Introduction
Cervical cancer is the fourth most common cancer in women and the fourth leading cause of cancer-associated mortality in women (with incidence and mortality rates of 6.6 and 7.5% considering the 10 most common cancers in women in 2018, respectively), with an estimated 570,000 cases and 311,000 deaths worldwide in 2018 (1). Moreover, in regions of the world with low (e.g. Niger, Chad, Sierra Leone, Gambia and Nigeria) and medium (Paraguay, Egypt, Vietnam, El Salvador, Nicaragua, Zambia and Pakistan) human development index, cervical cancer has the second highest incidence (18.2%) and mortality (12%) rates, behind breast cancer (1). In the USA 13,800 new cases and 4,290 cervical cancer-associated deaths have been estimated for 2020, and the probability, from birth to death, of developing invasive cervical cancer from 2014 to 2016 was of 0.6% (2). Since the mid-1970s cancer survival has improved in the USA for all of the most common cancers (e.g. lung and bronchus, colon and rectum, breast, prostate, oral cavity and pharynx, esophagus, stomach, pancreas, liver, kidney, urinary bladder, melanoma, ovary and thyroid), except those of the uterine cervix and uterine corpus (3). In cervical cancer the 5-year relative survival rate for all races, according to the stage at diagnosis, was 92%, 56% and 17%, for localized, regional and distant lesions, respectively (2). Moreover, cervical cancer remains the second leading cause of cancer-associated death in women aged 20 to 39 years in the USA (2).
Infection with a subtype of the human papillomavirus, known as high-risk (HR), is the most common cause of cervical cancer (4). However, not all women infected with HR-HPV will develop cervical cancer, as this disease develops over 2 decades following exposure to these viruses and might require additional contributing factors, including tobacco smoke (5), parity (6,7), estrogens (8), oral contraceptives (9), immune deficiencies, such as human immunodeficiency virus-seropositive women (10), genetic polymorphisms (11), epigenetic regulation of costimulatory factors Tim-3 and galectin-9 (12) and methylation sensitivity of the enhancer from HPV16 (13). Dietary factors also appear to play a role, as a diet rich in plant-based nutrients, which increases dietary fiber, vitamins -C, -E, -A, α- and β-carotenes and lutein, has been found to reduce the risk of cervical cancer (14,15).
HPVs are DNA viruses containing two genes, which are expressed early named E6 and E7, which in the high-risk viral types encode proteins that promote cervical carcinogenesis (16). E6 and E7 oncoproteins associate with and inactivate the tumor suppressor proteins p53 (17) and retinoblastoma (Rb) (18), respectively. Employing chemical carcinogens in the K14E7 transgenic mouse, E7 was found to promote the formation of benign tumors in the skin, while in the transgenic mice K14E6 treated with these carcinogens, E6 accelerate the progression of the benign tumors to the malignant state (16). However, in the cervix, E7 increased cell proliferation and generated micro-invasive cervical cancers, while E6 did not produce neoplasia or cancer even following 17β-estradiol (E2) treatment, which is a contributing factor for cervical cancer, and was delivered in continuous release pellets implanted in the dorsal skin (8,19). The co-expression of E6 and E7 in double transgenic mice (K14-E6:E7) revealed that E6 modulated the malignant phenotype produced by E7, causing micro-invasive cervical cancers to be dispersed over a larger area of cervical stromal tissue (19). E2 treatment is required not only for the genesis of cervical cancer (20,21) but also for its persistence and continued development (22).
The disassembly of the TJ, which is a cell-cell adhesion structure characteristic of epithelial cells, constitutes one of the earliest steps in cell transformation (23,24). However, its role in the development of cervical cancer has not been elucidated, therefore further investigation was performed in the present study. TJs are comprised of a complex group of molecules, including the peripheral proteins, cingulin, ZO-1, -2 and -3 and the integral proteins, claudins, occludin and JAMs. While numerous TJ proteins are downregulated during cell transformation (e.g. ZO-1, -2, and -3, occludin, JAM-A and MUPP1), others are overexpressed (e.g. claudins -2, -3 and -4), even though they are no longer present at the cell borders (23). With respect to claudins, there are differences in expression levels of the protein in different tissues and different types of cancer. For example, in lung squamous cell carcinomas claudins -1, -2, -4 and -7 are expressed, while claudins -2, -3, -4, -5 and -7 are expressed in lung adenocarcinomas, and claudin-4 was found to be expressed in lung large cell carcinomas express. Claudin-1 is expressed in esophageal squamous cell carcinomas, while in the adenocarcinoma form of the disease claudins -2, -3, -4, and -7 are expressed. Claudins -1, -4, -5 and -7 are expressed in the squamous cell carcinomas found in the oral cavity, while claudin-1 is expressed in the low mucoepidermoid carcinoma, and claudin-3 is expressed in the high mucoepidermoid carcinoma (23). Therefore the expression of claudins is a useful molecular tool for distinguishing between different types of tumors and for the prediction of patient survival rates. For example, claudin-6 is the most distinctive molecular marker of atypical teratoid/rhabdoid tumors, which are highly aggressive malignant central nervous system tumors of children, in comparison to other brain tumors, such as choroid plexus papilloma, ependymoma, large-cell medulloblastoma, classic medulloblastoma and pediatric glioblastoma (25).
