IL-6 promotes head and neck tumor metastasis by inducing epithelial-mesenchymal transition via the JAK-STAT3-SNAIL signaling pathway

Cell culture and reagents

Primary human dermal microvascular endothelial cells (ECs) were purchased from Lonza (Walkersville, MD). ECs were maintained in Endothelial Cell Basal Medium-2 (EBM-2) containing 5% FBS and growth supplements. Head and neck squamous carcinoma cell (HNSCC) line CAL27 (ATCC) was maintained in Dulbecco's Modified Eagle Medium (DMEM) supplemented with 10% FBS. Immortalized (HPV E6/E7) oral epithelial cells (IOE) were provided by Drs. William Foulkes and Ala-Eddin Al Moustafa (McGill University, Montreal, Quebec, Canada) (32) and cultured in keratinocyte serum-free medium (K-SFM, Invitrogen, Carlsbad, CA). Primary antibodies against E-Cadherin (Cat # 3195), Akt (Cat # 9272), pAkt (Cat # 9271), STAT3 (Cat # 9132), STAT3 (pY705) (Cat # 9131), STAT3 (pS727) (Cat # 9134), p44/42 MAPK (ERK1/2, Cat # 9102), pERK1/2 (Cat # 4377), FAK (Cat # 3285) and Tubulin (Cat # 2148) were purchased from Cell Signaling (Danvers, MA); Snail (Cat # AB70983), pFAK (Cat # AB4803), and Vimentin (Cat # AB20346) were from Abcam; pJAK1 (Cat # 44422G) and pJAK2 (Cat # 44426G) were from Invitrogen (Carlsbad, CA) and IL-6 (Cat # AF-206-NA) was from R&D Systems (Minneapolis, MN). Recombinant IL-6, IL-8 and CXCL1/GRO-alpha proteins were purchased from PeproTech Inc (Rocky Hill, NJ).

Transduction of endothelial cells with Bcl-2 and CAL27 and IOE cells with IL-6

Bcl-2 was introduced into human microvascular endothelial cells (ECs) as described previously (33). Interleukin-6 (IL-6) expression plasmid IL6- pTargeT (Promega, Madison, WI; a kind gift from Dr. Tushar Patel, The Ohio State University) containing full-length IL-6 (34) was amplified using Vent polymerase (New England Biolabs, Beverly, MA). To generate retroviral expression vector containing IL-6, the cDNA for full-length IL-6 was amplified using the forward: 5′-cggaattcTCCAGGAGCCCAGCTATG-3′ and the reverse: 5′-taggatccAATCTGAGGTGCCC ATG-3′ primers and the PCR product was sub cloned at the EcoRI/BamHI site of pLXSN vector (Clontech, Palo Alto, CA). We verified the authenticity of the expression vector by sequencing the plasmid DNA. IL-6 was introduced into CAL27 and IOE cells by using the same protocol as described above for Bcl-2 transduction.

Tumor and endothelial cell co-culture

1 × 10⁵ Endothelial cells overexpressing Bcl-2 (EC-Bcl-2) or endothelial cells transduced with vector alone (EC-VC cells) were layered on top of collagen coated 6-well transwell inserts (3 μM, Greiner Bio-One, Monroe, NC) and these transwell inserts containing endothelial cells were then carefully placed on top of 6-well plates containing tumor cells. After 72 hrs of co-culture, transwell inserts were removed and CAL27 cell lysate was prepared for Western Blotting. In the immunofluorescence staining experiments, CAL27 cells were cultured on top of cover slips in 6-well plates while the rest of the protocol was same as described above. For IL-6 neutralizing experiments, CAL27 cells were co-cultured with EC-Bcl-2 in presence of neutralizing anti-IL-6 antibody (50 ng/ml)) or isotype control antibody.

ELISA

Endothelial cells, CAL27 or IOE cells were cultured in 6-well plates till they were 80% confluent. Cells were washed and fresh media was added. After 24 hours, culture supernatants were collected from 3 separate experiments and cell number in each well was counted. Chemokines levels (IL-8, IL-6 and CXCL1/GRO-α) in culture supernatants were measured using Quantikine human ELISA kits (R&D Systems, Minneapolis, MN) as per manufacturer's instructions and normalized to 1×10⁵ cells.

