choice of sugar-phosphate links to
encode genetic information The oligomer assumes a duplex
with extended nucleotide repeats formed out of mostly
N-type sugar puckers With the exception of the 5¢-terminal
guanine that assumes the syn glycosyl conformation, all
other bases prefer the anti glycosyl conformation Base pairs
in the duplex exhibit slide ()1.96 A˚) and intermediate values
for X-displacement ()3.23 A˚), as in ADNA, while their
inclination to the helical axis is not prominent Major and
minor grooves display features intermediate to A and
BDNA The duplex structure of iso d(GGGGCCCC) may
therefore be best characterized as a hybrid of A and BDNA
Importantly, the results confirm that even 3¢ deoxy 2¢,5¢
DNA supports duplex formation only in the presence of distinct slide (‡)1.6 A˚) and X-displacement (‡ )2.5 A˚) for base pairs, and hence does not favor an ideal BDNA topology characterized by their near-zero values Such restrictions on base pair movements in 2¢,5¢ DNA, which are clearly absent in 3¢,5¢ DNA, are expected to impose con-straints on its ability for deformability of the kind observed
in DNA during its compaction and interaction with proteins
It is therefore conceivable that selection pressure relating to the optimization of topological features might have been a factor in the rejection of 2¢,5¢ links in preference to 3¢,5¢ links Keywords: structure of 2¢,5¢ DNA; evolution of 3¢,5¢ vs 2¢,5¢ links in nucleic acids; AB hybrid structure; restrained base pair movements; topological restrictions in 2¢,5¢ DNA
Nature’s selection of 3¢,5¢ linkages (instead of 2¢,5¢ linkages)
in nucleic acids, to encode genetic information, is intriguing
The fact that 2¢,5¢ links are formed in abundance and serve
as a template in nonenzymatic reactions suggest that they
might have been the ancestors of the biotic 3¢,5¢ links, which
could have evolved from a pool of 3¢,5¢ and 2¢,5¢ links [1]
Nucleic acids with 2¢,5¢ links satisfy one of the critical
features required for the fidelity of replication, namely that
they associate to form Watson and Crick base-paired
duplex structures [2–5], although with weaker affinity than
3¢,5¢-linked DNA strands However, detailed knowledge
about stereochemistry, polymorphism and topological
properties of 2¢,5¢ DNA duplexes, which may provide
insights into the factors that determine nature’s choice of
sugar-phosphate links from a stereochemical perspective, is
sparse [6–9] In fact, there are only two reports of NMR
structure determination – one on a 2¢,5¢ DNA fragment [10]
and one on a 2¢,5¢ RNA fragment [11] – both of which
suggest an A-type duplex structure with some
stereochem-ical details that differ from genomic DNA and RNA
duplexes In this context, it is relevant to recognize the results from recent modeling studies on 2¢,5¢ nucleic acids, which suggest that 2¢,5¢ DNA cannot form a 10-fold BDNA-like duplex (like 3¢,5¢ DNA) without the mandatory slide (‡)1.6 A˚) and X-displacement (‡ )2.5 A˚) [9] With a view to probe further into the structural properties of 2¢,5¢ DNA, we report here a high-resolution NMR study of the 2¢,5¢ DNA fragment that possesses a guanine tract followed
by a cytosine tract, to discern also possible sequence effects The results show that iso d(GGGGCCCC) [d(G4C4)] assumes a duplex that conforms to neither a canonical BDNA nor an ADNA family, but a duplex characterized
by features of both A and BDNA Possible implications
of this on the topological restrictions of 2¢,5¢ DNA, and its rejection by nature, are discussed
Materials and methods
DNA synthesis and NMR sample preparation The 2¢,5¢-linked 3¢ deoxy (GGGGCCCC) (iso DNA), was synthesized at 1 lmol scale on an in-house Applied Biosystem 391 automatic DNA synthesizer using solid-state phosphoramidite chemistry [12] The universal support (purchased from BioGene) was used as the solid support for the synthesis The standard concentration of phospho-ramidite was diluted with an equal volume of acetonitrile The products were cleaved off the column with 5 mL of 37% ammonium hydroxide containing 5% LiCl The
Correspondence to N Yathindra, Department of Crystallography and
Biophysics, University of Madras, Guindy Campus, Chennai-600 025,
India Fax: + 91 44 2230 0122,
2 Tel.: + 91 44 2235 1367,
E-mail: ny@vsnl.com
Abbreviations: d(G 4 C 4 ), d(GGGGCCCC); LALS, linked atom least
squares; RDC, residual dipolar couplings.
