Schwann Cells Augment Cell Spreading and Metastasis of Lung Cancer

Lung cancer is the leading cause of cancer-related mortality worldwide (1), with almost two million new cases annually, constituting nearly 13% of all newly diagnosed cancer patients per annum (2). Furthermore, the incidence of lung cancer has incessantly risen during the last decade in most countries.

Metastases are responsible for the cancer-related deaths of most patients with cancer and high lung cancer–related mortality is also due to the tendency of the malignant cells to spread to the lymph nodes and distant organs (3). More than 40% of the newly diagnosed patients with lung cancer have distant metastases, and the metastatic foci can be detected in up to 60% of patients at the time of the treatment initiation (4). Autopsy studies have revealed over 90% prevalence of lung cancer metastases to the lymph nodes, brain, bone, liver, and the adrenal glands (5); up to 50% of patients with advanced lung cancer develop brain metastases during the course of their disease (6).

Although the mechanisms of positive tropism of lung cancer cells for the nervous system is not understood, recent studies described how the malignant cells survive in the brain (7). Extravasation of circulating tumor cells into the brain parenchyma triggers the response from the stromal elements, such as pericytes, reactive glia, neural progenitor cells, neurons and oligodendrocytes. The interaction of activated glial cells, including astrocytes and microglia, with the malignant cells is reminiscent of the response to the brain injury (7). Interestingly, analysis of the gene expression data showed that the brain microenvironment prompts a broad reprogramming of metastasized cancerous cells, resulting in their gain of the neuronal cell characteristics and mimicking neurogenesis during development (8). However, the role of the peripheral nervous system (PNS) in the lung cancer progression and metastasis remains to be elucidated.

The lungs are densely innervated by the PNS, which control the airways and lung homeostasis (9). Nevertheless, the clinicopathologic significance of the innervation of lung cancers has not been investigated, although the neurogenesis (tumor neoinnervation), its putative regulatory mechanisms, and potential positive association with poor clinical outcomes have been reported in pancreatic, prostate, gastric, colorectal, and breast cancers (10).

Because of the abundant lung innervation, the spread of cancer cells along the nerves can be expected. Perineural invasion, that is, cancer cell dissemination in and along nerve bundles beyond the extent of any local invasion, is found in the head and neck, pancreatic, stomach, and colon cancers (11). The presence of cancer cells in the perineurium is mostly associated with poor prognosis and high recurrence in pancreatic, colorectal, and gastric cancers, but not in invasive breast carcinoma (12). Highly contradictory results have been reported on the “inherent positive neuronal tropism” of lung cancer and the prognostic value of the perineural invasion in the lungs (13, 14). The involvement of the PNS in the formation of distant lung cancer metastases has not been studied.

Malignant cells not only grow near the nerve fibers but also respond to signals provided by the PNS through an accelerated proliferation, dissemination, and survival (15, 16). For instance, we reported that the purified murine DRG neurons can be activated by melanoma cells and, in turn, augment melanoma growth in vivo (17). Understanding that the in vivo microenvironment of peripheral nerves is created and maintained by the nerve ensheathing Schwann cells (SC), the principal glia of the PNS (18), we provided theoretical evidence that SC may aid tumor growth by promoting tumor-favorable conditions through their interaction with malignant cells and the facilitation of the immunosuppressive microenvironment (19). Although the main function of SC is to maintain the integrity of axons, SCs have been shown to increase the integrin-dependent tumor invasion on laminin in the pancreatic and the prostate cancer microenvironment (20). SCs may also enhance the invasiveness of the salivary adenoid cystic carcinoma cells (21). At the perineural invasion sites of pancreatic cancer, SCs may contribute to the malignant cell colonization of the nerves by activating the mesenchymal-to-epithelial transition (MET) and reducing cell motility (22). It is critical to elucidate the influence of the tumor microenvironment on SCs and peripheral nerves as well as the mechanisms of the protumor and prometastatic activity of SCs in cancer. Although SCs have been identified in human lung cancer tissue (23), how SCs function in the microenvironment of lung cancer is still unknown.

