ABSTRACT
Mechanosensory hair cells located in the inner ear mediate the sensations of hearing and balance. If damaged, mammalian inner ear hair cells are unable to regenerate, resulting in permanent sensory deficits. Aquatic vertebrates like zebrafish (Danio rerio) have a specialized class of mechanosensory hair cells found in the lateral line system, allowing them to sense changes in water current. Unlike mammalian inner ear hair cells, lateral line hair cells can robustly regenerate following damage. In mammals, the transcription factor Foxg1 functions to promote normal development of the inner ear. Foxg1a is expressed in lateral line sensory organs in zebrafish larvae, but its function during lateral line development and regeneration has not been investigated. Our study demonstrates that mutation of foxg1a results in slower posterior lateral line primordium migration and delayed neuromast formation. In developing and regenerating neuromasts, we find that loss of Foxg1a function results in reduced hair cell numbers, as well as decreased proliferation of neuromast cells. Foxg1a specifically regulates the development and regeneration of Islet1-labeled hair cells. These data suggest that Foxg1 may be a valuable target for investigation of clinical hair cell regeneration.
INTRODUCTION
Auditory perception and balance depend on specialized mechanosensory hair cells in the inner ear (Caprara and Peng, 2022; Fettiplace, 2017; Thomas et al., 2015). Damage to these hair cells can occur through multiple mechanisms including genetic mutations, age, prolonged sound exposure, infection, and exposure to ototoxic drugs (Matsui and Cotanche, 2004), resulting in deafness and loss of vestibular function. Mammals are incapable of regenerating lost or damaged hair cells after development, except for small populations of vestibular hair cells (Burns and Stone, 2017). These injuries lead to permanent sensory-motor disability (Matsui and Cotanche, 2004; Roberson and Rubel, 1994). The World Health Organization estimates that as of 2018 hearing loss was the fourth leading cause of disability in humans with more than 466 million individuals diagnosed (World Health Organization, 2018). In turn, the cost of untreated hearing loss reported by the World Health Organization in 2017 was estimated to be greater than 750 billion dollars globally (World Health Organization, 2017). Thus, research into the development of these mechanosensory hair cells and surrounding tissue is an important field for human health and economic stability.
Contrary to their mammalian counterparts, non-mammalian vertebrates are capable of regenerating functional hair cells during development and throughout adulthood (Brignull et al., 2009; Ghysen and Dambly-Chaudiere, 2007; Kniss et al., 2016; Pinto-Teixeira et al., 2015). In addition to inner ear hair cells, zebrafish, as well as other aquatic vertebrates, have a lateral line mechanosensory system that uses hair cells to sense water motion (Brignull et al., 2009; Coombs et al., 2014). Zebrafish lateral line hair cells show developmental, morphological, and genetic conservation with mammalian inner ear hair cells making them an attractive model for studying human development and disease (Nicolson, 2005). The zebrafish lateral line mechanosensory system lends itself particularly well to experimentation due to its superficial localization on the surface of the body, allowing easy visualization and manipulation. The lateral line system is made up of small sensory organs, called neuromasts, arrayed on the surface of the fish. These neuromasts contain mechanosensory hair cells and surrounding support cells (Thomas et al., 2015). Many key molecular pathways, such as FGF, Notch, and Wnt are critical for the development of inner ear and zebrafish lateral line hair cells (Baek et al., 2022; Jiang et al., 2014; Lush et al., 2019). Support cells help form the surrounding tissue in which hair cells reside, provide trophic support for innervating neurons, and act as progenitors for hair cell regeneration (Cruz et al., 2015; Thomas et al., 2015). Support cells' ability to act as progenitors is regulated by FGF, Notch, and Wnt signaling (Lush et al., 2019; Lush and Piotrowski, 2014; Megerson et al., 2024). Zebrafish lateral line hair cells show dose-dependent damage in response to ototoxic drugs, as is also observed in mammalian models and human patients (Brignull et al., 2009; Namdaran et al., 2012). Taken together, the conservation between mammalian hair cells and zebrafish lateral line hair cells makes them a significant tool to investigate disease and damage as well as possible therapeutic interventions.
Forkhead box G1 (Foxg1) is a member of a large family of transcription factors that regulate multiple cellular processes including proliferation, differentiation, and survival (Clark et al., 1993). Foxg1 has been implicated during development to increase proliferation of progenitor cells in multiple tissues including the inner ear tissue (Ding et al., 2020; Hwang et al., 2009; Wong et al., 2019). Recent work shows that Foxg1 is involved in hair cell development and homeostasis in mammalian models (He et al., 2020; Hwang et al., 2009; Zhang et al., 2020). The loss of Foxg1 function results in morphological deformities of the cochlea and sensory cristae, as well as altered hair cell polarity and total hair cell numbers (Hwang et al., 2009; Pauley et al., 2006; Zhang et al., 2020). Foxg1 is also implicated in age related hair cell homeostasis through regulation of reactive oxygen species and autophagy (He et al., 2020). Foxg1 interacts directly with critical hair cell development pathways such as canonical Wnt, FGF, and Notch signaling (Akol et al., 2022; Ding et al., 2020). The possible functions of Foxg1 have not been studied in the developing zebrafish lateral line or in the context of hair cell regeneration. We seek to uncover the function of Foxg1 in the development and regeneration of the zebrafish lateral line.
Our current study investigates the function of Foxg1a using the foxg1aa266 (Thyme et al., 2019) mutant line to examine development and regeneration of hair cells and support cells in the zebrafish posterior lateral line (pLL). We show that loss of Foxg1a function results in slower pLL primordium (pLLP) migration during the development of the pLL and delayed neuromast formation. We found that significantly fewer hair cells form in the nascent pLL and there is a reduction in proliferating cells in the developing neuromasts. Following regeneration, we found a significant reduction in proliferating support cells, hair cells numbers, and α-Isl1-positive cells in foxg1aa266 mutant neuromasts. This work suggests that Foxg1a functions in the neuromast to enable appropriate proliferation and differentiation of cells in the zebrafish lateral line. Understanding the role of Foxg1a in hair cell biology may provide potential directions for future interventions in human hearing loss.