In our previous study claudin-4 expression was found to increase in the cervix of K14E7 transgenic mice compared with that in non-transgenic mice, particularly in the presence of E2 (26). In the present study the impact of the E7 onco-protein on the expression levels of claudins -1 and -10 in the murine cervix was investigated, as the expression of claudin-1 has been found to decrease in breast carcinomas, stage II colon cancer, lymph node metastasis in colon carcinomas, colon mucinous carcinomas, lung adenocarcinomas, prostate adenocarcinomas, pancreatic endocrine tumors, gastric cancer, type I endometrial carcinomas, hepatocellular carcinomas, kidney clear cell carcinomas, papillary urothelial neoplasms of low malignant potential and low grade urothelial cell carcinomas, thyroid undifferentiated carcinomas and thyroid medullary carcinomas, while it is overexpressed in colorectal cancer, in the early phase of carcinogenesis in cervical intraepithelial lesions, esophageal squamous cell carcinoma, squamous cell carcinoma of the tongue, low grade mucoepidermoid carcinoma of minor salivary glands, and in melanoma (23). With regards to claudin-10, its expression was reduced in the biliary tract, breast carcinomas and in lung adenocarcinomas of the invasive type (23,24); however, it was increased in chicken ovarian cancer (27), hepatocellular carcinoma (28) and papillary thyroid cancer (29). To further confirm the results, the impact of the stable expression of the E7 oncoprotein in Madin-Darby Canine Kidney (MDCK) cells was also investigated, including monolayer morphology, transepithelial electrical resistance (TER), claudin expression, invasive properties and cell stiffness. The results indicate that E7 perturbs the expression pattern of claudins and the degree of sealing of TJs in the cervix, and in MDCK cells, and also alters the cytoarchitecture of the cervix and MDCK monolayer, induces migration of MDCK cells in three-dimensional (3D) cultures and triggers the stiffening of their apical membrane.
Materials and methods
Mouse model and hormone treatment
The K14E7 trans-genic mice were a gift of Dr Paul F. Lambert (Department of Oncology, McArdle Laboratory for Cancer Research, University of Wisconsin, Madison, WI, USA) and were backcrossed in the FvB background, maintained and used as heterozygotes in further experiments. The FvB mice, used as the control group, were obtained from the animal facility of the Center for Research and Advanced Studies, Mexico. FvB and K14E7 transgenic female mice were housed in cages with double air filtration (HEPA level 99% of particles) and maintained at 20-22°C, under a 12 h dark-light cycles with food and water ad libitum. Animals were sacrificed by cervical dislocation at 2-, 4- and 7-months of age. Hormone treatment was performed with continuous release pellets delivering 0.05 mg of E2 over 60 days (cat. no. SE-121; Innovative Research of America). Pellets were implanted in the dorsal skin of 1-month-old virgin female transgenic and non-transgenic mice. Mice sacrificed at 4-months of age received two E2 pellets: one at the first month of age and the other at the 3rd month of age; whereas mice sacrificed at 7-months of age received an additional third pellet at the 5th month of age. A total of 42 mice were used in the study: 5 FvB mice, and 3 K14E7 at each age (2-, 4- and 7-months of age) were analyzed and 3 FvB and K14E7 mice treated with E2 were also used for each age group (2-, 4- and 7-months of age). The weight of the FvB and K14E7 mice was similar to what had been previously reported (20, 26 and 30, and 18, 23 and 26 g for 2-, 4- and 7-month-old FvB and K14E7 mice, respectively) (30).
Cell culture
Epithelial MDCK cells were obtained from the American Type Culture Collection (cat. no. CCL-34). Cells between the 60th and 90th passage were grown at 36.5°C in disposable plastic bottles in a humidified atmosphere with 5% CO2 in DMEM (cat. no. 12800-082; Gibco; Thermo Fisher Scientific, Inc.) with 100 U/ml penicillin-streptomycin (cat. no. A-01; In Vitro, S.A; www.invitro.com.mx) and 10% fetal bovine serum (cat. no. S1650-500; Biowest). Cells were harvested with trypsin-EDTA (cat. no. EN-005; In Vitro). Mycoplasma testing of parental and MDCK-E7 cells was performed using a mycoplasma detection kit for conventional PCR (cat. no. 11-8005; Vector® Gem OneStep; Minerva Biolabs GmbH) according to the manufacturer’s instructions.
Immunofluorescence
Reproductive female tracts containing the vagina, cervix and both uterine horns were removed from the FvB and K14E7 mice treated with or without E2. The tissue was then immersed in tissue freezing mounting media (cat. no. 14020108926, Tissue Tek; Leica Microsystems GmbH) and incubated at -70°C for 24 h. Next, 5-7-µm sections were cut using a Leica MC1510 cryostat (Leica Microsystems GmbH) and mounted on pre-cooled (-20°C) gelatin-coated slides. The slides were then incubated at -70°C for 24 h. Subsequently, the sections were fixed with 100% ethanol at -20°C for 20 min and the remaining protocol was performed as previously described (31). Using a confocal microscope (Leica TG5 SP8; Leica Microsystems GmbH) with a ×63 objective, images were captured of the squamous multilayered epithelium of the exocervix only, where the basal epithelial cells are the site of infection of HPV in women (32). Immunofluorescence of claudins in parental and MDCK-E7 cells at the TER peak (18 h) following the Ca-switch, was performed following a previously described protocol (31).
The following primary antibodies were used for the immunofluorescence experiments in both mouse cervix and MDCK cells: rabbit antibodies against claudin-1 (cat. no. 51-9000, dilution 1:100) and claudin-10 (cat. no. 38-8400; dilution 1:100) (both Invitrogen; Thermo Fisher Scientific, Inc.). In addition, in MDCK cells the following mouse monoclonal primary anti-bodies were used: against E7 protein of HPV16 (cat. no. ab30731; dilution 1:100; Abcam), claudin-2 (cat. no. sc-293233; dilution 1:100; Santa Cruz Biotechnology Inc.) and claudin-4 (cat. no. 32-9400; dilution 1:100; Invitrogen; Thermo Fisher Scientific, Inc.). The following secondary antibodies were used: donkey anti-rabbit IgG coupled to Alexa-Fluor 488 (cat. no. A21206; dilution 1:500), donkey anti-mouse coupled to Alexa-Fluor 594 (cat. no. A21203; dilution 1:1,000), or donkey anti-rabbit coupled to Alexa-Fluor 594 (cat. no. A21207; dilution 1:1,000) (all Invitrogen; Thermo Fisher Scientific, Inc.). In parental MDCK and MDCK-E7 cells actin was detected using rhodaminated phalloidin (cat. no. P1951; dilution 1:50; Sigma-Aldrich; Merck KGaA). Slides were mounted with Vectashield with DAPI (cat. no. H1200; Vector Laboratories, Inc; Maravai Life Science).
