Vol. 12 no. 1 1996
Pages 25-30
NAMOT2—a redesigned nucleic acid modeling
tool: construction of non-canonical DNA
structures
Eugene S.Carter, II and Chang-Shung Tung
Abstract
Using a new set of reduced coordinates developed for
describing regular and unusual nucleic acid structures, we
have revised our nucleic acid modeling tool NAMOT2.
NAMOT2 is general in terms of modeling different nucleic
acid structures. A set of modifiable libraries allows users to
customize their modeling environment. With this set of
libraries, NAMOT2 can be used to model non-canonical
structures such as parallel-stranded, triple-stranded and
quadruple-stranded nucleic acid molecules. For modeling
irregular structures (junctions, hairpin loops, etc.), we
introduce a structural recipe approach. The complete
procedure using NAMOT2 to construct the structure of a
specific molecule is treated as the recipe for that structural
motif. The existing recipes can be modified to generate new
recipes for different structural motifs. Several examples of
nucleic acids with non-canonical structures were modeled
using NAMOT2. These examples include a DNA-drug
complex, a DNA cube, a six-arm junction and a curved
DNA molecule.
Introduction
Since the release of NAMOT (a nucleic acid modeling
tool) about a year ago, we have developed a new set of
reduced parameters (Tung et al., 1994) for describing base
structures of nucleic acid molecules. One of the key
features for this new set is that it satisfies the inversion
rule. That is the parameters calculated from the two
strands of the molecule are identical except sign reversals
for tilt (T) and slide (Dx). Compared to the previous
parameter set (Soumpasis and Tung, 1988), the new set is
more versatile in terms of describing non-canonical
structures such as triple-stranded, quadruple-stranded
and parallel-stranded nucleic acid molecules. Using this
new parameter set, we have developed NAM0T2, a
redesigned version of our nucleic acid modeling tool.
NAM0T2 has all the existing features of NAMOT plus
new features that make the software suitable for designing
and modeling both regular and unusual nucleic acid
structures. A WWW (world wide web) site has been
Theoretical Biology and Biophysics (T-10), Theoretical Division, Los
Alamos National Laboratory, Los Alamos, NM 87545, USA
) Oxford University Press
established that hosts all information pertaining to
NAMOT2. To demonstrate the versatility of this
program, examples of modeling non-canonical nucleic
acid structures including a DNA-drug complex, a DNA
cube, a six-arm junction and a curved DNA molecule will
be presented. The coordinates of all these modeled
structures are available through our NAMOT WWW site.
Methods and results
NAM0T2 features
To accommodate the added functions in the program, we
have redesigned the user interface for NAMOT2. Figure 1
shows the layout of the NAMOT2 interface. The
NAMOT2 window consists of five basic areas. The top
bar contains a series of pull-down menus for executing a
set of frequently executed tasks such as open/close files,
alter images in the display (space filling or stick model),
on-line help, etc. The main canvas is designed to display
the molecules. According to their application, the size of
this canvas can be changed by the users with a simple
command. The command window accepts user-input
commands. The results of user actions are displayed in
the data window. The rightmost column has a series of
controls for geometric manipulation of the molecules
being modeled.
NAMOT2 utilizes both Cartesian coordinates and
reduced coordinates. For intramolecular structure manipulation, the set of reduced coordinates was used. The
specifier for the reduced representation is H:U:B (helix:
unit:base). The Cartesian coordinates are used for intermolecular structure manipulation including translations
and rotations of different molecules. The specifier for the
Cartesian representation is M:C:G:A (moleculexhain:
group:atom).
Expansibility and portability are two major goals in
developing NAMOT2. To make the software easily
portable, NAM0T2 has its own sphere (Porter, 1978)
and line rendering (Bresenham, 1965) routines as seen in
other molecular graphic softwares such as RASMOL
(Sayle and Bissell, 1992). NAMOT2 features dynamical
memory allocations and link list for both the molecular
(M:C:G:A and H:U:B) and library data structures. This
feature allows NAMOT2 to avoid setting an arbitrary
25
E.S.Carter, II and C.-S.Tung
namotf
file
Image
Pi mt ing
Advanced
Kelp | Hoi
House
Node
Normal ..j
Phosphate controls
ParanSM
.
