Notch Signaling Mediates Olfactory Multiciliated Cell Specification

Abstract

Epithelial multiciliated cells (MCCs) use motile cilia to direct external fluid flow, the disruption of which is associated with human diseases in a broad array of organs such as those in the respiratory, reproductive, and renal systems. While many of the signaling pathways that regulate MCC formation in these organ systems have been identified, similar characterization of MCC differentiation in the developing olfactory system has been lacking. Here, using live cell tracking, targeted cell ablation, and temporally-specific inhibition of the Notch signaling pathway, we identify the earliest time window of zebrafish olfactory MCC (OMCC) differentiation and demonstrate these cells’ derivation from peridermal cells. We also describe regionally segregated Notch signaling across time points of rapid OMCC differentiation and show that Notch signaling downregulation yields an increase in OMCCs, suggesting that OMCC fate is normally repressed in a region-specific manner during olfactory development. Finally, we describe Notch signaling’s regulation of the differentiation/ciliogenesis-associated genes foxj1a and foxj1b. Taken together, these findings provide new insights into the origins and developmental programming of OMCCs in vivo.

Graphical Abstract

1. Introduction

Multiciliated cells (MCCs) are specialized epithelial cells with motile cilia that are typically found on the apical side of their resident epithelium (Brooks and Wallingford, 2014; Spassky and Meunier, 2017). MCCs’ roles are wide-ranging across organ systems such as the airway epithelia, kidneys, and outer epidermis, and defects in MCC development and/or function have been associated with ciliopathies including primary ciliary dyskinesia, hydrocephalus, and subfertility (Mitchison and Valente, 2017; Praveen et al., 2015; Wallmeier et al., 2020). Additionally, in zebrafish, the olfactory epithelium (OE), responsible for the sense of smell, harbors its own unique MCC population that guides water flow past sensory neurons to facilitate odor detection (Reiten et al., 2017). These olfactory MCCs (OMCCs) offer an accessible in vivo model to visualize the dynamics of MCC development and maturation.

Previous work on OMCCs has focused predominantly on the complex process of ciliogenesis and the subsequent function of cilia (Pathak et al., 2007; Reiten et al., 2017), with only a handful of genes implicated in the earlier process of cell differentiation (Chong et al., 2018). In other organ systems, meanwhile, the Notch signaling pathway has been shown to play a role in MCC differentiation (Liu et al., 2007; Rock et al., 2011; Tasca et al., 2021), but a potential role for Notch signaling in OMCC differentiation has not been explored. Furthermore, the cellular origins and spatiotemporal mechanics of OMCC specification and differentiation remain entirely uncharacterized.

In vertebrates, the outermost epithelial tissue consists of a squamous layer, the periderm, that is typically present during early development before being eventually replaced by basal keratinocyte-like cells (Kimmel et al., 1990; Lee et al., 2014; Richardson et al., 2014). Recently, intriguing work demonstrated that in the zebrafish pharynx, peridermal cells invade the pharyngeal cavity (Teixeira Rosa et al., 2019) and are essential in guiding formation of the tooth placode (Oralova et al., 2020). The periderm is similarly found in close proximity to the developing OE, leaving open the possibility of a peridermal contribution to olfactory development.

Here, we applied time-lapse imaging, targeted cell ablation, and temporally-specific signaling inhibition techniques to uncover the spatial and cellular origins and developmental progression of zebrafish OMCCs. First, we identified peridermal cells differentiating into OMCCs and ablated the periderm to show that it is essential for OMCC differentiation. We also tracked Notch signaling output, identified its sustained suppression in OMCCs, and, upon temporally-specific knockdown of the Notch signaling pathway, observed a region-specific expansion of OMCCs consistent with an observed alteration in the expression patterns of the ciliogenic genes foxj1a and foxj1b. Taken together, we have identified the developmental time window for OMCC differentiation, characterized OMCCs’ origins and developmental progression, and suggested a regionally-localized role for Notch signaling in repressing OMCC cell fate until the appropriate time for differentiation.

