A non-tight junction function of claudin-7—Interaction with integrin signaling in suppressing lung cancer cell proliferation and detachment

Our results revealed that HCC827 claudin-7 KD cells became smaller in size, less spread out, and grew in an isolated patch pattern while the control cells were spread out and uniformly distributed over the plate (Fig. 1a). Claudin-7 immunofluorescence staining (Fig. 1b) and western blot (Fig. 1c) showed the successful knockdown of claudin-7 using #2 shRNA vector against claudin-7. HCC827 cells infected with #3 shRNA vector against claudin-7 are shown in Additional file 1: Figure S1 and these claudin-7 knockdown cells were designated as KD2 cells. Expression level of claudin-4 was increased in claudin-7 KD cells (Fig. 1d). There were no changes of expression levels of adherens junction protein E-cadherin and TJ proteins claudin-1 and −3 after claudin-7 was knocked down (Fig. 1d). There was no expression of claudin-2 in HCC827 lung cancer cells (data not shown). We have also knocked down claudin-7 expression in NCI-H358 (H358) human lung cancer cells (Additional file 2: Figure S2).

Because claudin-7 KD cells were smaller in size with high density in each patch, we performed a cell number count experiment to investigate the cell growth rates in both control and claudin-7 KD cells. We found that claudin-7 expression was associated with decreased cell proliferation. After claudin-7 was knocked down in HCC827 cells, the cell proliferation rate increased. Starting from day 4, the number of claudin-7 KD cells was twice more than that of control cells (Fig. 2a). Similarly, the claudin-7 KD2 cells (infected with #3 shRNA vector against claudin-7) also showed a significantly increased cell proliferation for all three days examined compared to the control cells (Additional file 3: Figure S3A). The cell doubling time for control and claudin-7 KD cells were 24 and 12 h, respectively. Cell cycle analysis revealed that a higher percentage of claudin-7 KD cells was in the DNA synthesis phase (S phase) and mitosis (G2/M) phase (Fig. 2b), consistent with the higher growth rate observed in the claudin-7 KD group.

We proceeded to examine several proteins involved in cell proliferation, cell survival and apoptosis by western blot. Our results revealed increased levels of phospho-ERK1/2, phospho-Bcl-2 and survivin and decreased level of cleaved PARP in claudin-7 KD cells (Fig. 2c). The claudin-7 KD2 cells also displayed an increased level of phospho-ERK1/2 and a decreased level of cleaved PARP as shown in Additional file 3: Figure S3B. Similar results were obtained in H358 claudin-7 KD cells in that these cells were higher in cell proliferation and had increased phospho-ERK1/2 level and decreased cleaved PARP level (Additional file 4: Figure S4).

Figure 2d shows the cell cycle analysis before and after 24 h of aphidicolin treatment. Aphidicolin blocks cells in S phase and prevents G1 cells from entering mitosis. Cyclin dependent kinase 2 (Cdk2) is required for G1 to S phase transition [20]. After 2.5 and 18 h of releasing the cells from aphidicolin, the cdk2 expression levels were higher in claudin-7 KD cells (Fig. 2e), indicating that more KD cells were transitioning from G1 to S phase. There were no significant differences in expression levels of cyclin B1 and cyclin E between control and claudin-7 KD cells.

We used TUNEL (Terminal deoxynucleotidyl transferase dUTP nick end labeling) staining method to detect the apoptosis. Although the total population of apoptotic cells in both control and claudin-7 KD cells was low, the claudin-7 KD cells showed a lower percentage of apoptotic cells than the control cells as shown in Additional file 5: Figure S5.

The wound healing assay was performed to examine the migration ability of both control and claudin-7 KD cells. We found that claudin-7 KD cells were peeled off easily along the edge of the wound each time after making a scratch to the monolayer (Fig. 3a, left). When the cells were plated on the uncoated coverslips, control cells adhered well and formed a monolayer on the coverslip while the KD cells grew on top of each other and formed spheroids (Fig. 3a, right). In addition, it took KD cells about 40 % less time to be completely lifted from the culture dish by trypsin than control cells (Fig. 3b). Cell attachment assay also revealed that only about 10 % KD cells were attached to the culture plate compared to the control cells four hours after plating (Fig. 3c, left). Similar results were also obtained in claudin-7 KD2 cells (Additional file 6: Figure S6A). The attached cell number was significantly reduced in claudin-7 KD2 cells compared to that of the control cells. In addition, lentivirus claudin-7 shRNA was used to knockdown claudin-7 expression in breast cancer cell line T-47D (Fig. 3c, right). Only 45 % T-47D claudin-7 KD cells were attached to the plate compared to the control cells four hours after plating. Electron microscopy revealed larger gaps between HCC827 claudin-7 KD cells and coverslip than that between control cells and coverslip while TJs were shown largely intact in both cell lines under high magnification (Fig. 3d, arrow). Transepithelial resistance (TER) measurement revealed that suppression of claudin-7 in HCC827 cells did not affect the barrier function of TJs (Fig. 3e). These results indicate that cell attachment machinery, not functions of TJs, was impaired in claudin-7 KD cells.

