Beruflich Dokumente
Kultur Dokumente
Understanding the identity of drug targets that are estimated to contain ~30,000 genes and, of these, ~3,000
encoded by the human genome is of great importance genes were suggested to be linked to disease based on
for the development of new pharmaceutical products the extrapolation of data from the number of antifungal
and the allocation of resources within academic and targets in the yeast genome. The overlap between the
industrial biomedical research. Currently marketed two sets of genes estimated to be ~6001,500 genes
drugs mediate their effects through only a small num- in total3 was suggested to represent the number of
ber of the potential human target proteins. Previously pharmacologically exploitabletargets.
published estimates of the number of current human In 2006, Imming etal.4 listed 218 drug targets for
drug targets range from ~200400, depending on the approved drugs, using by their own terms some-
method used and how the drug targets were defined and what arbitrary definitions for drug targets and drug
categorized. categories, which led to a lower estimate in compari-
In 1996, Drews and Ryser were the first to present son. Also in 2006, Overington etal.5 estimated that the
an overall estimate of the number of target proteins in human genome contained 266 proteins that could be
humans and pathogens when they put forward an esti- targeted by pharmacological agents. They were able to
mate of 483 drug targets, based on an analysis of drugs assign a total of 324molecular targets to 1,065 phar-
listed in the ninth edition of The Pharmacological Basis macological agents through a systematic review of the
of Therapeutics1. They also estimated that the number of US Food and Drug Administration (FDA) Orange Book
potential drug targets in the human genome was ~5,000 and the Center for Biologics Evaluation and Research
10,000, which was influenced by the view at the time that (CBER) website. This analysis led to the identification of
the human genome contained ~300,000 genes1,2. 21,000 active agents, but only 1,357 unique drugs when
*Department of Neuroscience, After the sequencing of the human genome, Hopkins redundant formulations and non-therapeutic agents
Functional Pharmacology, and Groom3 estimated that there were 120 drug targets were removed5.
Uppsala University, for marketed small-molecule drugs, by analysing the The publicly available DrugBank database, which
Uppsala Biomedical Center,
Investigational Drugs database and Pharmaprojects drew heavily from these earlier data sets, was launched
75124 Uppsala, Sweden.
Department of Neuroscience, databases. In total, they identified 399 targets for which in 2006. The DrugBank database not only has a sys-
Uppsala Biomedical Center, ligands with drug-like properties (regardless of whether tematic collection of drugprotein interactions but
BOX593, 75124 Uppsala, or not they were approved drugs) were known. By also contains associations of proteins with consensus
Sweden. expanding from members of the 130 protein families that genetic annotations, such as Swissprot. The DrugBank
Correspondence to H.B.S.
email: helgi.schioth@neuro.
were found to be drug targets, they also estimated database has been expanded by ~60% since its release to
uu.se that ~10% of all genes in the genome were pharma- include further FDA-approved and experimental drugs,
doi:10.1038/nrd3478 cologically tractable. In 2002, the human genome was as well as data for almost 1,000 additional drugtarget
interactions68. The database currently contains infor- drugtarget interactions were removed because the
mation on ~1,500 experimental, approved and with- majority of these interactions were either irrelevant from
drawn drugs, with up to 107 data fields for each drug a disease-treatment perspective or part of the side-effect
that contain information including current indications, profile of the pharmacological agent. We therefore per-
documented drugtarget interactions, target protein formed a manual curation of the database to achieve
accession numbers and pharmacologicalactions. a reproducible data set of human genes encoding the
In this article, we have analysed the complete data target proteins that are responsible for the desired phar-
set of pharmacological agents as of May 2009 from macological effects. We also manually curated entries for
the DrugBank database (see FIG.1 for an overview of the indications and classified proteins according to molecu-
analysis process). Annotations in the DrugBank data- lar function using their Swissprot accession numbers.
base that were related to the functional relevance of We retrieved drug approval dates from the FDA and
linked these to the drugs in our curated data set to iden-
tify trends in drug development. We manually added
0WODGT FDA-approved drugs and drug targets from 20072010
&TWIDCPMFCVCDCUG &TWIECVGIQT[
FTWIU 7PMPQYPOGEJCPKUO into our data set, to account for lag times between the
&KTGEV&0#KPVGTCEVKPI introduction of a new drug and its incorporation into the
#PVKDKQVKE DrugBank database. In contrast with previous publica-
4GOQXCNQHFTWIU FTWIU #OKPQCEKF tions in this field, we also provide a specific list of the
YKVJPQMPQYPVCTIGV drugs (see Supplementary information S1 (table)) to aid
&KGVCT[UWRRNGOGPV
1VJGT further analyses.
6QVCN
FTWIUCEVKPIQP Drugs and classes of drug targets
RTQVGKPVCTIGVU The initial curation of the data set involved the identi-
6CTIGVECVGIQT[ 0WODGT fication of drugs without known protein targets, such
$CEVGTKC as drugs that act directly on DNA, dietary supplements
4GOQXCNQHFTWIU
YKVJPQJWOCPVCTIGV
FTWIU #PKOCNU and drugs that do not have known protein targets. This
8KTWUGU resulted in a list of 1,317 drugs that act on 1,236 protein
2CTCUKVGU targets. We also identified drugs that were administered
(WPIK as prodrugs and were therefore already present in the
FTWIUCEVKPIQP 'WT[CTEJCGQVC data set in their active form, and we identified drugs
JWOCPRTQVGKPVCTIGVU
6QVCN that act on non-human target genes, such as antibiotics,
antiparasitics and antifungals.
4GOQXCNQHPQP This specification resulted in a list of 1,092 pharma-
VJGTCRGWVKEVCTIGVU PQPVJGTCRGWVKEVCTIGVU cological agents that act on 1,044 human protein targets.
We then identified the genes encoding non-therapeutic
FTWIUYKVJQWVMPQYP
protein targets (for example, albumin or cytochrome P450
FTWIUCEVKPIQP
VJGTCRGWVKEGGEV VJGTCRGWVKEVCTIGVU enzymes), as well as agents without therapeutic targets.
OGFKCVKPIVCTIGVU We used a row-by-row approach in favour of automated
methods in which each drug, drug target and interac-
tion was evaluated and validated using current medical
literature and publicly available databases. The main
effect-mediating targets were identified manually for each
6QRFTWIKPFKECVKQPU 0WODGT 6CTIGVENCUU 0WODGT drug and therapeutically irrelevant targets or side effect-
#PVKJ[RGTVGPUKXGCIGPVU 4GEGRVQTU mediating targets were identified in the data set. This was
#PVKPGQRNCUVKECIGPVU 'P\[OGU performed in an effort to create a data set with strictly
#PVKKPCOOCVQT[CIGPVU 6TCPURQTVGTU defined active targets, which could be useful for further
*[RPQVKEUCPFUGFCVKXGU 1VJGT bioinformatics analysis of the therapeutic portion of the
#PVKCNNGTIKECIGPVU druggable genome as well as for novel target prediction.
