Deep learning enables automated volumetric assessments of cardiac function in zebrafish

Bringing Artificial Intelligence (AI) into Environmental Toxicology Studies: A Perspective of AI-Enabled Zebrafish High-Throughput Screening

The booming development of artificial intelligence (AI) has brought excitement to many research fields that could benefit from its big data analysis capability for causative relationship establishment and knowledge generation. In toxicology studies using zebrafish, the microscopic images and videos that illustrate the developmental stages, phenotypic morphologies, and animal behaviors possess great potential to facilitate rapid hazard assessment and dissection of the toxicity mechanism of environmental pollutants. However, the traditional manual observation approach is both labor-intensive and time-consuming. In this Perspective, we aim to summarize the current AI-enabled image and video analysis tools to realize the full potential of AI. For image analysis, AI-based tools allow fast and objective determination of morphological features and extraction of quantitative information from images of various sorts. The advantages of providing accurate and reproducible results while avoiding human intervention play a critical role in speeding up the screening process. For video analysis, AI-based tools enable the tracking of dynamic changes in both microscopic cellular events and macroscopic animal behaviors. The subtle changes revealed by video analysis could serve as sensitive indicators of adverse outcomes. With AI-based toxicity analysis in its infancy, exciting developments and applications are expected to appear in the years to come.

Cilia function as calcium-mediated mechanosensors that instruct left-right asymmetry

The breaking of bilateral symmetry in most vertebrates is critically dependent upon the motile cilia of the embryonic left-right organizer (LRO), which generate a directional fluid flow; however, it remains unclear how this flow is sensed. Here, we demonstrated that immotile LRO cilia are mechanosensors for shear force using a methodological pipeline that combines optical tweezers, light sheet microscopy, and deep learning to permit in vivo analyses in zebrafish. Mechanical manipulation of immotile LRO cilia activated intraciliary calcium transients that required the cation channel Polycystin-2. Furthermore, mechanical force applied to LRO cilia was sufficient to rescue and reverse cardiac situs in zebrafish that lack motile cilia. Thus, LRO cilia are mechanosensitive cellular levers that convert biomechanical forces into calcium signals to instruct left-right asymmetry.

RBPMS2 Is a Myocardial-Enriched Splicing Regulator Required for Cardiac Function

BACKGROUND

RBPs (RNA-binding proteins) perform indispensable functions in the post-transcriptional regulation of gene expression. Numerous RBPs have been implicated in cardiac development or physiology based on gene knockout studies and the identification of pathogenic RBP gene mutations in monogenic heart disorders. The discovery and characterization of additional RBPs performing indispensable functions in the heart will advance basic and translational cardiovascular research.

METHODS

We performed a differential expression screen in zebrafish embryos to identify genes enriched in nkx2.5-positive cardiomyocytes or cardiopharyngeal progenitors compared to nkx2.5-negative cells from the same embryos. We investigated the myocardial-enriched gene RNA-binding protein with multiple splicing (variants) 2 [RBPMS2)] by generating and characterizing rbpms2 knockout zebrafish and human cardiomyocytes derived from RBPMS2-deficient induced pluripotent stem cells.

RESULTS

We identified 1848 genes enriched in the nkx2.5-positive population. Among the most highly enriched genes, most with well-established functions in the heart, we discovered the ohnologs rbpms2a and rbpms2b, which encode an evolutionarily conserved RBP. Rbpms2 localizes selectively to cardiomyocytes during zebrafish heart development and strong cardiomyocyte expression persists into adulthood. Rbpms2-deficient embryos suffer from early cardiac dysfunction characterized by reduced ejection fraction. The functional deficit is accompanied by myofibril disarray, altered calcium handling, and differential alternative splicing events in mutant cardiomyocytes. These phenotypes are also observed in RBPMS2-deficient human cardiomyocytes, indicative of conserved molecular and cellular function. RNA-sequencing and comparative analysis of genes mis-spliced in RBPMS2-deficient zebrafish and human cardiomyocytes uncovered a conserved network of 29 ortholog pairs that require RBPMS2 for alternative splicing regulation, including RBFOX2, SLC8A1, and MYBPC3.

