Light-sheet Fluorescence Microscopy to Capture 4-Dimensional Images of the Effects of Modulating Shear Stress on the Developing Zebrafish Heart

Summary

Here, we present a protocol to visualize developing hearts in zebrafish in 4-Dimensions (4-D). 4-D imaging, via light-sheet fluorescence microscopy (LSFM), takes 3-Dimensional (3-D) images over time, to reconstruct developing hearts. We show qualitatively and quantitatively that shear stress activates endocardial Notch signaling during chamber development, which promotes cardiac trabeculation.

Abstract

The hemodynamic forces experienced by the heart influence cardiac development, especially trabeculation, which forms a network of branching outgrowths from the myocardium. Genetic program defects in the Notch signaling cascade are involved in ventricular defects such as Left Ventricular Non-Compaction Cardiomyopathy or Hypoplastic Left Heart Syndrome. Using this protocol, it can be determined that shear stress driven trabeculation and Notch signaling are related to one another. Using Light-sheet Fluorescence Microscopy, visualization of the developing zebrafish heart was possible. In this manuscript, it was assessed whether hemodynamic forces modulate the initiation of trabeculation via Notch signaling and thus, influence contractile function occurs. For qualitative and quantitative shear stress analysis, 4-D (3-D+time) images were acquired during zebrafish cardiac morphogenesis, and integrated light-sheet fluorescence microscopy with 4-D synchronization captured the ventricular motion. Blood viscosity was reduced via gata1a-morpholino oligonucleotides (MO) micro-injection to decrease shear stress, thereby, down-regulating Notch signaling and attenuating trabeculation. Co-injection of Nrg1 mRNA with gata1a MO rescued Notch-related genes to restore trabeculation. To confirm shear stress driven Notch signaling influences trabeculation, cardiomyocyte contraction was further arrested via tnnt2a-MO to reduce hemodynamic forces, thereby, down-regulating Notch target genes to develop a non-trabeculated myocardium. Finally, corroboration of the expression patterns of shear stress-responsive Notch genes was conducted by subjecting endothelial cells to pulsatile flow. Thus, the 4-D light-sheet microscopy uncovered hemodynamic forces underlying Notch signaling and trabeculation with clinical relevance to non-compaction cardiomyopathy.

Introduction

Biomechanical forces, such as hemodynamic shear stress, are intimately involved in cardiac morphogenesis. In response to hemodynamic shear forces, myocardial ridges and grooves develop in a wave-like trabecular network in alignment with the direction of the shear stress across the atrioventricular (AV) valve¹. Cardiac trabeculation is necessary to increase contractile function and myocardial mass². Mutations in Notch signaling pathways result in congenital heart defects in humans and other vertebrates³. For example, gata1a⁴ and tnnt2a⁵ morpholino oligonucleotides (MO) have been shown to reduce erythropoiesis, while erythropoietin mRNA (EPO)⁶ and Isoproterenol (ISO)⁷ increase red blood cells and heart rate respectively, and therefore wall shear stress (WSS). Furthermore, ErbB2 signaling, downstream of Notch, promotes cardiomyocyte proliferation and differentiation to generate contractile force, which in turn activates Notch signaling^8,9. It is suggested that shear stress governs Notch signaling driven trabeculation for ventricular development. Currently, there are many studies that attempt to further understand the genetic programming events leading to congenital heart defects (CHD)^10,11,12, but very little are investigating how mechanical forces influence the forming heart.

In order to investigate the mechanical forces acting on the endocardium, close observation during the developmental period needs to be implemented. However, it is challenging to obtain good quality images of in vivo beating samples due to the inherence of traditional microscopy¹³. In order to observe the development over time within a sample, physical sectioning and staining, therefore, need to occur^13,14,15. Although confocal microscopy is widely used to image the 3-D structure of samples^14,16, these imaging systems' acquisition is still limited by slow scanning speeds.

Light-sheet fluorescence microscopy (LSFM) is a unique imaging technique that allows the visualization of in vivo dynamic events with long working distance¹³. This technique uses a light-sheet fluorescent microscopy to optically section a sample¹⁷. Due to illumination of only a thin sheet of light on the sample, there is a reduction in photo-bleaching and photo toxicity^13,18. The large field of view and long working distance allows for large samples to stay intact as they are imaged^13,14,17. The low magnification allows for a larger area to be imaged, while the long working distance allows for thicker samples to be imaged without compromising the signal-to-noise ratio. Many groups have used LSFM to image entire embryos¹⁷, brains^14,18, muscles and hearts¹⁹ among other tissues, showing the diverse types of samples that can be imaged.

