Vol. 10, No.

1 2013
Drug Discovery Today: Disease Models

Editors-in-Chief
Jan Tornell AstraZeneca, Sweden
DRUG DISCOVERY Andrew McCulloch University of California, SanDiego, USA
TODAY
DISEASE Zebrafish as a platform for in vivo drug discovery
MODELS

Zebrafish in pharmaceutical industry

research: finding the best fit
A. Fleming1,2,*, W.K. Alderton3
1
Department of Physiology, Development and Neuroscience, University of Cambridge, Downing Street, Cambridge CB2 3EG, UK4
2
Department of Medical Genetics, University of Cambridge, Cambridge Institute for Medical Research, Addenbrookes Hospital, Hills Road,
Cambridge CB2 0XY, UK5
3
CB1 Bio Ltd., Cambridge CB24 3AH, UK6

The growing literature on zebrafish disease models and

Section editors:
drug safety assessment suggests this organism may Calum A. MacRae Cardiovascular Division, Brigham and
have utility in the drug discovery process. Over the Womens Hospital, Boston, MA, USA.
Randall T. Peterson Cardiovascular Research Center,
past seven years, 24 zebrafish papers have been pub-
Massachusetts General Hospital, Charlestown, MA, USA.
lished with co-authors from pharmaceutical compa-
nies, suggesting that the model can be applied to (CROs) or academic groups to develop and validate relevant
industrial drug discovery. Here we review how the larval zebrafish assays.
pharmaceutical industry has used zebrafish to date
Where do zebrafish have utility in the drug discovery
and highlight the hurdles that currently prevent their
process?
wider acceptance in drug discovery research. Drug discovery and development is a long and expensive
process. It is estimated to take an average of 1015 years and
Introduction
up to $800 million to develop a new drug [2]. Furthermore, on
Over the past ten years, there has been a shift in the use of
average only five out of 250 compounds selected from in vitro
zebrafish (Danio rerio) purely for the academic study of verte-
drug discovery programs for pre-clinical testing will enter
brate development, to the modeling of disease processes and
clinical trials [2]. In the majority of cases this is because either
the use of such models in small molecule screens (reviewed in
the compound did not show sufficient therapeutic activity in
this issue by Peterson and MacCrae [1]). In addition, the use of
vivo or it had adverse effects and was therefore considered
larval zebrafish as a tool for studying toxicology and safety
unsafe. Because zebrafish can be used at multiple stages in
liabilities of pharmaceutical compounds is becoming estab-
drug discovery (Fig. 1), there is the potential to obtain in vivo
lished. This review focuses on how the pharmaceutical indus-
data on both efficacy and safety at the earliest possible oppor-
try has embraced the zebrafish as a tool for drug discovery and
tunity thereby potentially reducing attrition in the drug dis-
the barriers that currently limit the wider acceptance of this
covery process. Larval zebrafish assays have been utilized in the
model organism in drug discovery research. Here we review the
pharmaceutical industry in several different aspects of drug
published literature to present a summary of how pharmaceu-
discovery namely target validation, toxicology and safety
tical companies have set up zebrafish research in-house or have
pharmacology assessment, disease modeling and drug repro-
collaborated with zebrafish contract research organisations
filing (Fig. 1; Table 1). Toxicology and safety pharmacology are
*Corresponding author.: A. Fleming (af425@cam.ac.uk) areas that have gained most interest for the application of
4
http://www.pdn.cam.ac.uk/. larval zebrafish assays in the pharmaceutical industry. In 2008,
5
http://www.cimr.cam.ac.uk/.
6
http://www.cb1bio.com/. a survey of members of the Safety Pharmacology Society

1740-6757/$ .Crown Copyright 2012 Published by Elsevier Ltd. All rights reserved. DOI: 10.1016/j.ddmod.2012.02.006 e43
 Drug Discovery Today: Disease Models | Zebrafish as a platform for in vivo drug discovery Vol. 10, No. 1 2013

Gene Knockdown
Target Validation
Novel Target Validation

High Throughput Screening

Research
Efficacy and Safety Screening
Efficacy vs Safety: Therapeutic Window Lead Optimisation

Drug Safety
Safety Pharmacology, Toxicity Pre-Clinical Development

Development
Clinical Development

Reprofiling
New Indication Screening
Clinic
Drug Discovery Today: Disease Models

Figure 1. Schematic diagram illustrating where zebrafish assays can be used in drug discovery and development. Larval zebrafish assays can be used at multiple
points in the drug discovery process for assessment of target validation, efficacy and safety liabilities. Zebrafish assays can also be used for reprofiling/
repurposing/repositioning to identify new uses for approved or generic drugs and compounds that have failed in Phase 2 for reasons other than safety.