Defective cell attachment in claudin-7 KD cells indicates a non-TJ function of claudin-7 in cell-matrix adhesions. Integrins are the principal receptors responsible for binding the cells to extracellular matrix. After screening for major subunits, we found that integrin β1 was greatly decreased at both mRNA and protein levels (Fig. 4a and b). In addition, the phosphorylation level of focal adhesion kinase (FAK) was also greatly reduced in claudin-7 KD cells (Fig. 4b, left). The decreased expression levels of integrin β1 and phospho-FAK were observed in claudin-7 KD2 cells as well (Additional file 6: Figure S6B). Moreover, the reduced signals for integrin β1 and phospho-FAK were confirmed in the claudin-7 knockout mouse lungs (Fig. 4b, right). Immunofluorescence staining showed a strong basolateral staining of claudin-7 (Fig. 4c, top, Z-stack) that was co-localized with integrin β1. Knockdown of claudin-7 disrupted integrin β1 localization in that it was no longer at the cell membrane (Fig. 4c, bottom). Single staining of integrin β1 or claudin-7 as well as anti-secondary antibody only were shown in Additional file 7: Figure S7. Co-immunoprecipitation confirmed that claudin-7 and integrin β1 formed a protein complex in HCC827 cells (Fig. 4d). The decreased level of phospho-FAK in claudin-7 KD cells shown in Fig. 4b was consistent with the phenotype observed in claudin-7 KD cells in that they were less spread out than control cells (Fig. 1a).

To determine whether integrin β1 is the key regulator mediating cell-matrix adhesions, we used mouse anti-human integrin β1 adhesion-blocking antibody to treat both control and claudin-7 KD cells. Mouse IgG treatment was used as a negative control (Fig. 4e). Anti-integrin β1 antibody treatment significantly interfered with the cell attachment in both control and KD cells when compared to the mouse IgG treatment (Fig. 4e, right). The cell attachment assay revealed that the attached cells were significantly reduced in both control and claudin-7 KD cells after integrin β1 antibody treatment. This adhesion-blocking experiment indicates that integrin β1 is one of the key regulators in mediating cell attachment in HCC827 cells.

Importantly, H358 claudin-7 KD cells also formed spheroids when they were plated on the uncoated coverslips while control cells adhered well (Additional file 8: Figure S8A). Results from cell attachment assay indicated that only about 15 % H358 claudin-7 KD cells were attached to the culture plate when compared to the control cells (Additional file 8: Figure S8B). The expression levels of integrin β1 and phospho-FAK were also reduced in H358 claudin-7 KD cells (Additional file 9: Figure S9A). Immunofluorescent staining showed the partial co-localization of claudin-7 with integrin β1 (Additional file 9: Figure S9B). In addition, claudin-7 interacted with integrin β1 and formed a protein complex in H358 cells as well (Additional file 9: Figure S9C).

Cell-matrix adhesions involve not only cells and cell-matrix adhesion proteins such as integrin β1, but also involve extracellular matrix proteins. The most abundant extracellular proteins are collagens. Real-time RT-PCR revealed that collagen subunits 4α2, 5α1, 6α2, 15α1, and 16α2 were all significantly decreased at the transcriptional levels in claudin-7 KD cells (Fig. 5a). Thus, we cultured the claudin-7 KD cells on type IV collagen-coated plates to see whether a deficient collagen deposition was the primary consequence of claudin-7 suppression and whether type IV collagen was able to rescue the defects of claudin-7 KD cells. We found that claudin-7 KD cells grown on type IV collagen-coated plates became larger in size, more spread out (Fig. 5b), and showed partially improved cell attachment as determined by the prolonged trypsinization time when compared to the KD cells cultured on regular culture plates (data not shown). However, the cell proliferation rate was still significantly higher in claudin-7 KD cells when compared to control cells (Fig. 5c).

