Our work not only addressed how primordium migration is regulated but also how this migration is coupled to organ morphogenesis and size control. Elucidating how transient or stable multicellular neuromasts are organized into epithelial rosettes is not only important for lateral line morphogenesis but understanding neuromast rosette formation is also relevant for understanding neural stem cell niches, vertebrate pancreas development, Drosophila axis elongation and ommatidia development, just to name a few. However, little is known about the molecular mechanisms through which signaling pathways induce cytoskeletal changes leading to cell shape changes and organ size control. We discovered that Notch signaling cell-autonomously determines the number of cells that contribute to a rosette by upregulating cell adhesion and tight junction proteins independent of proliferation (Fig. 4, left panel, Kozlowskaja-Gumbriene et al., 2017). Prior to our study, organ size control in all tested model systems had been exclusively linked to cell proliferation (eg. via the Hippo pathway).
In summary, we have discovered a feedback loop between Wnt, Fgf and Notch signaling that coordinates all aspects of lateral line development: collective, directed cell migration, cell type specification and organ size control and morphogenesis. This has generated a deep level of knowledge about the lateral line that makes it one of the best characterized vertebrate organ systems with respect to morphogenesis. This initial characterization of the neuromast gene regulatory network lays down a robust foundation for a systematic cellular and molecular dissection of vertebrate morphogenesis not only by my laboratory, but also by our colleagues in the field.
Establishment of hair cell polarity
Hair cells are functionally polarized different hair cells perceive water movement coming from different directions. Therefore, the proper establishment of the planar cell polarity of hair cells during development and regeneration is essential. Loss of planar cell polarity proteins, such as vangl2 lead to randomization of hair cell polarity. We discovered that in wnt11r, glypican4 and fzd7a/7b mutants, hair cells show different defects. We are currently investigating what cellular processes are affected in these mutants, as they have not been previously linked to the establishment of cell polarity.
Stem cell control during development
The cell-cell interactions that orchestrate progenitor quiescence, activation and differentiation are not well understood but dynamic interactions between progenitors and specialized niche environments play key roles in regulating the properties of progenitor pools. Therefore, understanding niche-progenitor interactions at the cellular level is crucial for building a general understanding of this process.
Morphologically, two domains can be distinguished in the primordium: the leading region, which is unpatterned, and the posterior two thirds that are organized into 2-3 rosettes for future deposition (Fig. 4 right). Previous studies suggested that the leading Wnt-positive region in the primordium is a stem cell zone. However, it is not known whether stem cells are also present in the trailing region or whether primordium stem cells are related through lineage to the deposited stem cells of the neuromast and interneuromast cells. Because stem and amplifying progenitor cells are crucial for the development of sensory organs and the regeneration of organs in adult animals, we sought to trace the lineage of embryonic and adult stem/progenitor cells, and to test if the behavior of adult and embryonic stem cells is controlled by the same gene regulatory network.
To address these questions, we established a formal collaboration with Ajay Chitnis’ group at the NIH to fate map and track primordium cells (Nogare et al., 2016). Our studies revealed that the leading region of the primordium does not contain stem cells that asymmetrically divide and self-renew, but rather that progenitor cells exist throughout the primordium (Fig. 5).
Second, we determined where in the primordium interneuromast stem cells originate, as we had previously shown that interneuromast cells are stem cells and give rise to postembryonic neuromasts after the primordium has deposited the initial set (Grant et al., 2005). We mapped their location in the primordium (Nogare et al., 2016), Fig. 6A,B), and showed that interneuromast cells and mantle cells originate exclusively from the edges of the primordium, suggesting that they are either induced there or that they are unresponsive to distal differentiating signals secreted from central cells in the primordium. Having identified the location of pro-interneuromast stem cells in the primordium now enables us to search for the signaling molecules/niche that induce/maintain these stem cells.
