Development of a multiciliated cell

Abstract

Multiciliated cells (MCC) are evolutionary conserved, highly specialized cell types that contain dozens to hundreds of motile cilia that they use to propel fluid directionally. To template these cilia, each MCC produces between 30 and 500 basal bodies via a process termed centriole amplification. Much progress has been made in recent years in understanding the pathways involved in MCC fate determination, differentiation, and ciliogenesis. Recent studies using mammalian cell culture systems, mice, Xenopus, and other model organisms have started to uncover the mechanisms involved in centriole and cilia biogenesis. Yet, how MCC progenitor cells regulate the precise number of centrioles and cilia during their differentiation remains largely unknown. In this review, we will examine recent findings that address this fundamental question.

Introduction

Multiciliated cells are highly specialized epithelia that contain tens to hundreds of motile cilia on their apical surface (Figure 1a [1,2]). The cilia beat in a polarized and coordinated fashion to generate directional fluid flow over the surface of the cell and associated tissues. In humans, MCC are found in the respiratory tract, brain ventricles and segments of the male and female reproductive organs. In the airway, motile cilia of MCC beat directionally to propel mucus and inhaled contaminants out of the lungs. Ciliary motility of ependymal and choroid plexus MCC is key for movement of cerebrospinal fluid through the central nervous system, while MCC cilia in the reproductive tract play important roles in ovum and sperm transport [3,4]. Mutations that disrupt MCC differentiation, cilia assembly or motility can cause diseases which include primary ciliary dyskinesia, hydrocephalus and infertility, among others. Importantly, defects in the formation of the correct number of centrioles and cilia per cell, referred to as “reduced generation of multiple motile cilia” (RGMC), can also lead to these human pathological conditions [5–9]. Thus, the proper assembly of motile cilia is critical for the functions of MCC, and the overall health of the organism.

Figure 1. Variability in centriole and cilia abundance of multiciliated cells.

(a) Scanning Electron Microscopy images of mouse respiratory (trachea) and ependymal (brain ventricle) MCC. (b) Fluorescent microscopy images of cultured mouse airway basal progenitor cells and mature airway MCC (left and center), and en face image of ventricular ependymal MCC from mouse brain. (c) Schematic of the proliferation and differentiation steps of multiciliated epithelia. The key stages of centriole amplification from parental centrioles and deuterosomes (dark spheres) are depicted. (d) Quantification of centriole number and cell surface area of mouse airway MCC in vitro and in vivo., highlighting a linear relationship between the two. Data reproduced and modified from Ref. [27].

The development of a MCC from a progenitor stem cell is a process referred to as multiciliogenesis, and involves several key steps. First, it requires the activation of a unique transcriptional program that specifies cell fate and expression of genes essential for multiciliogenesis [10–12]. Next comes the formation of centrioles, barrel-shaped microtubule structures that act as the base upon which cilia are built. Finally, cilia are assembled onto the apically docked centrioles (also called basal bodies, BB), and begin beating in a polarized directional manner. Several excellent reviews have covered various aspects of MCC biology in depthd–their evolutionary conservation, fate specification, transcriptional program, cytoskeletal organization, and mechanisms regulating ciliary assembly and motility [1–4,10,11]. Here, we will focus on recent studies regarding the process of centriole biogenesis and the establishment of cilia abundance, a fundamental aspect of MCC biology that has long remained a mystery.

Centriole biogenesis

Like the majority of cell types in mammals, MCC progenitors contain a centrosome with a pair of centrioles (hereafter referred to as the parental centrioles; PC) and a solitary non-motile primary cilium (Figure 1b–c [13,14]). As the progenitors proliferate, cells maintain this centrosome-cilium number via the canonical centriole duplication pathway that is tightly linked to the cell cycle [15]. During their differentiation MCC become post-mitotic and, depending on the cell type, must then produce anywhere between ~ 30 and 500 additional centrioles that will mature into BBs and template the assembly of an equal number of motile cilia (Figure 1b). This process is termed centriole amplification and historically thought to occur through two parallel pathways: a centriole-dependent pathway whereby the original PC template the assembly of roughly 2–8 procentrioles each; and an acentriolar pathway mediated by dozens of deuterosomes which are spheroidal electron-dense structures that can nucleate tens of procentrioles each (Figure 1c [13,16–18]). Deuterosomes are transient structures that are absent in proliferating progenitor cells, start to form during early centriole amplification, and mostly disappear after centriologenesis is complete. Although both PC-dependent and deuterosome-dependent pathways have been observed since the late 1960s [13], the molecular mechanisms that govern the two pathways and their relative contributions to the total complement of centrioles has mostly remained enigmatic.