Although previous research demonstrated reduced hemodynamic shear force by occluding the inflow or outflow tracks of the zebrafish heart, the information is solely qualitative. It results in an abnormal third chamber, diminished cardiac looping, and impaired valve formation²⁰. The 4-D LSFM images give a new perspective into the way the hemodynamic shear forces affect the development of the cardiac tissue. These mechanical forces may activate force-sensitive signaling molecules and induce the formation of the trabecular ridges. Because of the added time aspect of 4-D imaging, one is able to track changes in development in real time, which could lead to new revelations that had gone unnoticed previously. The zebrafish is an ideal model for imaging because scientists can observe an entire vertebrate animal versus only cell-cell interactions. Oxygen can also diffuse through the entire embryo, which allows development to occur without depending on the vascular system, unlike in mammalian development. Even though the zebrafish heart lacks the pulmonary organs, which require a four-chambered heart, there is a large number of cardiac genes that are conserved between zebrafish and humans²¹.

In this manuscript, we describe how to use light-sheet fluorescence microscopy to image the developing trabeculae in zebrafish hearts under various circumstances. First, injection of gata1a⁴ or tnnt2a⁵ MOs were used to lower the blood viscosity, and therefore WSS. The morphology of the heart was then recorded. In a separate group of fish, we increased the WSS by administering EPO mRNA⁶ or isoproterenol⁷ and observed the results. We also conducted a cell study with different pulsatile or oscillatory flow rates. After imaging each group, we found that WSS sensed by the endocardium via Notch signaling initiates trabeculation.

Protocol

The following methods were performed in compliance with UTA and UCLA IACUC protocols. These experimental groups were used with transgenic Tg(cmlc2:gfp), the wea (weak atrium) or clo (cloche) mutants: (a) wild-type (WT) control, (b) gata1a MO, and (c) tnnt2a MO injections (Table 1).

Model	Name	Modified Genes	Phenotype	Reference
Control	Wild type	None	N/A
	Tg(cmlc2:gfp)	Cardiac myosin light chain	Green flourescence specifically expressed in the myocardium	34
Decreased Wall Shear Stress	Weak atrium (wea) mutant	Atrium-specific myosin heacy chain	Atrium cannot contract, compact ventricle, thick myocardial wall, narrow lumen, dialated atrium,	37
	Cloche (clo) mutant	N/A	Complete removal of endocardium, unaturally large atrium with small ventricle, nonexistent cardiac cushins	41
	gata1a MO	gata1a	Anemia from lack of red blood cells	4
	tnnt2a MO	tnnt2a		39
Increased Wall Shear Stress	EPO mRNA	None	Severe polycythemia, increase in number of circulating blood cells, increased viscosity	6
	Isoprotenerol	None	Increased cardiac rate	7

Table 1: Zebrafish descriptions. Definitions and descriptions of Zebrafish used in experiments.

1. Study Setup

1. Prepare Nrg1 and EPO mRNA for trabeculation rescue.
   1. Obtain Human Nrg1 cDNA and amplify from a donor plasmid.
      Note: The Human Nrg1 cDNA was gifted from William Talbot from Stanford University.
      1. Take out necessary reagents and materials, namely PCR mastermix (stored in -20 °C), primers (See Table 2) (stored in -20 °C), a PCR plate (stored at room temperature), and a 20 μL pipette (stored at room temperature). See Table of Materials for example products.
         
