Abstract

   The vestibular sensory epithelium of humans and mice may degenerate
   into a layer of flat cells, known as flat epithelium (FE), after a
   severe lesion. However, the pathogenesis of vestibular FE remains
   unclear. To determine whether the epithelial–mesenchymal transition
   (EMT) participates in the formation of vestibular FE, we used a
   well-established mouse model in which FE was induced in the utricle by
   an injection of streptomycin into the inner ear. The mesenchymal and
   epithelial cell markers and cell proliferation were examined using
   immunofluorescence staining and quantitative reverse transcription
   polymerase chain reaction (qRT-PCR). The function of the EMT was
   assessed through transcriptome microarray analysis. The results
   demonstrated that mesenchymal cell markers (α-SMA, S100A4, vimentin,
   and Fn1) were upregulated in vestibular FE compared with the normal
   utricle. Robust cell proliferation, which was absent in the normal
   status, was observed in the formation of FE. Microarray analysis
   identified 1,227 upregulated and 962 downregulated genes in vestibular
   FE. Gene Ontology (GO) analysis revealed that differentially expressed
   genes (DEGs) were highly associated with several EMT-related GO terms,
   such as cell adhesion, cell migration, and extracellular matrix.
   Pathway enrichment analysis revealed that DEGs were enriched in the
   EMT-related signaling pathways, including extracellular matrix
   (ECM)-receptor interaction, focal adhesion, PI3K/Akt signaling pathway
   and cell adhesion molecule. Protein–protein interaction networks
   screened 20 hub genes, which were Akt, Casp3, Col1a1, Col1a2, Fn1, Hgf,
   Igf1,Il1b, Irs1, Itga2, Itga5, Jun, Mapk1, Myc, Nras, Pdgfrb, Tgfb1,
   Thbs1, Trp53, and Col2a1. Most of these genes are reportedly involved
   in the EMT process in various tissues. The mRNA expression level of hub
   genes was validated using qRT-PCR. In conclusion, the present study
   indicates that EMT plays a significant role in the formation of
   vestibular FE and provides an overview of transcriptome characteristics
   in vestibular FE.

   Keywords: epithelial–mesenchymal transition, vestibular, microarray,
   cell proliferation, hair cell, supporting cell

Introduction

   Vestibular end organs, including the utricle, saccule, and cristae
   ampullae, are responsible for the perception of linear acceleration and
   head rotation. Sensory epithelia of vestibular end-organs consist of
   two kinds of highly differentiated cells: hair cells (HCs) and
   supporting cells (SCs). HCs and SCs are alternatively arranged in a
   special mosaic structure required for normal vestibular function.
   Various insults to the vestibular sensory epithelium could lead to
   vestibular dysfunction ([33]McCall et al., 2009; [34]Wang et al., 2015;
   [35]Brosel et al., 2016; [36]Isgrig et al., 2017; [37]You et al., 2018;
   [38]Zhang et al., 2020; [39]Fu et al., 2021). Severe lesions damage
   both vestibular HCs and SCs and induce the sensory epithelium to be
   replaced by a layer of flat cells, referred to as flat epithelium (FE)
   ([40]Wang et al., 2017). FE has been found in the inner ear of patients
   with severe deafness and/or vestibular dysfunction ([41]Nadol and
   Eddington, 2006; [42]Teufert et al., 2006; [43]McCall et al., 2009),
   suggesting that FE is an important pathological change in patients with
   inner ear diseases. However, the pathogenesis of vestibular FE remains
   unknown, and there is no biological intervention for patients with FE
   in the inner ear. Elucidation of the molecular mechanism underlying FE
   formation is significant for designing therapeutic strategies for
   vestibular dysfunction.

   The epithelial–mesenchymal transition (EMT) is a biological process
   (BP) that allows epithelial cells to acquire a mesenchymal cell
   phenotype, including migratory capacity, invasiveness, resistance to
   apoptosis, and increased production of extracellular matrix (ECM)
   components ([44]Kalluri and Weinberg, 2009). The EMT is integral in
   development and wound healing, and contributes pathologically to
   fibrosis and cancer progression ([45]Lamouille et al., 2014). In
   addition, the EMT participates in inner ear development and damage
   repair ([46]Simonneau et al., 2003; [47]Kobayashi et al., 2008;
   [48]Johnen et al., 2012; [49]Wu and Kelley, 2012). The EMT is involved
   in the formation of cochlear FE, which is characterized by a robust
   proliferative response, upregulation of mesenchymal cell markers, and
   cell migration ([50]Kim and Raphael, 2007; [51]Ladrech et al., 2017).
   Because the two components of the inner ear, cochlea and vestibular
   end-organs, share common embryonic origins and biological features, we
   hypothesize that the EMT also participates in the process of vestibular
   FE formation.

