Interleukin-6 blockade attenuates lung cancer tissue construction integrated by cancer stem cells

In the present study, we successfully generated lung cancer stem cell (CSC)-like cells by introducing a small set of transcription factors into a lung cancer cell line. In addition to properties that are conventionally referred to as CSC properties, the lung induced CSCs exhibited the ability to form lung cancer-like tissues in vitro with vascular cells and mesenchymal stem cells, which showed structures and immunohistological patterns that were similar to human lung cancer tissues. We named them “lung cancer organoids”. We found that interleukin-6 (IL-6), which was expressed in the lung induced CSCs, facilitates the formation of lung cancer organoids via the conversion of mesenchymal stem cells into alpha-smooth muscle actin (αSMA)-positive cells. Interestingly, the combination of anti-IL-6 antibody and cisplatin could destroy the lung cancer organoids, while cisplatin alone could not. Furthermore, IL-6 mRNA-positive cancer cells were found in clinical lung cancer samples. These results suggest that IL-6 could be a novel therapeutic target in lung cancer.

). To assess the sphere formation ability, which is considered to be a property of cancer stem cells, we cultured these cells on low attachment plates on days 10, 20, and 30 after transduction. The parental A549 cells and EGFP-A549 cells formed less than 3 spheres under this condition. In contrast, the number of spheres formed by the OSK-A549 cells was remarkably increased, especially on day 20 after transduction (Figs 1A, S1B). ii) OSK-A549 cells contained subsets with distinct morphology and trypsin sensitivity.
Dome-shaped colonies appeared only in the OSK-A549 cells from 10 to 15 days after transduction (Fig. 1B). The colonies were picked up and sub-cultured (Fig. 1C). During one week of continuous culture, these colonies grew larger and gave rise to spindle-shaped cells around themselves (Fig. 1D). We succeeded in collecting the colony cells and the spindle-shaped cells separately by using the different sensitivity of the cells to detachment by trypsin: the spindle-shaped cells detached earlier, while the colony cells detached later (Fig. 1E). We termed the colony cells, "OSK-A549-Colony cells" and the spindle-shaped mesenchyme-like cells, "OSK-A549-SN cells". iii) The induced OSK-A549-Colony cells exhibited chemo-resistance and a delayed cell cycle.
Since cancer stem cells are thought to be particularly resistant to anticancer drugs 6,7 , we assessed the viability of the parental A549 cells, the OSK-A549-Colony cells and the OSK-A549-SN cells following exposure to cisplatin. In comparison to the other two cell lines, the OSK-A549-Colony cells were significantly more resistant to cisplatin (Fig. 1F). We next performed cell cycle analyses by flow cytometry based on Ki67 (a proliferation marker) and Hoechst 33342 staining because quiescence is an important mechanism of drug resistance in stem cells [6][7][8] . The proportion of Ki67-negative G0-phase cells in the OSK-A549-Colony cells was significantly increased in comparison to the parental A549 cells (Figs 1G, S1C). A cell cycle assay using redox dye consistently showed the increased proportion of G0/G1-phase cells in the OSK-A549-Colony cells ( Figure S1D). iv) OSK-A549-SN cells recapitulate the epithelial mesenchyme transition (EMT) and enhance the invasion ability.
As cancer stem cells are considered to be responsible for cancer invasion 9,10 , we next investigated the EMT and the changes in the invasion ability of the induced cells. Immunocytochemistry of an epithelial marker (E-cadherin) showed the presence of E-cadherin-negative cells in OSK-A549-SN cells that had originated from OSK-A549-Colony cells; whereas, the parental A549 cells were almost all E-cadherin-positive (Fig. 1H). Flow cytometry to detect the expression of E-cadherin revealed that 38.3% of the OSK-A549-SN cells were E-cadherin-negative ( Figure S1E). A wound healing assay revealed that the migration ability of OSK-A549-SN cells was increased in comparison to the parental A549 cells ( Figure S1F). The double-layered collagen gel hemisphere (DLCGH) method 11,12 also showed that invasion ability of the OSK-A549-SN cells was significantly enhanced ( Figure S1G,H). Moreover, observation with a HoloMonitor M4 (Digital holographic microscopy) revealed that OSK-A549-SN cells had the highest motility and migration ability among the three types of cells ( Figure S2A,B).