Here, we describe how SCs augment the migration and invasion of lung cancer cells. We determined that SCs, reactivated by lung cancer, produce CXCL5 and CXCL1, which, via the CXCR2 and PI3K/Akt/GSK-3β signaling, induce the expression of epithelial-to-mesenchymal transition (EMT) regulators Snail and Twist in malignant cells. The Snail and Twist-dependent EMT of lung cancer cells increases their migration and invasiveness in vitro and their ability to form metastases in vivo. Our results demonstrate that the role of SCs in cancer progression is not limited to the perineural invasion, and likely includes other, yet unidentified, mechanisms by which SCs affect tumor growth and metastasis. A better understanding of the tumor–neuroglia interaction is essential for the development of novel cancer therapy targeting different elements of the tumor microenvironment.

C57BL/6 mice (7- to 8-week-old) from Taconic were housed in a pathogen-free facility under controlled temperature, humidity, and 12-hour light/dark cycle with a commercial rodent diet and water available ad libitum. All studies were conducted in accordance with the NIH guidelines for the Care and Use of Laboratory Animals and approved by the Institutional Animal Care and Use Committee of the University of Pittsburgh (Pittsburgh, PA).

Murine Lewis lung carcinoma (LLC) and lung squamous cell carcinoma KLN-205 cell lines (ATCC; authenticated, Mycoplasma tested, low passage, contaminant free) were maintained in a complete RPMI1640 medium (GIBCO BRL) supplemented with nonessential amino acids, 10% heat-inactivated FCS (Gemini Bio-Products), 2 mmol/L glutamine, 100 IU/mL penicillin, and 100 μg/mL streptomycin (Invitrogen Life Technologies). For the preparation of tumor-conditioned medium, subconfluent cultures were harvested, washed, and 1 × 10⁶ cells were cultured in 20 mL of complete RPMI1640 medium with 10% FCS for 24 hours. Then, the medium was replaced with medium supplemented with 2% FCS. After 48 hours, conditioned medium was collected and centrifuged at 700 × g at room temperature for 5 minutes to pellet the cells. The supernatant was collected and centrifuged at 2,000 × g at 4°C for 10 minutes to remove cell debris.

Tumor cells were also treated with the modified SC-conditioned medium, control medium, CXCR2 small-molecule inhibitor SB265610 (1 μmol/L, 24 hours, Tocris) and PI3K inhibitor LY294002 (10 μmol/L, 24 hours, Cayman Chemical). Appropriate concentration of the carrier, EtOH or DMSO (<0.1% v/v), served as controls.

Primary glial cells (SCs) were prepared from the sciatic nerve of adult mice according to established protocols (24) with some modifications. The distal whole sciatic nerves were dissected under the stereoscopic microscope (Leica) in sterile conditions. After 5-day incubation at 37°C in SC medium (ScienCell Research Laboratories), the sciatic nerves were dissociated using 1 mg/mL NB4G collagenase (Sigma) in 3 mL HBSS (37°C, 45 minutes). Then, the sciatic nerves were stripped free of epineurium, cut into 1-mm segments, and placed as explants into tissue culture dishes containing collagenase NB4G (37°C, 30 minutes). After centrifugation (400 × g, 5 minutes), the supernatant was discarded, the pellet was resuspended in 1 mL SC medium, and the nerves were triturated until they were completely dissociated. The cell suspension was diluted in SC medium supplemented with 2 μmol/L forskolin (Sigma) and 50 ng/mL heregulin-β1 (Sigma) and cultured in Poly-d-lysine precoated flasks for 4 days following by an additional 6–10-day culture in the medium without forskolin and heregulin. During this time, SCs were twice separated from fibroblasts using 300 μg/mL collagenase NB4G solution in 6 mL HBSS/75 cm² flask (37°C, 30 minutes). Prior to the experimentation, SC culture purity was established at 98%–99% based on the immunofluorescent cytochemistry and flow cytometry phenotyping (BD FACSAria) for SC-specific markers GFAP, p75^NGF, and S100B.