RESULTS
foxg1a is expressed in the developing and regenerating posterior lateral line
Foxg1 function in the development of mammalian inner ear led us to investigate its possible role in the zebrafish lateral line (Ding et al., 2020; Pauley et al., 2006). Work by others has suggested that foxg1a is expressed in lateral line neuromasts but as of yet, no functional role has been elucidated (Baek et al., 2022; Lush et al., 2019). We first sought to determine if foxg1a is expressed in the developing lateral line. Using wholemount RNA in situ hybridization (WISH) in wild-type embryos we show foxg1a is expressed in the migrating pLLP at 28 h post fertilization (hpf) (Fig. 1A; Fig. S1A). The migrating pLLP deposits clusters of cells in its wake that will continue to proliferate and differentiate to give rise to neuromasts containing mechanosensory hair cells and surrounding support cells (Brignull et al., 2009; Ma and Raible, 2009; Thomas et al., 2015). We see expression of foxg1a in newly deposited wild-type zebrafish neuromasts at 2 days post fertilization (dpf) (Fig. 1B; Fig. S1B) and in maturing neuromasts at 5 dpf (Fig. 1C; Fig. S1C) and 8 dpf (Fig. 1D; Fig. S1D). foxg1a expression is maintained in the regenerating neuromast at 3 h post hair cell ablation with the ototoxic aminoglycoside antibiotic neomycin (Fig. 1E; NEO), 1 day-post NEO exposure (Fig. 1F; 1 day-post NEO), and after regeneration is complete at 3 days-post NEO exposure (Fig. 1G; 3 days-post NEO). The continued expression of foxg1a in the neuromast during regeneration was also reported by single-cell RNA-sequencing (Baek et al., 2022). We next sought to analyze expression levels and cellular localization of foxg1a during homeostasis and regeneration. To do so we used hybridization chain reaction fluorescence in situ hybridization (HCR FISH) in wild-type larvae. We observed expression throughout the neuromast during homeostasis, and during regeneration at 1 day-post NEO and 3 days-post NEO (Fig. 1H-J′). We further characterized expression by comparing fluorescent intensity of the whole neuromast, hair cells, and non-hair cell support cells during homeostasis and regeneration. A significant increase in foxg1a expression is observed following hair cell ablation in the whole neuromast (Fig. 1K), hair cells (Fig. 1K′), and surrounding support cells (Fig. 1K″) 1 day-post NEO and 3 days-post NEO when compared to 5 dpf non-ablated larvae. Although we find increased foxg1a expression during regeneration, we do not find a bias toward specific cell types within the neuromast; foxg1a is expressed in hair cells and surrounding support cells at similar levels (Fig. 1K-K″). These data demonstrate that foxg1a expression occurs within the migrating pLL primordium and carries on through maturation of LL neuromasts with increased levels of expression during regeneration.
Foxg1a regulates posterior lateral line primordium migration and neuromast development
As we observed expression of foxg1a in the developing and regenerating lateral line, we sought to determine if it plays a functional role as well. During embryonic development, the zebrafish pLL follows a well characterized and stereotypical developmental pattern that involves the collective migration of the pLLP cells and deposition of proto-neuromasts along the trunk (Ma and Raible, 2009; Thomas et al., 2015). Mutations affecting pLL development result in phenotypes including truncated lateral line formation, supernumerary neuromasts, loss of neuromasts, altered hair cell numbers, and changes in neuromast size (Lush and Piotrowski, 2014; Thomas et al., 2015). The foxg1aa266 mutant allele is predicted to be a null mutation with a large deletion overlapping the Forkhead domain in the single exon of foxg1a (Thyme et al., 2019). Neuromasts of the zebrafish pLL form after being deposited in the wake of the migrating pLLP between ∼22 hpf and 48 hpf. Live time-lapse imaging of heterozygous and homozygous foxg1aa266 embryos expressing Tg(prim:lyn2mCherry), a transgene that labels the cell membranes of the migrating primordium and neuromasts, as well as other sensory tissues (Wang et al., 2018), shows pLLP migration velocity between 33-48 hpf is significantly slower in mutants as compared to controls (Fig. 2A-E; Movie 1). We confirmed this reduced velocity was not due to a reduction in the total number of primordia cells in foxg1aa266 larvae (Fig. S2A,A″,B,B″,C′), total area of the pLLP (Fig. S2C″), or a reduction of cellular proliferation in the primordia as observed by BrdU, a thymidine analog, incorporation into newly synthesized DNA during migration (Fig. S2A,A′,B,B′,C). At 2 dpf we found there was a small, but significant reduction in the number of deposited neuromasts in foxg1aa266 mutants as compared to heterozygous controls (Fig. 2F-H). There was no significant change in somite number, embryo length, or eye circumference at 2 dpf, indicating that the slower pLLP migration appears specific to pLL development as opposed to overall developmental delay (Fig. S3A-B′,C,D,E). However, this reduction was recovered by 5 dpf (Fig. 2I-K). These data demonstrate Foxg1a function plays a role in the timing of pLL development during primordium migration and neuromast deposition.