(Received 4 March 2004, revised 30 April 2004, accepted 21 May 2004)
Trang 2solution was incubated in a 55C water bath for 16 h and
then lyophilized The pellet from lyophilization was
dis-solved in 5% NaHCO3and purified by FPLC The collected
peak elution was lyophilized and the sample stored at
)20 C NMR samples (0.6 mM) of the 2¢,5¢ DNA fragment
were prepared in 20 mM potassium phosphate buffer
containing 0.5 mM EDTA and 100 mM KCl For
experi-ments in D2O, the samples were lyophilized and redissolved
in D2O UV melting studies show that the Tmof iso d(G4C4)
is 32 C under identical buffer conditions
NMR data acquisition
NMR experiments were carried out on a 600 MHz Varian
Unity-plus spectrometer 1D spectra in H2O were recorded
using the jump-and-return pulse sequence for H2O
sup-pression at different temperatures in the range of 2–45C
[13] 2D NOESY spectra in H2O were recorded at 2C with
mixing times of 80 ms and 300 ms Phase-sensitive NOESY
spectra in D2O [14] were recorded with mixing times of 70,
120, 150, 200, 250 and 300 ms; and TOCSY spectra [15]
were recorded with mixing times of 30 and 90 ms at 2C
The DQF-COSY spectrum [16,17] and 2D J-resolved
spectra [18] were recorded in D2O for 1H–1H coupling
constant estimation For the various experiments, the time
domain data consisted of 2048 complex points in t2 and
300–400 complex points in t1 dimension The relaxation
time delay was between 1 and 3 s for the different 2D
experiments
Experimental restraints
Data processing and analysis were carried out usingVNMR
andFELIXpackages [19] on a Silicon graphics work station
Based on the relative intensities and build-up, the cross
peaks in the NOESY spectra (obtained in D2O at various
mixing times), are classified as strong, medium-strong,
medium, and weak, and the interproton distances are
restrained, respectively, to the ranges 2–3 A˚, 2.5–3.5 A˚,
3–4.5 A˚, and 3.5–5.5 A˚ The narrow bounds are mostly
used for strong intranucleotide cross peaks, for which
distance ranges are small and known As the distance ranges
for the observable NOEs are not so large, the NOE distance
bounds used are considered to be realistic The interproton
distances involving the exchangeable protons in the H2O
NOESY spectra are restrained to the ranges 2–4 A˚ and 3.5–
5.5 A˚, corresponding to the strong and weak cross peaks,
respectively At this level of NOE intensity quantification,
spin diffusion is not expected to influence the distance
restraints to a significant extent Even so, the larger the
number of distance restraints, the better it is for internal
consistency, and the structures derived would be more
reliable A total of 162 NOE restraints were collected, of
which 115 were intranucleotide and 47 internucleotide NOEs
Base (H8/H6) sugar proton NOEs, especially to the H1¢, also enable deriving constraints on the glycosyl torsion angles The H8/H6–H1¢ distance is very short ( 2.3–2.5 A˚) for a syn conformation and relatively much longer ( 3.5– 4.0 A˚) for an anti conformation Thus, the H8/H6–H1¢ NOE will be very strong, even at short mixing times (such as 60–70 ms) if the glycosyl torsion angle is in the syn domain, whereas, under the same conditions, the peak will be nearly absent for an anti conformation We observe that G1 has a synconformation, while all others are in the anti domain (spectra presented in Results)
The 2D J-resolved spectra provides precise values of the J(H1¢–H2¢) coupling constants (Table 1) The observed coupling constants are very small, indicating that the sugar geometry belongs largely to the N domain (in the N domain this coupling constant is near 0–2 Hz, whereas it varies between 9 and 10 Hz in the S domain) A common practice is to consider the sugar geometry as an equilib-rium mixture of N and S types, and the coupling constants as weighted averages However, there are also reports in the literature [18] where the sugar ring is believed to be rigid, and is primarily of a single type, at least in the interior of the duplex In the present case, we observe that the terminal residues, for example, C8 and G2, where one would have expected greater dynamism, exhibit very small values ( 1.5Hz) for J(H1¢–H2¢) If one considers an equilibrium model, for a 10% contribution of the S domain, the contribution to the coupling constant would be around 1 Hz Moreover, it is clear from the steepness of the curve displaying the dependence of coupling constants on pseudorotation phase angle P (Fig 1), that the P range in the N domain is not going
to be very different regardless of whether the S contribu-tion is explicitly considered Thus, from the small values
of the coupling constants for the terminal residues, it is evident that the sugar geometries are dominantly in the N domain only This will be also true for the internal residues Now, in the N domain, especially in the P range 30–80, the dependence of H1¢–H2¢ coupling on P is very steep and this significantly narrows the range of permis-sible P-values for a given coupling constant value [18,20] Taking these factors into consideration, sugar puckers were restrained to the P ranges indicated in Table 1 and these were then converted to respective dihedral angle ranges in the sugar rings
The hydrogen bond restraints were given as two distances per hydrogen bond (a total of 36 restraints) for the central hexamer (see below) Based on the observation that the peak count for the duplex is the same as expected from a single strand in the various spectral data (indicating that the duplex is highly symmetric and the two strands are
Table 1 J(H1¢–H2¢) coupling and the corresponding ranges of phase angle of pseudorotation (P°).