Many of the key proteins involved in canonical centriole duplication play an essential role in both PC- and deuterosome-dependent centriole amplification pathways in MCC [14,19,20]. This indicates that centriole duplication in cycling cells and centriole amplification in MCC share common and conserved molecular mechanisms despite their morphological differences. However, recent studies using in vitro and in vivo models of human, mouse and frog MCC have identified proteins that are uniquely associated with deuterosomes. For example, Deup1 was recently characterized as a core deuterosome-specific protein that is essential for deuterosome biogenesis [21]. Several other deuterosome-enriched factors have since been identified and shown to play key roles in deuterosome formation and/or function [22–25]. These data support the theory that the PC dependent and deuterosome-dependent centriole amplification pathways act in parallel but may use unique molecular components to initiate the assembly of centrioles in a MCC. It has been estimated that the deuterosome-dependent pathway contributes the majority (80%–90%) of the total centrioles assembled during amplification, while the PC-dependent pathway contributes roughly 10%–20% to-wards the final number [3,13,21,26]. Yet it remains unclear how the final number of centrioles is established, and how such a wide range in number exists even within the same cell type in a tissue. For example, ependymal MCC typically average between 30 and 90 BBs and cilia per cell, a roughly 3-fold variation between cells [26]. Similarly, airway MCC can contain ~100–500 BBs and cilia, meaning neighboring cells can have up to a 5-fold variation in the amounts of each organelle (Figure 1b and d, [27]). How do progenitor cells that start with exactly two parental centrioles generate such a variable number following differentiation?

Role of templating structures

Since procentriole assembly appears to require both parental centrioles and deuterosomes, one obvious mode of regulation in controlling the abundance of BBs in a mature MCC are these nucleating structures. This raises the question regarding the origins of deuterosomes themselves, which have long been thought to form de novo. Intriguingly, live-cell imaging of differentiating mouse ependymal MCC showed that deuterosomes are nucleated from the original PC, suggesting that these two pathways may not be independent of each other after all [26]. This would indicate that all centrioles produced in MCC are ultimately dependent on the PC. However, using a combination of chemical and genetic methods to manipulate PC formation, multiple studies have now demonstrated that formation of deuterosomes can occur independently of the PC [27–29]. Knockdown and pharmacological approaches used to inhibit centriole duplication in dividing progenitor cells of both airway and ependymal MCC, rendering them devoid of PC prior to differentiation, did not block deuterosome biogenesis. In fact, cells formed slightly more deuterosomes on average and these were capable of nucleating procentrioles normally [27–29]. Importantly, loss of the PC did not alter the final complement of BBs produced by those MCC, suggesting that perhaps deuterosomes compensate for the loss of PC-dependent centriole biogenesis to establish the correct number per cell.

The next major question was whether deuterosomes are essential for centriole amplification and abundance. Inhibiting deuterosome biogenesis using gene knockdown approaches in mouse airway and Xenopus embryonic MCC disrupted centriole formation and number [21], suggesting that deuterosomes are indeed essential for this process. Yet surprisingly, a recently developed mouse model harboring a null allele of Deup1, which results in MCC that lack deuterosomes, showed that cells were able to produce the normal complement of BBs [30]. Deup1 mutant MCC underwent the normal stepwise assembly stages of centriole amplification with small variations; the parental centrioles nucleated more procentrioles each (which formed along their length) compared to wild-type cells, and there was an increase in procentriole nucleation adjacent to the PC [30]. This suggests that the PC may compensate for the loss of deuterosome-mediated centriole biogenesis. The authors tested this hypothesis by simultaneously eliminating both PC and deuterosomes, and discovered that clusters of procentrioles formed de novo within a cloud of fibrogranular and pericentriolar materials in the cytoplasm. Once again, there was no significant difference in the abundance of BBs produced by the cell [30], suggesting that BB number is not likely a function of these nucleating structures.

