Download as pdf or txt
Download as pdf or txt
You are on page 1of 10

Sensors and Actuators B 189 (2013) 11–20

Contents lists available at ScienceDirect

Sensors and Actuators B: Chemical


journal homepage: www.elsevier.com/locate/snb

Fish on chips: Microfluidic living embryo array for accelerated in vivo


angiogenesis assays
Jin Akagi a , Khashayar Khoshmanesh a,b , Chris J. Hall c , Jonathan M. Cooper d , Kathryn E. Crosier c ,
Philip S. Crosier c , Donald Wlodkowic a,e,∗
a
The BioMEMS Research Group, School of Chemical Sciences, University of Auckland, Auckland, New Zealand
b
School of Electrical and Computer Engineering, RMIT University, Melbourne, Australia
c
Department of Molecular Medicine and Pathology, University of Auckland, Auckland, New Zealand
d
School of Engineering, University of Glasgow, Glasgow, UK
e
School of Applied Sciences, RMIT University, Melbourne, Australia

a r t i c l e i n f o a b s t r a c t

Article history: Zebrafish (Danio rerio) has recently emerged as a powerful experimental model in molecular medicine
Received 14 September 2012 and drug discovery. Nevertheless, automated in situ analysis of zebrafish embryos is still in early devel-
Received in revised form 4 November 2012 opment. Currently available technologies do not allow for an automated loading, docking and spatial
Accepted 12 November 2012
address designation to large numbers of single embryos during imaging. Moreover, microperfusion treat-
Available online 21 November 2012
ment of large numbers of immobilized zebrafish embryos is still inaccessible. In this work, we describe the
proof-of-concept design of a 3D microfluidic embryo array for real-time developmental analysis of trans-
Keywords:
genic zebrafish embryos. The Lab-on-a-Chip system was fabricated directly in poly(methyl methacrylate)
Microfluidics
Lab-on-a-Chip
(PMMA) transparent thermoplastic using infrared laser micromachining. The multilayer chip contains a
Zebrafish trap-and-release immobilization manifold with an array of micro-mechanical traps that capture single
Embryo fish embryos. The docking is assisted by the combination of gravitation and low-pressure suction at the
Transgenic bottom plane of the device. The design achieves one-embryo-one-trap for convenient address designation
Bioassay and spatial encoding to each embryo and is capable of high-throughput docking and recovery of single
Angiogenesis embryos at a large scale. We also present data that the microfluidic embryo array can be readily applied to
kinetic analysis of investigational pharmacological agents inhibiting blood vessel growth (angiogenesis)
in transgenic Tg(fli1a:EGFP) zebrafish. The work provides a foundation for automated screening of intact
metazoan model organisms in drug discovery using Lab-on-a-Chip.
© 2012 Elsevier B.V. All rights reserved.

1. Introduction fields such as vascular, lymphatic, immune and neural biol-


ogy [5–7]. Mushrooming reports have validated a plethora of
Small multicellular model animals such as zebrafish (Danio fluorescent transgenic zebrafish models as surrogates for large-
rerio) have recently revolutionized the tool-box of modern scale chemical screening and chemogenomics in drug discovery
biomedical research [1–3]. They provide innovative analytical [8–10].
capabilities that cannot be easily replicated using isolated cells Despite the biomedical advantages of the zebrafish model sys-
and tissues, such as organogenesis, tissue regeneration, drug accu- tem, the dispensing, treatment and analysis of zebrafish embryos
mulation, metabolism and organ-specific toxicity in the living is still largely manual and labour-intensive [1–3]. Recent innova-
intact organism. They also bridge the gap between traditional tions include zebrafish screening in high-density microtiter plates
high-throughput cell-based assays (in vitro) and low-throughput, with embryos dispensed using robotic liquid handling systems
rodent in vivo tests [3–5]. The conserved vertebrate developmental [3,11]. The bioassays carried out under static microtiter plate con-
biology, transparent properties and straightforward transgene- ditions can be, however, inadequate to test many pharmaco-active
sis of zebrafish have facilitated accelerated bioanalysis in many compounds as a result of their adsorption, degradation, metabolic
inactivation, shortage of oxygen, and uncontrolled changes in
medium pH [12,13]. The ability to perform drug exposures under
perfusion without displacing the embryos during analysis and
∗ Corresponding author at: The BioMEMS Research Group, School of Chemical
disturbing their development is, therefore, of utmost importance
Sciences, University of Auckland, 23 Symonds Street, Building 301, Science Centre,
[1,2,12].
1142 Auckland, New Zealand. Tel.: +64 9 373 7599; fax: +64 9 373 7422.
E-mail addresses: d.wlodkowic@auckland.ac.nz, donald.wlodkowic@rmit.edu.au
The drawbacks associated with manual handling and analy-
(D. Wlodkowic). sis of developing fish embryos can be addressed by the emerging

0925-4005/$ – see front matter © 2012 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.snb.2012.11.036
12 J. Akagi et al. / Sensors and Actuators B 189 (2013) 11–20

field of microfluidic Lab-on-a-Chip (LOC) technologies [1,2]. As an 2.2. Chip fabrication