investigated attitudes towards zebrafish-based assays for safety relevance of the zebrafish model for certain safety evaluations
related endpoints [3]. 59% of respondents from pharmaceu- (http://www.fda.gov/AboutFDA/CentersOffices/NCTR/What-
tical/biotechnology companies indicated they were already WeDo/NCTRPublications/ucm197826.htm).
actively investigating zebrafish assays with CNS, cardiovascu-
lar, visual and auditory systems end-points highlighted as Toxicity
being of particular interest. Another indicator of the industrys Academic researchers in the environmental toxicity field
commitment to the zebrafish model system is the establish- have studied the effects of heavy metals, dioxins, polychlori-
ment of in-house zebrafish research facilities (see Table 1) with nated biphenyls and other pollutants on zebrafish for over 20
AstraZeneca, Pfizer, Bristol Myers Squibb and Novartis having years [4]. More recently, zebrafish have also been used to
made such investments. identify toxicity liabilities of pharmaceutical compounds
such as hepatotoxicity, reproductive toxicity/teratogenicity,
Zebrafish in drug safety assessment neurotoxicity, ototoxicity and acute toxicity.
Drug safety assessment is an integral part of pre-clinical Hepatotoxicants are usually identified during the clinical
development and is a highly regulated and conservative development programme of a new drug, thus causing attri-
environment where the adoption of new technologies is tion of clinical candidates. Recently, J&J and Evotec plc
slow. The emergence in the past 10 years of the use of together developed and validated a larval zebrafish assay
zebrafish larvae in non-GLP pre-clinical safety assessment for the identification of hepatoxicants which had 91% pre-
in the pharmaceutical industry has been incremental, but dictivity [5] (Table 2).
the zebrafish is not yet accepted as a standard tool in drug Zebrafish assays for embryotoxicity/teratogenicity have
safety evaluation. Because zebrafish assays are being used in recently been described by authors from Bristol-Myers Squibb
the drug discovery phase rather than regulated pre-clinical and Janssen R&D (J&J) [6] and assay methodology was dis-
studies, they will not necessarily be subject to Food and Drug cussed at a workshop sponsored by HESI (Health and Envir-
Agency (FDA) regulations for pre-clinical IND submissions. onmental Sciences Institute) with representation from
However, the National Center for Toxicological Research (a several pharmaceutical companies [7]. Over the past decade,
research component of the FDA) announced the opening of a pharmaceutical companies have been increasingly reluctant
zebrafish facility for high-throughput toxicity assessment in to validate and implement new technologies alone, therefore
March 2010, suggesting that the FDA does appreciate the consortia of several companies evaluating large numbers of

e44 www.drugdiscoverytoday.com
 Vol. 10, No. 1 2013 Drug Discovery Today: Disease Models | Zebrafish as a platform for in vivo drug discovery

Table 1. Publications by pharmaceutical companies using larval zebrafish assays (2006February 2012)
Pharmaceutical company Assay In-house facility/academic collaborator References
or zebrafish CRO
Abbott Cardiac function Academic collaborator [11]

AstraZeneca Visual function Zebrafish CRO [16]

AstraZeneca Seizure liability In-house [14]

AstraZeneca Tauopathy Academic collaborator [19]

AstraZeneca ADME Academic collaborator [29]

AstraZeneca Ototoxicity Academic collaborator [27]

Bristol-Myers Squibb Embryotoxicity/teratogenicity In-house [30,31]

Eli Lilly Bone formation Zebrafish CRO [18]

Eli Lilly Primordial germ cell culture Academic collaborator [32]

GlaxoSmithKline Embryotoxicity/teratogenicity Academic collaborator [33]

Johnson & Johnson; Pfizer Hepatotoxicity Zebrafish CRO [34]

Johnson & Johnson Embryotoxicity/teratogenicity Zebrafish CRO [6]

Merck KGaA Embryotoxicity/teratogenicity Academic collaborator [3538]

Novartis Developmental and molecular biology In-house [3941]

Novatis Gastrointestinal motility In-house and with academic collaborator [42]

Novartis Toxicology; determining mechanism of action Academic collaborator [43]

Pfizer Safety pharmacology assays Zebrafish CRO [10]