Range of phase angle
of pseudorotation (P)
49–58 35–47 55–64 49–58 57–66 37–48 68–77 33–45
Trang 3equivalent), NCS restraints were imposed to obtain
sym-metry between the two strands forming the duplex
Structure calculation
Structure calculation of the iso d(GGGGCCCC) was
carried out using X-PLOR 3.8.5 [21] The topology and
parameter files were appropriately modified to handle 2¢,5¢
linkages to obtain optimum geometry at the 2¢,5¢
phospho-diester linkage Ideal A- and B-type duplex models for iso
DNA (possessing helical parameters identical to those of the
canonical ADNA and BDNA), obtained previously [9]
using the linked atom least squares (LALS) refinement
approach [22], were used as the starting models for structure
calculation This is justified considering that the NMR
spectra in water clearly establish Watson and Crick base
pair formation between antiparallel strands The model iso
ADNA duplex is characterized by the same value of slide,
X-displacement and the helical parameters, as ADNA On
the other hand, the iso BDNA model, while possessing the
same helical parameters as BDNA, is distinguished by a
nonzero slide (‡)1.7 A˚) and X-displacement (‡ )2.5 A˚), in
sharp contrast to the ideal BDNA that is characterized by
zero values for them Nonzero slide and X-displacement are
found to be mandatory to generate a 10-fold 2¢,5¢ duplex,
even with 3¢ deoxy sugars [9] Thus, the iso BDNA and iso
ADNA models are very different from each other, and
choosing these two as initial models removes any starting
model bias in the results of calculation Such a strategy also
saves computational efforts compared to starting the
calculation from a completely extended structure In the
latter case, much effort is expended for the formation of
the base pair itself
Syn conformation was imposed for the 5¢ end guanine
(see below) The initial model was subjected to restrained
energy minimization using the conjugate gradient algorithm
and was guided by the experimental NOE distance restraints
as well as dihedral restraints A conformational search was
performed on the octamer duplex using the simulated
annealing protocol [23], followed by structure refinement
using the gentle refine protocol ofX-PLOR3.8.5 A distance-dependent dielectric constant was used throughout the structure calculation to mimic the presence of high dielectric solvent, typically for simulating water (when explicit water is not used) The starting structure was heated to 1000 K, and sets of 100 structures that are significantly different from one another were extracted during high-temperature dynamics Each of the structures was subjected to 18 ps of high-temperature dynamics followed by slow cooling to
100 K, at steps of 50 K During each cooling step the structures were subjected to 500 fs of molecular dynamics Finally, the structures were energy minimized using the conjugate gradient algorithm This was followed by a refinement using the gentle refine protocol, where each of the structures was subjected to 20 ps of molecular dynamics
at 300 K Average coordinates over the last 10 ps of molecular dynamics simulation were computed and then refined by conjugate gradient minimization The NOE distance restraints, hydrogen bond restraints (given as two distances per hydrogen bond), and dihedral restraints on the sugar conformation were applied throughout the entire calculation with force constants of 50 kcalÆmol)1ÆA˚ )2,
100 kcalÆmol)1ÆA˚)2 and 300 kcalÆmol)1ÆA˚)2, respectively NCS restraints with a force constant of 300 kcalÆmol)1ÆA˚)2 were imposed to obtain symmetry between the two strands
of the duplex
Results
1D and 2D proton spectra The 1D1H NMR spectrum (Fig 2A) of the octamer iso d(GGGGCCCC) displays three peaks corresponding to the imino protons at 13.25 (G4), 12.80 (G2) and 12.70 (G3) p.p.m., expected from Watson and Crick base pairs in
an antiparallel duplex Sequence-specific assignments for the exchangeable and nonexchangeable protons were made from the NOESY and TOCSY spectra following the procedures developed for 3¢,5¢ duplexes [24] The observa-tion of NOE changes from G imino to C amino protons of nonterminal base pairs in the NOESY water spectra (Fig 2B) further substantiates the formation of Watson and Crick base pairing between G and C The uninterrupted self and sequential connectivity from H8/H6 to H1¢ (Fig 3A), as well as H8/H6 to H2¢ (Fig 3B) in the NOESY spectra suggest a right-handed helical structure These sequential connectivities are consistent throughout the various regions of the spectra From the temperature dependence of the G imino resonances in 1D spectra in
H2O (data not shown), the melting temperature of the duplex was seen to be 30 C
Tables 2 and 3
3 show the chemical shift values for all the assigned sugar and base protons The stereospecific assign-ments involving the 3¢ and 3¢¢ protons were based on the 2¢)3¢ and 2¢)3¢¢ NOE intensities in the 70 ms NOESY spectrum As the H2¢–H3¢ proton separation is shorter than the H2¢–H3¢¢ separation, irrespective of the sugar confor-mation, the H2¢–H3¢ NOE intensity should be stronger at shorter mixing times The relative intensities of the cross-peaks of the interproton base to sugar NOEs in the NOESY spectrum (Fig 3C), at mixing times varying from 70 to
300 ms, indicate that the 5¢-terminal guanine exists in the
Fig 1 Plots showing the dependence of the 3-bond coupling constants
(J) on the phase angle of pseudorotation (P).
