Osteopontin-driven partial epithelial-mesenchymal transition governs the development of middle ear cholesteatoma

ABSTRACT

Cholesteatoma is a common disease of the middle ear. Currently, surgical removal is the only treatment option and patients face a high risk of relapse. The molecular basis of cholesteatoma remains largely unknown. Here, we show that Osteopontin (OPN), a predominantly secreted protein, plays a crucial role in the development of middle ear cholesteatoma. Global transcriptome analysis revealed the loss of epithelial features and an enhanced immune response in human cholesteatoma tissues. Quantitative RT-PCR and immunohistochemical staining of middle ear cholesteatoma validated the reduced expression of epithelial markers, as well as the elevated expression of mesenchymal markers including Vimentin and Fibronectin, but not N-Cadherin, α-smooth muscle actin (α-SMA) or ferroptosis suppressor protein 1 (FSP1), indicating a partial epithelial-mesenchymal transition (EMT) state. Besides, the expression of OPN was significantly elevated in human cholesteatoma tissues. Treatment with OPN promoted cell proliferation, survival and migration and led to a partial EMT in immortalized human keratinocyte cells. Importantly, blockade of OPN signaling could remarkably improve the cholesteatoma-like symptoms in SD rats. Our mechanistic study demonstrated that the AKT-zinc finger E-box binding homeobox 2 (ZEB2) axis mediated the effects of OPN. Overall, these findings suggest that targeting the OPN signaling represents a promising strategy for the treatment of middle ear cholesteatoma.

Introduction

Cholesteatoma is a disease characterized by excessive proliferation of keratinized squamous epithelial cells in the middle ear. Based on the underlying pathology, middle ear cholesteatoma can be categorized into congenital and acquired types. Congenital cholesteatoma is considered a developmental defect originating from the accumulation of epithelial remnants behind the tympanic membrane [1,2]. By contrast, acquired cholesteatoma is the consequence of an uncontrolled growth of the keratinized squamous epithelial cells migrating from the damaged tympanic membrane due to perforation, invagination, infection or other tissue damages, to the middle ear [3,4]. Whereas the lesions are benign, middle ear cholesteatoma can erode the nearby bone structures, leading to conductive and/or sensorineural hearing loss, labyrinthine fistula, facial paralysis, and intracranial complications [5–8]. The annual incidence of acquired cholesteatoma was estimated to be 6.8 to 10 per 100,000 in adults [9–11]. To date, surgical removal remains the only treatment option for this disease and patients face a high risk of relapse, given that the 5-year recurrence rates are approximately 15% in adults and 37% in children [12], and the cumulative recurrence rate at 10 years is greater than 40% [9]. Therefore, middle ear cholesteatoma contributes a significant amount of burden to health care.

Histologically, the lesion in middle ear cholesteatoma can be divided into three layers: the cystic content, the matrix, and the perimatrix [13]. The cystic content consists mainly of keratinized debris, lipid secretions and necrotic squamous epithelial cells. The matrix is principally composed of hyperproliferative stratified squamous epithelium, while the perimatrix is an inflamed subepithelial connective tissue containing collagen fibers, fibrocytes, as well as inflammatory cells [14]. Recent work is beginning to understand the molecular mechanisms of middle ear cholesteatoma, demonstrating that it is associated with aberrant cell proliferation and apoptosis, as well as inflammation [5,14,15]. Besides, these works showed that a variety of signaling pathways or molecules, including the keratinocyte growth factor [16,17], epidermal growth factor receptor [18,19], Yes-associated protein (YAP) [20,21], inhibitor of DNA-binding 1 [5,22,23], stem cell [24], as well as epigenetics [15], were involved in the development of cholesteatoma. However, the underlying molecular mechanisms remain incompletely understood.

In the current study, we profiled the transcriptomes of human acquired middle ear cholesteatoma and normal skin tissues. We show that cholesteatoma exhibited a partial epithelial-mesenchymal transition (EMT) and an overexpression of Osteopontin (OPN). Treatment with OPN promoted cell proliferation, survival as well as migration, and led to a partial EMT. Importantly, blockade of the OPN signaling ameliorated the cholesteatoma-like symptoms in SD rats. Hence, the OPN signaling may represent a relevant target for the management of middle ear cholesteatoma.

Materials and methods

Patient samples

Acquired middle ear cholesteatoma tissues were obtained from patients undergoing cholesteatoma surgery. Meanwhile, normal skin biopsies from postauricular incision were collected as controls. Tissue samples were preserved in liquid nitrogen immediately after dissection for RNA analysis, or in 4% paraformaldehyde for immunohistochemical staining. This study was approved by the Institutional Review Board of Tongji Hospital, Huazhong University of Science and Technology and all patients signed informed consent forms.

RNA sequencing

Total RNAs were extracted from 4 normal postauricular skin tissues and 3 middle ear cholesteatoma tissues using TRIzol reagent (Thermo Fisher, Waltham, MA). The concentration and integrity of RNA were assessed using a Bioanalyzer 2100 system with the RNA Nano 6000 Assay Kit (Agilent Technologies, Santa Clara, CA). Subsequently, double-stranded cDNA libraries were prepared using the NEBNext Ultra RNA Library Prep Kit for Illumina (New England Biolabs, Ipswich, MA) and sequenced using the NovaSeq 6000 system (Illumina, San Diego, CA).

