Displacement Analysis of Myocardial Mechanical Deformation (DIAMOND) Reveals Segmental Heterogeneity of Cardiac Function in Embryonic Zebrafish

Summary

The goal of this protocol is to detail a novel method for the assessment of segmental cardiac function in embryonic zebrafish under both physiological and pathological conditions.

Abstract

Zebrafish are increasingly utilized as a model organism for cardiomyopathies and regeneration. Current methods evaluating cardiac function fail to reliably detect segmental mechanics and are not readily feasible in zebrafish. Here we present a semiautomated, open-source method for the quantitative assessment of four-dimensional (4D) segmental cardiac function: displacement analysis of myocardial mechanical deformation (DIAMOND). Transgenic embryonic zebrafish were imaged in vivo using a light-sheet fluorescence microscopy system with 4D cardiac motion synchronization. Acquired 3D digital hearts were reconstructed at end-systole and end-diastole, and the ventricle was manually segmented into binary datasets. Then, the heart was reoriented and isotropically resampled along the true short axis, and the ventricle was evenly divided into eight portions (I–VIII) along the short axis. Due to the different resampling planes and matrices at end-systole and end-diastole, a transformation matrix was applied for image registration to restore the original spatial relationship between the resampled systolic and diastolic image matrices. After image registration, the displacement vector of each segment from end-systole to end-diastole was calculated based on the displacement of mass centroids in three dimensions (3D). DIAMOND shows that basal myocardial segments adjacent to the atrioventricular canal undergo the highest mechanical deformation and are the most susceptible to doxorubicin-induced cardiac injury. Overall, DIAMOND provides novel insights into segmental cardiac mechanics in zebrafish embryos beyond traditional ejection fraction (EF) under both physiological and pathological conditions.

Introduction

Chemotherapy-induced cardiac toxicity and ensuing heart failure are one of the main reasons for chemotherapy discontinuation¹. Therefore, cardiac functional assessment plays a crucial role in the identification of cardiac toxicity and, more importantly, in the prediction of early cardiac injury following chemotherapy². However, current approaches for cardiac functional assessment encounter limitations. Methods such as left ventricular ejection fraction (LVEF) provide only global and often delayed cardiac mechanics after injury^3,4. Tissue Doppler imaging provides segmental myocardial deformation information but suffers from significant intraobserver and interobserver variability, in part due to ultrasound beam angle dependency⁵. Two-dimensional (2D) speckle tracking utilizes the B-mode of echocardiography, which theoretically eliminates the angle dependency, but its accuracy is limited by out-of-plane motion⁶. Therefore, a rigorous approach for quantifying segmental cardiac function is lacking in both research and clinical settings.

In this context, we developed a 4D quantification method for the analysis of segmental cardiac function that we named displacement analysis of myocardial mechanical deformation (DIAMOND), to determine the displacement vectors of myocardial mass centroids in 3D space. We applied DIAMOND for the in vivo assessment of cardiac function and doxorubicin-induced cardiac toxicity with zebrafish (Danio rerio) as the animal model, chosen due to their regenerating myocardium and highly conserved developmental genes⁷. We further compared segmental DIAMOND displacement with global ejection fraction (EF) determination and 2D strain following doxorubicin treatment. By integrating DIAMOND displacement with 4D light-sheet fluorescent microscopy (LSFM) acquired rendering of embryonic zebrafish hearts, DIAMOND shows that the basal myocardial segments adjacent to the atrioventricular canal undergo the highest mechanical deformation and are the most susceptible to acute doxorubicin cardiac injury⁸.

   

Protocol

All methods described here have been approved by the UCLA Institutional Animal Care and Use Committee (IACUC), and experiments were performed in compliance with protocols approved by the UCLA Office of Animal Research.

1. Breeding Tg(cmlc2:mCherry) zebrafish and collection of embryos

1. Follow the housing, breeding, and embryo collection procedures as described in previously established husbandry and breeding practices. For details, see Messerschmidt et al.⁹.
2. Treat the collected embryos with 0.003% 1-phenyl-2-thiourea (PTU) in E3 medium 18 h postfertilization to maintain the transparency of the embryos for LSFM imaging.

2. Doxorubicin treatment to induce cardiac injury

1. At 3 days postfertilization (dpf), treat the embryos with doxorubicin at a concentration of 10 μM in E3 fish water medium. After a 24 h treatment to 4 dpf, replace the doxorubicin medium with fresh E3 medium.
   CAUTION: Doxorubicin is a chemotherapy medication. Appropriate personal protective equipment (PPE) is required and the waste should be disposed of in biohazard waste containers.