   The EMT is characterized by a change in cell phenotype from epithelial
   to mesenchymal cells with upregulation of mesenchymal cell markers
   (vimentin, α-SMA, S100A4, fibronectin, N-cadherin, etc.) and
   downregulation of epithelial cell markers (E-cadherin, cytokeratin, and
   ZO-1, etc.). Thus, these factors are usually used as biomarkers to
   define the involvement of EMT ([52]Kalluri and Weinberg, 2009).
   Recently, high-throughput screening, such as microarray and RNA-seq
   technologies, has enabled researchers to identify gene expression
   profiles in various diseases, rendering exploration of the underlying
   molecular mechanisms less difficult. The role of the EMT in diseases
   and the specific genes or signaling pathways involved have been
   explored using these techniques in the past decades ([53]Puram et al.,
   2018). However, whether EMT participates in the inner flattening
   process of vestibular sensory epithelium has not been identified.

   To determine the role of the EMT in the formation of vestibular FE, a
   high dose of streptomycin was inoculated into the mouse inner ear to
   induce FE in the utricle ([54]Wang et al., 2017). Mesenchymal and
   epithelial cell markers and cell proliferation were assessed in normal
   utricle and vestibular FE using immunofluorescence staining. Then, the
   mRNA expression profile was examined using microarray analysis.
   Bioinformatics analysis was used to further analyze the biological
   functions of differentially expressed genes (DEGs). Finally, the
   representative DEGs were validated using quantitative reverse
   transcription polymerase chain reaction (qRT-PCR). In the present
   study, the role of EMT in vestibular FE formation was investigated, and
   the potential mechanisms underlying this process were explored.

Results

Expression of Mesenchymal and Epithelial Cell Markers in Utricular Flat
Epithelium

   To determine the potential mechanisms underlying FE formation after the
   loss of nearly all original epithelial cells, the expression of
   mesenchymal and epithelial cell markers was examined using
   immunofluorescence staining and qRT-PCR in the normal utricle and
   utricular FE samples. As shown in [55]Figures 1A–D″, mesenchymal cell
   markers α-SMA and S100A4 were poorly expressed in normal utricle but
   highly expressed in FE. In contrast, epithelial cell marker ZO-1 was
   significantly expressed in the normal samples but weakly expressed in
   FE ([56]Figures 1E–F″). Furthermore, the mRNA expression levels of
   mesenchymal cell markers, S100A4, α-SMA, vimentin, and fibronectin 1
   (Fn1) were significantly higher in FE than in the normal utricle
   ([57]Figure 1G). The expression of epithelial cell markers (E-cadherin,
   ZO-1, keratin 5, and keratin 8) was not significantly different between
   the normal utricle and FE ([58]Figure 1H).

FIGURE 1.

   [59]FIGURE 1
   [60]Open in a new tab

   Expression of epithelial and mesenchymal cell markers in the normal
   utricle and flat epithelium (FE). Immunofluorescence staining of α-SMA
   (A–B″), S100A4 (C–D″), and ZO-1 (E–F″) showing the expression of α-SMA
   and S100A4 is upregulated and ZO-1 expression is downregulated in FE.
   High magnification images of square areas in (B,D,F) are shown in
   (B′–B″,D′–D″,F′–F″), respectively. Scale bars: (A) (applies to B–D,F),
   50 μm; (B′) (applies to B″,D′,D″,E,F′, F″), 20 μm. (G) qRT-PCR results
   revealing that the mRNA expression levels of mesenchymal cell markers
   (α-SMA, S100A4, vimentin, and Fn1) are significantly increased in FE
   compared with the normal utricle. (H) mRNA expression levels of
   epithelial cell markers (E-cadherin, ZO-1, keratin 5, and keratin 8)
   are not significantly different between FE and normal utricle. *P <
   0.05 and ^**P < 0.01 according to Student’s t-test.

Robust Mitosis in Adult Mouse Utricle After Severe Damage

   To evaluate if the utricular sensory epithelium possesses proliferation
   capacity during FE formation, the normal utricle and FE were stained
   with EdU to observe mitosis in the cells and with the epithelial cell
   marker E-cadherin to label the actin cytoskeleton. At 3 days after
   streptomycin injection, a few EdU-positive cells were observed
   ([61]Figures 2A–A′). At 7 days after the lesion, the number of
   EdU-positive cells was increased ([62]Figures 2B–B′). At 11 days after
   the lesion, most of the original sensory epithelium areas expressed
   E-cadherin, and EdU-positive cells were extensively distributed
   throughout the FE, indicating robust cell proliferation in the
   utricular FE during the early period of FE formation ([63]Figures
   2C–C′,E–E″′). At 22 days after the lesion, the epithelial cytoskeleton
   was completely formed, and the number of EdU-positive cells was
   dramatically decreased in the epithelial layer ([64]Figures 2D–D′).