All the antibodies used reacted with mouse and dog tissues according to the manufacturer's instructions. The two exceptions are the monoclonal antibody against claudin-2 and the anti-claudin-10 polyclonal antibody, in which no details were provided regarding the reactivity with dog tissue. However, the monoclonal antibody against claudin-2, was raised against amino acids 29-80 of the human claudin-2 protein, and a blast sequence analysis revealed a 96% identity and 98% similarity between human and dog claudin-2 (data not shown). The poly-clonal antibody against claudin-10 was successfully used in MDCK cells in a previous study (33).
Relative mean fluorescence intensity measurements
The relative mean fluorescence intensity measurements of claudins -1 and -10 in the cervix of the FvB and K14E7 mice treated with or without E2, and of claudins -1, -2, -4 and -10 in MDCK cells were obtained using ImageJ software (v1.52n, National Institutes of Health) using the freehand function. The integrated density feature of ImageJ was used to record pixel intensities per area. Data were derived from three images of the cervix for all the experimental groups: FvB and K14E7 mice treated with or without E2 at 2-, 4- and 7-months of age.
Transmission electron microscopy (TEM)
The reproductive female tracts containing the vagina, cervix and both uterine horns were removed from 2- and 7-month-old female FvB and transgenic K14E7 mice treated with or without E2. Subsequently 1-mm width discs from the middle of the exocervix were excised. Samples were fixed at room temperature with 2.5% (v/v) glutaraldehyde in 0.1 M sodium cacodylate buffer, pH 7.2, for 60 min. Samples were then treated at room temperature for 60 min with a solution of 1% (w/v) osmium tetroxide in 0.1 M sodium cacodylate buffer, containing 0.5 mg/ml ruthenium red. Following dehydration with increasing concentrations of ethanol and propylene oxide, samples were embedded in Polybed epoxy resins and polymerized at 60°C for 24 h. Thin sections (60-nm) were stained at room temperature for 20 min with uranyl acetate and subsequently for 2 min with lead citrate prior to examination at a magnification of ×20,000, using a Jeol JEM-1011 transmission electron microscope.
The permeability of the TJ present in the superficial cells of the multi-stratified epithelium of the cervix of mice was evaluated using the paracellular pathway marker ruthenium red, as previously reported (34,35). Thus, if the ruthenium red stain only highlighted the apical surface of the superficial cells in the cervix, this indicated that passage through the paracellular pathway was blocked due to TJ sealing. Instead, if the staining was present along the paracellular pathway, below the TJ region, in the superficial cells of cervix, the TJ was opened.
Stable transfection of MDCK cells with E7 oncoprotein
MDCK cells at 70% confluence were transfected with 3 µg of pcDNA3E7 plasmid as previously described (36) using Lipofectamine® 2000 (cat. no. 11668019; Invitrogen; Thermo Fisher Scientific, Inc.) according to the manufacturer's instructions. To generate the stable MDCK-E7 clones, the cells were harvested using trypsinization and plated at a density of 3×104 cells/cm2 on 100-mm culture dishes with DMEM containing 800 µg/ml geneticin (G418; cat. no. 11811-031; Gibco; Thermo Fisher Scientific, Inc.), 48 h following transfection. Resistant clones were selected and re-cloned manually using the cloning ring technique, after 2-3 weeks (37). Stable colonies were maintained in the presence of 200 µg/ml G418.
PCR amplification of E7
Total RNA was extracted from parental and E7 MDCK cells using TRIzol® reagent (cat. no. 15596026; Invitrogen; Thermo Fisher Scientific, Inc.) following the manufacturer's instructions. RNA purity and concentration were measured using a Nanodrop™ ND-8000 spectrophotometer (cat. no. ND-8000-GL; Thermo Fisher Scientific, Inc.) and its integrity was determined using electrophoresis in 1% agarose gels, by resolving the 28S and 18S ribosomal RNA bands. cDNA was synthesized from RNA using SuperScript One-Step reverse transcription (RT)-PCR with platinum Taq DNA polymerase (cat. no. 10928042; Invitrogen; Thermo Fisher Scientific, Inc.), following the manufacturer's instructions and a Biometra GmbH Professional Basic Gradient thermocycler (cat. no. 070-601; Analytik, Jena US LLC). The PCR for HPV16 E7 was performed in a final 25 µl volume containing: 1 µg template RNA, 0.2 µM forward and reverse primers for HPV16 E7, 1U RT/Platinum Taq mix, 1.2 mM MgCl2 and 200 µM dNTPs. The primers were designed according to E7 gene sequence obtained from GenBank (accession no. AF125673; www.ncbi.nlm.nih.gov/nuccore/4927719); and had the following sequence: Forward, 5′-CTCAGAGGAGGAGGATGAAATAG-3′; and reverse, 5′-CTGAGAACAGATGGGGCACAC-3′. cDNA synthesis was performed at 50°C for 30 min and at 94°C for 2 min. The following thermocycling conditions were used: Initial denaturation at 94°C for 15 sec followed by 25 cycles of 94°C for 1 min, 60°C for 30 sec and 72°C for 1 min. The PCR products were separated on 1.5% agarose gels and images were obtained following ethidium bromide staining. The expected amplicon size was 196 bp.