Pfizer ADME In-house/Zebrafish CRO [22]

compounds have been a favored approach for validating new one laboratory, rigorously tested with 100 compounds and
technologies such as the zebrafish. One such group investi- will be amenable to an automated screening platform. The
gating zebrafish teratogenicity assays is the cross-pharma consortium is expected to complete this project in early 2012.
consortium of Bristol-Myers Squibb, Pfizer, Amgen, AstraZe- Zebrafish have also proved useful for predicting acute
neca and the zebrafish CRO Lampire Biological Laboratories toxicity of novel compounds in the lead optimization phase
[8]. The goal of this consortium is to generate a harmonized of a drug discovery project at Pfizer, UK, guiding medicinal
zebrafish teratogenicity assay to be used across the pharma- chemistry synthesis of less toxic derivatives (personal com-
ceutical industry. The assay will be validated in more than munication, Paul Butler Pfizer, 2011).

Table 2. Predictivity of larval zebrafish screens in drug safety testing

Assay Endpoint No. of compounds Predictivity References
Hepatotoxicity Phenotypic 50 91% [5]

Embryotoxicity/teratogenicity Phenotypic 31 87% [30]

15 75% [6]

Cardiac function 2:1 A:V decoupling 23 78% [12]

18 90% [11]

9 78% [10]

Visual function Optomotor response 27 70% [16]

Optomotor response 9 78% [10]

Convulsant activity Locomotor activity 25 72% [14]

www.drugdiscoverytoday.com e45
 Drug Discovery Today: Disease Models | Zebrafish as a platform for in vivo drug discovery Vol. 10, No. 1 2013

Safety pharmacology 10 compounds per week and may not be amenable to sig-
Zebrafish have been applied to in vivo pharmacological assess- nificant scale up and automation.
ment to identify a wide range of safety liabilities including
cardiac, gastrointestinal, renal, visual and auditory function Zebrafish efficacy models
and CNS assessments such as cognitive impairment, abuse There is great potential for the use of zebrafish in disease
potential and seizure liability [9]. Larval zebrafish assays for modeling and in high content small molecule screening [1].
drug safety assessment are most applicable where there is Academic groups have spearheaded this approach, with
good predictive validity and automation capacity for a parti- numerous zebrafish screens identifying compounds that have
cular safety area and where no mammalian cell-based assays subsequently had their activity validated in rodent assays,
are possible. For example, QT prolongation has been com- and there is at least one example of a compound identified in
prehensively addressed by in vitro hERG assays, which have a primary screen in larval zebrafish that is now in clinical
>90% predictivity and now cost approximately $1 a well. trials (reviewed in [17]). However, as is apparent from Table 1,
While zebrafish larvae at 3 d.p.f. have been investigated for there has been less interest from the pharmaceutical industry
predicting QT prolongation by measuring a distinctive heart in zebrafish efficacy than in drug safety screening. To under-
arrhythmia [1012], this assay is unable to compete in terms stand this imbalance, one needs to look at the traditional
of cost or throughput with in vitro hERG assays. However, drug discovery process favored by pharmaceutical research
zebrafish may be suitable for assessing integrated cardiac groups (Fig. 1). Typically, drug discovery starts with the
function. Two areas of drug safety assessment where the identification of a novel target and the validation of this
zebrafish show potential for industry adoption are visual target. Compounds are screened for their efficacy against this
function and seizure liability. target and through a series of iterations, lead compounds are
identified with increasing specificity and efficacy. By con-
Seizure liability trast, zebrafish assays are blackbox screens, where the target
Clinical candidates are routinely assessed for the potential to is not known but a library is screened to identify biological
cause tonic/clonic convulsions or for having a proconvulsant compounds that rescue a disease phenotype. While this type
effect by reducing the threshold to seizure-inducing triggers. of screen may be attractive to pharmaceutical companies for
Models using cells or tissues do not replicate the complex reprofiling (also known as repurposing or repositioning)
phenomenon of seizures in integrated systems. Zebrafish drugs it does not fit the traditional drug discovery project
larvae at 7 d.p.f. respond to convulsants such as pentylenete- model.
trazole with a distinct series of movements evolving from For the pharmaceutical industry to adopt a model for
increased swimming activity to rapid whirlpool-like beha- efficacy screening, it needs to meet several criteria: the model
vior and tectal whole-field recordings produce electrical activ- should show pathological changes that are comparable to
ity characteristic of convulsions in human [13]. A study by those seen in humans; the severity of disease should be
AstraZeneca, using locomotor tracking to calculate the num- quantifiable; the disease should be ameliorated by treatment
ber of high speed movements as a read-out of seizure-like with a positive control compound; the model should be
behavior, achieved a predictivity of 72% using a validation set validated with any drugs currently used in the clinic for
of compounds with known seizure liability and negative the disease indication; the model should be amenable to
controls [14] (Table 2). automation or high-throughput screening. One final point
to consider is the false-negative rate for any given assay.
Visual function While academic groups will usually accept a high false-nega-
Because zebrafish have rich color vision, they offer a distinct tive rate because the aim of the work is to identify a small
advantage over visual function testing in nocturnal rodents. number of hits for subsequent analysis, traditional industry
By 5 d.p.f., the visual system is well developed and assays such drug discovery screens require higher confidence. For zebra-
as the optokinetic response (OKR) and optomotor response fish models to gain traction with the pharmaceutical indus-
(OMR) take advantage of inherent visual reflexes and have try, it is important to understand whether a compound fails
been developed for compound testing [15]. An AstraZeneca owing to lack of efficacy or for other reasons (e.g. poor
and DanioLabsLtd study in the OMR assay evaluated whether compound uptake, low homology with the human target).
8 d.p.f. zebrafish could be useful in predicting the adverse It is therefore worth considering not where zebrafish models
effects of drugs on visual function in human [16]. The OMR could be of benefit in the drug discovery process, but where
assay showed a good concordance between the effects of they offer an advantage over the conventional in vivo mam-
compounds on zebrafish larvae with the data available from malian efficacy assays:
the clinic or from other in vivo models with a predictivity
of 70% (Table 2). However, while the OMR assay had accep- 1) Time benefit for screening in zebrafish: One major advantage
table predictivity, the throughput was only approximately of using zebrafish assays over traditional mammalian