Sequence reads were mapped to the reference human genome sequence (build 38) using HISAT2 (version 2.0.5). Read counts were determined using FeatureCounts (version 1.5.0-p3). Differential gene expression analysis was performed using the DESeq2 package (version 1.20.0) in R Studio. Fold change was calculated as the ratio between the means of normalized gene reads of cholesteatoma and normal skin tissues. The p values were adjusted (padj) using the Benjamini and Hochberg’s approach for controlling the false discovery rate. Genes with fold change >3 (which means fold change >3 for upregulation or fold change <1/3 for downregulation) and padj < 0.01 were considered differentially expressed. Gene ontology (GO) analysis was performed using the clusterProfiler R package (version 3.8.1) on differentially expressed genes.

Cell culture and transfection

HaCaT cells were cultured in RPMI 1640 medium containing 10% fetal bovine serum and penicillin-streptomycin (penicillin: 100 U/ml, streptomycin: 100 μg/ml) at 37°C with 5% CO₂. For OPN treatment, HaCaT cells were incubated in medium with or without OPN (100 ng/ml, Cat# HY-P70499, MedChemExpress, Shanghai, China). The cells were harvested 24 or 36 hrs later for RNA or Western blot analysis, respectively. To suppress ZEB2 expression, an siRNA mixture (RiboBio, Guangzhou, China) targeting human ZEB2 was transfected into HaCaT cells using Lipofectamine 3000 transfection reagent (Thermo Fisher). Scramble RNAs (RiboBio) were also transfected into HaCaT cells, separately, to be used as experimental controls. The siRNA sequences targeting ZEB2 were as followings: 5’-GGA GUU ACU UCU CCU AAU A-3’, 5’-GAA GCU ACG UAC UUU AAU A-3’, and 5’-GCA CUA GUC CCU UUA UGA A-3’. Twenty-four hours after transfection, the cells were incubated in fresh culture medium with or without OPN (100 ng/ml). The cells were cultured in the incubator for an additional 24 or 36 hrs for RNA or Western blot analysis.

Quantitative RT-PCR

Total RNAs of HaCaT cells and tissues were extracted using TRIzol reagent (Thermo Fisher). The concentration and quality of RNA were determined using a NanoDrop ND-2000 spectrophotometer (Thermo Fisher). Total RNAs were then reverse-transcribed to cDNA using M-MLV reverse transcriptase (Promega, Madison, WI). After combining cDNAs, gene-specific primers (Supplementary Table S1), and power SYBR green PCR master mix (Thermo Fisher), PCR was carried out on an ABI 7900HT real-time PCR system (Thermo Fisher). Relative gene expression levels were calculated using the 2^−ΔCt method, where ΔCt is the difference between the Ct value of a given gene and that of GAPDH control.

Immunohistochemistry

Tissue sections at the thickness of 5 μm were deparaffinized in xylene and rehydrated in graded alcohol. Antigen retrieval was carried out in sodium citrate buffer (10 mM, pH 6.0). The endogenous peroxidase activity was quenched using 3% hydrogen peroxide at ambient temperature for 20 min, and then the tissue sections were blocked using 5% normal goat serum for 30 min. Tissue sections were incubated with primary antibodies at 4°C overnight. The primary antibodies were mouse monoclonal antibody against Ki67 (Cat# ZM-0166, 1:100, Zsbio) or α-SMA (Cat# ZM-0003, 1:200, Zsbio), and rabbit polyclonal antibodies against E-Cadherin (Cat# 20874–1-AP, 1:100, Proteintech), ZO-1 (Cat# ab216880, 1:100, Abcam), Occludin (Cat# ab222691, 1:200, Abcam), Claudin 1 (Cat# AF0127, 1:100, Affinity Biosciences), Vimentin (Cat# 10366–1-AP, 1:200, Proteintech) or OPN (Cat# 22952–1-AP, 1:100, Proteintech). Subsequently, tissue sections were incubated with biotin-conjugated secondary antibodies and HRP-conjugated streptavidin. The antibody-biotin-HRP complexes were visualized using 3,3’-diaminobenzidine (DAB). Images were acquired using a brightfield microscope (Olympus, Tokyo, Japan), and were analyzed using ImageJ (NIH, Bethesda, MD). The staining density and the percentage of protein-expressing cells were considered [25].

Western blot

Cells were harvested and lysed in RIPA buffer. Protein concentration was determined using a BCA assay. Proteins were separated on SDS-PAGE gels, and then were transferred to PVDF membranes. The membranes were blocked using 5% nonfat milk, and subsequently were incubated with primary antibodies at 4°C overnight. The following primary antibodies were used: E-Cadherin (Cat# 20874–1-AP, 1:5000, Proteintech), Vimentin (Cat# 10366–1-AP, 1:2500, Proteintech), p-AKT (Cat# 4060S, 1:1000, Cell Signaling Technology), AKT (Cat# 2938S, 1:2000, Cell Signaling Technology), ZEB2 (Cat# ab138222, 1:500, Abcam) and GAPDH (Cat# 10494–1-AP, 1:10000, Proteintech). After washing, the blots were incubated with HRP-conjugated secondary antibodies at room temperature for 1 hr. The membranes were then exposed to the supersignal west femto maximum sensitivity substrate (Thermo Fisher), and chemiluminescence was acquired. Densitometry of the blots was performed using the ImageJ software. Uncropped blots can be found in the supplemental information (Supplementary Figure S3–5).