Trang 4synconformation, while other bases favor the anti confor-mation This is a recurring feature found in 2¢,5¢-linked dimers [25–27] and oligomers [10,11] In the crystal struc-tures of 2¢,5¢-linked dinucleoside monophosphates, the syn conformation is stabilized by an intramolecular hydrogen
Fig 2 NMR spectra and the NOESY spectrum (A) 1D H 2 O
exchangeable NMR spectra of iso d(GGGGCCCC) in 100 m M KCl,
pH 7.0, and at 2 C, showing the imino and amino proton signals (B)
Selected region of the NOESY spectrum (mixing time 300 ms) in H 2 O
solution showing NOE correlations from G imino to C amino protons.
CNH 2(i) and CNH 2(e) refer to the internal (H-bonded) and external
(free) amino protons of the cytosine base.
Fig 3 (H8/H6)–H1¢ cross-peak region of a 300 ms 2D NOESY spectrum of iso d(GGGGCCCC) in D 2 O solution at 2 °C, showing the uninterrupted sequential connectivities from (A) (H8/H6) to H1¢ protons (B) (H8/H6) to H2¢ protons (C) Stacked plot of the H8/H6 to H1¢ region showing a high intensity for the H8–H1¢ cross-peak of G1, suggesting syn glycosyl conformation for the terminal G1 residue.
Trang 5bond between the purine N3 and O5¢H of the sugar residue,
besides sugar O4¢–base (syn) base interaction [9,25–27]
The (H1¢–H2¢) coupling constants derived from 2D
J-resolved spectra clearly indicate that all of the 3¢ deoxy
sugars belong to the N type, except for C7 which has a
slightly higher coupling constant (5.1 Hz)
Structural features of 2¢,5¢ d(GGGGCCCC)
The 3D structure of 2¢,5¢ d(GGGGCCCC) was obtained by
simulated annealing molecular dynamics usingX-PLOR3.8.5
[21] Experimental restraints and structure convergence
parameters are listed in Table 4 The convergent structures
are clustered into families: BFI (Fig 4A) with 39 structures,
and BFII (Fig 4B) with 20 structures when the starting
model was ideally iso BDNA; and AFI (Fig 4C) with 85
structures and AFII (Fig 4D) with 15 structures when the
starting model was iso ADNA Structures represented
by BFI and AFI families (FI) differ considerably in their
overall topologies from the structures represented by BFII
and AFII families (FII) The root mean square
devi-ation (rmsd) between FI and FII is greater than 3 A˚, while
it is less than 1 A˚ for structures within FI or FII The
structures were selected using standard criteria on the basis
of proper covalent geometry, the least number of distance
and dihedral violations, symmetry and low energy
The duplex model AFI (Fig 5A), closely resembles BFI
(Fig 5B) The rmsd between the average structure of AFI
(Fig 5A) and BFI (Fig 5B) is 0.8 A˚ Thus, in spite of the
large rmsd (> 4 A˚) in the starting structures, the final
structures fall into similar families, indicating that the
structures are not biased by the choice of the initial model
This also indicates that the experimental restraints are
sufficient and consistent to define good convergent struc-tures In view of this, it is believed that there is no need for any further refinements using residual dipolar couplings (RDCs), as often performed in longer DNA stretches [28–30] Likewise, we also did not feel the need for any relaxation matrix refinement, which takes into account spin diffusion explicitly, which may be required if the NOE data set is very small At the same time, relaxation matrix refinement puts a greater demand on the accuracy of NOE quantification
In the final structures, the terminal GC pairs are not well defined owing to insufficient NOEs Hence, structural features manifested in the central hexamer of iso d(G4C4), corresponding to the GGGCCC duplex in the family AFI, which has the highest population of converged structures and also has very good convergence, are considered for detailed discussion
Calculated values of X-displacement and slide for the base pairs in AFI are given in Table 5 Average values of X-displacement and slide of GC base pairs at the GG step (Fig 5A) are)3.25 A˚ and )1.62 A˚ (Table 5), respectively
On the other hand, slide for the GC pair at the GC step that links the G stretch with the C stretch is rather high ()3.32 A˚)
The nature of the base stacking interaction in the iso d(GGGGCCCC) duplex, as seen in AFI, is shown in Fig 6A Stacking at the G2G3 and G3G4 steps involves overlap of the six-membered ring of one guanine with the imidazole ring of the adjacent guanine, while there is only
Table 3 Chemical shifts (p.p.m) for iso d(GGGGCCCC) 2
exchange-able protons.
Base H 1 H 22 /H 42 (e) H 21 /H 41 (i)
Table 2 Chemical shifts (p.p.m) for iso d(GGGGCCCC) 2
non-exchangeable protons.