Pa i ..J j
Start pj"
NAMOT2
Eod
^1 I
Chmgcr"
act background white
set f0105round black
Base controls
Parmeter
Sx
Start HT 1:2
1-2
_J j
Viwj
j
Change f^"
Fig. I. The layout of the NAMOT2 interface.
maximum molecular size for modeling. Users are only
limited by the capabilities of their computers. NAMOT2
has been compiled successfully on AIX, IRIX, LINUX,
OSF1 and SunOS (4.1.3 and 5.4). Both MOTIF and
XView toolkit versions exist, allowing NAM0T2 to be
compiled on almost any Unix platform. NAM0T2 can be
compiled with either ANSI (ANSI, 1989) or traditional
(Kernighan and Ritchie, 1978) C compilers.
To increase the capabilities for constructing and
manipulating structures, NAMOT2 features a usermodifiable library for base, unit, helix, sugar and phosphates. Currently, this library features 29 DNA base pairs
deduced from non-empirical calculations (Hobza and
Sandorfy, 1987) plus one triplex (T-A-T; Radhakrishnan
and Patel, 1994) and one quadruplex (G-G-G-G; Smith
and Feigon, 1992). This library enables users to build their
customized environment in modeling both regular and
26
unusual structures. Besides being more versatile in
modeling nucleic acid structures, NAMOT2 also features
added graphic capabilities. Using a specifier language,
users can selectively display, color and/or space-fill the
molecule. The exported graphic image can be in
PostScript™, PPM or TIFF™ formats.
Using NAM0T2 as a tool, complete procedures
(structural recipes) for constructing different structural
motifs for nucleic acids can be developed. One of the
advantages for using structural recipes is that this method
uses less information to store different structures. Motifs
can also be assembled together to form new structures.
Due to the fact that all parameters used by NAM0T2 to
construct nucleic acid structures have intuitive physical
meanings, it is likely that one can easily develop the recipe
for a new structure by modifying a recipe in the existing
pool. All the recipes for the non-canonical structures of
NAMOT2—a redesigned nucleic acid modeling tool
nucleic acid molecules being modeled in this paper are
available through the NAMOT WWW site.
DNA-ditercalinium
complex
In our earlier work (Tung and Carter, 1994), we
demonstrated the capability of NAMOT to generate a
cavity inside a DNA duplex for drug intercalation. As
NAM0T2 is able to dock small ligands against the nucleic
acid molecule, we modeled the structure of a D N A ditercalinium complex. Ditercalinium is an anticancer
drug that binds to DNA by dis-intercalation. Instead of
inhibition of DNA replication or transcription processes,
ditercalinium induces malfunction of DNA repair systems
(Lambert et al., 1990) that in turn causes cell death. The
structure of the DNA-ditercalinium complex was solved
to 1.7 A resolution using X-ray crystallography (Gao et
al., 1991; Williams and Gao, 1992).
We used the crystal structure of the drug in the complex
(DDD030 in NDB; Berman et al., 1992) as the structure
for ditercalinium in this study. DNA molecule for this
study is an eightmer, [d(CGCGCGCG)] 2 , in a canonical
B-conformation (Arnott and Hukins, 1972). Two intercalation sites were introduced to the oligomer at C 3 -G 4
Fig. 2. The modeled structure of the DNA-ditercalinium complex
superimposed with the structure solved using X-ray crystallography. The
two complexes were aligned using only the atoms in ditercalinium. The
modeled structure is drawn in thick lines while the crystal structure is drawn
in thin lines. The two DNA structures have a RMS difference of 1.0 A.
and C 5 -G 6 . The intercalation sites were generated by
simply extended the rise (Dz) of the base-pair step to 6.8 A
and reduced the twist (tt) to 20°. In the docking procedure,
both the drug and the DNA molecules were treated as
rigid bodies. The ditercalinium was docked to DNA from
the major groove side of the molecule. The objective of the
procedure is to dock the drug molecule into the
intercalation sites without introducing steric clashes. For
the purpose of monitoring the surface contacts, during the
docking process, both the DNA and the drug molecules
were displayed using space-filling models. Surface matching between the two molecules were achieved by simply
viewing the structure of the complex while rotating and
translating the drug molecule with respect to the DNA
molecule. The modeled structure of the complex (as shown
in Figure 2), when compared to the crystal structure (using
only the middle four base pairs of the DNA and the
ditercalinium molecule), has a RMS difference of 0.7 A.