3. Notch pathway modulation

1. Treat zebrafish embryos with the Notch pathway inhibitor (2S)-N-[(3,5-difluorophenyl)acetyl]-L-alanyl-2-phenyl]glycine 1,1-dimethylethyl ester (DAPT) at a concentration of 10 μM in E3 fish water medium from 3–6 dpf.
2. Microinject the Notch downstream effectors Notch intracellular domain (NICD) and Neuregulin-1 (Nrg-1) mRNA at concentrations of 10 pg/nL and 5 pg/nL, respectively, into the 1-cell stage zebrafish embryos^8,10.
   NOTE: The microinjection is performed under a microscope with the support of an air pump to accurately control the volume injected. The mRNA microinjection into the cell is done when the fertilized egg is at the first cell stage. For details on the preparation and sequence of the mRNAs, see Chen et al.⁸. For details on the microinjection and preparation of injection needles, see Rosen et al.¹⁰.

4. LSFM imaging and post-imaging synchronization

1. For the LSFM imaging techniques and post-imaging synchronization algorithm, see details in previous publications^9,11.
   NOTE: Briefly, our system utilizes a continuous-wave laser as the illumination source to image all transgenic zebrafish lines. The detection module is composed of two scientific complementary metal-oxide semiconductor (sCMOS) cameras and two sets of filters for dual-channel imaging. The detection module is perpendicularly installed to the illumination plane. Each LSFM frame is acquired within a 20 msec exposure time, while the resolving power in cross section is ~0.65 μm and the step size between consecutive frames is ~2 μm. A 589 nm laser was used to excite mCherry fluorescent signals.

5. Reconstruction of the 3D systolic and diastolic heart

1. Open the folder created by the post synchronization algorithm, then open the "Output" folder. Select the middle plane of the heart and load the entire folder into ImageJ. Find the first diastolic and systolic phase and record the frame number.
2. Open the "Output/By State" folder and find the folders that have the same numbers as the frame numbers just recorded. Convert the images in the folder into 3D TIFF (tagged image file format) files and name them "diastole.tif" and "systole.tif".

6. Segmentation of the ventricle

1. Open the image analyzing software (see Table of Materials). Click File | Open data, and load "diastole.tif" and "systole.tif". Enter the voxel size according to the imaging settings.
   NOTE: For the LSFM system used, the typical voxel size is 0.65 µm x 0.65 µm x 2 µm.
2. Click the "SEGMENTATION" panel and manually segment out the ventricle part of the heart. The built-in "Threshold" tool that can select all the regions above a certain intensity can facilitate this process. The ventricle is the thicker chamber with a stronger fluorescence.
   NOTE: Make sure to remove the atrioventricular canal and the outflow tract in the segmented ventricle, because this affects the displacement analysis.
3. After the segmentation is done, click the "Project" panel. Right click the "diastole.Labels.tif" and "systole.Labels.tif" tabs in the console and click "Export Data as" to save the data as 3D TIFF files.

7. Creation of rectangular parallelepipeds for image registration

1. Run "prepImage_1.m" in the programming environment (see Table of Materials). Open "prepImage_1.m", "ImPath" in line 5 so the folder contains the original and segmented TIFF files, and change "slice" in line 4 to the number of slices of the 3D tif files.
2. After running the code, it will generate five new 3D TIFF files ("test.tif", "diastole_200.tif", "systole_200.tif", "diaLabel.tif", and "sysLabel200.tif") as well as two new folders ("resample_dia" and "resample_sys").

8. Resample systolic and diastolic 3D hearts along the short axis plane

1. Import all five 3D TIFF files into the image analyzing software (see Table of Materials).
   NOTE: The voxel size is unchanged.
2. Go to the MULTIPLANAR panel. Choose "diastole_200.tif" as the primary data. Align the X-axis (the green line in the XY plane) with the vertical long axis of the ventricle, and align the Z-axis (the red line in the YZ plane) with the horizontal long axis of the ventricle.
   NOTE: The vertical long axis is determined by finding the longest axis connecting the apex and the outflow tract in the XY plane, and the horizontal long axis is determined by finding the longest axis connecting the apex and the outflow tract in the YZ plane. Rotate the axis by placing the cursor at the end of the axis.
3. Choose three random points from the oblique YZ plane (the short axis plane) in a counterclockwise manner and record their 3D position coordinates.
   NOTE: Make sure the points are chosen in a counterclockwise manner.
4. Repeat steps 8.2 and 8.3 for "systole_200.tif".
5. Click the "PROJECT" panel. Create a "Slice" object for "diastole_200.tif" by right clicking on "diastole_200.tif" and searching for "Slice" object. Left click the Slice object just created, and in the Properties panel | Options, check "Set Plane" and choose three points in "Plane Definition". Enter the coordinates of the three points from step 7.3.
6. Repeat step 8.5 for "systole_200.tif".
   NOTE: The slice object created should have the name "Slice 2".
7. Right click "diastole_200.tif" and search for "Resample Transformed Image" and create the object. In the Properties panel, choose "Slice" as the "Reference" and click Apply. This should generate an object named "diastole_200.transformed".
8. Right click "diastole_200.transformed" and search for "Resample" and create the object. Choose "Voxel Size" as the "Mode" and change "Voxel Size" to be x = 1, y = 1, and z = 1 in the Properties panel.
9. Click "Apply". This should generate an object named "diastole_200.resampled". Right click "diastole_200.resampled" and save it as a 3D TIFF file.
10. Repeat the same step for "diaLabel.tif" and "test.tif". Save "diaLabel.resampled" and "test.resampled" as 3D TIFF files. Repeat the same step for "systole_200.tif", "sysLabel.tif", and "test.tif" using "Slice 2" as a reference, and save "systole_200.resampled", "sysLable.resampled", and "test2.resampled" as 3D TIFF files.
    NOTE: Make sure there are a total of six TIFF files saved in this step.