Residue H6/H8 H1¢ H2¢ H3¢ H3¢¢ H4¢ H5¢/H5¢¢ H5
G1 7.97 5.98 5.18 2.53 2.37 4.58 3.88,3.69 –
G2 7.78 5.83 4.7 2.45 2.31 4.61 4.18,4.08 –
G3 7.62 5.96 4.91 2.56 2.48 4.78 4.51,4.14 –
G4 7.58 5.99 4.61 2.48 2.38 4.77 – –
C5 7.5 6.08 4.62 2.44 – 4.76 4.39,4.07 5.10
C6 7.88 5.94 4.5 2.31 – 4.71 4.10 5.48
C7 7.73 6.03 4.66 2.49 2.31 4.04 4.27 5.52
C8 7.84 5.66 4.26 1.84 1.82 4.56 4.39,4.0 5.61
Table 4 NMR restraints for iso d(GGGGCCCC) 2 NOE distance restraints (per strand)
Non-exchangeable NOE restraints 140 Exchangeable NOE restraints 22 Total restraints 162
Sugar dihedral restraints (per strand) 40 Hydrogen bond restraints 36 BFI (model obtained when iso BDNA is used as the starting duplex) Number of convergent structures 39
rmsd from the average structure 0.5 A˚ )1.0 A˚ NOE violation > 0.2 A˚ 1
Dihedral angle violation > 5 Nil BFII (model obtained when iso BDNA is used as the starting duplex) Number of convergent structures 20
rmsd from the average structure 0.3 A˚ )1.0 A˚ NOE violation > 0.2 A˚ 1
Dihedral angle violation > 5 Nil AFI (model obtained when ADNA is used as the starting duplex) Number of convergent structures 85
rmsd from the average structure 0.1 A˚ )0.6 A˚ NOE violation > 0.2 A˚ 1
Dihedral angle violation > 5 Nil AFII (model obtained when ADNA is used as the starting duplex) Number of convergent structures 15
rmsd from the average structure 0.1 A˚ )0.5 A˚ NOE violation > 0.2 A˚ 1
Dihedral angle violation > 5 Nil
Trang 6minimal stacking between cytosines Likewise, stacking at
the GC step, which links the G stretch with the C stretch, is
minimal owing to a larger slide ()3.32 A˚) Superposition of
the base pairs of the (GGGCCC)2 fragment of the iso
d(G4C4) duplex with the ideal iso BDNA (Fig 7),
demon-strates a strong resemblance in the stacking patterns
An estimate of the dimensions of major and minor
grooves is obtained by generating a 12mer duplex using the
central hexamer of the average structure (AFI) as the repeat
using the programFREEHELIX[31] The groove topologies of
AFI show significantly different features from the ideal
duplex models The major groove is wide (17 A˚), while its
minor groove is narrow (10.3 A˚)
The 3¢ deoxy sugars in iso d(G4C4) favor N-type pucker,
corresponding to the C4¢ exo conformational domain (P ¼
38–64), except in the residue C7, which favors C4¢ exo/O4¢
endo pucker, corresponding to P¼ 54–90 (Table 1) in
AFI In any case, none of the sugars shows a tendency for
S-type sugar conformation
Base pairs in AFI are slightly overwound, and the duplex
has 9 bp per turn, with an average helical twist of 38.4 and
a rise of 3.76 A˚ (Table 5) The average helical twist at the
GG and CC steps is 41, while it is 28 at the GC step Slight underwinding at this step is accompanied by a higher slide of)3.32 A˚ Base pairs are nearly perpendicular to the helix axis (inclination angle 3) The two central base pairs
of the duplex are practically planar and they do not exhibit significant propeller twist (Table 6), while the base pairs flanking them possess a larger value of)22 Phosphodi-ester conformations at all the GG steps, as well as at the GC step, conform to the (g–,g–) domain, while they correspond
to the (t,g–) at the CC steps (Table 7)
Discussion
It is now well established that nucleic acids, even with 2¢,5¢ linkages, associate to form Watson and Crick paired duplexes [2–5,10,11,32–37] They also selectively associate with DNA and RNA with a varying degree of stability Interestingly, it has been shown recently that 2¢,5¢ RNA fragments form even hairpins with a stability comparable to RNA hairpins [38] In an effort to obtain a comprehensive understanding of the stereochemistry that govern the structures of 2¢,5¢ nucleic acids, we recently reported the
Fig 5 Stereo plot of the average structure of iso d (G 4 C 4 ) (A) AFI and (B) BFI.
Fig 4 Stereo plot of the families of converged structures of iso
d(GGGGCCCC) 2 (A) BFI (39 structures), (B) BFII (20 structures),
(C) AFI (85 structures), and (D) AFII (15 structures).
Table 5 Base-step parameters in the average structure (AFI) of the iso d(GGGGCCCC) duplex.
