Chipping away the mystery of drug responses
JC Rockett
162
lated with, observed effects. Interani-
mal variation is an important con-
sideration in the selection of such dose
levels. Preliminary studies by the
authors indicated that interanimal
variation exceeded experimental vari-
ation, and this was used to support
pooling of samples for further analysis.
Although this approach may be appli-
cable for mechanistic studies, there
would appear to be less value from the
pharmacogenomic standpoint. Indi-
vidual variation, which exists even in
inbred strains, lies at the very heart of
pharmacogenomics, and should be
embraced rather than smoothed away.
Interestingly, the authors also pointed
out that some genes were hypervari-
able among individual animals. Poss-
ible reasons for this were not offered,
but are most likely caused by polymor-
phisms in the gene regulatory regions.
One of the challenges for pharmaco-
genomics, therefore, is to identify and
characterize the source of such differ-
ences and how they relate to drug
metabolism.
The most common polymorphism is
the single nucleotide polymorphism
(SNP), and it has been estimated that
there may be as many as 200–300 000
of them in the protein coding
sequences of the human population. A
large school of thought supports the
contention that characterizing these
SNPs is the key to deciphering the gen-
etic basis of complex disease. However,
a more realistic first step for pharmaco-
genomics is to properly decipher drug
response at the gene level. For
example, the CYP family of genes is of
central importance in drug meta-
bolism, and several members have
gene sequences over 90% homolo-
gous. Furthermore, many of the CYP
genes have multiple allelic variants
that are responsible for a high pro-
portion of observed variations in drug
response. It is therefore a significant
challenge to characterize which of
these genes and alleles are regulated in
response to specific drug exposures.
RT-PCR is probably the best current
PCR is a relatively low-throughput
technology and researchers are seeking
ways to replace it with higher through-
put microarrays. However, there is
concern that the widely used cDNA
arrays are unable to discriminate such
closely related genes. There is a move,
therefore, towards arrays composed of
short (50–80 base) oligonucleotides.
These can, with careful selection, be
constructed from unique regions of
the chosen genes to theoretically pro-
vide good discriminatory power. How-
ever, since the overall efficacy of this
length of probe is still under scrutiny,
the Affymetrix platform of multiple,
short oligonucleotides per gene
remains the current gold standard.
Gerhold’s paper demonstrates the
ability of Affymetrix chips to dis-
tinguish members of the CYP family
with Ͼ90% sequence homology. This
kind of resolving power will be hard to
beat without a jump in technology,
and may permit Affymetrix to develop
a very firm grip on this area of the
market.
Those working with arrays learn
quickly that deciphering the data is
the key to the treasures that lie hidden
within, and that the key is hard to
turn. Not surprisingly then, these
industrially based researchers took the
logical step of designing a custom dat-
abase to hold and mine their data.
This is all well and good for private
companies with proprietary concerns,
but raises the wider issue of just how
much useful information remains
locked up in the archives of such priv-
ate entities, untapped by their curators
because of lack of knowledge or inter-
est. The development of a national or
international gene expression database
is currently a difficult prospect due to
such factors as lack of cross-platform
concordance, the disarrayed state of
gene nomenclature and annotation,
and a jealous desire to guard data from
competitors. However, it can be
argued that it would be in the best
interests of medical research for gene
expression laboratories to pool data in
a central database(s), perhaps after
they have extracted information that
interests them. In this vein, the Euro-
pean Bioinformatics Institute has set
up the Microarray Gene Expression
Database group,11,12 a cross-organiza-
tion entity whose goal is ‘to facilitate
the establishing of gene expression
data repositories, comparability of
gene expression data from different
sources and interoperability of differ-
ent gene expression databases and
data analysis software’. It should also
be of no surprise if pharmaceutical
companies soon start scrambling to
form strategic alliances with one
another in order to reduce experi-
mental costs and mine the untapped
wealth of one another’s data.
Although achieving the goal of
using gene expression profiles for
rapid and early screening of new drugs
is still a little way off, the data pro-
vided by Gerhold et al and others like
it is clearly encouraging, and indicates
that the time when this approach will
be used routinely to aid in drug devel-
opment is not that far away.
ACKNOWLEDGEMENTS
This document has been reviewed in accord-
ance with the US Environmental Protection
Agency policy and approved for publication.
Mention of trade names or commercial pro-
ducts does not constitute endorsement or rec-
ommendation for use. Thanks to Drs Mitch
Rosen (US EPA) and Ian Dix (AstraZeneca) for
scientific review of this manuscript prior to
submission.
DUALITY OF INTEREST
None declared.
Correspondence should be sent to
JC Rockett, Reproductive Toxicology Division
(MD-72), National Health and Environmental
Effects Research Laboratory, US EPA, Research
Triangle Park, NC 27711, USA.
Tel: ϩ1 919 541 2678
Fax: ϩ1 919 541 4017
E-mail: rockett.johnȰepa.gov
1
2
Lazarou J et al. JAMA 1998; 279: 1200–
1205.
Gerhold D et al. Physiol Genomics 2001; 5:
org/cgi/content/full/5/4/161
Nguyen L et al. Drug Metab Dispos 2000;
28: 987–993.
Jelinsky S, Samson L. Proc Natl Acad Sci USA
1999; 96: 1486–1491.
Giaever G et al. Nat Genet 1999; 21: 278–
283.
Scherf U et al. Nat Genet 2000; 24: 236–
3
4
5
6
method
for
discriminating
the
expression of such drug metabolizing
genes, as it provides both quantitative
expression data and can differentiate
allelic variants.10 Unfortunately, RT-
244.
The Pharmacogenomics Journal