Western blot analysis
Western blots of lysates from the cervix of FvB and K14E7 mice treated with or without E2, of 2-, 4- and 7-months of age and of parental and MDCK-E7 cells at the TER peak (18 h) following the Ca-switch, was performed following standard procedures, as previously described (31). The following primary antibodies were used: rabbit polyclonal anti-claudin-1 (cat. no. 51-9000; dilution 1:500; Invitrogen; Thermo Fisher Scientific, Inc.), claudin-10 (cat. no. 38-8400; dilution, 1:500; Invitrogen, Thermo Fisher Scientific, Inc.) and GAPDH (cat. no. RPCA-GAPDH; dilution 1:40,000; Encor Biotechnology Inc.). In addition, for MDCK cells the following primary antibodies were used: rabbit polyclonal anti-claudin-2 (cat. no. 51-6100, dilution 1:1,000; Invitrogen; Thermo Fisher Scientific, Inc.), and claudin-4 (cat. no. 36-4800; dilution 1:500; Invitrogen; Thermo Fisher Scientific, Inc.), and a mouse monoclonal against actin (generated and provided by Dr Manuel Hernández, Department of Cell Biology, Center for Research and Advanced Studies, Mexico city, Mexico). The following secondary antibodies were used: goat anti-rabbit IgG conjugated to horseradish peroxidase (HRP; cat. no. 6111620; dilution 1:20,000; Invitrogen; Thermo Fisher Scientific, Inc.), or goat anti-mouse IgG conjugated to HRP (cat. no. 626520; dilution 1:10,000; Invitrogen; Thermo Fisher Scientific, Inc.). The proteins were visualized using a Immobilon chemiluminescence detection (cat. no. WVKLS 0500; EMD Millipore).
All the antibodies used react with mouse and dog tissues according the manufacturer's instructions. The only exception was the polyclonal anti-claudin-10 antibody, in which there was no information regarding its reactivity on dog tissue. However, this antibody was successfully used in MDCK cells in a previous study (33).
TER
Parental and MDCK-E7 cells were cultured at a density of 1.2×105 cells/cm2 on 1.12 cm2 Transwell poly-ester membrane clear inserts (cat. no. 3460; pore size 0.4 µm; Corning Inc.). Confluent monolayers of control and MDCK-E7 cells were transferred from normal calcium (NC; 1.8 mM Ca2+; cat. no. 12800-082; Gibco; Thermo Fisher Scientific Inc.) containing media to low calcium (LC; 1-5 µM Ca2+; cat. no. D9800-10.10; United States Biological) media for 22 h. Subsequently, the monolayers with no TER (time 0) were changed to NC media (Ca-switch) to trigger TJ formation and TER development. TER was measured continuously and without interruptions for 63 h from each insert using an automated cell monitoring system (cellZscope version CZS0101909; nanoAnalytics GmbH). TER values were obtained using the CellZscope software (version 2.2.0.16827; nanoAnalytics GmbH). Statistical analysis using the Student's t-test was used to investigate the TER observed in parental and MDCK-E7 cells at the peak of resistance obtained at 18 h.
Cell migration assay in a 3D matrix
MDCK and MDCK-E7 cells were plated at a density of 2.4×105 cells/cm2 on 12-well inserts containing Alvetex® 3-dimentional polystyrene scaffold (cat. no. AVP005-12; Reinnervate Ltd.) precoated at room temperature for 2 h with basement membrane matrix BD Matrigel™ (cat. no. 356230, BD Biosciences) at a concentration of 0.8 mg/ml, according to the manufacturer's instructions. Every second day DMEM media with 10% fetal calf serum was changed from the upper and lower chambers. The migration of MDCK and MDCK-E7 cells was evaluated using confocal immunofluorescence in 10-day-old cultures where the cells were stained with rhodaminated phalloidin (cat. no. P1951; Sigma-Aldrich; Merck KGaA) for 2 h at room temperature, and mounted with Vectashield (cat. no. H-1200; Vector Laboratories; Maravai LifeSciences) antifade mounting medium containing DAPI, which fluoresces blue when bound to DNA. Images were captured using a confocal microscope (Leica TG5 SP8) with a ×63 objective in the z plane using LAS AF X software (version 3.0; Leica Microsystems GmbH). A depth coded z series image was generated from red to blue, where warmer colors (red) correspond to cells that have not migrated to the bottom of the scaffold and are still located on the surface, whereas colder colors (blue) represent the cells that have migrated deeper in the scaffold.
Atomic force microscopy
The nanomechanical properties of the apical surface of parental and MDCK-E7 cells present as either isolated cells or as islands of cells was analyzed using Atomic force microscopy. The cells were plated at a density of 1.5×104 cells/cm2 on ultra flat silicon wafers placed in 12-well plates and incubated at 36.5°C in a humidified incubator with 5% CO2 with DMEM (cat. no. 12800-082; Gibco; Thermo Fisher Scientific, Inc.) supplemented with 10% fetal bovine serum (cat. no. S1650-500; Biowest). The cells were fixed for 10 min with 4% paraformaldehyde at room temperature, after 8 h. All measurements were performed using a Solver Next Atomic Force Microscope from NT-MDT Spectrum Instruments. Elastic Modulus were measured based on the deviation force lateral (DFL) curves (displacement of cantilever vs. distance). A cantilever contact type CSG10, was used with a curvature radius of 10 nm. Scanning conditions for the cells were as follows: A ramp size set to 80 µm for islets and 50 µm for isolated cells, with a gain of 0.18 and a rate of 0.3 Hz with 256 points. Experimental data were analyzed using the Image Analysis Software (v3.5; NT-MDT Spectrum Instruments) to determine Young’s modulus in kPa.