Cell proliferation assay

HaCaT cells were seeded into a 96-well plate, incubated overnight, and then were treated with or without OPN (100 ng/ml). In a separate experiment, HaCaT cells were transfected with scramble RNAs or siZEB2 RNAs. After 24 hrs, cell culture medium was replaced by fresh one with or without OPN (100 ng/ml). Cell proliferation was assessed every 24 hrs using a CCK-8 assay kit (Cat# E606335, Sangon Biotech, Shanghai, China) according to the manufacturer’s instructions. Briefly, cells were supplemented with CCK-8 solution and were then incubated at 37°C, 5% carbon dioxide for 1 hr. Thereafter, the optical density at 450 nm was measured using a microplate reader.

TdT-mediated dUTP nick end labeling (TUNEL) assay

HaCaT cells were seeded on coverslips and incubated at 37°C, 5% carbon dioxide for 24 hrs. Cells were then treated with or without OPN (100 ng/ml) for an additional 48 hrs. To inhibit ZEB2 expression, HaCaT cells were transfected with scramble RNAs or siZEB2 RNAs for 24 hrs. Cell culture media were then replaced by fresh one with or without OPN (100 ng/ml) and the cells were incubated at 37°C, 5% carbon dioxide for another 48 hrs. Cell apoptosis was assessed using a CoraLite 488 TUNEL assay kit (Cat# PF00006, Proteintech, Wuhan, China) following the manufacturer’s instructions. Briefly, cells were fixed with 4% paraformaldehyde at 4°C for 30 min, and then were permeabilized with 70% ethanol at −20°C for 4 hrs. Subsequently, cells were incubated in equilibration buffer at room temperature for 5 min, and the reaction mixture at 37°C for 1 hr. After washing, cells were counterstained with DAPI to reveal cell nucleus. Image acquisition was performed using a fluorescence microscope (Olympus, Tokyo, Japan). The percentage of TUNEL-positive cells were assessed.

Cell migration assay

HaCaT cells were seeded into 6-well plates and were then incubated at 37°C, 5% carbon dioxide for 24 hrs. Subsequently, uniform scratches were made on the monolayer cell cultures and the media were replaced by fresh one supplemented with or without OPN (100 ng/ml). Images of cells were recorded using a microscope and then gap widths were determined. To suppress ZEB2 expression, HaCaT cells were transfected with siZEB2 RNAs, or scramble RNAs as experimental control, for 24 hrs. Scratches were generated and the media were replaced by fresh one supplemented with or without OPN (100 ng/ml). Images of cells were then recorded and gap widths were determined.

Rat cholesteatoma model

Six weeks old male Sprague-Dawley (SD) rats were randomly assigned to one of three treatment groups. Rats in group 1 was the normal control without any intervention. In groups 2 and 3, both ears of rats underwent intratympanic injections of 0.2 ml of 100% propylene glycol (Cat# A610450, Sangon Biotech) on days 1, 8, and 15 to induce cholesteatoma in the middle ear. On the 12th and 13th weeks of the experiment, 0.2 ml of saline or verbascoside (1.87 μg/ear, Cat# HY-N0021, MedChemExpress) [26], a CD44 inhibitor, was administered into the middle ears of rats in groups 2 or 3, respectively. During these procedures, rats were anaesthetized by intraperitoneal injection of avertin (300 mg/kg, Sigma-Aldrich, St. Louis, MO). The animals were sacrificed in the 15th week of the experiment and both tympanic bullas were dissected out. Animal experiment was approved by the IACUC of Huazhong University of Science and Technology.

The bullas were fixed with 4% formaldehyde, decalcified with 10% EDTA, and then were embedded in paraffin. Tissues were sectioned at the thickness of 5 μm, and stained with hematoxylin and eosin. Images were acquired with a brightfield microscope.

Statistical analysis

Data are presented as mean ± SEM or box and whisker plots. Data distribution was assumed to be normal but this has not been formally tested. The comparisons between two groups were performed with two-tailed Student’s t-test. Comparisons among four groups were performed using one-way ANOVA with Bonferroni’s post hoc test. Cell growth curve and migration data were analyzed using two-way ANOVA with Bonferroni’s post hoc test. A p value less than 0.05 was considered statistically significant.

Results

Transcriptomic analysis reveals the differentially expressed genes between human middle ear cholesteatoma and the normal skin tissues

In order to investigate the mechanism of cholesteatoma development, we profiled the transcriptomes of human middle ear cholesteatoma and normal postauricular skin tissues (as control) using RNA sequencing. As shown in Figure 1(a,b), 2.17% or 6.55% of genes were significantly upregulated or downregulated in cholesteatoma compared with normal skin tissue, respectively. Notably, expression of OPN, which has a pleotropic function in physiological and pathological processes, was significantly upregulated in cholesteatoma tissue Figure 1(b). The gene ontology (GO) analysis indicated that the downregulated genes could affect various biological processes, including epidermis development, keratinocyte differentiation, keratinization and epidermal cell differentiation, and the upregulated genes were involved in immune cell activity, including neutrophil-mediated immunity, leukocyte migration and phagocytes Figure 1(c,d). Molecular function analysis showed that the downregulated genes were enriched in channel activity, intermediate filament binding and cell adhesion molecule binding, and that the upregulated genes were enriched in chemokine activity, chemokine receptor binding as well as cytokine activity Figures 1(e,f). The cellular component analysis demonstrated that the downregulated genes were enriched in intermediate filament, keratin filament, extracellular matrix and cell-cell adherent junction, while the upregulated genes were enriched in the tertiary granule, secretory granule lumen, primary lysosome, endocytic vesicle and lysosomal membrane (Supplementary Figure S1). Together, these data suggest that human middle ear cholesteatoma is associated with severe inflammation, aberrant keratinocyte differentiation, and alterations in cellular structure.