v) The sphere formation ability of OSK-A549-Colony cells is enhanced.
We further investigated whether the induced OSK-A549-Colony cells had sphere formation ability. Phase contrast microscopy showed obvious cellular aggregation in the OSK-A549-Colony cells, but not in the parental A549 cells (Fig. 1I). Hematoxylin and eosin (HE) staining revealed that the OSK-A549-Colony cells could form dense spheres with a glandular ductal structure, whereas the parental A549 cells could not (Fig. 1I). Alcian blue-PAS staining revealed the presence of polarized glandular ductal structures with mucin secretion in the OSK-A549-Colony cell-derived spheres ( Figure S1I).

vi) The tumorigenicity of OSK-A549-Colony cells is enhanced.
To examine the tumorigenicity in vivo, we subcutaneously transplanted 1 × 10 6 parental A549 cells or OSK-A549-Colony cells into immunocompromised mice and measured the volume of the tumors that developed. The tumors derived from OSK-A549-colony cells were significantly larger than those derived from parental A549 cells (Fig. 1J).
Based on the above findings described in i) to vi), we considered the OSK-A549-Colony cells to be lung cancer stem-like cells.

OSK-A549-Colony cells could form lung cancer organoids by co-culture with HUVECs and
MSCs. It is reported that liver organ buds can be generated by the co-culture of induced pluripotent stem cell-derived hepatocytes, human umbilical vein endothelial cells (HUVECs) and human mesenchymal stem cells (MSCs) 13 . The cellular interaction is considered to be important for this self-organization 14 . Not only normal tissues but also cancer growth has been thought to depend on dynamic interaction with adjacent stromal cells that compromise a niche 15 . We therefore investigated whether OSK-A549-Colony cells could induce self-organization by interacting with HUVECs and MSCs. We co-cultured the parental A549 or OSK-A549-Colony cells with HUVECs and MSCs on low attachment plates ( Fig. 2A). cells were found around the OSK-A549-Colony cells (indicated as white arrows). (I) Phase contrast microscopy of the spheres (upper panels) and HE staining images (lower panels). Phase contrast microscopy of the OSK-A549-Colony cells showed obvious cellular aggregation, but not the parental A549 cells on the 24-well low attachment plate. (J) A comparison of the tumorigenicity in vivo between the parental A549 and OSK-A549-Colony cells (n = 10, **P < 0.01; repeated measures ANOVA).
Phase contrast microscopy revealed no obvious differences between the parental A549 cell-derived and the OSK-A549-Colony cell-derived cellular aggregates (Fig. 2B). However, HE staining and immunostaining for alpha smooth muscle actin (αSMA) and Cytokeratin 7 (CK7) demonstrated that only the OSK-A549-Colony cells could form cohesive cell nests that were similar to human lung cancer tissues (Fig. 2C), and we called these nests "lung cancer organoids". To visualize the cellular composition of the organoids, we transduced EGFP or mCherry into OSK-A549-colony cells or HUVECs, respectively and predicted the cellular viability by detecting the fluorescence intensity. HUVECs could survive in the OSK-A549-Colony cell-derived organoids but not in the parental A549 cell-derived organoids ( Figure S3A,B), and an increased number of cluster of differentiation 31 (CD31; a marker of endothelial cells)-positive cells were found in the OSK-A549-Colony cell-derived organoids than in the parental A549 cell-derived organoids ( Figure S3C).
In human lung cancer tissue specimens, Ki67-positive cells were more frequently detected on the outer side of the cancer cell clusters than on the inner side ( Figure S3D). We found the peripheral-dominant distribution of Ki67-positive cells in the OSK-A549-Colony-derived organoids, but not in the parental A549-derived organoids ( Figure S3D    The mean fluorescence intensity of GFP in the EGFP-labeled parental A549 cell-derived organoids. The addition of IL-6 significantly strengthened the mean fluorescence intensity (n = 3, **P < 0.01; two-tailed paired t-test). (C) HE and αSMA staining showed that the addition of IL-6 (10 ng/ml) increased the organoid forming ability of A549 + HUVEC + MSC by collaborating with the αSMA-positive cells. (D) The proportion of αSMA-positive cells in the parental A549 derived-organoids was increased by the addition of IL-6 (10 ng/ml) (n = 3, **P < 0.01; two-tailed paired t-test). (E) Phase contrast and fluorescence images of the organoids consisting of EGFP-labeled OSK-A549-colony cells, HUVECs and MSCs. (GFP shows cancer cells). (F) The mean fluorescence intensity of GFP in the EGFP-labeled OSK-A549-Colony cell-derived organoids. The addition of IL-6 antibodies did not attenuate the mean fluorescence intensity (n = 3, two-tailed paired t-test). (G) αSMA and HE staining showed that IL-6 signal blockade attenuated the organoid construction ability of OSK-A549-Colony + HUVEC + MSC, first, we compared the gene expression of the parental A549 cells and the OSK-A549-Colony cells. We found that 575 genes were differentially expressed in OSK-A549-Colony cells to a significant extent (P < 0.05; Fold change, >2.0). Among these genes, we compared the gene expression in the OSK-A549-Colony and OSK-A549-SN cells; the results are shown in a volcano plot (Fig. 3A). It showed that 54 of the genes that were differentially expressed between the OSK-A549-Colony cells and the OSK-A549-SN cells (P < 0.05; Fold change, >2.0; red plots in Fig. 3A). The expression of previously-reported candidate marker genes, such as CD133, CD44 and ABCG2, which are argued to be related to cancer stem cells, and the transduced genes OCT3/4, SOX2 and KLF4 were not significantly different among the parental A549, OSK-A549-Colony and OSK-A549-SN cells ( Figure S4A,B). Although the expression of NANOG, a pluripotency marker, in OSK-A549 colony and OSK-A549-SN cells was significantly higher than in parental A549 cells, the absolute values of NANOG expression were small across all three kinds of cells.
Interleukin-6 (IL-6) was the gene that was most differentially expressed in OSK-A549-Colony cells in comparison to OSK-A549-SN cells (Fig. 3A). We performed a quantitative RT-PCR and confirmed that IL-6 was significantly upregulated in the OSK-A549-Colony cells in comparison to the other two cell types (Fig. 3B). Thus the present study focused on IL-6.
Interestingly, dual staining of IL-6 mRNA and CAM5.2 revealed that IL-6 mRNA was detected in both CAM5.2 positive and negative cells in the OSK-A549-Colony-derived organoids, but not in the parental A549-derived aggregates (Fig. 3C).