Statistical analysis
The statistical analysis between multiple groups in different experimental conditions (Figs. 2B and C, and 4B and C) was performed using three-way ANOVA followed by Tukey's post hoc test. The statistical analysis between 2 groups (Figs. 3B, 5B, 8A, C and D, and 9B) was performed with Student's t-test. Statistical analysis was performed using Prism GraphPad v6 (GraphPad Software Inc.). Data are presented as mean ± standard deviation. Number of repeats are indicated in the legend of each figure.
Results
Transgenic K14E7 mice treated with E<sub>2</sub> develop invasive cancer with a reduced expression of claudin-1
The expression of claudin-1 in the multi-stratified epithelium of the cervix of control FvB and transgenic K14E7 mice treated with or without E2 was investigated. In control FvB mice, at 2 months of age, claudin-1 is present at the border of cells from the suprabasal region up to the most superficial cell layer facing the cervix lumen where it is expressed with higher intensity, while no claudin-1 expression is present in the basal cell layer (Fig. 1). In mice at 4- and 7-months of age the expression of claudin-1 was present at the basal cell layer augments and neither, treatment of FvB mice with E2 nor the expression of E7 in the transgenic K14E7 mice changed this pattern of expression (Figs. 2A and S1-4). Quantitative analysis revealed no change in claudin-1 immunofluorescence intensity in the cervix between FvB mice at 2-, 4-, and 7-months of age with or without E2 treatment (Fig. 2B). In 2-month-old K14E7 mice without E2 treatment, there was an increase in claudin-1 expression compared with that in FvB mice, and FvB and K14E7 mice treated with E2 (Fig. 2B). Western blot analysis of claudin-1 in the cervix of the aforementioned mice confirmed the increase in claudin-1 protein expression in 2-month-old K14E7 mice without treatment with E2 compared with that in 4- and 7-month-old K14E7 mice without E2 treatment, and in 2-month-old FvB mice without E2 and K14E7 mice with E2 (Fig. 2C).
Treatment of transgenic K14E7 mice with E2 induces the appearance of papillomas which develop progressively with age. In carcinomas that project into the cervical stroma of 7-month-old mice, the expression of claudin-1 was reduced compared with that in the rest of the cervix (Fig. 3A and B).
Claudin-10 localizes at an earlier age in the upper layer of the cervix in FvB mice treated with E<sub>2</sub> and K14E7 mice treated with or without E<sub>2</sub>
Subsequently, the expression of claudin-10 in the cervix of control FvB and transgenic K14E7 mice treated with or without E2 was investigated (Figs. S5-8). Claudin-10 was expressed in a diffuse cytoplasmic pattern, from the basal to the uppermost layers of the cervix in FvB mice without E2 treatment at 2- and 4-months of age (Fig. 4A-a and -e); however, it was found to be expressed at the border of cells in the superficial cell layers in 7-month-old mice (Fig. 4A-i). The expression of claudin-10 in FvB mice treated with E2 and in K14E7 mice treated with and without E2 was located in the border of cells in the in the upper cell layers from the second month of life (Fig. 4A--b-d, -f-h, and -j-l). Quantitative analysis revealed a significant higher fluorescence intensity of claudin-10 in the cervix of FvB mice treated with E2 at 2-months of age compared with that in mice at 4- and 7-months of age, and these values were similar to those in K14E7 mice treated with or without E2 at 2-, 4- and 7-months of age (Fig. 4B). In addition, the quantitative fluorescence intensity of claudin-10 in 2-month-old FvB mice was higher compared with that observed in 7-month-old K14E7 mice without E2 treatment and in 2- and 7-month-old K14E7 mice treated with E2 (Fig. 4B). The western blot analysis of claudin-10 protein expression in the cervix of the aforementioned mice decreased in a time-dependent manner in both FvB and K14E7 mice treated with or without E2, whereas in K14E7 mice treated with E2 the level of claudin-10 at 4- and 7-month of age was lower compared with that in K14E7 mice without E2 treatment, which were 2-months old (Fig. 4C).
Furthermore, in 7-month-old transgenic K14E7 mice treated with E2, the papillomas which can invade the stroma showed a low expression levels of claudin-10, similar to that observed in the lower layers of the rest of the cervix (Fig. 5A and B).
Opening of TJs at the cervix superficial cell layer is induced by E<sub>2</sub> or E7
To determine if the changes observed in the expression of claudins -1 and -10 were accompanied with an increase in the paracellular permeability of the cervix, further analysis was performed. Fig. 6A shows a semi-thin section of the cervix in a 2-month-old FvB mouse without E2 treatment, where the ruthenium red stain was found to be located in the lumen, which was in contact with the apical surface of the superficial layer of cells in the multistratified epithelium. Subsequently, using transmission electron microscopy, the thin sections found in the cervix of 2-month-old FvB mice, without E2 treatment, were impermeable to the paracellular electrodense marker, ruthenium red, which was found to be maintained in FvB mice at 7-months-old (Fig. 6B-a and -e). In contrast, in FvB mice at 2- and 7-months-old treated with E2, 29 and 36% of TJ were permeable to ruthenium red, respectively (Fig. 6B-b and -f). There percentage permeability in K14E7 mice treated without E2 was 67 and 55%, at 2- and 7-months of age, respectively (Fig. 6B-c and -g), which was similar to that found in K14E7 mice treated with E2 (Fig. 6B-d and -h).
Therefore, it was concluded that the increase in TJ perme-ability in the cervix of FvB mice treated with E2 occurs at the same age (2-months old) when a change in the expression pattern of claudin-10 was observed.