e46 www.drugdiscoverytoday.com
 Vol. 10, No. 1 2013 Drug Discovery Today: Disease Models | Zebrafish as a platform for in vivo drug discovery

models is the short duration of the assays. For example, effect is transient and gene expression may have returned to
work by Eli Lilly and DanioLabs Ltd. established that normal levels if the disease of interest is modeled in larvae
zebrafish larvae could be used to assess bone formation older than 3 d.p.f. However, the emergence of zinc-finger
and bone loss and could provide a useful tool to investi- nuclease technologies [20] and the recent advances in
gate compounds for the treatment of osteoporosis [18]. In whole exome sequencing of mutagenised zebrafish [21],
rodents, the standard model for osteoporosis involves offer other approaches for making or identifying fish carry-
ovariectomy followed by a period of between 0 and 10 ing null mutations in a gene of interest.
weeks to allow bone loss to occur, then from 2 to 56 weeks
of compound treatment. By contrast, the larval zebrafish
assay was performed in 7 days offering significant time Hurdles to the acceptance of zebrafish assays by the
benefits and reduced compound usage. Another area pharma industry
where studies in zebrafish larvae may significantly reduce Despite the promising zebrafish safety, toxicology and effi-
assay length is in neurodegeneration research. Using a cacy studies described above, there are still significant hurdles
transgenic zebrafish model of tauopathy, zebrafish to be addressed before the zebrafish is accepted as a widely
researchers at Deutsches Zentrum fur Neurodegenerative used model organism in the drug discovery process.
Erkrankungen and the Ludwig-Maximilians-University,
Munich, Germany in collaboration with AstraZeneca, Lack of understanding of compound uptake, distribution and
demonstrated that pathological hyperphosphorylated metabolism in zebrafish larvae
tau could be detected as early as 24 h.p.f. and accumulated In conventional in vivo experiments, a known and measur-
up to 7 d.p.f. in larval transgenic zebrafish [19]. Changes in able amount of drug is administered to the animal. However,
these disease markers were successfully used to evaluate the non-uniform and unpredictable uptake from the well
novel GSK3b inhibitors. In comparison, the same patho- water into the zebrafish larvae can make it difficult to inter-
logical hyperphosphorylated forms of tau only appear in pret screening results and make comparisons between com-
rodent models of tauopathy from 4 months old. pounds. A study on cardiac effects of compounds in zebrafish
2) Ethical advantages: One potential driver for the growing [11] noted significant right shifted pharmacology when com-
use of the zebrafish model is the pressure on both indus- paring the well concentrations of compounds to results in
trial and academic researchers to reduce, replace or refine mammalian assays because compound uptake by the larvae
experiments using vertebrates and to look for more was not taken into consideration. In a study by Pfizer and
humane alternatives for in vivo research. Because zebrafish DanioLabs/Summit plc to measure compound absorption
are considered lower vertebrates by some legislative and metabolism in larval zebrafish, large differences were
authorities, the replacement of mammalian models with observed in the concentration of drug in the larvae compared
zebrafish alternatives may be advantageous while still to that of the media after 3 h of drug exposure with some
offering the advantages of using an in vivo model. drugs concentrated within the larval tissue and others not
3) Drug reprofiling: Reprofiling is the identification of new detectable [22]. In a separate study, this group also found
uses for existing drugs and clinical candidate molecules. significant changes in uptake at different larval ages for
This is an attractive way for pharmaceutical researchers certain compounds [10]. A further Pfizer study of 45 com-
extend the reach and lifetime of their patents by finding pounds concluded that no single physico-chemical property
additional uses for drugs in their portfolio. By using (e.g. c Log P, c Log D, pKa, molecular weight) could accurately
libraries of compounds with known safety and toxicity predict compound uptake [23], making it extremely difficult
data, this approach can reduce the time, cost and risk of to predict how any novel compound is likely to behave
drug development. Although there are no published stu- within the system. In addition, our understanding of the
dies by pharmaceutical companies employing such metabolism of compounds by zebrafish is still in its infancy
screens, this may reflect the commercial sensitivity of their with a few publications to date on a limited number of
libraries rather than a reluctance to use the zebrafish metabolites (reviewed in this issue Hill [24]).
model. A further limitation is the lack of knowledge about the
4) Target validation: Although genomic and proteomic bloodbrain barrier in zebrafish larvae. Studies have shown
approaches can be employed to validate that a particular evidence of a bloodbrain and bloodretinal barrier in the
biological target is relevant to a disease, the generation of a larval zebrafish as early as 3 d.p.f., co-incident with the
mouse knock-out is considered the gold standard for target expression of tight junction markers such as claudin5a and
validation. The use of morpholino oligonucleotides (MOs) ZO1 [25]. However, further work is required to understand
for transient gene knockdown in zebrafish may be advanta- whether drugs show similar exclusion to that in mammals
geous in this area in terms of time, cost and ethical con- and whether active uptake and efflux mechanisms exist.
siderations. The major caveat to the use of MOs is that their Therefore, the assessment of CNS-mediated effects in larvae