         Table 2: Primers for human Nrg1 cloning.
      2. Prepare the mastermix for each primer set in triplicate using the 20 μL pipette. For 1 set of triplicates, use 37.5 μL of the mastermix, 3 μL of DNA-free water, 4.5 μL of primer. For more triplicates, simply multiply by the number of samples.
      3. Dilute the working solution Human Nrg1 cDNA (20 ng/μg concentration) in the primer strip with DNA-free water so that there is 10 μL total per well per sample.
         Note: Ensure that each solution under step 1.1.1 is evenly distributed. Do this by mixing each solution by pipetting up and down. To prevent cross-contamination, use one pipette tip for only one sample.
      4. Aliquot the Nrg1 cDNA with the 10 μL multi-channel pipette to desired wells in the PCR plate.
      5. Add 14.5 μL of the mastermix to the cDNA in the wells. Apply a sticky cover to the PCR plate to seal the wells, and place into the PCR machine.
      6. Start the PCR machine and ensure that the cycle reaches appropriate temperatures according to the enzymes and primers used.
   2. Clone the Nrg1 cDNA into the plasmid pCS2⁺ at the BamHI and EcoRI sites using excess enzymes BamHI and EcoRI to ensure cDNA is ligated in all plasmids.
      1. Mix the Nrg1 cDNA with the donor plasmid pCS2⁺, commercially available master mix solution and primers (Table 2).
         Note: The total reaction volume should be 50 μL with 25 μL of master mix, 3 μL of primers and 1 μL of 20 ng/μL donor plasmid with Nrg1 cDNA.
      2. Place the mix from step 1.1.2.1 in the PCR machine and set the parameters to meet the following specifications: 98 °C for 10 s, 55 °C for 5 or 15 s, and 72 °C for 5 s/kb.
      3. Purify the PCR product using a purification kit following the manufacturer's instructions. Elute the product into 50 μL of elution buffer.
      4. Digest the purified PCR product and 5 μg of pCS2⁺ with BamHI and EcoRI separately at 37 °C for 2 h in a 200 μL reaction volume. Purify the resulting digestion with a commercial kit and elute in 20 μL and 100 μL of Tris-HCl (pH 8.5), respectively.
      5. Ligate 4 μL of Nrg1 cDNA and 2 μL of the digested pCS2⁺ plasmid in a 25 μL reaction with 1 μL of a ligase catalyst at 16 °C overnight.
         Note: The procedure can be stopped here for overnight incubation.
      6. Use 2 μL of the ligation solution from step 1.1.2.5 and transform into 50 μL of E. coli bacteria cells suitable for cloning (see Table of Materials).
   3. Confirm that the transformation took place by screening the Nrg1 cDNA clones using PCR. Choose clones at random and add to a solution of specific primers, polymerase, and deoxyribonucleotide triphosphates (dNTPs).
      Note: The controls were the vector (pCS2⁺) with and without the insert (human cDNA). Do not pick too big of a colony because excessive bacteria can inhibit PCR.
      1. Pick 8 colonies from the transformation and screen pCS2⁺-Nrg1 clones using the PCR primers from Table 2.
      2. Culture one clone with Nrg1 cDNA to isolate the pCS2⁺-Nrg1 plasmid DNA. Inoculate with 100 μL of E. coli bacteria with pCS2⁺-Nrg1 plasmid into 100 mL of LB media, and culture by shaking (160 - 225 RPM) at 37 °C.
   4. Transfect purified pCS2⁺-Nrg1 plasmid into HEK-293 cells in a 6-well plate using a commercial transfection reagent following the manufacturer's instructions.
   5. 24 h after transfection, lyse the cells using a lysis buffer and run through an SDS-PAGE gel, transfer to a gel membrane, and then stain with anti-Nrg1 antibody²². Verify Nrg1 protein expression using a Western Blot.
      Note: The control is an empty pCS2⁺ plasmid. Can be paused here if gel membrane transfer occurs overnight.
   6. Synthesize Nrg1 mRNA using a commercial product similar to that in the Table of Materials and follow the manufacturer's instructions.
      1. Take 5 μL of the pCS2⁺-Nrg1 DNA from the bacteria isolated culture, and digest with NotI in a 200 μL reaction for 2 h at 37 °C to linearize the plasmid.
      2. Purify the linearized plasmid with a commercial kit and elute in 20 μL of Tris-HCl (pH 8.5).
      3. Conduct in vitro transcription with a commercial RNA isolation kit in a 20 μL reaction using 5 μL of the linearized plasmid following the manufacturer's instructions.
      4. Add the in vitro transcribed Nrg1 to 350 μL of RNA lysis buffer and then purify with an RNA isolation kit. Elute the RNA into 60 μL of Tris-HCl (pH 8.5).
      5. Measure the RNA concentration and store at -80 °C for future use.
   7. Follow steps 1.1.6.1 - 1.1.6.5 above for the EPO cDNA and mRNA preparation but clone the cDNA into the pCS2+ plasmid at EcoRI and XhoI sites instead.
2. Inject Morpholino into zebrafish
   1. Design²³ the MO injections using an online tool (Table of Materials) against the ATG sequence of gata1a⁴ (5′-CTGCAAGTGTAGTATTGAAGATGTC-3′) and tnnt2a⁵ (5′-CATGTTTGCTCTGATCTGACACGCA-3′).
   2. Add gata1a and tnnt2a MO separately to the nuclease-free water to make the final concentrations of 8 ng/nL and 4 ng/nL, respectively. Repeat with EPO mRNA with a final concentration of 20 pg/nL. Ensure that the final volume is 1 mL.
   3. Inject 1 nL of the 8 ng/nL of gata1a MO and 1 nL of the 4 ng/nL of tnnt2a MO into separate zebrafish embryos at the 1 to the 4-cell stage to reduce ventricular wall shear stress (WSS)²⁴. To increase WSS, inject 1 nL of 20 pg/nL of EPO mRNA⁶ into the zebrafish embryos at the 1- to 4- cell stage.
3. Chemically treat zebrafish to inhibit trabeculation
   1. Using a 20 mL pipette, dilute the 10 mg/mL AG1478 in 1% DMSO in E3 medium to a final concentration of 5 μM at 30 hpf into a 15 mL tube. As a control, treat each fish with 1% DMSO only.
   2. Add N-[N-(3,5-Difluorophenactyl)-L-Alanyl]-S-Phenylglycine t-butyl ester (DAPT) in 1% DMSO (100 μM) to E3 medium in a 15 mL tube to inhibit Notch signaling at 40 hpf in a similar fashion. As a control, treat with only 1% DMSO.

