Multilayer mounting enables long-term imaging of zebrafish development in a light sheet microscope

Fluorescence microscopy is an invaluable tool for in vivo studies of developmental processes. However, excessive light exposure in commonly employed techniques, such as confocal, two-photon or spinning disc systems, causes phototoxicity and fluorophore bleaching, and only short intervals of developmental processes can be imaged. By contrast, in light sheet microscopy, such as selective plane illumination microscopy (SPIM) (Huisken et al., 2004) or digital scanned laser light sheet fluorescence microscopy (DSLM) (Keller et al., 2008), the sample is illuminated with a thin laser light sheet in the focal plane of the detection objective and imaged at high speed with a sensitive camera. The energy input into the sample is minimized, making light sheet microscopy ideally suited for time-lapse imaging of biological processes over long periods of time (Huisken and Stainier, 2009; Weber and Huisken, 2011), such as heart function and morphogenesis in zebrafish (Scherz et al., 2008; Arrenberg et al., 2010), neuronal cell migration in C. elegans (Wu et al., 2011), or cell lineage tracing during lateral line formation in zebrafish (Swoger et al., 2011).

Imaging a developing organism over many hours or a few days requires careful optimization of the mounting technique. In SPIM and DSLM, sample preparation is radically different from the dish or slide mounting used for conventional microscopy. Immersing a vertically oriented sample in an aqueous solution facilitates rotation of the fragile sample without deformation and provides unobstructed views from all angles (Huisken and Stainier, 2009).

The optimal mounting for imaging developmental processes should provide enough space for the sample to grow, while keeping it in a fixed position. For in vivo imaging in SPIM and DSLM, specimens are commonly embedded in a low-melting-point agarose cylinder (Huisken et al., 2004; Keller et al., 2010). The transparent agarose matches the refractive index of water (1.33) and biological tissue, and concentrations of 1.0-1.5% provide enough mechanical stability to reproducibly move the sample. However, such agarose embedding is only suitable for specimens that do not alter in shape or size during the experiment, such as Drosophila embryos (Huisken and Stainier, 2009; Preibisch et al., 2010).

The translucent zebrafish embryo is ideal for imaging vertebrate development (Godinho, 2011). Since embryo length increases by a factor of four within 3 days (Kimmel et al., 1995), time-lapse imaging of embryos embedded in a rigid scaffold, such as agarose, leads to severe growth restrictions and developmental defects (Fig. 1A,A′) (Keller et al., 2008; Keller et al., 2010). Thus, the potential of long-term imaging with light sheet microscopy is challenged by the lack of physiological mounting techniques.

We have evaluated the development of zebrafish embryos in different mounting media and tested their suitability for SPIM. We provide mounting protocols for the first days of zebrafish development for light sheet microscopy and other long-term imaging techniques. Mounting the sample in low concentrations of agarose or methylcellulose in refractive-index-matched fluorinated ethylene propylene (FEP) tubes ensures both immobility and growth of the embryo as well as optimal image quality from all sides.

Zebrafish (Danio rerio) adults and embryos were kept at 28.5°C and were handled according to established protocols (Nüsslein-Volhard and Dahm, 2002) and in accordance with EU directive 2011/63/EU as well as the German Animal Welfare Act. Transgenic lines Tg(kdrl:GFP) (Jin et al., 2005) and Tg(myl7:GFP) (Huang et al., 2003) were used. Tnnt2 morpholino injection was performed as described (Sehnert et al., 2002).

FEP tubes (Bola, S1815-04, refractive index 1.338, inner diameter 0.8 mm, outer diameter 1.6 mm) were flushed with 1 M NaOH (Merck), placed in 0.5 M NaOH and ultrasonicated for 10 minutes. They were flushed with double-distilled H₂O and with 70% ethanol. Tubes were then placed in 70% ethanol and ultrasonicated for 10 minutes. Finally, tubes were flushed with and stored in double-distilled H₂O. All solutions had been degassed and filtered using a syringe filter (Millex-HV PVDF 0.45 μm).

Low-melting-point agarose (Sigma) and methylcellulose (Sigma) were diluted in E3 (Nüsslein-Volhard and Dahm, 2002). For embedding in 1.5% (w/v) agarose, embryos were placed in liquid agarose, drawn into a glass capillary with Teflon-coated plungers (BRAND) and extruded after polymerization. For mounting in liquid media, FEP tubes were used as an enclosure. Embryos were placed in the mounting medium and drawn into the tube with a needle (100 Sterican, blunt) and syringe. The tube was plugged with solid 1.5% agarose. For long-term time-lapse imaging, the tube was coated with 3.0% methylcellulose containing 200 mg/l Tricaine (Aldrich). Methylcellulose was slowly drawn into the tube and out again. The coated tube was rinsed once with E3 containing 200 mg/l Tricaine.