#PVKEQPXWNUCPVU Through this strict manual curation of the DrugBank
#PVKCTT[VJOKCCIGPVU database (FIG.1), we identified 435 effect-mediating drug
#PVKRU[EJQVKECIGPVU targets in the human genome. These structures are the
#PVKFGRTGUUCPVU targets of 989 unique drugs, through 2,242 drugtarget
#PCNIGUKEU interactions (see Supplementary informationS1 (table)).
The DrugBank database lists indications for each
Figure 1 | Data curation process. Data were collected from the DrugBank database drug, and these were also validated in the curation
0CVWTG4GXKGYU^&TWI&KUEQXGT[
as of May 2009 and curated manually to identify drugs without known targets and drugs
process. The most common indication for the drugs
that target gene products of pathogens, such as antiparasitics, antivirals and antibiotics.
Specific human gene products that encode therapeutic drug targets were also identified in our data set is antihypertensive drugs, followed by
manually among the drugtarget interactions. Primary and secondary indications of the antineoplastic and anti-inflammatory agents. Each
drugs were identified by manually curating interaction entries. Drugs and drug targets target-encoding gene was assigned a class according
that were approved by the US Food and Drug Administration between 2007 and 2010, to the molecular function of the gene product using
and were not included in the Drugbank database, were added manually. the Swissprot accession numbers. Our curated data set
shows that receptors make up the largest group of drug genes, comprising 29% of all human drug targets (FIG.2).
targets: 193 proteins (44% of the human drug targets) Hydrolases (EC3) are the most common class of enzy-
are receptors, and 82 (19%) of these are G protein-cou- matic drug targets, comprising 42% of all human enzyme
pled receptors (GPCRs). In the overall data set, ~36% drug targets. They are followed by oxidoreductases (EC1)
of drugs target GPCRs (FIG.2). GPCRs have been com- and transferases (EC2), which comprise 27% and 19% of
monly targeted by antihypertensive and anti-allergic all human enzyme drug targets, respectively. In addition,
drugs. Ligand-gated ion channels, which are the second the majority (78%) of enzyme targets are soluble proteins,
largest receptor target class, are most commonly targeted not membrane-associated proteins. Anti-inflammatory
by hypnotic drugs and sedatives. The third largest recep- treatments most frequently act on enzyme targets; for
tor target class, receptor tyrosine kinases such as the example, the most studied anti-inflammatory target path-
epidermal growth factor receptor have been targeted way the eicosanoid metabolic pathway is modulated
frequently by anticancer drugs (intracellular non-recep- via the enzyme targets cyclooxygenase 1 and cyclooxyge-
tor kinases, such as ABL, the active site of which is the nase 2, which belong to the oxidoreductase family and are
target of imatinib,are classifiedas enzymes). targeted by acetylsalicylic acid. Other common enzyme
Enzymes are the second largest group of target pro- targets for drugs include DNA polymerases, angiotensin-
teins in the human genome, with 124 target-encoding converting enzyme and the monoamine oxidases.
'P\[OGU
#PVKKPCOOCVQT[CPVKPGQRNCUVKE
'%1ZKFQTGFWEVCUGU
#PVKKPCOOCVQT[CPVKPGQRNCUVKE
'%6TCPUHGTCUGU
#PVKPGQRNCUVKEDKURJQURJQPCVG
'%*[FTQNCUGU
#PVKJ[RGTVGPUKXGXCUQFKNCVQT
'%.[CUGU
#PVKJ[RGTVGPUKXGFKWTGVKE
'%+UQOGTCUGU
#PVKPGQRNCUVKEKOOWPQUWRRTGUUKXG
'%.KICUGU
#PVKPGQRNCUVKECPVKDTKPQN[VKE
/WNVKRNG'%ITQWRU
#PVKPGQRNCUVKECPVKCFTGPCN
1VJGT #PVKKPCOOCVQT[CPVKPGQRNCUVKE
'P\[OGKPVGTCEVKPIRTQVGKPU #PVKKPCOOCVQT[INWEQEQTVKEQKF
5VTWEVWTCNCPFCFJGUKQPRTQVGKPU #PVKPGQRNCUVKE
.KICPFU #PVKTJGWOCVKE
1VJGT #PVKKPCOOCVQT[CPVKPGQRNCUVKE
Figure 2 | Classification of the currently utilized drug targets that are encoded by 0CVWTG4GXKGYU^&TWI&KUEQXGT[
the human genome. The class
of drug target was determined by using the Swissprot accession number of the corresponding therapeutic target of each
drug. The most common indications for each class are listed, as well as the corresponding number of drugs that target
each protein class. *Enzymes are further classified as soluble or membrane-associated; the number in the brackets
corresponds to the number of membrane-associated enzymes in each class.
2003) were approved. As noted above, there was a stable kinase that is present in patients with chronic myeloid
rate of introduction of NTDs that target GPCRs, hydro- leukaemia dramatically improved disease treatment,
lases, isomerases and oxidoreductases in all three inno- and demonstrated the potential of molecular targets in
vation peaks, and ~50% of all of the drugs that have been anticancer drug discovery, which are now being widely
approved since 1982 fall into one of these categories. pursued10.
Closer inspection of the NTDs reveals that small-mol- GPCRs are clearly the most commonly exploited class
ecule drugs were the most common class of agents acting of drug targets (FIG.2) approximately 36% of all drugs in
on novel targets, comprising 60% of all NTDs, but it also our data set target GPCRs and they also represent the
reveals an increase in the development of biotechnology- largest class of target structures for NTDs that have been
based NTDs such as monoclonal antibodies and fusion approved since 1983; approximately 17% of all targets for
proteins during the past two decades (TABLE1). The first which NTDs have been introduced since 1982 have been
monoclonal antibody drug that was approved for sale GPCRs. According to data from IMS Health MIDAS, six
in the United States was muromonabCD3, which is an out of the 20 drugs (30%) with the highest global sales in
immunosuppressant that was approved by the FDA in 2010 clopidogrel (Plavix; Sanofi/Bristol-Myers Squibb),
1986. Since then, the number of monoclonal antibody quetiapine (Seroquel; AstraZeneca), olanzapine (Zyprexa;
drugs has steadily increased, and they now comprise Lilly), montelukast (Singulair; Merck), aripiprazole
a considerable proportion of all NTDs. In the mid- to (Abilify; Bristol-Myers Squibb) and valsartan (Diovan;
late1990s (19942000) monoclonal antibody drugs Novartis) target GPCRs. In addition, 63 out of the 200
comprised 24% of all NTDs, and between 2001 and 2010 drugs (~32%) with the highest sales in the United States
they comprised 20% of all NTDs. Monoclonal antibodies in 2009 target GPCRs (see the Drugs.com website). The
have also been a major commercial success, with sales majority of pharmacologically exploited GPCRs belong to
rising dramatically in the past decade. For example, the the Rhodopsin family (reviewed in REF.11).