www.drugdiscoverytoday.com e47
 Drug Discovery Today: Disease Models | Zebrafish as a platform for in vivo drug discovery Vol. 10, No. 1 2013

(a) (b)

3 2 3 2
4 1 4
1

5 7
6 7
6

(c)

8 10
9

11
13 12

Drug Discovery Today: Disease Models

Figure 2. Assessing multiple endpoints in embryotoxicity/teratogenicity assays. In embryotoxicity/teratogenicity assays, embryos and larvae are visually
scored for changes in multiple (up to 40) morphological features at different timepoints. In control embryos at 24 h.p.f., the eye (1), ventricles (2), midbrain/
hindbrain boundary (3), otic vesicle (4), notochord (5), somites (6) and tail (7) can easily be visualized. In embryos treated with the embryotoxic compound,
methyl mercury chloride, defects in the eye (1), ventricles (2) and midbrain/hindbrain boundary (3) readily apparent. The otic vesicle (4) is absent, the
somites (6) appear rounded rather than chevron-shaped and the tail (7) is reduced in length. Teratogenicity assays typically include analysis of larvae at a
later timepoint, for example, 5 d.p.f. (C), where additional morphological features such as pigment (8), muscle (9), fins (10), cloaca and uritogenital opening
(11), heart (12) and jaw (13) can be assessed. Although advances have been made in image recognition software, the automated analysis of multiple features
required for embryotoxicity assays, when features may be absent, malformed or misplaced is likely to be technically challenging.