Figure 1.

Global transcriptome profiling reveals the differentially expressed genes between middle ear cholesteatoma and normal skin tissues. (a) pie chart showing the percentages of upregulated, downregulated, and unchanged genes in cholesteatoma versus normal skin tissue. (b) volcano plot depicting the differentially expressed genes (DEGs) between cholesteatoma and normal skin tissues. C, D. the gene ontology (GO): biological process analysis of downregulated (c) and upregulated genes (d) E, F. the GO: molecular function analysis of downregulated (e) and upregulated genes (f).

Partial epithelial-mesenchymal transition and elevated expression of OPN in the middle ear cholesteatoma

Next, we validated the differentially expressed genes using quantitative real-time PCR. The data demonstrate that the expression of epithelial marker genes, such as E-Cadherin, ZO-1, Occludin, Claudin 1, Desmoglein 1, Desmocollin 1 and Desmoplakin 1, were significantly downregulated Figure 2(a–g). By contrast, the expression of mesenchymal marker genes, such as Vimentin, Fibronectin, and TGF-β1, were significantly upregulated in cholesteatoma tissues in comparison with the normal controls Figure 2(h,k,n). Notably, the mRNA levels of N-Cadherin, α-SMA and FSP1 were not significantly impacted in cholesteatoma Figure 2(i,j,l). These data suggest that a partial epithelial-mesenchymal transition (EMT) occurs in the cholesteatoma of middle ear. The EMT refers to the process in which epithelial cells gradually lose their characteristics and functions, and acquire the characteristics of mesenchymal cell [27,28], and the partial EMT refers to the state in which a cell loses the properties of epithelial cell, and has not fully obtained the properties of mesenchymal cell [29]. Consistent with the RNA-seq data, the mRNA level of OPN was markedly increased in cholesteatoma Figure 2(m), suggesting a role in its development.

Figure 2.

Partial EMT and overexpression of OPN in human middle ear cholesteatoma. A-G. the relative mRNA levels of epithelial marker genes, including E-Cadherin (a), ZO-1 (b), Occludin (c), Claudin 1 (d), Desmoglein 1 (e), desmocollin 1 (f) and Desmoplakin 1 (g), were assessed between cholesteatoma and normal skin tissues by qRT-PCR. n = 5 per group. H-L. the relative mRNA levels of mesenchymal marker genes, including Vimentin (h), N-Cadherin (i), α-SMA (j), Fibronectin (k) and FSP1 (l), were assessed between cholesteatoma and normal skin tissues by qRT-PCR. n = 5 per group. M, N. the relative mRNA levels of OPN (m) and TGF-β1 (n) were assessed in cholesteatoma versus normal skin tissues by qRT-PCR. n = 5 per group.

Data are presented as box and whisker plots. Lower and upper hinges, first and third quartiles; the horizontal line, median; the whiskers extend from minimal to maximal values. *P < 0.05, **P < 0.01, two-tailed Student’s t-test.

Subsequently, we verified the RNA-seq data using immunohistochemical (IHC) staining. We assessed cell proliferation using IHC for Ki67, a marker for cell proliferation. As expected, the number of Ki67-positive cells was markedly elevated in the cholesteatoma tissues in comparison with that in normal skin tissues Figure 3(a,b). Besides, in agreement with qRT-PCR data, the protein levels of E-Cadherin, ZO-1, Occludin and Claudin 1 were significantly reduced, whereas the protein levels of Vimentin and OPN were dramatically elevated in cholesteatoma tissues comparing with that in normal skin tissues Figure 3(a–d,g,h). Of note, the protein level of α-SMA was not significantly impacted in cholesteatoma Figure 3(e,f). Overall, these data suggest that there exist a partial EMT and an elevated expression of OPN in the middle ear cholesteatoma.

Figure 3.

(a) immunohistochemical staining for Ki67, E-Cadherin, ZO-1, Occludin and Claudin 1 of normal skin and human middle ear cholesteatoma tissues. Scale bars, 40 μm. (b) box and whisker plots showing the relative protein levels of Ki67, E-Cadherin, ZO-1, Occludin and Claudin 1 in middle ear cholesteatoma relative to the normal skin tissues. n = 12 ~ 20 (normal skin) or 9 ~ 15 (cholesteatoma). C, D Representative images of immunohistochemical staining for vimentin of cholesteatoma versus normal skin tissue (c). In panel d, the relative protein level of vimentin is displayed. Dashed lines indicate the border of matrix and perimatrix. Scale bars, 40 μm or 10 μm (magnified images). n = 19 (normal skin) or 11 (cholesteatoma). E, F. Representative images of immunohistochemical staining for α-SMA (e) and the quantification of immunostaining (f) Scale bars, 40 μm. NS, not significant. n = 18 (normal skin) or 12 (cholesteatoma). G, H. Representative images of immunostaining for OPN (g) and the quantification of immunostaining (h) Scale bars, 40 μm or 10 μm (magnified images). n = 19 (normal skin) or 14 (cholesteatoma).

