Full text of DOI:10.1038/sj.tpj.6500013

Postgenomic prospects of success
A Rane
6
developmental window during which
the pathway is turned on.
human study samples.2,3 Further, the
ability to produce transgenic and
knock-out mutants and the possibility
to create congenic strains facilitate
functional studies of mutated genes.
Similarly, Drosophila provides an extra-
ordinary opportunity for ‘functional
genomics’ analysis of complex mech-
anisms like those involved in neural
development. One hundred years of
classical genetics has provided an
unprecedented collection of mutant
alleles of many different loci affec-
ting development and behavior
(www.flybase.org). Further, as can be
done with the mouse genome, tar-
geted mutations can be introduced
into the Drosophila genome by hom-
ologous recombination.4
Although experimental species carry
great value, final evidence for the
involvement of these genes in human
diseases must come from extensive
epidemiological studies, preferentially
in different populations. Statistical
methods adapting information of mul-
tiple variants and quantitative trait
components of complex diseases are
being developed.5 We still have too
many loci and very few genes for com-
plex traits. The debate over gene
identification strategies is ongoing—
what populations should be studied,
what the optimal strategies for statisti-
cal analyses will be (taking advantage
of both vast genetic data and quanti-
tative phenotype information) and
how any association between a genetic
marker and a clinical phenotype will
finally be validated. Large epidemio-
logical study samples collected from
multiple, different populations will
most probably be the best strategy.
Predicted minor effects of multiple
genes in complex traits will require
efficient collaboration, data sharing
and pooled data analyses among
research groups around the world.
Only in such settings can the fruits of
the genome project be best harvested.
Many protein networks have already
been identified by a variety of biocom-
puting-based strategies, including
phylogenetic profiling, the Rosetta
Stone (or domain fusion) method and
the gene neighbor method, using indi-
cators
identified
via
homology
searches and motif finding.1 All these
strategies, although in need of further
developmental work, will occupy gen-
erations of biocomputing specialists
and provide experimentalists with
novel hypotheses of disease mech-
anisms.
DUALITY OF INTEREST
None declared.
Correspondence should be sent to
L Peltonen, Department of Human Genetics,
UCLA School of Medicine, 695 Charles E
Young Drive South, Suite 6506, Los Angeles,
CA 90095-7088, USA.
GENES IN COMPLEX DISEASES
The great challenge in the future of
genetics lies in identifying the genes
involved in oligo- and polygenic
human diseases. Many diagnostic fea-
tures of complex diseases represent
QTL traits, regulated by several genes
and influenced by lifestyle and
environment. Well-established mouse
phenotypes combined with the rap-
idly evolving mouse sequence will be
of greatest significance in the dissec-
tion of complex traits. In mice, the
positional cloning of a disease gene is
a relatively straightforward approach
due to the ability to backcross the mice
and to restrict genetically the sus-
pected chromosomal regions to an
interval which can be sequenced and
in which the mutation gene identified.
The corresponding gene can then be
analyzed for functional variants in
E-mail: lpeltonenȰmednet.ucla.edu
REFERENCES
1
Eisenberg D, Marcotte EM, Xenarios I, Yates
TO. Protein function in the post-genomic
era. Nature 2000; 405: 823–826.
2
Stoll M et al. New targets for human hyper-
tension via comparative genomics. Genome
Res 2000; 10: 473–482.
3
4
Lusis J. Atherosclerosis. Nature 2000; 407:
233–241.
Rong YS, Golic KH. Gene targeting by hom-
ologous recombination in Drosophila.
Science 2000; 288: 2013–2018.
5
6
Blangero J, Almasy L. Multipoint oligogenic
linkage analysis of oligogenic traits. Genet
Epidemiol 1997; 14: 959–964.
Peltonen L, McKusick VA. Dissecting human
diseases in the postgenomic era. Science
2001; 291: 1224–1229.
ified genes and 3.5 billion base
pairs.
Postgenomic prospects of success in
drug development and
pharmacotherapy
This marks the start of the new era
of functional genomics including
pharmacogenomics. The latter term
denotes the influence of genetic vari-
ation on the disposition of drugs, as
well as the response and adverse reac-
tions to drugs.
