Identifying CpG Islands: Sliding Window and Hidden Markov Model Approaches - Mathematical Concepts and Methods in Modern Biology - page 266

which give rise to sperm and eggs, must express their housekeeping genes, promoters
of housekeeping genes would retain their CpG dinucleotides and appear as CGIs [ 7 ].
However, CGIs are not confined to the promoters of housekeeping genes. CGIs
are found in the promoters and protein coding regions (exons) of about 40% of
mammalian genes [ 8 ]. Since CGIs seem to be located at the promoter regions of many
known genes, identifying CGIs may be a useful method for identifying new genes.

9.1.3 DNA Methylation in Cancer
DNA methylation is a powerful mechanism for gene silencing. Genes which are
methylated and repressed in vertebrate organisms are effectively shut off. For example,
hemoglobin genes should be methylated (and silent) in skin cells but unmethylated
(and actively expressed) in red blood cell precursors.
The generation of cancer requires multiple changes in gene structure and function.
A single mutation is not enough to transform a normal cell into a cancerous cell;
between three and seven mutations or other genetic insults are required [ 9 , 10 ]. The
affected genes—those which are involved in the progression from normal cell to
cancer cell—fall into two different categories: oncogenes and tumor suppressors.
Oncogenes must be activated, and tumor suppressors must be inactivated, in order
for cancer to develop. Tumor suppressors may be inactivated through mutation or
through gene silencing.
Mutations of tumor suppressors were identified first. In hereditary retinoblastoma,
a cancer of the retina, affected children have inherited a defective copy of the gene
for the tumor suppressor Rb from one of their parents. Tumors often arise in both
eyes of these children following the loss of the second copy of Rb (inherited from
the other parent) [ 11 ]. Additional tumor suppressor gene mutations were discovered
by examining the DNA of many different tumors and comparing that DNA to that of
normal tissue. For example, the tumor suppressor p16 , a cell cycle control protein,
was found to be deleted or to have sufferedmutations inmany different types of cancer
[ 12 ]. Tumor suppressor genes do not need to be mutated to contribute to the genesis
of cancer, however. They merely need to be silenced. If the CGIs in the promoters of
p16 or Rb genes are inappropriately methylated and the genes are turned off, then the
cell in question will be one step closer to the development of a cancer.

9.2 QUANTITATIVE CHARACTERISTICS OF THE CpG ISLAND
REGIONS AND SLIDING WINDOWS
ALGORITHMS
CGIs were first characterized quantitatively 25 years ago by Gardiner-Garden and
Frommer [ 13 ] and their definition is still widely in use today. The terms percent
combined C

+ G content ( % C + G ) and observed over expected CpG ratio

( introduced by the authors provide a way to identify genomic regions

with higher frequencies of C and G nucleotides and CpG dinucleotides. The % C

O / ECpG )

+ G