Introduction
Slits are large secreted proteins (up to 200 kDa) that regulate neuronal orientation and branching during nervous system development by signaling through cognate Robo receptors that consist of four known members (Robo1-4). In mammals, three Slit genes, Slit1, Slit2 and Slit3, have been identified. All the Slit proteins contain four leucine-rich repeat domains, six EGF domains, a laminin G domain and a cysteine-rich domain (1). All three Slit proteins are proteolytically cleaved into a large N-terminal fragment (140 kDa) and a shorter C-terminal fragment (50–60 kDa) (2,3). The majority of the N-terminal fragment and full-length Slits are membrane-associated, which mediates their function. The C-terminal fragment is more diffuse and although its function remains largely unknown, Condac et al recently reported the pro-coagulation effects of the Slit3 C-terminal fragment by neutralizing the anticoagulant activity of heparin (4).
Slit1 expression is specific to neurons, while Slit2 and Slit3 are detected in various tissues, such as the lung, kidney and heart (5,6). Compared with normal tissues, Slit genes are epigenetically silenced in a wide variety of cancers, indicating that Slits are potent tumor suppressors (7,8). To date, most studies involving the effects of Slits on tumors have focused on Slit2, including epigenetic modifications (9,10), expression in various types of cancer and inhibition of tumor growth, metastasis (11–17) and angiogenesis (18,19). Although most results support the inhibitory effects of Slit2 on tumor growth and metastasis, the role of Slit2 in growth and metastasis is still controversial. Several studies have validated a stimulative role of Slit2 on tumor growth and metastasis (20–22), and this discrepancy reflects the complexity of Slit/Robo signaling in different types of cancers.
Slit3 shares a high level of homology with Slit2, both with the same proteolytic cleavage site, TSPCD, at the beginning of the 6th EGF repeat (2). Slit3 is detected in various normal tissues, with skin, brain cerebellum and lung having the highest expression levels (8). Genetic inactivation of Slit3 in mice revealed multiple roles of Slit3 in organ development, particularly in the diaphragm and kidney (23,24). In the vascular system, Slit3 is secreted by endothelial cells and vascular smooth muscle cells to stimulate angiogenesis by binding to the Robo4 receptor (25,26). Similar to Slit2, frequent hypermethylation of promoters is the main cause of Slit3 gene silencing in most cancers (8,27,28). Lung carcinomas appear to be an exception. The frequency of Slit3 methylation in lung cancers is far less than Slit2 (8,29), indicating that an alternative mechanism apart from methylation is responsible for the downregulation of Slit3. Recombinant Slit3 was reported to inhibit the migration of melanoma cells (30); the second leucine-rich repeat domain is sufficient to induce such repressive effects (31). Reduction of miR-218, which resides in the introns of the Slit2 and Slit3 genes, causes downregulation of Slit3, resulting in promotion of metastasis in thyroid cancer (32). Slit3 overexpression in human breast carcinoma cells resulted in tumor growth by suppressing CXCR4 expression (17). These results suggest the negative regulation of Slit3 on tumor growth and metastasis, similar to the most reported functions of Slit2. However, the functions of Slit3 in lung carcinoma have not been widely reported.
In the present study, we investigated the effects of Slit3 on lung carcinoma cells by Slit3 gene silencing in A549 cells, which have a high level of Slit3 expression compared to other cancer cell lines. We validated the inhibitory effects of Slit3 on proliferation, migration and invasion in the cells, confirming a potential tumor-suppressor role of slit3 in lung carcinoma.
Materials and methods
Cell culture
Human glioblastoma U87MG cells, and human normal lung fibroblasts (HFL-1) were purchased from the National Platform of Experimental Cell Resources for Sci-Tech (Beijing, China). Human lung adenocarcinoma A549, Spc-a-1 and Ltep-a-2 cells were purchased from the Cell Resource Center at the Institute of Life Sciences, Chinese Academy of Science (Shanghai, China). All cells were cultured in high-glucose Dulbecco’s modified Eagle’s medium (DMEM; HyClone, Logan, UT, USA) supplemented with 10% fetal bovine serum (FBS; Gibco-BRL, Carlsbad, CA, USA).