investigative tool, LOCs represent a new direction that is poised
to miniaturize and revolutionize research on physiology in vivo. The chip was designed and modeled using CorelDraw X4 (Corel
Zebrafish embryo and larvae analysis was previously shown on Corporation, Ontario, Canada) and SolidWorks 2011 (Dassault Sys-
chip-based systems [14–16]. These noteworthy reports were, how- temes SolidWorks Corp, Concord, MA, USA) CAD packages. Physical
ever, designed for automatic manipulation of hatched larvae or prototyping of integrated 3D microfluidic devices was performed
manual handling of chorionated embryos for developmental anal- in poly(methyl methacrylate) (PMMA) transparent thermoplastic
ysis [14,17,18]. The latter chip-based systems completely lacked using a non-contact 30 W infrared laser micromachining system
integrated and automated operation and thus scalability for drug with a 50 ␮m elliptical beam spot (Universal Laser Systems, AZ,
discovery [15,16,19]. In this regard we have recently demonstrated USA). All layers were optically aligned and thermally bonded at
the passive trapping and immobilization of zebrafish embryos 110 ◦ C for up to 2 h. Leak free fluidic connections were accom-
using a simple fluidic trap on a chip-based device [2]. The differ- plished by direct laser machining of fluidic interconnects to fit the
ential resistance between the main channel and trapping channel 1/16 in. PTFE tubing (Cole-Parmer, IL, USA) with an internal diam-
allowed immobilizing large numbers of zebrafish embryos [2]. This eter (ID) 1.5 mm ID. The tubing was reversibly plugged into the
passive hydrodynamic designs suffered, however, from subopti- devices fluidic ports and connected to miniaturized stepper-motor
mal trapping efficiencies that were highly susceptible to operating peristaltic (Kamoer, China) and piezoelectric ultrasonic (Bartels
conditions and proper chip priming. Also that design was not Mikrotechnic, Germany) pumps.
susceptible for high-levels of automation, throughput and user
friendliness [2]. 2.3. Computational fluid dynamics (CFD) simulations
Here, we for the first time report on inherently scalable 3D,
multi-layer microfluidic system for automated and high efficiency 2D and 3D models of the device were created with virtual
trapping and immobilization of living zebrafish embryos. In con- embryos as spherical structures inside the traps. The simulation
trast to any previously described technologies, our design exploits was performed using Gambit 2.3 software (Fluent, Lebanon, NH,
the gravitational-induced sedimentation of embryos combined USA) to create the geometry and mesh generation. Finite-volume
with low-pressure suction at the bottom plane of the device to based Fluent 6.3 software (Fluent) was subsequently used to solve
rapidly trap embryos. The 3D array design features considerable the associated differential equations governing the balance of mass,
improvement in the efficiency of trapping compared to the pas- momentum, chemical species, as given in Supplementary Data.
sive fluidic trap while substantially reducing the complexity of
operation and user engagement [2]. The monolithic device creates 2.4. Imaging
a dynamic living embryo array which enabled us to: (i) trans-
port embryos, (ii) immobilize them for convenient imaging, (iii) Time-lapse imaging of developing embryos cultured on
continuously deliver reagents and drugs while under continuous chip-based devices was performed using the Leica MZ7.5 stereo-
real-time observation, (iv) support on-chip embryo development microscope equipped with a Leica DFC295 camera and running
and also (v) retrieve specimens of interest for further processing. under the LAS Multi-time software (Leica Microsystems, Germany).
It also includes integrated heating manifold to provide stable tem- A Nikon SMZ1500 fluorescent stereomicroscope equipped with a
perature for embryo development on chip. This proof-of-concept DS-U2/L2 camera and standard FITC/GFP filter cube was used to
technology provides a new rationale for rapid and automated acquire fluorescence images of developing Tg(fli1a:EGFP) embryos.
manipulation of developing zebrafish embryos at a large scale in An infrared camera (FLIR P-Series IR, FLIR Systems Inc., USA) was
drug discovery. Moreover, in contrast to our prior hydrodynamic used to obtain the thermal distribution inside the chip-based
traps, the current design enables a modification to provide inde- device. The IR camera was positioned at a distance of 0.5 m from
pendent actuation of every single trap. This can be used for rapid the surface of the microfluidic system to measure its temperature.
and selective un-docking of immobilized embryos by applying pos-
itive pressure though the suction manifold at the bottom plane of 2.5. Data and statistical analysis
the device. Recovered embryos are then collected from input or
output ports connected to the main channel. Such feature can be Data analysis and presentation was performed using
prospectively using in phenotype-based sorting of large numbers the LAS (Leica Microsystems); ImageJ (freely available at
of embryos. http://rsb.info.nih.gov/ij/web page) and GraphPad Prism (Graph-
Pad Software, Inc., CA, USA). The Student’s t-test was applied for
2. Experimental comparison between groups using MS Excel (Microsoft, USA) and
GraphPad Prism (GraphPad Software) with significance set at
2.1. Zebrafish husbandry and embryo culture p < 0.05. All control measurements are provided in detail in the
Figure legends, where appropriate, but in general involved making
Wild type (AB line; Zebrafish International Resource Center, direct comparisons between the chip-based devices and static
Oregon, Eugene, OR, USA) and transgenic Tg(fli1a:EGFP) adult 24-well microtiter plates or 60 mm Petri dishes (Nalgene Nunc
zebrafish were used [6,8,20]. Embryos obtained from natural Inc., NY, USA).
spawning were then collected in embryo medium E3, kept at
28.5 ± 0.5 ◦ C and developmentally staged as described earlier [8]. 3. Results
To inhibit angiogenesis, Tg(fli1a:EGFP) embryos at 16 hpf (hours
post fertilization) stages were loaded on a chip. Following dock- 3.1. Microfluidic system validation and performance
ing, embryos were perfused with the E3 solution supplemented
with anti-angiogenic drugs such as: VEGFR1-3 inhibitor AV951 The 3D (4-layer) design was fabricated in a biologically compat-
(Tivozanib); VEGFR2/PDGFR␤ inhibitor Sunitinib; broad range Bcl- ible and optically transparent PMMA polymer using infrared laser
2 inhibitor TW37; biphenolic Akt inhibitor Honokiol or DMSO micromachining. The chip consisted of four integrated modules: (i)
vehicle (Selleckchem, TX, USA). Animal research was conducted the main channel (1.7 mm × 1.5 mm × 55 mm) for embryo loading
with approval from The University of Auckland Animal Ethics Com- and medium perfusion, (ii) a linear array of up to 20 embryo traps,
mittee (approval ID R903). (iii) a suction manifold (0.7 mm in depth) that created a drag force
J. Akagi et al. / Sensors and Actuators B 189 (2013) 11–20 13

Fig. 1. Microfluidic living embryo array. (A) 3D chip-based device with four integrated modules: (i) the main channel for embryo loading and microperfusion, (ii) a linear array
of up to 20 embryo traps, (iii) a suction manifold, and (iv) a heating manifold circulating warm water around the array of traps. The chip based device can be actuated by applying
suction in three regimens: (i) at the main channel outlet; (ii) at the suction manifold outlet; (iii) at both main channel and suction manifold outlets providing fine control
of microperfusion conditions (for details refer to text). (B) A solid-state piezoelectric ultrasonic microdiaphragm pump with a footprint of only 15 mm × 30 mm × 3.8 mm
provides a pulseless flow due to the low displacement volumes of maximum 1 ␮l. (C) Cross-sectional view of the device across the embryo trapping array. Blue arrows denote
the direction of the fluid flow. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.)

to trap and immobilize embryos and (iv) a heating manifold that gravitational-induced sedimentation of embryos combined with
consisted of a U-shaped channel (1.5 mm × 3 mm × 100 mm) cir- the low-pressure suction at the bottom plane of the device to
culating warm medium (29 ◦ C) around the linear trapping array rapidly trap embryos (Fig. 2C and D). Namely, in the presence
(Fig. 1A–C). Water to the heating manifold was aspirated and of gravity, embryos are deflected under the combined effect of
recirculated in a closed-loop perfusion at a flow rate of up to suction flow and gravity towards the traps (Fig. 2C and D). Impor-
3 ml/min from an external water bath. The thin wall (0.2 mm) tantly, the size and shape of the traps were designed to assure: (i)
between the heating manifold and the trapping array heating man- single embryo occupancy, and (ii) unobstructed passage of other
ifold facilitated the rapid heat transfer directly to immobilized embryos in the rectangular main channel following docking. Sub-
embryos. The device was also interfaced with a miniaturized 6 sequent embryos introduced into the device rolled freely on top of
VDC stepper-motor peristaltic and piezoelectric ultrasonic micro- the immobilized embryos towards the next available trap (Fig. 2C
diaphragm pumps that realized adjustable flow rates between 0.01 and D; Supplementary Movie 1). The process was repeated until
and 4 ml/min (Fig. 1B). all traps were occupied. Embryo trapping experiments were per-
The LOC device was optimized specifically for trapping, culture formed by flowing in fixed number of embryos that corresponded
and analysis of embryos that remain within the oval chorion struc- to the number of traps. Trapping efficiency was then calculated as
ture (Fig. 2A and B). To achieve this, a linear array of traps was a number of captured embryos divided by the number of embryos
ablated using laser raster mode to the depth of 1.6 mm. Each trap injected into the LOC device. The system achieved 100% trapping
had a conical geometry with a diameter of 1.8 mm on the top plane efficiency when actuated at a total flow rate of up to 0.6 ml/min
and a diameter of 1.50 mm on the bottom plane. Subsequently, (Fig. 2E). The deterioration of the trapping efficiency at higher flow
channels with a diameter of 0.5 mm were laser drilled on the bot- rates was observed due to increased embryo velocities and their
tom plane of each well using a 50 ␮m laser cutting beam. These high momentum that could not be compensated by suction-based
channels are seen as small apertures at the bottom of microwells trapping (Fig. 2E). Over 99% ± 1 of trapped embryos retained their
and interconnect the trapping array with a suction manifold located position during the course of even long-term experiments (ca. 72 h)
underneath (Fig. 1A and B). The chip based device can be actuated with no dislodgement observed when suction manifold was actu-
by applying suction in three regimens: (i) at the main channel out- ated between 0.1 and 0.4 ml/min. Noteworthy, the design allows
let to support rapid device priming and cleaning; (ii) at the suction for a very straightforward recovery of immobilized embryos. They
manifold outlet providing low-pressure suction at the bottom plane can be un-docked by applying positive pressure through the suction
of the device to support embryo trapping and rapid drug perfusion; manifold at the bottom plane of the device. Recovered embryos are
(iii) at both main channel and suction manifold outlets providing then collected from input or output ports connected to the main
fine control of microperfusion conditions. channel.
The major obstacle against manipulation and arraying of To assess the hydrodynamic conditions we initially performed
millimeter-sized embryos in perfusion chip-based devices is linked a simulation of the pressure distribution around the embryo traps.
to their substantial mass (up to 1 mg for a single zebrafish At the flow rates of main/suction outlets of 0.2/0.0 ml/min, the
embryo), which leads to rapid gravitational-induced sedimentation pressure difference across the top and bottom surfaces of embryos
and high momentums of translational and rotational movements. remained almost constant ∼0.02 Pa for all embryos (Fig. 3A and B).
To overcome these issues, our innovative design exploited the In this case, the embryos experienced a slightly higher pressure
14 J. Akagi et al. / Sensors and Actuators B 189 (2013) 11–20

Fig. 2. Operational principles of the device. (A) 3D streamlines of flow through the last two traps of the grooved vacuum chip colored by velocity (m/s), obtained by CFD
simulations. The result is obtained when suction flow rates of 0.1 ml/min are obtained at both of main and suction outlets. (B) Cross sectional cut across the mid plane of
the device depicting the immobilized embryos inside the trapping array. (C) 3D cartoon showing the embryo trapping principles. (D) Experimental data showing embryo
immobilization sequence. Embryos are deflected towards the traps under the combined effect of suction flow and gravity. Embryos are trapped and immobilized one by one
in sequence. (E) Embryo trapping efficiency characterized by the number of embryos captured divided by the number of embryos injected into the LOC device. Trapping was
performed at varying flow rates in the main channel and suction manifold.Inset shows docked and actively immobilized zebrafish embryos (traps No. 6–11).
J. Akagi et al. / Sensors and Actuators B 189 (2013) 11–20 15