DNA cube
Because of the nature of the molecule that the complementary strands form a double helical structure, plus the
fact that the chemistry for synthesis is straightforward and
well developed, DNA is an ideal material for nanotechnology. The proposed applications of DNA as biomaterial
(Seeman, 1991) include: the construction of macromolecular zeolites, the assembly of new catalysts, delivery of
drugs and the assembly of molecular electronic devices.
One such effort towards the goal of developing DNA into
useful biomaterials is the design and synthesis of a DNA
molecule with the connectivity of a cube (Chen and
Seeman, 1991). Each side of this cube is a 20 base-pair
duplex. The connectivity of the molecule was confirmed
using both systematic digestion and gel electrophoresis
analysis.
Given the lack of experimental information available
for the conformation, we choose to model this DNA
molecule as a square cube to take advantages of the
symmetry relationship existing in both the edges and the
vertices of the cube. Each of the eight vertices is modeled
as an orthogonal three-way junction, while each of the 12
sides is modeled as a regular double helix in the Bconformation.
The template structure of an orthogonal three-way
junction is constructed. For the purpose of taking
advantage of the symmetry relationship, the junction is
built with the vertex lying on the origin and the helical axes
of the three arms coinciding with the x, y and z axes. To
satisfy these constraints, each of the three base pairs
forming the junction is allowed to move with only one
translational and one rotational degree of freedom as
shown in Figure 3(b). Our goal is to find the conformation
27
E.S.Carter, II and C.-S.Tung
Recipe for constructing
an orthogonal 3-way junction
set WCAdfirsI on
generate d d b c
write pdh eg
close
generate (I (I I) I
write pdh la
close
generate d d l> a
write pdb at
close
load pdb rigid eg
load pdh rigid ta
load pdh rigid at
rolorig I 2 ISO.
rolorig I I 91).
rolorig I 3 -"Ml.
rotorig I 1 -32.2
trans I H.S4 I). 0.
rotorig 2 2 I HO.
rotorig 2 1 90.
rolorig 2 2 67.8
trans 2 0.11.03 0.
rotorig 3 3 -49.2
Irans300 12.75
write pdb 3-way-jc
•.V,
~A'
V;
Fig. 3. The structure of the vertices in the cube. The conformation of the
three base-pairs that form the three-arm junction of a vertice is shown in
(a). The phosphate linkages between the pairs of sugar are shown as
dashed lines. The arrangement of the eight vertices such that the 12 sides
of cube are approximately equal is shown in (b).
of the three base pairs such that the distances for the three
phosphate linkages between the bases are optimal (3.5 A;
Tung, 1993). With only six degrees of freedom, a short run
of Metropolis Monte Carlo simulation (Metropolis el al.,
1953) quickly finds the conformation of the three base
pairs that optimize the distances for the phosphate
linkages. The resultant translational and rotational
parameters for the three base pairs are listed in Table I.
The complete procedure for using NAMOT2 to construct
an orthogonal three-way junction is treated as a structural
recipe shown in Figure 3(a). This recipe can be used to
construct all the vertices in the cube.
One important feature of this three-way junction is that
the three sides of the junction are not the same. Both the
amount of translations and rotations for the three basepairs that form the junction are not identical. If a, b and c
Fig. 4. The modeled structure of a orthogonal DNA cube in atomic detail.
are the distances from the origin to the three base-pairs on
the x, y and z axes respectively, we have c > b > a. Since
a + c is approximately equal to b + b, we were able to
arrange the eight vertices such that the 12 sides of the cube
are approximately equal, as shown in Figure 3(c).
After the structures of all the eight vertices are
assembled, the structure of the 12 duplexes that connect
the eight vertices can be constructed. Due to the
differences in the rotation of the three base-pairs in the
junction, the total twist angles for the 18 base-pairs that
connect the base-pairs in the vertices to form edges of the
cube are different. A small adjustment has to be
introduced to the twist angles of these helices such that
the connections between the helices and junctions are
possible. The four duplexes that connect the b A- b arms of
the junctions have a twist angle of 35.5 A, while the
remaining eight duplexes have a twist angle of 37.5 A. The
final structure of the modeled DNA cube is shown in
Figure 4.
