From: AAAI Technical Report SS-99-01. Compilation copyright © 1999, AAAI (www.aaai.org). All rights reserved.
A Graphical Technique for Preliminary Assessment of Effects on
DNA Sequences from Toxic Substances
A Nandy1, C Raychaudhury1 and S C Basak2
1
Computer Division, Indian Institute of Chemical Biology, 4 Raja S C Mullick Road, Calcutta 700 032, INDIA. Email: anandy@giascl01.vsnl.net.in
2
Natural Resources Research Institute, University of Minnesota, 5013 Miller Trunk Highway, Duluth, Minn. 55811,
USA. E-mail: sbasak@sage.nrri.umn.edu
Abstract
Some toxic substances are known to bind
preferentially to specific segments in a DNA sequence
while others such as copper preferentially affect DNA
constituents such as guanine. In view of the complex
nature of large molecules such as DNAs and the
possibility of these toxic chemicals affecting
homologous segments, techniques for identifying the
possible DNA sites that may be affected assume
significance. While chemical and laboratory tests
remain the basic tools of such investigations, rapid
computer-based searches that may help to minimise
the possible locations of such toxic damages would
be a useful supplement.
We have observed using a two-dimensional
Cartesian representation
technique
for DNA
sequences that individual genes have unique graph
signatures. These arise from the base arrangements and
distribution in the coding sequences that are specific
to each gene type and therefore have characteristic
graphical representations. For conserved genes the
entire gene including coding and non-coding regions
are seen to have unique shapes. The characteristic
shapes retain their shape similarity for homologous
genes making visual identification possible from a
library of gene graph signatures. Use of automatic
pattern recognition programs can make such
identifications simple and fast and lead to more
efficient scanning of long DNA sequences. A
quantitative indexing scheme recently proposed by u s
estimates the
dispersion
of the
graphical
representation and provides quantitative estimates of
graph similarity. This will help to further narrow
down the possible matches against the library
catalogue.
The basic benefit we derive from a twodimensional graphical technique is that minor
deviations in base sequences do not alter the
characteristic shape and hence homologous sequences
generate shape similarity, which we may term as shape
homology; identification of homologous sequences i n
the normal character-based representation present a
more formidable problem. In the case of effects of toxic
substances on specific DNA sequences, we can trace
the pattern of the sequence segment in the twodimensional
representation,
and
homologous
sequences that have shape homology with this
pattern can be expected to be receptive to such toxic
chemicals. A rapid visual or computer search of a
complete DNA sequence for such specific patterns can
lead to identification of possible sites which can then
be investigated further by more rigorous means. A s
another example of the use of these graphs, high levels
of copper toxicity leading to depletion of guanine
component of a DNA sequence will show up in these
graphs as a compression along the horizontal axis
There are many other information that can be read
off from these graphs that make them a useful tool i n
analysis of DNA sequences; several of these could be
profitably employed in the search for effects of toxic
substances. Systematic differences are seen in the
characteristic shapes for a set of homologous
conserved genes, and these can be used as a guide for
estimating significant changes. Closer inspection of
the graphical shapes provides indications of local
base dominances and evidences of repetitive segments
and therefore possible extent of damages that may
accrue from high toxicity levels. Thus the graphical
technique provides to a first approximation a new and
useful predictive and diagnostic tools.
Introduction
The wide use of different chemical substances and
growing library of new synthetic chemicals have led to
increased possibility of toxic effects on tissues and cells of
living beings (Basak et al 1998). Several toxic substances
are known to cross the cellular barrier and affect intracellular matter; e.g., arsenic toxicity arising from the
ability of arsenite [As(III)] to bind protein thiols (Hu , Su
and Snow 1998), furazolidone or N-(5-nitro-2furfurylidene)-3-amino-2-oxazolidone, one of the members
of the group of synthetic nitrofurans at a concentration level
of 0.5 µg/ml inhibiting DNA synthesis (Maiti et al 1983),
Cr genotoxicity manifesting as gene mutations, several
types of DNA lesions and inhibition of macromolecular
synthesis (Singh et al 1998), a series of synthetic
chemicals such as methyl and ethyl quaternary
pyridiniumtetrahydrocarbazoles active in inhibiting DNA
synthesis in Ehrlich ascites cells (Ferlin et al 1998),
norfloxacin and its inducing metabolites in vitro with S9
rat liver mutagenic to hisG48 strains TA102 and TA104
(Arriaga-Alba et al 1998), airborne particulates inducting
sister chromatid exchanges (SCE) in human tracheal
epithelial cells (Hornberg et al 1998), and others.
