JNK signaling maintains the mesenchymal properties of multi-drug resistant human epidermoid carcinoma KB cells through snail and twist1

Rabbit monoclonal phospho-JNK1/2 and c-Jun were purchased from Cell Signaling Technology (Beverly, MA). Rabbit monoclonal antiserum against JNK1/2, c-Jun, E-cadherin, N-cadherin, vimentin, snail, twist1, GAPDH, and β-actin were obtained from Santa Cruz Biotechnology (Santa Cruz, CA); SP600125 was purchased from Sigma (St. Louis, MO). JNK1/2 shRNA (5^′-CCGGAAAGAATGTCCTACCTTCTTTCTCGAGAAAGAAGGTAGGACATTCTTTTTTTTG-3^′) and shRNA control cloned into pMAGIC 7.1 lentiviral system were purchased from Sunbio (Shanghai, China). The c-Jun siRNA (5^′-GCGGGAGGCAUCUUAAUUATT-3^′) and control were obtained from GenePharma (Shanghai, China)

KB human epidermoid carcinoma cells and human epithelial kidney 293 T cells were obtained from the American Type Culture Collection (Manassas, VA) and maintained in Dulbecco’s modified Eagle’s medium (Sigma; St. Louis, MO) containing 10% fetal calf serum. The VCR-selected multiple-drug tolerant KB/VCR subline was obtained from Zhongshan University of Medical Sciences (Guangzhou, China) and routinely cultured in medium containing VCR (200 ng/ml). KB/VCR cells were cultured in VCR-free medium for at least 3–7 days prior to be used for experiments to avoid drug-associated secondary effects, and were cultured in absence of VCR for no more than 15 days to keep the MDR phenotype. The KB/VCR resistant cells were authenticated by comparing their fold resistance with that of the parental cells and examining the expression level of ABC transporters (ABCB1 and ABCG2) in KB/VCR cells cultured in the presence, or absence of VCR for about 15 days.

The c-Jun siRNA and control were transfected to KB/VCR cells using lipefectamine 2000 (Invitrogen; Carlsbad, CA) as the instructions provided by the manufacturer.

The JNK1/2 shRNA and control were transfected to 293 T cells using lipefectamine 2000 (Invitrogen; Carlsbad, CA) as the instructions provided by the manufacturer. Viral stocks were prepared and infections performed as previously reported [16].

Migration assay and invasion assay were determined using a transwell system (Corning Costar; Acton, MA) with an 8-μm pore size membrane coated with fibronectin for migration assay or with matrigel for invasion assay as described by Li et al. [17]. Briefly, 100 μl (5 × 10⁴ cells) of KB or KB/VCR cells was added to the upper wells and 600 μl of DMEM with or without 10% FBS, which was used as a chemoattractant, was added to the lower wells. After incubation for 8 h at 37°C, the migrated cells or invaded cells were fixed with 90% EtOH and then stained with 0.1% crystal violet in 0.1 mol/L borate and 2% EtOH (pH 9.0). The stained cells were subsequently extracted with 10% acetic acid. The absorbance values were determined at 570 nm with a Spectrophotometer (BioTek; Winooski, VT).

Total RNA was isolated with the RNAiso Plus Kit from TaKaRa (Dalian, China) as the instructions provided by the manufacturer. Total RNA was reversely transcribed using Moloney Murine Leukemia Virus (M-MLV) reverse transcriptase (TaKaRa; Dalian, China) and cDNAs were used for PCR with the following primers (Invitrogen; Shanghai, China): Snail: 5^′-GAGGCGGTGGCAGACTAG-3^′, 5^′-GACACATCGGTCAGACCAG-3^′; twist1: 5^′-GGAGTCCGCAGTCTTACGAG-3^′, 5^′-TCTGGAGGACCTGGTAGAGG-3^′; HIF-1: 5^′-CAGCTATTTGCGTGTGAGGA-3^′,5^′-CCAAGCAGGTCATAGGTGGT-3^′; NF-κB: 5^′-GGCGAGCAACTCAATAAAGC-3^′, 5^′-GAGCAAAGGACTGCCAAGAC-3^′; Foxo2: 5^′-GATCACCTTGAACGGCATCT-3^′, 5^′-ACCTTGACGAAGCACTCGTT-3^′; slug: 5^′-CTTTTTCTTGCCCTCACTGC-3^′, 5^′-ACAGCAGCCAGATTCCTCAT-3^′; ZEB1: 5^′-GAGAAGCGGAAGAACGTGAC-3^′, 5^′-GCTTGACTTTCAGCCCTGTC-3^′; ZEB2: 5^′-TTCCTGGGCTACGACCATAC-3^′, 5^′-GCCTTGAGTGCTCGATAAGG-3^′; Stat3: 5^′-ACATTCTGGGCACAAACACA-3^′, 5^′-CACACCAGGTCCCAAGAGTT-3^′; Stat5a: 5^′-ACATTTGAGGAGCTGCGACT-3^′, 5^′-CCTCCAGAGACACCTGCTTC-3^′; TATA: 5^′-ACCCTTCACCAATGACTCCTATG-3^′, 5^′-TGACTGCAGCAAATCGCTTGG-3^′. The PCR products were analyzed using agarose gel electrophoresis.