Transient transfection with siRNA

CAL27 overexpressing IL-6 (CAL27-IL-6) and IOE cells overexpressing IL-6 (IOE-IL-6) cells were transfected with siRNA for Akt, STAT3 or p42/42 MAPK (ERK1/2) using Signal Silence siRNA's from Cell Signaling according to the manufacturer's instructions. Seventy two hours post transfection, cells were either used for migration experiments or whole cell lysates were prepared for Western blotting.

Western Blot Analysis

Whole cell lysates were separated by 4-12% NuPAGE Bis-Tris gels (Invitrogen, Carlsbad, CA) and transferred onto PVDF (GE Healthcare, Waukesha, WI) membranes. Nonspecific binding was blocked by incubating the blots with 3% BSA in Tris buffered saline containing 0.1% Tween-20 (TBST) for 1 hr at room temperature (RT). The blots were then incubated with primary antibody in TBST + 3% BSA at 4°C overnight. After washing with TBST, the blots were incubated with horseradish peroxidase-conjugated sheep anti-mouse IgG (1:10,000) or with donkey anti-rabbit IgG (1:10,000) for 1 hr at RT. An ECL-plus detection system (Amersham Life Sciences, Piscataway, NJ) was used to detect specific protein bands. Protein loading in all the experiments was normalized by stripping the blots and then re-probing with anti-tubulin antibody. Alpha Innotech (San Leandro, CA) imaging software was used to quantify Western blot bands.

Immunofluorescent staining

Endothelial cells, CAL27 cells or IOE cells were cultured in labtech chambers. CAL27 and IOE cells were treated with recombinant IL-6 (50 ng/ml) for different time points. At the end of incubation, cells were fixed with 4% paraformaldehyde for 15 minutes at RT and permeablized by treating with 100% methanol for 10 minutes at -20°C. Next, slides were washed with PBS, blocked with normal goat IgG for 1 hr at room temperature and incubated overnight at 4°C with rabbit anti-E-cadherin and mouse anti-vimentin antibodies. After washing with PBS, chamber slides were incubated with secondary antibodies (goat anti-mouse-IgG-Alexa Fluor 488 and goat anti-rabbit-IgG-Alexa Fluor 594). Chamber slides were then mounted with Prolong gold antifade Reagent with DAPI (Invitrogen). The fluorescent images were captured using Nikon Eclipse 80i microscope with DS-Ri1 camera at 600× magnification and overlaid using NIS-Elements-Basic Research software (Nikon, Melville, NY).

Tumor cell motility assay

Tumor cell motility was examined by Xcelligence system using the RTCA DP instrument (Roche, Mannheim, Germany) as per manufacturer's instructions. In brief, 160 μl of media (DMEM for CAL27 cells and K-SFM for IOE) containing 10% FBS was added to the lower chambers. The upper chamber (sensor surface facing down) was then carefully assembled on top of lower chamber and 50 μl of serum free media was added to the wells. After 1 hr of equilibration with media, 100 μl of cell suspension (50,000 cells/well in serum free media) was added to each well. Cell migration to lower chamber was monitored and expressed as cell migration index at 24 hrs.

Tumor metastasis models

CAL27 transduced with vector alone (CAL27-VC) or CAL27-IL-6 (1×10⁶) and endothelial cells (1×10⁶) were mixed with 100 μl of Matrigel and injected subcutaneously in the flanks of SCID mice as described previously (25). Tumor volume measurements began on day 6 and continued twice a week until the end of the study. The length and width of the tumors were measured using a digital caliper and tumor volumes were calculated using the formula, volume (mm³) = L × W²/2 (length L, mm; width W, mm). After 40 days, primary tumors, regional lymph nodes, and lungs were carefully removed. To label and identify flank draining lymph nodes, 25 μl of 5% Evans Blue dye (Sigma, St. Louis, MO) was injected in the foot pads of animals 30 minutes before euthanizing (35). Primary tumors, lymph nodes and lungs were fixed with 4% paraformaldehyde and then processed for immunohistochemistry. Tumor tissue sections were deparaffinized and antigen retrieval was achieved by pressure cooking in a Decloaking chamber at 120°C for 20 minutes (36). Slides were then washed with TBST, blocked with normal goat IgG for 1 hr, and incubated overnight at 4°C with rabbit anti-pSTAT3 or Rabbit anti-snail or rabbit anti-E-cadherin and mouse anti-vimentin antibodies. After washing with PBS, slides were incubated with secondary antibodies (goat anti-mouse-IgG-Alexa Fluor 488 and goat anti-rabbit-IgG-Alexa Fluor 594). Slides were then mounted with Prolong gold antifade reagent with DAPI (Invitrogen, Carlsbad, CA). The fluorescent images were captured using Nikon Eclipse 80i microscope with DS-Ri1 camera at 400× or 600× magnification and overlaid using NIS-Elements-Basic Research software (Nikon, Melville, NY). Lymph nodes were stained with mouse anti-pan cytokeratin antibody (ab7753, Abcam).

Statistical analysis

Data from all the experiments are expressed as mean ± SEM. Statistical differences were determined by two-way analysis of variance and Student's t test. A p value of <0.05 was considered significant.

Bcl-2 expressing endothelial cells (EC-Bcl-2) promote EMT changes via the secretion of IL-6

We have previously shown that Bcl-2 expression is significantly elevated in tumor-associated blood vessels of head and neck cancer patients as compared to matched control samples (27). Recently, we demonstrated that up-regulation of Bcl-2 in tumor-associated endothelial cells is sufficient to enhance tumor metastasis, in vivo (25). However, the molecular mechanism by which EC-Bcl-2 promotes tumor metastasis is poorly understood. In this study, we examined if secreted factors from EC-Bcl-2 could induce EMT-related changes in head and neck tumor cells. Indeed, co-culture of EC-Bcl-2 along with tumor cells (CAL27) markedly enhanced vimentin expression (76% increase) while significantly decreasing E-cadherin expression (52% decrease) in CAL27 cells (Fig. 1A-B). In addition, co-culture of EC-Bcl-2 with CAL27 also upregulated snail expression, a key repressor of E-cadherin (Fig. 1B). We next examined the chemokine profile of EC-Bcl-2 cells. Culture supernatants from EC-Bcl-2 cells contained significantly higher levels of IL-8, IL-6 and Gro-α as compared to endothelial cells transduced with vector alone (EC-VC) (Fig. 1C). To further examine the role of these chemokines in EMT process, CAL27 cells were treated with recombinant IL-8, IL-6, or Gro-α for 3 days and E-cadherin and vimentin levels were analyzed by Western blotting. Interestingly, only IL-6 was able to significantly induce the expression of vimentin (63%) while repressing E-cadherin expression in CAL27 (51%, Fig. 1D). Similarly, co-culture of CAL27 cells and EC-Bcl-2 in the presence of neutralizing IL-6 antibody significantly reversed IL-6-mediated EMT changes (Fig. 1E-F).

IL-6 promotes EMT changes in head and neck squamous cells and immortalized oral epithelial cells

To further examine the role of IL-6 in mediating EMT changes in head and neck squamous carcinoma cells, we cultured CAL27 cells for 3 days in the presence or absence of IL-6. IL-6 treatment of CAL27 cells markedly increased vimentin expression (62%) while at the same time decreasing the expression of E-cadherin (61%) (Fig. 2A-B). IL-6 treatment also significantly increased snail expression (51%, Fig. 2B). Similarly, IL-6 treatment of IOE cells significantly upregulated the expression of vimentin (95%) and snail (64%) while reducing the E-cadherin expression (Fig 2C-D). Next, we examined the IL-6 mediated signaling cascade in CAL27 cells by treating these cells with IL-6 for different time points (0-240 min). IL-6 predominantly activated JAK/STAT3, MAPK, and PI3K/Akt signaling pathways. JAK1/2 activation peaked 10 minutes post IL-6 treatment, whereas STAT3 (pY705 and pS727), ERK1/2 and AKT activation peaked 30 minutes post IL-6 treatment (Fig. 2E).