Base step Slide (A˚) X-disp (A˚) Twist () Rise (A˚) G2-G3 )1.53 )3.36 42.3 3.68 G3-G4 )1.71 )3.13 39.67 3.61 G4-C5 )3.32 )3.19 28.02 4.23 C5-C6 )1.71 )3.13 39.63 3.61 C6-C7 )1.53 )3.23 42.37 3.68 Average )1.96 )3.20 38.39 3.76
Trang 7NMR structure of a 2¢,5¢ RNA fragment [11] that exhibited
interesting features which supported our predictions from
modeling studies [8,9] We report here the results of
high-resolution NMR structure of a 2¢,5¢-linked DNA fragment
d(GGGGCCCC)
The structural model, AFI, that emerged from NOE
and other NMR data, exhibit slide ()1.96 A˚) and
intermediate X-displacement ()3.32 A˚) for the base pairs,
a feature normally seen only in ADNA duplexes
However, the magnitude of X-displacement observed here
is lower () 4.7 A˚) than that found in ADNA
Interest-ingly, the slide ()3.32 A˚) at the lone GC step, linking the
G stretch with the C stretch, is found to be nearly twice
that found at the GG steps ()1.62 A˚), indicating possible
sequence effects A comparison of the stacking pattern
observed at the GG steps of the present structure with those in ideal ADNA, iso ADNA and iso BDNA duplexes (Fig 6B) brings out a strong similarity It is interesting that the similarity in stacking persists, notwith-standing different values for X displacement that charac-terizes these duplexes (Table 5) However, it should be noted that all of them possess nearly the same slide ()1.7 A˚) Thus, the base stacking pattern in iso d(G4C4) is like that in ADNA, except at the GC step where a large slide causes adjacent bases to move away, resulting in minimal overlap between them
Another unusual feature is the predominance of N-type pucker in nearly all the 3¢ deoxy sugars in 2¢,5¢ d(G4C4) This is in sharp contrast to the S-type puckers preferred
by 2¢ deoxy sugars in DNA duplexes This has been
Fig 6 Base stacking at different steps in the AFI duplex of iso (G 4 C 4 ) and the GG steps of iso BDNA: iso ADNA and ADNA Note the identical base stacking at the GG steps of AFI and ideal duplexes Figures were drawn using 3 DNA v1.5 [47].
Trang 8anticipated in view of certain stereochemical arguments
[8,9] Exclusive preference for the N-type sugar puckers
has, in fact, been indicated by the early NMR studies on
2¢,5¢-AAA [39] and crystal structures of 3¢
deoxynucleo-sides [25–27] Such preference for N-type pucker has also
been confirmed by recent1H NMR analysis on a number
of 3¢ deoxynucleosides and stereo-electronic arguments
[40,41] Unconstrained molecular dynamics simulations of
a 2¢,5¢ DNA duplex, lasting a few nanoseconds, have also
demonstrated the retention of N-type pucker for the sugar
[42] It should be recognized at this juncture that the
consequence of N-type sugar pucker is to render the
preferred nucleotide conformation to be extended in 2¢,5¢
DNA and compact in 3¢,5¢ DNA [8,9] It is well known
that the extended nucleotide repeats lead to an extended
BDNA, and the compact nucleotide repeat leads to a
compact ADNA type of duplexes (Fig 8) The 2¢,5¢ DNA
fragment d(G4C4) is thus composed of extended
nucleo-tide repeats that are normally part of BDNA but with a
distinct X-displacement, slide and base stacking like in ADNA Thus, the duplex model AFI of 2¢,5¢ d(G4C4), possesses composite features of both A and BDNA In view of these, it is perhaps appropriate to regard the structure of iso d(G4C4) as a hybrid structure of A and B forms
It is gratifying that the more populated AFI family of iso d(G4C4) resembles the ideal iso BDNA-like duplex [9], which is also characterized by similar values of slide, intermediate displacement, base stacking pattern and extended nucleotide repeat formed out of N-type sugar puckers (Table 5) Furthermore, the overall groove topol-ogies of iso d(G4C4) resemble BDNA, with the widths of the major groove and the minor groove having values of 17 A˚ and 10.3 A˚, respectively (Table 8)
It has been demonstrated from modeling investigations that 2¢,5¢ isomers, even with 3¢ deoxyriboses, cannot form duplexes without base pair displacements [9] Results of CD and FTIR investigations on iso DNA fragments comprising
a variety of base sequences also seem to converge to suggest that they favor A-type rather than B-type duplexes (S Raja
& N Yathindra, unpublished observation) Furthermore, it has been found [43] that iso d(CGCGCG) does not associate
to form left-handed ZDNA This has been attributed to the inaccessibility [42] to form the well-known water-mediated hydrogen bond stabilization interaction between the amino group of the syn guanine and the anion oxygen of the phosphate group [44] These clearly point to the constraint
on the range of duplex helical structures possible for nucleic acids with 2¢,5¢ linkages
The lateral slide of the sugar-phosphate chain from the periphery (as in 3¢,5¢ links) towards the helix interior in
Fig 7 Superposition of the G 3 C 3 fragment of
AFI with ideal iso BDNA Root mean square
deviation with respect to base pairs is 0.6 A˚.