Table I. Parameters for making an orthogonal three-way junction
Base-pair a
Base-pair b
Base-pair c
Six-arm junction
Rr 180
R\ 90
R.-90
/?,. 180
Rx 90
Rr 67.8
R.-49.2
Rx -32.2
T\3.84, 0, 0
TO, 11.03,0
TO ,0, 12.75
The six-arm junction is yet another structural motif that
can be used as a biomaterial in nanotechnology. This
structural motif can serve as the connecting part in a cubic
lattice. It was shown that the six-arm junction is not as
stable as junctions containing three or four arms (Wang el
al., 1991). Longer arms (16 bp) were required to make a
stable six-arm junction. A six-stranded molecule (JYL6G,
as shown in Figure 5) that forms a six-arm junction was
constructed by Wang et al. (1991). Based on Ferguson
Base-pair a, perpendicular to the .Y-axis; base-pair b, perpendicular to the
j'-axis; base-pair c, perpendicular to the z-axis; Rx, rotation with respect
to the .v-axis; Rv, rotation with respect to the v-axis: R:, rotation with
respect to the z-axis; 7^ translation.
All translations are in Angstroms and all rotations are in degrees.
28
NAMOT2—a redesigned nucleic acid modeling tool
analysis, indicates that the backbone of this particular
strand follows a continuous helical conformation at the
junction.
Without experimental data to indicate the exact
arrangement of the six arms in the molecule, we chose to
model the molecule as an orthogonal six-arm junction.
The proposed structure of the six-arm junction (as shown
in Figure 6a) has the four arms (I, II, III and IV) lying on a
plane, while arms V and VI run perpendicular to the plane.
Arms I, II, III and IV form an open, four-way junction
with a structure similar to the one proposed by von
Kitzing el al. (1990). Arms V and VI run through the
cavity at the center of the four-way junction. The structure
of the molecule in the vicinity of the junction was
determined such that the stereochemistry is maintained
(i.e. all bonds are intact and no steric clashes observed).
The modeled structure of the six arm junction is shown in
Figure 6b.
Curved DNA
Fig. 5. The base sequences of the six strands of DNA that form a six-arm
junction as proposed by Wang el al. (1991).
analysis (Ferguson, 1964), it was shown that the six arms
of this molecule (JYL6G) form an extended structure that
share a common central site. The lack of protection of the
strand V at thejunction, as shown in the hydroxyl radical
Whether intrinsic or induced, some DNA molecules do
exist in a curved conformation (Klug el al., 1980; Wu and
Crothers, 1984). Here, we show the modeling of a
superhelical DNA molecule that mimics the nucleosome
structure where the DNA wraps around a histone core in
a left-handed superhelical conformation. The basic
approach of modeling a superhelical molecule involves
the following steps, (i) Generating a helical curve using a
specified pitch and diameter. In the case of nucleosomal
III
II
V
a
u
Fig. 6. The structure of the modeled six-arm junction. The arrangement of the six strands of DNA in three-dimensional to form thejunction is shown in
(a). The atomic modeled structure of the six-arm junction is shown in (b).
29
E.S.Carter, II and C.-S.Tung
References
Fig. 7. A superhelical DNA that mimics a 146 bp nucleosomal DNA
wrapped around a histone core.
DNA, we used a 28 A pitch and a 106 A diameter as
observed in crystal data, (ii) Digitizing the curve into 3.4 A
segments to represent the base-pair steps in the molecule,
(iii) Mapping the segments into a set of base-pair
parameters for generating the structure of all bases in
the molecule, (iv) Completing the molecular structure by
constructing the sugar-phosphate linkages. The details of
the procedure for generating a curved DNA molecule that
follows any smooth spatial curves (including the superhelices) will be published elsewhere. The complete
modeled structure of a 146bp nucleosomal DNA is
shown in Figure 7.
Discussion
Based on a new set of reduced coordinates, we have
developed our nucleic acid modeling tool (NAMOT) into
a new redesigned software: NAMOT2. NAMOT2 is a
general tool capable of modeling both regular and noncanonical nucleic acid molecules. The four examples
shown above have demonstrated the versatility of
NAMOT2.