2. Imaging Techniques

1. 4-D cardiac LSFM Imaging with synchronization algorithm
   1. Take 2-D images of the entire fish embryo over multiple cardiac beating cycles using the steps below (Figure 1) . Ensure that the light-sheet thickness is approximately 5 μm. Take 500 x-y frames with an exposure time of 10 ms. Set the z-scanning to 1 μm for lossless digital sampling according to the Nyquist sampling principle²⁵.
      1. Create a 1% (w/v) agarose gel by adding 1 g of agarose to 100 mL and heating it until all agarose is dissolved. Load a small plastic tube with the embryo and agarose using a 20 μL pipette and secure it on the stage.
      2. Open the image viewing software and click Live to see the live view of the sample. Adjust the objective using the knob so that the sample is in focus. Similarly, move the stage so that the live view of the sample shows the top of the embryo.
      3. Take images 500 times for 5 s at 10X magnification using the microscope's software²⁶.
      4. Move the stage using the motor to a new layer (1 μm in the z-axis) and repeat the imaging procedure.
      5. Repeat above 2 steps until entire heart is fully imaged.
   2. Ensure data integrity by disregarding the first and last cardiac cycle images. Find the period of the cardiac cycle by matching images that were taken at the same time point in the cardiac cycle, but in different periods. Use the following equation that was developed to find the period:
      (1)
      where D is the cost function used for fitting a period hypothesis, I_m the captured image, τ' the time the image was captured, z_k the z-direction index of the image, x_m is the vector of the pixel index, and T' is the periodic hypothesis.
   3. Use the following equation to find the relative shift between two pictures in similar stages:
      (2)
      where Q_k',k denotes a cost function for the relative shift, R is the possible spatial neighborhood, L is the total time of capturing the image, x is the pixel index, z_k and z_k' are the first and second z-direction slices, t is time, and s is the relative shift hypothesis.
   4. Convert the relative shift to absolute shift with respect to the first image and solve the linear equation using the pseudo-inverse approach, to find the absolute relation to align the next section of images as previously described²⁷.
   5. The final 4-D images were processed using image processing software (Table of Materials).
      1. Stacking 2-D images into 3-D
         1. Open a commercial image processing software.
         2. Click Open Data.
         3. Select all the images to be stacked into 3-D > Click Load.
         4. Add in the voxel size of 0.65 x 0.65 x 1 > Click OK.
         5. Click the Multi-Planar View box to visualize the 3-D image.
         6. Right-click the blue slice_1.tiff box > Select Display > Select Volren.
         7. Click Edit > Select Options > Select Edit Colormap.
         8. Adjust the color using the boxes outlined in red > Click OK.
      2. Converting 3-D to 4-D
         1. Click "File > Open Time Series Data.
         2. Select the 3-D tiffs that were just made > Click Load.
         3. Enter the same voxel size as before.
         4. Right click the blue slice_1.tiff box > Select "Display" > Select Volren.
         5. Click Edit > Select Options > Select Edit Colormap. Adjust the color using the boxes > Click OK.
         6. Right click Time Series Control > Click Movie maker > Press the Play button in order to watch the movie.
         7. Add a file name, frame size, frame rate, quality equal to 1, type Monoscopic > Click Apply
         8. Export the video. Open the video in an appropriate software.

3. qRT-PCR analysis

1. Zebrafish heart RNA Isolation to quantify Notch ligands, receptors, and target genes' expressions
   1. Sacrifice the zebrafish by subjecting them to an overdose of tricaine methylate^28,29. Using a previously described protocol³⁰, the zebrafish hearts were excised and prepared for RNA isolation.
   2. Isolate the total RNA, and synthesize the cDNA using the cDNA synthesis kit following the manufacturer's instructions.
   3. Design the PCR primers for the Notch ligands Jag1, Jag2, Dll4, receptor Notch1b, and signaling related genes Nrg1 and ErbB2. See Table 3 for the primers we used.
   4. Add the primers from step 3.1.3, the isolated mRNA from step 3.1.2, and a commercial mastermix to the PCR plate. Run the qRT-PCR at the appropriate temperatures and times according to the enzymes used. Normalize the results to zebrafish α-actin.
      Note: This will calculate the expression levels for each of the ligands, receptors, and genes.

4. In vitro Human Aortic Endothelial Cell (HAEC) Experiments

1. Dynamic Shear Stress Model
   1. Culture HAEC cells with HAEC culture media. Once fully confluent, expose the cells to laminar flow and pulsatile flow of 23 x 10^-5 N at the rate of 1 Hz for 24 h.
      1. Warm up EC media/DMEM 10% FBS in a water bath for 20 min.
      2. Retrieve HAEC cells from liquid nitrogen tank and place the vial into the water bath until melted.
      3. Using a 1,000 μL pipette, remove the thawed cells and put them in a 15 mL tube with 3-5 mL of media. Centrifuge the cell solution at 53 x g for 3 min.
      4. Aspirate the solution off and add enough media for a total volume of 10 mL. Add the cell solution to a sterile T75 flask, cap the flask, and distribute the cells evenly on the bottom of the plate.
      5. Mark the flask with initials, date, cell type, and passage number. Incubate the flask at 37 °C, 5% CO₂, until cells are 80% confluent. Remember to change the media every 2-3 days.
      6. Using a commercial pump, attach the tubing to the flask, and set the parameters mentioned above in step 4.1.1^31,32,33.
   2. Add GI254023X to 50 mL of HAEC medium at a final concentration of 5 μM for 30 min.
   3. With the pre-mixed GI254023X, an ADAM10 inhibitor which suppresses Notch signaling, conduct the same laminar flow or pulsatile flow experiments as in 4.1.1.
   4. Quantify Notch ligands (Jag1, Jag2, and Dll4) and Notch target genes (hairy and enhancer of split [Hes]) using qRT-PCR for four groups of HAEC samples (laminar flow, laminar flow + GI254023X, pulsatile flow, pulsatile flow + GI254023X) described previously³².