2. Materials and Methods

2.1. Zebrafish

Animals were treated and cared for in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory animals. All experiments were approved by the University of Illinois at Chicago Institutional Animal Care and Use Committee. Embryos were grown, staged, and harvested as previously described (Kimmel et al., 1995; Westerfield, 2000). Embryos at time points older than 24 hours post-fertilization (hpf) were treated with 1-phenyl-2-thiourea to prevent pigmentation from interfering with imaging. Transgenic lines used and their abbreviations in this manuscript are: Tg(OMP2k:lyn-mRFP)/rw035 (Sato et al., 2005) = OMP:RFP; Tg(Bactin:HRAS-EGFP) (Cooper et al., 2005) = actb2:GFP; Tg(z1044-Gal4; UAS:NTR-mCherry) (Eisenhoffer et al., 2017; Pisharath and Parsons, 2009) = GET-Periderm; Tg(Tp1bglob:eGFP) (Parsons et al., 2009) = Tp1:EGFP; Tg(hsp70l:DN-MAML-GFP) (Zhao et al., 2014) = HS:dnMAML.

2.2. Immunostaining

Embryos were collected, fixed with 4% PFA, washed in PBST (1% BSA, 1% DMSO, 0.1% Triton X-100 in PBS, pH 7.4), blocked in 10% normal donkey or goat serum/PBDT, and incubated overnight at 4°C with a primary antibody targeting acetylated α-Tubulin (acTub; 1:1000; Sigma-Aldrich, T7451). Further PBST washes and blocking were followed by secondary antibodies (Invitrogen) overnight at 4°C. DAPI was added to stain nuclei. After further PBDT and PBS washes, embryos/larvae were mounted to image the left olfactory epithelium in each sample. Confocal microscopy was performed using a Zeiss LSM 800 microscope with a 40x/1.1 W objective.

2.3. In situ Hybridization Chain Reaction (HCR)

Zebrafish embryos and larvae were fixed using 4% (w/v) PFA in DEPC-treated PBS. Hybridization chain reaction (HCR) probes were designed to detect foxj1a and foxj1b mRNA, and HCR was performed as previously described (Choi et al., 2018). Samples were counterstained using DAPI. Probes, fluorescent hairpins, and buffers were purchased from Molecular Instruments (www.molecularinstruments.com).

2.4. Targeted Cell Ablation

To ablate periderm cells, Tg(z1044A-Gal4; UAS:NTR-mCherry) embryos were immersed in 15 mM Metronidazole (MTZ) solution in egg water with 1.5% DMSO as done previously (Zhu et al., 2019) at 44–54 hpf, followed by fixation at 54 hpf and immunostaining. Control embryos were immersed in egg water with 1.5% DMSO.

2.5. Notch Signaling Inhibition

Notch inhibition studies were performed using N-[N-(3,5-Difluorophenacetyl-L-alanyl)]-S-phenylglycine t-Butyl Ester (DAPT, Sigma # 565770; chemical inhibitor of γ-secretase) or the transgenic zebrafish line Tg(hsp70l:DN-MAML-GFP) (dominant negative mastermind-like 1). For chemical knockdown, zebrafish clutchmates were incubated from 48 hpf to 72 hpf in 100μM DAPT + 1% DMSO in egg water or 1% DMSO only in egg water, followed by fixation at 72 hpf, or the embryos were then washed with egg water and fixed at 96 hpf. For genetic knockdown, heat shock was induced at 48 hpf by incubating Tg(hsp70l:DN-MAML-GFP) heterozygotes or wild-type clutchmates at 37°C for 30’, followed by 5’ at 40°C. Embryos were then sorted for the presence or absence of global dnMAML–GFP fluorescence before fixation at 72 hpf.

2.6. Imaging and Quantitative Analysis

Embryos were prepared and mounted for imaging as previously described (Rajan et al., 2018; Saxena et al., 2013). Confocal time-lapse microscopy was performed on a Zeiss LSM 800 microscope with a 40x/1.1 W objective. Collected data were processed and analyzed using Imaris (Bitplane, Inc.) and exported as PNG images or MP4 movies. For surface analysis, a region of interest comprising the acTub signal or HCR signal was selected and Imaris’s ‘Surface’ extension was used for automated signal detection and thresholding across control and experimental samples to compute signal volume and area. For GET-Periderm cell number quantitation, 5.5 µm diameter spots were placed with uniform thresholds to automatically detect GET-Periderm+ cells. Calculations to determine statistical significance were performed using Student’s t-test of variables (two-sample t-test assuming unequal variances). Graphs and figure legends include n values for experiments.