Total RNA was subjected to reverse transcription with the kit from TaKaRa (Dalian, China) according to the manufacturer’s instructions. Then the cDNAs were amplified by Real-time PCR (iQ5; Bio-Rad) with the SYBR-Green kit (TaKaRa, Dalian, China) with the primers (Invitrogene; Shanghai, China) mentioned above. The alteration of mRNA expression in cells was assessed using the iQ5 optical system software by delta delta Ct method.

Cells were lysed in lysis buffer (50 mM Tris–HCl, 150 mM NaCl, 1% Nonidet P-40) supplemented with protease inhibitors (0.5 mM phenylmethylsulfonyl fluoride, 1 μg/ml aprotinin and leupeptin) for 15 min on ice. Equal amounts of protein were subjected to SDS-polyacrylamide gel electrophoresis and transferred onto a polyvinylidene difluoride membrane (Immobilon P; Millipore). The membranes were then incubated with the appropriate antibodies as indicated.

The data of the experimental studies were expressed as the average ± s.d. Statistical differences were analyzed by the two-tailed Student’s t test and P < 0.05 was considered as significant.

We employed the MDR cell line KB/VCR, a well established MDR cell line [15], to firstly explore whether the MDR cells possess mesenchymal properties. We observed that the KB/VCR cells exhibited elongated, fibroblastic morphology (Figure 1A). On the contrary, the parental cells KB exhibited epithelial cobblestone phenotype (Figure 1A). These observations suggested that the KB/VCR cells lost the epithelial morphology and acquired mesenchymal phenotype when compared to parental KB cells. We next evaluated the expression of markers for epithelial and mesenchymal characteristics in KB/VCR cells and parental KB cells with western blot analysis. In line with the alterations in morphology between KB/VCR and KB cells, we found that the expression of the epithelial marker E-cardherin was decreased in KB/VCR cells compared to KB cells, while the expression of mesenchymal markers including N-cadherin and vimentin was increased when compared to KB cells (Figure 1B). We then sought to compare the motile behavior including migration and invasion abilities between KB cells and KB/VCR cells utilizing a transwell system. Our results revealed that KB/VCR cells migrated and invaded more efficiently than KB cells in response to 10% serum, which was used as a chemoattractant (Figure 1C-D). These data together indicate that KB/VCR cells undergo epithelial-mesenchymal transition and acquire more powerful motile capacity in comparison to KB cells.

JNK signaling has been implicated in the occurrence of MDR during chemotherapy [13, 14], whereas there is little knowledge regarding the contribution of JNK signaling to EMT properties and motile ability of MDR cells. We then set out to explore the influence of JNK signaling on the mesenchymal phenotypes of MDR cells. To this end, we firstly assessed whether JNK signaling is activated in KB/VCR cells by evaluating the phosphorylation status of JNK1/2 and c-Jun, a key transcription factor downstream of JNK1/2. Western blot analysis revealed that JNK1/2 and c-Jun were obviously phosphorylated when compared to that in KB cells, whereas there was no alteration in the total protein expression for JNK1/2 and c-Jun. These results suggest that JNK signaling is activated in KB/VCR cells when compared to KB cells (Figure 2A). Keeping this in mind, we then asked whether interference with JNK signaling would disrupt the mesenchymal properties of KB/VCR cells. We began by blockade of the JNK1/2 signaling system through targeting JNK1/2 with a small molecular compound SP600125 (SP), a specific and widely used antagonist of JNKs [18], to inhibit the JNK1/2 activity in KB/VCR cells. As expected, treatment of KB/VCR cells with SP significantly suppressed the phosphorylation of JNK1/2 and c-Jun in KB/VCR cells (Figure 2B), confirming that SP can be useful for investigating the contribution of JNK1/2 in maintaining the mesenchymal phenotype of KB/VCR cells. After treated with SP 10 μM for 24 h, the KB/VCR cells lost the fibroblatic mesenchymal morphology and recovered the epithelial cobblestone phenotype (Figure 2C) similar to that possessed by KB cells as shown in Figure 1A. Meanwhile, exposure of the KB/VCR cells with SP 10 μM for 24 h obviously reduced the expression of mesenchymal markers N-cadherin and vimentin in KB/VCR cells, accompanying with elevated expression of epithelial marker E-cadherin (Figure 2D). Moreover, we observed that the expression of epithelial and mesenchymal markers in KB/VCR cells after treated with SP was almost recovered to that in KB cells (Figure 2D). In line with these alterations in the expression of epithelial and mesenchymal markers, we found that after treatment with SP, the migration (Figure 2E) and invasion (Figure 2F) of KB/VCR cells in response to serum were abundantly suppressed. Taken together, these observations support that JNK1/2 activation is required for maintaining the mesenchymal properties of KB/VCR cells.

