Download Watermarking sexually reproducing diploid organisms

BIOINFORMATICS APPLICATIONS NOTE
Vol. 24 no. 17 2008, pages 1961–1962
doi:10.1093/bioinformatics/btn342
Genetics and population analysis
Watermarking sexually reproducing diploid organisms
Dominik Heider, Daniel Kessler and Angelika Barnekow∗
Department of Experimental Tumorbiology, University of Muenster, Badestrasse 9, 48149 Muenster, Germany
Received on April 30, 2008; revised and accepted on July 2, 2008
Advance Access publication July 3, 2008
Associate Editor: Martin Bishop
ABSTRACT
Summary: DNA watermarks are used for hiding messages or
for authenticating genetically modified organisms. Recently, we
presented an algorithm called DNA-Crypt for generating DNAbased watermarks that can be integrated into the genome
by using the characteristics of the degenerative genetic code.
DNA-Crypt generates the watermark by replacing single bases
and thus creating synonymous codons that encrypt the hidden
information. Mutations within the integrated DNA sequence can
be corrected using several mutation correction codes, to keep
the hidden information intact. This method has successfully
been tested in asexually replicating organisms like bacteria
or yeast, where the watermark is duplicated with every cell
division. It has been shown that DNA watermarks produced by
DNA-Crypt do not influence the transcription or translation of
a protein. In sexually reproducing diploid organisms, additional
problems can occur, e.g. recombination events can destroy hidden
information. Using population predictions as well as statistical
analyses we identified a coupled Y-chromosomal/mitochondrial
DNA watermarking procedure as the most appropriate for diploid
organisms. We developed a mitochondria adapted version of
DNA-Crypt, which is called Project Mito that can be used in
combination with the original program.
Availability: http://www.uni-muenster.de/Biologie.NeuroVer/
Tumorbiologie/DNA-Crypt/index.html
Contact: [email protected]
Password: WWUTB
Requirements: Java 5.0 or higher
Supplementary information: Supplementary data are available at
Bioinformatics online.
1
INTRODUCTION
DNA watermarks have recently been used for authentication or
identification of genetically modified organisms (GMOs) (Heider
and Barnekow, 2007). It has been shown in silico, in vitro and in vivo
that these watermarks do not affect the translation of mRNA and that
the resulting protein is functionally intact (Heider and Barnekow,
2007, 2008).
Watermarks in asexually reproducing organisms are passed on
with every cycle of cell division. The only mechanism by which a
watermark can be lost is in the case of mutation events, which can
be avoided using mutation correction codes (Heider and Barnekow,
2007).
∗ To
whom correspondence should be addressed.
Using sexually replicating organisms instead of asexually
dividing individuals new problems evolve for DNA watermarking.
Diploid chromosome sets represent a new challenge. In
addition, recombination and crossover events can destroy
integrated watermarks.
2
APPROACH
Population predictions revealed that coincidental mating with a
genetic drift of non-watermarked alleles leads to a reduction or
loss of watermarks. An autosomal transmitted watermark can be
disrupted from one generation to the next by recombination events.
Y-chromosomal linked heredity leads to a male watermarked line,
female offspring cannot be watermarked using the Y-chromsome.
In contrast to the Y-chromosomal heredity, a mitochondrial
transmissed watermark leads to a female watermarked line, as
mammalian mitochondrial DNA (mtDNA) is mainly maternally and
only 0.01—0.1% paternally inherited (Schwartz and Vissing, 2003).
To combine the positive effects of all transmission modes and
to suppress the negative impacts it is necessary to link different
watermark constellations. An X-/Y-chromosomal linked inheritance
leads to a reduction of watermark containing individuals starting
with the third generation (Fig. 1). A Y-chromosomal linked heredity
combined with a mitochondrial transmission would be the best
alternative to preserve the watermark (Fig. 1).
Both the X-/Y-chromosomal linked transmission of watermarks
and the Y-/mitochondrial mode of transmission result in a decreasing
number of watermarked individuals from one generation to the next
regarding the whole population (limn→∞ = 0).
The survival of the watermark using an X-/Y-chromosomal
linked procedure depends on the 50% probability to transmit the
watermarked X chromosome from mother to son. In the worst case
constellation (with respect to the maintenance of the watermark), the
watermarked X chromosome is never inherited and therefore, the
worst case of an X-/Y-chromosomal linked watermark procedure
is the minimum number (R) of watermarked individuals within a
generation n:
1
Rw (n) = n
(1)
2
and in the whole population (G)
n+10
,n ≥ 5
Gw (n) = n
k
k=1 2
(2)
Compared to the X-/Y-chromosomal mode of transmission, the
Y-/mitochondrial-based transmission of watermarks yields a partial
stabilization of a female and a male watermarked line, because
© The Author 2008. Published by Oxford University Press. All rights reserved. For Permissions, please email: [email protected]
1961
D.Heider et al.