Fig. 3. Computer simulation of normal pressure exerted on embryos immobilized within traps. (A) Contours of pressure (Pa) over the surface of the last four embryos at
the flow rates of main/suction outlets = 0.2/0.0 ml/min. (B) Variations of pressure (Pa) across the trapped embryos at the flow rates of main/suction outlets = 0.2/0.0 ml/min.
(C) Contours of pressure (Pa) over the surface of the last four embryos at the flow rates of main/suction outlets = 0.1/0.1 ml/min. (D) Variations of pressure (Pa) across the
trapped embryos at the flow rates of main/suction outlets = 0.1/0.1 ml/min.

at their bottom surface (Fig. 3A). In comparison, at the flow rates (Fig. 4C and D). We found that dye freely entered all occupied traps
of main/suction outlets of 0.1/0.1 ml/min, the pressure difference and the complete dye exchange across the whole device occurred
across the top and bottom surfaces of embryos consistently within 150–180 s (Fig. 4C). This could be further accelerated to
increased along the main channel, increasing from 0.03 Pa for the below 30 s simply by increasing the flow rate at the suction out-
first embryo to 1.2 Pa for the last one (Fig. 3C and D). This time, the put. We also validated the dye delivery to each trapped embryo by
embryos experienced a higher pressure at their top surface (Fig. 3C). perfusing the chip with a solution of 0.04% Trypan Blue and subse-
Next, we validated the uniformity of mass transfer to each quently quantifying the intensity of the stained embryos (Fig. 4D)
embryo across the array (Fig. 4A–D). The design of 3D embryo traps after a washing step with E3 medium. We observed uniform label-
featured additional small grooves ablated at the periphery of the ing with occasional higher intensities due to the heterogeneous
microwells. The grooves were designed to facilitate the transverse sizes within the embryo population (Fig. 4D) on our chip-based
flow of the medium through the traps, especially when occupied device. This was confirmed by results obtained from staining of
by embryos (Fig. 2A and B). We first simulated the variations of zebrafish embryos under static (Petri dish) conditions. Namely they
dye concentration over the surface of embryos trapped across the have achieved similar Trypan Blue staining baseline as obtained on
chip at different times (Fig. 4A and B). The CFD simulations indi- a chip-based device.
cated that the embryos close to the inlet and outlet were quickly Our data showed also a strong correlation with the compu-
exposed to drug while the ones in the middle were exposed more tational models that guided the design of the device providing
slowly (Fig. 4A). Despite of this, the drug concentration over the further evidence that the 3D suction trap principle allows for
surface of embryos No. 1–20 reached an average value of 57.5% both robust immobilization of the fish embryos inside the traps
after 30 s, 82.1% after 60 s and 91.8% after 150 s when flow rates and also efficient medium exchange and uniform drug delivery
of 0.1 ml/min were applied at both of main channel and suction across the miniaturized array. The device has the ability to pulse
outlets (Fig. 4B). Accordingly, the microgrooves increased the mass the embryos with the investigational drug followed by the rapid
transfer of up to 40% as compared with the design that did not fea- medium exchange without disturbing the embryo position. This
ture additional grooves (data not shown). The variations of Péclet cannot be achieved on a static Petri dish and is of particular impor-
number (indicating the ratio of convective to diffusive transport) tance to precisely control elements such as the transcriptional
across different embryos justified the slow exposure of the middle activation of target genes using inducible transgenes.
embryos. The Péclet number was less than 5 around the middle
embryos, indicating the diffusive mass transport. In comparison, 3.2. Embryo development inside the microfluidic environment
the Péclet number was more than 5 around the embryos close to
the inlet and outlet, indicating the convective mass transport that Next, we set to validate the compatibility of the chip microenvi-
occurred fast (Fig. 4A). When the bottom plane suction manifold ronment for the zebrafish embryo culture over extended periods of
is offline the drug penetration towards and inside the traps will time. For this purpose, we first performed numerical simulations
slowly occur because of diffusion from the main channel. to estimate both the flow velocity and the extent of shear stress
We next validated the CFD results by perfusing the chip with exerted over embryos docked in 3D array. The embryos were sim-
a solution of 0.04% Trypan Blue and subsequently quantifying the ulated as rigid and not deformable spheres (Fig. 5A and B). Flow
intensity of stained embryos after a washing step with E3 medium rates of main/suction outlets of 0.2/0.0 ml/min led to a maximum
16 J. Akagi et al. / Sensors and Actuators B 189 (2013) 11–20

Fig. 4. Drug delivery inside the microfluidic embryo array. (A) CFD simulation showing the variations of Péclet number across the array. The Péclet number is less than 5
around the middle embryos, indicating the diffusive mass transport while more than 5 around the embryos close to the inlet and outlet, indicating the convective mass
transport. Inset represents the magnified graph data for variations of Péclet number across the traps 1–10. (B) Variations of drug concentration over the surface of embryos
obtained by CFD simulations. The drug concentration over the embryos No. 17–20 reaches an average value of 57.5% after 30 s (left panel) and 91.8% after 150 s (right panel).
The results are obtained by applying flow rates of 0.1 ml/min at both of main and suction outlets. (C) Validation of the uniformity of dye delivery to each embryo across the
microfluidic array. Chip was perfused with a 0.04% Trypan Blue dye and then the dye was washed away with E3 medium. Only 1–17 embryos are shown due to imaging camera
optics limitations. The results from traps 18–20 did not exhibit any differences from the trend. (D) Intensity of dye across each embryo obtained by image analysis of the
experiments after 150 s of E3 wash as depicted in (C). Red line denotes the average staining intensity. Results obtained from staining of zebrafish embryos under static (Petri
dish) conditions achieved similar Trypan Blue staining baseline as obtained on a chip-based device. White arrows depict the direction of fluid flow and embryo movement.
Asterisks denote traps that are first to completely exchange the drug due to a higher flow rate passing through their apertures. (For interpretation of the references to color
in this figure legend, the reader is referred to the web version of the article.)