Cell viability/proliferation was determined by the MTT assay as described previously (25). Briefly, LLC cells were plated in 24-well plates (50 × 10³ cells/mL/well) in a complete medium with or without SC medium or LLC-conditioned SC medium (cell-free supernatants; 10%–15% v/v) and cultured overnight. MTT reagent (1.0 mg/mL in PBS) was added to each well for 4 hours before lysing with a detergent (500 μL DMSO, dimethyl sulfoxide). Mixed solution was transferred to 96-well plate in triplicates and absorbance was measured on a plate reader (Tecan GENios) at 570 nm. All the experiments were repeated at least twice.

The spreading and migration capabilities of LLC and KLN-205 cells were assessed by measure the expansion of cells on surfaces. Tumor cells were seeded on 10-cm tissue culture dishes (∼0.5 × 10⁶ cells/mL) and cultured for 24 hours in complete medium containing 10% FBS to approximately 70%–80% confluence. Three parallel linear scratches were generated in the monolayer with a sterile 1-mL plastic pipette tip. After creating the gap (time point zero), tumor cells were cultured in complete medium (containing 2% FBS) with either 10% (v/v) SC conditioned medium or SC control medium. For coincubation experiments, LLC and KLN-205 cells at the culture dish edges were removed by the scratcher and replaced with either empty (control) or SC-containing coverslips. After 24 hours of incubation, tumor cells were fixed with 4% paraformaldehyde for 15 minutes. Three representative images from each scratched area for each condition were obtained to estimate the relative migration of the cells. The data were analyzed using ImageJ software (NIH, Bethesda, MD). The characteristic measurement was the wound closure speed during the linear growth phase of cells, quantitatively evaluated as the gap distance or the percentage of cell-covered area. The centerpiece approximation established alterations of the cell-covered area (in %) per time unit. The experiments were performed in duplicates and repeated at least twice.

The effect of SCs on the invasiveness of LLC and KLN-205 cells was evaluated with Transwell Migration Matrigel assay. Polycarbonate membranes (8-μm pore size, Corning) were precoated with growth factor–reduced Matrigel (1.0 mg/mL, BD Biosciences): 30–50 μL of Matrigel was added to a 24-well transwell insert and solidified in 37°C for 30 minutes to form a thin gel layer. Tumor cells (10,000/insert) resuspended in 200 μL complete RPMI1640 medium (1% FCS) were added to the top chamber; 750 μL of SC medium (1% FCS), LLC-conditioned SC medium or SCs (1–2 × 10⁵ cells) were added to the bottom chamber. After 18-hour incubation, noninvading cells were removed from the top surface of the membrane by “scrubbing” with cotton swabs. Then insert was removed from the chamber and gently submerged in PBS several times to remove unattached cells. Cells that penetrated through the Matrigel to the underside surfaces of the membranes were fixed and stained with Diff-Quick kit. A few drops of Permont were added onto the membrane on a glass. Migrated cells were counted on a grid in 10 high-power fields (HPF). Each assay was carried out in duplicate and all experiments were repeated at least three times.

Cells were lysed in NE-PER Nuclear and Cytoplasmic Extraction Reagents (Cytoplasmic Extraction Reagent I and II, and Nuclear Extraction Reagent, Thermo Fisher Scientific) with a protease inhibitor cocktail (Roche), separated on 4%–12% SDS-PAGE, transferred to polyvinylidene difluoride membrane (Invitrogen) and analyzed with antibodies recognizing Snail (1:1,000, Cell Signaling Technology), Twist (1:1,000, Abcam), phospho-AKT (S473, 1:2,000, Cell Signaling Technology), AKT (Pan, 1:1,000, Cell Signaling Technology), phospho-GSK-3β (S9, 1:1,000, Cell Signaling Technology), GSK-3β (27C10, 1:1,000, Cell Signaling Technology), phospho-MAPK (T202/Y204, 1:1,000, Cell Signaling Technology), and MAPK (3A7, 1:1,000, Cell Signaling Technology). For quantitation, gels were scanned and analyzed by ScanIt or ImageJ software. The bands were quantified by pixel density using β-actin as a housekeeping control. The results are presented as relative protein expression calculated for each experimental group versus corresponding controls.