We already have made significant progress in determining how interneuromast cell behavior is controlled once they are deposited. We demonstrated that neighboring glial cells induce quiescence and inhibit interneuromast progenitor cells from proliferating. Consequently, mutations in the Erbb pathway that cause a loss of Schwann cells, lead to the precocious proliferation and differentiation of postembryonic neuromasts (Grant et al., 2005, Lush and Piotrowski, 2014, Perlin et al., 2011). We discovered that Erbb signaling is not only involved in instructing Schwann cell migration along the lateral line nerve, but that Erbb signaling in Schwann cells also non cell-autonomously inhibits mitogenic Wnt/b-catenin and Fgf signaling in interneuromast cells (Fig. 7). Importantly, our studies showed that postembryonic neuromast development is in many aspects similar but not identical to embryonic neuromast development in the primordium. This provides novel mechanistic insight by showing how Erbb signaling controls neural progenitor quiescence via the inhibition of Wnt/b-catenin signaling.
These findings illustrate the intricate manner in which diverse signaling pathways coordinate distinct aspects of the niche-progenitor interaction needed to maintain the proper balance and timing of this dynamic cell population. As the signaling interactions that control postembryonic neural progenitor quiescence and activation are not well understood in developing nervous tissue, our studies may have relevance in other contexts.
Our research program aims to dissect complex morphogenetic and regenerative processes both in vivo and at high resolution. During embryonic development cells acquire fate and organize themselves into distinct and intricate three-dimensional organs or tissues. In order for proper morphogenesis to occur, a variety of cellular behaviors have to be tightly coordinated (e.g., cell migration, cell fate specification, cell-cell adhesion, cell proliferation, interactions with the environment, cell shape changes). Such complexity has hindered the elucidation of the molecular and cell biological mechanisms that regulate embryonic morphogenesis. Even less understood are the mechanisms that allow reintegration of regenerating cells into a pre-existing, mature tissue.
We have identified the sensory lateral line of the zebrafish as a relatively simple system that has allowed us to begin to shed light onto the molecular and cellular basis of complex vertebrate morphogenesis. The lateral line is an ideal organ to mechanistically dissect embryonic and post-embryonic morphogenetic pro¬cesses because of 1) the accessibility of the sensory organs to direct observation and manipulation; 2) the relative simplicity of the lateral line system; 3) the similarity between lateral line hair cells and inner ear hair cells; 4) their ability to regenerate; and 5) the genetic tools available in zebrafish to molecularly dissect lateral line development.
We have developed tools to investigate the functional consequences of pathway and gene manipulations on cell behavior at unprec¬edented levels of detail (Nogare et al., 2016, Venero Galanternik et al., 2016b, Venero Galanternik et al., 2016c). We have systematically applied these new tools to define the fundamental signaling interactions that control both lateral line development and regeneration. As such, we have begun to identify similarities and differences in gene regulation between embryonic development and regeneration. By genetically testing these differences, we aim to understand the permissive factors that allow regeneration in an adult vertebrate, a fundamental aspect of regenerative biology and medicine that has yet to be satisfactorily addressed.
Zebrafish Sensory Lateral Line
The lateral line is a sensory system for the detection of water movements, which aids the animals with capturing prey, avoiding predators and schooling. The lateral line consists of mechanosensory organs (neuromasts) distributed in lines on the head and along the flanks of the animal. These neuromasts contain hair cells that are very similar to the hair cells of the ear and vestibular system of mammals, such as ourselves. Despite the unusual location of the hair cells in the skin, lateral line and ear hair cells develop by similar mechanisms and are derived from cephalic placodes. However, in contrast to the otic placode, the lateral line placode (also called primordium) undergoes a remarkable posterior migration towards the tail tip. This migration is a dynamic event that involves the primordium periodically depositing neuromasts until it reaches the tail tip, patterning the future lateral line. We have discovered that the primordium also deposits a trail of interneuromast cells in between these neuromasts which represent stem cells that give rise to postembryonic neuromasts (Grant et al., Neuron 2005, Lush et al., eLife 2014). The complex cellular signaling interactions resulting in the integration of stem cell regulation, migration and hair cell regeneration remain a mystery and form a central focus for our research.
Understanding how neuromasts develop within the primordium is not only essential to mechanistically grasp their morphogenesis, but also necessary to reveal the genetic similarities and differences underpinning neuromast regeneration. Such formal comparison between embryonic and postembryonic development may ultimately help define key components that make vertebrate tissues and organs regeneration competent.