The stable expression of E7 alters the monolayer architecture of epithelial MDCK cells
To improve the understanding of E7 on epithelial cell transformation, a stable transfection of HPV16 E7 oncoprotein in epithelial MDCK cells was created, which is a cell line characterized for exhibiting well developed TJs (38), and has frequently been used to investigate the effect of viruses and viral proteins on TJs (39-43). Fig. 7A revealed that E7 was expressed in clones 1 and 5 of MDCK-E7 cells but not in parental MDCK cells using RT-PCR. As the expression of E7 was the same in all clones, clone 5 was selected to be used in further experimentation. Fig. 7B shows that the E7 protein is expressed in a diffuse manner in MDCK-E7 cells, while there is no expression in parental MDCK cells, using immunofluorescence. Light microscopy images of semi-thin sections reveal that MDCK-E7 cells grow in an abnormal manner characterized by the presence of some cells growing on top of one another and by the widening of intercellular spaces (Fig. 7C). These observations were further confirmed by TEM (Fig. 7D).
E7 induces the development of an acute peak of TER in MDCK monolayers and changes the expression pattern of claudins-1 and -10
Subsequently, if the development of TER was altered by the expression of E7 in MDCK cells was investigated and the Ca-switch procedure to trigger TJ formation and TER development was used. Fig. 8A shows that MDCK-E7 cells reached a much higher peak of TER compared with that in the control cells. However, at 40 h following the Ca-switch, when the TER stabilizes in both cell lines, both experimental groups display similar TER values. Next, the expression of claudins-1 and -10 at the TER peak (18 h) following the Ca-switch was investigated. In addition, the expression levels of claudins-2 and -4 were also investigated, as the former forms paracellular cationic pores (44,45) which increases TJ permeability (46) and decreases TER (47), while the latter exerts the opposite effect, functioning as a cationic barrier (48) or an anionic pore (49). Using immunofluorescence, the expression of claudins-1, and -10 was found to be reduced, while that of claudin-2 was reduced at the cell borders and appeared in a diffuse pattern in the cytoplasm at the peak of TER in MDCK-E7 cells compared with that in parental cells. On the other hand, claudin-4 expression was increased (Fig. 8B and C). These results were further confirmed using western blot analysis (Fig. 8D).
Taken together, the results suggest that the higher peak of TER found in MDCK-E7 cells compared with that in parental monolayers, was due to an altered expression of claudins.
E7 enhances the migrate ability of MDCK cells through a 3D matrix and induces cell stiffening and stress fiber formation
Next, if the presence of E7 induced the migration of MDCK cells was investigated, as this characteristic is important for cancerous cells to invade the underlying stroma and metastasize. A 3D in vitro model was used, as it replicates the tissue organization compared with that in 2D models (50). MDCK and MDCK-E7 cells were plated on top of a Matrigel® coated Alvetex®Scaffold, which is a porous and inert polystyrene platform with large voids that create 3D spaces where cells can grow. Fig. 9A shows that MDCK cells migrated through the scaffold between 20 and 30 µm, whereas MDCK-E7 cells migrated to a distance of 50 µm, thus revealing that E7 enhanced the ability of MDCK cells to migrate.
It has been previously shown that cells undergo a stiffening stage prior to acquiring malignant features (51). Therefore, the apical surface elastic Young's module in parental and MDCK-E7 cells was measured using standard nanoindentation force microscopy. The experimental data are supported by their theoretical analysis, which is based on the description of the tip-sample interactions using the Euler-Bernoulli equation (www.efunda.com/formulae/solid_mechanics/beams/theory.cfm) and the asymptotic solutions of the oscillatory tip behavior during its interaction with the sample. Fig. 9B shows the rigidity of the apical surface of MDCK-E7 increases compared with that in parental cells by 5.64- and 1.99-fold in isolated and in cell islands, respectively. As stiffening is characterized by the accumulation of stress fibers (51), if E7 promoted the expression of stress fibers in MDCK cells was investigated. Fig. 9C shows a proliferation of stress fibers in MDCK-E7 compared with that in control MDCK cells.
Taken together, these results indicate that the E7 oncoprotein promotes cell stiffening and invasion.
Discussion
In the present study a previously characterized transgenic mouse containing the hK14HPV16E7 transgene, where the expression of E7 oncoprotein from HR-HPV16 is regulated by the promoter for human keratin 14 (hK14) (52) was used. The activity of this promoter is restricted to the basal cell layer of multi-stratified epithelia, to guarantee that the expression of E7 is directed to the cell types where HPV infection is targeted. K14E7 transgenic mice display several characteristic phenotypes, including wrinkled skin, thickened ears and loss of hair in adults, and develop skin tumors, which begins at 9-months of age (52). In addition, animals exhibit stunted growth and a high mortality rate during the first 2 weeks of life caused by the incapacity to feed due to esophagus hyperplasia. This hyperplasia is characterized by an expansion of the keratin 10-positive basal layer of cells and is found in several additional sites, including the skin, palate, forestomach and exocervix (52).
The expression of claudin-1 in the cervix of FvB mice, without E2 treatment, was higher at the most superficial cell layer facing the lumen, while the presence at the basal cell layers augments, as the age of the mice increases from 2- to 7-months of age, in the present study. This same pattern of expression was found in FvB mice treated with E2, and in K14E7 mice treated with or without E2. However, in 2-month-old K14E7 mice treated without E2 the amount of claudin-1 found in the cervix was higher compared with that in the other experimental groups. On the other hand, the expression of claudin-1 was reduced compared with that in the rest of the cervix in the invasive papillomas that developed in 7-month-old K14E7 mice treated with E2. These results are consistent with previous research which shows that claudin-1 is markedly expressed in grades I, II and III of human intraepithelial neoplasias, and in situ carcinomas (53-55), but decreases with progression to the invasive phenotype (56).