Table 6 Propeller twist (°) of base pairs in the average structure (AFI)
of the iso d(GGGGCCCC) duplex.
Base pair Propeller twist ()
Table 7 Conformation angles (°) in the average structure (AFI) of the iso d(GGGGCCCC) 2 duplex.
Residue
a (P-O5¢)
b (O5¢-C5)
c (C4¢-C5¢)
n (C2¢-O2¢)
f (P-O2¢)
v
(C1¢-N) P
Trang 92¢,5¢ nucleic acids causes the base pairs to slide, resulting in
the intrinsic requirement of slide, and hence
X-displace-ment, that manifest in all 2¢,5¢ nucleic acid duplexes This
limits the access to a lower range of values of both slide
(<)1.5 A˚) and X-displacement (< )2.5 A˚) in 2¢,5¢ nucleic
acids In contrast, nucleic acids with 3¢,5¢ links have a wider
range of access for both slide (0–2.5 A˚) and X-displacement
(0–4.7 A˚) that includes ranges forbidden for the 2¢,5¢ isomer This enables 3¢,5¢-linked nucleic acids to assume a variety of duplexes with distinct topological features and also afford other capabilities, such as bending, kinking and curvature, which form the basis for nucleic acid compaction and specificity of interaction with proteins It is therefore anticipated that the restricting factors in 2¢,5¢ nucleic acids,
Fig 8 Shape and dimension (adjacent P–P separations) of the repeating nucleotide units in 2¢,5¢- and 3¢,5¢-linked nucleic acids An equatorial (e) link renders the adjacent phosphates to be proximal, leading to a compact nucleotide (P–P 5.9 A˚), while an axial (a) link renders them to be distal, leading to an extended nucleotide (P–P 7.0 A˚).
Table 8 Comparison of structural features of the iso d(GGGGCCCC) 2 duplex (AFI) with the ideal A and B types of duplexes formed by 3¢,5¢ and 2¢,5¢ links.
Features/parameters BDNA ADNA iso BDNA iso ADNA AFI
Sugar pucker S type N type N type S type N type
C2¢endo C3¢endo C3¢endo C2¢endo C4¢exo
Trang 10which are mentioned above, probably impose additional
constraints limiting these capabilities Also, it has been
shown from modeling consideration that the lateral slide
of the sugar-phosphate chain leads to overwinding of the
2¢,5¢-linked single-stranded helix to enhance the adjacent
base–base or sugar–base stabilizing interactions [9,42,45]
Tighter winding of the 2¢,5¢ single-stranded DNA helix,
compared with 3¢,5¢ DNA, probably offers restrictions to
the folding abilities of even single-stranded 2¢,5¢ DNA
Hence, it may be argued that topological restrictions
inherent to the 2¢,5¢-linked helical duplexes might have also
contributed towards their rejection It is worth mentioning
that the inherent low thermal stability of 2¢,5¢ links might
have been another factor involved in nature’s selection of
the 3¢,5¢ links Thus, optimization of the topology of duplex
helix, besides the optimization of base pair stability [46],
must have been important in the chemical etiology of
nucleic acid structures
Conclusions
Systematic investigations of 2¢,5¢ nucleic acids have provided
new perspectives on the stereochemical details pertaining to
their ability, or lack of it, to form duplex structures akin to
their naturally occurring 3¢,5¢ isomers In parallel to our
finding [8,9] of the critical features that distinguish the
shapes and dimensions of the repeating nucleotides of 3¢,5¢
and 2¢,5¢ isomers, we have provided structural details of an
isoRNA [11] and an iso DNA duplex fragment (present
work) from NMR studies Together, these should provide a
structural basis for understanding much of the experimental
data from solution studies concerning the associations of
2¢,5¢ nucleic acids and also with DNA and RNA
Compar-ison of the structure deduced for iso d(GGGGCCCC), from
the current study, and that of iso d(CGGCGCCG) [10]
suggest that even 2¢,5¢ DNAs are prone to sequence effects,
as evidenced by some differences seen in structures of the
two sequences The former sequence assumes a hybrid
structure of A and BDNA duplexes, while the latter assumes
an ADNA-like duplex with mixed C2¢ endo and C3¢ endo
sugar puckers for the central hexamer The fact that both
these sequences, studied by NMR, point to a non-BDNA
duplex structure, suggest a constrained nature of base pair
movements in 2¢,5¢ nucleic acids vis-a`-vis their 3¢,5¢ isomers
This is in complete conformity with the modeling studies
[8,9] which indicate that slide and X-displacement of base
pairs lower than )1.7 A˚ and )2.5 A˚, respectively, are
inaccessible owing to the inherent chemistry of the
2¢,5¢-linked sugar-phosphate backbone It seems, then, that a
need for greater topological flexibility of DNA helices might
have had a bearing on the selection of 3¢,5¢ links over 2¢,5¢
links during the course of evolution
Acknowledgements
NMR and computational facilities, provided by the National Facility
for High Resolution NMR at the Tata Institute of Fundamental
Research, Mumbai, are gratefully acknowledged N.Y and B.J.P.
thank DST and CSIR for a research grant and senior fellowship,
respectively S.R thanks CSIR for a Senior Research Fellowship UGC
and DST are thanked for the financial support to the Department
under DSA (UGC) and FIST (DST) programs.