CONCLUSIONS

Our study identifies RBPMS2 as a conserved regulator of alternative splicing, myofibrillar organization, and calcium handling in zebrafish and human cardiomyocytes.

Intrinsic myocardial defects underlie an Rbfox-deficient zebrafish model of hypoplastic left heart syndrome

Hypoplastic left heart syndrome (HLHS) is characterized by underdevelopment of left sided structures including the ventricle, valves, and aorta. Prevailing paradigm suggests that HLHS is a multigenic disease of co-occurring phenotypes. Here, we report that zebrafish lacking two orthologs of the RNA binding protein RBFOX2, a gene linked to HLHS in humans, display cardiovascular defects overlapping those in HLHS patients including ventricular, valve, and aortic deficiencies. In contrast to current models, we demonstrate that these structural deficits arise secondary to impaired pump function as these phenotypes are rescued when Rbfox is specifically expressed in the myocardium. Mechanistically, we find diminished expression and alternative splicing of sarcomere and mitochondrial components that compromise sarcomere assembly and mitochondrial respiration, respectively. Injection of human RBFOX2 mRNA restores cardiovascular development in rbfox mutant zebrafish, while HLHS-linked RBFOX2 variants fail to rescue. This work supports an emerging paradigm for HLHS pathogenesis that centers on myocardial intrinsic defects.

Using DeepLabCut as a Real-Time and Markerless Tool for Cardiac Physiology Assessment in Zebrafish

Simple Summary

With the advancement of existing technology, artificial intelligence is widely applied in various fields of research, including cardiovascular studies. In this study, we explored the feasibility of conducting a markerless cardiac physiology assessment in zebrafish embryos by using DeepLabCut (DLC), a deep learning tool for motion analysis. Several cardiac parameters, such as heart rate, diastolic–systolic volumes (EDV/ESV), stroke volume, cardiac output, shortening fraction, and ejection fraction were obtained by the DLC-trained model and then compared to the previous published methods, Time Series Analysis and Kymograph. This new method has several advantages, having full automation, precise detection, and real-time labelling. This network was also trained to analyze zebrafish with cardiovascular defects (pericardial edema) induced by chemical treatments with ethanol and ponatinib. It was revealed that the heart rate, EDV/ESV, stroke volume, and cardiac output from both the ethanol and ponatinib groups displayed significant reductions compared with the control. Hopefully, this trained DLC network can contribute to a better understanding and investigation of the existing cardiovascular system and abnormalities.

Abstract

DeepLabCut (DLC) is a deep learning-based tool initially invented for markerless pose estimation in mammals. In this study, we explored the possibility of adopting this tool for conducting markerless cardiac physiology assessment in an important aquatic toxicology model of zebrafish (Danio rerio). Initially, high-definition videography was applied to capture heartbeat information at a frame rate of 30 frames per second (fps). Next, 20 videos from different individuals were used to perform convolutional neural network training by labeling the heart chamber (ventricle) with eight landmarks. Using Residual Network (ResNet) 152, a neural network with 152 convolutional neural network layers with 500,000 iterations, we successfully obtained a trained model that can track the heart chamber in a real-time manner. Later, we validated DLC performance with the previously published ImageJ Time Series Analysis (TSA) and Kymograph (KYM) methods. We also evaluated DLC performance by challenging experimental animals with ethanol and ponatinib to induce cardiac abnormality and heartbeat irregularity. The results showed that DLC is more accurate than the TSA method in several parameters tested. The DLC-trained model also detected the ventricle of zebrafish embryos even in the occurrence of heart abnormalities, such as pericardial edema. We believe that this tool is beneficial for research studies, especially for cardiac physiology assessment in zebrafish embryos.