Embryos at 24 hours postfertilization (hpf) were dechorionated and placed, one embryo per well, in 96-well plates (Nunc) filled with mounting medium. Control embryos were placed in E3 or 200 mg/l Tricaine. Each well was imaged every hour for 48 hours at 28.5°C using a Zeiss Axiovert 200M microscope. Embryo length was determined at 24 and 72 hpf using Fiji software (Schindelin et al., 2012) and compared with that in E3. Immobility was measured with Fiji by superimposing all images of the same well and subtracting the minimum from the maximum intensity projection. The mean gray value of the resulting image is an indicator of the movement of one embryo. Immobility in E3 and Tricaine was considered 0% and 100%, respectively.

Embryos at 24 hpf were placed in 96-well plates with different concentrations of Tricaine and incubated for 48 hours. Movies were acquired at 72 hpf using an Olympus SZX16 microscope and a Panasonic Lumix G2 camera. The average heart rate was determined by manual counting over a period of 8 seconds.

Embryos at 24 hpf were mounted in FEP tubes with 15 μl E3, 0.1% agarose or 3.0% methylcellulose, which were placed in a reaction tube filled with E3. Every 12 hours, six embryos were carefully removed from the tube, imaged and measured (Zeiss Axiovert 200M, Fiji). A different set of embryos was used for each time point to prevent damage of the embryos. Additional images were taken at 72 hpf (Olympus SZX16, AVT Stingray camera).

A multidirectional SPIM (mSPIM) setup was used as previously described (Huisken and Stainier, 2007). Long-term time-lapse images of developing embryos were acquired using a Leica HCX APO 10×/0.3W objective, Coherent Sapphire 488 nm laser and Andor iXon 885 camera. Embryos were imaged every 10 minutes or incubated at 28.5°C and imaged every 6 hours. Fluorescence and transmission images were projected using maximum intensity and Gaussian-based stack focuser algorithms (Fiji plug-in by J. Schindelin, University of Wisconsin-Madison), respectively.

Fluorescent beads (Estapor F-Y050, 500 nm, 1:2000 dilution) were imaged by mSPIM (Leica HCX APO 20×/0.5W objective, Chroma LP575 emission filter) in 1.5% agarose, 1.5% agarose in FEP tubes, and 1.5% agarose in glass capillaries (BRAND). Six samples per mounting condition were imaged in the center of each cylinder (147 planes, 0.5 μm step size) and cropped to the central 25%. Beads were detected in three dimensions with subpixel localization (Preibisch et al., 2010), registered and averaged. Lateral x and y and axial z line profiles were measured in the central slice of the resulting image stack using Fiji. In MATLAB (MathWorks), a Gaussian fit was performed on the measured intensity profile and the full width at half maximum (FWHM) was computed. Intensity traces in x, y and z for all beads extracted from one sample per mounting condition were plotted.

Because embedding in rigid agarose restricts the growth and morphogenesis of zebrafish embryos (Fig. 1A,A′), we examined alternative, physiological mounting methods. Embryos were mounted in different concentrations of agarose, methylcellulose and Tricaine at 24 hpf, and their development was monitored until 72 hpf. Growth and immobility were determined as criteria for proper development and spatial confinement for precise imaging.

Zebrafish embryos can be immobilized mechanically by embedding in a rigid scaffold or chemically by applying anesthetizing drugs such as Tricaine, which inhibits muscle contractions (Frazier and Narahashi, 1975). Tricaine concentrations ranging from 100 mg/l (Tsutsui et al., 2010) up to 750 mg/l (Truong et al., 2011) have been used as sedative, although as little as 30 mg/l Tricaine can influence embryonic development (Kimmel et al., 1995). Tricaine suppresses the contraction of not only skeletal muscles but also of smooth and cardiac muscles, resulting in reduced heart rate (Muntean et al., 2010) and affecting developmental processes that are influenced by hemodynamics (Culver and Dickinson, 2010).

We examined heart rate and morphology (Fig. 1B) after exposing embryos to Tricaine from 24-72 hpf. We used Tricaine concentrations of up to 800 mg/l, which was lethal after 2 days. At concentrations below 200 mg/l the heart rate fluctuated by less than 10%, which is consistent with a previous study (Craig et al., 2006). However, heart edemas were observed at concentrations as low as 100 mg/l (Fig. 1B, lower part). Thus, for long-term imaging, we recommend Tricaine concentrations that are as low as possible, not exceeding 200 mg/l or even 100 mg/l for cardiac studies. This concentration guarantees 80% immobilization on our scale. Additional stabilization of the sample is needed to achieve accurate and reproducible imaging over time.

Zebrafish embryos were examined in 0-1.5% agarose (Fig. 1C) and 0-3% methylcellulose (Fig. 1C′) from 24-72 hpf in order to determine the concentration that balances growth and immobility. These mounting media have already been used in various microscopy techniques (Renaud et al., 2011), are available in most laboratories, do not expose the sample to heat or mechanical stress during the mounting, and their refractive index is comparable to that of water. Agarose allows unperturbed growth of the embryos at a concentration as low as 0.1% (Fig. 1C) with an immobility of 80%. Agarose concentrations higher than 0.4% completely inhibit growth of the embryo. These results indicate that the commonly employed concentrations of agarose are non-physiological. In the case of methylcellulose, 85% immobility and 68% growth were achieved at a concentration of 3%. However, in contrast to solid 1.5% agarose, the viscous 0.1% agarose and 3.0% methylcellulose require an enclosure to hold the specimen in the microscope.