shear stress of 0.01 Pa at the upper surface of trapped embryos, reached all developmental staging criteria such as tail detachment,
which remained uniform across all embryos (Fig. 5A). In compar- tail and fin morphology, accumulation of melanocytes, eye and
ison, flow rates of main/suction outlets of 0.1/0.1 ml/min led to a lens formation and formation of intersegmental blood vessels (ISV)
maximum shear stress of 0.01 Pa at the upper surface of the first together with cardiovascular function (heart rate and blood flow)
four embryos while a maximum shear stress of 0.07 Pa at the lower that were statistically comparable with static Petri dish control
surface of the last four embryos (data not shown). Finally, flow rates experiments (Table 1).
of main/suction outlets of 0.0/0.2 ml/min led to a maximum shear
stress of 0.01 Pa at the upper surface of the first four embryos while 3.3. On-chip angiogenesis assay
a maximum shear stress of 0.14 Pa at the lower surface of the last
four embryos (Fig. 5B). The sharp increase in shear stress values in Zebrafish transgenic lines have recently been reported as
the last embryos is consistent with the abrupt increase of the trans- convenient surrogate bioassays to perform primary screens of
verse flow of medium through the traps No. 14–20 (Fig. 4C; data investigational anti-angiogenic compounds that diffuse into the
not shown). embryo and induce dose dependent inhibition of ISV formation.
We experimentally validated the influence of both the sheer We set to explore the applicability of the microfluidic embryo
stress conditions and polymeric chip structure on zebrafish embryo array for the analysis of a panel of small-molecule compounds
development by performing a 72-h culture of developing wild- with anti-angiogenic properties using the transgenic zebrafish
type zebrafish embryos at varying flow rates (Fig. 5C and D). We line Tg(fli1a:EGFP). This particular strain can be used to non-
observed the normal and very uniform development of all embryos invasively observe the development of blood vessels expressing
immobilized across the array of all traps when perfused at a total enhanced green fluorescent protein (EGFP) in all endothelial cells.
flow rate ranging from 0.1 to 1 ml/min (Fig. 5C, Table 1). The cumu- The Tg(fli1a:EGFP) embryos were loaded onto a chip at 16 hpf
lative survival of embryos cultured on a chip-based device for up stage before sprouting of intersegmental and head vessels had
to 72 h was over 95 ± 5% with the exception of chips kept at a static begun. The embryos were then continuously perfused in a closed-
regimen or perfused at flow rates lower than 0.1 ml/min (Fig. 5A). loop perfusion at the flow rate of 0.2 ml/min with E3 media
The decreased survival at ultra-low flow rates was directly con- containing VEGFR1-3 inhibitor AV951 (Tivozanib; 0.1–1 ␮M);
nected to depletion of oxygen inside the chip when insufficient VEGFR2/PDGFR␤ inhibitor Sunitinib (1–100 ␮M); broad range Bcl-
exchange of medium inside the gas non-permeable PMMA device 2 inhibitor TW37 (1–100 ␮M); biphenolic Akt inhibitor Honokiol
was present (Fig. 5C and D). Importantly, the embryo development (1000 ␮M). Images of developing embryos immobilized on chip-
and viability was uniform across the traps 1–20 irrespectively of the based devices were acquired at 0, 24 and 48 h intervals that
calculated shear stress condition discussed above (Fig. 5D). Further- corresponded to 16, 40 and 64 hpf developmental stage (Fig. 6A).
more, analysis of developmental and teratogenic endpoints during Our data show that the non-stimulated (control) Tg(fli1a:EGFP)
72 h of zebrafish embryo culture on microfluidic chip-based device embryos developed normal vasculature during the course of the
did not result in any discernible phenotypic effects irrespectively experiments as evidenced by the presence of characteristic pat-
of the magnitude of flow rates (Table 1). The developing embryos terns of ISVs (Fig. 6A). Moreover, the embryos grown on a chip
J. Akagi et al. / Sensors and Actuators B 189 (2013) 11–20 17

Fig. 5. Embryo development inside the microfluidic environment. (A) Contours of shear stress over the surface of embryos at the flow rates of main/suction out-
lets = 0.2/0.0 ml/min. (B) Contours of shear stress over the surface of embryos at the flow rates of main/suction outlets = 0.0/0.2 ml/min. (C) Cumulative survival of developing
zebrafish embryos immobilized on a chip. Embryos were perfused at varying volumetric flow rates. Control denotes static 60 mm Petri Dish experiments. The deaths at ultra-
low flow rates relate to depletion of oxygen inside the non-gas permeable PMMA device. Inset depicts the time-lapse imaging of developing zebrafish embryo immobilized
on the array. (D) Survival of developing zebrafish embryos under shear stress condition depicted in (A) and (B). Control (green), 0.2/00 ml/min (blue) and 0.0/0.2 ml/min
(black). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.)

in the continuous presence of anti-angiogenic drugs provided a we have for the first time evaluated both natural biphenolic Akt
real-time bioassay principle in an intact vertebrate organism. The inhibitor Honokiol and broad range Bcl-2 inhibitor TW37 in the
VEGFR1-3 inhibitor Tivozanib proved to be the most potent drug Tg(Fli1a:EGFP) angiogenesis model. The latter compound achieved
achieving 100% of ISV growth inhibition at 1 ␮M concentration. The nearly 35% inhibition of ISVs when used at a concentration of
VEGFR2/PDGFR␤ inhibitor Sunitinib achieved similar results but at 100 ␮M. This new data supports the notion of the role of Bcl-2
only at 100× higher concentration (Fig. 6A and B). Interestingly, family of proteins in the angiogenesis as postulated recently by
18 J. Akagi et al. / Sensors and Actuators B 189 (2013) 11–20

Table 1
Analysis of lethal and sublethal developmental endpoints during 72 h of zebrafish embryo culture on microfluidic chip-based device.