Data are presented as box and whisker plots. Lower and upper hinges, first and third quartiles; the horizontal line, median; the whiskers extend from minimal to maximal values. *P < 0.05, two-tailed Student’s t-test.

OPN treatment leads to a partial EMT state, and promotes cell proliferation and survival in HaCaT cells

OPN, also known as secreted phosphoprotein 1 (SPP1), has various functions in physiological and disease conditions. To establish a role of OPN in cholesteatoma development, we assessed its effect on HaCaT cells, an immortalized human keratinocyte cell line. We treated HaCaT cells with vehicle or OPN for 24 hours, and then assessed the mRNA levels of epithelial and mesenchymal marker genes. The qRT-PCR data showed that OPN treatment could decrease the mRNA levels of E-Cadherin, ZO-1, Occludin and Claudin 1, whereas increase the mRNA levels of Vimentin and Fibronectin Figure 4(a–e, h). Treatment with OPN did not significantly impact the expression of N-Cadherin, α-SMA or FSP1 Figure 4(f,g,i). These data suggest that an acute treatment with OPN was sufficient to elicit a partial EMT state in HaCaT cells. In line with the qRT-PCR results, Western blot analysis showed that E-Cadherin protein level was reduced, and Vimentin protein level was elevated following OPN treatment Figure 4(j,k), suggesting the occurrence of partial EMT. The CCK-8 and TUNEL assays displayed that OPN treatment could promote cell proliferation and inhibit cell apoptosis in HaCaT cells Figure 4(l–n). Besides, OPN treatment could enhance cell migration in HaCaT cells Figure 4(o). Together, OPN treatment leads to a partial EMT state, and promotes cell proliferation and survival in immortalized human keratinocytes.

Figure 4.

Treatment with OPN leads to a partial EMT state, and promotes cell proliferation, survival as well as migration in HaCaT cells. A-I. HaCaT cells were incubated with or without OPN (100 ng/ml) for 24 hrs. The relative mRNA levels of E-Cadherin (a), ZO-1 (b), Occludin (c), Claudin 1 (d), Vimentin (e), N-Cadherin (f), α-SMA (g), Fibronectin (h), and FSP1 (i) were then analyzed. Lower and upper hinges, first and third quartiles; the horizontal line, median; the whiskers extend from minimal to maximal values. n = 6 (control) or 5 (Osteopontin). J, K. HaCaT cells were incubated with or without OPN (100 ng/ml) for 36 hrs, and then were harvested. Total proteins were prepared and Western blots for E-Cadherin and Vimentin were performed (j) and quantified (k). GAPDH was used as a loading control. n = 3 per group. (l). growth curve of HaCaT cells treated with or without OPN (100 ng/ml) was assessed using a CCK-8 assay. Day 0, time immediately before OPN treatment. n = 5 per group. M, N. HaCaT cells were incubated with or without OPN (100 ng/ml) for 48 hrs. TUNEL assay (green) was then performed to reveal apoptotic cells (m), and the percentage of TUNEL-positive cell was determined (n). Scale bars, 100 μm. n = 4 per group. (o). treatment with OPN promoted the migratory ability of HaCaT cells. HaCaT cells were seeded in 6-well plates. Twenty-four hours later, uniform scratches were made on the bottom of cell cultures and medium was replaced by fresh one with or without OPN. Gap width was determined using a brightfield microscope. n = 4 per group.

Data are presented as box and whisker plots (a-i). Lower and upper hinges, first and third quartiles; the horizontal line, median; the whiskers extend from minimal to maximal values. In panels k-o, data are presented as mean ± SEM. *P < 0.05, **P < 0.01, two-tailed Student’s t-test (a-k, n), two-way ANOVA with Bonferroni’s post hoc test (l, o).

ZEB2 mediated the effects of OPN on HaCaT cells

Subsequently, we asked how OPN elicits partial EMT and impacts the proliferation and survival in HaCaT cells. OPN is known to bind CD44 receptors [30–33]. Within the cell, CD44 regulates the activity of AKT [34,35], which plays a critical role in EMT via controlling the expression of zinc finger E-box binding homeobox 1 (ZEB1) and ZEB2. Thereby, we hypothesized that the AKT-ZEB signaling mediates the effects of OPN on HaCaT cells. To initially examine this possibility, we assessed the expression of ZEB1 and ZEB2 in cholesteatoma and HaCaT cells. As shown in Figure 5(a,b), in comparison with the normal skin tissues, the mRNA level of ZEB2 was elevated in cholesteatoma tissues, whereas there was no significant difference for ZEB1 between these two groups. Moreover, after treatment with OPN, the mRNA level of ZEB2, but not ZEB1, was markedly increased Figure 5(c, d). Subsequently, we assessed the protein levels of p-AKT, AKT1, and ZEB2 in HaCaT cells with or without OPN treatment for 36 hours. As demonstrated in Figure 5(e,g), ZEB2 protein level was increased in HaCaT cells following OPN treatment. In addition, OPN treatment could activate AKT in HaCaT cells, as demonstrated by the increased phosphorylation level of AKT at Ser473 Figure 5(e,f).