References
1 Prakash, T.P., Roberts, C & Switzer, C (1997) Activity of 2¢,5¢ linked RNA in the template-directed oligomerization of mono-nucleotides Angew Chem Int Ed Engl 36, 1522–1523.
2 Doughtery, J.P., Rizzo, C.J & Breslow, R (1992) Oligodeoxy-nucleotides that contain 2¢,5¢ linkages: synthesis and hybridization properties J Am Chem Soc 114, 6254–6255.
3 Kierzek, R., He L & Turner, D.H (1992) Association of 2¢,5¢ oligoribonucleotides Nucleic Acids Res 20, 1685–1690.
4 Hashimoto, H & Switzer C (1992) Self association of 2¢,5¢ linked deoxynucleotides meta-DNA J Am Chem Soc 114, 6255–6256.
5 Giannaris, P.A & Damha, M.J (1993) Oligoribonucleotides containing 2¢,5¢-phosphodiester linkages exhibit binding selec-tivity for 3¢,5¢ RNA over 3¢,5¢ ssDNA Nucleic Acids Res 21, 4742–4749.
6 Srinivasan, A.R & Olson, W.K (1986) Conformational studies of (2¢,5¢) polynucleotides: theoretical computations of energy, base morphology, helical structure and duplex formation Nucleic Acids Res 14, 5461–5479.
7 Anukanth, A & Ponnuswamy, P.K (1986) 2¢,5¢-linked poly-nucleotides do form a double stranded helical structure: a result from the energy minimization study of A2¢p5¢A Biopolymers 25, 729–752.
8 Lalitha, V & Yathindra, N (1995) Even nucleic acids with 2¢,5¢ linkages facilitate duplexes and structural polymorphism: pros-pects of 2¢,5¢ oligonucleotides as antigene/antisense tool in gene regulation Curr Sci 68, 68–75.
9 Premraj, B.J & Yathindra, N (1998) Stereochemistry of 2¢,5¢ nucleic acids and their constituents J Biomol Struct Dyn 16, 313–327.
10 Robinson, H., Jung, K.-E., Switzer, C & Wang, A.H.-J (1995) DNA with 2¢,5¢ phosphodiester bonds forms a duplex in the A-type conformation J Am Chem Soc 117, 837–838.
11 Premraj, B.J., Patel, P.K., Kandimalla, E.R., Agrawal, S., Hosur, R.V & Yathindra, N (2001) NMR structure of a 2¢,5¢ RNA favors A type duplex with compact C2¢ endo nucleotide repeat Biochem Biophys Res Commun 283, 537–543.
12 Shi, K (1999)
4 Crystal structures of nucleic acid fragments and their complexes with anticancer drugs PhD Thesis, The Ohio State University, Columbus, OH, USA.
13 Plateau, P & Gueron, M (1982) Exchangeable proton NMR without base-line distortion using new strong-pulse sequences.
J Am Chem Soc 104, 7310–7311.
14 Jeener, J., Meier, B.H., Bachmann, P & Ernst, R.R (1979) Investigation of exchange processes by two-dimensional NMR spectroscopy J Chem Phys 71, 4546–4554.
15 Braunschwiler, L & Ernst, R.R (1983) Coherence transfer by isotropic mixing: application to proton correlation spectroscopy.
J Magn Reson 53, 521–528.
16 Rance, M., Sorensen, O.W., Bodenhausen, G., Wagner, G., Ernst, R.R & Wuthrich, K (1983) Improved spectral resolution in COSY 1H NMR spectra of proteins via double quantum filtering Biochem Biophys Res Commun 117, 479–485.
17 Neuhaus, D., Wagner, G., Vasek, M., Kagi, J.H.R & Wuthrich,
K (1985) Systematic application of high-resolution phase-sensi-tive two dimensional 1H-NMR techniques for the identification
of the amino-acid-proton spin systems in proteins Rabbit metallothionein-2 Eur J Biochem 151, 257–273.
18 Wuthrich, K (1986) NMR of Proteins and Nucleic Acids John Wiley and Sons, New York.
19 Biosym Inc (1993) Insight II Biosym Inc., San Diego CA, USA.
20 Hosur, R.V., Ravikumar, M., Chary, K.V.R., Sheth, A., Govil, G.
& Tan Zu-Kun and Miles, T.H (1986) Solution structure of d-GAATTCGAATTC by 2D NMR FEBS Lett 205, 71–76.