Morphological scoring Flow rate (ml/min)

Control Main/suction 0.2/0.0 Main/suction 0.0/0.2

Body shape N N N
Notochord morphology N N N
Tail detachment + + +
Tail morphology N N N
Fin morphology N N N
Pigmentation N N N
Eye formation N N N
Lens formation N N N
Intersegmental blood vessel (ISV) formation N N N
Cardiovascular function (heart rate and blood flow) N N N
Body twisting + + +

N: normal; A: abnormal (% denote number of abnormal embryos); (+) present.

Fig. 6. On-chip real-time angiogenesis assay. (A) On-chip angiogenesis assay using transgenic the zebrafish line Tg(fli1a:EGFP). Transgenic embryos were arrayed and
immobilized at 16 hpf and continuously perfused with E3 media containing vehicle control (DMSO) or selected small molecule drugs (Honokiol, Sunitinib, Tivozanib and
TW37). The spatial embryo encoding allows for automated microscopic visualization of patterns of intersegmental vessels (ISV). White arrows and red arrows denote partial
and complete ISV growth inhibition, respectively. (B) Quantification of average ISVs growth inhibition as shown in (A). The growth inhibition percentages were calculated
from fluorescence area values. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.)

some authors (Fig. 6A and B) [21–23]. Also the prospective applica- with no moving parts. This design greatly facilitates: (i) staining
tion of BH3 mimetics such as TW37 as anti-angiogenic compounds or treatment without displacing the embryos; (ii) analysis of indi-
warrants further exploration [21,23]. vidual developing embryos under continuous perfusion; (iii) devel-
opment of customized image and data analysis software, allowing
4. Discussion address designation to each embryo; and (iv) spatial segregation of
developing embryos to avoid embryo-to-embryo interaction. In the
Zebrafish has recently emerged as a small model organism par- latter case, embryo-to-embryo interaction can be eliminated when
ticularly suitable for drug discovery and pathology. Automated device is operated under open-loop perfusion and only suction
and high-throughput fish embryo assays are, however, still largely manifold is actuated. As a result the adjacent embryos will never
unavailable [1,2,11]. Recent noteworthy reports have shown that be in contact with compounds/enzymes released, e.g. from dying
fish embryos can develop in a confined microfluidic environ- embryos. This is a very important consideration in drug discov-
ment and that manual LOC devices hold a substantial promise for ery and drug screening routines that can help eliminate bystander
miniaturized toxicological analysis [14,15]. The highly integrated, effects present in bulk embryo culture under static conditions.
automated and reproducible loading, positioning, long-term Our data support the notion that due to the nature of the flow
immobilization of large number of single zebrafish embryos inside inside the 3D trapping system, the embryos will be kept within a
a precisely controllable fluidic microenvironment is, however, still low shear stress micro-environment. In this regard, we presented
unrealized. In contrast to any previously described technologies, both CFD simulations and embryo developmental analysis under
our current work introduces a user-friendly and highly repro- different flow regimens. We have recently reported on signaling
ducible system for automatic manipulation of single embryos events associated with single cells under a range of flow induced
inside unique 3D multilayer chip-based devices. The microfluidic mechanical loads [24,25]. Our current data show that the embryos
system was designed for automatic loading, perfusion, analysis and docked in 3D traps are exposed to a shear stress at least two
recovery of specimens all in a single, monolithic integrated device orders of magnitude lower than values reported to trigger signaling
J. Akagi et al. / Sensors and Actuators B 189 (2013) 11–20 19