Lentiviral infection
shRNA Slit3 oligos were annealed and cloned into the lentiviral vector pLentiLox 3.7 (a gift from Professor Jiahuai Han, State Key Laboratory of Cellular Stress Biology, School of Life Sciences, Xiamen university, China). The Slit3 N-terminal (33aa–1120aa, ~140 kDa) was amplified from pSecTag2B-hSlit3 (gift from the Institut National de la Santé et de la Recherche Médicale, Paris, France) and subcloned into the lentiviral expression vector pBobi (gift from Professor Jiahuai Han) with a myc tag at the carboxyl terminus, and the same vector encoding GFP was used as a control. Virus stocks were prepared by co-transfecting pll3.7 or pBobi with two packaging plasmids (pHR and pVSVG) into 293T cells. Viral supernatants were harvested after 48 h, filtered and centrifuged (90 min at 75,000 × g). A549 cells were infected with pll3.7-Slit3-shRNA in the presence of 8 μg/ml Polybrene (Sigma, St. Louis, MO, USA) for two days. Infected A549 cells were sorted by clone picking.
Real-time quantitative PCR
Total RNA was extracted with TRIzol reagent (Life Technologies, Carlsbad, CA, USA). cDNA for the PCR template was generated using the ReverTra Ace qPCR RT kit (FSQ-101; Toyobo, Osaka, Japan) as recommended by the manufacturer’s protocols. Primer sequences are listed in Table I. Real-time PCR was performed in an MJ Mini System (Bio-Rad) using Thunderbird SYBR qPCR Mix (QPS-201; Toyobo). Amplification was performed under the following conditions: 95°C for 1 min followed by 40 cycles at 95°C for 15 sec, and 62°C for 45 sec. To normalize the real-time PCR results, GAPDH was used as an internal control. Relative genes expression in various tumor cell lines was calculated using the Livak and Schmittgen (33) method and compared with the U87MG cells.
Immunofluorescence
Cells were fixed with 4% paraformal-dehyde/phosphate-buffered saline (PBS) for 15 min at room temperature. Fixed cells were washed twice with PBS and permeabilized with 0.25% Triton X-100/PBS for 15 min. Cells were blocked with 1.0% FBS/PBS for 1 h and labeled with rabbit polyclonal primary antibody for tubulin (T3526; Sigma) at a 1:80 dilution in 0.1% BSA/PBS buffer with 0.05% Tween-20 (PBST) at 37°C for 1 h. After incubation, the cells were washed 3 times with PBST and incubated with Alexa Fluor 594-conjugated goat anti-rabbit secondary antibody (A11037; Invitrogen) at a 1:1,000 dilution at room temperature for 1 h in the dark. After washing with PBST, the cells were stained with 4′,6-diamidino-2-phenylindole (DAPI) for 5 min. After extensive washing, the cells were examined and recorded with an inverted fluorescence microscope (IX71; Olympus, Japan) equipped with a SPOT Insight QE CCD camera.
Cell proliferation assay
A total of 500 cells were seeded/well in 96-well microplates (Sunub, Shanghai, China) and cultured for 4 consecutive days. The medium was changed every other day. On each day, one group of culture media was replaced with 100 μl serum-free fresh medium containing 10% MTT (Sigma) and maintained at 37°C for 4 h. After discarding the medium, 150 μl of dimethylsulfoxide (DMSO) was added to each well to dissolve MTT formazan. Optical densities were measured at 560 nm with a microplate reader (Multiskan MK3; Thermo Fisher Scientific, USA).
Cell migration and invasion assays
Cell migration was assessed by a wound healing recovery assay. Cells that were 90% confluent in a 24-well plate (Sunub) were scratched with a 200-μl pipette tip to form wound gaps. After washing out cell debris, the wound gaps were photographed under a microscope at a magnification of ×40 to acquire a baseline image. The cells were then cultured in DMEM for the indicated time periods and photographed to obtain the second set of images. Gap width was measured using Image-Pro Plus 6.0 software. Cell migration was quantitatively analyzed by subtracting the gap width of the second image from the baseline image.