total market for the four most successful monoclonal Other common receptor target structures for NTDs
antibodies infliximab (Remicade; Centocor Ortho in the period that was studied include cytokine receptors,
Biotech), rituximab (Rituxan; Genentech/Roche/Biogen receptor tyrosine kinases, ligand-gated ion channels and
Idec), adalimumab (Humira; Abbott) and bevacizumab integrins. Enzymes represent the second largest group
(Avastin; Genentech/Roche) increased by a fac- of novel target structures and most commonly belong
tor of 3.90 from 2004 to 2010, whereas the total global to the hydrolase, oxidoreductase or transferase protein
drug market during the same period increased by fac- families (EC3, EC1 and EC2, respectively). Novel
tor of 1.53, according to data from IMS Health MIDAS hydrolase targets have been dominated by peptidases
(December 2008, 2009 and 2010). and ester hydrolases (EC3.4 and EC3.1). Examples
The majority of drugs (14 out of 15) with the highest of such targets include renin, which is a peptidase
global sales in 2009 were approved by the FDA during that is targeted by the antihypertensive drug aliskiren
the past 15years, during the second innovation peak, (Tekturna/Rasilez; Novartis) (approved in 2007) and
which is unsurprising as these are the products that still phosphodiesterase 5, which is an ester hydrolase that
have patent protection and have also had sufficient time is targeted by the erectile dysfunction drug sildenafil
to become established in the market. A common feature (Viagra; Pfizer) (approved in 1998). Novel targets in
of all of these drugs is that they target defined protein the transferase family have been dominated by phos-
structures that are encoded by the human genome. This phorous-containing group-transferases (EC2.7) such
is comparable to 65% of drugs in the DrugBank data- as 5-AMP-activated protein kinase, which is indirectly
base that target effect-mediating structures encoded activated by metformin (approved in 1995). However,
by the human genome. Other target categories include the exact mechanism of activation of 5-AMP-activated
targets that are produced by pathogens, such as viral protein kinase by metformin remains to be estab-
and bacterial proteins, and non-genomic structures lished12. NTDs that modulate oxidoreductases often
such as extracellular macromolecules and DNA. The target enzymes that act on the CHOH group of donors,
DrugBank database also includes various other agents, with NAD+ or NADP+ as the acceptor in redox reactions
such as osmotic agents, sclerosing agents, nutrient sup- (EC1.1.1). Some examples include: 3hydroxy3meth-
plements and inactive diagnosticagents. ylglutaryl-CoA (HMG-CoA) reductase, which is
The most common indication for NTDs is cancer targeted by the statin class of cardiovascular drugs
treatment: ~19% of all NTDs that have been approved to reduce high levels of low-density lipoprotein cho-
since 1983 are anticancer drugs. Advances in the under- lesterol; alcohol dehydrogenase, which is targeted by
standing of the molecular basis of cancer during this disulfiram for the treatment of alcohol dependency; and
period have contributed substantially to this trend, and inosine 5-monophosphate dehydrogenases, which are
have resulted in a shift from traditional cytotoxic thera- targeted by the immunosuppressive drug mycopheno-
pies to the development of drugs that specifically target late mofetil (CellCept; Roche).
particular proteins that are linked to various cancers,
such as small-molecule kinase inhibitors and monoclo- Drugtarget networks
nal antibodies that bind to proteins on the surface of Protein networks or modules are increasingly being
cancer cells. One notable example is the first marketed studied in the field of network biology using methods
small-molecule kinase inhibitor, imatinib (Gleevec; from graph theory, which is a growing field within com-
Novartis), which by inhibiting the mutant BCRABL puter science. Much of network biology has focused on
6JGTCRGWVKECEVKQPU 6CTIGVENCUU
)KCPVEQORQPGPV
%#%0#&
%#%0#$ 0+64'0&+2+0'
%.'8+&+2+0'
.'4%#0+&+2+0'
#/.1&+2+0'
&+.6+#<'/
-%0'
#PVKCTT[VJOKEU
)#$#2'06+0
4GEGRVQTU
0+/1&+2+0'
%#%0)
0+('&+2+0' #%'6;.&+)+61:+0
&'5.#015+&'
#./+64+0 '
+0#/+&'
('.1&+2+0'
#PVKCNNGTI[CIGPVU 'P\[OGU
&+)1:+0&+)+61:+0
+54#&+2+0'
%#%0#&
#62#
.'815+/'0�
17#$#+0
24')#$#.+0
%#%0$
0+%#4&+2+0'
#PCNIGUKEU 6TCPURQTVGTU
%#%0$ -%03
%#%0#5
%#%0#%
%#%0#&
0+51.&+2+0'
600%
%#%0#(
%#%0$ %#%0##
%#%0$
#PVKKPCOOCVQTKGU 1VJGTVCTIGVU
8'4#2#/+.
+$76+.+& '
-%0-
$'24+&+.
2#4#/'6*#&+10' #24+0&+0'
-%0,
&1('6+.+&'
%#%0#+
%#./ $'8#061.1.
*[RPQVKEUCPFUGFCVKXGU
516#.1. +51'6*#4+0'
1245
#PVKEQPXWNUCPVU
&;%.10+0' $+51241.1. 14%+24'0#.+0'
/'61241.1. 5#.$76#/1.
241%#+0' 24#%61.1.
-%0* %#%0#*
241%#6'41.
.'81$72+8#%#+0' $72+8#%#+0' #6'01.1. 6+/1.1.
'5/1.1 . 0#&1.1 .1:24'01.1. %.'0$76'41.(14/16'41.
2412#4#%#+0' 412+8#%#+0'
241%#+0#/+&' 0'$+81.1 .
'0%#+0+&' 24124#01.1.
1:;$7241%#+0' 4#01.#<+0' +512416'4'01.
2412#('010' )4#0+5'6410 &1$76#/+0' 2+0&1.1. 2'0$761.1. 6'4$76#.+0'4+61&4+0'
&':641/'6*142*#0
2#.1015'6410
)4+0$
#PVKFGRTGUUCPVU
5%0# %*.141241%#+0'
&+$7%#+0'
+0&'%#+0+&'
$'0<10#6#6'
(152*'0;61+0 10'6410
#%'$761.1.
/'6+24#01.1. #4$76#/+0'
$'6#:1.1.
-'6#/+0'
&1.#5'6410 ('016'41.
5%0#
$'0<1%#+0' .+&1%#+0 '
%#%0#)
*':;.%#+0'
#.15'6410
(.70#4+<+0'
#/+1ƚ' #&4$ #&4$ %#46'1.1.
#4(14/16'41. /'/#06+0'
)4+0#
&+512;4#/+&'
&410'ƚ' 5#./'6'41.
('.$#/#6'
5%0#
64+/'6*#&+10'
)4+0#
/'2+8#%#+0' /':+.'6+0'
*64# #.24'01.1.
#PVKJ[RGTVGPUKXGU
.'81$701.1.
%#4$#/#<'2+0'
'6*157:+/+&'
'2+0'2*4+0'
37+0+&+0 ' 5+.1&15+0 %#48'&+.1.