may erroneously identify safety liabilities or efficacy in com- Challenges in scale-up and automation
pounds that are excluded from the brain in older fish and in Larval zebrafish assays require a throughput of 100s to 1000s
mammals. of compounds a week to leverage the advantages of this
model system and for it to be deployed early in drug dis-
Lack of large scale validation studies covery. Ideally, the zebrafish assay should be run in 96 well
Table 2 shows the predictivity of zebrafish safety assays in plate format to achieve equivalent throughput to in vitro
published validation studies by the pharmaceutical industry assays. Recent technological advances have overcome some
in partnership with zebrafish CROs. While the predictivity of the existing bottlenecks such as large scale embryo pro-
established in a limited number of assays to date is good (70 duction and embryo dispensing into multi-well plates. How-
91%), more extensive studies, such as that being undertaken ever some assays, such as embryotoxicity (which requires
by the teratology cross-pharma consortium described above, visual scoring under a microscope of up to 40 endpoints),
are required for wider acceptance of the zebrafish model. In have been challenging to automate (Fig. 2). The OMR and
the field of drug safety assessment, there is industry-wide OKR assays for visual function also have limited throughput
agreement on the standard set of compounds with which a [15,16]. The development of multi-arena behavioral analysis
new assay should be tested, making it easier to perform software for movement detection and high-throughput
validation studies from which the results can be compared microscopy imaging hardware and recognition software sug-
to existing assays. By contrast, efficacy assays tend to be gests however, that these problems are tractable.
validated with different sets of compounds, depending on
the targets of interest for each pharmaceutical company. Comparison between industry-accepted in vitro assays and novel
Furthermore, it is often hard to find positive controls to zebrafish assays
use in efficacy assays because there may be no or few mar- Proponents of the zebrafish model would argue that a major
keted drugs for a particular disease indication. advantage of their use of is that, in vivo, multiple targets are

e48 www.drugdiscoverytoday.com
 Vol. 10, No. 1 2013 Drug Discovery Today: Disease Models | Zebrafish as a platform for in vivo drug discovery

tested in a single assay. By contrast, it is commonly the case pharmaceutical industry and academic groups with specific
that a single in vitro assay measures activity of a compound areas of expertise.
against a single target. However, this potentially creates a
problem in the interpretation and extrapolation between the Conclusion
two types of assay (e.g. in the calculation of predictivity, see Over the past ten years, there has been sustained interest in
Table 2). As discussed earlier, in vitro hERG binding and patch- the zebrafish as a discovery tool in the pharmaceutical indus-
clamp assays are accepted by both the pharmaceutical indus- try. However, further large scale validation studies and clar-
try and regulatory bodies as suitable methods for investigat- ification on compound uptake measurement are vital to gain
ing the potential of compounds to cause QT prolongation. full acceptance and widespread use of the model in industry.
However, these assays only measure the affinity of com- The larval zebrafish teratogenicity assay may deliver in 2012
pounds for the hERG channel. By comparison, the distinctive in this respect. Without such studies zebrafish assays are only
2:1 atrial:ventricular arrhythmia observed in larval zebrafish likely to be adopted in niche areas such as drug reprofiling, or
can be used to identify compounds which block not only the for where no cell based equivalent assay can recapitulate the
hERG channel but also other cardiac ion channels, for exam- system, for example, seizure liability and visual function
ple, YS-035, the L-type calcium channel blocker causes the assays for drug safety assessment.
same 2:1 atrial:ventricular arrhythmia [26]. While this could
be argued as advantageous, because the zebrafish assay can Conflict of interest statement
identify arrhythmias arising from blockage of a range of The authors declare they have no conflicts of interest.