3. Results

3.1. Timeline of olfactory multiciliated cell (OMCC) development

OMCCs have been previously characterized in zebrafish larvae at 4 days post-fertilization (dpf), when they are present around the rim of the olfactory pit (Reiten et al., 2017). To determine when and where OMCCs first appear, we performed time-lapse imaging starting on 1 dpf with transgenic zebrafish lines. At 41 hpf, we found that the line OMP:RFP, known to label ciliated olfactory sensory neurons at earlier time points (Sato et al., 2005), now also labeled cells along the lateral edge of the OE (Fig. 1A–B) that had wider and more oblong cell bodies in comparison to nearby neurons (Fig. 1C, asterisk and arrowheads, respectively). Over several hours, these cells gradually took on a cuboidal shape typical of OMCCs, as shown in Fig. 1C–E (asterisk), and antibody staining for acetylated α-Tubulin (acTub) at 48 hpf detected acTub+; OMP:RFP+ cilia that had rapidly formed on the laterally-located cells’ apical surfaces (Fig. 1F–F’’, bracket in F’). The number of cilia increased substantially by 72 hpf (Fig. 1G–G’), and OMCCs with motile cilia could be clearly identified along the apicolateral edge of the OE (Supplementary Video 1).

Fig. 1.

Timeline of olfactory multiciliated cell (OMCC) development. (A) Schematic of zebrafish embryo showing the olfactory epithelium (OE) and olfactory bulb (OB); insets demonstrate the increasing numbers of OMCCs adjacent to olfactory sensory neurons (OSNs) at 41–72 hpf. (B) 15 µm thick anterolateral view of a representative OMP:RFP+ embryo at 41 hours post-fertilization (hpf). (C-E) Magnified images of embryo in (B, box) demonstrating the differentiation of an OMCC (asterisk) on the lateral side of the OE at 41 hpf (C), 43 hpf (D), and 45 hpf (E). Arrowheads in (C) indicate OSNs. (F-G’’) 3D-projection anterolateral views of a representative OMP:RFP+ embryo with immunofluorescence staining for acetylated α-Tubulin (acTub) at 48 hpf (F-F’’) and 72 hpf (G-G’’). Bracket in (F’) indicates acTub signal at the apicolateral edge of the OE. (F’’,G’’) show 3 µm thick optical slices demonstrating colocalization of acTub with OMP:RFP+ cilia. OMP:RFP: magenta; acTub: cyan; DAPI: grey. Orientation arrows: L: Lateral; M: Medial; D: dorsal; V: ventral. Scale bars: 10 µm.

3.2. Contribution of periderm to OMCCs

Given the proximity of peridermal cells to the olfactory pit, we performed time-lapse imaging of the Gal4 enhancer trap line GET-Periderm (Eisenhoffer et al., 2017; Franco et al., 2019; Teixeira Rosa et al., 2019), which mosaically labels a large subset of Krt4+ periderm cells by driving the expression of UAS:NTR-mCherry. At 48–53 hpf, we found mCherry+ GET-Periderm cells along the OE’s apicolateral edge that differentiated into cuboidal OMCCs exhibiting multiple cilia (Fig. 2A–F, arrowhead). We took advantage of the UAS:NTR-mCherry system to add metronidazole (MTZ), as previously described (Pisharath and Parsons, 2009; Zhu et al., 2019), and induce spatiotemporally-specific apoptosis in peridermal cells. We initiated MTZ treatment at 44 hpf, 3 hours after the first OMCCs were observed (Fig. 1B), and quantitated the resulting GET-Periderm cell ablation at 54 hpf (Fig. 2I). The decrease in GET-Periderm cell number yielded an absence of periderm-derived OMCCs and a drastic reduction in acTub signal in comparison to control clutchmates (Fig. 2G,G’,H,H’; Supplementary Videos 2 & 3); some acTub staining remained, perhaps due to the GET-Periderm line’s known mosaicism (Eisenhoffer et al., 2017). We quantitated acTub signal volume and found a statistically significant decrease (Fig. 2G’’,H’’,J), suggesting that periderm ablation results in reduced OMCC differentiation. These findings indicate that during the earliest window of OMCC development, peridermal cells are an essential source of OMCCs.