Zebrafish Models of Cardiac Disease: From Fortuitous Mutants to Precision Medicine

Heart disease is the leading cause of death worldwide. Despite decades of research, most heart pathologies have limited treatments, and often the only curative approach is heart transplantation. Thus, there is an urgent need to develop new therapeutic approaches for treating cardiac diseases. Animal models that reproduce the human pathophysiology are essential to uncovering the biology of diseases and discovering therapies. Traditionally, mammals have been used as models of cardiac disease, but the cost of generating and maintaining new models is exorbitant, and the studies have very low throughput. In the last decade, the zebrafish has emerged as a tractable model for cardiac diseases, owing to several characteristics that made this animal popular among developmental biologists. Zebrafish fertilization and development are external; embryos can be obtained in high numbers, are cheap and easy to maintain, and can be manipulated to create new genetic models. Moreover, zebrafish exhibit an exceptional ability to regenerate their heart after injury. This review summarizes 25 years of research using the zebrafish to study the heart, from the classical forward screenings to the contemporary methods to model mutations found in patients with cardiac disease. We discuss the advantages and limitations of this model organism and introduce the experimental approaches exploited in zebrafish, including forward and reverse genetics and chemical screenings. Last, we review the models used to induce cardiac injury and essential ideas derived from studying natural regeneration. Studies using zebrafish have the potential to accelerate the discovery of new strategies to treat cardiac diseases.

Zebrafish Embryos and Larvae as Alternative Animal Models for Toxicity Testing

Prerequisite to any biological laboratory assay employing living animals is consideration about its necessity, feasibility, ethics and the potential harm caused during an experiment. The imperative of these thoughts has led to the formulation of the 3R-principle, which today is a pivotal scientific standard of animal experimentation worldwide. The rising amount of laboratory investigations utilizing living animals throughout the last decades, either for regulatory concerns or for basic science, demands the development of alternative methods in accordance with 3R to help reduce experiments in mammals. This demand has resulted in investigation of additional vertebrate species displaying favourable biological properties. One prominent species among these is the zebrafish (Danio rerio), as these small laboratory ray-finned fish are well established in science today and feature outstanding biological characteristics. In this review, we highlight the advantages and general prerequisites of zebrafish embryos and larvae before free-feeding stages for toxicological testing, with a particular focus on cardio-, neuro, hepato- and nephrotoxicity. Furthermore, we discuss toxicokinetics, current advances in utilizing zebrafish for organ toxicity testing and highlight how advanced laboratory methods (such as automation, advanced imaging and genetic techniques) can refine future toxicological studies in this species.

Deep learning is widely applicable to phenotyping embryonic development and disease

Genome editing simplifies the generation of new animal models for congenital disorders. However, the detailed and unbiased phenotypic assessment of altered embryonic development remains a challenge. Here, we explore how deep learning (U-Net) can automate segmentation tasks in various imaging modalities, and we quantify phenotypes of altered renal, neural and craniofacial development in Xenopus embryos in comparison with normal variability. We demonstrate the utility of this approach in embryos with polycystic kidneys (pkd1 and pkd2) and craniofacial dysmorphia (six1). We highlight how in toto light-sheet microscopy facilitates accurate reconstruction of brain and craniofacial structures within X. tropicalis embryos upon dyrk1a and six1 loss of function or treatment with retinoic acid inhibitors. These tools increase the sensitivity and throughput of evaluating developmental malformations caused by chemical or genetic disruption. Furthermore, we provide a library of pre-trained networks and detailed instructions for applying deep learning to the reader's own datasets. We demonstrate the versatility, precision and scalability of deep neural network phenotyping on embryonic disease models. By combining light-sheet microscopy and deep learning, we provide a framework for higher-throughput characterization of embryonic model organisms. This article has an associated 'The people behind the papers' interview.