Overall, these studies highlight the robust nature of the centriole amplification program and the resilience of mammalian MCC in forming the correct number of BBs and cilia, even in the absence of both PC and deuterosomes. It also establishes the existence of a third route for centriole formation, namely de novo biogenesis that can function independently of the other two pathways and compensate for their loss. This is reminiscent of several multiciliated cell types that do not require one, or both, of these pathways to produce multiple BBs. For example, MCC in the pronephric ducts and olfactory placode of zebrafish lack deuterosomes yet undergo centriole amplification to produce ~10–20 BBs and cilia [31,32]. Similarly, mammalian olfactory sensory neurons amplify their centrioles in a deuterosome-independent manner to generate an average of roughly 16 BBs and cilia per cell [33,34]. Moreover, the freshwater planarian Schmidtea mediterranea lacks centrosomes and deuterosomes yet undergoes de novo centriole amplification to produce all its BBs [35]. So, if the templating structures are not the predominant factor that establishes BB abundance in mammalian MCC, then what instructs the differentiating progenitor cell to form the correct complement during centriole amplification?

Cell size and centriole number

Scaling of organelle size and abundance relative to cell volume is a biologically conserved phenomenon, and been previously shown to regulate intrinsic properties of both centrosomes and cilia [36–40]. Recent studies have shed further light on this relationship in multiciliated cells. Nanjundappa et al. [27] identified a direct relationship between centriole number and surface area in airway MCC, as larger cells appear to contain more centrioles compared to smaller neighboring cells (Figure 1b, d). This correlation was observed both in vitro and in vivo. To determine whether cell size–dnamely surface area or volume–was the major determinant of centriole number, the authors altered the shape of progenitor cells (by modulating the extracellular matrix) prior to their differentiation and induction of centriole amplification. The manipulation resulted in mature MCC that displayed a larger surface area, but were shallower in depth, resulting in a similar overall cell volume. Quantification of centriole number showed a proportional increase with the larger apical surface area, but not volume, of fully mature cells [27]. Thus, surface area size appears to dictate centriole abundance during amplification in airway MCC, and suggests the presence of molecular pathways that link the two cellular compartments.

Similarly, analysis of centriole amplification during radial intercalation of MCC progenitors in developing Xenopus embryos showed a direct correlation between centriole number and the apical surface area [41]. In these cells, expansion of the cell surface area during apical intercalation occurs in conjunction with the centriole amplification process. Increasing or decreasing surface area size using genetic or biomechanical approaches resulted in a corresponding elevation or reduction in centriole abundance, respectively. Alternatively, modulation of centriole number during amplification caused a relative change in apical surface area, highlighting the presence of feedback mechanisms that regulate the two processes. The authors speculated that mechanosensory proteins that can sense apical surface stretching may play a role, and found that the cation channel Peizo1 was localized adjacent to centrioles at the apical membrane. Depletion of Piezo1 uncoupled the relationship between apical surface area and centriole abundance; it caused a decrease in centriole number but a corresponding increase in surface area [41]. One possibility is that mechanical stretch can activate Piezo1, leading to an influx of Ca²⁺ from the extracellular environment, which may modulate centriole amplification via effects on gene transcription, cytoskeletal organization, or other pathways. Altogether, these studies indicate the presence of molecular mechanisms that help to calibrate centriole and cilia abundance in relation to apical surface area during the development of an MCC.