FIGURE 2.

   [65]FIGURE 2
   [66]Open in a new tab

   Cell division at different timepoints after severe damage to the
   utricular sensory epithelium. (A–A′) At 3 days after the lesion, the
   actin cytoskeleton disappeared in most areas of the epithelial layer,
   with a few cells labeled by EdU. (B–B′) The number of EdU-positive
   cells increased at 7 days. (C–C′) Robust proliferation of EdU-positive
   cells was detected in flat epithelium (FE) at 11 days. (D–D′)
   EdU-positive cells were not observed in the epithelial layer at 22
   days. (E–E″′) High-magnification view of the square area in (C) showing
   EdU labeling of the nuclei of FE cells. Scale bars: (A) (applies to
   A′–D′), 50 μm; (E) (applies to E′–E″′), 20 μm. (F) qRT-PCR results
   revealing that the mRNA expression levels of the cell proliferation
   marker Ki-67 were significantly increased in FE compared with the
   normal utricle, and the p27^kip1 expression level was decreased at 14
   days after damage. *P < 0.05 and ^***P < 0.001 according to Student’s
   t-test.

   The expression levels of proliferation markers Ki-67 and MCM2 ([67]Chow
   et al., 2016; [68]Yousef et al., 2017), as well as the cell cycle
   inhibitor p27^kip1 ([69]Kim and Raphael, 2007), were evaluated and
   compared between the normal utricle and the utricular FE at 14 days
   after streptomycin injection. As shown in [70]Figure 2F, the mRNA level
   of Ki-67 was significantly increased and that of p27^kip1 decreased in
   FE compared with the control groups ([71]Figure 2F).

Microarray Analysis

   To further determine the characteristics of FE transcriptomes and how
   the EMT is involved in the repair process of utricular sensory
   epithelium after severe damage, microarray analysis was performed using
   the Affymetrix mouse Clariom S array to analyze the transcriptomic
   differences between the normal utricle and FE. A total of 22,206 genes
   were extracted from each sample. When comparing transcripts between the
   normal utricle and FE, 2,189 transcripts differentially expressed (fold
   change > 2, P < 0.05) in FE were identified. [72]Figures 3A,B show a
   volcano plot and hierarchical cluster analysis of the DEGs between the
   two groups; 1,227 upregulated and 962 downregulated genes were detected
   in FE samples, and heatmap analysis showed distinct differences in the
   mRNA expression profiles of the normal utricle and FE.

FIGURE 3.

   [73]FIGURE 3
   [74]Open in a new tab

   Microarray analysis of the normal utricle and flat epithelium (FE). (A)
   Volcano plot representing the whole transcriptome changes in FE
   compared with the normal utricle. (B) Hierarchical clustering showing
   the differentially expressed genes (DEGs). Each group has five
   replicates. Yellow represents the upregulated genes and blue represents
   the downregulated genes. (C) Top 100 upregulated genes in FE compared
   with the normal utricle. (D) Top 100 downregulated genes in FE compared
   with the normal utricle. The horizontal axis represents the expression
   value. The # symbol indicates genes associated with
   epithelial–mesenchymal transition (EMT).

   To characterize the genes most significantly differentially expressed
   between FE and normal utricle, the top 100 upregulated and
   downregulated genes were selected and are listed in [75]Figures 3C,D.
   Among the DEGs, those previously reported to be associated with EMT
   were labeled with the # symbol; Ibsp, Fn1, Gdf10, Lcn2, Loxl2, Htra1,
   C3, Lox, Postn, Aspn, Ncf4, Bmp5, Slpi, Anxa3, Mir675, Nkd2, Cd36,
   Timp1, Sulf2, Acp5, Csf1r, and Tgfb1 were upregulated in FE, whereas
   Bdnf and Wdr66 were downregulated in FE.

Gene Ontology Analysis

   Gene Ontology (GO) enrichment analysis was performed based on the DEGs.
   Among the upregulated genes, 616 significant BP, 121 cellular component
   (CC), and 153 molecular function (MF) GO categories were detected (P <
   0.01; [76]Supplementary Table 1). According to the BP category results,
   the DEGs were significantly associated with cell adhesion and
   migration. In the CC category, DEGs were mainly associated with
   extracellular components. Among the downregulated genes, 129
   significant BP, 59 CC, and 36 MF categories were detected (P < 0.01;
   [77]Supplementary Table 2). In the BP category, DEGs were mostly
   associated with inner ear development and function. In the CC category,
   DEGs were associated with membrane, cilium, and synapse. The top 20
   upregulated and the top 20 downregulated GO terms are shown in
   [78]Figure 4. Among these GO terms, 34 were associated with EMT.