To further address the direct role of JNK1/2 activation in maintaining the mesenchymal characteristics of KB/VCR cells, we set out to knockdown JNK1/2 using RNA interference approach. We designed a shRNA sequence targeting both JNK1 and JNK2 and expressed it stably in KB/VCR cells using a lentiviral system. The western blot analysis revealed that the expression of JNK1/2 in KB/VCR cells infected with lentiviral mediated JNK1/2 shRNA was significantly decreased when compared to that of cells infected with the shRNA control (Figure 3A). Consequently, we observed that the phosphorylation of c-Jun was also reduced as anticipated (Figure 3A). Meanwhile, we found that limiting expression of JNK1/2 resulted in KB/VCR cells re-gaining an epithelial cobblestone phenotype (Figure 3B), which was similar to that of KB cells (Figure 1A) and contrast to the elongated, fibroblastic morphology of KB/VCR cells infected with shRNA control (Figure 3B). In alignment with the results obtained by the pharmacological tool SP, limiting the expression of JNK1/2 led to increase the expression of epithelial marker E-cadherin and to decrease the expression of mesenchymal markers N-cadherin, vimentin, when compared to that of the KB/VCR cells infected with shRNA control (Figure 3C). Furthermore, we observed that knockdown the expression of JNK1/2 obviously inhibited the migration and invasion of KB/VCR cells responding to serum (Figure 3D, E). These observations together with those obtained by small molecular antagonist SP demonstrate that JNK1/2 activation is quite important for maintaining the mesenchymal phenotype of KB/VCR cells.

The oncogenic function of JNKs is mostly based on their ability to phosphorylate c-Jun [11]. We then further investigated the efficiency of c-Jun on the mesenchymal phenotype of KB/VCR cells by eliminating the expression of c-Jun with small RNA interference approach. As expected, limiting the expression of c-Jun (Figure 4A) caused elevation of the expressions of epithelial marker E-cadherin and concomitantly reduction in the expression of mesenchymal markers N-cadherin and vimentin in KB/VCR cells (Figure 4A), paralleling to the observations achieved via interference with JNK1/2 by pharmacological and shRNA approaches. Consistent with the changes in the expression of epithelial and mesenchymal markers, knocking down the expression of c-Jun also resulted in preventing the migration and invasion of KB/VCR cells in response to serum (Figure 4B, C). These data together indicate that c-Jun, as well as JNK1/2, may be involved in maintaining the mesenchymal phenotype of KB/VCR cells, further supporting the argument that the JNK signaling is strictly required for maintaining the mesenchymal properties of KB/VCR cells.

A wide range of transcriptional factors activated by extracellular signaling have been directly or indirectly implicated in controlling the expression of epithelial and mesenchymal markers and eventually the phenotypes of EMT [10]. We thus sought to explore the potential transcriptional factors involved in the mesenchymal phenotypes maintained by JNK signaling in KB/VCR cells. Using RT-PCR approach, we firstly detected the expression of a panel of transcriptional factors including snail, slug, twist1, HIF1α, NF-κB, Foxo2, ZEB1, ZEB2, stat3 and stat5a, which all have been well characterized in inducing of EMT [10], at transcript level in KB/VCR cells compared to parental ones. Of interest, we observed that there were no alterations in the expression at transcript levels of the majority of those transcriptional factors, with exemption of snail and twist1, whose expression at mRNA level was obviously elevated in KB/VCR cells compared to that of parental KB cells (Figure 5A). This observation was further validated by real time PCR and western blot analysis at mRNA and protein levels, respectively (Figure 5B-C), thus suggesting the possibility of those two transcriptional factors being involved in mediating the EMT controlled by JNK signaling in KB/VCR cells. We then set out to gain direct evidences regarding the involvement of snail and twist1 in the mesenchymal properties controlled by JNK signaling in KB/VCR cells. Blockade of JNK1/2 activation with pharmacological tool SP (Figure 5D) or JNK1/2 shRNA (Figure 5E) approaches caused an obviously reduction in the expression of snail and twist1 when compared to that of control KB/VCR cells. Similar results were observed by mean of silencing the expression of c-Jun with siRNA (Figure 5F). Collectively, these observations suggest that JNK signaling may control the mesenchymal properties of KB/VCR cells through acting on both snail and twist1.