IDENTIFICATION OF THE GENE REGULATORY NETWORK AFFECTING FUNDAMENTAL DEVELOPMENTAL AND MORPHOGENETIC MECHANISMS
Signaling interactions that coordinate migration with organ morphogenesis
Cell migration is a fundamental, tightly coordinated developmental and organ morphogenetic process (Aman and Piotrowski, 2010). Much has been learned from in vitro studies about how individual cells migrate, yet the mechanisms integrating migration and morphogenesis of groups of cells in vivo are among the least understood processes in developmental biology. We have identified essential cell-cell feedback interactions between cells in the leading and trailing zones of a migrating cell cluster (Fig.2; (Aman and Piotrowski, 2008). Inhibitory interactions between the Wnt/β-catenin and Fgf signaling pathways serve to restrict activation of these pathways into mutually exclusive domains. The network is based on localized activation of the Wnt/β-catenin pathway in the leading zone of the primordium, which induces and restricts Fgf signaling to trailing cells. Loss of this polarity causes disruption of chemokine signaling and random tumbling of cells leading to primordium stalling. In addition, Fgf signaling is crucial for sensory organ development. Therefore, by inducing Fgf signaling in trailing cells and inhibiting this pathway in the leading cells, Wnt signaling restricts maturation and deposition of sensory organs to the trailing portion of the primordium.
Since then we have mechanistically expanded the components of the Fgf/Wnt feedback loop and the precise orchestration and integration of cell-autonomous mechanisms and environmental signals that regulate collective cell migration with organ morphogenesis in significant ways. For example, we have made a breakthrough in our understanding of how the distribution of signaling molecules across the primordium and their ability to activate signal transduction within individual cells is regulated by extracellular matrix molecules (Fig. 3; (Venero Galanternik et al., 2015, Venero Galanternik et al., 2016a).
II. IDENTIFICATION OF THE GENE REGULATORY NETWORK AFFECTING CELL BEHAVIOR DURING HOMEOSTASIS AND REGENERATION
The overarching goal of this part of our research is to gain a detailed understanding of the gene regulatory network that controls hair cell regeneration in the zebrafish lateral line. We will use this knowledge to determine how gene regulatory networks evolved between zebrafish, chicken and mice. Combined these studies will not only reveal how hair cells regenerate in a non-mammalian vertebrate but will also be very informative for the future interpretation and design of pathway manipulations aimed at inducing regeneration in the mouse.
As mammals age, inner ear hair cells progressively lose hair cells which leads to deafness. In contrast, non-mammalian vertebrates such as fish, birds and reptiles efficiently turn over hair cells during homeostasis and regenerate their sensory hair cells after injury throughout life. Even though hair cell development is fairly well understood and the cell biology of hair cell regeneration has been investigated in chicks and amphibians, the lack of genetics and non-synchronous hair cell death in these systems has precluded greater understanding of the underlying molecular processes. Our research of hair cell regeneration in mature neuromasts has established the lateral line as a unique model system to study adult hair cell regeneration at the single cell level in vivo. In addition, the ability to manipulate multiple signaling pathways and monitor their impact on cell identities and behaviors in vivo make the lateral line system an ideal model for the elucidation and functional characterization of the complex cross-talk between signaling networks. Such an approach is not possible in other regenerating species, such as the chick or in neonatal mice that show very limited regeneration.
In order to exploit the advantages of the lateral line, we are pursuing a three-pronged approach to define and functionally characterize the molecular and cellular interactions occurring during hair cell regeneration:
1. Cell biology of stem/progenitor cells: We developed a powerful assay that allows the functional interrogation of gene interactions that regulate the balance of progenitor self-renewal and differentiation at the single cell level (Romero-Carvajal et al., 2015, Venero Galanternik et al., 2016b). Using transgenic animals, we mapped the topography of stem/progenitor cells and identified all possible cell lineages in homeostatic and regenerating neuromasts by combining both long-term in vivo time lapse analyses and manual tracking of all dividing cells with cell fate analyses (Fig. 1N-Q).