How are the variations in cell surface area during differentiation communicated to the centriole amplification pathway to establish centriole number? There are several possible ways we envision this could occur. First, larger cells might increase transcription or translation of genes essential for centriole and cilia assembly. This would be analogous to the ‘limiting component’ model of organelle abundance [36–40], where a fixed quantity of a precursor protein(s) would be expressed then “used up” as centriole assembly occurs. In this scenario, the number of centrioles assembled would stop once the limiting component is no longer available. Nanjundappa et al. did note an increase in deuterosome number in cells with enlarged surface area [27], suggesting that transcriptional/translational output of at least some precursors is likely elevated. In addition, some of the transcription factors involved in the multiciliogenesis program, such as E2F4, have been shown to shuttle between the nucleus and the cytoplasm where it interacts with proteins essential for deuterosome biogenesis and centriole amplification [24]. This provides a mode of communication between the centriole amplification machinery and the transcriptional program of MCC, potentially fine-tuning the amount of precursor protein that is expressed in each cell. The transcription coactivator YAP is expressed in airway MCC and regulates ciliogenesis, several YAP interactors localize to BBs, and YAP nuclear translocation occurs during tissue regeneration of the multiciliated epithelium [42–45]. This is another example of a pathway that could integrate signals from various cellular compartments to transcriptionally influence MCC development and centriologenesis.

A second possibility is that cells with larger surface areas might spend longer periods of time in stages of centriole assembly compared to smaller cells. In this scenario, the rate of centriole assembly would be the same in cells of different size, but ones with larger surface areas would spend a longer time in stages of procentriole amplification and growth to generate more total centrioles. In support of this model, recent studies have shown that differentiating, non-dividing MCC repurpose the mitotic regulatory circuitry involving CDK1, CDK2, PLK1, and APC-C to control timely progression of centriole amplification, maturation, and motile ciliogenesis while avoiding reentry into mitosis [46,47]. Other mitotic regulatory proteins, such as CDC20B, were recently shown to influence this process by localizing to deuterosomes and regulating the release of centrioles during disengagement [25]. Essentially, these cell cycle regulatory proteins control how long the cell spends in each stage of centriole biogenesis: centriole nucleation, elongation, disengagement, migration, and docking. Hence, these temporal regulators of centriole formation in MCC may help relay information about cell size to the centriole assembly machinery, and act as internal autonomous clocks that dynamically modulate the length of time spent in assembly to achieve the desired final number of BBs, as has been proposed for other organelles and cellular structures [48].

Finally, it is possible that components of the MCC cytoskeleton directly or indirectly influence centriole and cilia abundance. It is well established that a network of apical and subapical actin in MCC play critical roles in modifying surface area size, regulating centriole migration and docking at the plasma membrane, ensuring the even distribution of centrioles across the cell surface, orienting (planar polarization) of mature BBs, and conferring mechanical resistance to sustain the shear stress imposed by ciliary beating [49–55]. What remains unclear is whether this actin network influences the centriole amplification machinery to regulate BB number, and how this might occur. Post-translationally modified microtubules at the apical surface of MCC connect BBs to the cell cortex and play important roles in their spacing and orientation [56]. Similarly, an apical network of intermediate filaments is found in close proximity to the BBs [57], although their precise functions here are unknown. A recent study in Xenopus discovered that, during radial intercalation, centrioles accumulate high levels of acetylated microtubules that are apically oriented [58]. Cells with more centrioles showed higher accumulation of these microtubules, and underwent more rapid apical insertion, thus providing evidence for cross-talk between centrioles, the cytoskeleton and cell surface dynamics. Yet, it remains unclear whether any of these cytoskeletal components regulate the abundance of BBs that are formed during centriole amplification.

Conclusions

Rapid advances in the field of multiciliated cell biology have shed light on the pathways involved in controlling their development across different tissues and organisms. One fundamental question that remains unresolved is how the variation in centriole–cilia abundance between cells is established. Even in the same organism, for example humans, different MCC contain vastly different numbers of centrioles and cilia, and can display large variability within the same cell type. This indicates that, although the principles that govern MCC specification and differentiation are generally conserved, there are bound to be cell-intrinsic differences at the molecular level. Recent studies have defined surface areadependent processes can establish centriole abundance in certain MCC. In addition, roles for mitotic regulators, transcription factors, cytoskeletal proteins and mechanosensory channels have also been identified. Yet how these pathways work together to modulate centriole abundance remains an open question. As is often the case in biology, it is likely a combination of some or all of these factors. Understanding this scaling phenomenon will be a fascinating area of research with both fundamental and clinical implications.