)4+0&
'6*161+0
1:%#4$#<'2+0' #24#%.10+&+0'
%.10+&+0 ' '2*'&4+0'
25'7&1'2*'&4+0'
61%#+0+&' &':/'&'61/+&+0'
/14+%+<+0'
.#$'6#.1.
'4)1.1+&
0+%'4)1.+0' /'2*'06'4/+0'
(.'%#+0+&' 24+.1%#+0'
2*'061.#/+0'
&+2+8'(4+0
6+<#0+&+0' /'5;.#6' /'6#4#/+01.
)4+0$
#PVKPGQRNCUVKECIGPVU
/'2*'0;61+0 2*'01:;$'0<#/+0'
/'6*;.&12# 1:;/'6#<1.+0'
$'0<2*'6#/+0' 61.#<1.+0'
/+&1&4+0 '
)7#0(#%+0' /'6*#/2*'6#/+0' 142*'0#&4+0'
/'6*1:#/+0'
)#$$4
)7#0#$'0<
#&4##
2*'0;.2412#01.#/+0' 2*'0;.'2*4+0'
2*'0&+/'64#<+0'
64+241.+&+0'
&':$41/2*'0+4#/+0'
&'5.14#6#&+0'
%+00#4+<+0'
)4+0
2*'0;61+0 $4+/10+&+0'
.1(':+&+0'
6#/57.15+0
#&4##
6'4#<15+ 0
2412'4+%+#<+0'
#PVKRU[EJQVKEU
014'2+0'2*4+0' &1:#<15+0 %.'/#56+0' $41/2*'0+4#/+0'
;1*+/$+0 ' #.(7<15+ 0 '/'8+0' %*.12*'&+#01.
#&4#%
$#%.1('0 24#<15+0 /'6*&+.#<+0'
&+/'0*;&4+0#6'
#&4#$ &4&+2*4*
&'$4+5137+0
/+46#<#2+0' .'81%#$#56+0' #56'/+<1.'
2*'0#%'/+&' #/2*'6#/+0'
64+(.712'4#<+0'
/+#05'4+0 64+/'24#<+0'
-'616+('0
#&4#&
$'6*#0+&+0' 64+2'.'00#/+0'
)#$$4
<10+5#/+&'
#&4#$ %;%.1$'0<#24+0'
.+5&':#/('6#/+0'
*64# 5'46+0&1.'
$41/1&+2*'0*;&4#/+0' /'%.+<+0'
#%'241/'6#<+0'
*4*
%;241*'26#&+0'
1VJGTKPFKECVKQPENCUUGU
$752+410'
5%0#
/'61%.124#/+&' &1:;.#/+0'
'5%+6#.124#/ ('01.&12#/ 6*+14+&#<+0' &'5+24#/+0'
64#<1&10' 1.12#6#&+0'
)7#0#&4'. 64+/+24#/+0' +/+24#/+0' 0'(#<1&10' <+24#5+&10' /1.+0&10' %*.142*'0+4#/+0'
57.(#6' %#$'4)1.+0' &+2*'0;.2;4#.+0'
5+$764#/+0' #21/142*+0' 37'6+#2+0' #<'.#56+0' #<#6#&+0'
5%0#
.+574+& '
5%0#
%1%#+0 ' %.1<#2+0 '
*64#
&+'6*;.2412+10 416+)16+0'
%*.142416*+:'0' 2+/1<+&'
/#2416+.+0' 241%*.142'4#<+0' 6'4('0#&+0'
2'/+41.#56
<7%.12'06*+:1.
&4&
%*.14/'<#010' #61/1:'6+0'
5%0#
/'6*;5'4)+&'
#4+2+24#<1.' 2'42*'0#<+0' #.%#(6#&+0' 2*'0+4#/+0'
#/1:#2+0'
24164+26;.+0'
5.%# *64% .'81&12#
&412'4+&1.
2412+1/#<+0'
241/#<+0 ' &1/2'4+&10' $'216#56+0'
.14#6#&+0'
/'37+6#<+0' '2+0#56+0'
(1524121(1. (.70+64#<'2#/ 8'0.#(#:+0'
&':('0(.74#/+0'
&4& /'514+&#<+0'
1.#0<#2+0' +.12'4+&10'
241)#$+&' )7#0'6*+&+0'
5.%#5.%#
&4& &4&
&4& 2'4)1.+& ' &+2*'0*;&4#/+0'
%;%.+<+0 ' (#/16+&+0'
*64
24121(1. <#.'2.10 /'6*164+/'24#<+0'
/+.0#%+24#0 %#4$+01:#/+0'
/+6+).+0+&'
'61/+'
(.781:#/+0' &7.1:'6+0'
$72412+1 001464+26;.+0'
/+0#24+0 '
4+52'4+&10' 6*+'6*;.2'4#<+0'
(.752+4+.'0'
%*4/
%*4/
$7%.+<+0'
*4*
#/+64+26;.+0' 2*'0/'64#<+0' %#42*'0#<+0' $'6#<1.'
-%0,
61.$76#/+&'
)4+- *64$
412+0+41.' 2#.+2'4+&10' 241/'6*#<+0'
)#$4$
/+01:+&+. /'6*1*':+6#. %+6#.124#/
)#$4$ &1:'2+ 0 /'6*;.2*'0+' .1:#2+0' $'0<37+0#/+&'
/'6*#<1.#/+&' %.1/+24#/+0' 2*'06'4/+0'
#5'0#2+0 '
0#6').+0+&'
$<42
)#$44
)#$4)
1:#<'2#/ /+&#<1.#/ (.72'06*+:1. #%'612*'0#<+0'
(.72*'0#<+0' 6412+%#/+&'
)#$4$
).+%.#<+&' /'6*;.'4)1018+0' 64+(.7241/#<+0' *#.12'4+&1.
37#<'2#/
*#.#<'2#/ 6'/#<'2#/
%*.14&+#<'21:+&'
#.24#<1.#/
5.%# 5'464#.+0' &'58'0.#(#:+0'
24#/+2':1.' #6412+0' *1/#6412+0'
/'6+:'0'
/'6*;.$41/+&'
4#0+6+&+0'
4'2#).+0+&'
-%0, &+%*.142*'0#/+&'
)#$4& (.74#<'2#/ <12+%.10 '
(.7/#<'0+. 2#41:'6+0'
&':641#/2*'6#/+0'
('0%#/(#/+0'
#/+57.24+&'
57.2+4+& '
%*4/
-%0, %*4/
%*.142412#/+&' 6*+#/;.#. '56#<1.#/
612+4#/#6' .14#<'2#/
%.1$#<#/ (.71:'6+0' 0+<#6+&+0'
4'/1:+24+&' $'0<6412+0' 241%;%.+&+0' (.#81:#6'
6+16412+7/
)#$44
).+$'0%.#/+&' 64+#<1.#/
%.10#<'2#/ $41/1%4+26+0' 5%121.#/+0' %+/'6+&+0'
%#
6*+12'06#.