Fig. 2.

Contribution of periderm to OMCCs. (A-F) 1.93 µm thick anterior views of a representative actb:GFP+; GET-Periderm+ embryo at 48 hpf (A), 49 hpf (B), 50 hpf (C), 51 hpf (D), 52 hpf (E), and 53 hpf (F) showing the appearance of multiple cilia (F, arrowhead) on the apical side of peridermal cells. (G-H’’) 3D projection anterior views of representative GET-periderm+ embryos at 54 hpf after treatment with DMSO (negative control; G-G’’) or 15mM MTZ/DMSO (cell ablation; H-H’’) at 44–54 hpf and subsequent acTub immunostaining. (G’’) and (H’’) show surfaces constructed around acTub signal to compute volume using Imaris’s ‘Surface Analysis’. (I) Number of GET-Periderm+ cells for DMSO- or MTZ-treated (44–54 hpf) embryos at 54 hpf. (J) acTub staining volume for DMSO- or MTZ-treated (44–54 hpf) embryos at 54 hpf. Horizontal bars: Mean +/− 95% confidence intervals. Statistical significance: **p<0.01, ****p<0.0001. n-values indicate number of embryos assayed. actb:GFP: magenta; GET-periderm: cyan; acTub: magenta; DAPI: grey. Orientation arrows: L: Lateral; M: Medial; D: dorsal; V: ventral. Scale bars: (A-F): 10 µm; (G-H’’): 15 µm.

3.3. Effect of Notch signaling on OMCC differentiation

The Notch signaling pathway plays key roles in the development and maturation of MCCs in many epithelial tissues (Liu et al., 2007; Rock et al., 2011; Tasca et al., 2021). To determine if Notch signaling might similarly play a role in OMCC differentiation, we tracked expression of the Notch signaling reporter Tp1:EGFP, which labels cells with active and/or recent Notch signaling (Ninov et al., 2012; Parsons et al., 2009). OMCCs (arrowheads) in Tp1:EGFP+; OMP:RFP+ embryos at 46, 56, and 66 hpf had nearly undetectable levels of Notch signaling in comparison to other cell types in the OE (Fig. 3A–C’). Hypothesizing that lower levels of Notch signaling may be necessary for MCC specification, we treated Tp1:EGFP+ zebrafish embryos with the γ-secretase inhibitor DAPT (Liu et al., 2007) to inhibit Notch signaling at 48–72 hpf, soon after the first OMCCs appear and while rapid OMCC differentiation is underway. At 72 hpf, Notch signaling-inhibited embryos exhibited a decrease in Tp1:EGFP levels (Fig. 3D–E), a robust, complementary increase in acTub staining (Fig. 3D–F), and an increased number of OMP:RFP+ OMCCs (Fig. S1A–B) in comparison to control clutchmates. Post-treatment recovery until 96 hpf yielded similar results (Fig. S2). To confirm the DAPT treatment-induced, temporally-specific results genetically, we used the heat shock-inducible transgenic zebrafish line HS:dnMAML (Zhao et al., 2014), which inhibits Notch signaling via a dominant negative form of the downstream transcriptional coactivator Mastermind like-1 (MAML1) (Zhao et al., 2014). Heat shock of heterozygous HS:dnMAML embryos and their wildtype clutchmates at 48 hpf followed by analysis at 72 hpf recapitulated the DAPT treatment results (Fig. 3G–I). Of note, the medial aspect of the olfactory pit, where Tp1:EGFP+ cells are normally present (Fig. 3D), was populated by OMCCs in the absence of Notch signaling (Fig. 3E). These data suggest that the suppression of Notch signaling in precursor cells drives OMCC differentiation in vivo.

Fig. 3.