IL-6 overexpression in head and neck squamous cells and normal oral epithelial cells induces EMT changes and enhances tumor cell scattering and motility

To further understand the role of IL-6 in the EMT process, we stably overexpressed IL-6 in CAL27 cells (CAL27-IL-6) and IOE cells (IOE-IL-6) using a retroviral vector. IL-6 expression levels in CAL27 and IOE cells were verified by ELISA and Western blotting. CAL27-IL-6 produced more than 8 fold IL-6 as compared to CAL27 cells transduced by vector alone (CAL27-VC) (Supplementary Fig. 1A). CAL27-IL-6 cells also showed significantly higher STAT3 phosphorylation (pY705) as compared to CAL27-VC cells (Supplementary Fig. 1B) thereby suggesting that IL-6 produced by CAL27-IL-6 cells is biologically active. Interestingly, overexpression of IL-6 in CAL27 was also associated with scattering effect (Fig. 3Aa), a hallmark of EMT process. In addition, CAL27-IL-6 cells showed significantly higher vimentin (54%) and snail expression (51%) whereas E-cadherin expression was significantly reduced (53%, Fig. 3B). Furthermore, IL-6 overexpression in CAL27 cells significantly enhanced tumor cell migration (Fig. 3C).

IOE cells produced very low baseline levels of IL-6 and overexpression of IL-6 markedly increased IL-6 production (>30 fold) as compared to IOE-VC (Supplementary Fig. 2A). Similar to CAL27, IL-6 overexpression induced scattering effect in IOE cells (Fig. 3Da) and significantly increased vimentin (63%) and snail expression (70%) while downregulating E-cadherin levels (77%, Fig. 3E). In addition, IL-6 overexpression significantly enhanced IOE cell motility (Fig. 3F).

IL-6 mediates EMT via the activation of JAK-STAT3-snail pathway

Our signaling experiments showed that IL-6 activates 3 distinct signaling pathways: JAK/STAT3, MAPK, and PI3K/Akt (Fig. 2E). To examine which of these pathways is used by IL-6 to mediate its EMT-related effects, we knocked down Akt, STAT3, and ERK1/2 in CAL27 and IOE cells by siRNA treatment. Down regulation of Akt, STAT3, and ERK1/2 protein levels were verified by Western blotting (Fig. 4A-B). Knocking down of STAT3 in CAL27-IL-6 (Fig. 4A and C) and IOE-IL-6 cells (Fig. 4B and D) significantly reversed IL-6 mediated upregulation of vimentin (52% and 48% in CAL27-IL-6 and IOE-IL-6 respectively) and snail protein levels (58% and 64%), whereas Akt and ERK knock down did not significantly alter vimentin and snail levels. Similarly, STAT3 knock down significantly reversed IL-6-mediated loss of E-cadherin expression (67% and 45%), while Akt knock down had no effect and ERK knock down had only partial effect on E-cadherin expression (Fig. 4).

STAT3 knock-down significantly reverses IL-6 mediated CAL27 and IOE motility by inhibiting FAK activation

We further examined the role of different signaling pathways in IL-6 mediated tumor cell and IOE cell migration. As observed with EMT changes, knocking down of STAT3 significantly decreased CAL27-IL-6 and IOE-IL-6 cell migration (Fig. 5A and C). Knocking down of ERK had partial effect while Akt knock down had the least effect on CAL27 and IOE cell migration (Fig. 5A and C). To further understand how STAT3 knock down may be inhibiting CAL27 and IOE motility, we examined the activation profile of focal adhesion kinase (FAK), a key player in cell migration. Interestingly, STAT3 knockdown in CAL27-IL-6 and IOE-IL-6 cells markedly reduced FAK phosphorylation whereas Akt and ERK knockdown had no effect on FAK activation (Fig. 5B and D).

IL-6 promotes tumor growth, EMT and tumor metastasis, in vivo

To further investigate the role of IL-6 in EMT process and tumor metastasis, we used a SCID mouse xenograft model. CAL27-IL-6 tumors showed a moderate but significant increase in tumor growth as compared to CAL27-VC cells (Fig. 6A). CAL27-IL-6 tumor cells exhibited fibroblastic morphology (spindle shape) in contrast to cobblestone morphology of CAL27-VC cells (Fig. 6B). In addition, CAL27-IL-6 tumors showed markedly elevated levels of activated STAT3 (pY705, Fig. 6C) and snail proteins (Fig. 6D) that were predominantly localized in the nucleus. CAL27-IL-6 tumors also showed decreased E-cadherin and increased vimentin expression (Fig. 6E-F).

Lymph nodes from animals carrying CAL27-VC tumors were negative for metastatic disease whereas 60% of lymph nodes from animals carrying CAL27-IL-6 tumors were positive for metastatic disease (Fig. 7A-C). Similarly, lungs from animals with CAL27-IL-6 tumors showed marked increase in metastatic nodes (Fig. 7D-E).