)#$4#
.74#5+&10'
)#$4#
)#$4# %*4/
$41/#<'2#/
0+64#<'2#/ /'6*;.2*'01$#4$+6#.
61.#<#/+&'
).+/'2+4+&'
.'8'6+4#%'6#/
)#$4#
%+01.#<'2#/
&+#<'2#/
)#$4#
%.16+#<'2#/
%.14#<'2#6' 2+.1%#42+0'
1:;2*'0%;%.+/+0' %;%.12'061.#6'
)#$43
24+/+&10 '
5'%1$#4$+6#.
)4+-
2*'01$#4$+6#.
)#$4)
%.+&+0+7/
)#$4)
2'061$#4$+6#.
)#$4#
24#<'2#/
(.7&+#<'2#/ 64#/#&1. +('0#%+0
)#$44
1:;$76;0+0 ('516'41&+0'
#&+0#<1.#/ 61.6'41&+0'
#%'61*':#/+&' '6*1:<1.#/+&'
#%'6#<1.#/+&' <1.2+&'/ $76#$#4$+6#.
$76#.$+6#.
/'241$#/#6' &+2*'0+&1. 2+4'0<'2+0'
).+2+<+&'
%# )#$42 %*40#
$76'6*#.
#0+516412+0'
64+&+*':'6*;.
+24#6412+7/
)#$4' %*40# *64&
).;%1&+#<+0' 57/#64+26#0 /'6*;.$41/+&' &+2*'/#0+.2412#06*'.+0'
57%%+0;.%*1.+0'
%*40#
/'6*;24;.10 (418#64+26#0 &+%;%.1/+0' 64152+7/
0+%16+0' %*.14+&'
/+8#%74+7/ /'6*#06*'.+0'
%*.1416*+#<+&' ).4# +51(.74#0' 4+<#64+26#0
&+*;&41'4)16#/+0'
$+2'4+&'0
%;%4+/+0 '
'6*12412#<+0'
$'0<6*+#<+&'
).4#
/60& #$#6
2'061.+0+7/
$76142*#01.
%1&'+0 '
%+5#64#%74+7/
$'5;.#6'
5.%#
24121:;2*'0'
$724'0142*+0' 1:;%1&10' )#..#/+0'
*;&41/142*10' /'%#/;.#/+0' 64+'6*+1&+&'
*;&41%*.1416*+#<+&'
%*.14<1:#<10'
#62& 0#.64':10'
*;&41%1&10'
/142*+0'
0#.$72*+0'
/'61%74+0'
&'%#/'6*10+7/
+1&+&'
#64#%74+7/
%41/1).+%#6' &+8#.241': 2#0%7410+7/
$'0&41(.7/'6*+#<+&' 51&+7/ 64+/'6*#2*#0
&#.(#/24+&+0'
124& *64 /'61%74+0' 67$1%74#4+0'
21.;6*+#<+&'
37+0'6*#<10' /'61.#<10'
*#.16*#0'
8#.241+%
#%+&8+)#$#64+0
.'8#..142*#0
/'6*#&;.
124/ 2+2'%7410+7/
/'6*1:;(.74#0' #%'6#6'
0#.1:10'
4'/+('06#0+.
57('06#0+.
#.8+/12#0
#0+.'4+&+0' ('06#0;.
5.%#
%*.14/'41&4+0 .'81/'6*#&;.
614#5'/+&' &+2*'01:;.#6' %+5#24+&'
#%'6#6' #.('06#0+.
.'8142*#01.
1:;/142*10'
6#2'06#&1. /'6*;.0#.64':10'
*;&41(.7/'6*+#<+&'
/'6*;%.16*+#<+&' /'6*#&10'$41/+&'
.12'4#/+&'
%*.146*#.+&10'
(7415'/+&' '6*#%4;0+%
#%+&
&;&41)'56'410'
/+('24+5610'
$7/'6#0+&'
(.6 014'6*+0&410' &4152+4'010'
4#(
(.6 7.+24+56#.
5.%#5.%#
514#('0+$
'54 2)4
'6101)'564'.
5.%# 5+.&'0#(+. &+(.724'&0#6' #/$'010+7/
()(4 $4#( -&4 '2124156'01. 6'0'%6'2.#5' 64+/'64':#6' 6#/1:+('0
()(4
8#4&'0#(+. $7&'510+&' '&412*10+7/ 2#.+('4/+0
;'5
#0+5+0&+10'2*;610#&+10' 0'156+)/+0' &'51)'564'.
7&'0#(+. &+0124156 #/+01%#241+% 4#.1:+('0'
641/'6*#/+0' #%+&
24'&0+%#4$#6' %.1$'6#51. &10'2'<+.
'2*# 570+6+0+ $
(.6
24#.#64':#6'
'564#&+1 .
014)'564'.
/')'5641. (7.8'564#06
.% - -+ 6
*;&41%146#/#6' 564'261-+0#5' .'81014)'564'.
014)'56+/#6'
*#.1$'6#51. 2*;5156+)/+0'
2'06156#6+0 2412+10#6' /+51241561.
26)+4 2.) &*(4 241)'56'410'
2&'#2&'$2&'#
'6*;01&+1.
()(4 56#6$
614'/+('0'
54% (;0
2#2#8'4+0' /'6*;.24'&0+51.10' &'51:+/'6#510' 2'/'64':'&
+.124156 '564#&+1.
%*.14164+#0+5'0'/'564#01.
2&'% +/#6+0+$
#0:# )#46
4'6'2.#5'
#%'01%17/#41. 9#4(#4+0
2&'# &':#/'6*#510'
(.7/'6*#510'
2.#)# (.7141/'6*1.10' &'/'%#4+7/ 0(-$
6*#.+&1/+&'
'56410'
2&'$
'0241(;..+0' 37+0'5641.
( 2415
2*'0241%17/10 8++# �#<1.
6*'12*;..+0'
64+#/%+01.10' (.74#0&4'01.+&'
26)'4
26)'4 2.) 6;/5
(.1:74+&+0'
241%
(10#4+07: 2#4#/'6*#510' $41/+&'
&')#4'.+:
*'2#4+0
4')#&'01510
#$#4'.+:
(
.'7241.+&'
5'42+0%
'01:+/10' %+%.'510+&'
57.1&':+&'
51&+7/
%+.156#<1.
6'64#&'%;.
57.(#6' $+8#.+47&+0
:+/'.#)#64#0
5'42+0&
#4&'2#4+ 0
&#$+)#64#0
.'2+47&+0 '01:#2#4+0
*20
/'.1:+%#/
1:#241<+0
%#4241('0 2+1).+6#<10' /'6;415+0'
('.;24'55+0
&'5/124'55+0
(1..+6412+0
$'6#
0+64124755+&' 0+64+%
1:+&'
%*& $'06+41/+&' 4#44'5 #62$ $%. &1%'6#:'.
(15+0124+..+5+0124+.
37+0#24+ .
*;&41:;74'#
.7/+4#%1:+$ &'5'42+&+0'
64'6+01+0 2#%.+6#:'.