Fig. 1. Transmission of watermarks. (A) X-/Y-chromosomal transmitted
watermark; (B) Y-chromosomal/mitochondrial transmitted watermark.
Watermarked chromosomes and mitochondria (*) are bold and underlined.
a male individuum always inherits the watermarked Y chromosome
from his father and every offspring always inherits the watermarked
mt DNA from its mother, which leads to the following number (R)
of watermarked individuals in generation n:
3
R(n) = n
2
and
(3)
4+3×(n−2)
,n ≥ 5
(4)
n
k
k=1 2
in the whole population (G).
The detailed calculations of the formulas 1 to 4 can be found in
the Supplementary material.
Thus we recommend the use of Y-/mitochondrial inheritence for
more stably watermarking sexually reproducing organisms.
As mtDNA differs from nuclear DNA in some aspects,
watermarking procedures have to be adjusted to these conditions.
Today, most of the genetic information needed for mitochondrial
function is coded within the nucleus and the gene products are
transported into mitochondria (Gabaldon and Huynen, 2004; Kessler
et al., 2007). The mtDNA itself still encodes some components of the
mitochondrial translation machinery (rRNA and tRNA) as well as
structural components of the oxidative chain (Garesse and Vallejo,
2001). The tRNA molecules are necessary to use a mitochondriaspecific genetic code that slightly differs from the universal code. In
addition, policistronic regions offer a special challenge. The mtDNA
displays a higher mutation rate compared to cellular genes, which
emphasizes the necessity for a mutation correction code (Heider and
Barnekow, 2007).
In an in silico analysis we identified a region within the human
mtDNA, which is suitable for inserting a watermark. This region
G(n) =
1962
within the Cytochrome c oxidase subunit I gene spans the range
from 5904–7445 bp containing no overlapping gene regions. The
watermarking algorithms had to be modified to meet the special
requirements of mtDNA. Therefore, we developed a program called
Project Mito for creating mitochondrial watermarks. Project Mito is
derived from the original DNA-Crypt and can be used in combination
with the DNA-Crypt algorithm. The binary file, which is thought to
be encrypted into DNA is first modified by a mutation correction
code, the Hamming-code, to correct mutations within the DNA
sequences. A header, containing the length of the original file,
is produced and linked to the modified binary sequence. The
composition of the header and the file is translated into a DNA
sequence, which is embedded into the mtDNA using synonymous
codons as described (Heider and Barnekow, 2007). To extract the
information within the DNA watermark, Project Mito reads out the
first synonymous codons within the target gene and reconstructs
the binary header, which notifies Project Mito about the length of
the intergrated information. Possible mutations occuring in the DNA
sequences are corrected by Project Mito itself and the DNA sequence
is translated into the binary file.
Users should follow the instructions of the manual, which is
provided on the project’s homepage.
Using Project Mito we were able to design a first potential
watermark sequence that could be inserted into the above described
region within human mtDNA. Analyses of the Cytochrome c oxidase
subunit I gene using Project Mito shows that the gene can carry a
maximum information of 60 bytes. In our experiments the integrated
watermark sequence contains the message ‘Project Mito’. A pairwise
sequence alignment displays an identity of about 95% between the
watermark sequence and the original sequence. These differences in
the DNA sequence do not alter the resulting protein as described.
Taken together we presented and tested in silico a watermarking
procedure that can be used for authentication of genetically modified
mammalian cells or even in animal breeding processes based on a
Y-/mitochondrial transmission.
ACKNOWLEDGEMENTS
The authors thank Daniel Tenbrinck (Institute of Computer Science,
University of Muenster) for verifying the statistical calculations.
This work is part of the PhD thesis of D.H.
Conflict of Interest: none declared.
REFERENCES
Gabaldon,T. and Huynen,M.A. (2004) Shaping the mitochondrial proteome. Biochim.
Biophys. Acta, 1659, 212–220.
Garesse,R. and Vallejo,C.G. (2001) Animal mitochondrial biogenesis and function: a
regulatory cross-talk between two genomes. Gene, 263, 1–16.
Heider,D. and Barnekow,A. (2007) DNA-based watermarks using the DNA-crypt
algorithm. BMC Bioinformatics, 8, 176.
Heider,D. and Barnekow,A. (2008) DNA watermarks: a proof of concept. BMC Mol.
Biol., 9, 40.
Kessler,D. et al. (2007) The DNA binding parvulin par17 is targeted to the mitochondrial
matrix by a recently evolved prepeptide uniquely present in hominidae. BMC Biol.,
5, 37.
Schwartz,M. and Vissing,J. (2003) New patterns of inheritance in mitochondrial disease
Biochem. Biophys. Res. Commun., 310, 247–251.