Hair cell numbers are reduced in foxg1aa266 mutant neuromasts during development and regeneration
Cells in the maturing zebrafish neuromasts form mechanosensory hair cells and surrounding support cells (Coombs et al., 2014; Ghysen and Dambly-Chaudiere, 2007; Thomas et al., 2015). As Foxg1a function seems necessary for early lateral line development and is expressed in maturing neuromasts, we next asked if it was necessary for hair cell development. During pLLP migration, the first hair cells will start to form before being deposited with developing neuromasts (Itoh and Chitnis, 2001). Using the Tg(mysoin6b:GFP)w186 (myo6:GFP) transgenic zebrafish, which will label hair cells, we are able to visualize and quantify hair cell numbers. At 28 hpf we can see the formation of early hair cells in the trailing pLLP with no significant difference in quantity when comparing foxg1aa266 embryos to heterozygous controls (Fig. S2D,D′,F,F,G). Looking at later developmental time points we find on average significantly fewer hair cells in foxg1aa266 mutant pLL neuromasts compared to heterozygous siblings (Fig. 3A,B-D′) at 5 dpf when the majority of hair cells in the developing neuromast have matured. The reduction in hair cells is also observed at 8 dpf, suggesting it is not developmental delay (Fig. 3E,F-H″). To determine if hair cells are being lost to apoptosis, we conducted TUNEL assays at 4 dpf and 8 dpf and found no significant increase in cell death (Fig. S4A-H). We quantified the population of hair cells with functional mechanoelectrical transduction (MET) channels using the fixable fluorescent vital dye FM1-43FX, which enters hair cells through MET channels and can give insight into functionality of hair cells (Fettiplace, 2017). We observe there is also a significant reduction in FM1-43FX labeled hair cells in foxg1aa266 larvae as compared to controls at 5 dpf (Fig. 3B,B″,C,C″,D′) and that the percent of FM1-43FX-labeled hair cells is significantly reduced in mutant larvae (Fig. 3D″). Looking later in development we conducted this same experiment at 8 dpf and observed the same significant reduction in myo6:GFP and FM1-43FX-labeled hair cells in foxg1aa266 larvae as compared to heterozygous siblings (Fig. 3F,F″,G,G″,H′), however, the percent of FM1-43FX labeled cells is no longer reduced in the 8 dpf mutant population (Fig. 3H″). These data demonstrate that Foxg1a function is necessary for the appropriate number of hair cells to develop in the zebrafish lateral line.
Many of the molecular and cellular mechanisms that drive hair cell development in the zebrafish lateral line are also active during regeneration (Cruz et al., 2015; Kniss et al., 2016; Pinto-Teixeira et al., 2015). For that reason, we asked if Foxg1a function is required for regeneration of hair cells. We used NEO-exposure to ablate hair cells in 5 dpf zebrafish larvae when the majority of hair cells are mature (Harris et al., 2003) and assessed regeneration at 8 dpf (Fig. 3I). At 3 days-post NEO we again used myo6:GFP to label total hair cell numbers and found that foxg1aa266 larvae had fewer average hair cells per neuromast as compared to heterozygous siblings (Fig. 3J,J′,K,K′,L). When we use FM1-43FX to measure hair cells with functional MET channels 3 days-post NEO, we find a significant reduction in average hair cells per neuromast in foxg1aa266 larvae as compared to controls (Fig. 3J,J″,K,K″,L′), which recapitulates observations of homeostatic conditions at 5 dpf and 8 dpf. The percent of FM1-43FX labeled foxg1aa266 hair cells is not significantly different 3 days-post NEO (Fig. 3L″). A TUNEL assay at 18 h-post NEO, when proliferation is at its peak during regeneration (Ma et al., 2008), showed no significant difference in labeling between heterozygous sibling and foxg1aa266 mutants (Fig. S4I-K), suggesting cell death is not the major cause of the reduction of regenerated hair cells in foxg1aa266 mutant larvae. These data indicate that Foxg1a function is necessary for the development and regeneration of neuromast hair cells, and that that these hair cells have functional MET channels. We must note that these experiments do not fully confirm functional transduction of signals between hair cells and innervating neurons. Mutations affecting hair cell and neuromast development can also cause disruptions in innervation and hair cell polarity (Kindt and Sheets, 2018; Navajas Acedo et al., 2019). Using Tg(neuroD:eGFP) transgenic fish, which labels the axons of innervating neuromast neurons and other tissue (Obholzer et al., 2008), we assessed innervation of foxg1aa266 neuromasts and noted no significant difference when compared to control larvae (Fig. S5A,A″,B,B″,C′). Similarly, we observed no significant difference in hair cell polarity when comparing foxg1aa266 larvae to heterozygous siblings (Fig. S5D-F).
Total neuromast cell numbers are not reduced in foxg1aa266 mutants
As loss of Foxg1a function results in fewer hair cells, we asked whether there was also a reduction in the non-sensory support cells within the developing neuromast. To look at total support cell populations we used α-Sox2 antibody labeling and noted no significant difference in total support cells when comparing foxg1aa266 mutants to heterozygous siblings at 5 dpf (Fig. 3M,M″,N,N″,O′), though there was still a reduction in hair cells labeled with α-Otoferlin (α-Oto; Fig. 3M,M′,N,N′,O). We also noted there was no change in total DAPI-labeled cells in foxg1aa266 mutants as compared to heterozygous controls (Fig. 3M,M′″,N,N′″,O″). We reasoned there may be a significant difference in support cell populations during regeneration with loss of Foxg1a function. We used NEO exposure to ablate hair cells at 5 dpf then allowed regeneration before fixation and labeling with α-Sox2, and α-Oto to measure any loss of total support cells. We found that at 3 days-post NEO there was no significant decrease in α-Sox2 labeled cells when comparing foxg1aa266 larvae to heterozygous siblings (Fig. 3P,P″,Q,Q″,R′). Similar to observations at 5 dpf during homeostatic analyses, we also saw no significant decrease in total cell numbers as labeled with DAPI (Fig. 3P,P′″,Q,Q′″R″), but still observe the reduced hair cell phenotype when comparing foxg1aa266 mutants to controls (Fig. 3P,P′,Q,Q′,R).