Notch signaling inhibition increases OMCC differentiation. (A-C’) 7.6 µm thick anterior views of representative Tp1:EGFP+; OMP:RFP+ embryos at 46 hpf (A,A’), 56 hpf (B,B’), and 66 hpf (C,C’) showing the absence of Tp1:EGFP expression (arrowheads) on the lateral side of the OE where OMCCs differentiate. (D,E,G,H) 3D projection anterior views of the OEs of representative embryos after treatment with DMSO (negative control; D) or 100 µM DAPT/DMSO (Notch signaling (γ-secretase) inhibitor; E) at 48–72 hpf, or of embryos heat shocked at 48 hpf without (G) or with (H) the heat shock-inducible transgene HS:dnMAML. All embryos were fixed at 72 hpf, followed by acTub immunostaining. (F,I) acTub staining volume quantitated at 72 hpf using Imaris’s ‘Surface Analysis’ after chemical treatment (F) or heat shock-based genetic perturbation (I). Horizontal bars: Mean +/− 95% confidence intervals. Statistical significance: ***p<0.001, ****p<0.0001. n-values indicate number of embryos assayed. OMP:RFP: magenta; Tp1:EGFP: green; acTub: magenta. Orientation arrows: L: Lateral; M: Medial; D: dorsal; V: ventral. Scale bars: 15 µm.

3.4. Notch signaling regulates expression of foxj1a and foxj1b

The transcription factor FOXJ1 has been shown to induce the differentiation of human airway epithelium progenitors into MCCs, and in the frog epidermis, Foxj1 is associated with MCC fate change (Didon et al., 2013; Tasca et al., 2021). In zebrafish, orthologues foxj1a and foxj1b regulate motile ciliogenesis in Kupffer’s vesicle and kidney tubules and, additionally, are expressed in the developing OE (Chong et al., 2018; Yu et al., 2008). At 72 hpf, the near-absence of Notch signaling in OMCCs along the lateral edge of the OE (Fig. 3D) correlated with high levels of foxj1a and foxj1b mRNA expression in those cells (Fig. 4A–A”). Upon DAPT treatment at 48–72 hpf to inhibit Notch signaling (Fig. 4B–C’’), both foxj1a and foxj1b expression expanded broadly throughout the OE and into the medial side of the epithelium (Fig. 4B–E), pointing to a role for Notch signaling in spatially restricting the expression of these genes during OMCC differentiation.

Fig. 4.

Notch signaling regulates expression of foxj1a and foxj1b. (A-A’) 1.77 µm thick anterior views of a representative actb:GFP+ embryo at 72 hpf showing foxj1a and foxj1b mRNA expression on the lateral side of the OE (arrowhead). Colocalization in OMCCs (A-A’, box) is shown magnified in A’’. (B-C’’) 3D projection anterior views of representative OEs after treatment with DMSO (negative control; B-B’’) or 100µM DAPT/DMSO (Notch signaling inhibitor; C-C’’) at 48–72 hpf, followed by fixation at 72 hpf and whole mount in situ hybridization chain reaction (HCR) with probes targeting foxj1a (B,B’,C,C’) or foxj1b mRNA (B,B’’,C,C’’). (D,E) Total surface area of HCR signal per OE quantitated using Imaris’s ‘Surface Analysis’ for foxj1a (D) and foxj1b (E). Horizontal bars: Mean +/− 95% confidence intervals. Statistical significance: ***p<0.001, ****p<0.0001. n-values indicate number of embryos assayed. foxj1a mRNA: green; foxj1b mRNA: magenta; actb:GFP: orange; DAPI: grey. Orientation arrows: L: Lateral; M: Medial; D: dorsal; V: ventral. Scale bars: (A,A’,B-C’’): 15 µm; (A’’):5 µm.

4. Discussion

Our findings outline the spatiotemporal dynamics of OMCC development, identify the earliest time points of OMCC specification, reveal an essential contribution of peridermal cells to the OMCC population, and demonstrate the role of Notch signaling in regulating OMCC differentiation. OMCC specification occurs only in a subset of cells that exhibit, as per an established Notch signaling reporter line (Parsons et al., 2009), nearly undetectable Notch signaling levels in comparison to that found in nearby cells in the OE. Temporally-specific Notch signaling knockdown, both chemical and genetic, yields a striking increase in OMCCs and upregulates the expression of Foxj1 transcription factors that have been previously implicated in MCC fate specification and differentiation (Didon et al., 2013; Tasca et al., 2021).