While some toxic substances can affect DNA
molecules binding preferentially to specific segments in a
DNA sequence, others like copper preferentially affect DNA
constituents such as guanine. The nature of the interaction
of Cu(II) ions with DNA has been found to be dominant in
two ways which may take place simultaneously at low
ionic strengths of the solvent: type I interaction prevalent
at low values of Cu(II) ions/DNA ratio has the overall effect
of stabilising (Richard Daune and Schreiber 1973) or nondenaturing (Foerster et al 1979) the DNA; type II
interaction is predominant at higher values (>0.3) of the
Cu(II) ions/DNA ratio and has the effect of destabilising
the DNA by initiating breakage of G-C bonds or mutation
or depletion of the Gs (Richard Daune and Schreiber 1973,
Foerster et al 1979). Similarly, several other chemicals are
suspected of causing gene mutations leading to inherited
genetic defects leading to necessity of techniques to
predict and identify such effects.
In view of the complex nature of large molecules
such as DNAs and the possibility of these toxic chemicals
affecting homologous segments, methods for identifying
the possible DNA sequences that may be affected become
significant. While chemical and laboratory tests remain the
basic tools of such investigations, and new techniques are
constantly being innovated (see, e.g., Mitchelmore et al
1998), rapid computer-based searches that may help to
minimise the search for possible locations of such toxic
damages would be a useful supplement. In this paper we
report application of a graphical technique (Nandy 1994)
that can be used for this purpose.
adenosine gets relatively enhanced. This is evident visually
from as low as a 5% depletion of the guanine. In terms of
the graph radius as defined in our earlier paper, we find that
the graph radius for the three plots change from 96.17 for
the unaltered human beta globin gene to 97.61 for a 5%
random depletion of guanine and 100.29 for a 10%
depletion. Thus, a comparison of beta globin gene
sequence from a standard reference sample and from
contaminated samples can readily be tested for random
mutation of bases. Our studies on histones, kinases,
thymidines and other samples of conserved gene sequences
also show similar variations with changes in one or more
bases that may have been induced by the action of toxic
chemicals; the comparison can be extended similarly to any
identified sequence segment.
-250
-150
-50
50
50
C
A
G
Method
0
-50
The method of representing DNA sequences
graphically using a 2-dimensional Cartesian coordinate
system has been explained elsewhere (Nandy 1994, Ray
Raychaudhury and Nandy 1998). The shapes of these DNA
graphs depend on the base distribution in the sequence.
The plot is generated by moving one step in the positive
x-direction for a guanine (G) in the sequence, the negative
x-direction for an adenosine (A), the positive y-direction for
a cytosine (C ) and the negative y-direction for thymine
(T), the succession of such steps producing a shape
characteristic of the sequence. We have shown (Nandy
1994) that for conserved genes such plots are shape similar
thereby making identification of a new sequence of the
family possible rapidly and easily by visual inspection
alone; elsewhere we have shown that one can read off base
preferences and local abundances directly from the shape of
these graphs (Nandy 1995), or identify coding and noncoding sequences (Nandy 1996a). Changes in base
distribution and composition induce changes in the visual
plots of the DNA sequences; for the same gene for different
species we have noticed systematic drifts in the sequence
pattern which has been attributed to evolutionary changes
(Nandy 1996b). This has been estimated quantitatively by
using a new algorithm that sums the co-ordinate values of
the graph points and normalizes to per nucleotide as has
been explained in Raychaudhury and Nandy 1998.
Results and Discussions
As a preliminary exploration of the technique we
have used the complete normal human beta globin gene
sequence inclusive of introns and exons as the base
sequence. A random alteration in the guanine bases in the
sequence was performed programmatically to simulate the
effect of high dose of copper (Cu(III)) toxicity. The plot of
the beta globin gene sequence is seen to shift to the left as
the guanine gets depleted and the abundance of the
-100
-150
3
-200
1
2
T
-250
Fig: Human Beta globin gene from the beta globin
region on chromosome 11 and variations plotted
on the ACGT axes. 1: The normal beta globin
gene sequence; 2: beta globin gene sequence with
5% random mutation in guanine; 3: same with
10% random mutations in guanine.