Zebrafish disease models in drug discovery: from preclinical modelling to clinical trials

Numerous drug treatments that have recently entered the clinic or clinical trials have their genesis in zebrafish. Zebrafish are well established for their contribution to developmental biology and have now emerged as a powerful preclinical model for human disease, as their disease characteristics, aetiology and progression, and molecular mechanisms are clinically relevant and highly conserved. Zebrafish respond to small molecules and drug treatments at physiologically relevant dose ranges and, when combined with cell-specific or tissue-specific reporters and gene editing technologies, drug activity can be studied at single-cell resolution within the complexity of a whole animal, across tissues and over an extended timescale. These features enable high-throughput and high-content phenotypic drug screening, repurposing of available drugs for personalized and compassionate use, and even the development of new drug classes. Often, drugs and drug leads explored in zebrafish have an inter-organ mechanism of action and would otherwise not be identified through targeted screening approaches. Here, we discuss how zebrafish is an important model for drug discovery, the process of how these discoveries emerge and future opportunities for maximizing zebrafish potential in medical discoveries.

A hybrid of light-field and light-sheet imaging to study myocardial function and intracardiac blood flow during zebrafish development

Biomechanical forces intimately contribute to cardiac morphogenesis. However, volumetric imaging to investigate the cardiac mechanics with high temporal and spatial resolution remains an imaging challenge. We hereby integrated light-field microscopy (LFM) with light-sheet fluorescence microscopy (LSFM), coupled with a retrospective gating method, to simultaneously access myocardial contraction and intracardiac blood flow at 200 volumes per second. While LSFM allows for the reconstruction of the myocardial function, LFM enables instantaneous acquisition of the intracardiac blood cells traversing across the valves. We further adopted deformable image registration to quantify the ventricular wall displacement and particle tracking velocimetry to monitor intracardiac blood flow. The integration of LFM and LSFM enabled the time-dependent tracking of the individual blood cells and the differential rates of segmental wall displacement during a cardiac cycle. Taken together, we demonstrated a hybrid system, coupled with our image analysis pipeline, to simultaneously capture the myocardial wall motion with intracardiac blood flow during cardiac development.

During the conception of the heart, cardiac muscular contraction and blood flow generate biomechanical forces to influence the functional and structural development. To elucidate the underlying biomechanical mechanisms, we have embraced the zebrafish system for the ease of genetic and pharmacological manipulations and its rapidity for organ development. However, acquiring the dynamic processes (space + time domain) in the small beating zebrafish heart remains a challenge. In the presence of a rapid heartbeat, microscopy is confined by temporal resolution to image the cardiac contraction and blood flow. In this context, we demonstrated an integrated light-sheet and light-field imaging system to visualize cardiac contraction along with the flowing blood cells inside the cardiac chambers. Assuming the periodicity of the cardiac cycle, we synchronized the image data in post-processing for 3-D reconstruction. We further quantified the velocity of the various regions of cardiac muscular contraction, and tracked the individual blood cells during the cardiac cycles. The time-dependent velocity maps allow for uncovering differential segments of cardiac contraction and relaxation, and for revealing the patterns of blood flow. Thus, our integrated light-sheet and light-field imaging system provides an experimental basis to further investigate cardiac function and development.

Automatic Segmentation and Cardiac Mechanics Analysis of Evolving Zebrafish Using Deep Learning