Proliferation is reduced in foxg1aa266 mutant neuromasts during development
Since we observed a reduction in hair cells with no apparent increase in cell death, we reasoned that the hair cells may not be generated due to effects on cellular proliferation in foxg1aa266 mutant larvae. To investigate this, we used BrdU incorporation in 24-h pulses between 2-5 dpf (Fig. 4A), during the initial maturation of pLL neuromasts. We find significantly fewer cells labeled by BrdU incorporation and fewer hair cells in foxg1aa266 mutant neuromasts as compared to heterozygous controls in larvae exposed to BrdU between 2-3 dpf (Fig. 4B-B″,C,C″,D′) and between 3-4 dpf (Fig. 4E,E″,F,F″,G′). During these same time points we again observe a significant reduction in average hair cell number per neuromast in foxg1aa266 larvae as compared to controls (Fig. 4D,G). At 2-3 dpf the percent of hair cells labeled with BrdU is not significantly different (Fig. 4D″); however, both mutant and control fish pulsed with BrdU from 3-4 dpf demonstrated a reduced percentage of BrdU labeled hair cells compared to the 2-3 dpf (Fig. 4D″,G″). Of the fish pulsed with BrdU 3-4 dpf, a significantly larger percent of BrdU-labeled hair cells was observed when comparing foxg1aa266 embryos to heterozygous siblings (Fig. 4G″). The total number of cells labeled with BrdU incorporation during the 3-4 dpf time point was significantly reduced in the foxg1aa266 embryos as compared to heterozygous siblings (Fig. 4G′), as was observed during the 2-3 dpf time period. We take these data to suggest that loss of Foxg1a function results in fewer hair cells as a consequence of reduced proliferation during neuromast development. To confirm we did not see a recovery or prolonged BrdU incorporation in foxg1aa266 mutants, we pulsed larvae from 4-5 dpf and observed no significant difference in number of or percent of BrdU incorporation in foxg1aa266 larvae as compared to heterozygous siblings while still showing a reduction in hair cells (Fig. 4H-J″).
Proliferation is reduced during regeneration in foxg1aa266 neuromasts
As loss of Foxg1a function results in reduced proliferation in developing neuromasts, we asked if there is a similar reduction during regeneration. To assess proliferation, we incubated larvae in BrdU for the first 24 h immediately following hair cell ablation with NEO when proliferative regeneration of hair cells is known to occur (Ma et al., 2008). Larvae were then allowed a further 2 days of regeneration before fixation and analysis (Fig. 4K). We observed a significant decrease in BrdU-labeled cells in the foxg1aa266 neuromasts as compared to heterozygous siblings (Fig. 4L,L″,M,M″,N′). We again questioned if there was a delay or recovery of the proliferation phenotype during regeneration. Following NEO-exposure, we pulsed cohorts of larvae with BrdU from 5-6 dpf, 6-7 dpf, or 7-8 dpf (Fig. S6A) then fixed all the larvae at 8 dpf when regeneration was complete. BrdU incorporation showed a significant decrease during the first 24 h pulse (Fig. S6B-C′″,E) demonstrating our initial observations (Fig. 4N′); however, no significant difference was observed in BrdU incorporation of foxg1aa266 larvae as compared to controls during the 6-7 dpf, or the 7-8 dpf (Fig. S6G-O). Though the number of cells observed to have BrdU incorporation was reduced in foxg1aa266 larvae, the percentage of BrdU-labeled hair cells was not significantly different between foxg1aa266 mutant neuromasts and heterozygous controls following regeneration (Fig. 4L,L″,M,M″,N″). We take these data to indicate that loss of Foxg1a function does not impede the proliferative regeneration of hair cells, but it does result in an overall reduction in proliferation and hair cell regeneration.
Support cell populations are intact in foxg1aa266 mutant neuromasts
During regeneration, differing support cell subpopulations give rise to new hair cells and/or replenish support cells based on location within the zebrafish neuromast (Lush et al., 2019; Thomas and Raible, 2019). Work by others demonstrated that regenerating hair cells arise predominantly from dorsoventral support cells (Romero-Carvajal et al., 2015; Thomas and Raible, 2019) and tend to be located more central to the neuromast as opposed to the periphery (Lush et al., 2019). We asked if there may be a difference in these support cell sub-populations during development and regeneration. These subpopulations are distinguished based on their transcriptomic profile and spatial organization within the neuromast (Fig. 5A) (Lush et al., 2019; Thomas and Raible, 2019). We examined these neuromast support cell sub-populations using Tg(sfrp1a:nlsEos)w217 (sfrp1a:nlsEos) to label the peripheral mantle cells (Thomas and Raible, 2019) (Fig. 5G-H) and Tg(sost:nlsEos)w215 (sost:nlsEos) to label dorsoventral support cells (Fig. 5C-D′) (Thomas and Raible, 2019). These transgenes express nuclear-localized Eos (nlsEos) under the control of the endogenous sfrp1a and sost promoters (Thomas and Raible, 2019). As a photoconvertible fluorophore Eos will permanently shift from green fluorescent emission to red fluorescent emission when exposed to UV light, allowing for lineage tracing (Wiedenmann et al., 2004). We first quantified the mantle cell population expressing sfrp1a:nlsEos and dorsoventral population expressing sost:nlsEos at 5 dpf by counting the total photoconverted cells under homeostatic conditions (Fig. 5B). We found no significant difference in the number of labeled cells in foxg1aa266 fish as compared to heterozygous siblings (Fig. 5C-F) when looking at the labeled mantle cells or dorsoventral cells. Using HCR FISH, we quantified the number of cells expressing an anteroposterior support cell marker tnfsf10l3 at 5 dpf (Fig. 5B) and found no significant difference in the number of labeled cells between foxg1aa266 and heterozygous sibling larvae (Fig. 5G,H). To look at these same cell populations during regeneration we photoconverted sfrp1a:nlsEos-positive cells or sost:nlsEos-positive cells prior to NEO-exposure, then ablated hair cells with NEO, and traced the fate of nlsEos-positive cells following regeneration (Fig. 5I). We found no significant difference in the total number of either sfrp1a:nlsEos-labeled cells or sost:nlsEos-labeled cells following regeneration when comparing heterozygous and foxg1aa266 larvae (Fig. 5J-M). To assess whether the anteroposterior support cell population was perturbed during regeneration with loss of Foxg1a we again used HCR FISH of tnfsf10l3 and found no significant difference in the number of cells showing fluorescence in mutant larvae compared to controls (Fig. 5N,O). These data suggest that the mantle cell, dorsoventral, and anteroposterior support cell populations are not sensitive to loss of Foxg1a function.
mRNA expression patterns are largely unchanged in foxg1aa266 neuromasts
During development, cells deposited in maturing neuromasts will give rise to hair cells, surrounding support cells, and quiescent mantle cells (Brignull et al., 2009; Ma and Raible, 2009). Previous work has demonstrated there are subpopulations of neuromast support cells that show differential function as well as genetic expression (Baek et al., 2022; Lush et al., 2019; Megerson et al., 2024; Thomas and Raible, 2019). Using WISH at 5 dpf, we compared the expression profiles of different genes based on their roles in proliferation and generation of hair cells, cell population localization, and signaling pathways to better understand how loss of Foxg1a function is causing a reduction in hair cells. Based on scRNA-sequencing screens from the Piotrowski and Burgess labs, we investigated four separate genes corresponding to cells central to the neuromast, isl1a, six1a, six1b, and gata2a (Jimenez et al., 2022; Lush et al., 2019). At 5 dpf we observe similar patterns of expression for six1a, six1b, and gata2a in foxg1aa266 larvae when compared to controls (Fig. 6B-D′). Surprisingly, though the other central cell markers appeared similar, we noted a stark reduction in expression patterns of isl1a in foxg1aa266 larvae at 5 dpf when compared to heterozygous siblings (Fig. 6A,A′). Dorsoventral and anteroposterior support cells were labeled with sost and tnfs10l3, respectively, with sost marking regions important for hair cell regeneration (Lush et al., 2019; Thomas and Raible, 2019). Expression of sost and tnsf10l3 was similar in control and foxg1aa266 neuromasts (Fig. 6E-F′) recapitulating our earlier data of cell counts of sost:nlsEos-positive cells and HCR FISH of tnfsf10l3 (Fig. 5E-H). We next looked for changes in genes linked to hair cell development and regeneration, specifically in the Wnt and Notch/Delta signaling pathways (Megerson et al., 2024; Romero-Carvajal et al., 2015). We noted that expression patterns for Notch/Delta ligands and markers notch3, deltaD, her2, her4.1, and lfng appeared similar in foxg1aa266 larvae compared to controls (Fig. 6G-K′). We also see no obvious expression differences when comparing expression of the hair cell fate marker atoh1a in foxg1aa266 larvae to controls (Fig. 6L,L′). Similarly, when we looked at expression of different Wnt ligands and β-catenin in foxg1aa266 larvae we observe no obvious differences as compared to heterozygous controls (Fig. 5M-O′). Together, these expression profiles suggest that isl1a expression is regulated by Foxg1a function in posterior lateral line neuromasts.
α-Isl1-labeled cells are reduced in foxg1aa266 mutants during development
The transcription factor islet1a (isl1a) is expressed in central support cells and is predicted to be in hair cell progenitor cells within zebrafish posterior lateral line neuromasts (Baek et al., 2022). The reduced expression of isl1a in 5 dpf foxg1aa266 mutants we observed (Fig. 6A-A′) lead us to ask if we would see a concomitant reduction in the number of Isl1+ cells in the foxg1aa266 mutant neuromast. Using an α-Isl1 antibody, we first looked at the migrating pLLP, where the first hair cells begin to form (Itoh and Chitnis, 2001). At 28 hpf, we found that a small number of nascent myo6:GFP+ hair cells in both mutant and control embryos which were not significantly different when comparing heterozygous control and foxg1aa266 mutant primordia (Fig. S3D-E′,F). All the myo6:GFP+ cells were also labeled with α-Isl1 (Fig. S3F″), though overall, there was a significant reduction in total α-Isl1-labeled cells in foxg1aa266 mutants as compared to controls (Fig. S3D-E″,F′). We next examined α-Isl1 antibody labeling in maturing pLL neuromasts and observed that at 2 dpf and 5 dpf foxg1aa266 fish exhibit reduced numbers of myo6:GFP+ hair cells and α-Isl1+ cells (Fig. 7A-E″,C,C′,F,F′). We also note there are significantly fewer myo6:GFP+/α-Isl1+ co-labeled cells at both the 2 dpf and 5 dpf time points in foxg1aa266 mutants as compared to controls (Fig. 7A,-E″,C″,F″). Interestingly, when we quantified α-Isl1+, but myo6:GFP- cells, we noted that while there were fewer on average, there was no significant difference in mutant neuromasts as compared to heterozygous siblings at either the 2 dpf, or 5 dpf time points (Fig. 7A-E″,C′″,F′″). These data indicate that the reduced complement of α-Isl1+ cells may correspond to the reduced myo6:GFP+ hair cell population.
foxg1aa266 neuromasts have fewer Isl1-labeled regenerating hair cells
We next sought to determine which cells are labeled by α-Isl1 antibody in the regenerating neuromast and if they are altered in foxg1aa266 mutants. To assess this we fixed and labeled mutant and control fish expressing myo6:GFP with α-Isl1antibody at multiple time points during regeneration to better separate these populations (Fig. 7G). We first looked at α-Isl1 labeling 3 h-post NEO when most hair cells have been ablated, but new hair cells have not regenerated. At this time point we observed that there was still a large population of α-Isl1+ cells in both the heterozygous siblings and foxg1aa266 mutants 3 h-post NEO (Fig. 7H-I″,J′) indicating that there is a subset of α-Isl1+ cells distinct from hair cells, and that there are significantly fewer α-Isl1+ cells in foxg1aa266 larvae as compared to controls (Fig. 7J′-J″). As regeneration progresses, we see that foxg1aa266 larvae form fewer hair cells per neuromast as compared to heterozygous siblings at 1 day-post NEO (Fig. 7K,K′,L,L′,M), 2 days-post NEO (Fig. 7N,N′,O,O′,P), and 3 days-post NEO (Fig. 7Q,Q′,R,R′,S). During these regeneration periods the total number of α-Isl1a+ cells in foxg1aa266 larvae are significantly reduced compared to heterozygous controls (Fig. 7H-R”,J′,M′,P′,S’) at each time point. Fewer α-Isl1+/myo6:GFP+ co-labeled regenerated hair cells are present in foxg1aa266 larvae 2 days-post NEO (Fig. 7P″) and 3 days-post NEO (Fig. 7S″), whereas the α-Isl1+ non-hair cell population only exhibits a significant difference in mutants at 3 h-post NEO as compared to controls (Fig. 7J″′). No significant difference in the number of α-Isl1+ non-hair cells is observed at 1 day-post NEO (Fig. 7M″′), 2 days-post NEO (Fig. 7P′″), or 3 days-post NEO (Fig. 7S′″). These data indicate that α-Isl1+ labels a subset of neuromast cells, and that some of those α-Isl1+ cells include myo6:GFP hair cells. The hair cells which are also α-Isl1+ cells appear to be sensitive to Foxg1a function.
α-Isl1+ cells represent a distinct population from sost:nlsEos+ cells
Genetic expression analyses using scRNA-sequencing and cell lineage experiments suggest that neuromast cells are made up of distinct sub-populations that can contribute to hair cell regeneration differently based on expression profile and location (Baek et al., 2022; Lush et al., 2019; Thomas and Raible, 2019). The work by Thomas et al. demonstrated that a dorsoventral population labeled with sost:nlsEos+ will give rise to the majority of hair cells regenerated following hair cell ablation (Thomas and Raible, 2019). We questioned how much overlap there may be between the α-Isl1+ support cell, α-Isl1+ hair cell, and the dorsoventral sost:nlsEos+ populations and the effect of Foxg1a function on these cells. Using lineage tracing we looked at homeostatic development and regeneration by photoconverting sost:nlsEos in fish co-expressing myo6:GFP at 5 dpf then fixing at 8 dpf and then labeling with α-Isl1 antibody (Fig. 8A,E). Fish used for regeneration were exposed to NEO immediately after photoconversion at 5 dpf to allow lineage tracing of how sost:nlsEos+ cells contribute to hair cell regrowth (Fig. 8E). We observed a decrease in the number of myo6:GFP+ hair cells, α-Isl1+ cells, and myo6:GFP/α-Isl1+ co-labeled cells under both homeostatic conditions (Fig. 8B-B″,C-C″,Di,Dii,Div) and during regeneration (Fig. 8F-F″,G-G″,Hi,Hii,Hiv) in foxg1aa266 mutant larvae as compared to heterozygous siblings. The number of sost:nlsEos+ cells at 8 dpf under homeostatic conditions and following regeneration showed no significant difference in foxg1aa266 larvae compared to controls (Fig. 8F′″,G,G′″,Hi). We were surprised the see that under both homeostatic and regenerative conditions sost:nlsEos and α-Isl1 label distinct groups of cells in the neuromast (Fig. 8B-C′″,Dv-Dvi,F-G′″,Hv-Hvi). In particular, we find that only a small subset of myo6:GFP+ hair cells are co-labeled with α-Isl1 and sost:nlsEos during homeostasis, and the number is significantly lower in foxg1aa266 mutants (Fig. 8Dvii). Following regeneration, we find a small increase in hair cells co-labeled with α-Isl1 and sost:nlsEos (Fig. 8Hvii), though this population is smaller than total regenerated hair cells (Fig. 8Hi) with significantly fewer in foxg1aa266 mutants as compared to heterozygous siblings (Fig. 8Hvii). Together, these data suggest that within the neuromast there are distinct subpopulations of cells which are labeled by sost:nlsEos, α-Isl1. Foxg1a function appears to specifically regulate α-Isl1+ hair cells during development and regeneration.
DISCUSSION
Foxg1 is a transcription factor that has largely been studied for its role in neural development and subsequent developmental defects accompanying mutations within its gene (Wong et al., 2019; Zhao et al., 2009). The zebrafish homolog foxg1a has been well studied in the context of forebrain development, but our study is the first to look specifically at lateral line development and regeneration (Raj et al., 2020; Toresson et al., 1998; Zhao et al., 2009). Though recent research suggests foxg1 is required for the proper development of electroreceptors in the lateral line of paddlefish (Minarik et al., 2024). Our work demonstrates that foxg1a is necessary for the proper development and regeneration of the zebrafish posterior lateral line. Loss of Foxg1a function results in delayed posterior lateral line development, a reduction in the proliferation of neuromast cells, and fewer hair cells. During regeneration, Foxg1a is required for proliferation, formation of new hair cells, and the number of α-Isl1-positive cells in the neuromast. We find that α-Isl1-labeled hair cells are specifically reliant on Foxg1a function during development and regeneration in the zebrafish posterior lateral line.
Foxg1a functions to promote proper development of mechanosensory systems
In studies of the mammalian inner ear, Foxg1 is demonstrated to be necessary for proper morphological formation, innervation, and sensory cell development (Ding et al., 2020; Hwang et al., 2009). Foxg1 knockdown in mouse embryos results in shortened cochlea, greater proportion of inner ear hair cells, and a reduction in inner ear hair cell innervation (Ding et al., 2020; Hwang et al., 2009; Pauley et al., 2006; Zhang et al., 2020). When we looked at early pLL development in foxg1aa266 mutant zebrafish, we noted a reduced rate of migration of the posterior lateral line primordium and delayed neuromast formation, though we found recovery of neuromast numbers by 5 dpf. It is interesting to note that murine models investigating the knockout of Foxg1 demonstrated an increase in hair cells (Pauley et al., 2006), in contrast to the zebrafish lateral line where we see a decrease in hair cell development. In mice, the loss or conditional knockdown of Foxg1 results in significant disruption to innervation of hair cells and the cochlea (Pauley et al., 2006; Zhang et al., 2020). When we examined innervation of hair cells in the foxg1aa266, we found a similar pattern to controls, suggesting the axonal extension of the sensory neurons associated with pLL may not be regulated by Foxg1a. More work is needed to better distinguish the precise role of Foxg1 mechanosensory system development as there is evident conservation of function within the tissue though possibly differing effects between species.
Foxg1a is necessary for proliferative development and regeneration of sensory tissue
Hair and support cells of the zebrafish lateral line develop and regenerate mostly through proliferation and differentiation of support cell progenitors (Coombs et al., 2014; Ghysen and Dambly-Chaudiere, 2007; Kniss et al., 2016; Ma and Raible, 2009; Thomas et al., 2015). As mammalian Foxg1 is implicated in proliferative development of nervous tissue we reasoned the zebrafish Foxg1a may also regulate proliferation in the lateral line (Akol et al., 2022; Wong et al., 2019). Our work shows that loss of Foxg1a function results in reduced proliferation both during development and regeneration of the zebrafish pLL without an increase in cell death. Our work suggests a conserved role for Foxg1 in different support cell's ability to generate new hair cells via proliferation (Akol et al., 2022; Ding et al., 2020; Hwang et al., 2009). The timing of Foxg1 function may also be important, as murine work with loss of function (Pauley et al., 2006) or conditional knockdown (Zhang et al., 2020) showed differing effects during different developmental time points. This is also important as regeneration in murine models is observed during a short window in early development, while zebrafish hair cells retain the capacity to regenerate through the life of the animal (Brignull et al., 2009; Matsui and Cotanche, 2004).
Islet1 in the posterior lateral line neuromasts
Islet1 is a LIM-homeodomain transcription factor known to regulate development of nervous tissue, neural epithelia, and otic tissue (Liang et al., 2011; Sun et al., 2008). In mice its expression is noted in developing support cell populations and nascent hair cells, with its expression in hair cells fading away as differentiation continues (Radde-Gallwitz et al., 2004). Work by others has shown isl1a is expressed in the central support cells of zebrafish neuromasts (Lush et al., 2019). Our work shows that α-Isl1-labels a subset of hair cells and additionally that loss of Foxg1a results in a decrease in the numbers of this cell population. It is also interesting that the isl1a expressing cells appear to be distinct from the dorsoventral cells sost:nlsEos cells and represent a separate group of Foxg1a sensitive hair cells. These data indicate there is possibly a role for isl1a in support cell progression to a hair cell fate that is at least in part regulated by foxg1a. Future work is needed to address linage tracing of central support cells to confirm their contribution to regenerating hair cells and to demonstrate the precise function of Foxg1a in these α-Isl1+ cells, particularly the target genes regulated by Foxg1a transcriptional function. It would also be interesting to investigate other known cellular functions of Foxg1a like autophagy, regulation of ROS, and cell cycle.
Concluding remarks
Our study reveals a novel function of Foxg1a in regulating cellular proliferation in a of neuromast cells during development and regeneration, and in regulating the proper number of mechanosensory hair cells. We believe this work provides greater insight into the cellular and molecular mechanisms driving vertebrate mechanosensory tissue development and may elucidate new opportunities in mammalian hair cell regeneration.
METHODS AND MATERIALS
Zebrafish lines and maintenance
The following D. rerio (zebrafish) lines were used: wild-type*AB (ZIRC; http://zebrafish.org), foxg1aa266 (Thyme et al., 2019), Tg(sost:nlsEos)w215 (Thomas and Raible, 2019), Tg(myosin6b:GFP)w186 (Thomas and Raible, 2019), Tg(sfrp1a:nlsEos)w217 (Thomas and Raible, 2019), Tg(prim:lyn2mCherry) (Wang et al., 2018), TgBAC(neurod:EGFP)nl1 (Obholzer et al., 2008), and (−4.9sox10:EGFP)ba2 (Dutton et al., 2009). Zebrafish were maintained and staged according to standard protocols (Kimmel et al., 1995). Experiments reported in this study were conducted on larvae between 24 hpf, and 8 dpf. Larvae were kept in E3 embryo medium (14.97 mM NaCl, 500 μM KCL, 42 μM Na2HPO4, 150 μM KH2PO4, 1 mM CaCl2 dihydrate, 1 mM MgSO4, 0.714 mM NaHCO3, pH 7.2). For all experiments, larvae were treated with tricaine (Syndel) prior to fixation in 4% paraformaldehyde/PBS (Thermo Fisher Scientific). All research was performed in accordance with the McGraw laboratory protocol #45344 approved by the UMKC IACUC committee. In laboratory zebrafish lines, sexual determination and differentiation has been shown to occur at ∼25 dpf (Kossack and Draper, 2019), after the timepoints analyzed in this study.
Wholemount RNA in situ hybridization
Wholemount RNA in situ hybridization (WISH) was carried out using established protocols (Thisse and Thisse, 2008), modified with a 5-min Proteinase K (Thermo Fisher Scientific) treatment to preserve neuromast integrity. The probes used were: atoh1a (Itoh and Chitnis, 2001), notch3 (Itoh and Chitnis, 2001), deltaD (Haddon et al., 1998), foxg1a, tnfsf10l3, sost, sfrp1a, lfng, gata2a, her4.1, her2, wnt2, wnt10a, ctnnb1, six1a, six1b and isl1a. Antisense probes were generated using established protocols (Thisse and Thisse, 2008) or using a PCR-based protocol (Logel et al., 1992).
HCR fluorescent RNA in situ hybridization
HCR FISH was carried out following the manufacturer's protocol (Molecular Instruments). The probes used were foxg1a-B2 (4 pmol) and tnfsf10l3-B2 (4 pmol), with the amplifier B2-647 (Molecular Instruments). Larvae were subsequently labeled with DAPI and mounted using Fluorescent Mounting Media (EMD Millipore) to prevent fading.
Immunohistochemistry, FM1-43FX and DASPEI labeling
Wholemount immunolabeling was performed using established protocols (Ungos et al., 2003) for all antibodies except when noted. The primary antibodies used were: α-Otoferlin antibody (α-Oto; mouse monoclonal, 1:200, DSHB, University of Iowa), α-Sox 2 antibody (rabbit polyclonal, 1:100, Invitrogen), α-Islet1 antibody (α-Isl1; mouse monoclonal, 1:100, DSHB, University of Iowa), and α-BrdU antibody (mouse monoclonal, 1:100, BD Biosciences). For α-Isl1 antibody labeling, larvae were fixed for 1-h at room temperature in %PFA and then incubated in water overnight at 4°C. Larvae in α-Isl1antibody were incubated for 2 days at room temperature. Secondary antibodies used were as follows: goat α-rabbit Alexa-647 antibody (1:1000, Invitrogen), goat α-mouse Alexa-568 antibody (1:1000, Invitrogen), and goat α-mouse Alexa-647 antibody (1:1000, Invitrogen). Mature hair cells were labeled by a 1-min incubation in 3 µM FM1-43FX (Invitrogen; Owens et al., 2008). Nuclei were labeled with 30 mM DAPI (Thermo Fisher Scientific). Antibody block was made per protocol using 2% goat serum. Hair cells were visualized in live larvae using 2-(4-(dimethylamino)styryl)-N-ethylpyridinium iodide (DASPEI; Invitrogen) according to established protocols (Harris et al., 2003).
TUNEL Labeling
TUNEL labeling was conducted using the Click-iT Plus TUNEL Assay Alexa Fluor 594 (Invitrogen, C10618). The kit protocol was adapted for larval zebrafish; larvae were fixed in 4% paraformaldehyde in PBS for 1 h at room temperature. Proteinase K digestion was done with 10 ug/mL proteinase K in PBS for 10 min. All wash steps were doubled to reduce background.
Neomycin exposure and regeneration
For hair cell ablation, 5 dpf larvae were incubated in 400 µM neomycin (NEO, Millipore-Sigma) in embryo medium for 0.5 h and then washed three times in fresh embryo medium. In experiments analyzing complete regeneration, larvae were collected 3 days post-NEO exposure. Shorter time periods of regeneration were also conducted per experimental need. For positive control TUNEL labeling following neomycin exposure, larvae were collected 1 h post NEO. For RNA in situ hybridization, larvae were collected 28 hpf, 2 dpf, and 5 dpf; for regeneration, larvae were collected at 3 h post NEO, 1 day post NEO, or 3 days post NEO.
BrdU incorporation
Cellular proliferation was analyzed using 5-bromo-2′-deoxyuridine (BrdU; Millipore Sigma). BrdU incorporation was carried out using established protocols (Harris et al., 2003; Laguerre et al., 2005); larvae were incubated in 10 mM BrdU for 24 h at set times during development or following NEO-induced ablation, larvae were exposed to BrdU immediately after NEO for 24 h and then transferred to fresh E3 for 2 days and fixed at 8 dpf.
Photoconversion
For photoconversion experiments using Tg(sost:nlsEos)w215 or Tg(sfrp1a:nlsEos)w217 fish, 5 dpf larvae were placed in a shallow depression slide and exposed to 405 nm light for 20 s using a Zeiss Imager.D2 compound microscope and a 10x objective. For regeneration experiments, photoconversion was carried out prior to NEO exposure. For live imaging of Tg(sost:nlsEos)w215 fish, larvae were anesthetized using tricaine and embedded in 1.2% low melt agarose/E3 embryo medium.
Image collection
For imaging of RNA in situ hybridization and immunohistochemistry, processed larvae were placed in 50% glycerol/PBS and mounted on slides. For imaging of HCR in FISH, larvae were mounted on slides in Fluorescent Mounting Media (Calbiochem) and imaged within 3 days of processing to prevent signal loss. Images were collected using a Zeiss 510 Meta LSM confocal microscope with Zen 2009 software or a Nikon AXR resonant scanning confocal microscope with NIS-Elements software. Images were processed using Fiji software (Schindelin et al., 2012) and brightness and contrast were adjusted using Photoshop (Adobe).
Quantification and statistical analysis
For quantification of neuromast cell numbers, we took the mean of individual neuromasts from multiple fish and conducted appropriate analyses comparing control and mutant samples. Neuromasts from L1-L4 were analyzed for each fish and cells were manually counted. All statistical analysis was carried out using GraphPad Prism 10 (GraphPad Prism version 10.0.0 for Mac, GraphPad Software, San Diego, California USA, www.graphpad.com). The data presented represents discrete variables so use of nonparametric tests determined. A Mann–Whitney U two-tailed nonparametric test was used for pair-wise comparisons, a Kruskal–Wallis test with Dunn's multiple comparison tests between multiple conditions, and a Fisher's exact test was used to compare condition across groups. Significance was set at P<0.05. All data are presented as±standard deviation (s.d.). A power analysis was conducted from initial samples of hair cell averages with an alpha of 0.05, and beta of 0.2, which established a minimum sample size of six per group for counts. Power calculated via ClinCal.com sample size calculator for two independent samples.
Acknowledgements
The authors would like to thank Dr. David Raible for providing the Tg(myosin6b:GFP)w186, Tg(sost:nlsEos)w215, and Tg(sfrp1a:nlsEos)w217 zebrafish lines. The authors would like to thank Dr. Alexander Schier for the foxg1aa266 zebrafish line. Animal care provided by the McGraw lab zebrafish facility.
Footnotes
Author contributions
Conceptualization: J.M.B., H.F.M.; Methodology: J.M.B.; Validation: J.M.B., H.F.M.; Formal analysis: J.M.B., H.F.M.; Investigation: J.M.B., E.M.T., M.M.V., C.B., R.H., H.F.M.; Resources: H.F.M.; Writing - original draft: J.M.B.; Writing - review & editing: J.M.B., H.F.M.; Visualization: J.M.B., H.F.M.; Supervision: H.F.M.; Project administration: H.F.M.; Funding acquisition: H.F.M.
Funding
Funding provided by National Institute of Health, NIGMS 1R16GM146690 (HFM). Open Access funding provided by the National Institute of General Medicine 1R6GM146690. Deposited in PMC for immediate release.
Parts of this manuscript and data are also published in portions of Jon M Bell's doctoral thesis.
Bell, J. M. (2024). Regulation of the development and regeneration of the zebrafish posterior lateral line (Order No. 31486729). Available from Dissertations & Theses at University of Missouri Kansas City. (3086769566). Retrieved from http://proxy.library.umkc.edu/login?url=https://www.proquest.com/dissertations-theses/regulation-development-regeneration-zebrafish/docview/3086769566/se-2.
Data availability
All relevant data can be found within the article and its supplementary information.
References
Competing interests
The authors declare no competing or financial interests.