5. Rescue and over-expression of Notch signaling

1. Inject 1 nL of the prepared Nrg1 mRNA at a concentration of 5 pg/nL (from step 1.1.6) at the 1- to 4-cell stage with the gata1a MO in order to overexpress Notch target genes²⁴.
2. Inject 1 nL of the Nrg1 mRNA into wea mutant in a similar way²⁴.
3. Double the concentration of Nrg1 mRNA (10 pg/nL) and inject²⁴ 1 nL into zebrafish to observe the ventricular morphology.
4. Inject 1 nL of the 20 pg/nL of EPO mRNA⁶ into the zebrafish embryos at the 1- to 4- cell stage to augment hematopoiesis to increase ventricular WSS as previously shown²⁴.
5. Inject 1 nL of the 50 μM Isoproterenol⁷, which increases the rate of contractility, to the E3 medium for 24 h while zebrafish are cultured as previously shown²⁴.
6. Perform 4-D LSFM imaging for each group. Follow the instructions in steps 2.1.1-2.1.5.2.8.

6. Quantification of Fractional Shortening and volume change over time

1. Perform 4-D LSFM imaging for each group at 50, 75, and 100 hpf. Follow instructions in steps 2.1.1-2.1.5.2.8. Divide each image so each has 600 nodes for replication of the cardiac wall motion.
2. Measure the change in diameter of the ventricle during diastole and systole using the following formula:
     (3)
   Load 3-D ventricular images at every 0.1 s with visualization software and measure the volume.
3. Plot the volume change over time for each experimental group at 75 hpf and 100 hpf.

Representative Results

LSFM was used in this manuscript in order to acquire high-resolution 2-D and 3-D pictures. As seen in Figure 1A and 1B, the illumination lens directs the light sheet at the sample. Because of the thinness of the light sheet, only a single plane is illuminated. The detection lens is positioned perpendicular to the illumination lens and is focused on the illuminated plane (Figure 1B). The light sheet from the illumination lens then scans the sample from left to right (Figure 1C). The scanning allows less photo-bleaching, photo-toxicity, and has a lower signal-to-noise ratio.

LSFM was applied to Tg(cmlc2a:gfp) embryos to visualize the entire cardiac structure³⁴ by rapid scanning (> 30 s) with a voxel resolution of 0.65 x 0.65 x 1 µm. Due to a single plane illumination, background noise was significantly reduced (Figure 1). Trabecular ridges protruded into the ventricular lumen at 75 hpf, and a trabecular network clearly was shown at 100 hpf (Figure 2A, B). Gata1a MO micro-injection reduced hematopoiesis and viscosity by 90%^4,35. This decreased viscosity attenuated hemodynamic shear stress, resulting in a delayed initiation and density of trabecular network at 75 hpf and 100 hpf when compared to control group (Figure 2C, D). Furthermore, Notch ligands (Dll4, Jag1, and Jag2), receptor (Notch1b), and downstream signaling components (Nrg1 and ErbB2) were down-regulated in response to gata1a MO injection (p < 0.05, n=5) (Figure 2K)^2,8,36. Co-injection of gata1a MO with 5 μM of Nrg1 mRNA up-regulated Notch signaling-related gene expression and rescued trabecular formation at 75 hpf and 100 hpf (Figure 2E, F, K). Thus, gata1a MO reduced hemodynamic forces leading to down-regulation of Notch signaling, whereas Nrg1 mRNA rescue up-regulated Notch-related genes and re-initiated trabeculation.

In order to further prove our hypothesis that changing hemodynamic forces and thus trabeculation via Notch signaling, we introduced the wea mutant that develops a non-contractile atrium³⁷. In the absence of ventricular inflow and hemodynamic forces during atrial systole, wea mutants express lower cardiac mRNA levels of Notch signaling component genes and target genes compared to controls, and harbor non-trabeculated, small and weakly contracting ventricles (Figure 2G, L). However, Nrg1 mRNA micro-injection into the wea mutants up-regulated the Notch signaling pathway, accompanied by the partial rescue of trabecular ridges, leading to a more pronounced contractile ventricle despite a non-contractile atrium (Figure 2H, L)². In this context, the wea mutants corroborated the relation between the non-contractile atrium and a non-trabeculated ventricle, underscoring the hemodynamic modulation of trabeculation via Notch signaling.