A Rane
Karolinska Institutet, Department of Medical Laboratory Sciences and Technology,
Division of Clinical Pharmacology, Huddinge University Hospital, Stockholm, Sweden
FROM GENOTYPE TO
PHENOTYPE
The society cherishes great expec-
tations of the impact of this know-
ledge on the possibilities to find new
ways of identification and pharmaco-
The human genome has now been
mapped by virtue of enormous aca-
demic and financial investments
over the past decade.1,2 We now
know the dimension of the genome
which contains about 30 000 ident-
The Pharmacogenomics Journal

Postgenomic prospects of success
A Rane
7
logical treatment of disease. The possi-
bility to predict protein structures
from the genome gives promise for
determining which drug molecules
they may interact with and what the
function is—an enzyme, a receptor, a
transporter protein etc.3 Proteomics,
the science about the pattern and
function of gene products, will be very
important in the future. However, the
in silico prediction of the function is
not sufficient and complementary in
vitro tests are therefore important. As
the majority of proteins are deeply
integrated with cellular lipid mem-
branes, their function must also be
tested in vivo, in intact cells or in the
intact organism. Modulation of the
expression by transfection and/or anti-
sense nucleotide treatment will help to
confirm the identity of the protein and
its function. However, it must be born
in mind that the gene expression as
well as the function of the gene pro-
duct may be different in different
tissues and organs.
The reductionist research paradigm
is not in itself sufficient to benefit
from the new knowledge. In order to
test new pharmacological principles, it
is mandatory to establish collabor-
ation between molecular genetics and
the physiological, pharmacological
and clinical sciences.
The sequencing of the human gen-
ome has revealed a large number of
points of variation such as single
nucleotide polymorphisms (SNPs).
The vast majority of the 1.42 million
SNPs4 are probably silent without
implications for the phenotype as
most of them occur in non-coding
regions. Even among SNPs in the
coding regions (cSNPs), a substantial
fraction will not lead to changes in the
encoded proteins (‘synonymous’ in
contrast to ‘non-synonymous’ nucleo-
tide substitutions). Therefore, rela-
tively few isolated genomic changes
will have functional implications in a
biomedical or clinical sense.
In the search for pharmacological
modulators of disease-related gene
products it is important first to know
what kind of effects they represent.
The high throughput screening (HTS)
concept utilizes expression models or
bioassays which are useful for screen-
ing of specific effects in vitro, but the
relevance in the intact organism is
often difficult to predict. As pointed
out above, the ultimate proof of con-
cept will not be achieved until the
drug is tested in animal experiments
and in clinical trials. Only then is it
possible to assess the prospects of
pharmacological treatment.
Currently, only about 500 proteins
are targets for pharmacological treat-
ment. Considering the existence of at
least 30000 genes, any scenario to
investigate the majority of these pro-
ducts is fictive today. The long way
from in silico prediction of protein
structure from the genome, via in vitro
HTS expression, preclinical in vivo
studies and further to clinical studies
will require massive financial invest-
ments. It is not plausible that this
approach will pay off substantially
within the next 10 years or so. There-
fore, intelligent searches for relevant
genes are necessary before embarking
on such genetics-based drug develop-
ment programs.
ping of metabolic pathways of new
drug candidates8 and phenotyping
may also be performed in vitro.9 Identi-
ficiation of ethnic differences in drug
kinetics has also led to discoveries of
new allelic variants of drug metabolis-
ing enzyme genes.10
However, the inheritance of the
majority
of
drug
metabolising
enzymes and receptors is more com-
plex. It seems that many disorders and
features are influenced by several
genes as well as by ill-understood
effects of the environment. We know
that genetic traits may be influenced
by environmental factors such as diet,
chemical pollutants, drugs, or by con-
stitutional factors such as age or preg-
nancy.11
A different approach for
identification of the genetic back-
ground of variation in drug metab-
olism or drug action is therefore
required. If the ‘normal’ phenotypic
variants are known, ‘abnormal’
phenotypes in respect of drug
response, or absorption, metabolism,
and excretion of drugs have to be
searched for. With access to large
cohorts of patients it is possible to find
such outliers and to look for associ-
ations between the deviating pheno-
types and specific SNPs or constel-
lations of SNPs. Obviously, the clinical
endpoints, whether they be related to
drug response, non-response, ADRs
etc, need to be very robust and well
defined. This is a critical, often neg-
lected but limiting factor in functional
pharmacogenomics on a clinical level.
The inheritance of most enzymes
and receptors is probably polygenic.
The polygenic large cohort approach
requires that a relevant set of genes is
selected. The human genomic SNP
map4 will help us to select the study
sites in the genes of interest. Non-syn-
onymous cSNPs will be of primary
interest. Sometimes, silent SNPs may
also be of interest,12 or SNPs in the
regulatory regions of the gene.13 How-
ever, it is believed that the concerted
action of several SNPs or specific con-
stellations of SNPs may be more
important than single SNPs for the
phenotype of many disorders and
other features. Thus, it has been dem-
onstrated that certain haplotypes in
the ␤2-adrenergic receptor gene have a
FROM PHENOTYPE TO
GENOTYPE
The identification of outliers in phar-
macological response or adverse drug
reaction (ADR) studies will remain a
useful approach to discover genes of
important drug metabolising enzymes,
receptor systems etc.5 Such obser-
vations have often guided research
into new avenues with discoveries of
new pharmacologic treatment prin-
ciples and polymorphisms.6,7 How-
ever, if several target molecules affect
the disease or drug response, the
chance of identifying the second, third
etc drug target will diminish.
Studies of monogenic traits, for
example in the field of drug metab-
olism have typically been carried out
in panels of phenotypically and geno-
typically defined individuals, eg
‘extensive’ or ‘poor’ metabolisers. This
approach has been successful for map-
What do we do with all this gen-
omic information? The drug industry
has already mortgaged for the expec-
tations to find new and valuable drug
targets, ie enzymes or receptors whose
function is directly or indirectly
related to disease or disease treatment.
www.nature.com/tpj

Postgenomic prospects of success
A Rane
8
Department of Medical Laboratory Sciences
and Technology, Division of Clinical
Pharmacology, Huddinge University Hospital,
SE-141 86 Stockholm, Sweden.
Tel: +46 8 585 810 51
higher predictive value than single
SNPs in the prediction of response to
agonist therapy.14 This complicates
the search for genetic differences
will find useful relationships between
the disease, drug response, etc and the
genomic variation. In the end this
‘neo-Linne´an approach’ may lead to
generation of ideas and hypotheses
that may serve as a starting point of
new, and more traditional hypothesis-
based projects.
between
phenotypes
and
the
Fax: +46 8 585 810 70
E-mail: Anders.RaneȰlabtek.ki.se
interpretation of such differences.
Advanced and efficient technology
for screening of large numbers of SNPs
is now available and includes DNA
microarrays, pyrosequencing tech-
niques, etc. It is mandatory to involve
bioinformatics which is a discipline
that evolved in response to the need
for handling and interpretation of the
massive amount of data.
Functional pharmacogenomics at
the clinical level will be important to
identify known genetic variants of rel-
evance in pharmacotherapeutics such
as safety and response rate. Safety is
one of the mainstays in drug therapy.
New drugs with undesirable or unac-
ceptable ADR profiles are seldom
accepted no matter how effective they
are. The aim is to identify patients at
risk of adverse reactions to drug treat-
ment. The dimension of the problem
of ADRs in drug therapy has recently
been highlighted.15 In the US, ADRs
constitute the 5th–6th most important
cause of death (Ͼ100000 in 1994), and
there were 2.2 million patients with
serious ADRs in US hospitals in 1994.
Thus, ADRs result in enormous costs
to society.
Response rate is another major con-
cern in clinical use of drugs as well as
in drug development. Lack of effi-
cacious treatment is extremely costly.
Identification of non-responders prior
to clinical trials would rationalise and
economise drug development. There-
fore, genetic featuring will also be
employed in the selection of patient
study groups in clinical drug develop-
ment and in routine health care.
Research based on the human genome
will generate massive information on
genetic patterns and variation in a
multitude of genes in relation to the
studied disorders. This is sometimes
denoted as ‘omics’ research.16 ie gener-
ation of data without knowledge in
advance of which bits of information
or which correlations will prove use-
ful. To some extent our studies are
hypothesis-driven whereas other parts
are based on the hypothesis that we
REFERENCES
1
International Human Genome Sequencing
Consortium. Initial sequencing and analysis
of the human genome. Nature 2001; 409:
860–921.
CONCLUSIONS
The mapping of the human genome
now marks the beginning of a never-
ending era of functional genomics
2
3
Venter JC et al. The sequence of the human
genome. Science 2000; 291: 1304–1351.
Bray D. Reductionisms for biochemists: how
to survive in the protein jungle. Trends
Biochem Sci 1997; 22: 325–330.
or—in
a pharmacotherapeutic con-
text—pharmacogenomics.
Successful research in this field
needs a number of prerequisites. The
technical advancements have already
been employed in genome mapping.
High-throughput genotyping and
gene expression profiling method-
ologies are necessary. More powerful
methods and tools to deal with the
massive data information are needed
in the bioinformatics science. And in
clinical pharmacogenomics, a number
of additional prerequisites are manda-
tory. First, robust clinical endpoints
for clear identification of ‘response’,
‘non-response’, ADRs, etc and interac-
tion with national or international
morbidity registries will be extremely
important. Second, DNA/tissue bank-
ing needs to be developed and sub-
jected to quality assurance. The ethical
issues of biobanking are important to
consider. Third, networking and col-
laboration, preferably over the Inter-
net, will be the basis of successful clini-
cal data capture.
The progress towards the clinical
benefit of the knowledge of the
human genome may be slow. In order
to make use of it, a broad collaboration
all the way from molecular to clin-
ical sciences is mandatory. The
reductionist research paradigm has to
be reoriented to the physiological and
clinical sciences in order to explain the
meaning of the genomic variation
between, and within species.
4
The International SNP Map Working Group.
A map of human genome sequence vari-
ation containing 1.4 million single nucleo-
tide polymorphisms. Nature 2001; 409:
928–933.
5
6
7
Evans WE, Relling MV. Pharmacogenomics:
translating functional genomics into
rational therapeutics. Science 1999; 286:
487–491.
Mahgoub A, Idle JR, Dring DG, Lancaster R,
Smith RL. Polymorphic hydroxylation of
debrisoquine in man. Lancet 1997; 2:
584–586.
Eichelbaum J, Spannbrucker N, Steinke B,
Dengler HJ. Defective N-oxidation of spar-
teine in man:
a
new pharmacogenetic
defect. Eur
J Clin Pharmacol 1979; 16:
183–187.
8
9
Brynne N, Dale´n P, Alva´n G, Bertilsson L,
Gabrielsson J. Influence of CYP2D6 poly-
morphism on the pharmacokinetics and
pharmacodyanamics of tolterodine. Clin
Pharmacol Ther 1998; 63: 529–539.
¨
Mortimer O et al. Polymorphic formation of
morphine from codeine in poor and exten-
sive metabolisers of dextromethorphan.
Relationship to the presence of immuno-
identified cytochrome P-450IID1. Clin Phar-
macol Ther 1990; 47: 27–35.
10 Johansson I et al. Genetic analysis of the
Chinese CYP2D locus. Characterization of
variant CYP2D6 genes present in subjects
with diminished capacity for debrisoquine
hydroxylation. Mol Pharmacol 1994; 46:
452–459.
11 Wadelius M, Darj E, Frenne G, Rane A.
Induction of CYP2D6 in pregnancy. Clin
Pharmacol Ther 1997; 62: 400–407.
12 Hoffmeyer
S et al. Functional polymor-
phisms of the human multidrug-resistance
gene: multiple sequence variations and cor-
relation of one allele with P-glycoprotein
expression and activity in vivo. Proc Natl
Acad Sci USA 2000; 97: 3473–3478.
13 Rebbeck TR, Jaffe JM, Walker AH, Wein AJ,
Makowicz SB. Modification of clinical pres-
entation of prostate tumors by a novel gen-
etic variant in CYP3A4. J Nat Canc Inst
1998; 90: 1225–1229.
14 Drysdale CM et al. Complex promoter and
coding region ␤2-adrenergic receptor
haplotypes alter receptor expression and
DUALITY OF INTEREST
None declared.
Correspondence should be sent to
A Rane, Professor and Chair of Clinical
Pharmacology, Karolinska Institutet,
The Pharmacogenomics Journal

Sudden impact?
DA Collier
9
predict in vivo responsiveness. Proc Natl
Acad Sci USA 2000; 97: 10483–10488.
15 Lazarou J, Pomeranz BH, Corey PN. Inci-
dence of adverse drug reactions in hospi-
talized patients. A meta-analysis of prospec-
tive studies. JAMA 1998; 279: 1200–1205.
16 Weinstein JN et al. An information-intensive
approach to the molecular pharmacology
of cancer. Science 1997; 275: 343–349.
ing is unresolved. Since the genetic
variation which generates human
diversity is hypothesised to mainly
reside with these polymorphisms, they
are rapidly becoming the marker of
choice for complex disease geneticists.
Despite the success of linkage, the
identification of the susceptibility
genes lying within has been frustrat-
ingly slow. Complex disorders with
heritability in excess of 80%, such as
schizophrenia and bipolar disorder,
have resisted all attempts to conclus-
ively identify genes. Even so, a mini-
industry of complex disease genetics
has developed over the past decade,
with thousands of papers addressing
the molecular genetics of psychiatric
disorders alone. Despite there being
few concrete findings which can be
translated into clinical medicine, this
effort has not been in vain, as frus-
tration has catalysed many important
advances in statistical, clinical and
molecular methodologies.
Even when armed with knowledge
of the rough whereabouts of a com-
plex disorder susceptibility gene (say
within about 10 million of the 3
billion base-pair genome), the remain-
ing task of identifying the guilty gene
from within this region is daunting.
The pre-genome sequence isolation of
disease genes for Mendelian diseases
was extremely difficult and very
expensive (several hundred million
dollars for the cystic fibrosis gene)
because of the huge tracts of unknown
DNA which needed cloning and care-
ful analysis, but the process could still
be achieved without any biochemical
knowledge about the causative gene.
The situation for complex diseases,
where there will not be a direct geno-
type-phenotype relationship and the
linked region is much larger, is at least
an order of magnitude more difficult.
Before the human genome sequence,
analysing these linked regions for sus-
ceptibility genes was nearly imposs-
ible, not least because of the number
of genes within (there might be
Sudden impact? The human genome
sequence and the pace of gene
discovery in complex diseases
DA Collier
Senior Lecturer in Molecular Genetics, The Institute of Psychiatry, London, UK
the 1990s, when human genome
analysis was a largely academic pur-
suit, to the present day, where many
billions of research and development
dollars are dedicated to its study. Of
course the human genome project will
not find genes for complex disorders
for us, but in the short term, there is
no doubt that the human genome pro-
ject will accelerate the process of dis-
covery, simply by making the practical
effort substantially easier.
The main success in complex disease
genetics, albeit over at least a decade of
effort, has been with linkage analysis:
despite many false starts we now have
a good idea of where many loci for
complex diseases are in the genome.
Even for schizophrenia (which has a
history of conflicting results) there are
now regions of the genome (such as
chromosome 13q) where disease genes
clearly lie.1,2 This success was depen-
dent on a previous great advance in
human genomics, the isolation and
positioning of many hundreds of
microsatellite genetic markers which
replaced the unreliable, slow and DNA
hungry use of Southern blotting.3
These 30000 or so repeat markers are
now taken for granted, and more can
be isolated at will simply by scanning
the human genome for simple
sequence repeats. Ironically, just when
sequence repeat markers have become
abundant and easier than ever to ana-
lyse, attention has switched to the use
of single nucleotide polymorphisms
(SNPs) as genetic markers, for which
the most efficient method of genotyp-
With the flurry of publicity on the suc-
cess of the nearly-finished human gen-
ome project, the public can be for-
given for expecting major medical
breakthroughs in the very near future.
As many have commented, the gen-
ome sequence only marks the begin-
ning of our understanding. The break-
throughs to expect in the immediate
future will be scientific rather than
medical, helping our understanding of
human biology and diseases. The pro-
ject is expected to catalyse the cloning
of complex disease susceptibility
genes, drive forward the developing
field of pharmacogenomics, and insti-
gate a new era of pharmaceutical drug
discovery based on the rapid discovery
of new peptides and proteins which
are potential drug targets.
Major advantages of the human
genome project for the complex dis-
ease geneticist are that it will allow
the structural, (and consequently
functional) characterisation of even
the most elusive of human genes. Cur-
rently, almost nothing is known about
the biological function of the majority
of genes, and the genome project will
help reveal whole new pathways of rel-
evance to areas such as brain develop-
ment, appetite, and cardiovascular
physiology. Equally important is the
ability to catalogue millions of poly-
morphic DNA sequence variants,
many thousands of which will have
consequences for human disease. The
perception of potential for genomics is
illustrated by the change from the
idealistic early days at the beginning of
www.nature.com/tpj