Cell invasion was assessed by the Matrigel invasion assay. Chambers (8 μm; Millipore, Billerica, MA, USA) were coated with 20 μl diluted Matrigel (0.1 mg protein/ml; BD) for 30 min at 37°C and inserted into a 24-well plate. A total of 1×104 cells/chamber in 100 μl serum-free DMEM were plated on the upper Matrigel-coated chambers. Medium in the lower chambers contained 10% FBS as the source of chemoattractant. The plates were incubated for 24 h at 37°C. Cells that had invaded the lower surface at 37°C were fixed with methanol and stained with crystal violet. Three random fields were counted under a light microscope.
Gelatin zymography
A total of 2 million cells were cultured in a 6-cm dish (Sunub) with 5 ml DMEM (10% FBS) for 8 h. Media were replaced with 2 ml serum-free DMEM and incubated for an additional 12 h. The conditioned media was collected and filtered with a 0.22-μm filter (Millipore). A total of 10 μl of conditioned media was mixed with 10 μl sample buffer [0.25 M Tris-HCl, pH 6.0, 8.5% glycerol, 4% sodium dodecyl sulfate (SDS) and 0.01% bromophenol blue] and electrophoresed on a 7.5% SDS-polyacrylamide gel containing 2 mg/ml gelatin (Sigma). After electrophoresis, the gel was washed 3 times for 10 min in 2.5% Triton X-100 and placed in incubation buffer (50 mM Tris-HCl, pH 7.6, 10 mM CaCl2, 50 mM NaCl and 0.05% Brij-35) over night at 37°C. After incubation, the gel was stained with a solution of 0.25% Coomassie blue R250, 40% methanol and 10% acetic acid for 1 h at room temperature and destained with 40% methanol and 10% acetic acid until the protein bands were apparent.
Immunoblotting
To detect E-cadherin and vimentin, the cells were washed with cold PBS (pH 7.4), incubated with cold RIPA lysis buffer (P0013; Beyotime, China) on ice for 10 min, and collected with a scraper. Lysates were then centrifuged at 12,000 × g for 10 min. To detect membrane-associated Slit3 protein, a Qproteome Cell Compartment kit (Qiagen, Hilden, Germany) was used to isolate the protein. Protein samples were boiled at a ratio of 3:1 with sample buffer (250 mM Tris-HCl, pH 6.8, 40% glycerol, 20% mercaptoethanol, 8% SDS and 0.04% bromophenol blue) and separated by SDS-PAGE. Following electrophoresis, proteins on the gel were transferred to an Immobilon-P membrane (PVDF; 0.45 μm; Merck Millipore, Molsheim, France), which was blocked with 5% skim milk powder dissolved in TBS buffer with 0.05% Tween-20 (TBST). After washing, the membranes were incubated with rabbit polyclonal primary antibody for Slit3 (AB5703P; Merck Millipore) at a 1:500 dilution, rabbit monoclonal primary antibody for E-cadherin (#3195; Cell Signaling) at a 1:1,000 dilution, rabbit polyclonal primary antibody for vimentin (sc-5565; Santa Cruz) at a 1:1,000 dilution, mouse monoclonal antibody for α-tubulin (T6074; Sigma) at a 1:4,000 dilution and appropriate horseradish peroxidase-linked secondary antibodies at a 1:2,000 dilution. After washing 3 times with TBST, the bound antibody was developed using the ECL Plus Western Blotting Detection System (Thermo, USA).
Statistical analysis
GraphPad Prism software was used for all statistical analysis. Quantitative data are expressed as the mean ± SD and were compared using the unpaired Student’s t-test.
Results
Slit and Robo expression in the lung carcinoma cell lines
The Oncomine database indicates that Slit3 expression is downregulated in lung carcinoma compared to normal tissues (34,35). To verify the results in the tumor cell lines, we examined Slit and Robo expression in three lung adenocarcinoma cell lines (A549, Spc-a-1 and Ltep-a-2). Since Slit3 has been reported to be overexpressed in gliomas and is considered a biomarker for gliomas (36), we used the glioblastoma cell line u87MG as a positive control for Slit3.