Tnnt2a MO was delivered to stop cardiac contraction by inhibiting cardiac troponin T^38,39. The tnnt2a MO-injected fish were completely devoid of circulation and ventricular WSS due to significant down-regulated Notch signaling, leading to a smooth ventricular surface and thin wall when compared to control group (Figure 2I, M). To investigate whether WSS on the endocardium was a requirement for trabeculation, we also have cloche (clo) mutants for which endocardium and the endothelial lining of the heart do not develop^40,41. The clo mutant failed to develop cardiac trabeculae and showed smaller sized ventricle and incomplete cardiac looping (Figure 2J). Thus, the endocardial lining is necessary to sense hemodynamic shear stress to activate Notch signaling (Figure 2M). However, when the hemodynamic shear stress was increased by increasing hematopoiesis or treatment of isoproterenol to increase contractility, the expression levels of Notch-related genes did not increase, and the trabecular network morphology did not change (Figure 3).

To further demonstrate the relationship between shear stress and Notch signaling, HAECs were used for in vitro tests. A pulsatile flow was exerted on confluent cells to simulate the hemodynamic shear stress observed by endothelial cells. It is shown that compared to static culture, pulsatile culture increases expression of Notch-related genes (Figure 4). Furthermore, treatment with the ADAM10 inhibitor significantly reduced Notch signaling (Figure 4).

Next, we calculated ventricular fractional shortening, and ventricular cavity change in volume over time for the wild-type, gata1a MO, AG1478, and those rescued by Nrg1 injection groups at 50 hpf, 75 hpf, and 100 hpf (Figure 5A-C). Treatment with ErbB signaling inhibitor, AG1478, significantly delayed and reduced fractional shortening during ventricular diastole at 100 hpf (Figure 5A). Both gata1a MO and AG1478 treatment reduced ventricular chamber size during contraction, as assessed by the changes in 4-D LSFM (Figure 5E, F). Both groups developed an increased end-systolic and -diastolic volume when compared to the wild-type. Co-injection of Nrg1 mRNA with gata1a MO restored both fractional shortening, and Ejection Fraction toward those of the wild type.

Figure 1: Fluorescent light sheet overview. (A) Representative LSFM system where the sample is placed in the intersection of the light sheet from the illumination lens (IL) and the detection path of the detection lens (DL). (B) The cylindrical lens behind the IL creates the micron-sized light sheet (purple) within the sample, which excites a thin sheet of the sample. The detection lens records the emitted fluorescent signal. (C) A schematic of the thin laser sheet (purple) optically sectioning the zebrafish embryo which moves right to left. This figure has been modified from Lee et al¹⁷. Please click here to view a larger version of this figure.

Figure 2: Ventricular morphologies of genetically manipulated Tg(cmlc:gfp) zebrafish with Notch Ligand, receptor, and target gene expression. (A) The trabecular network is shown at 75 hpf of control zebrafish. (B) The trabecular network at 100 hpf of control zebrafish. (C) Injection of gata1a MO at the 1- to 4-cell stage showed little to no trabeculation at 75 hpf. (D) Injection of gata1a MO into zebrafish had delayed trabecular network formation. (E) Co-injection of gata1a MO and human Nrg1 mRNA partially restored trabeculation at 75 hpf. (F) Co-injection of gata1a MO and human Nrg1 mRNA almost completely restored the trabecular network at 100 hpf. (G) wea mutant at 100 hpf shows no trabeculation and a smooth ventricular wall. (H) Injection of human Nrg1 into wea mutant zebrafish partially restored trabeculation at 100 hpf. (I) Injection of tnnt2a MO showed no trabeculation at 75 hpf (not shown) and 100 hpf. (J)clo mutant showed no trabeculation at 100 hpf. (K) Injection of the gata1a MO significantly reduced mRNA expression of Notch1, Nrg1, and Jag2 (t-test, *P < 0.05, n=5 vs. control). Co-injection of gata1a MO and human Nrg1 mRNA significantly increased the expression of Notch1b, Nrg1, ErbB, and Jag1 compared to control. (L) The wea mutant had the significantly lower expression of Notch-related genes (t-test, *P < 0.05, n=5 vs. control). Injection of human Nrg1 increased the expression of Notch-related genes; Jag1 and Jag2 were significantly higher than control. (M) The tnnt2a MO injection and clo mutant showed a significantly lower expression of Notch-related genes (t-test, *P < 0.05, n=5 vs. control). Scale bars: 50 μm. Data are shown as the mean standard deviation. This figure has been modified from Lee et al¹⁷. Please click here to view a larger version of this figure.

Figure 3: Increasing WSS in manipulated Tg(cmlc:gfp) zebrafish with Notch ligand, receptor and target gene expression in vivo. Increasing the WSS via over-expression of hematopoiesis with the injection of EPO(A) or addition of ISO (B) via increasing heart rate did not increase trabeculation and had a trabecular network similar to that of the control. (C) The injection of EPO mRNA or Isoproterenol did not change the expression of Notch ligand, receptor or target genes (t-test, *P < 0.05, n=5 vs. control). This figure has been modified from Lee et al¹⁷. Please click here to view a larger version of this figure.