NAMOT2 is written in C/Xlib. With the existence of
both Motif™ and Xview™ interfaces, NAMOT2 is
portable at most Unix workstations. A WWW site
(http://namot.lanl.gov) has been established to host all
the information (e.g. demos, update information, an
official copy of the user manual, etc.) pertaining to
NAM0T2. NAM0T2 resides on an anonymous ftp site,
namol.lanl.gov.
Acknowledgements
We would like to thank Dr Angel Garcia for providing us with
subroutines to generate sugar structures and to align molecules based
on minimum RMS difference. Dr Dikeos M.Soumpasis of the Max
Planck Institute and Dr Gerhard Hummer are acknowledged for the
collaboration in developing the set of reduced coordinates used in
NAMOT2. Ms Marilyn L.Kwan is acknowledged for her help in
constructing the mismatch library. Mr Reinhard Klement of the Max
Planck Institute and Dr Craig T.Martin of the University of
Massachusetts provided us with useful suggestions and comments
pertaining to NAM0T2 usages. This work is supported by the US
Department of Energy.
30
ANSI (1989) Programing Language C, X3.159-1989.
Arnott.S. and Hukins,W.L. (1972) Biochem. Biophvs. Res. Comnnm.. 47,
1504-1509.
Berman,H.M., Olson,W.K., Beveridge,D.L, Westbrook.J., Gelbin.A.,
Demeny,T., Hsieh.S.-H., Srinivasan.A.R. and Schneider.B. (1992)
Biophvs.J., 63, 751-759.
Bresenham,J.E. (1965) IBM Syst. J., 4, 25-30.
Chen,J. and Seeman,N.C. (1991) Nature, 350, 631-633.
Ferguson,K..A. (1964) Metabolism, 13, 985-1002.
Gao.Q., Williams,L.D., Egli,M., Rabinovich.D., Chen,S.-L., Quigley,
G.J. and Rich, A. (1991) Proc. Nail. Acad. Sci. USA, 88, 2422-2426.
Kernighan.B.W. and Ritchie.D.M. (1978) The C Programming Language, Prentice Hall, Englewood Cliffs, NJ.
Klug,A., Rhodes.D., Smith,J., Finch.J.T. and Thomas.J.O. (1980)
Nature, 287, 509-516.
Lambert,B., Segal-Bendirdjian,E., Esnault.C, Le PecqJ.B., Roques,
B.P., Jones.B. and Yeung,A.T. (1990) Anti-Cancer Drug Des., 5, 4 3 53.
Metropolis.N., Rosenbluth.A.W., Rosenbluth,M.N., Teller.A.H. and
Teller.E. (1953) J. Chem. Phvs., 21, 1087-1092.
PorterJ.K. (1978) Comput. Graphics, 12, 282-285.
Radhakrishnan,I. and Patel,D. (1994) J. Mol. Biol., 241, 600-619.
Sayle.R. and Bissell,A. (1992) In Proc. 10th Eurographics UK 1992
Conference, University of Edinburgh, Scotland, April 1992.
Seeman,N.C. (1991) NAMOTechnology, 2, 149-158.
Smith.F.W. and Feigon,J. (1992) Nature, 356, 164-168.
Soumpasis,D.M. and Tung,C.-S. (1988) J. Biomol. Struct. Dvn., 6, 397420.
Tung,C.-S. and Carter.E.S., II (1994) Comput. Applic. Biosci.. 10, 427433.
Tung,C.-S., Soumpasis,D.M. and Hummer,G. (1994) J. Biomol. Struct.
Dyn., 11, 1327-1344.
von Kitzing,E., Lilley.D.M.J. and Diekmann.S. (1990) Nucleic Acids
Res., 18,2671-2683.
Wang,Y., Mueller,J.E., Kemper,B. and Seeman.N.C. (1991) Biochemistry, 30, 5667-5674.
Williams.L.D. and Gao,Q. (1992) Biochemistry, 31, 4315-4324.
Wu,H.-M. and Crothers.D.M. (1984) Nature, 308, 509-513.
Received on June 27. 1995; accepted on August 31. 1995