We discovered that support cells exclusively self-renew in the dorso-ventral poles, are quiescent in the anterior-posterior poles and differentiate into hair cells only in the center of the sensory organs (Fig. 2A-B’’). Our bulk RNA-Seq analyses and in situ screen of regenerating neuromasts revealed that the Wnt and Notch pathways respond to hair cell death and that the Wnt and Notch pathways are expressed in the poles and central neuromast compartments, respectively. Functional analyses of these pathways showed that downregulation of Notch signaling is required for Wnt induced proliferation of support cells and hair cell differentiation during regeneration (Fig. 2C-I). Altogether, our findings demonstrate that the majority of support cells are competent to respond to injury and that localized activation of different signaling pathways balances stem cell maintenance with differentiation, thus ensuring the lifelong ability of fish to regenerate the lateral line sensory organ. The importance of these discoveries is that they define not only the key signaling pathways modulated by homeostasis and regeneration, but definitively identify the distribution, competency, and potency dynamics of the cellular agents (i.e., stem cells) effecting these activities. It is relevant to note that recent studies in the mouse have shown that the signaling pathway interactions we have identified in the lateral line are also operating in the mouse inner ear sensory epithelium (Li et al., 2015), indicating evolutionary conservation and supporting our premise that our studies will significantly inform the mammalian condition.
We also discovered that peripheral mantle support cells are quiescent stem cells that do not respond to hair cell death but are mobilized upon severe loss of support cells. Such quiescent or reserve stem cells have been identified in many other regenerating tissues, such as the blood, intestine and hair follicles. Therefore, our work indicates that the lateral line sensory organs are not only a powerful model to study different stem cell populations and stem cell behavior during hair cell regeneration, but also serve as a powerful paradigm to inform stem cell biology dynamics in other vertebrate organs. We and our colleagues in the hair cell regeneration field are currently building upon this knowledge to determine how other signaling pathways fit into this network as these pathway interactions are conserved between zebrafish and mice.
Our work established a basic network between Wnt/Fgf and Notch signaling that is fundamentally important for maintaining the balance of progenitor cell self-renewal and differentiation and that ensures the life-long ability to regenerate hair cells. We are currently building upon this knowledge to determine how other signaling pathways/genes that are modulated during regeneration fit into this network (Jiang et al., 2014).
2. Expression Analyses
2a. Bulk RNA-Seq analysis of regenerating neuromasts: To identify genes and pathways involved in triggering and executing hair cell regeneration, we successfully carried out RNA-Seq analyses on FACS isolated support cells at several time points after hair cell death. Rather than focusing on the most highly up- and downregulated genes to identify novel pathways, we characterized the behavior of signaling pathways known to be crucial for hair cell development. The rationale behind our approach was that many of these pathways will be likely re-deployed during regeneration. For example, the manipulation of the Notch and Wnt/β-catenin pathways during mouse inner ear development causes extra hair cell formation. Unfortunately, after injury the manipulation of Wnt and Notch do not lead to the restoration of a fully functional epithelium. One of the reasons for the limited success of these studies is that the precise timing of when these pathways are activated or inhibited, and how these pathways interact during regeneration remains unknown. In contrast to hair cell regeneration in chicken in which hair cell death can take several hours, zebrafish hair cells die within minutes and regeneration is relatively synchronous. This provided a unique opportunity to generate a high resolution and detailed analysis of the temporal progress of events during regeneration. We determined the activation status of signaling pathways at 1, 3 and 5h post hair cell death, and discovered that the Notch and Fgf pathways are downregulated immediately, but that Notch signaling is re-activated between 3-5h to limit hair cell differentiation and maintain self-renewing progenitors (Fig. 2J,K). Even though in many other organ systems Wnt/β-catenin signaling is described as being crucial for regeneration, our studies show that Wnt/β-catenin signaling is activated relatively late and is not involved in triggering the regenerative response. This dataset allowed us to build a dynamic model of pathway interactions that are currently being functionally tested. Elucidating the dynamic interaction network between pathways will also aid in devising experimental approaches to trigger successful regeneration in mammals.
Cross-species comparison to identify divergence or similarities during the evolution of gene regulatory networks underlying hair cell development and regeneration in zebrafish, chicken and mice.
Tatjana is a founding member of the Hearing Regeneration Consortium under the auspice of the Hearing Health foundation (http://hearinghealthfoundation.org/hearing_restoration_project). The Consortium consists of 14 members who work on mouse, chick and zebrafish hair cell regeneration. The aim of the consortium is to make significant progress in our understanding of hair cell regeneration and to eventually trigger regeneration in mammalian ears. This collaboration provides us with access to chick and mouse RNA-Seq data, allowing us to study the evolution of gene regulatory networks.