To investigate whether slowing down the cell proliferation rate is able to rescue the phenotype of KD cells, we treated control and claudin-7 KD cells with genistein or aphidicolin for 24 h. Our results showed that claudin-7 KD cells responded to both drug treatments in that the cell size became enlarged when compared to untreated cells (Fig. 5d&e). However, culturing KD cells on type IV collagen-coated plates or treating KD cells with genistein or aphidicolin did not increase integrin β1 protein expression level in claudin-7 KD cells (Fig. 5f).

To confirm that claudin-7 is the key factor in regulating cell proliferation and cell attachment in HCC827 lung cancer cells, we transfected claudin-7 tagged with myc into the claudin-7 KD cells to investigate whether we can rescue the phenotype of claudin-7 KD cells (Fig. 6a). Exogenous claudin-7-myc expression not only increased integrin β1 and cleaved PARP protein levels (Fig. 6b), but also reversed the morphology observed in claudin-7 KD cells in that they became larger in size, more spread out (Fig. 6a), and showed the significantly slower cell proliferation rate compared to the KD cells without claudin-7 transfection (Fig. 6c). Moreover, after we transfected claudin-7 or integrin β1 back to the KD cells and then plated them on uncoated glass coverslip, claudin-7 or integrin β1 transfected KD cells were adhered well and formed a monolayer while the vector-transfected KD cells still formed spheroid structures (Fig. 6d). These results demonstrate that claudin-7 is critical in regulating cell proliferation and cell attachment and the decreased integrin β1 level is the direct consequence of claudin-7 suppression that is responsible for weakened cell attachment.

To investigate whether claudin-7 can inhibit cell growth in vivo, we injected HCC827 control or claudin-7 KD cells into the flanks of the nude mice. As shown in Fig. 7a, the tumor induced by control cells was much smaller than that induced by claudin-7 KD cells. The inserts are the tumors taken out from the left and right sides of the nude mice, representing the control cells-induced and claudin-7 KD cells-induced tumors, respectively. Western blot results showed that claudin-7 KD cells-induced tumors have higher expression levels of phospho-ERK1/2, survivin, and cleaved PARP than those of control cells-induced tumors (Fig. 7b). Fig. 7c and d show the quantitative analysis confirming that the sites inoculated with claudin-7 KD cells developed the tumors significantly larger in size and heavier in weight than those inoculated with control cells, demonstrating that claudin-7 inhibited the tumor growth in vivo. Fig. 7e and f show the growth curves of ten control cells-inoculated tumors and ten claudin-7 KD cells-inoculated tumors, respectively. Seven of the ten nude mice inoculated with control cells developed tumors while nine of the ten nude mice inoculated with claudin-7 KD cells developed tumors. Six control tumors grew significantly slower than their paired claudin-7 KD tumors and the slopes of these paired tumor growth curves are significantly different.

The rabbit polyclonal anti-claudin-1 (Cat. 51–9000), −2 (Cat. 516100), −3 (Cat. 341700), −4 (Cat. 364800), ZO-1 (Cat. 61–7300), and mouse monoclonal anti-myc (Cat. 46–0603) antibodies were from Invitrogen (Carlsbad, CA). The rabbit polyclonal anti-claudin-7 antibody (Cat. 18875) was obtained from Immuno-Biological Laboratories (Japan). The anti-E-cadherin antibody (Cat. 610182, clone: 36) was purchased from BD Biosciences (San Jose, CA). The rabbit polyclonal anti-phospho-Bcl-2 (Cat. A7025, Ser56) was from Assay Biotechnology (San Francisco, CA). The rabbit polyclonal anti-phospho-FAK (Cat. 3283S) and anti-phospho-ERK1/2 (Cat. 9101S) were from Cell Signaling Technology (Beverly, MA). The mouse monoclonal and the goat polyclonal anti-integrin β1 antibodies were purchased from BD Biosciences (Cat. 610467) and Santa Cruz Biotechnology (Cat. sc-6622, Santa Cruz, CA), respectively. The mouse anti-integrin β1 adhesion-blocking antibody was from Chemicon (Cat. MAB22253Z, Millipore, MA).

Lung adenocarcinoma cell line HCC827, lung carcinoma cell line NCI-H358 (H358) and mammary ductal carcinoma cell line T-47D were purchased from ATCC (Manassas, VA) and grown in RPMI 1640 culture medium containing 10 % FBS, 100 units/ml of penicillin, and 100 μg/ml streptomycin in a humidified air (5 % CO₂ atmosphere) at 37 °C. Each of three lentivirus claudin-7 shRNA vectors (sequences #1: 5′-TTCCAAGGAGTATGTGTGA-3′; #2: 5′-GGCTATGGGAGTGTCTAGA-3′; #3: 5′-TCCCTACCAACATTAAGTA-3′) or one off-target shRNA control vector were transfected into HCC827, H358 or T-47D cells. The cells infected with #2 lentivirus claudin-7 shRNA vector are designated as KD cells, and the cells infected with #3 shRNA vector are designated as KD2 cells. All lentivirus vectors contain a GFP expression sequence. After 48 h incubation, transfected cells were selected by 1 μg/ml puromycin.