cascade in single cells [24,25]. In stark contrast to cells, however, [8] C.M. Buchanan, J.H. Shih, J.W. Astin, G.W. Rewcastle, J.U. Flanagan, P.S. Crosier,
the fish embryos are protected by an elastic chorion membrane and P.R. Shepherd, DMXAA (Vadimezan, ASA404) is a multi-kinase inhibitor target-
ing VEGFR2 in particular, Clinical Science 122 (2012) 449–457.
therefore the actual shear stress effects on the developing tissues [9] C. Chakraborty, C.H. Hsu, Z.H. Wen, C.S. Lin, G. Agoramoorthy, Zebrafish: a com-
are likely negligible. plete animal model for in vivo drug discovery and development, Current Drug
This proof-of-concept 3D LOC technology allowed us for the Metabolism 10 (2009) 116–124.
[10] T.P. Barros, W.K. Alderton, H.M. Reynolds, A.G. Roach, S. Berghmans, Zebrafish:
first time to rapidly investigate a panel of small-molecule drugs. an emerging technology for in vivo pharmacological assessment to identify
The embryo immobilization allowed for developmental analysis potential safety liabilities in early drug discovery, British Journal of Pharma-
without the need for re-focusing and specimen re-positioning. cology 154 (2008) 1400–1413.
[11] A. Letamendia, C. Quevedo, I. Ibarbia, J.M. Virto, O. Holgado, M. Diez, J.C. Izpisua
Address designation to each embryo during analysis substan-
Belmonte, C. Callol-Massot, Development and validation of an automated high-
tially accelerated the data acquisition in contrast to conventional throughput system for zebrafish in vivo screenings, PLoS ONE 7 (2012) e36690.
Petri dish assays. This was particularly advantageous as immobi- [12] E. Lammer, H.G. Kamp, V. Hisgen, M. Koch, D. Reinhard, E.R. Salinas, K. Wendler,
S. Zok, T. Braunbeck, Development of a flow-through system for the fish embryo
lized embryos were pulsed with a 0.2 mg/ml of tricaine mesylate
toxicity test (FET) with the zebrafish (Danio rerio), Toxicology In Vitro: an inter-
buffered solution before image acquisition to provide temporary national journal published in association with BIBRA 23 (2009) 1436–1442.
anesthesia and inhibit the intrinsic embryo movements during [13] S. Weigt, N. Huebler, R. Strecker, T. Braunbeck, T.H. Broschard, Zebrafish (Danio
the fluorescent imaging. The drug exchange and delivery were rerio) embryos as a model for testing proteratogens, Toxicology 281 (2011)
25–36.
performed automatically within seconds without embryo dislodge- [14] T.Y. Chang, C. Pardo-Martin, A. Allalou, C. Wahlby, M.F. Yanik, Fully auto-
ment or need for repetitive pipetting. mated cellular-resolution vertebrate screening platform with parallel animal
processing, Lab on a Chip 12 (2012) 711–716.
[15] D. Choudhury, D. van Noort, C. Iliescu, B. Zheng, K.L. Poon, S. Korzh, V. Korzh, H.
5. Conclusions Yu, Fish and chips: a microfluidic perfusion platform for monitoring zebrafish
development, Lab on a Chip 12 (2012) 892–900.
[16] E.M. Wielhouwer, S. Ali, A. Al-Afandi, M.T. Blom, M.B. Riekerink, C. Poelma,
We envisage that microfluidic LOC systems where most tasks J. Westerweel, J. Oonk, E.X. Vrouwe, W. Buesink, H.G. vanMil, J. Chicken, R.
are performed automatically without disturbing the embryo, and van’t Oever, M.K. Richardson, Zebrafish embryo development in a microfluidic
without sudden changes to the embryo environment, will demon- flow-through system, Lab on a Chip 11 (2011) 1815–1824.
[17] C. Pardo-Martin, T.Y. Chang, B.K. Koo, C.L. Gilleland, S.C. Wasserman, M.F.
strate a better performance and be capable of generating more Yanik, High-throughput in vivo vertebrate screening, Nature Methods 7 (2010)
reproducible data than conventional static and manual zebrafish 634–636.
embryo bioassays. Our work provides a new rationale for rapid and [18] C.B. Rohde, F. Zeng, R. Gonzalez-Rubio, M. Angel, M.F. Yanik, Microfluidic system
for on-chip high-throughput whole-animal sorting and screening at subcellu-
automated manipulation of developing zebrafish embryos inside lar resolution, Proceedings of the National Academy of Sciences of the United
the new class of miniaturized devices. States of America 104 (2007) 13891–13895.
[19] A. Funfak, A. Brosing, M. Brand, J.M. Kohler, Micro fluid segment technique for
screening and development studies on Danio rerio embryos, Lab on a Chip 7
Acknowledgements (2007) 1132–1138.
[20] G.N. Serbedzija, E. Flynn, C.E. Willett, Zebrafish angiogenesis: a new model for
drug screening, Angiogenesis 3 (1999) 353–359.
Supported by the Faculty Research Development Fund (FRDF),
[21] B.D. Zeitlin, E. Joo, Z. Dong, K. Warner, G. Wang, Z. Nikolovska-Coleska, S. Wang,
University of Auckland (JA, DW); Early Career Research Excellence J.E. Nor, Antiangiogenic effect of TW37, a small-molecule inhibitor of Bcl-2,
Award 2012 (ECREA 2012), University of Auckland (DW); Ministry Cancer Research 66 (2006) 8698–8706.
of Science & Innovation New Zealand (JA, CH, KC, PC, DW); The [22] E. Karl, K. Warner, B. Zeitlin, T. Kaneko, L. Wurtzel, T. Jin, J. Chang, S. Wang, C.Y.
Wang, R.M. Strieter, G. Nunez, P.J. Polverini, J.E. Nor, Bcl-2 acts in a proangio-
Australia Endeavour Awards, Department of Education, Employ- genic signaling pathway through nuclear factor-kappaB and CXC chemokines,
ment and Workplace Relations, Australia (KK); The Australian Cancer Research 65 (2005) 5063–5069.
Research Council’s Discovery Early Career Researcher Award fund- [23] N. Ashimori, B.D. Zeitlin, Z. Zhang, K. Warner, I.M. Turkienicz, A.C. Spalding, T.N.
Teknos, S. Wang, J.E. Nor, TW-37 a small-molecule inhibitor of Bcl-2, mediates
ing scheme (ARC DE120101402) (KK); Biotechnology and Biological S-phase cell cycle arrest and suppresses head and neck tumor angiogenesis,
Sciences Research Council (BBSRC), Engineering and Physical Sci- Molecular Cancer Therapeutics 8 (2009) 893–903.
ences Research Council (EPSRC) and Scottish Funding Council UK, [24] H. Yin, X. Zhang, N. Pattrick, N. Klauke, H.C. Cordingley, S.J. Haswell, J.M.
Cooper, Influence of hydrodynamic conditions on quantitative cellular assays
funded under the RASOR (Radical Solutions for Researching the in microfluidic systems, Analytical Chemistry 79 (2007) 7139–7144.
proteome) Program (JMC, DW). [25] D. Wlodkowic, S. Faley, M. Zagnoni, J.P. Wikswo, J.M. Cooper, Microfluidic
single-cell array cytometry for the analysis of tumor apoptosis, Analytical
Chemistry 81 (2009) 5517–5523.
Appendix A. Supplementary data
Biographies
Supplementary data associated with this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.snb.2012.11.036.
Jin Akagi received his BSc Hon in Chemistry from the University of Auckland, Auck-
land, New Zealand in 2009. Currently he is a senior PhD Student at the BioMEMS
References Research Group, University of Auckland. His research interests concentrate on devel-
opment of highly automated Lab-on-a-Chip technologies for manipulation and
[1] D. Wlodkowic, K. Khoshmanesh, J. Akagi, D.E. Williams, J.M. Cooper, analysis of small model organisms and also applications of chip-based microflow
Wormometry-on-a-chip: innovative technologies for in situ analysis of small cytometry for anti-cancer drug screening.
multicellular organisms, Cytometry A 79 (2011) 799–813.
Khashayar Khoshmanesh received a PhD degree in Biomechanical Engineering
[2] K. Khoshmanesh, J. Akagi, C.J. Hall, K.E. Crosier, P.S. Crosier, J.M. Cooper, D.
from Deakin University, Australia in 2010 with a focus on microfluidics and dielec-
Wlodkowic, New rationale for large metazoan embryo manipulations on chip-
trophoresis. He was awarded the 2010 Endeavour Fellowship to conduct research
based devices, Biomicrofluidics 6 (2012) 24102–2410214.
at the BioMEMS Research Group, University of Auckland, New Zealand and also
[3] G.N. Wheeler, A.W. Brandli, Simple vertebrate models for chemical genetics and
the 2012 American-Australian Association Fellowship to conduct research at Stan-
drug discovery screens: lessons from zebrafish and Xenopus, Developmental
ford Microfluidics Laboratory, Stanford University, USA. As the recipient of the
Dynamics 238 (2009) 1287–1308.
2012–2015 Discovery Early Career Researcher Award by the Australian Research
[4] L.I. Zon, R.T. Peterson, In vivo drug discovery in the zebrafish, Nature Reviews
Council, he is a Research Fellow at RMIT University, Australia. His research inter-
Drug Discovery 4 (2005) 35–44.
ests include manipulation of bio-particles in microfluidics under mechanical and
[5] K.R. Kidd, B.M. Weinstein, Fishing for novel angiogenic therapies, British Journal
electrical forces.
of Pharmacology 140 (2003) 585–594.
[6] N.D. Lawson, B.M. Weinstein, In vivo imaging of embryonic vascular develop- Chris J. Hall received his PhD in Molecular Medicine from The University of Auckland
ment using transgenic zebrafish, Developmental Biology 248 (2002) 307–318. in 2003. He then completed a short postdoctoral position at Max Planck Institute for
[7] K.S. Okuda, J.W. Astin, J.P. Misa, M.V. Flores, K.E. Crosier, P.S. Crosier, lyve1 Molecular Genetics, Berlin, Germany, working with the German Gene Trap Consor-
expression reveals novel lymphatic vessels and new mechanisms for lymphatic tium. Since 2004 he has worked (currently as a Senior Research Fellow) at the School
vessel development in zebrafish, Development 139 (2012) 2381–2391. of Medical Sciences at the University of Auckland in the Developmental and Cancer
20 J. Akagi et al. / Sensors and Actuators B 189 (2013) 11–20

Genetics group headed by Professors Phil and Kathy Crosier. His research interests lie Philip S. Crosier is a Professor of Molecular Medicine at the School of Medical Sci-
in exploiting the experimental tractability of the zebrafish model system to further ences, The University of Auckland. He graduated MSc (Hons) from The University
understand the immune system and as a drug discovery platform. of Auckland and obtained his PhD from the University of Otago. He has worked at
the Institut de Récherches sur les Mâladies du Sang, Paris where he was the Asso-
Jonathan M. Cooper is a Professor of Bioelectronics and Bioengineering and a head ciation Claude Bernard Fellow and as a visiting scientist at Genetics Institute, USA.
of the Division of Biomedical Engineering at the University of Glasgow. In 2004 His major research interest is in the regulation of cell growth and differentiation,
he was elected as a Fellow of the Royal Academy of Engineering and in 2001 a and lineage commitment. He applies the transgenic zebrafish model systems to
Fellow of the Royal Society of Edinburgh. He was appointed to the Wolfson Chair investigate the regulation of lymphangiogenesis and the development of the innate
in Biomedical Engineering in 2009 and awarded a Royal Society Merit Award in immune systems.
2010. His research interests encompass the advantages of miniaturization in the
biomedical sciences, and have an internationally renowned activity in biosensors, Donald Wlodkowic received his PhD in Biomedicine from the University of Kuopio,
lab-on-a-chip, nanophotonics and acoustics. Kuopio, Finland in 2007. He subsequently completed post-doctoral training at the
University of Glasgow and Glasgow Caledonian University (UK). Since 2010 he is the
Kathryn E. Crosier received her medical degree (MBChB) from the University of Director of the BioMEMS Research Group at the School of Chemical Sciences, Univer-
Otago in 1978 and PhD from The University of Auckland, New Zealand in 1989. She sity of Auckland. He is also a founder of the iRobotiX Group and an Adjunct Associate
holds Fellowships of the Royal Australasian College of Physicians (FRACP) and the Professor at the RMIT University, Melbourne, Australia. His current research inter-
Royal College of Pathologists of Australasia (FRCPA). From 1989 to 1991 she under- ests include the interface between robotics, biomedicine and BioMEMS. He has
took a Fellowship at the Children’s Hospital and Harvard Medical School, Boston. pioneered many microfluidic technologies for miniaturized bio-analysis of living
Kathryn is a Professor of Molecular Medicine at The University of Auckland. Her cells, embryos and small model organisms.
research is focused on haematopoietic stem cell biology and the genetics of myeloid
malignancies. In both of these areas she utilizes the zebrafish model system.

You might also like