Visualizing voltage dynamics in zebrafish heart

Abstract

The zebrafish heart provides a useful vertebrate cardiovascular model with outstanding advantages, including genetic manipulatability, optical accessibility and rapid development. In addition, an emerging topic in cardiotoxicity assay and drug discovery is its use in phenotype-based chemical screening. Here, we report a technique that allows non-invasive voltage mapping in beating heart using a genetically encoded probe for transmembrane potential. Application of the anti-allergy compound astemizole resulted in aberrant propagation of excitation, which accounted for a lack of ventricular contraction. This optical method will provide new opportunities in broad areas of physiological, developmental and pharmacological cardiovascular research.

Introduction

Many key findings of molecular and cellular mechanisms underlying cardiovascular functions have been unearthed through research on the zebrafish (Danio rerio), which possesses outstanding advantages as a vertebrate model including genetic manipulatability, optical accessibility and rapid development. An additional advantage of the zebrafish heart is its suitability in high-throughput phenotype based chemical screenings. Small fish embryos are amenable to large-scale micro-well assays to identify bioactive molecules that perturb phenotypic traits in heart (Burns et al. 2005).

Propagation of electrical excitation over myocardial cells is the primary physiological process in heart contraction. Electrical activities in zebrafish heart have been successfully probed using microelectrodes (Milan et al. 2006). This approach, however, does not give spatial information of the activities in detail. Moreover, such a measurement requires skillful manipulations, thus limiting its use in large-scale high-throughput assays. Accordingly, function of zebrafish heart has been mostly assessed in terms of indirect aspects such as morphology, contraction rate, Ca²⁺ elevation and blood flow.

We have been interested in optical probing of membrane voltage in excitable cells, tissues and organs. We recently developed a genetically encoded fluorescent probe for transmembrane potential, named ‘mermaid’ (Tsutsui et al. 2008). Change in voltage elicits conformational change of the voltage-sensing domain (S1–S4) derived from tunicate voltage-sensitive phosphatase, which alters the efficiency of fluorescence resonance energy transfer (FRET) between green- and orange-emitting fluorescence proteins: mUKG and mKOκ, respectively. The ratiometric readout of mermaid enables measurement in motile samples. By taking advantage of mermaid as well as the genetic manipulatability and optical accessibility in zebrafish, we attempted to establish a convenient, non-invasive method that allows in vivo imaging of voltage dynamics in a whole heart.

Methods

Fish

A transgenic zebrafish line expressing the voltage probe mermaid (Tsutsui et al. 2008) specifically in the myocardial cells was generated as follows. The promoter for zebrafish cardiac myosin light chain 2 (cmlc2) gene was amplified by polymerase chain reactions based on the published sequence (Huang et al. 2003). The cmlc2 promoter, mermaid and the SV40 poly-adenylation signal were placed in this order in pT2KXIGΔin, a vector carrying the Tol2 transposable element (Urasaki et al. 2006). Generation of transgenic fish with the Tol2-based method was carried out as described previously (Urasaki et al. 2006). Zebrafish adults, embryos and larvae were maintained at 28°C. All the procedures were performed in compliance with the policies and regulations of The Journal of Physiology as described by Drummond (2009) as well as the guidelines approved by the animal care and use committees of the Osaka University, RIKEN, and National Institutes of Natural Sciences.

Imaging

An epifluorescence inverted microscope (IX71, Olympus) with a power-stabilized 75 W xenon lamp (Ushio) was used. In order to minimize loss of fluorescence during detection, and to perform rapid and simultaneous acquisitions of FRET donor and acceptor images, two CCD cameras (CoolSNAP-HQ, Photometrics) were operated in parallel. Fluorescence was collected with a water immersion objective (×20 Olympus, NA 0.50) and split with a dichroic mirror with high surface accuracy (DM545, Olympus) into the donor and acceptor signals. The signals were bandpass filtered via BP475-540 and BP565-635 for donor and acceptor signals, respectively, and projected to the two CCDs using individual tube lenses. The CCD was exposed for ∼22–24 ms for each frame. This configuration allowed simultaneous acquisitions of donor and acceptor primary images at ∼40–43 frames per second. An embryo at 70–80 h post fertilization (hpf) was placed in a tiny well made with agarose gel (1.0%) on a glass-bottomed dish, and was viewed ventrally. Data acquisition and analysis were performed with MetaMorph (Molecular Devices, Sunnyvale, CA, USA) and IDL (Research Systems, Boulder, CO, USA), respectively.

Results

Voltage mapping in beating heart

The fluorescence pattern in the cmlc2::mermaid transgenic line (Fig. 1A,B) was consistent with that seen in the cmlc2::GFP transgenic line described previously (Huang et al. 2003); robust fluorescence was visible in the heart tubes after ∼20 h hpf. Anaesthesia (0.01% tricaine methanesulfonate) and/or gel-embedding prior to imaging were not always necessary because embryos at 70–80 hpf move intermittently and, in many experiments, kept still for several seconds within the field of view of the microscope objective (Supplementary video 1). In this way, membrane voltage dynamics under physiological conditions was clearly visualized as the fluorescence emission ratio (mKOκ/mUKG), which increases upon depolarization in a single scan, non-time-averaged dataset. The excitation propagated from the junction between the atrium and the blood vessel to the ventricle periodically (Fig. 1C; Supplementary video 2).