cardiac ion channels, it raises problems when investigators
attempt to compare their results with those from single target Acknowledgements
in vitro assays. There is a danger that inconsistencies between We are indebted to the following for insightful comments:
the in vivo and in vitro datasets are likely to be interpreted as Paul Butler and Mike Aleo (Pfizer), Derek Leishman (Eli Lilly),
false-positives when they may actually be caused by the drug Will Redfern and Matt Winter (Astra Zeneca), Alan Roach
acting on a target not represented in the in vitro model. (Vastrata Ltd), Frances Richards (Cancer Research UK) and
Stephane Berghmans (European Science Foundation). A.F. is
Strategies to facilitate the use of zebrafish in funded by a Medical Research Council Skills Gap Award.
pharmaceutical research
References
Many of these hurdles are tractable and the challenge now is
1 Peterson, R.T. and MacRae, C.A. (2012) Drug screening in the zebrafish: an
how best to address these problems. The teratogenicity con- overview. Drug Discov. Today: Disease Models (in press)
sortia described above [8] is an excellent way of achieving 2 Margolis, J. and Plowman, G.D. (2004) Overcoming the gridlock in
discovery research. Nat. Biotechnol. 22, 522524
large-scale validation and cross-company agreement on
3 Redfern, W.S. et al. (2008) Zebrafish assays as early safety pharmacology
methods. A second strategy is to facilitate interactions with screens: paradigm shift or red herring? J. Pharmacol. Toxicol. Methods 58,
academic groups. All the validation studies listed in Table 2 110117
were performed either by pharmaceutical companies in- 4 Scholz, S. et al. (2008) The zebrafish embryo model in environmental risk
assessment applications beyond acute toxicity testing. Environ. Sci. Pollut.
house or in collaboration with zebrafish CROs. Comments Res. Int. 15, 394404
from researchers in the pharmaceutical industry suggest that 5 Hill, A.J. et al. (2010) A review of early discovery hepatotoxicity screening
they view academic zebrafish researchers as experts in disease using the in vivo zebrafish model. 31st Annual Meeting of the American
College of Toxicology
modeling but interactions between the two groups have been
6 Van den Bulck, K. et al. (2011) Zebrafish developmental toxicity assay: a
limited. A new approach to fostering such interactions is the fishy solution to reproductive toxicity screening, or just a red herring?
creation of academic screening facilities such as the MRC Reprod. Toxicol. 32, 213219
7 Chapin, R. et al. (2008) State of the art in developmental toxicity screening
Centre for Developmental and Biomedical Genetics (CDBG)
methods and a way forward: a meeting report addressing embryonic stem
Screening Unit at Sheffield University (http://cdbg.shef. cells, whole embryo culture, and zebrafish. Birth Defects Res. B Dev. Reprod.
ac.uk/screening/zebrafish.html) and the Zebrafish Screening Toxicol. 83, 446456
and Stock Facility at Karlsruhe Institute of Technology 8 Gustafson, A.L. et al. (2012) Inter-laboratory assessment of a harmonized
zebrafish developmental toxicology assay-progress report on phase I.
(http://www.itg.kit.edu/248.php). The CDBG has already Reprod. Toxicol., doi:10.1016/j.reprotox.2011.12.004
completed several screens and other studies with industry 9 Barros, T.P. et al. (2008) Zebrafish: an emerging technology for in vivo
partners [27], with further studies expected to be published pharmacological assessment to identify potential safety liabilities in early
drug discovery. Br. J. Pharmacol. 154, 14001413
later in 2012 (Stephen Renshaw, personal communications).
10 Berghmans, S. et al. (2008) Zebrafish based assays for the assessment of
The increased funding available for academic/industrial part- cardiac, visual and gut function potential safety screens for early drug
nerships (e.g. MRC Industrial Collaboration Award; Well- discovery. J. Pharmacol. Toxicol. Methods 58, 5968
11 Mittelstadt, S.W. et al. (2008) Evaluation of zebrafish embryos as a model for
come Trust Health Innovation Challenge Fund) and the
assessing inhibition of hERG. J. Pharmacol. Toxicol. Methods 57, 100105
NIH strategy to support more risky drug discovery projects 12 Milan, D.J. et al. (2003) Drugs that induce repolarization abnormalities
[28] may help to facilitate collaborations between the cause bradycardia in zebrafish. Circulation 107, 13551358