This method, both quantitatively and in visual
graph system, is also useful in the case of contaminants
that more generally affect pyrimidines or purines. In the
former case the plot will show change in the extent in the
vertical y-axis, while the latter will alter the extent and
spread along the x-axis. Development of a suitable program
for pattern recognition and comparison from a set of
standard library of sequence shapes will automate the
processes and allow this technique to be used as a ready
reckoner to measure the intensity of toxicity of newly
synthesized as well as already available chemicals and
products. Such a technique could be especially useful for
measurement of the toxic effects on life forms where other
clues are not readily available as in the case of mussels and
other aquatic fauna.
Conclusion
We have thus shown that a 2D-graphical method
of DNA sequence representation can be used to provide
visual clues to sequence mutation and alteration brought
on by toxic substances like Cu(III) in relatively high
concentrations. This technique can be used to measure the
extent of toxicity by quantitative estimation of the degree
of mutation so as to set a comparison standard.
Development of a pattern recognition software to scan and
identify regions of deviations from a reference library of
sequence plots would help in rapid identification of
toxicity effects and quantification of levels of toxicity in
newly synthesized and existing toxic chemicals.
References
Arriaga-Alba M, Barron-Moreno F, Flores-Paz R, GarciaJimenez E,
Rivera-Sanchez R 1998, Genotoxic
evaluation of norfloxacin and pipemidic acid with the
Escherichia coli Pol A-/Pol A+ and the ames test, Arch
Med Res 29(3):235-40
Basak S C, Grunwald G D, Host G E, Niemi G J and
Bradbury S P 1998, A comparative study of molecular
similarity, statistical, and neural methods for predicting
toxic modes of action, Environ Toxic Chem,
17(6):1056-1064
Ferlin M G, Chiarelotto G, Marzano C, Severin E,
Baccichetti F, Carlassare F, Simonato M and Bordin F
1998, Synthesis and biological properties of a new series
of N-pyrido substituted tetrahydrocarbazoles, Farmacol
53(6): 431-437
Foerster W, Bauer E, Schitz H, Akimenko N M,
Minchenkova L E, Evolokimov Y M and Varshakovsky
Y M 1979, Bioploymers 18: 625
Hornberg C, Maciuleviciute L, Seemayer NH, Kainka E
1998, Induction of sister chromatid exchanges (SCE) in
human tracheal epithelial cells by the fractions PM-10
and PM-2.5 of airborne particulates, Toxicol Lett 9697:215-20
Hu Y, Su L, Snow ET 1998, Arsenic toxicity is enzyme
specific and its affects on ligation are not caused by the
direct inhibition of DNA repair enzymes, Mutat Res
408(3):203-18
Maiti, M, Ghosh, S, Chatterjee, A and Chatterjee, S N
1983, Thermal stability of DNA interacting with
furazolidone and Cu(II) ions, Z Naturforsch, 38c: 290293
Mitchelmore C L, Birmelin C, Livingstone D R and
Chipman J K 1998, Detection of DNA strand breaks in
isolated mussel (Mytilus edulis L.) digestive glans cells
using the ÒCometÓ assay, Ecotoxicol Environ Saf
41(1):51-58
Nandy, A 1994, A new graphical representation and
analysis of DNA sequence structure: I. Methodology
and
Application
to Globin Genes, Current Sc
66(4):309-314
Nandy A and Nandy P 1995, Graphical analysis of DNA
sequence structure: II. Relative abundances of
nucleotides in
DNAs, gene
evolution and
duplication, Current Sc 68(1), 75-85.
Nandy, A 1996a: Two
dimensional
graphical
representation of DNA sequences and intron-exon
discrimination in intron-rich sequences, Comput Appl
Biosci, 12(1):55-62
Nandy, A 1996b: Graphical analysis of DNA sequence
structure: III. Indications of evolutionary distinctions
and characteristics of Introns and Exons, Current Sc
70(7): 661-668
Ray, A, Raychaudhury, C and Nandy, A 1998, Novel
Techniques of Graphical Representation and Analysis of
DNA Sequences - A Review, J Biosc 23(1):55-71
Raychaudhury, C and Nandy, A 1998, ÒIndexing Scheme
and Similarity Measures for Macromolecular SequencesÓ,
J Chem Info Comput Sc, in press
Richard H, Daune M and Schreiber J P 1973, Biopolymers
12: 1
Singh J, Carlisle D L, Pritchard D K, Patierno S R 1998,
Chromium-induced
genotoxicity
and
apoptosis:
relationship to chromium carcinogenesis, Oncol Rep
5(6): 1307-1318
Tinkler J, Gott D and Bootman J, Risk assessment of
dithiocarbamate accelerator residues in latex-based
medical devices: genotoxicity considerations 1998, Food
Chem Toxicol 36(9-10):849-866