The pcDNA3.1 myc or pcDNA3.1-claudin-7-myc (mouse claudin-7 cDNA) was transfected into HCC827 claudin-7 KD cells using Amaxa Nucleofector device (Amaxa Biosystems, Cologne, Germany). Geneticin (G418) was added to select the transfected cells.

Five × 10³ HCC827 or H358 control or claudin-7 KD cells were seeded into each well of 24-well culture plates and then trypsinized on days 2, 4, 6 after plating, respectively. One × 10⁵ 6 micron AlignFlow Plus beads (Molecular Probes, Eugene, OR) were added to each sample and relative ratio of beads versus cells was obtained by the flow cytometer. The total cell number for each sample was then calculated.

HCC827 control or claudin-7 KD cells untreated or treated with 5 μg/ml aphidicolin for 24 h were harvested, washed with PBS, and fixed in 70 % ethanol. Prior to analysis, cells were washed with PBS and then resuspended in the propidium iodide staining solution. Samples were analyzed on the flow cytometer.

Total RNA was isolated from HCC827 control and claudin-7 KD cells using Qiagen RNeasy Kit (Qiagen, Valencia, CA) following the manufacturer’s instructions and reverse transcripted into cDNA using RT² First Strand Kit (Qiagen). Real-time RT-PCR was performed using the RT² SYBR Green qPCR Mix (Qiagen). The relative changes in gene expression were analyzed using 2^-ΔΔCt method [16].

HCC827 or H358 control or claudin-7 KD cells grown on poly-D-lysine coated glass coverslips (BD Biosciences, Bedford MA) were fixed in 100 % methanol for 8 min at −20 °C and washed with PBS for 5 min before blocked in 5 % bovine serum albumin for 60 min at room temperature. After blocking, cells were incubated with primary antibodies. All antibodies were diluted in PBS containing 2.5 % BSA. After washing, cells were incubated with corresponding secondary antibody for 45 min at room temperature. Coverslips were mounted with ProLong Antifade Kit (Molecular Probes Inc., OR). Samples were photographed using a Zeiss Axiovert S100 or Zeiss LSM 510 laser confocal scanning microscopy (Carl Zeiss Inc., Thornwood, NY) and analyzed by MetaMorph software (Molecular Devices, Sunnyvale, CA).

HCC827 or H358 control cells were washed three times with ice-cold PBS and then lysed in RIPA buffer. After centrifugation, the supernatants were incubated with either anti-claudin-7 or anti-integrin β1 antibody at 4 °C overnight. Protein A or protein G beads were then added to the mixture, followed by incubation at 4 °C for 3 h. The beads were washed twice with RIPA buffer, once with high salt buffer (0.5 M NaCl), and once with Tris buffer (10 mM Tris, pH 7.4). Bound proteins were eluted from the beads in SDS sample buffer and analyzed by western blot.

For TER measurement, HCC827 control or claudin-7 KD cells were plated on collagen-coated Transwell filters with a pore size of 0.4 μm (Corning Costar, Cambridge, MA). A millicell-ERS volt-ohm meter (Millipore, Bedford, MA) was used to determine the TER value. All the TER values were normalized for the area of the filter and obtained after background subtraction.

Five-week old male athymic nude mice were obtained from Charles River Laboratory (Wilmington, MA) and used for human tumor xenografts. Two × 10⁶ HCC827 control or claudin-7 KD cells were suspended in the culture medium and injected subcutaneously into the left and right flanks of each nude mouse, respectively. All the mice were sacrificed eight weeks after injection. The tumors were removed and weighed. The animal experiments were performed according to the animal use protocol (AUP) approved by East Carolina University.

Statistical analysis was performed using Origin50 (OriginLab, MA) or SigmaPlot (SPSS Science, Chicago, IL) software. For each in vitro experiment, at least three independent experiments were performed. All data was recorded and expressed as means ± SE. The differences between two groups were analyzed using the unpaired Student’s t-test. The data from the nude mice tumor xnenograft experiment was analyzed using Wilcoxon Rank-Sum Test. All statistical tests were two-sided and a value of P < 0.05 was considered significant (*).