9. Division of the resampled heart

1. Import all six resampled files from step 8 to ImageJ. Select a slice of "systole_200.resampled" in which the atrioventricular canal is clearly visualized. Record the number of the slice.
   1. Use the "Image | Transform | Rotate" function of ImageJ so that the atrioventricular canal is vertical. Apply the same rotation to all files. Close all windows and save all changes.
   2. Move "diastole_200.resampled", "diaLabel.resampled", and "test.resampled" to the "resample_dia" folder, and move "systole_200.resampled", "sysLable.resampled", and "test2.resampled" to the "resample_sys" folder.
2. Open "divider_2_8_pieces.m". Change "ImPath" in line 5 and "ImPath" in line 395 to the image directory. Change the variable "Middle" in line 22 and line 411 to the slice numbers where the atrioventricular canal is clearly visualized in "systole_200.resampled" and "diastole_200.resampled".
3. Run the code, and in the prompted windows click once at the center of the ventricle and click once at the center of the atrioventricular canal. This needs to be done twice for both systole and diastole images.

10. Registration of systolic and diastolic image matrices

1. Open "register_3.m" and change "ImPath" in line 4 to the image folder path. It might take 5–20 min to run this code depending on the computation power of the system.
   NOTE: The artificially created rectangular parallelepipeds in step 7 are used for 3D rigid registration that preserves the distance between two points and angles subscribed by three points. When the end-diastole rectangular parallelepiped (red) is registered to the end-systole rectangular parallelepiped (green), the ensuing discrepant 3D location permits the derivation of a unique matrix of rigid transformation consisting of rotation and translation from the end-diastole matrix to the end-systole matrix (Figure 1H). We perform the registration and regularized energy minimization to denoise the matrix after the transformation using an image processing toolbox (see Table of Materials). For a detailed mathematical description, please see Chen et al.⁸.

11. Output of the displacement vectors

1. Open "displacement_4.m" and change "ImPath" in line 4 to the image folder path.
2. Run "displacement_4.m", which generates a "vector8.txt" file in the "vectors" folder. Once the "vector8.txt" file is open, there will be an 8 x 4 matrix. Each row of the matrix has four numbers, which are the magnitudes of the X component, Y component, Z component, and the SUM magnitude of the displacement vector of a specific segment of the ventricle.
   NOTE: The displacement vector is obtained by calculating the displacement of the mass centroid of each segment in 3D space. We calculate the 3D mass centroid (P_S and P_D) coordinates (where k indicates the X, Y, or Z coordinate, respectively) of each segment (I-VI) in the segmentation dataset from systole to diastole (Figure 1J). We define the mass centroid in 3D space as follows:
   
   where C_x = X, C_y = Y, and C_z = Z, M_i = the mass of each segment (I ≤ i ≤ VI), m = the number of voxels of each segment, and ρ = the density function as the segmented region is 1 whereas the rest is 0. The L2-norm of the sub-displacement vectors along the X-, Y-, and Z-axes and the sum displacement vector are calculated during the cardiac cycle. There are a total of eight rows in the matrix. The first row and the eighth row contain the atrioventricular canal and are thus ignored in our analysis. Segments I to VI are represented by the second row to the seventh row.

Results

The process by which DIAMOND was developed to assess 3D segmental cardiac function is presented in Figure 1. Following LSFM image acquisition and reconstruction in 3D of the embryonic zebrafish heart (Figure 1A), the true short axis plane was determined as the plane perpendicular to the vertical and horizontal long axes, both of which are determined in a multiplane viewer (Figure 1B). The heart was ...

Discussion

A rigorous strategy for quantification of segmental myocardial function is critical to assess cardiac mechanics beyond traditional EF, known to be an insensitive and delayed indicator of myocardial injury^1,4,12. Hence, there has been a growing interest in markers of early myocardial changes, and a growing body of literature supports myocardial deformation parameters as an early indicator to forecast ventricular dysfunction

Materials

List of materials used in this article

Name	Company	Catalog Number	Comments
Amira6	FEI		Image analyzing software
DAPT	Millipore Sigma	D5942-5MG
Doxorubicin hydrochloride	Millipore Sigma	D1515-10MG
Ethyl 3-aminobenzoate methanesulfonate	Millipore Sigma	E10521-10G	Tricaine
MATLAB	MathWorks		Programming environment
MATLAB Image Processing Toolbox	MathWorks		Image processing toolbox