415+).+6#<10'
#824# #824$
+51514$+&'
&+0+64#6' 20.+2 6#<#416'0' 4#4) /1':+24+.
'614+%1:+$ &+(.70+5# .
57241('0 6* '2+47$+%+0 4#$'24#<1.'
4'5%+00#/+0'
44/
57.+0&#%
22#4) 62* 741(1..+6412+0
(5*4
024 0+641).;%'4+0 21&1(+.1 : 6#/+$#416'0'
4#4$
8+0$.#56+0'
67$$
-'61241('0
4:4$
6+#241('0+% 0#241:'0 2'4+0&124+.
#%'
8#5124'55+0
#%+& &+%.1('0#% %*14+1)10#&16412+0 2#06124#<1.'
+/$+%+ 0 14.+56#6
0#$7/'610' 6'4.+24'55+0 #.(# +51514$+&'/1010+64#6' ##.'0' +:#$'2+.10' 52+4#24+ .
#$%)
%#$#<+6#:'.
4:4)
-'6141.#% 6'64#*;&41$+126'4+0
67$$#
%.1(#4#$+0'
26)5 (.74$+241('0 %10+8#26#0 /+61:#06410'
4#4# #62#
.#05124#<1.' %#0&1:#64+.
//'
26)+5 2*'0;.$76#<10' /'('0#/+%
#%+& 5#.5#.#6' 64'24156+0+. /'016412+05 '4;6*4+6;. 612# #%+64'6+0
#.+64'6+01+0 8+0%4+56+0' %#26124+ . (.7#$+0'
.*%)4
+$7241('0
61./'6+0 )15'4'.+0 0'5+4+6+&'
6'01:+%#/ $':#416'0'
024 &1:147$+%+0 '51/'24#<1.' 64#0&1.#24+.
$'0#<'24+.
'0#.#24+ .
21.#
#%'6#/+012*'0
5#.+%;.+%
+0&1/'6*#%+0
/'%.1('0#/+%
#%+&
%.12+&1)4'.
2#* +5164'6+01+0 4:4# 1/'24#<1.' 8+014'.$+0'
'61&1.#%
#%+&/'5#.#<+0'
0'2#('0#%
24; 024
6'0+215+&'
'61215+& '
$41/('0#%
#.1:
('01241('0 .76412+0
#52+4+0 61.8#26#0 #.(#
5#24126'4+0
6+%.12+&+0' 24#57)4'.
2+41:+%#/ <+.'7610
'4$$
/1%.1$'/+&' 81).+$15' .15#46#0
8#.5#46#0
(.76#/+&' %;2416'410' (.78#56#6+0
2+6#8#56#6+0 4+5'&410#6'
<1.'&410#6' /106'.7-#56 #0#5641<1.' %#.%+64+1.
+06'4('410
#.(#%10
$#5+.+:+/#$ &'0+.'7-+0
&+(6+61: *&#% $+/#6124156 +$4+67/1/#$ #.5'41:;.10 2+/'%41.+/75 %#4$+/#<1.'
$'21'6+0
#.(#
#<#%+6+&+0'
64#567<7/#$
)## 6#515#46#0
+06'4('410
+06'4('410
#.(#$ #.(## *&#%
/'6*1:;
21.;'6*;.'0'
/#1# *&#%
64#8124156 615+67/1/#$ 6'64#$'0#<+0' 6#%41.+/75 2412;.6*+174#%+. +(15(#/+&'
64#0;.%;241/+0'
.#2#6+0+$
5+
5#24+5#46#0 0+.76#/+&' (.71:;/'56'410'
.18#56#6+0
4'%1/$+0#06
+(0#4 +.4#
4'%1/$+0#06 814+0156#6
26)(4 /5# 5.%# (-$2# 621
).;%1.'21'6+0
$'6#
'214 &0/6 756'-+07/#$
*&#% *&#%
)'(+6+0+$ +06'4('410 &#%.+<7/#$
#.(#0
+.$
&41/156#01.10'
5'.')+.+0'
/#1$ 2*'0'.<+0'
%'67:+/#$
')(4 #/;#
0#0&41.10'
41578#56#6+0
#.'0&410#6'
$'6#$
+06'4('410 #.&'5.'7-+0
.#6#0124156 4+67:+/# $ 4'5'42+0 ' 5+41.+/7 5 /'6*+/#<1.'
'21'6+0
#.(#
&'%+6#$+0'
*5&$ %5(4$
5.%#
#$%% %26#
%.%0-#
)24#
+(0)4
(
(%'4) %&
+.4 &'51:;%146+%156'410' (.7&41%146+510'
5%00# %1.#
%04 25/& 54&# ()(
642/ //2 #&*% 8#/2 4#/2 8#24'16+&' 61.%#210' 6'5#/14'.+0 6'/5+41.+/75
51/#6412+0
4'%1/$+0#06
5+6#).+26+0 41/+2.156+/ 4'/+-+4'0 4#0+$+<7/#$
2+8#.#6'
25/$ (+0#56'4+&'
6#%4 %1/6 )*4*4 (4#2 )*4 &22 /2. 4'0 8')(
5%00) %1.# #/+.14+& ' %1..#)'0#5'
4#/2 $167.+07/
57.(+02;4#<10'
64+.156#0' 5#4)4#/156+/
241$'0'%+&
2'4*':+.+0'
0+(.7/+%
+06'4('410
/'2'0<1.#6'
#%+&
$'0<;.2'0+%+..1;.
&4164'%1)+0
)#//#$ #.(# 21.;.;5+0'
#0#-+04 #$#6#%'26
#
04%
0#$+.10'
25/& ()( 6428 //2 #&*# 5;6
4#/2
$146'<1/+$ &76#56'4+&'
2'0615#0
21.;57.(#6'
/'06*1. /#4+/#56#6 (1/'2+<1.' 61:+0
6;2'$
24#/.+06+&'
5#./10
%#.%+610+0
25/$ 54&#
()( 642# //2 #&*$ 8#/2
*+561.;6+%7/ 4#/2 #24'2+6#06 '06#%#210' 5'4/14'.+0 '8'41.+/75 2')8+51/#06 5#:#).+26+0 '.641/$12#) #.+5-+4' 0 $'8#%+<7/#$
*5&$ %5(4#
5.%#
#$%% %262.#
)24$
+(0)4
(
(%'4#
+.4%& 52+4101.#%610' 4+/10#$#06
'2.'4'010'
5%00& %1.#
/#4+01 .
2')(+.)4#56+/
.+4#).76+&'
+&7457.(#5' ('01(+$4#6' ('$7:156#6 %15;06412+0 .#04'16+&'
/;%12*'01.#6'
.'816*;41:+0' #/$4+5'06#0 #$%+:+/# $
612 612/6/604#/604$
64#0':#/+%
2.#6 61%+.+<7/#$
+.4 6+.7&410#6'
2620 65*46*;416412+0
26*4 #%#6 2)#
6'4+2#4#6+&' 57.(#5#.#<+0' 57%4#.(#6' 5'%4'6+0
5%64 4+.10#%'26
+.# %:%4 24-## %%-$4 8')(
2.'4+:#(14 2*'0(14/+0 2'06#)#564+0 2')#26#0+$ 1:;61%+ 0
1:6 152#
%& +.4# 124'.8'-+0 0+6+5+010'
*2&
%5(4 ).24
/24$2 22#4# :&* /%4
/1('6+. 61216'%#0 4#/'.6'10 #%+& #.(# .+212416'+0
'26+(+$#6+&'
$167.+07/
+6)#
0#6#.+<7/#$
%&'
/741/10#$
7)%)
/+).756#6 /'6;4#210'
0# 24-#$ *246 +)(4
/'6(14/+0 /'4%#261274+0' /'%#5'4/+0 /#4#8+41%
%%4 %.%0 &*1&* +054
.7$+2415610' .'(.701/+&' +057.+ 0
+//70'
).1$7.+0 %3# 6.4 $&-4$ )%)4
+/+37+/1 & +%#6+$#0 6
).7%#)10
4'%1/$+0#06
)'/67<7/#$
1<1)#/+%+0 %& 524 02%. +6)#.
(+0)1.+/1& '<'6+/+$ ' '(#.+<7/#$ '%7.+<7/#$
% '%#..#06+&'
-.- #.&* 60(5(
&+57.(+4#/ &'0157/#$ ʁ.'0'
4;4 21.# 21.$
&#%#4$#<+0' %;6#4#$+0'
(:;& 2224 #$%$ %#54
%;%.16*+#<+&' %;%.15214+0' %1.%*+%+0' %+0#%#.%'6 %+.#<#24+.
#%' &2'2 %%-#4
%+.#56#6+0 %'47.'6+&' %#4/756+0'
)54 %#4$+&12#
&&% 1:64
%#4$'61%+0 %#241/# $
25/# %#0#-+07/#$
+.$ 50#261:+0
6;2'#
$.'1/;%+ 0
.+)
+-$-$ 6:04&%'#%#/
#74#01(+0
%&#
#45'0+%
64+1:+&'
#4%+67/1/#$
#06+6*;/1%;6'
).1$7.+0
#0+564'2.#5'
2. ) #.22.
#/+(156+0'
#.2*#2416'+0#5'
+0*+$+614 '.#
#.'/67<7/#$
%& #.'(#%'26
%&
Figure 4 | Drugtarget network: summary. This figure provides a graphical representation of all of the interactions
between drugs and therapeutic drug targets in our curated data set. Nodes represent targets (shown as triangles) and
0CVWTG4GXKGYU^&TWI&KUEQXGT[
drugs (shown as circles). Novel target drugs (NTDs) are represented with large circles and drugs with established targets
are represented with small circles. Interacting drugs and targets are connected with lines. This network was created using
Cytoscape. Several isolated components have been identified in our data set, the largest of which we have called the giant
component. The giant component includes 49% of all drugs and 30% of all drug targets in our data set. See Supplementary
information S3 (figure) to view this drugtarget network in greater detail, as well as the labels at full size.
interactions between proteins, which have been called target functional class and novelty (that is, whether the
interactome networks. Drugtarget interactions have drug targeted a previously unexploited target). The net-
also been studied in this context 9. The field of interac- work was viewed in Cytoscape and analysed using local
tomics represents a new way to understand the success Pythonscripts.
of drug targets in the context of their functional milieu. The drugtarget network consists of many subnet-
Topographical analysis of the complex network of inter- works; that is, clusters of connected drug targets. The
cellular protein interactions may also lead to new ave- largest of these components, which we have called the
nues for target prediction13,14. giant component (FIG.4), contains 489 drugs (49% of
We created a drugtarget network from our curated the total) and 131 targets (30% of the total). The giant
data set (which includes 989 drugs and 435 effect-medi- component includes 50% of all drugs from the top 10
ating protein targets) by connecting drugs with their indication classes. The giant component is also biased
known targets (FIGS4,5,6). The drugtarget network was in terms of the type of targets; 42% of all receptor tar-
integrated with annotation concerning drug indication, gets and 66% of all transporter targets are found in the
C %GVWZKOCD D 5KNFGPCN
7FGPCN 8CTFGPCN
2CPKVWOWOCD 2GPVQUVCVKP
'4$$
.CRCVKPKD 2&'%
2&'$
6TCUVW\WOCD
')(4 #&# &KR[TKFCOQNG &[RJ[NNKPG 2&'#
6CFCNCN 2&'#
'TNQVKPKD 2GPVQZKH[NNKPG
2&'&
2&'# 'PRTQH[NNKPG
)GVKPKD 2&'# 2CRCXGTKPG
4#( (.6
2&'$
(.6 6JGQRJ[NNKPG %CGKPG
/KNTKPQPG
$4#( 5QTCHGPKD -&4 #&14#
#&14#$ #OTKPQPG
;'5
'2*# -+6 5WPKVKPKD
(.6 #&14##
.%- 2&'#
2&)(4$ 'PQZKOQPG 4GICFGPQUQP
56#6$ &CUCVKPKD %5(4
%KNQUVC\QN
064-
4'6
6JGTCRGWVKECEVKQPU 6CTIGVENCUU
#$.
+OCVKPKD 2&)(4#
(;0 #PVKPGQRNCUVKECIGPVU 4GEGRVQTU
54%
$GECRNGTOKP 1VJGTKPFKECVKQPENCUUGU 'P\[OGU
1VJGTVCTIGVU
0KNQVKPKD #$. &&4
Figure 5 | Drugtarget network: selected networks. Networks of particular interest from the overall network are shown in
0CVWTG4GXKGYU^&TWI&KUEQXGT[
this figure. a | The interaction networks of receptor tyrosine kinase inhibitors for cancer treatment are depicted. These drugs
form two networks that are centred around two receptor tyrosine kinases: the epidermal growth factor receptor (EGFR) and
the platelet-derived growth factor receptor subunits A and B (PDGFRA and PDGFRB). Becaplermin, which is approved for the
topical treatment of foot ulcers in patients with diabetes, is the only agent in these networks that is not an anticancer drug.