In addition, the expression of claudin-1 was found to be low in aggressive colorectal carcinomas (57,58) and in gastric cancer (59,60). In the latter (61), as well as in oral squamous cell carcinomas (62) and thyroid cancer (63), the absence of claudin-1 is associated with poorly differentiated tumors. Both papillary urothelial neoplasms of low malignant potential (PUNLMP) and low-grade urothelial cell carcinoma have a decreased expression level of claudin-1 compared with that in inverted urothelial neoplasms, which almost always exhibit a benign behavior (64). The low expression level of claudin-1 was found to be a predictor of recurrence and poor patient survival in colon cancer (65-67), hepatocellular carcinoma (68,69), thyroid (63) and prostate (70) cancer, lung adenocarcinoma (71), non-small cell lung cancer (72) and PUNLMPs (64). With respect to breast cancer, the expression of claudin-1 was initially associated with the basal-like type, which has a very poor prognosis (73,74). However, a previous study has found that a claudin-1 negative phenotype predicts a high risk of recurrence and mortality in triple-negative breast cancers (75). In addition, a reduction or the total loss of the expression levels of expression claudin-1 was also observed in the group of patients with recurrent breast cancer in comparison to the non-recurrent group, as well as in the lymph node metastasis-positive group (76,77). Taken together, the findings from previous research studies are consistent with the results of the present study, that there is a reduced expression level of claudin-1 in the invasive papillomas in 7-month-old K14E7 mice treated with E2.
Nevertheless, claudin-1 has also been reported to be overexpressed in several cancerous tissues, including high-grade cervical intraepithelial neoplasias (78) and invasive cervical cancer (79), where it has been considered as a biomarker of the disease (80), with a prognosis potential (81). Claudin-1 was also found to be overexpressed in colorectal tumors (82-84), mucoepidermoid carcinomas of the salivary gland (85), thyroid papillary carcinomas, papillary microcarcinomas primary tumors and lymph node metastases (86), and esophageal (87,88), hypopharyngeal (89,90), tongue (91) and oral (92) squamous cell carcinomas, where claudin-1 was found to be associated with cell invasion and disease recurrence (93). Claudin-1 expression in melanoma cells promotes the secretion and activation of metalloproteinase 2, which enhances cell motility (94,95) and mediates tumor necrosis factor α-induced cell migration in gastric cancer cells (96). Claudin-1 was also found to be a prognostic marker and shorter patient survival in kidney clear cell carcinomas (97), intestinal-type gastric cancer (98), lung adenocarcinoma (99) and stage N2 non-small cell lung cancer (100).
The localization of claudin-1 is also altered in cancerous cells. For example, in colorectal cancer, the membrane staining intensity of claudin-1 was reduced compared with that in adjacent non-neoplastic tissue, while a significant increase in claudin-1 cytoplasmic staining was found in colorectal cancer tissue (101). Similarly, in follicular thyroid carcinoma there was negative claudin-1 staining in the membrane and an increase in the nucleus, resulting in increased cell migration and invasion (102). This expression is abnormal as benign thyroid tissue and peritumoral non-malignant thyroid tissue are negative for claudin-1 staining (103,104) and in normal epithelia, in which claudin-1 is expressed [such as the skin (105), kidney (106) and intestine (107,108)], the expression is negative in the nucleus and concentrates at the TJ in the lateral membrane. Thus, evidence is emerging showing that in human cancer, the expression and localization of claudin-1 is altered when compared with that in normal tissue, which was consistent with the results in the murine cervical cancer model in the present study.
It was found in the current study that the expression of claudin-10 in the cervix of FvB mice, without E2 treatment, was reduced after 2-months of age and was found to be increased in the border of cells in the upper cell layers from the 2-month of age in FvB mice treated with E2, and in K14E7 mice with and without E2 treatment. The quantitative analysis of the fluorescence revealed lower values of claudin-10 in K14E7 mice treated with or without E2 in the second month of life compared with that in FvB mice treated with E2; however, this effect was not confirmed using western blot analysis. The reason for this discrepancy remains unknown, although it could be due to the greater statistical variance of the western blots results, which makes them not statistically significant.
Furthermore, in 7-month-old transgenic K14E7 mice treated with E2, the expression level of claudin-10 in the papillomas, which invades the stroma, was not detectable. This was consistent with previous research which found that the expression of claudin-10 diminishes in breast (109) and biliary tract carcinomas compared with that in normal tissues (110). Moreover, the expression of claudin-10 was associated with the overall survival of patients with lung adenocarcinoma, as it was found to be lower in invasive lepidic adenocarcinoma compared with that in the in situ lung adenocarcinomas (111).
In contrast, the mRNA expression level of claudin-10 was found to be associated with the recurrence of primary hepatocellular carcinoma following curative hepatectomy (112). In addition, type III hepatic tumors with a high malignant potential, display cytoplasmic staining of claudin-10 around the nucleus (113). In the thyroid, claudin-10 was found to be negatively expressed in follicular thyroid carcinomas, and overexpressed in papillary thyroid carcinomas (114), where it serves as a biomarker to distinguish tumors from benign thyroid lesions (115).
With respect to the paracellular permeability of the cervix, treatment of FvB mice with E2 increased the permeability of TJ to ruthenium red. However, the effect was markedly higher in K14E7 mice treated with or without E2, which suggest that both E2 and E7 induce the development of weaker TJs. In this respect, we have previously found that in the skin of ovariectomized FvB mice, E2 treatment diminishes the expression of TJ proteins occludin and ZO-2 (39). Notably, in mice the blockade of ruthenium red paracellular passage can be established at the most superficial layer of cells in the cervix, while in the human cervix the epithelial junctions that restrict the diffusion of paracellular markers are localized in the cells located three or four layers below the lumen (116).