Slit1 expression was not detected in any of the cell lines analyzed. Slit2 was detected in all of the cell lines and in general maintained a relatively lower expression when compared with that in the U87MG cells (Fig. 1A). Slit3 was only detected in the A549 and U87MG cells. A549 cells showed high Slit3 expression levels that were nearly 9-fold more than that of the U87MG cells. Of the three lung adenocarcinoma cell lines, high Slit3 expression was only detected in the A549 cells, indicating cell-specific Slit3 expression. However, high Slit3 expression in the A549 cells was not evident at the protein level. Although Slit3 is a secreted protein, both ELISA of the cell supernatant and immunoblotting of the cell lysates (up to 80 μg of total loading protein) did not detect Slit3 (data not shown). We theorized that the membrane-associated characteristics of Slit and the relatively low amount of Slit3 may account for the lack of Slit3 detection in the supernatant and cell lysates. Unexpectedly, immunofluorescence only detected a very weak signal on the A549 cell surface (data not shown).
Robo receptors are heterogeneously expressed by tumor cell lines. While Robo2 was not detected in any of the cell lines assayed, Robo1 was detected in all of the cell lines. A549 cells maintained a relatively high level of Robo1 expression compared with the other cell lines. Robo3 and Robo4 were only detected in the U87MG and A549 cells, yet not in the Spc-a-1 and Ltep-a-2 cells (Fig. 1B).
Slit3 knockdown in A549 cells
To evaluate the role of Slit3, A549 cells were transduced with the VSVG pseudotyped lentivirus plasmid (pLL3.7) for stable Slit3 silencing (A549-shRNA) or an empty control plasmid without Slit3-shRNA insertion (A549-shctrl). Silencing of Slit3 mRNA in the A549 cells was first verified by real-time quantitative PCR. Slit3 mRNA expression in the A549 cells was significantly decreased, with a knockdown efficiency of 78 and 69%, respectively (Fig. 2A). Given that Slit3 is predominantly membrane-associated, we isolated the membrane-associated proteins in the A549 cells using a Qproteome Cell Compartment kit to detect the changes in Slit3 protein levels. We did not detect Slit3 protein bands when using 60–120 μg of total loading protein, yet detected it when using 180 μg of total sample protein (Fig. 2B). The discordance between high Slit3 mRNA expression and protein abundance suggests that post-translational mechanisms influenced Slit3 abundance in the A549 cells. The molecular size of the Slit3 protein band detected in the A549 cells was slightly larger than the theoretically calculated value of N-terminal Slit3 (~140 kDa), which reflects cellular glycosylation of the slit3 protein. The results are consistent with the findings that the majority of Slits are secreted in a membrane-associated manner. Notably, full-length Slit3 (~200 kDa) was not detected, indicating that proteolytically processed N-terminal Slit3 is the primary form of secreted Slit3 in the A549 cells.
Slit3 silencing induces EMT in A549 cells
Slit3 silencing induced significant morphologic changes in the A549 cells as demonstrated by an elongated fibroblast-like morphology, a phenomenon reminiscent of EMT (Fig. 3). We further examined the effects of Slit3 on the expression of molecular hallmarks of EMT, including the epithelial marker E-cadherin and the mesenchymal marker vimentin. Immunoblotting analysis confirmed that E-cadherin protein expression was obviously reduced in the A549-shRNA cells when compared with the controls. Corresponding to reduced E-cadherin expression, vimentin expression was significantly increased.
Slit3 silencing enhances the proliferation, migration and invasion of A549 cells
To evaluate the effects of Slit3 on the proliferation of A549 cells with Slit3 knockdown, the MTT assay was performed. Compared with the controls, the Slit3 knockdown significantly stimulated the proliferation of the A549 cells (Fig. 4). Cell migration was measured by a wound healing assay in which the Slit3 knockdown significantly promoted the migration of the A549 cells when compared to that of the controls (Fig. 5A). Similar results were also observed in the Matrigel invasion assay. Slit3 knockdown enhanced the ability of A549 cells to traverse the Matrigel-coated membrane (Fig. 5B). These results suggest that Slit3 exerts negative regulation on the proliferation and invasive behavior of A549 cells, thereby confirming the role of slit3 as a tumor suppressor.