In conventional microscopy, glass slides provide a platform for imaging from one side. For light sheet microscopy, an enclosure for vertical mounting and multiview imaging is needed. We found a tube of FEP to be a promising support. FEP is well suited for imaging aqueous samples from all sides because of its matching refractive index of 1.338. FEP tubes have been used for short-term observations of zebrafish embryos in fluorescent widefield microscopy (Petzold et al., 2010) and optical projection tomography (OPT) (Bassi et al., 2011; McGinty et al., 2011).

We chose FEP tubes with an inner diameter matching the size of the fish, but still allowing for proper growth over a few days, and evaluated the development of embryos mounted in our media of choice. The FEP enclosure had no influence on the shape of the embryonic growth curve (Fig. 2A), and the differences in length at the end of our experiments were minor (Fig. 2B). Embryos mounted in a tube filled with E3 or 0.1% agarose grew 8% less than freely swimming siblings, possibly owing to reduced oxygen supply or a lack of movement. When 3% methylcellulose was used as mounting medium, the tube enclosure had no significant effect on growth. The morphology of the embryos suggests that development was delayed but not otherwise impaired (Fig. 2C). Embryos were raised and no difference in the survival rate was observed. We conclude that FEP tubes are physiological enclosures for non-rigid mounting media.

To quantify the optical performance of FEP tubes, fluorescent beads were embedded in 1.5% agarose inside a tube and imaged in SPIM. For comparison, beads were also imaged in a glass capillary and without an enclosure. For each condition, intensity profiles of beads were plotted (supplementary material Fig. S1). They showed little variance and were averaged to quantify aberrations caused by the surrounding material (maximum intensity projection in Fig. 3C). FEP does not distort the bead image compared with no enclosure, whereas a glass capillary causes severe aberrations (Fig. 3D). Therefore, FEP is well suited as an enclosure for non-rigid mounting media in light sheet microscopy and other microscopy systems.

Combining the results from the individual experiments, we have established protocols for the multilayer mounting of embryos for imaging their development over many days. In a direct comparison with a previous method using 1.5% agarose, our techniques constitute a major improvement. Whereas traditionally embedded embryos were unable to undergo morphological changes such as outgrowth of the head and died at 50 hpf (supplementary material Fig. S2), our mounting techniques allowed normal development.

To demonstrate the superior physiological and optical performance of the sample mounting technique, transgenic Tg(kdrl:GFP) embryos were embedded in 0.1% agarose enclosed in FEP at 24 hpf and imaged with SPIM for 3 days. The tubes had been coated with 3% methylcellulose before embedding to prevent adhesion of the tail to the tube. Transmission and fluorescence images show that development was undisturbed and finest detail in the developing vasculature was resolved (Fig. 3E; supplementary material Movie 1).

The ideal mounting conditions depend on the organ of interest, the developmental stage and the duration of the experiment (Fig. 4). Agarose at 1.5% represents a quick and stable mounting medium for acquiring a snapshot, and fiducial markers for multiview reconstructions can easily be added (Preibisch et al., 2010) (supplementary material Movie 2). Optionally, the sample can be enclosed in FEP for extra stability and distortion-free high-speed rotations. For long-term imaging, we suggest the use of 3% methylcellulose or 0.1% agarose in an FEP tube to ensure proper growth and immobilization of the embryo. Methylcellulose should be used when imaging for up to 1 day and for Tricaine-sensitive organs, such as the heart, as it provides stronger immobilization (supplementary material Movie 3). Otherwise, we recommend 0.1% agarose with 200 mg/l Tricaine as mounting medium in an FEP tube coated with 3.0% methylcellulose. This multilayer mounting ensures immobilization of the embryo over several days with only minor influences on development (supplementary material Movie 1).

During early stages of development, prior to 24 hpf, the embryo can be left in the chorion, which has no influence on the optical quality when imaged by SPIM. The embryos develop normally in the chorion and therefore the rigidity of the mounting medium is irrelevant (supplementary material Movie 4).

Light sheet microscopy is a promising technique for non-invasive imaging of fragile samples over long periods of time, which up to now has been limited by insufficient mounting techniques. This study provides techniques for mounting zebrafish and potentially other small vertebrates, unleashing the full potential of SPIM for long-term imaging. Protocols for light sheet microscopy will become even more important with the release of commercial microscopes. The mounting protocols presented here have also been successfully tested on the first commercial multiview light sheet microscope (Zeiss). Our results will also be applicable to other multi-angle imaging techniques, such as tomography (e.g. OPT). We have shown that dedicated sample mounting techniques are a key component in the application of current and future developments in microscopy.

We thank S. Preibisch and C. Perez-Campos for evaluating the bead images; J. Schindelin for the Fiji plug-in; B. Borgonovo for help with the FEP tube cleaning; and H.-O. K. Lee and D. Richmond for comments on the manuscript.