Background: In the study of early cardiac development, it is essential to acquire accurate volume changes of the heart chambers. Although advanced imaging techniques, such as light-sheet fluorescent microscopy (LSFM), provide an accurate procedure for analyzing the heart structure, rapid, and robust segmentation is required to reduce laborious time and accurately quantify developmental cardiac mechanics. Methods: The traditional biomedical analysis involving segmentation of the intracardiac volume occurs manually, presenting bottlenecks due to enormous data volume at high axial resolution. Our advanced deep-learning techniques provide a robust method to segment the volume within a few minutes. Our U-net-based segmentation adopted manually segmented intracardiac volume changes as training data and automatically produced the other LSFM zebrafish cardiac motion images. Results: Three cardiac cycles from 2 to 5 days postfertilization (dpf) were successfully segmented by our U-net-based network providing volume changes over time. In addition to understanding each of the two chambers' cardiac function, the ventricle and atrium were separated by 3D erode morphology methods. Therefore, cardiac mechanical properties were measured rapidly and demonstrated incremental volume changes of both chambers separately. Interestingly, stroke volume (SV) remains similar in the atrium while that of the ventricle increases SV gradually. Conclusion: Our U-net-based segmentation provides a delicate method to segment the intricate inner volume of the zebrafish heart during development, thus providing an accurate, robust, and efficient algorithm to accelerate cardiac research by bypassing the labor-intensive task as well as improving the consistency in the results.

Tissue clearing and imaging methods for cardiovascular development

Tissue imaging in 3D using visible light is limited and various clearing techniques were developed to increase imaging depth, but none provides universal solution for all tissues at all developmental stages. In this review, we focus on different tissue clearing methods for 3D imaging of heart and vasculature, based on chemical composition (solvent-based, simple immersion, hyperhydration, and hydrogel embedding techniques). We discuss in detail compatibility of various tissue clearing techniques with visualization methods: fluorescence preservation, immunohistochemistry, nuclear staining, and fluorescent dyes vascular perfusion. We also discuss myocardium visualization using autofluorescence, tissue shrinking, and expansion. Then we overview imaging methods used to study cardiovascular system and live imaging. We discuss heart and vessels segmentation methods and image analysis. The review covers the whole process of cardiovascular system 3D imaging, starting from tissue clearing and its compatibility with various visualization methods to the types of imaging methods and resulting image analysis.

Mending a broken heart: In vitro, in vivo and in silico models of congenital heart disease

Birth defects contribute to ∼0.3% of global infant mortality in the first month of life, and congenital heart disease (CHD) is the most common birth defect among newborns worldwide. Despite the significant impact on human health, most treatments available for this heterogenous group of disorders are palliative at best. For this reason, the complex process of cardiogenesis, governed by multiple interlinked and dose-dependent pathways, is well investigated. Tissue, animal and, more recently, computerized models of the developing heart have facilitated important discoveries that are helping us to understand the genetic, epigenetic and mechanobiological contributors to CHD aetiology. In this Review, we discuss the strengths and limitations of different models of normal and abnormal cardiogenesis, ranging from single-cell systems and 3D cardiac organoids, to small and large animals and organ-level computational models. These investigative tools have revealed a diversity of pathogenic mechanisms that contribute to CHD, including genetic pathways, epigenetic regulators and shear wall stresses, paving the way for new strategies for screening and non-surgical treatment of CHD. As we discuss in this Review, one of the most-valuable advances in recent years has been the creation of highly personalized platforms with which to study individual diseases in clinically relevant settings.

Deep Learning for Biospectroscopy and Biospectral Imaging: State-of-the-Art and Perspectives

With the advances in instrumentation and sampling techniques, there is an explosive growth of data from molecular and cellular samples. The call to extract more information from the large data sets has greatly challenged the conventional chemometrics method. Deep learning, which utilizes very large data sets for finding hidden features therein and for making accurate predictions for a wide range of applications, has been applied in an unbelievable pace in biospectroscopy and biospectral imaging in the recent 3 years. In this Feature, we first introduce the background and basic knowledge of deep learning. We then focus on the emerging applications of deep learning in the data preprocessing, feature detection, and modeling of the biological samples for spectral analysis and spectroscopic imaging. Finally, we highlight the challenges and limitations in deep learning and the outlook for future directions.