Slit3 silencing upregulates MMP2 and MMP9 expression
To explore the mechanisms underlying the stimulative effects of Slit3 silencing on the invasion of A549 cells, we detected MMP2 and MMP9 expression, which are key enzymes for degradation of type IV collagen, a major component of the basement membrane. Quantitative PCR results showed that silencing of Slit3 in the A549 cells caused significant upregulation of both MMP2 and MMP9 (Fig. 6A). To verify the effects of Slit3 silencing on MMP2 and MMP9 secretion, gelatin zymography was performed. Supernatants were prepared by incubating the cells with serum-free DMEM for 12 h to decrease the effects of cell proliferation on MMP production. As shown in Fig. 6B, A549 control cells maintained very weak MMP2 secretion, whereas MMP9 production was not detected. Slit3 silencing in the A549 cells induced an increase in MMP2 production. Although MMP9 was not detected (Fig. 6B, left), Slit3 silencing in the A549 cells caused significant upregulation of MMP9 at the mRNA level (Fig. 6A). Lack of MMP9 detection could be due to low MMP9 expression in A549 cells. Quantitative PCR confirmed that MMP9 expression was significantly lower than MMP2 in the A549 cells (data not shown). To determine whether weak MMP2 and MMP9 expression is common in lung carcinoma cell lines, culture supernatants from the Spc-a-1 and Ltep-a-2 cell lines were also assayed using gelatin zymography. Human normal lung fibroblasts (HFL-1) were used as a positive control. HFL-1 maintained high levels of MMP2, while the two lung carcinoma cell lines showed weak MMP2 production. MMP9 was not detected in any of the cell lines (Fig. 6B, right). These results suggest that relatively weak MMP2 and MMP9 expression may be a common characteristic in the lung carcinoma cell lines analyzed.
Discussion
Although a number of studies have explored the roles of Slit2 in tumor progression, relatively little progress has been made in defining the effects of Slit3 on tumors. In contrast to Slit2, which was reported to be upregulated in various tumors (18,20), few results concerning Slit3 upregulation in tumor tissues have been reported. Although microarray analysis revealed Slit3 upregulation in separated invasive hepatocellular carcinoma cells, no evidence of protein level upregulation has been provided (37). Moreover, Slit3 upregulation at the mRNA level does not necessarily lead to an increase in Slit3 at the protein level. Chen et al examined the correlation between mRNA expression and protein levels in lung adenocarcinomas and found a significant correlation in only a small subset of the proteins studied (38). Towner et al recently reported high Slit3 expression on the surface of high-grade human gliomas compared to low-grade gliomas and normal brain tissues using immunohistochemistry and suggested Slit3 as a potent biomarker for human high-grade gliomas in clinical diagnosis (36). These results are in contrast to reports that Slit3 is frequently methylated in gliomas (8), indicating heterogeneous Slit3 expression in gliomas. The present study confirmed Slit3 expression in the u87MG glioblastoma cell line and detected higher Slit3 expression in lung carcinoma A549 cells. These results are consistent with reports that the frequency of promoter methylation is lower in lung carcinomas (8,29) and partially explained high Slit3 expression in A549 cells. However, the high slit3 expression in A549 cells was not confirmed in two other lung adenocarcinoma cell lines (Spc-a-1 and Ltep-a-2), indicating cell-specific Slit3 expression that was not apparent at the protein level.
As in the u87MG cells, Slit3 proteins were not detected in the supernatant using ELISA. However, immunofluorescence detected a very weak signal on the cell surface, indicating that Slit3 in A549 cells remains mostly membrane-associated. Immunoblotting further validated the trace amounts of Slit3 associated with the cell membrane. These results suggest that post-translational mechanisms influence Slit3 protein abundance in cells despite mRNA expression. Given that Slit3 functions as a latent tumor suppressor, we suggest that post-translational regulation of Slit3 confers a self-protective advantage against the tumor-suppressor effects of Slit3.