Total RNA from LLC and KLN-205 cells were prepared using the RNeasy kit (Qiagen). For reverse transcription, 1 mg of total RNA was primed with oligoDT (Roche Applied Science) for 10 minutes at 65°C prior to the generation of cDNA via the addition of a master mix of reverse transcriptase 200 units MMLV-RT (Gibco), 1 mmol/L dNTP (Promega), and 10 mmol/L dithiothreitol per reaction for 1 hour at 37°C. The cDNA was then utilized for PCR using standard PCR procedures with the specific primers for Snail and Twist (IDT Integrated DNA Technologies). RPLP0 transcript was used as an internal control. The PCR products were analyzed in 1.2% agarose gel containing ethidium bromide. A DNA ladder II (GeneChoice, Inc.) was used as a size marker. Bands were visualized under UV light (Cell Biosciences), images were recorded, scanned, and densitometry was used for image analysis (ScanIT) utilizing RPLP0 as a housekeeping control. The results are presented as relative mRNA expression calculated for each experimental group versus corresponding controls.

Inhibition of Twist and Snail expression in LLC and KLN-205 cells was done as described previously (26) with some modifications. Target-specific siRNA gene silencers, consisting pools of three target specific 19–25 nucleotide-long double stranded RNA molecules, as well as control siRNA (sc37007) were prepared by Santa Cruz Biotechnology. Tumor cells were cultured in 6-well plates in antibiotic- and serum-free medium to reach 30%–40% confluence. siRNA solutions (50 pmol siRNA/transfection) were added into each well for 7 hours followed by the addition of fresh medium (2 mL), and 24 hours later cells were used for experiments. Dilution buffer, siRNA transfection reagent, siRNA transfection medium, and control siRNAs were from Santa Cruz Biotechnology. For independent verification of target gene silencing results, individual siRNA duplex components were also used. The transfection efficiency was assessed by the Western blot analysis. The siRNA against mouse CXCL5 was from Thermo Fisher Scientific. Transfection efficacy in SCs was verified by the ELISA.

The lungs were harvested from mice and fixed in 2% paraformaldehyde following 24-hour treatment in 30% sucrose. Sections (50 μm) were incubated overnight at room temperature with primary antibody p75^NTR (1:100, Abcam) diluted in 0.1% Triton X-100 and 3% serum. Next, fluorescent secondary antibodies (Donkey anti-chicken IgY polyclonal, FITC, Jackson ImmunoResearch #703-096-155) were centrifuged (10,000 × g, 10 minutes), diluted 1:200 in 0.1% Triton X-100/3% serum and applied to slides. Sections were incubated (room temperature, 4–6 hours), rinsed in PBS, and wet mounted with VECTASHIELD HardSet Antifade Mounting Medium (Vector Labs). Images (20×) of stained sections were taken using Nikon A1 confocal microscope.

The human lung cancer sections were obtained from two patients with lung adenocarcinoma (Shanghai Chest Hospital Tissue Bank). Informed written and signed consent was obtained from patients, all studies were conducted in accordance with recognized ethical guidelines (Declaration of Helsinki), and were IRB approved. Tissue specimens were fixed in 2% paraformaldehyde. All sections were permeabilized and blocked in 3% horse serum, 0.1% Triton X-100/PBS for 1 hour. Sections were incubated overnight at room temperature with primary antibody GFAP (Millipore, 1:100) and S100B (Abcam,1:100) diluted in 0.1% horse serum/PBS. Secondary staining was performed using appropriate fluorescent antibodies: Alexa Fluor 488 (1:500, Invitrogen) and Alexa Fluor 568 (1:500, Invitrogen). Samples were mounted in DAPI containing Vectashield mounting medium. Images (20×) of stained sections were taken using Leica TCS SP8 STED 3X confocal microscope.

For evaluation of CXCL1 and CXCL5 secretion, SCs, 3LL, or KLN-205 cells (insert) were cocultured with SCs (bottom chamber) in 6-well plate transwell system with 0.4-μm pore size inserts (Corning). After 48 hours, cells were harvested, washed, plated in fresh medium, and conditioned medium was collected at different time points to assess CXCL5 and CXCL1 concentrations by ELISA (R&D Systems and LifeSpan BioSciences, Inc, respectively).

GFP-expressing LLC cells (2 × 10⁶/100 μL PBS) were inoculated subcutaneously in a right flank of mice. Prior to the inoculation, LLC-GFP cells were pretreated with the (i) control SC medium, (ii) SC-conditioned medium (10% v/v), and (iii) LLC-preactivated SC-conditioned medium (10% v/v) for 48 hours. For the preparation of SC-conditioned medium, cultured SCs (75% confluence) were washed, cultured in complete RPMI1640 medium (2% FCS) for 48 hours, and cell-free supernatant were collected and aliquoted. For the preparation of LLC-preactivated SC-conditioned medium, SCs were first cultured in LLC-conditioned medium (2% FCS; 10% v/v, 48 hours), washed, and then cultured again for the generation of the conditioned medium as described above.

Animals were sacrificed within 30 days after tumor cell inoculation, and regional lymph nodes were harvested. Single-cell suspensions from 2–3 lymph nodes were prepared and the presence of GFP-positive cells was determined by flow cytometry. The lymph nodes from tumor-free mice served as controls. All groups included 6–7 mice and all experiments were repeated twice.

For a single comparison of two groups, the Student t test was used after the evaluation of normality. If data distribution was not normal, a Mann–Whitney rank-sum test was performed. For the comparison of multiple groups, ANOVA was applied. SigmaStat Software was used for data analysis (SyStat Software, Inc.). For all statistical analyses, P < 0.05 was considered significant. All experiments were repeated at least two times. Data are presented as the mean ± SEM.

Since the initial description of SCs in the lungs of cats (27) and further demonstrations of a close association between SCs and axons of the pulmonary nerves in various vertebrates (28, 29), there are still no reports describing the participation of PNS neuroglial cells in lung cancer development or progression. Early descriptions of lung SCs were based on the electron microscopy, while the more recent studies utilized IHC with the glia-specific markers S100B (or S100β) and GFAP (29). Using another SC marker, the low-affinity NGF receptor (also known as p75 neurotrophin receptor, p75^NTR), we have developed an immunofluorescence staining to demonstrate the presence of SCs in the murine lungs. Figure 1A (top) shows elongated cells that are seen most prominently around the medium to large airways, as expected for glia accompanying the pulmonary nerves (30). In addition, we revealed the presence of GFAP⁺S100B⁺ SCs in human lung adenocarcinoma tissue (Fig. 1A, bottom).

Using the scratch assay, we demonstrated that SCs significantly accelerate the in vitro rate of migration of tumor cells: the addition of the SC-conditioned medium to LLC cell monolayers decreased the gap area of cell-covered field from 85.7% ± 6.7% to 65.3% ± 7.3% (P < 0.05, n = 4; Fig. 1B). An even stronger effect of SCs on malignant cell motility was observed with KLN-205 cells: 66.1% ± 9.7% versus 27.7% ± 4.5% (P < 0.05, n = 3; Fig. 1C). Next, using the Matrigel transwell migration assay, we demonstrated that both the SCs and SC-conditioned medium attract LLC and KLN-205 cells 3- to 4-fold stronger than the control medium (P < 0.05, n = 4; Fig. 1D). Importantly, SCs did not directly affect LLC and KLN-205 cell proliferation in vitro (Fig. 1E). Taken together, our results suggest that in vitro SCs promote lung cancer cell migration and invasion and attract malignant cells without affecting their proliferation rate.

We tested whether SCs affect the expression of the epithelial-to-mesenchymal transition regulators (Snail, Slug, Twist, E-cadherin, ZEB, and N-cadherin; ref. 31) in lung cancer cells, as EMT accompanies tumor cell capacity to invade (31). The mRNA and protein levels of Snail and Twist in both LLC and KLN-205 cells were significantly increased after the exposure to the SC-conditioned medium (P < 0.05, Fig. 2A and B). Importantly, pretreatment of SCs with tumor-conditioned medium markedly increased SCs ability to upregulate Snail and Twist expressions in tumor cells (Fig. 2A and B).

To verify the role of Snail and Twist in SC-dependent enhancement of the tumor cell–migratory potential, we blocked the expression of these EMT regulators in tumor cells using specific siRNA (Fig. 2C). The inhibition of Snail and Twist in both LLC and KLN-205 cells abrogated the increase of tumor cell migration induced by SCs (P < 0.05, Fig. 2D and E). Thus, our results suggest that tumor-activated SCs may augment the migratory capacity of lung cancer cells at least in part by stimulating cancer cells’ expression of the EMT regulators Snail and Twist.

Next, we identified SC factors that are upregulated by tumor cells and are involved in increasing the expression of Snail and Twist by malignant cells. We measured the production of chemokines by lung cancer cell–stimulated SCs using Mouse Cytokine Array (Proteome Profiler). Among several increased chemokines (including CCL3, 4, 13 and CXCL2, 13), CXCL1 and CXCL5 were selected for further analysis because of their known ability to control expression of EMT regulators Twist and Snail (32, 33). Figure 3A and B show that tumor-activated SCs secrete significantly higher levels of CXCL1 and CXCL5 compared with control SCs (P < 0.05). In contrast, coculturing LLC and KLN-205 cells with SCs does not alter tumor cells’ ability to produce these chemokines (Fig. 3C and D). Of note, both tumor cell lines express high levels of CXCL1, while CXCL5 is produced by KLN-205 but not LLC cells. Taken together, these results demonstrate that lung cancer cell lines increase the production of several chemokines by SCs, which may be responsible for the observed SC-dependent increase in malignant cell invasiveness.

We next studied the effect of SC-derived chemokines on lung cancer cell invasiveness. Both recombinant CXCL5 and CXCL1 increase the expression of Twist (up to 3-fold, P < 0.05) and Snail (up to 2-fold, P < 0.05) proteins in LLC cells in a dose- and time-dependent manner (Fig. 4A and B). In addition, CXCL5 and CXCL1 significantly increase the transwell migration of LLC cells in a dose-dependent manner (Fig. 4C).

To determine whether SC-derived chemokines have a similar effect on lung cancer cells as the recombinant CXCL1 and CXCL5, we either inhibited the chemokine expression in SCs by specific siRNAs or blocked the chemokine receptors with an antagonist. Tumor-activated SCs produce both CXCL5 and CXCL1 (Fig. 3A and B). LLC and KLN-205 cells produce CXCL1 while only KLN-205 cells produce CXCL5 (Fig. 3C and D). Therefore, we concentrated on SC-derived CXCL5 signaling to LLC cells.

Blocking the production of CXCL5 by naïve and LLC-activated SCs with CXCL5-specific siRNA (P < 0.05; Fig. 4D) abrogates their ability to upregulate the expression of Twist and Snail in LLC cells (P < 0.05, Fig. 4E and F). Importantly, inhibiting SC expression of CXCL5 attenuates their ability to stimulate LLC cell migration (P < 0.05, Fig. 4G). The above results were confirmed using SB265610 to selectively inhibit CXCR2, the CXCL5 receptor, on LLC cells (Fig. 4E). Figure 4F and G show that blocking CXCR2 abrogates (P < 0.05) SC-dependent upregulation of Twist and Snail by LLC cells as well as their transwell migration capacity. Taken together, these results suggest that SCs in the tumor microenvironment can be activated to produce CXCL5, which causes the malignant cells to express EMT regulators Twist and Snail and stimulates cancer cell invasion in vitro.

The involvement of either PI3K-Akt or MAPK (Erk1/2) pathways in CXCL5/CXCR2 activation and cell motility have recently been reported in liver, colon, and breast cancer cells but not in lung cancer (32, 34). We found that SC-treated lung cancer cells exhibit an increased phosphorylation of serine/threonine protein kinase Akt (protein kinase B) and glycogen synthase kinase 3β (GSK-3β), but not MAPK (Fig. 5A, P < 0.05). Blocking CXCL5 expression in SCs or inhibiting CXCR2 with SB265610 on malignant cells significantly decreases the phosphorylation of Akt and GSK-3β in SC-treated LLC cells. Akt and GSK-3β are known downstream effectors of PI3K; therefore, we confirmed that PI3K inhibitor LY294002 blocks the activation of Akt/GSK-3β signaling in SC-conditioned LLC cells (Fig. 5B, P < 0.05).

We also confirmed that PI3K/Akt/GSK-3β signaling is involved in the expression of EMT regulators and the motility of LLC cells. Figure 5C shows that the inhibition of PI3K in LLC cells prevents SC-dependent increase in the expression of Snail and Twist (P < 0.05). PI3K inhibitor LY294002 also blocks SC-induced transwell migration capacity of LLC cells (Fig. 5D, P < 0.05). Taken together, our results demonstrate that SC-derived CXCL5 stimulates the EMT and the motility of LLC cells via the CXCR2/PI3K/Akt/GSK-3β/Snail-Twist axis.

As SCs increase LLC invasiveness in vitro, we tested whether SCs could alter the formation of the metastases by lung cancer cells in vivo. Figure 6A shows that SC-treated LLC cells migrate significantly more to the regional lymph nodes when subcutaneously injected into syngeneic mice (P < 0.05). In vitro cancer cell–preconditioned SCs stimulate the metastasis formation by LLC cells significantly more compared with naïve SCs (P < 0.05, Fig. 6A). Importantly, when the CXCL5 production in SCs was blocked with siRNA as above, the ability of both naïve and cancer cell–preactivated SCs to stimulate the formation of metastases by SC-treated LLC cells was abrogated (P < 0.05, Fig. 6B). Taken together, our results show for the first time that SCs within the tumor microenvironment can stimulate lung cancer invasion and metastasis through their release of CXCL5 and the activation of CXCR2/PI3K/Akt/GSK-3β/Snail-Twist signaling in malignant cells.

The lungs are densely innervated by the PNS (9). Recent studies demonstrated that the nerve fibers are found in and around tumors and the tumor denervation may alter its growth, suggesting the involvement of the PNS neurons in carcinogenesis and tumor progression (35, 36). Furthermore, intratumor nerve density may be a marker of an aggressive behavior and a poor clinical prognosis for several types of cancer (15, 37).

Neural regulation of cancer involves a complex network of neuromediators, such as catecholamines and acetylcholine, which can affect proliferation and migration of malignant cells (38, 39). Neurotransmitters can enhance the perineural pathway for the spread of malignant cells (40) and the presence of perineural invasion is an independent prognostic factor in some cancers (41). The nervous system also modulates angiogenesis (42), the tumor microenvironment (17), immune reactions (43), and the inflammatory pathways, influencing malignant cell growth, invasion, and metastasis formation (44).

Despite the growing evidence for the role of neurotransmitters and neuropeptides in the tumor milieu, relatively little is known about the involvement of the PNS neuroglia in cancer progression. SCs, the principal glial cells of the PNS involved in nerve myelination, survival, and regeneration (18), may enable cancer progression in a contact-dependent manner, mediated by the expression of the neural cell adhesion molecule 1 (NCAM1; ref. 45). SCs can direct malignant cells to migrate toward the nerves, promoting perineural invasion (46). In the pancreatic and prostate cancers, SCs have been shown to increase integrin-dependent tumor invasion along the axons (20). Furthermore, pancreatic ductal adenocarcinoma cells secrete CXCL12, attracting SCs and inducing perineural invasion in early carcinogenesis (47). Although SCs have been identified in the human lung cancer tissue (23), nothing is known about their function in lung cancer.

Here, we report that SCs directly stimulate lung cancer cell invasiveness and migration with secreted chemokines. SC-dependent activation of tumor cells is associated with the increase in the expression of the EMT transcription factors Snail and Twist. Although EMT is well characterized (31), the involvement of SCs in the EMT process has never been demonstrated. Using rat SCs and human salivary adenoid cystic carcinoma (SACC) cell line cocultures, Shan and colleagues have reported that SCs, probably via brain-derived neurotrophic factor, repress the expression of E-cadherin, used as a biomarker of the EMT, in SACC cells (21). However, the utilization of a xenogeneic system, an incomplete blockade in their functional studies, an undetermined role of the EMT regulators and an unknown signal transduction limit the significance of their reported observations.

EMT inducers such as TGFβ, VEGF, FGFs, TNFα, HGF, IGF1, PDGF, and EGF have been identified and investigated (31), but SCs have never been implicated as a source of soluble factors, which may directly control the EMT in cancer. Fernando and colleagues demonstrated that the transition of human tumor cells from an epithelial-to-mesenchymal-like phenotype is associated with the autocrine signaling of several chemokines, such as CXCL8, which promote tumor growth, motility, and invasion of various types of carcinomas (48). Tumor-derived CXCL5 promotes tumor cell spreading by inducing the EMT in hepatocellular carcinoma and colorectal cancer (32, 49). Chemokines CXCL8 and CXCL5 from nonmalignant cells could also function in a paracrine mode to induce the epithelial cancer cells to undergo the EMT (50, 51). CXCL5 derived from breast cancer–associated osteoblasts can also induce the EMT, migration, and the invasion of the malignant cells (52). We report here that SC-derived CXCL5 may induce the EMT in lung cancer: by binding to its receptor CXCR2 on lung cancer cells, CXCL5 produced by SCs activates the PI3K/Akt/GSK-3β signaling and causes an increase in the expression of the EMT regulators Snail and Twist associated with an increased cell migration.

CXCL5-CXCR2 and ERK1/2 pathways play critical roles in the regulation of liver cancer cell migration and invasion (34). The effect of the osteoblast-derived CXCL5 on the EMT of breast cancer cells is associated with an increase in Raf/MEK/ERK activation, mitogen- and stress-activated protein kinase 1 (MSK1), and Elk-1 phosphorylation and Snail upregulation (52). In hepatocellular carcinoma, CXCR2/CXCL5 promotes cell spreading by inducing the EMT through the activation of the PI3K/Akt/GSK-3β/Snail signaling pathway (49). Tumor-derived CXCL5 enhances migration and invasion of colorectal cancer cells by inducing ERK/Elk-1/Snail and AKT/GSK-3β/β-catenin–dependent EMT (32). We show that either preventing PI3K activation or silencing Snail and Twist attenuates CXCL5/CXCR2-dependent lung cancer cell migration and invasion in vitro.

Although several studies proved that CXCL5/CXCR2 axis is oncogenic in various human cancers, no data on CXCL5/CXCR2 signaling exist for lung cancer. Only recently has the clinical significance of CXCL5 in lung cancer been suggested. The serum level of CXCL5 is significantly elevated in patients with non–small cell lung carcinoma (NSCLC) and CXCL5 immunoreactivity is enhanced in lung cancer tissue (53). Furthermore, CXCL5 expression correlates with the histologic grade, tumor size, lymph node metastasis, overall survival, and progression-free survival in patients with NSCLC (53). These results were independently confirmed by the studies evaluating the expression of prognostic markers in early stage NSCLC (54). Eighteen genes encoding primarily chemokines and cytokines were found to be significantly altered, and among all of the altered genes, only CXCL5 was found to statistically significantly correlate with both the overall and disease-free survival (54). Interestingly, although CXCL5 is considered a therapeutic target in liver cancer (55), few CXCL5-targeting drugs have been developed and tested in preclinical studies. Our results suggest that these studies should be expanded to lung cancer models as well. SCs within the tumor microenvironment should also be considered as a new therapeutic target, although additional studies are needed to demonstrate the feasibility of this novel approach.

No potential conflicts of interest were disclosed.

Conception and design: Y. Zhou, G.V. Shurin, H. Zhong, M.R. Shurin

Development of methodology: Y. Zhou, G.V. Shurin, H. Zhong, Y.L. Bunimovich

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): Y. Zhou, H. Zhong, B. Han, Y.L. Bunimovich

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): Y. Zhou, H. Zhong, M.R. Shurin

Writing, review, and/or revision of the manuscript: Y. Zhou, H. Zhong, Y.L. Bunimovich, M.R. Shurin

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): Y. Zhou, H. Zhong, M.R. Shurin

Study supervision: Y. Zhou, G.V. Shurin, H. Zhong, B. Han, M.R. Shurin

This work was supported by LF grant (to M.R. Shurin), National Natural Science Foundation of China (No. 81472642 to H. Zhong), and UPCI Melanoma and SPORE in Skin Cancer Career Enhancement Program Award NIH P50CA121973 (to Y.L. Bunimovich). This project used the University of Pittsburgh Imaging Core that is supported in part by award P01HL114453.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.