Cardiovascular Performance Measurement in Water Fleas by Utilizing High-Speed Videography and ImageJ Software and Its Application for Pesticide Toxicity Assessment

Water fleas are a good model for ecotoxicity studies, and were proposed for this purpose by the United States Environmental Protection Agency, due to their easy culture, body transparency, and high sensitivity to chemical pollution. Cardiovascular function parameters are usually used as an indicator of toxicity evaluation. However, due to the nature of the heart and blood flow, and the speed of the heartbeat, it is difficult to perform precise heartbeat and blood flow measurements with a low level of bias. In addition, the other cardiovascular parameters, including stroke volume, cardiac output, fractional shortening, and ejection fraction, have seldom been carefully addressed in previous studies. In this paper, high-speed videography and ImageJ-based methods were adopted to analyze cardiovascular function in water fleas. The heartbeat and blood flow for three water flea species, Daphnia magna, Daphnia silimis, and Moina sp., were captured by high-speed videography and analyzed using open-source ImageJ software. We found the heartbeat is species-dependent but not size-dependent in water fleas. Among the three water fleas tested, D. magna was identified as having the most robust heartbeat and blood flow rate, and is therefore suitable for the ecotoxicity test. Moreover, by calculating the diameter of the heart, we succeeded in measuring other cardiovascular parameters. D. magna were challenged with temperature changes and a pesticide (imidacloprid) to analyze variations in its cardiovascular function. We found that the heartbeat of D. magna was temperature-dependent, since the heartbeat was increasing with temperature. A similar result was shown in the cardiac output parameter. We also observed that the heartbeat, cardiac output, and heartbeat regularity are significantly reduced when exposed to imidacloprid at a low dose of 1 ppb (parts per billion). The blood flow rate, stroke volume, ejection fraction, and fractional shortening, on the contrary, did not display significant changes. In conclusion, in this study, we report a simple, highly accurate, and cost-effective method to perform physiological and toxicological assessments in water fleas.

An Overview of Methods for Cardiac Rhythm Detection in Zebrafish

The heart is the most important muscular organ of the cardiovascular system, which pumps blood and circulates, supplying oxygen and nutrients to peripheral tissues. Zebrafish have been widely explored in cardiotoxicity research. For example, the zebrafish embryo has been used as a human heart model due to its body transparency, surviving several days without circulation, and facilitating mutant identification to recapitulate human diseases. On the other hand, adult zebrafish can exhibit the amazing regenerative heart muscle capacity, while adult mammalian hearts lack this potential. This review paper offers a brief description of the major methodologies used to detect zebrafish cardiac rhythm at both embryonic and adult stages. The dynamic pixel change method was mostly performed for the embryonic stage. Other techniques, such as kymography, laser confocal microscopy, artificial intelligence, and electrocardiography (ECG) have also been applied to study heartbeat in zebrafish embryos. Nevertheless, ECG is widely used for heartbeat detection in adult zebrafish since ECG waveforms' similarity between zebrafish and humans is prominent. High-frequency ultrasound imaging (echocardiography) and modern electronic sensor tag also have been proposed. Despite the fact that each method has its benefits and limitations, it is proved that zebrafish have become a promising animal model for human cardiovascular disease, drug pharmaceutical, and toxicological research. Using those tools, we conclude that zebrafish behaviors as an excellent small animal model to perform real-time monitoring for the developmental heart process with transparent body appearance, to conduct the in vivo cardiovascular performance and gene function assays, as well as to perform high-throughput/high content drug screening.

First person – Alexander Akerberg

First Person is a series of interviews with the first authors of a selection of papers published in Disease Models & Mechanisms (DMM), helping early-career researchers promote themselves alongside their papers. Alexander Akerberg is first author on ‘ Deep learning enables automated volumetric assessments of cardiac function in zebrafish’, published in DMM. Alexander is a postdoctoral research fellow in the lab of Caroline E. Burns and C. Geoffrey Burns at Boston Children's Hospital, MA, USA, where he enjoys building new tools while investigating the molecular mechanisms that govern cardiac form and function.