Figure 5.

The AKT-ZEB2 signaling mediates the effects of OPN on HaCaT cells. A, B. the mRNA level of ZEB2 (b), but not ZEB1 (a) was elevated in human middle ear cholesteatoma comparing with the normal skin tissue. n = 5 per group. C, D. HaCaT cells were incubated with or without OPN (100 ng/ml) for 24 hrs. The mRNA levels of ZEB1 (c) and ZEB2 (d) were then determined using qRT-PCR. n = 6 (control) or 5 (osteopontin). E-G. HaCaT cells were treated with or without OPN (100 ng/ml) for 36 hrs. Western blots for p-AKT (S473), AKT1 and ZEB2 were then performed (e) and quantified (F, G). GAPDH was used as a loading control. n = 3 per group. H-K. HaCaT cells were transfected with scramble RNAs as experimental control, or the siRNAs targeting ZEB2 (siZEB2) for 24 hrs, and then were treated with or without OPN (100 ng/ml) for an additional 24 hrs. The mRNA levels of E-Cadherin (h), ZO-1 (i), Vimentin (j) and Fibronectin (k) were analyzed using qRT-PCR. n = 6 per group. L, M. HaCaT cells were transfected with scramble RNAs or the siRNAs targeting ZEB2 (siZEB2) for 24 hrs. Subsequently, the cells were treated with or without OPN (100 ng/ml) for 36 hrs. Western blots were performed to determine the protein levels of E-Cadherin, Vimentin, p-AKT (S473), AKT1 and ZEB2 (l). GAPDH was used as a loading control. In panel M, quantifications of the blots are presented. NS, not significant. n = 3 per group. (n). HaCaT cells were transfected with scramble RNAs or siRNAs targeting ZEB2 (siZEB2) for 24 hrs, and then were treated with or without OPN (100 ng/ml) for 72 hrs. Growth curve was assessed using the CCK-8 assay. Day 0, time immediately before OPN treatment. n = 5 (control) or 6 (Osteopontin). (o). HaCaT cells were transfected with scramble RNAs or siZEB2 RNAs for 24 hrs. The cells were then incubated with or without OPN (100 ng/ml) for 48 hrs. TUNEL assay was carried out to reveal apoptotic cells and the percentage of TUNEL-positive cells is shown. NS, not significant. n = 3 (scramble RNA, ctrl) or 4 (all other groups) per group. (p). HaCaT cells were transfected with scramble RNAs or siRNAs targeting ZEB2 (siZEB2) for 24 hrs. Thereafter, uniform scratches were made on the bottom of the cell cultures and the medium was replaced by fresh one with or without OPN. Gap width was determined using a brightfield microscope. n = 4 per group.

Data are presented as box and whisker plots (a-d, h-k). Lower and upper hinges, first and third quartiles; the horizontal line, median; the whiskers extend from minimal to maximal values. Data are presented as mean ± SEM (f, g, m-p). *P < 0.05, two-tailed Student’s t-test (a-g), one-way (h-k, m, o) or two-way (n, p) ANOVA with Bonferroni’s post hoc test. In panels n and p, asterisks indicate the comparisons between scramble RNA, OPN and other groups.

After the expression of ZEB2 in HaCaT cells was suppressed using an siRNA mixture targeting the mRNA of this gene, OPN-induced downregulation of E-Cadherin and ZO-1, as well as the upregulation of Vimentin and Fibronectin were significantly abolished Figure 5(h–k). Further, Western blot analysis supported the qRT-PCR results Figure 5(l,m). In cell biology, the CCK-8 assay demonstrated that knockdown of ZEB2 largely attenuated the pro-proliferative effect of OPN on HaCaT cells Figure 5(n). The results of TUNEL and migration assays indicated that knockdown of ZEB2 blunted the pro-survival as well as the pro-migration effects of OPN on HaCaT cells Figure 5(o,p), (Supplementary Figure S2). Hence, these data suggest that the AKT-ZEB2 signaling mediates the effects of OPN on HaCaT cells.

Blockade of OPN signaling inhibits cholesteatoma development in rats

Next, we asked whether blockade of OPN signaling could suppress the development of middle ear cholesteatoma in SD rats. As CD44 is a prominent receptor for OPN, we chose to inhibit this molecule via intratympanic injection of verbascoside [26]. Intratympanic injection of propylene glycol could successfully induce cholesteatoma in male adult SD rats [36,37], as demonstrated by the presence of keratinized debris and accumulation of leukocytes in the middle ear cavity (Figure 6, middle panel). Treatment with verbascoside could remarkably ameliorate these symptoms (Figure 6, right panel). Thus, blockade of OPN-CD44 signaling could ameliorate the development of middle ear cholesteatoma in vivo.

Figure 6.

Blockade of CD44 inhibits the development of middle ear cholesteatoma in SD rats. Cholesteatoma was induced in male adult SD rats using propylene glycol (middle and right panels). Saline or verbascoside, a CD44 inhibitor, was administered into the middle ear cavity of rats. After treatment, bullas were dissected out and H&E staining was performed. Left panel, the middle ear of SD rat without any intervention. Middle panel, the middle ear of SD rat administered with saline. Right panel, the middle ear of SD rat treated with the CD44 inhibitor. Black stars indicate the middle ear cavity. Yellow stars indicate the pronounced inflammation in the middle ear cavity. Note the accumulation of leukocytes (middle panel). Red arrows point to the cochlea. Dark arrows indicate cholesteatoma characterized by the presence of keratinized debris in the middle ear cavity. n = 6 per group.

Discussion

Cholesteatoma is a common middle ear disease, yet, the molecular mechanisms of its pathogenesis are less well understood. Here, we show that there exist a partial EMT and an overexpression of OPN in human cholesteatoma tissues. Treatment with OPN leads to partial EMT and promotes cell proliferation, survival and migration of HaCaT cells, an immortalized human keratinocyte cell line. We demonstrate that these effects are mediated by the AKT-ZEB2 signaling. Importantly, blockade of OPN signaling improves the cholesteatoma-like symptoms in rats. Thus, our findings suggest that the OPN signaling plays a critical role in the development of middle ear cholesteatoma.

Recent progress in understanding the pathogenesis of middle ear cholesteatoma

Over the past few years, significant research efforts have been directed toward understanding the pathogenesis of middle ear cholesteatoma. For example, a previous study has shown that the number of Wnt5a-positive cells had increased and YAP nuclear translocation was detected in epithelial and mesenchymal cells in human middle ear cholesteatoma tissues [20]. ROCK (Rho-kinase) protein level and kinase activity were both reduced in human middle ear cholesteatoma [38]. In addition, the expression of prostaglandin E2 receptor EP4 was significantly inhibited in human acquired middle ear cholesteatoma [39]. The expression of β-Catenin, a component of the Wnt (wingless-related integration site) signaling pathway, was evidently increased in human middle ear cholesteatoma tissues [40]. Furthermore, treatment with Wnt3a promoted while the inhibition of Wnt signaling suppressed cholesteatoma cell growth [40].

Using RNA-seq or 16S rRNA-seq, prior studies have demonstrated that cholesteatoma was associated with dysregulations in macrophages [41], m⁶A-modified mRNAs [42], and bacteria [43]. Besides, the production of nitrogen oxide was reduced in human middle ear cholesteatoma [44]. A study profiled the proteome of cholesteatoma and demonstrated that α-synuclein played a role in the development of middle ear cholesteatoma [45]. Additionally, chronic inflammation seemed to have a crucial role in the recurrence of middle ear cholesteatoma [46].

Previous studies have also assessed the role of keratinocyte growth factor (KGF) in the development of middle ear cholesteatoma. KGF could stimulate stem cell proliferation in the epidermis via enhancing p63 phosphorylation [47]. Additionally, KGF was involved in the formation of cholesteatoma by promoting the growth of p75-positive neural crest lineage cells [48]. The YAP protein [21] and FOXC2 [49] exhibited significant roles in the development of cholesteatoma.

Partial EMT and the development of middle ear cholesteatoma

The EMT refers to the process in which epithelial cells gradually lose their characteristics and functions, and acquire the characteristics of mesenchymal cell [27,28]. Epithelial cells undergo a series of changes during EMT. The apical-basal polarity is gradually lost and the front-rear polarity is established. Intercellular junctions, including tight junctions, adhesive junctions, desmosomes, and hemidesmosomes, are disintegrated. Cells gradually change from a pebble-like shape to a spindle-like shape. The expression of epithelial cell markers, such as E-Cadherin, cytokeratin, and ZO-1, are reduced. By contrast, expression of mesenchymal cell markers, such as N-Cadherin, Vimentin and Fibronectin, are elevated. Besides, cells exhibit increased capabilities of proliferation, migration, and invasion. In physiology, EMT occurs in the process of embryonic development. It also plays a role in the processes of wound healing and tissue fibrosis [50]. When cancer cells obtain the properties of mesenchymal cell, they tend to resist apoptosis, and have increased capacities to invade and metastasize [51]. Previous work has demonstrated that various signaling molecules/pathways are involved in the EMT process, such as epidermal growth factor (EGF), fibroblast growth factor (FGF), hepatocyte growth factor (HGF), the Wnt signaling pathway, and the Notch signaling pathway. Besides, transcription factors including TWIST1, ZEB1 and ZEB2 [52], which inhibit the expression of E-Cadherin and promote the switch to the mesenchymal cell state, play a critical role in EMT.

Recent studies demonstrate that EMT is not a binary process. Indeed, cells can enter a partial EMT state during the processes of development or wound healing. Partial EMT, also known as hybrid EMT, refers to the state in which a cell loses the properties of epithelial cell, and has not fully obtained the properties of mesenchymal cell [29]. Cancer cells in a partial EMT state have increased capacities of cell proliferation and metastasis [29,53,54]. In the current study, we show that, the expression of epithelial cell markers including E-Cadherin, Occludin, ZO-1, Claudin, Desmoglein 1, Desmocollin 1, and Desmoplakin 1, was significantly reduced in cholesteatoma tissues in comparison with the normal postauricular skin tissues. By contrast, the expression of the mesenchymal cell markers including Vimentin and Fibronectin, but not N-Cadherin, α-SMA or FSP1, were significantly increased. These data demonstrate that there exists a partial EMT in human middle ear cholesteatoma.