FIGURE 4.

   [79]FIGURE 4
   [80]Open in a new tab

   Gene Ontology (GO) enrichment analysis of differentially expressed
   genes (DEGs) between the normal utricle and flat epithelium (FE). (A)
   Top 20 upregulated GO terms associated with the biological process (BP;
   red), cellular component (CC; yellow), and molecular function (MF;
   blue). (B) Top 20 downregulated GO terms. The # symbol indicates GO
   terms associated with the epithelial–mesenchymal transition (EMT).

Pathway Enrichment Analysis and Pathway Interaction Network Analysis

   Pathway enrichment analysis was performed based on the KEGG (Kyoto
   Encyclopedia of Genes and Genomes) database. Based on the upregulated
   and downregulated genes, 98 and 34 signaling pathways were detected,
   respectively (P < 0.05; [81]Supplementary Tables 3, [82]4). Among the
   top 40 significantly enriched signaling pathways ([83]Figures 5A,B), 4
   were associated with the EMT, including ECM–receptor interaction
   (mmu04512) ([84]Gonzalez and Medici, 2014), focal adhesion (mmu04510)
   ([85]Ji et al., 2019), PI3K/Akt signaling pathway (mmu04151) ([86]Xu et
   al., 2015), and cell adhesion molecules (mmu04514) ([87]Keller et al.,
   2019).

FIGURE 5.

   [88]FIGURE 5
   [89]Open in a new tab

   Pathway enrichment analysis and pathway interaction network analysis.
   (A,B) Pathway enrichment analysis showing the top 20 upregulated and
   top 20 downregulated signaling pathways. The # symbol indicates
   pathways associated with the epithelial–mesenchymal transition (EMT).
   (C) Pathway interaction network analysis. Nodes represent pathways, and
   the arrows represent an interaction target between pathways.

   Next, pathway interaction network analysis was performed to generate an
   interaction network encompassing 44 significantly altered pathways;
   each pathway in the network was measured by counting the upstream and
   downstream pathways ([90]Supplementary Table 5). A group of EMT-related
   signaling pathways was found to be closely associated with other
   pathways, including the MAPK signaling pathway (degree = 54), PI3K/Akt
   signaling pathway (degree = 41), TGF-β signaling pathway (degree = 17),
   NF-κB signaling pathway (degree = 16), regulation of actin cytoskeleton
   (degree = 16), and focal adhesion (degree = 16) ([91]Figure 5C).

Construction of the Protein–Protein Interaction Network and Screening of Hub
Genes

   The Search Tool for the Retrieval of Interacting Genes (STRING)
   database was used to construct a protein–protein interaction (PPI)
   network of selected genes. Genes involved in EMT-related signaling
   pathways ([92]Figure 5) were selected to build a network using
   Cytoscape (v3.7.2). All the nodes and edges were mapped in the PPI
   network, as shown in [93]Figure 6A. To screen hub genes from the entire
   PPI network, the Cytoscape plugin cytoHubba was used. A total of 20 hub
   genes were screened using the maximum neighborhood component (MNC)
   algorithm: Akt, Casp3, Col1a1, Col1a2, Fn1, Hgf, Igf1, Il1b, Irs1,
   Itga2, Itga5, Jun, Mapk1, Myc, Nras, Pdgfrb, Tgfb1, Thbs1, Trp53, and
   Col2a1 ([94]Figure 6B). Among those genes, 19 have been shown to
   participate in the EMT process in other tissues; however, an
   association of Col2a1 with EMT has not been reported ([95]Table 1).

FIGURE 6.

   [96]FIGURE 6
   [97]Open in a new tab

   Visualization of the protein–protein interaction (PPI) network and the
   candidate genes. (A) Entire PPI network. The edges indicate the PPIs in
   the Search Tool for the Retrieval of Interacting Genes (STRING)
   database. (B) Identification of the candidate genes from the entire PPI
   network using the maximum neighborhood component (MNC) algorithm. Edges
   represent the protein–protein associations. The orange nodes represent
   the genes that have been reported to be associated with the
   epithelial–mesenchymal transition (EMT), and gray nodes represent genes
   that have not been reported previously.

TABLE 1.

   mRNAs involved in epithelial–mesenchymal transition (EMT) and their
   associated diseases or potential mechanisms.
   Gene symbol Description (disease/mechanisms) References