2b. Single cell RNA-Seq of homeostatic and regenerating neuromasts (unpublished): The in situ gene expression analysis of candidate genes and cell behaviors showed that support cells are heterogeneous. Bulk RNA-Seq analyses do not allow the study of cell populations in isolation and the measured gene expression levels are an average across all cell types (Jiang et al., 2014, PNAS). This averaging also prevents the elucidation of the transcriptional dynamics of a temporal process such as cell differentiation. In collaboration with the Flow cytometry, the Molecular biology and Computational biology cores, we have successfully established single cell RNA-Seq of sorted lateral line cells. We have successfully performed gene expression analyses at the single support cell level during homeostasis to determine a) how many support cell populations exist and where they are located in a homeostatic neuromast, b) what signaling pathways are co-activated in a given cell and which signals a cell is producing, c) the transcriptional dynamics of the process of differentiation and progenitor maintenance during regeneration and d) how experimental manipulation of signaling pathways affects gene expression and cell identities. We will be able to distinguish cells on different trajectories, temporally order them along those trajectories, and identify new regulatory factors controlling differentiation. Single cell expression profiling will also reveal if pathways act cell-autonomously or non-cell-autonomously. Importantly, the data will be informative for the interpretation and design of strategies to induce regeneration in the mouse.
3. Mutagenesis screen to identify genes affecting hair cell regeneration (unpublished): Forward mutagenesis is a complementary and powerful approach to identify genes involved in hair cell regeneration. The power of the hair cell regeneration screen lies in its simplicity and efficiency. The screen is performed on the same embryos being screened for developmental defects. We have identified mutations in 26 genes that perturb regeneration (hair cell regeneration: 4 genes; hair cell degeneration w/o neomycin: 6 genes, 12 alleles); development (extra hair cells: 2 genes, 4 alleles; fewer hair cells: 3 genes -one is atoh1a-, 3 alleles; primordium migration defects: 4 genes, 9 alleles; neuromast deposition defects: 7 genes, 10 alleles).
The mutational analysis will identify genes that are potentially not components of the candidate pathways that we are already testing. In addition, the mutants are a powerful tools to test the function of genes in directing cell behaviors and how they interact with other pathways, such as Wnt and Notch. We are presently cloning at least one allele per gene by RNA-Seq analysis and are investigating if mutations cause regeneration-specific defects or whether they affect global developmental processes, such as proliferation, cell death or cell migration.
Summary: The results of our 3-pronged approach coupled with our detailed understanding hair cell development allowed us to gain significant insight into our understanding of the molecular and cellular basis of hair cell regeneration. This knowledge serves as a baseline, not only for all of our future experiments but also the experiments of our colleagues in the lateral line field. Importantly, our findings will also aid to determine why mammals fail to regenerate as they age.
a) Comparison of scRNA-Seq data of zebrafish, chick and mouse hair and support cells (in collaboration with Stefan Heller, Stanford University). Using homeostatic tissues we will a) determine if the two regenerating species (zebrafish and chicken) possess similar support cell populations based on their expression profiles and b) if in mammals these populations differentiate during ageing. A subsequent aim will be to compare the transcriptional responses of zebrafish and chicken support cells to hair cell death to determine how conserved the molecular mechanisms of hair cell regeneration are.
b) Identify enhancers by ATAC-Seq, H3K27ac ChIP-Seq and RNA-Seq during zebrafish development and regeneration.
Enhancers are instrumental in determining a gene’s tissue-specific and developmental stage-specific expression. To expand upon our extensive characterization of the gene regulatory networks impacting lateral line development and regeneration based on expression profiling and mutant analyses, we plan to identify some of the critical cis-regulatory regions involved in modulating the relevant expression profiles. As a first step, we will use ATAC-Seq, which monitors the relative openness or accessibility for chromatin states, in combination with epigenetic marks for enhancers (e.g. H3K27ac, H3K4me1) or co-activators such as p300. These approaches require relatively few cells and should allow us to monitor the dynamics of changes in cis-regions correlated with alterations in the expression profiles of specific cell populations during development and regeneration. As neonatal mouse sensory epithelia show some regenerative ability that is lost postnatally, it is very likely that either the chromatin accessibility changes at a global level or at particular enhancers are repressed or fail to be activated. To shed light on these questions it will be useful to begin to characterize the enhancers employed in regenerating species, such as the zebrafish.
Homeostasis & Regeneration