Figure 1. Voltage dynamics of a beating heart in zebrafish.

A, transgenic zebrafish (cmlc2::mermaid). An unanaesthetized, unrestrained fish was imaged ventrally. A donor fluorescence image (mUKG channel; in green) was imposed with transmission image (in grey). A low magnification objective (×4, Olympus) was used here. B, a representive donor channel fluorescence image. a: atrium; v: ventricle; bv: blood vessel. C and D, pseudocoloured ratio (mKOκ/mUKG) images representing a single cycle of heart contraction in the cmlc2::mermaid (C) and mermaid-null (D) transgenic fish. Propagation of excitation from the junction of vessel and atrium to the ventricle was highlighted with arrows in C. E and F, plot of total fluorescence intensity from a whole beating heart versus time in the mermaid-null (E) and mermaid (F) transgenic fish. Green = mUKG (donor) channel. Red = mKOκ (acceptor) channel. Black = ratio (mKOκ/mUKG). Bar = 500 μm (A); 100 μm (B–D).

It is generally difficult to completely preclude motion artifacts even with ratiometry, because factors such as contamination of the sample's autofluorescence and imperfect linearity in fluorescence detection should result in motion-dependent apparent ratio changes. We therefore needed to see to what extent the ratio signals in our measurements were contributed by factors other than the membrane voltage. During the development of mermaid, we have found that voltage-dependent FRET change, the nature of mermaid, is completely removed by D129R mutation at least for the physiological range of membrane potential (Supplementary Fig. 1). Thus, the D129R mutant can be used as a negative control probe, which we named ‘mermaid-null’. We generated a cmlc2::mermaid-null transgenic line and performed imaging in the same way. We did not observed significant ratio changes in the beating hearts of the mermaid-null fish (Fig. 1D), which verified that the ratio signal in the mermaid fish mostly reported voltage dynamics. To statistically compare signal of mermaid-null and mermaid, we analysed total fluorescence from a whole beating heart. Intensity signals of mUKG and mKOκ channels of mermaid-null showed positively correlated fluctuations with amplitude of a few per cent (Fig. 1E), which resulted from focusing and defocusing of the heart accompanied by its beat-motions. That fluctuation cancelled well in ratio, and the root mean square (RMS) amplitude of the ratio fluctuation was 0.124 ± 0.034% (mean ±s.e.m.; n= 6). By contrast, the donor- and acceptor-signals of mermaid contained both common and reciprocal component, which resulted from motion induced focusing/defocusing and voltage-dependent FRET, respectively (Fig. 1F). The RMS amplitude of the mermaid's ratio signal reached 2.34 ± 0.18% (n= 6). It was thus estimated that the fraction of the voltage-independent component in mermaid's ratio signal was as small as one-twentieth.

Electrical basis for compound induced cardiac dysfunction

The ether-a-go-go-related gene (ERG) channel is responsible for the cardiac delayed rectifier potassium current. Inhibition of the human ERG (hERG) channel is often related to drug-induced cardiotoxic effects, such as QT prolongation or torsades de pointes. It should be noted that the zebrafish ERG channel exhibits remarkable homology with the hERG channel: within the pore domain, the sequence similarity between them reaches 99%. It has been known that the zebrafish heart responds to chemicals that exhibit cardiotoxicity in humans, including drugs that have been withdrawn from the market due to potentially lethal side effects (Milan et al. 2003; Langheinrich et al. 2003; Taglialatela et al. 1998), even though direct evidence on whether the same mechanisms underlie the dysfunctions in zebrafish and human have been scarce. We tested one such drug, astemizole, which was originally developed as a second-generation histamine H1 receptor blocker but afterward was found to inhibit hERG (Taglialatela et al. 1998). Astemizole was bath applied to zebrafish. In consistent with the previous report (Langheinrich et al. 2003), we observed substantial cardiac abnormalities, including decreased heart rate, aberrant atrial:ventricular contraction ratio, and complete absence of ventricular contraction, depending on the dose and time after application (data not shown). Complete blockade of ventricular contraction was elicited by the application of 5 μm astemizole for 15 min (Supplementary video 3). In these fish, circulation was severely impaired and blood cells accumulated in the chambers. We then performed voltage imaging to reveal an electrical basis for this dysfunction. Before the application, we observed normal propagation of membrane excitation from the vessel–atrium junction to the ventricle (Fig. 2A, Supplementary video 4). In the astemizole-treated hearts that lacked ventricular contractions, we were surprised to find that electrical excitation first occurred at near the atrium–ventricle (A-V) boundary and then propagated backward into the atrium (Fig. 2B, Supplementary video 4). In the normally beating hearts, the ratio peak at the A-V boundary and at the ventricle followed that at the atrium after 57 ± 5.8 (mean ±s.e.m.; n= 5) and 85 ± 6.2 ms (n= 5) (Fig. 2C), while in the the astemizole-treated hearts without ventricular contractions, they preceded it by 52 ± 4.9 (n= 5) and 95 ± 4.3 ms (n= 5), respectively (Fig. 2D). Our voltage imaging technique in the zebrafish heart thus revealed an altered mode of excitation propagation that reflects drug-induced cardiac dysfunction.