Following the investigation of the cervix in the trans-genic mice, further analysis was performed to determine if the expression of E7 altered the TJs and transformed the characteristics of the normal epithelia, which required transfection of E7 into an epithelial cell line. A non-cancerous cell line of cervix from a mouse or other mammal was not used as to the best of our knowledge these are not currently commercially available. In addition, from the seminal study of TJs in MDCK cells performed by Cereijido et al (117) in 1978, the cell line has been widely used to investigate the electrical properties of TJs, the permeability of the paracellular pathway (118-120), the changes in the ultrastructure of TJs visualized in freeze-fracture replicas (38,121), the molecular composition of TJs (46,122-126), as well as the response of TJs to a wide variety of factors, including temperature (127), ions (128-130), signaling cascades (131,132), toxins (133) and growth factors (134,135). Moreover, the role of claudins, in particular claudins -1 (136,137), -2 (46,125,138), -4 (46,125,126) and -10 (33,139) has been extensively studied in MDCK cells. Furthermore, the effect of numerous viruses and viral proteins on TJs has also been investigated in MDCK cells (39-43,140-160). Therefore, the MDCK cell line was selected as it is an ideal in vitro model system to investigate factors that regulate or have a harmful effect on TJs.
A stable MDCK cell line was created which expressed E7, and it was found that the monolayers had widened intercellular spaces and had areas in which some cells were growing on top of one another. A similar phenotype was observed in MDCK monolayers where the expression of the TJ protein ZO-2 was knocked down (161,162), which suggests that the E7 oncoprotein exerts a harmful effect on TJs. However, when the development of TER in MDCK-E7 monolayers was analyzed it was found that they achieved a higher peak of TER compared with that in parental cells. This unexpected result led to the investigation into the expression pattern of claudins at the time where TER reaches its highest values. It was found that the protein expression level of claudins -1, -2 and -10 decreased, while that of claudin-4 increased in MDCK-E7 monolayers. The alteration of a single type of claudin can modify, in a significant manner, the permeability and transepithelial electrical resistance of a tissue (163). The increased protein expression level of claudin-4 in MDCK-E7 cells was found to be important, as transfection of this protein can function as a cation barrier in MDCK cells and induce a significant decrease in permeability and an increase in TER (46,48). The decreased expression level of claudin-2 in MDCK-E7 monolayers was expected to have a significant effect on TER, as this claudin, which is highly expressed in leaky epithelia, such as the proximal tubule of the kidney (164) and the intestinal crypts (165), functions as a high conductance cation-permeable pore (44,47). Claudin-1 was found to be ubiquitously expressed claudin and in vitro overexpression studies reveal that it acts as a barrier, which increases TER (136,137) therefore, decreased expression would not be expected to contribute to the increased TER observed in MDCK-E7 cells. Claudin-10 has been found to be expressed in numerous tissues, including breast (109), biliary tract (110), lung (111), kidney (33) and liver (113). The function of the two major claudin-10 isoforms revealed that while claudin-10a acts as an anion pore in MDCK II cells (166), claudin-10b has no strong ion selectivity in MDCK II cells which exhibit a high expression of claudin-2 that forms cation pores, but instead claudin-10b acts as a strong cation-permeating pore in high resistance MDCK-C7 cells (33). Claudin-10a expression was found to be restricted to the kidney (33), therefore the remaining tissues, including the cervix and those where the mRNA for claudin-10b has been detected (heart, brain, spleen, lung, liver, skeletal muscle, testis, placenta, eye, lymph node, smooth muscle, prostate, thymus, stomach and uterus) are expected to express claudin-10b. However, as none of the available antibodies, including the one used in the present study, discriminate between claudin-10a and -10b, this was not confirmed in the murine cervix or in MDCK cells.
In MDCK cells it was found that the stable expression of E7 also increases cell migration through a 3D scaffold, stiffens the apical cell membrane, induces the appearance of stress fibers, decreases the expression level of claudins -1, -2 and -10, and augments the expression of claudin-4. All these changes suggest that the cell phenotype was transformed. With respect to increased cell migration and altered claudin expression, previous research has found that migration augments upon overexpression of claudin-2 in lung cancerous cells (167), claudin-4 in ovarian tumor cells (168), claudin-7 in ovarian cancer (169), claudin-8 in prostate cancer cells (170), claudin-10 in papillary thyroid cancer cells (29) and in hepatocellular carcinomas cells (171), and claudin-17 in hepatic cells (172); or upon silencing of claudin-3 (173) and claudin-4 (173) in ovarian cells OV2008, of claudin-6 in human breast epithelium cell line HBL-100 (174), of claudin-7 in clear cell renal cell carcinoma (175), and of claudin-11 in nasopharyngeal carcinoma (176). In MDCK cells migration augments following claudin-2 silencing (156), and in normal mammary epithelial cells as well as in breast and ovarian tumor cells, blocking the second extracellullar loop in claudin-4, with a peptide, which mimics a conserved sequence in this claudin loop, inhibited cell mobility (167,177,178).
With respect to cell stiffening, tumors have long been characterized for being harder compared with that in normal tissue, and while stiffening of the stroma is a crucial factor that favors tumorigenesis (179,180), changes in the tension of the cell have also been observed during cell transformation (51). MDCK cells also acquire a more rigid apical membrane when the TJ proteins ZO-1 and ZO-2 are both knocked out (181). Moreover, cell stiffening, accompanied by the appearance of stress fibers, has been found to be essential to drive breast tumor growth during premalignant stages (51).
In summary, the results from the present study indicate that the oncogenic protein E7 derived from HPV16 induces epithelial cells changes, including in the expression of claudins -1, -2, -4 and -10 and TJ sealing, which are accompanied by changes in cell stiffness, motility and cytoarchitecture, which could be important in the development of tumorigenesis.