Figure 4: Endothelial cells subject to oscillatory or pulsatile shear stress in vitro. HAECs subject to the pulsatile shear stress of τ_average = 30 x 10^-5 N/s at 1 Hz had upregulated Notch-related gene expressions and downregulated of Notch-related gene expressions when the ADAM10 inhibitor was applied (t-test, *P < 0.05, n=5 vs. control). This figure has been modified from Lee et al¹⁷. Please click here to view a larger version of this figure.

Figure 5: Effect of trabeculation of fractional shortening and ejection fraction. (A-C) Addition of AG1478 and gata1a MO significantly decreased the fractional shortening at 100 hpf, but co-injection of human Nrg1 and gata1a MO improved it. (D) gata1a injection and AG1478 significantly decreased the fractional shortening at 100 hpf (t-test, *P < 0.05, n=5 vs. control). Co-injection of gata1a and Nrg1 mRNA improved the fractional shortening. (E-F) Integrating the 4-D synchronization algorithm with LSFM showed that AG1478 and gata1a MO had increased end systolic and diastolic volumes at 75 hpf (E) and 100 hpf (F). Data are shown as the mean standard deviation. This figure has been modified from Lee et al¹⁷. Please click here to view a larger version of this figure.

Zebrafish Jag1	Forward	CCGCGTATGTTTGAAGGAGTATCAGTCG
	Backward	CAGCACGATCCGGGTTTTGTCG
Zebrafish Jag2	Forward	AGCCCTAGCAAAACGAGCGACG
	Backward	GCGTGAATGTGCCGTTCGATCAA
Zebrafish Dll4	Forward	CAAAGTGGGAAGCAGACAGAGCTAAGG
	Backward	CGGTCATCCCTGGGTGTGCATT
Zebrafish BMP10	Forward	GCATCAAGGGGCCACTCGTGTAGA
	Backward	TCGTCTCACTCCACTAGGTCCCATACTG
Zebrafish ErBb2	Forward	GATCAGGACTGCCAAACATTGACGTCT
	Backward	AGCAGCACACTGAACATGGCAGCA
Zebrafish Nrg-1	Forward	GTGTGTTTGTCCCTGTGGACGCGT
	Backward	CCTCCTGGAGCTTCCCCTCAAACA
Zebrafish Notch1b	Forward	CAGAGAGTGGAGGCACAGTGCAATCC
	Backward	GCCGTCCCATTCACACTCTGCATT

Table 3: Primers used for Zebrafish screening.

Discussion

In this protocol, we have shown that 4-D imaging can be used to track the development of a trabecular network in response to changes in biomechanical forces. In particular, the shear stress experienced by endothelial cells initiates the Notch signaling cascade, which in turn promotes trabeculation. In this manuscript, we have shown that (1) gata1a MO injection decreased hematopoiesis and therefore it reduced wall shear stress, (2) tnnt2a MO injection inhibited ventricular contractile function to reduce wall shear stress, and (3) wea mutants lacked atrial contraction, which then decreased the hemodynamic force applied to the ventricle. By reducing the shear stress on the cardiac walls, we quantified the Notch signaling via qRT-PCR and found that with reduced WSS, there was reduced Notch signaling. To further prove our hypothesis, clo mutants that lacked endocardium did not form trabeculae, and subjecting endothelial cells to pulsatile shear stress with ADAM10 present showed that shear stress activates Notch signaling.

The other imaging techniques such as confocal microscopy can be used to image samples in 3-D¹⁴. However, this technique and others have many drawbacks. For example, the confocal imaging cannot penetrate deeply into samples, has the low axial resolution, and slow scanning speeds^19,42. Because of the penetration depth limit in confocal imaging, small samples, such as zebrafish embryos, can only be imaged in its entirety. Two-photon confocal microscopy improves the penetration depth, but the resolution, slow scanning speed, and high cost make this technique undesirable¹³. There have been studies that combine two-photon microscopy and LSFM successfully¹⁸, however, the drawbacks are still prominent⁴².

LSFM, on the other hand, has the least photo-bleaching and damage^13,18,42, fast acquisition speed, and similar penetration depth¹³. Additionally, because of the dual lenses and excitation of a single plane (Figure 1), the background noise is substantially reduced⁴³. However, the samples need to be embedded into gels to immobilize the sample, which makes it difficult to apply to in vivo models other than zebrafish^14,17,19. Also, utilization of the fluorescent light limits the depth penetration to a few millimeters. Even though LSFM is a great method for visualization, it is critical that it is supplemented with quantitative data. Because LSFM is only qualitative, it can be subject to various interpretations. Using qRT-PCR, we could confirm that the LFSM 4-D images were a correct representation of the events occurring in the live samples. Occasionally in larger samples, stripes and shadows can form in the images. However, dual-sided illumination or mSPIM can be utilized to reduce these artifacts⁴². Because LSFM is so versatile, its capability of capturing high resolution and overall better-quality images makes the future of LSFM promising. Mentioned previously, LSFM can be combined with other imaging systems for successful imaging^18,42.