Figure 2. An electrical basis for the astemizole induced cardiac dysfunction.

A and B, pseudocoloured ratio (mKOκ/mUKG) images before (A) and 15 min after (B) application of astemizole (5 μm in bath). Fish were gel-embedded during imaging to ensure similar angle of view between the two datasets. Excitation propagating backwardly from the ventricle to the atrium was highlighted with arrows in B. Bar = 100 μm. See Supplementary video 4 for full movies. C and D, a representive plot of the mermaid's ratio at the atrium (green), ventricle (blue) and A-V boundary (red) in the normally beating heart (C) and in the astemiozole treated heart (D). Square regions of interest were manually registered in each frame as shown in the example images. The arrows indicate relative time of ratio peak at the ventricle (blue arrow) and the A-V boundary (red arrow) with respect to the peak at the atrium. Averaged time for three cycles of contractions was used as data for one heart.

Discussion

This is the first report that successfully used a genuine protein-based, genetically encoded voltage probe to visualize voltage dynamics of excitable tissue in vivo. While several groups have developed such genetically encoded voltage probes for more than a decade (see Baker et al. 2008 for review), it has been difficult to probe spatiotemporal propagation of membrane voltage in vivo due to its low signal-to-noise ratios. This study thus validates in vivo performance of the recently developed voltage probe, mermaid (Tsutsui et al. 2008).

For teleost heart, it has been suggested that the primary pacemaking site is located at the venous–atrial junction (Haverinen & Vornanen, 2007). The present result was in agreement with this view. We observed depolarization which first arises near this junction in the atrial diastole, and then propagates through the atrium to the ventricle accompanied by coordinated contractions. Similar propagation of excitation has been reported in isolated hearts of adult zebrafish using organic voltage-sensitive dye imaging (Sedmera et al. 2003). In general, spontaneous rhythmicity is potentially intrinsic to all myocardial cells. Localized pacemaking is a result of integrated action of cell-dependent expression profiles of various voltage-gated channels and electrical couplings between the myocardial cells. Under chemical or genetic perturbations, the pacemaker may become obscure or spatially shifted, leading to abnormal beatings. In this study, we observed perturbation by astemizole, which resulted in aberrant propagation of excitation. While inhibitory action at the ERG channel might be one possible mechanism for this drug-induced dysfunction in zebrafish, the present supporting evidence may not be inadequate.

Zebrafish and human heart share many molecules that play key roles in cardiac functions, the zebrafish heart thus providing a low cost, optically accessible small-sized system that can be used to test various chemical or genetic perturbations. Of course, such screens would not provide conclusive results directly applicable to human hearts, but instead, we may have rough but significantly fast and economical surveys of chemical or mutant libraries. An additional remarkable feature that is specific to zebrafish is that they can regenerate cardiac tissue after heart injury. Little is known about the cellular and molecular process underlying the regeneration. This voltage imaging technique may provide a unique tool to monitor electrical activities during the regenerative process of cardiac functions.

Further technical innovation is desirable. Because a heart is a sophisticated three-dimensional organ, voltage mapping should ideally be done in three dimensions. Although it may not be easy to acquire a three-dimensional dataset of FRET signals with a temporal resolution sufficient to resolve heartbeats, averaging of signals by an equivalent phase within the beat cycle may be performed to reconstruct a four-dimensional dataset (i.e. volume + phase) of the voltage dynamics. Such a reconstruction technique has been successfully applied to the morphological context of zebrafish hearts (Liebling et al. 2006).

In summary, we have here established an in vivo voltage mapping technique of a beating heart using a transgenic zebrafish that expresses mermaid under the control of a myocardial cell specific promoter. Combined with a variety of advantages of zebrafish including the powerful genetics, this optical method will provide new opportunities in broad areas of physiological, developmental and pharmacological cardiovascular research.

Glossary

Abbreviations

A-V

atrium-ventricle

CMLC

cardiac myosin light chain

ERG

ether-a-go-go-related gene

FRET

fluorescence resonance energy transfer

RMS

root mean square

Author contributions

H.T. designed the experiments. S.H. generated transgenic fish at Okazaki Institute. H.T. performed molecular biology, imaging and analysis at Osaka University and RIKEN. This project was supervised by A.M. and Y.O. All of the authors approved the final version to be published.

Supplemental material

Supplementary Figure 1

Supplementary video 1

Supplementary video 2

Supplementary video 3

Supplementary video 4

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer-reviewed and may be re-organized for online delivery, but are not copy-edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors