What we can learn from a tadpole about ciliopathies and airway diseases – Using systems biology in Xenopus to study cilia and mucociliary epithelia

The left-right organizer (LRO)

The LRO is a mono-ciliated epithelium that generates asymmetric fluid flow and induces the asymmetry of the left-right body axis (Blum et al., 2014; Schweickert et al., 2007). This epithelium is the gastrocoel roof-plate (GRP) in amphibians (Fig.1 A) and some basal fish (e.g. sturgeon), and is homologous to the embryonic node in mice and rabbits, or the Kupffer’s vesicle in teleost fish (Blum et al., 2007). The GRP forms one day post fertilization (LRO in the mouse is found at E8.0), is relatively large (approximately 3-4 fold larger than the mouse node), and can be easily dissected and used for live cell imaging of fluid flow or cilia movement in culture, or fixed and stained to investigate gene expression, cilia formation and morphology (Blum et al., 2009; Boskovski et al., 2013). Additionally, the ability to target gain- and loss-of-function manipulations on either the left, the right or both sides, which is unique among established vertebrate model organisms, has proved to be a powerful tool to analyze the molecular functions of components and mechanisms of left-right axis specification (Schweickert et al., 2010; Vick et al., 2009; Walentek et al., 2012).

The neural tube and the brain

As in other vertebrates, the developing neural tube in Xenopus produces cells that harbor a primary cilium (Fig.1 B). These primary cilia on neural progenitor cells are required to mediate cell signaling (mainly Hedgehog signaling), which patterns the neural tube (Dessaud et al., 2008; Wong and Reiter, 2008). The analysis of Hedgehog-dependent neural gene expression (e.g. nkx2.2 or pax6) by whole mount in situ hybridization offers a quick and easy assay to test if primary cilia formation or function is disturbed after manipulation (e.g. Chung et al., 2012; Walentek et al., 2016). In cases where neural signaling defects are recorded, primary cilia formation can be further analyzed using immunofluorescence on tissue sections (Fig.1 B), and again, the ability to target microinjections provides a useful tool for experimentation, i.e. the possibility to compare one manipulated side of the neural tube to a control injected side within the same individual.

Xenopus is also suitable to examine formation and function of motile cilia of the central nervous system, i.e. motile mono-cilia of the floor plate of the spinal cord, as well as motile mono-ciliated cells and MCCs of the brain (Schweickert and Feistel, 2015; Walentek et al., 2015a). The large size of the tadpole brain and its ventricles makes them amenable to manipulation and analysis of cerebrospinal fluid (CSF) flow, e.g. fluorescent microbeads can be injected directly into the large 4^th ventricle and their movement can be imaged and analyzed in vivo (Hagenlocher et al., 2013). Through this technique, patterns of CSF were recently visualized in the developing Xenopus brain, which revealed a previously underappreciated complexity as well as the contribution of different cilia beating modes in driving dorsal versus ventral CSF flow (Hagenlocher et al., 2013).

The mucociliary embryonic epidermis

Most prominently, mucociliary epithelia line the mammalian conducting airways and the epidermis of many vertebrate and invertebrate embryos (Fig.1 C) (Bartsch et al., 1997; Ganesan et al., 2013; Kemp, 1996; Verdugo, 1982; Webb and Northcutt, 1991). Although these tissues are derived from different germ layers (the ectoderm and endoderm), and hence cannot be considered strictly homologous, the mucociliary epithelia they form are remarkably similar in morphology, cell type composition and function. Mucociliary epithelia are pseudostratified columnar epithelia, characterized by the presence of secretory cell types that release mucus (goblet-like cells) or antimicrobial substances and factors that influence the behavior of the other cell types of the epithelium (e.g. Club and Pulmonary neuroendocrine cells in the airways, and small secretory cells in the Xenopus mucociliary epidermis) (Hayes et al., 2007; Quigley et al., 2011; Rock et al., 2010; Walentek et al., 2014a). Additionally, MCCs generate fluid flow and propagate distribution of released substances (Brooks and Wallingford, 2014; Todd, 1836). In the airways, these epithelia clear the respiratory tract of harmful substances and microorganisms to prevent damage and infection, while epidermal mucociliary epithelia prevent colonization and infection of the embryo by bacteria, as demonstrated by the production of factors with anti-microbial properties and functional assays (Dubaissi et al., 2014; Hayes et al., 2007). The mucociliary epithelium also likely contributes to proper oxygenation of the embryo by transporting fresh water along the epidermis, the major tissue for gas exchange prior to the development of gills and lungs. The mucociliary epidermis evolved much earlier than the lung or trachea, suggesting that it represents the first mucociliary epithelium in the vertebrates, which was likely co-opted and modified in the conducting airways or the reproductive system later on in evolution. Therefore, the Xenopus mucociliary epidermis could be regarded as a “prototypic” mucociliary epithelium, and while likely somewhat diverged from more derived organ epithelia, may be used to reveal the principal mechanisms of mucociliary development, regeneration and function (Hayes et al., 2007; Werner and Mitchell, 2012).

The use of the Xenopus mucociliary epidermis has several major advantages as compared to the more commonly used models for mucociliary research, i.e. the mouse trachea and human airway epithelial cells (HAECs), which can be differentiated in vitro (Fulcher et al., 2005; Rock et al., 2009): (1) The mucociliary epidermis develops rapidly into a mature mucociliary epithelium within only two days (Fig.3 B), while the airway epithelium in the mouse reaches maturity 1-2 weeks after birth, and the differentiation of HAECs requires a protocol of 3-4 weeks. (2) The exterior localization of the mucociliary epidermis allows in vivo manipulations (e.g. drug treatments) and analysis by live-cell imaging. Over the past years, advanced imaging techniques have been developed for use in the Xenopus epidermis that take advantage of the large size of Xenopus MCCs, as well as the ease of visualizing specific components using mRNA injections of fluorescently tagged constructs, e.g. for cilia, basal body and polarity-complex proteins (Fig.1 B,C) (Brooks and Wallingford, 2015; Werner and Mitchell, 2013). Furthermore, simple imaging methods were developed for the quantitative and qualitative analysis of extracellular fluid flow, which can be used for rapid functional screening and prediction of MCC cilia phenotypes (Walentek et al., 2014a; Walentek et al., 2015b; Walentek et al., 2016). (3) Most importantly, the molecular mechanisms, transcription factors, and expression profiles found in Xenopus MCCs share a deep conservation with mammalian airway MCCs (Quigley and Kintner, 2016; Treutlein et al., 2014). Therefore, the Xenopus mucociliary epidermis is an emerging model for airway mucociliary research with outstanding potential to facilitate novel approaches and important novel findings in ciliary and mucociliary biology.

Additional sites of ciliation in Xenopus embryos and tadpoles

Although the three tissues described above are most frequently employed for cilia research in Xenopus, additional tissues form ciliated cells in embryos and tadpoles, and are being actively used or could be used for basic and biomedical research (Schweickert and Feistel, 2015). Among these, the embryonic pronephric kidney is a popular model to study the principles of kidney development and function (Lienkamp et al., 2010; Wessely and Tran, 2011; Zhou and Vize, 2004). In Xenopus, the nephrostomes contain MCCs that transport fluid into the proximal tubules of the nephron, which is lined with cells presenting a primary cilium into the lumen, as in the mammalian metanephric kidney, where they sense fluid flow (Berbari et al., 2009; Schweickert and Feistel, 2015). MCCs can be also found at the tadpole nasal pits, where they drive fluid flow towards the embryo, which could help with sampling of the environment. As aquatic vertebrates, frogs also develop a lateral line organ to sense fluid motion around them. This is mediated by neuromasts, which use a primary cilium for mechanosensation of water currents (Elkan and Murray, 1951; Schweickert and Feistel, 2015). Similar hair cells can also be found in the developing inner ear (Chung et al., 2014a; Schweickert and Feistel, 2015). Frog tadpoles also develop a ciliated gut, i.e. the esophagus and the stomach are both covered by MCCs, which mediate fluid transport from anterior to posterior, suggesting that MCCs could be employed in early feeding or transport of bacteria to colonize the intestine (Chalmers and Slack, 1998; Walentek et al., 2015a). Interestingly, the conducting airways of adult Xenopus are not densely ciliated like in the mammalian airways (our unpublished observation), further supporting the idea that the mucociliary epidermis represents the more ancestral mucociliary tissue in vertebrates.

Transcriptional and post-transcriptional regulation of MCCs

The transcriptional control of MCCs relies on the step-wise activation of multiple layers of transcription factors, which collectively regulate cell type specification, massive centriole amplification and cilia formation (Fig.2 A). Efforts over the past years have led to an exceptionally deep understanding of the gene regulatory network that underlies these processes. Data generated in Xenopus revealed that upon Notch signaling inhibition in MCC progenitors, a trimeric complex of Multicilin (MCIDAS), E2F4 and TfDp1 is activated, which initiates the expression of ciliary transcription factors, e.g. foxj1 and rfx2 (Ma et al., 2014; Stubbs et al., 2012). The Multicilin complex, together with ciliary transcription factors, regulate a highly conserved set of core MCC genes (~950) required for vertebrate multi-ciliogenesis (Quigley and Kintner, 2016). There is a surprising degree of positive regulation between ciliary transcription factors, leading to multiple layers of amplification and potential redundancy (Fig.2 A), which likely confer robustness to cell fate choice and ciliary gene expression. Ciliary transcription factors also cooperate extensively to activate core MCC genes, which might lead to more precise and robust gene regulation (Chung et al., 2012; Chung et al., 2014b; Ma et al., 2014; Nemajerova et al., 2016; Pan et al., 2014; Quigley and Kintner, 2016; Stubbs et al., 2008; Stubbs et al., 2012; Takahashi et al., 2015; Tan et al., 2013; Zhou et al., 2015). This interpretation is supported by data on Myb and Foxn4, which regulate centriole amplification and activation of motile cilia genes in MCCs, respectively. Inactivation of either Myb (in mouse) or Foxn4 (in mouse and frog) leads to MCC ciliation defects at stages where MCCs are first formed, but these MCCs recover later during development and are eventually able to form normal cilia (Tan et al., 2013; Campell et al., 2016). This suggests that Myb and Foxn4 support timely and robust gene expression, which is mainly achieved by the master transcription factors for MCCs (the Multicilin complex) and motile ciliogenesis (Foxj1 and RFX2). Tp73, another transcription factor used in MCC ciliogenesis, is upstream of RFX2/3 and Foxj1, but does not seem to bind to the promoters of the vast majority of MCC genes (Marshall et al., 2016; Nemajerova et al., 2016). By contrast, Foxj1 and RFX2 bind a majority of MCC gene promoters, although Foxj1 in particular may do so indirectly (Quigley and Kintner, 2016). Recent work shows that Foxj1-binding DNA motifs are preferentially found at distal regulatory elements/enhancers, but rarely at core MCC gene promoters, while RFX2 sites or RFX motifs allowing for other RFX protein binding are found at both enhancers and promoters. Interaction of Foxj1 with core MCC gene promoters largely depends on the presence of RFX2, which forms homo- and hetero-dimers with other RFXs. These data hint that RFX2 or other RFX proteins may be required for Foxj1-mediated motile cilia gene expression by facilitating chromatin looping and interaction of enhancer-bound Foxj1 with specific promoters (Quigley and Kintner, 2016). It is tempting to speculate that other RFX factors, implicated in many other developmental and disease contexts, could fulfill essentially the same function of maneuvering distal transcriptional machinery to promoters of target genes (Choksi et al., 2014; Walentek, 2016). Taken together, these data suggest that some upstream combination of E2f4, TfDp1, and Multicilin, along with Tp73, regulates Foxj1 and RFX2. In turn, Foxj1 and RFX proteins perform the heavy transcriptional lifting, with minor assistance from Myb and Foxn4, of the hundreds of genes required to make motile cilia in MCCs.

In addition to transcriptional control, the MCC gene regulatory network also employs post-transcriptional mechanisms for gene regulation (Fig.2 B). The miR-34/449 family of microRNAs (miRs) is highly expressed in MCCs, required for ciliogenesis in human, mouse, zebrafish and Xenopus MCCs, and dysregulated in various airway diseases (Graff et al., 2012; Lizé et al., 2010; Marcet et al., 2011; Song et al., 2014; Walentek et al., 2014b; Wang et al., 2013b). Studies in Xenopus and mouse have shown that miR-34/449 are directly regulated by ciliary transcription factors (Nemajerova et al., 2016; Walentek et al., 2016), and required for successful ciliogenesis in MCCs through negative regulation of multiple targets, including Notch signaling components, actin regulators and the centriolar protein Cp110 (Cibois et al., 2015; Marcet et al., 2011; Song et al., 2014; Walentek et al., 2016). The case of Cp110 level regulation by miR-34/449 is particularly interesting, as it demonstrates the requirement for precise titration of protein levels during ciliogenesis. Cp110 is able to promote or to inhibit cilia formation in MCCs in a dose-dependent manner, and too little or too much Cp110 equally inhibits cilia formation, though through different mechanisms (Fig.2 B) (Walentek et al., 2016). To achieve robust activation of expression in MCCs, cp110 is directly activated by ciliary transcription factors (the Multicilin complex, Foxj1 and RFX2). To prevent excess accumulation of Cp110, five of the six miR-34/449 family members are also directly controlled by ciliary transcription factors (the Multicilin complex, Tp73, Foxj1 and RFX2) at two distinct genomic loci, and their high-level expression represses cp110 to achieve optimal Cp110 protein levels during ciliogenesis stages (Walentek et al., 2016). These data exemplify how miRNAs are employed within a gene regulatory network, and reveal a versatile mechanism of protein level control in MCCs.

These examples demonstrate how valuable Xenopus is for detection and functional testing of ciliary genes, and how systems biology approaches in Xenopus can be successfully used to gain insight into the gene regulatory networks that regulate cell fate specification and differentiation. It is likely that mechanisms similar to the ones described for MCCs are employed in other cell types and are dysregulated in diseases, e.g. cancer (Walentek, 2016; Walentek et al., 2014b). Future cilia research in Xenopus will therefore not only reveal conserved mechanisms in cilia formation, but also shed light on general principles of gene regulation and cell function in development and disease.

Signaling in mucociliary cell type specification and morphogenesis

The activation of transcriptional networks, such as the one described for MCCs (Fig.2 A), is achieved through signaling control by cell signaling cascades, e.g. Notch, BMP and Wnt. Knowing how cell signaling controls mucociliary development and regeneration is imperative to understand the changes that occur in chronic lung diseases, which are often characterized by tissue remodeling (i.e. changes in cell type composition) or impaired differentiation (e.g. in dysplasia) (Beers and Morrisey, 2011; Ganesan and Sajjan, 2013; Pain et al., 2014). A major challenge in understanding the signaling control of complex tissue formation and morphogenesis is the reiterative and pleiotropic nature of cell signaling pathways, which makes it hard to distinguish between primary and secondary effects on cell types or a given process. In contrast to the airway mucociliary epithelium, which develops late and requires many developmental steps to occur correctly before the epithelium is formed, the Xenopus mucociliary epidermis is specified in a relative simple process very early on in development (the epidermal ectoderm is specified by about 12h post fertilization in the animal hemisphere of the embryo; Fig.3 A) (Heasman, 1997; Herriges and Morrisey, 2014). Therefore, the mucociliary epidermis allows investigation of mucociliary development directly, without the prerequisite of successful lung formation. Furthermore, the cell type composition of the Xenopus mucociliary epidermis is fully resolved, while it remains uncertain how many rare cell types exist in the mammalian airway epithelium (Brooks and Wallingford, 2014; Crystal et al., 2008; Hayes et al., 2007; Quigley et al., 2011; Walentek et al., 2014a). In addition, the master transcription factors and transcriptional profiles for all epithelial cell types are either published or currently being resolved. These data facilitate analysis of the effects of signaling manipulations on cell type specific gene regulatory networks over developmental time, which will allow us to dissect the different functions of cell signaling in mucociliary epithelia. This understanding will likely provide key insights into mammalian airway epithelial development and disease mechanisms, as roles of major signaling pathways seem to be conserved between the mucociliary epidermis and airway cells.

Notch signaling negatively regulates MCC cell fate, while promoting secretory cell formation (Fig.3 B). This role of Notch signaling was found early on in the Xenopus mucociliary epidermis (Deblandre et al., 1999), and holds true for the mouse airways and human airway stem cells, indicating a deeply conserved function (Gomi et al., 2015; Tsao et al., 2009). Additionally, Notch signaling is also active during a second wave of signaling to subdivide progenitors into more distinct cell lineages, which was shown for Xenopus ion secreting cells and airways cell of the mouse (Fig.3 B) (Morimoto et al., 2012; Quigley et al., 2011; Rock et al., 2011).

A recent study has also compared the effects of BMP signaling on Xenopus and human mucociliary cells, and revealed a conserved requirement for BMP inhibition in the specification of MCCs and other cells derived from subepithelial progenitor cells (Fig.3 B) (Cibois et al., 2015). In later stages of development, BMP signaling is also required for intercalation of differentiating cells into the epithelium as well as for the expression of goblet cell markers (Fig.3 B), providing another case for reiterative signaling in addition to Notch (Cibois et al., 2015).

Lastly, Wnt/β-catenin signaling is implicated in several processes in mucociliary development, regeneration and disease, but its precise roles are not completely understood (De Langhe and Reynolds, 2008; Pongracz and Stockley, 2006). Experiments in the Xenopus mucociliary epidermis indicate that virtually all cell types require Wnt/β-catenin signaling at some point during specification and differentiation (Fig.3 B). Specifically, foxj1 expression in MCCs requires Wnt/β-catenin signaling in Xenopus and zebrafish, where it was shown to be directly regulated by β-catenin (Caron et al., 2012; Walentek et al., 2012; Walentek et al., 2015b). Tp63 is also directly regulated by Wnt/β-catenin signaling in Xenopus and humans, and is an important regulator of stem/progenitor cell proliferation and differentiation (Kjolby and Harland, 2016; Ruptier et al., 2011). Small secretory cells and goblet-like cells also require Wnt/β-catenin signaling for expression of foxa1 (which induces small secretory cells; Walentek et al., 2015b) and normal secretion (goblet-like cells; our unpublished results). Similar to the situation in Xenopus, Wnt/β-catenin signaling was implicated in mucus secretion (Mucenski et al., 2005; Okubo and Hogan, 2004), basal stem cell proliferation and differentiation (Brechbuhl et al., 2011; Ghosh et al., 2013; Hsu et al., 2014), and possibly FOXA1 expression (Lamouille et al., 2014; Wang et al., 2013a) in mice and human cells. Collectively, these findings suggest that a conserved Wnt/β-catenin dependent gene regulatory network exists in mucociliary epithelia, and contributes to the regulation of cell fate specification and specific differentiation processes in multiple cell types. It is therefore important to resolve the evolutionarily conserved functions of Wnt/β-catenin signaling in mucociliary epithelia, as well as the functions of other signaling pathways.

These examples all reinforce the argument that the Xenopus mucociliary epidermis can serve as a powerful model to study systems-wide effects of mucociliary signaling, which will help to dissect the reiterative and pleiotropic roles of signaling in airway mucociliary regeneration and disease.

In summary, Xenopus embryos provide an outstanding model for the investigation of cilia and ciliated epithelia, and the combination of classic advantages of the frog with new technologies for systems biology approaches opens new avenues to address important unresolved questions in the field. These features are further complemented by excellent resources for Xenopus research, including genomes, databases, national and international stock centers, and transgenic lines, which are provided through a highly collaborative and dedicated community of researchers (e.g. Grant et al., 2015; Harland and Grainger, 2011; Hellsten et al., 2010; Karpinka et al., 2014; Session et al., 2016). We look forward to more studies that use Xenopus systems biology to address important questions in cilia and mucociliary biology, which will help us to understand how gene regulatory networks function, how complex tissue formation is achieved, and how ciliopathies and airway diseases could be cured in the future.