Our direct observations of periderm differentiation into OMCCs and a drastic reduction in OMCCs in response to peridermal cell ablation point to a critical role for the periderm during olfactory development. Previous work on MCCs’ origins has focused primarily on internal organs such as the kidney (Liu et al., 2007) and trachea (Rock et al., 2011) that are distant from the periderm, and even in the outer epidermis of frog embryos, MCCs have been shown to arise from basal layer progenitors (Deblandre et al., 1999). The OE, on the other hand, is an early-forming superficial organ that is already susceptible to damage from environmental toxins or other agents during early stages of development. Given the close association of the periderm with the OE, it is possible that peridermal cells harbor the developmental plasticity to act as transient OMCC precursors that can quickly replace damaged MCCs and thus maintain homeostasis in developing larvae.

Consistent with our findings, it has been shown that olfactory ciliated non-sensory cells in actinopterygians such as sturgeons are derived from the outer epidermis (Zeiske et al., 2003). However, anatomical characterization of the developing zebrafish OE via electron microscopy has suggested the parsimonious possibility that subepidermal cells give rise to all cell types within the olfactory organ, including OMCCs (Hansen and Zeiske, 1993). Although our cell tracking data did not yield evidence of this hypothesis, our findings of a peridermal origin for OMCCs do not necessarily contradict it either -- it is possible that OMCCs derive from multiple progenitor cell types, whether in early development and/or at later stages. Along those lines, the periderm is lost entirely in zebrafish by 42 dpf (Lee et al., 2014), suggesting that the derivation of replacement OMCCs throughout an animal’s lifespan is dependent on at least one other source of precursor cells. Thus, it may be that both outer epidermal/peridermal and subepidermal cells have varying capabilities to form OMCCs across species. As with many aspects of olfactory development, further work is needed to fully understand developmental versus regenerative processes in this dynamically maintained organ.

Our data showing both an absence of OMCC formation in high-Notch signaling regions of the OE and increased OMCC differentiation upon Notch signaling knockdown are consistent with Notch signaling’s role in suppressing MCC differentiation in other tissues such as the zebrafish pronephros (Liu et al., 2007), mouse trachea (Rock et al., 2011), and frog epidermis (Tasca et al., 2021). Notch signaling is also known to regulate the expression of foxj1a in kidney tubules (Jurisch-Yaksi et al., 2013), while a connection to foxj1b has not been previously shown. Our finding that Notch signaling knockdown upregulates expression of both foxj1a and foxj1b throughout the olfactory pit may indicate that these factors are important for spatially restricting OMCC differentiation. Intriguingly, our Notch signaling knockdown experiments resulted in ectopic differentiation of OMCCs in regions of reduced Notch signaling (Fig. 3D–E) along the medial edge of the OE, thus disrupting a stereotypical gap in OMCC distribution and perhaps indicating a fate change in adjacent cells. Previous work has shown that OMCCs along the lateral edge of the OE guide appropriate directional fluid flow (Reiten et al., 2017), and therefore, the additional OMCCs along the medial edge in our perturbation experiments may end up disrupting this important organ-wide function.

Of note, Notch signaling inhibition has been shown to help differentiate human induced pluripotent stem cells into MCCs in vitro (Firth et al., 2014). Moving forward, zebrafish OMCCs offer a tractable model system to complement such approaches, and it may be illuminating to compare Notch signaling’s effects on induced pluripotent stem cells in vitro to its effects in vivo on periderm cells differentiating into OMCCs. Broadly, investigating peridermal differentiation into OMCCs may yield new ideas for how to drive a variety of precursors to take on an MCC fate in a therapeutic context.

Highlights.

Krt4+ peridermal cells give rise to the earliest zebrafish olfactory multiciliated cells (OMCCs)

Sustained Notch signaling inhibits OMCC differentiation

Notch signaling restricts regional expression of the ciliogenic genes foxj1a and foxj1b