However, it has been associated with an increased risk of cancer, and now has a black box warning on its label. b | The
interaction network of phosphodiesterase (PDE) inhibitors is depicted; the indications of drugs in this network encompass the
treatment of asthma and chronic obstructive pulmonary disease (xanthine derivatives theophylline, dyphylline and
enprofylline), erectile dysfunction (vardenafil, sildenafil and tadalafil) and congestive heart failure (amrinone, milrinone and
enoximone). Platelet aggregation inhibitors (cilostazol, dipyridamole and pentoxifylline), the spasmolytic drug papaverine and
the familiar stimulant caffeine are also included in this network. As in Figure 4, nodes represent targets (shown as triangles) and
drugs (shown as circles). Drugs with novel targets are represented with large circles and drugs with established targets are
represented with small circles. Interacting drugs and targets are connected with lines. The original network was created using
Cytoscape. ADA, adenosine deaminase; ADORA1, adenosine receptor A1; CSF1R, macrophagecolony-stimulating factor 1
receptor; DDR1, discoidin domain receptor tyrosine kinase 1; EPHA2, ephrin type A receptor 2; FLT1, fms-like tyrosine kinase 1;
KDR, kinase insert domain receptor; LCK, lymphocyte-specific protein tyrosine kinase; NTRK1, neurotrophic tyrosine kinase
receptor type 1; STAT5B, signal transducer and activator of transcription 5; YES1, Yamaguchi sarcoma viral oncogene homolog 1.
giant component, whereas only 5% of enzyme targets only 12 are NTDs. Hence, it is clear that almost half of all
and 6% of miscellaneous targets are found here. Overall, drugs on the market which represent the indication
58% of the targets in the giant component are receptors classes in which a large research effort has been made
and all but one of these are GPCRs or ligand-gated ion share a similar target interaction profile and exploit a
channels two very traditional classes of pharmaco- limited part of the proteome. In addition, the main inno-
logical targets. The drugs in the giant component have a vations within this cluster of successful drugs occurred
significantly higher number of target interactions than before the one drug, one target focus of recent times.
drugs in the smaller components (P<0.001; Wilcoxon The success of some of the drugs in the giant component
signed-rank test). Drugs in the giant component have an could be due to the fact that they utilize multiple targets.
average of three targets, whereas drugs outside the giant This highlights the importance of polypharmacology
component have an average of 1.5 targets. and an understanding of the target profile in relation to
Targets within the giant component have a much the interactome for the discovery of newdrugs.
longer history than the network as a whole, consider-
ing the FDA approval dates of drugs within this compo- Discussion
nent. The drugs in the giant component have a median In this analysis, we observed an approximately steady
approval year before 1982, which is the cut-off date for rate of introduction of new drugs targeting molecules
drug approvals in our data set. Out of all of the drugs encoded by the human genome, and an approxi-
that were approved after 1982 in the giant component, mately steady rate of introduction of NTDs into the US
the development of new antibodies within fields such to be performed on an even larger scale and in closer
as cancer and inflammation, in which the drug target association with detailed clinical phenotyping, which
is often expressed at higher levels in the diseasestate. could help to validate the therapeutic relevance of such
These new opportunities highlight the need to gain targets, which arguably remains one of the crucial chal-
a deeper understanding of disease processes at the lenges in drug discovery research. Simultaneous exploi-
molecular level for a wider range of targets, as noted in tation of several targets for disease treatment also poses
an early paper on drug targets17. In recent years, a range a major challenge owing to the complexity of cellular
of new candidate targets have been identified from signalling and drugprotein interactions in the human
genome-wide association studies for several common body. The burgeoning concept of network pharma-
complex disorders such as obesity, diabetes, cancer and cology aims to quantify these interactions and could
cardiovascular disease. New technologies such as next- provide a general tool for identifying new targets for
generation sequencing will allow these types of studies rational drug discovery.
1. Goodman, L.S., Gilman, A., Hardman, J.G., Gilman, This was one of the first papers to present a 16. Kaitin, K.I. Deconstructing the drug development
A.G. & Limbird, L.E. Goodman & Gilmans The functional view on drugtarget interactions by process: the new face of innovation. Clin. Pharmacol.
Pharmacological Basis of Therapeutics 9th edn illustrating them as networks. This paper also Ther. 87, 356361 (2010).
(McGraw-Hill, New York,1996). integrates and analyses drug targets in the context 17. Drews, J. Drug discovery: a historical perspective.
2. Drews, J. Genomic sciences and the medicine of of disease genetics and proteinprotein Science 287, 19601964 (2000).
tomorrow. Nature Biotech. 14, 15161518 interactomics.
(1996). 10. Stegmeier, F., Warmuth, M., Sellers, W.R. & Dorsch, M. Acknowledgements
3. Hopkins, A.L. & Groom, C.R. The druggable genome. Targeted cancer therapies in the twenty-first century: The authors would like to thank J. Weigelt, Karolinska
Nature Rev. Drug Discov. 1, 727730 (2002). lessons from imatinib. Clin. Pharmacol. Ther. 87, Institute, Stockholm, for providing valuable comments on an
4. Imming, P., Sinning, C. & Meyer, A. Drugs, their 543552 (2010). early version of the manuscript. These studies were sup-
targets and the nature and number of drug targets. 11. Lagerstrom, M.C. & Schioth, H.B. Structural diversity ported by the Swedish Research Council, the Novo Nordisk
Nature Rev. Drug Discov. 5, 821834 (2006). of G protein-coupled receptors and significance for Foundation and the Swedish Brain Research Foundation.
5. Overington, J.P., Al-Lazikani, B. & Hopkins, A.L. How drug discovery. Nature Rev. Drug Discov. 7, 339357
many drug targets are there? Nature Rev. Drug (2008). Competing interests statement
Discov. 5, 993996 (2006). 12. Hardie, D.G. Role of AMP-activated protein kinase in The authors declare no competing financial interests.
6. Wishart, D.S. etal. DrugBank: a knowledgebase for the metabolic syndrome and in heart disease. FEBS
drugs, drug actions and drug targets. Nucleic Acids Lett. 582, 8189 (2008).
Res. 36, D901D906 (2008). 13. Zhu, M. etal. The analysis of the drugtargets based
7. Wishart, D.S. etal. DrugBank: a comprehensive on the topological properties in the human protein FURTHER INFORMATION
resource for insilico drug discovery and exploration. protein interaction network. J.Drug Target. 17, Drugs@FDA website: http://www.accessdata.fda.gov/
Nucleic Acids Res. 34, D668D672 (2006). 524532 (2009). scripts/cder/drugsatfda/
8. Knox, C. etal. DrugBank 3.0: a comprehensive This paper uses topological methodology to DrugBank database website: http://www.drugbank.ca
resource for omics research on drugs. Nucleic Acids analyse the properties of drug targets in the Drugs.com website: http://www.drugs.com
Res. 39, D1035D1041 (2011). context of proteinprotein interaction networks. FDA Orange book: http://www.accessdata.fda.gov/scripts/
This paper describes the most recent version of the This information is then used for target prediction. cder/ob/default.cfm
DrugBank database, which is one of the most 14. Hopkins, A.L. Network pharmacology: the next IMS Health: http://www.imshealth.com
important sources of information on drugs and drug paradigm in drug discovery. Nature Chem. Biol. 4,
targets. 682690 (2008). SUPPLEMENTARY INFORMATION
9. Yildirim, M.A., Goh, K.I., Cusick, M.E., Barabasi, 15. Paul, S.M. etal. How to improve R&D productivity: See online article: S1 (table) | S2 (box) | S3 (figure)
A.L. & Vidal, M. Drugtarget network. Nature the pharmaceutical industrys grand challenge. Nature ALL LINKS ARE ACTIVE IN THE ONLINE PDF
Biotech. 25, 11191126 (2007). Rev. Drug Discov. 9, 203214 (2010).