www.drugdiscoverytoday.com e49
 Drug Discovery Today: Disease Models | Zebrafish as a platform for in vivo drug discovery Vol. 10, No. 1 2013

13 Baraban, S.C. et al. (2005) Pentylenetetrazole induced changes in zebrafish 30 Brannen, K.C. et al. (2010) Development of a zebrafish embryo
behavior, neural activity and c-fos expression. Neuroscience 131, teratogenicity assay and quantitative prediction model. Birth Defects Res. B
759768 Dev. Reprod. Toxicol. 89, 6677
14 Winter, M.J. et al. (2008) Validation of a larval zebrafish locomotor assay 31 Panzica-Kelly, J.M. et al. (2010) Morphological score assignment
for assessing the seizure liability of early-stage development drugs. J. guidelines for the dechorionated zebrafish teratogenicity assay. Birth
Pharmacol. Toxicol. Methods 57, 176187 Defects Res. B Dev. Reprod. Toxicol. 89, 382395
15 Alderton, W. (2011) Assessment of effects on visual function in zebrafish. 32 Fan, L. et al. (2008) Zebrafish primordial germ cell cultures derived from
In Zebrafish: Methods for Assessing Drug Safety and Toxicity (McGrath, P., vasa::RFP transgenic embryos. Stem Cells Dev. 17, 585597
ed.), pp. 191204, Wiley 33 Donaldson, L.R. et al. (2011) Rapid synthesis and zebrafish evaluation of a
16 Richards, F.M. et al. (2008) Validation of the use of zebrafish larvae in phenanthridine-based small molecule library. Org. Biomol. Chem. 9,
visual safety assessment. J. Pharmacol. Toxicol. Methods 58, 5058 22332239
17 Bowman, T.V. and Zon, L.I. (2010) Swimming into the future of drug 34 Hill, A. et al. (2012) Comparisons between in vitro whole cell imaging and
discovery: in vivo chemical screens in zebrafish. ACS Chem. Biol. 5, 159161 in vivo zebrafish-based approaches for identifying potential human
18 Fleming, A. et al. (2005) High-throughput in vivo screening for bone hepatotoxicants earlier in pharmaceutical development. Drug Metab. Rev.
anabolic compounds with zebrafish. J. Biomol. Screen. 10, 823831 44, 127140
19 Paquet, D. et al. (2009) A zebrafish model of tauopathy allows in vivo 35 Busquet, F. et al. (2008) Development of a new screening assay to identify
imaging of neuronal cell death and drug evaluation. J. Clin. Invest. 119, proteratogenic substances using zebrafish Danio rerio embryo combined
13821395 with an exogenous mammalian metabolic activation system (mDarT).
20 Ekker, S.C. (2008) Zinc finger-based knockout punches for zebrafish genes. Toxicol. Sci. 104, 177188
Zebrafish 5, 121123 36 Weigt, S. et al. (2010) Zebrafish teratogenicity test with metabolic
21 Kettleborough, R. et al. (2011) The zebrafish mutation project (ZMP): activation (mDarT): effects of phase I activation of acetaminophen on
whole exome sequencing of mutagenised zebrafish to identify a non-sense zebrafish Danio rerio embryos. Toxicology 275, 3649
allele in every protein coding gene. 7th European Zebrafish Meeting 37 Weigt, S. et al. (2011) Zebrafish (Danio rerio) embryos as a model for testing
22 Alderton, W. et al. (2010) Accumulation and metabolism of drugs and CYP proteratogens. Toxicology 281, 2536
probe substrates in zebrafish larvae. Xenobiotica 40, 547557 38 Weigt, S. et al. (2011) Developmental effects of coumarin and the
23 Doshna, C. et al. (2009) Multi-phase analysis of uptake and toxicity in anticoagulant coumarin derivative warfarin on zebrafish (Danio rerio)
zebrafish: relationship to compound physicalchemical properties. Society embryos. Reprod. Toxicol.
of Toxicology, 48th Annual Meeting 39 Shawi, M. and Serluca, F.C. (2008) Identification of a BMP7 homolog in
24 Diekmann, H. and Hill, A.J. (2012) ADMETox in the zebrafish. Drug Discov. zebrafish expressed in developing organ systems. Gene Expr. Patterns 8,
Today: Disease Models doi:10.1016/j.ddmod.2012.02.005 369375
25 Abdelilah-Seyfried, S. (2010) Claudin-5a in developing zebrafish brain 40 Serluca, F.C. (2008) Development of the proepicardial organ in the
barriers: another brick in the wall. Bioessays 32, 768776 zebrafish. Dev. Biol. 315, 1827
26 Langheinrich, U. et al. (2003) Zebrafish embryos express an orthologue of 41 Schebesta, M. and Serluca, F.C. (2009) olig1 Expression identifies
HERG and are sensitive toward a range of QT-prolonging drugs inducing developing oligodendrocytes in zebrafish and requires hedgehog and
severe arrhythmia. Toxicol. Appl. Pharmacol. 193, 370382 notch signaling. Dev. Dyn. 238, 887898
27 Buck, L.M. et al. (2011) Ototoxin-induced cellular damage in neuromasts 42 Field, H.A. et al. (2009) Analysis of gastrointestinal physiology using a
disrupts lateral line function in larval zebrafish. Hear Res. novel intestinal transit assay in zebrafish. Neurogastroenterol. Motil. 21,
28 Collins, F. (2011) An audience with. Francis Collins. Interviewed by Asher 304312
Mullard. Nat. Rev. Drug Discov. 10, 14 43 Oggier, D.M. et al. (2011) Effects of the protein kinase inhibitor PKC412 on
29 Jones, H.S. et al. (2010) Oxidative and conjugative xenobiotic metabolism gene expression and link to physiological effects in zebrafish Danio rerio
in zebrafish larvae in vivo. Zebrafish 7, 2330 eleuthero-embryos. Toxicol. Sci. 119, 104115

e50 www.drugdiscoverytoday.com