In contrast to the findings that both full length and N-terminal fragments of Slit are present in neuronal cells (39) or tissues (40), only the N-terminal of the Slit3 protein (~140 kDa) was detected at the A549 cell surface, indicating that the extent of proteolytic cleavage of the Slit3 protein differs by tissue type. The cleaved Slit fragment also exerts different functions. Nguyen et al demonstrated that full length Slit2 neutralized the branching of dorsal root ganglia axons induced by N-terminal Slit2 (41), suggesting that the function of Slit was partly associated with its structure. In the present study, lack of full length Slit3 in A549 cells may imply the loss of an unknown regulatory mechanism that balances the effects of the N-terminal Slit3 fragment.
Recombinant Slit3 was reported to inhibit the migration of malignant melanoma cells through downregulation of AP-1 (30). Slit3 overexpression in human breast carcinoma cells inhibits tumor growth by suppressing Sdf1 and Cxcr4 expression (17). The present study indicated that Slit3 silencing in A549 cells promoted proliferation, migration and invasion of the cells even though only trace amounts of Slit3 protein were detected. These results are evidence that Slit3 exerts negative regulatory effects on tumor proliferation and migration. Results are in accordance with reported suppressive effects of Slit2 on growth and migration in various types of tumor. In nervous system development, Slits function as repellent factors that prevent axon growth into particular regions (42). Increased migration due to Slit3 silencing supports the idea that repellent effects may also influence tumor cell migration.
EMT plays a critical role in tumor metastasis by which tumor cells weaken E-cadherin-dependent intercellular adhesion and enhance motility, facilitating tumor cells to invade surrounding tissues. Chen et al reported an inhibitory effect of Slit2 on EMT wherein Slit2 overexpression in colorectal cancer cells inhibited cell migration as a result of increasing E-cadherin levels, while Slit2 knockdown led to opposite results (11). However, Slit2 promotion of EMT was also reported in colorectal cancer cells (21), the different cell lines analyzed accounted for conflicting results. The results of the present study confirmed the inhibitory effects of Slit3 on EMT in lung carcinoma A549 cells given that Slit3 silencing induced an elongated fibroblast-like morphology, downregulation of E-cadherin and upregulation of vimentin.
MMPs play a critical role in tumor metastasis. The prometastatic or antimetastatic effects of Slit are likely associated with MMP expression or protein activities. Qi et al recently reported upregulation of MMP2, yet not MMP9 in skin tissue of Slit2-transgenic mice and proposed that MMP2 upregulation accounted for loss of the basement membrane in Slit2-transgenic mice, indicating a positive correlation between Slit2 and MMP2 expression (20). However, Prasad et al reported that Slit2 inhibits the CXCL12-induced activities of MMP2 and MMP9 in breast cancer cells (43). Our results showed that MMP2 and MMP9 expression was negatively associated with alteration of Slit3 expression. Slit3 silencing promoted MMP2 and MMP9 expression in A549 cells, whereas Slit3 overexpression reduced expression, supporting the negative correlation between Slit3 and MMP2 and MMP9 expression, even though only MMP2 secretion was detected. Conflicting results, however, were recently reported in primary amnion and myometrial cells wherein Slit3 knockdown with siRNA decreased MMP9 gene expression and secretion (44). Mechanisms underlying these results may be associated with heterogeneous expression of Robo receptors in various tissues. Of the four Robo receptors, Robo4 is structurally divergent from the other three members both in extracellular and intracellular domains and may stimulate signaling pathways that differ from other receptors, resulting in a functional difference. For example, Slit2/Robo4 and Slit2/Robo1 signaling pathways induce opposite effects on the migration of endothelial cells (45). In the present study, Robo1 and Robo4 were detected in A549 cells and Robo1 maintained a relatively high level of expression. Therefore, we suggest that Slit3 most likely exerts its effects in A549 cells by binding to Robo1.
In summary, these results demonstrated that Slit3 functions as a tumor suppressor in lung carcinoma. Due to promoter hypermethylation, overall expression of Slit3 is downregulated in most tumors. Even if hypomethylation of the promoter occurs in some cancer tissues, such as lung carcinoma, and results in high slit3 expression, transcription of the slit3 protein is still strictly controlled. Slit3 represses migration and invasion of A549 cells by blocking EMT and suppressing both MMP2 and MMP9 expression and secretion. Further studies are needed to decipher the mechanisms underlying the post-translational regulation of Slit3 and suppression of MMP2 and MMP9 expression and secretion.