Emerging evidence suggests that partial EMT plays a vital role in tumorigenesis, invasion, metastasis, and tumor stemness. For example, there was partial EMT in the leading edge of head and neck cancer, and this was postulated to be involved in the invasion process of tumor cells [55]. The partial EMT state also existed in skin and mammary primary tumors and was associated with an increased metastatic potential of tumor cells [56]. Furthermore, a previous study demonstrated that in mouse models of skin squamous cell carcinoma and lung tumor, disruption of Fat1 promoted a partial EMT phenotype [57]. The cancer cells exhibited increased tumor stemness and spontaneous metastasis. Notably, partial EMT also occurs in nonmalignant diseases. Indeed, renal epithelial cells undergo a partial EMT during the course of kidney fibrosis [58]. Using mouse model, a prior study demonstrated that BMP-7 could ameliorate the partial EMT in the diabetic kidney disease [59]. In the liver, hepatocyte-specific expression of Smad-7 attenuated the partial EMT associated with CCl₄-induced fibrosis [60]. Within the scleroderma skin, there also existed a partial EMT, albeit that the underlying mechanism of this phenomenon remains to be elucidated [61]. In our current study, we show that human middle ear cholesteatoma exhibits a partial EMT. Treatment with OPN, which is overexpressed in the cholesteatoma tissues, resulted in a partial EMT in HaCaT cells, an immortalized human keratinocyte cell line. Together, our current and others’ previous works highlight an important role of partial EMT in nonmalignant diseases. This emerging field warrants further investigation.

OPN and the development of middle ear cholesteatoma

OPN, also known as secreted phosphoprotein 1 (SPP1), is named for its role as a bridge between cells and hydroxyapatite through the Arg-Gly-Asp (RGD) and polyaspartic acid motifs [62]. There are two forms of OPN, i.e. secreted OPN, which is fully modified by post-translational modifications, and intracellular OPN, which is not or only partially modified by post-translational modifications. OPN is expressed in the cells of kidney, ovary, the immune system, as well as the gastrointestinal tract [63], and has pleotropic functions, including cell adhesion, migration, and survival, as well as immunomodulation. Pathologically, overexpression of OPN has been revealed in cancer, diabetes, cardiovascular diseases and renal disorders. OPN can bind to CD44 receptors and other receptors. CD44, a cell surface glycoprotein, is involved in inflammation, angiogenesis, and metabolism [64]. The binding of OPN to CD44 can help to prevent pathogen infection, recruit and activate neutrophils, macrophages, and granulocytes during the immune response [65]. Further, OPN is able to regulate fibroblast migration by binding to CD44 and the ERM (ezrin, radixin, moesin) proteins [66]. In this study, we show that OPN is overexpressed in the matrix of human middle ear cholesteatoma tissues. Treatment with OPN leads to a partial EMT and promotes cell proliferation, survival and migration in HaCaT cells. Our findings thus unveil a previously unappreciated role of OPN, that is to say, OPN plays a critical role in the development of cholesteatoma.

Previous studies have experimentally assessed the effects of the blockade of OPN signaling on disease progression. For example, antibody-mediated blockade of OPN inhibited angiogenesis in the breast cancer xenografts in mice [67]. Colon tumor growth was suppressed following an OPN blockade immunotherapy using OPN neutralization monoclonal antibodies [68]. Besides, a high affinity antibody against OPN could restrict the growth of large metastatic tumors in the lung [69]. An OPN RNA aptamer ablated the binding of OPN to CD44 and αvβ3 integrin receptors, thereby decreasing the progression and distal metastasis of a xenograft model of breast cancer [70]. Antibody-mediated blockade of OPN had a beneficial role in renal ischemia-reperfusion [71]. In addition, OPN was important in the initiation and persistence of CD8⁺ T cell-mediated graft-versus-host disease and anti-OPN treatment was able to retain the graft-versus-leukemia effect of alloreactive CD8⁺ T cells [72]. In the current study, we have successfully established a rat model for middle ear cholesteatoma. Pharmacological inhibition of CD44 remarkably ameliorated the cholesteatoma-like symptoms in SD rats. Thus, targeting the OPN-CD44 axis may represent a promising strategy for the treatment of middle ear cholesteatoma.

In summary, our current study demonstrates that there exist a partial EMT and an overexpression of OPN in human middle ear cholesteatoma. Treatment with OPN promotes cell proliferation, survival, as well as migration, and leads to a partial EMT in human keratinocyte cell line. Blockade of the OPN signaling improves the cholesteatoma-like symptoms in rats. The mechanistic study indicates that the AKT-ZEB2 axis critically mediates the effects of OPN. These findings suggest that the OPN-CD44 signaling is a promising target for the treatment of human middle ear cholesteatoma.

Funding Statement

This study was supported by the National Natural Science Foundation of China [81600806, to W.Z.].

Disclosure statement

No potential conflict of interest was reported by the author(s).

Author contributions

L.Z. conceived the study, designed and performed the experiments, analyzed the data, and drafted the manuscript. L.X., J.H. and C.H. provided technical support. A.L. and X.L. analyzed the data. W.Z. edited the manuscript and obtained grant support.

Data availability statement

The data that support the conclusion of this study have been included in this manuscript or available from the corresponding authors upon reasonable request.

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/15384101.2024.2345481