We demonstrated previously that applying a pulsatile shear stress of 23 x 10^-5 N at 1 Hz for 24 h substantially upregulates Notch signaling in vitro(Figure 4)¹⁷. We further demonstrated that shear forces activate Notch signaling by genetic manipulation of hematopoiesis (gata1a MO)⁴, cardiac contraction (tnnt1a MO)⁵, atrial contraction (wea mutant), endocardial deletion (clo mutant), and localization of Notch signaling (Tg[flk:mCherry;tp1:gfp]). We describe here that by lowering the wall shear stress in the ventricle, that Notch-related genes were down-regulated (Figure 2K, L, M). Furthermore, we were able to demonstrate that Nrg1 mRNA was able to rescue trabeculation, restore Notch signaling, and improve ventricular fractional shortening and ejection fraction (Figure 2E, F, H, K-M, Figure 5). The restoration of cardiac function was shown by the increase in contractile function-mediated hemodynamic forces that activated Notch signaling (Figure 2K-M).

We also determined that the shear stress-activated Notch signaling is endocardial dependent. By using gata1a MO, tnnt2a MO, wea mutant or AG1478 treatment, trabeculation was inhibited and Notch signaling was down-regulated (Figure 2K-M). However, Nrg1 rescued trabeculation and Notch signaling in wea mutants and when co-injected with gata1a MO (Figure 2D, H, K, L). When wall shear stress (EPO injection) or Notch signaling (10 pg/nL Nrg1 injection) was increased, there was no change in the EPO group in terms of morphology, but abnormal ventricular morphogenesis was present in the Nrg1 increased dose (Figure 3).

In conclusion, we have shown that the mechanotransduction of shear stress activates the Notch signaling pathway. By applying 4-D LSFM and genetically engineered zebrafish, we developed a new model system that shows how critical blood flow is during cardiac development to its morphology, contractility, and overall function.

Materials

Name	Company	Catalog Number	Comments
Clontech Hifi PCR pre-mix	Takara	639298	PCR mastermix 1.1.1.1, 1.1.1.2, 1.1.1.5, 1.1.2.1, 3.1.4
Human Nrg1 cDNA	Gift from William Talbot, Stanford University, Stanford, California, USA	N/A	Used for trabeculation rescue 1.1.1.3, 1.1.1.4, 1.1.2.1
CFX Connect™ Real-Time PCR Detection System	Bio-Rad	1855201	PCR Machine 1.1.1.5, 1.1.1.6, 1.1.2.2, 1.1.3.2
pCS2+	GE Health		Plasmid used to synthesize mRNA 1.1.2.1, 1.1.2.4, 1.1.2.5
Nucleospin purification kit	Clontech	740609.25	DNA Purification 1.1.2.3, 1.1.2.4, 1.1.6.2
T4 DNA ligase	Clontech	2011A	PCR Ligation solution 1.1.2.5
Stellar competent cells	Clontech	636763	E. coli cells used for transformation 1.1.2.6, 1.1.3.1, 1.1.3.2
Lipofectamine 2000 transfection reagent	Life Technologies	11668027	Transfection reagent 1.1.4
mMessage SP6 kit	Invitrogen	AM1340	Kit used to synthesize mRNA 1.1.6.3
Aurum Total RNA Mini Kit	Bio-Rad	7326820	Purifies RNA  1.1.6.4, 3.1.2
GeneTools 4.3.8	GeneTools	N/A	Software for primer design 1.2.1, 3.1.3
EPO cDNA	Creative Biogene	CDFH006026	Increases WSS 1.2.2, 1.2.3, 1.1.7
AG1478	Sigma-Aldrich	T4182	ErbB inhibitor 1.3.1
E3 medium			To grow embryos 1.3.1, 1.3.2, 5.1.5
DAPT	Sigma-Aldrich	D5942	γ-secretase inhibitor 1.3.2
Agarose	Sigma-Aldrich	A9539	Used for mounting embryos 2.1.1.1
ORCA-Flash4.0 LT Digital CMOS camera	Hamamatsu Photonics	C11440-42U	Used to capture Images 2.1.1.2, 2.1.1.3
Amira Software	FEI Software	N/A	Visualized and Analysed images into 3D, and 4D 2.1.5.1.1-2.1.5.2.8
Tricaone mesylate	Sigma-Aldrich	886-86-2	Used to humanely sedated or sacrifice embryos 3.1.1
iScript cDNA Synthesis Kit	Bio-Rad	1708890	Synthesizes cDNA 3.1.2
Eppendorf 5424 microcentrifuge	Eppendorf	05-400-005	Microcentrifuge 4.1.1.3
GI254023X	Sigma-Aldrich	260264-93-5	ADAM10 inhibitor 4.1.2, 4.1.3
Isoprenaline hydrochloride	Sigma-Aldrich	I5627	Isoproterenol increases WSS 5.1.5
MATLAB	Mathworks	N/A	Cardiac mechanics analysis