IL-6 blockade enhanced the chemosensitivity of OSK-A549-Colony organoids.
We next investigated the effects of IL-6 blockade on the chemosensitivity of the organoids by evaluating the fluorescence (EGFP) intensity of cancer cells and performing a histological examination of the organoids. The parental A549 cell-derived organoids were sensitive to 5 μM cisplatin (CDDP) (Fig. 3D). On the contrary, the OSK-A549-Colony cell-derived organoids were not (Fig. 3D). Notably, the evaluation of the mean fluorescence intensity of the cancer cells (Fig. 3D,E) and a histological examination (Fig. 3D,F) revealed that the OSK-A549-Colony cell-derived organoids were sensitive to the combination of cisplatin and anti-IL-6 antibodies (1 μg/ml).

IL-6 plays an important role in the construction of lung cancer organoids.
We evaluated the effects of IL-6 on the lung cancer organoid construction ability. The addition of IL-6 (10 ng/ml) strengthened the lung cancer organoid construction ability of the parental A549 cells (Fig. 4A,B). αSMA staining showed that the proportion of αSMA-positive cells in the parental A549 cell-derived organoid was significantly increased by the addition of IL-6 ( Fig. 4C,D). In contrast, IL-6 blockade using anti-IL-6 antibodies (1 μg/ml) attenuated the proportion of αSMA-positive cells in the OSK-A549-Colony cell-derived organoids (Fig. 4G,H). Importantly, IL-6 blockade alone could not kill the lung cancer stem-like cells (Fig. 4E,F). HE and αSMA staining showed that the proportion of αSMA-positive cells in the OSK-A549-colony cell-derived organoids was significantly reduced by IL-6 blockade (Fig. 4G,H). Next, we confirmed that the addition of IL-6 could induce αSMA-positive cells from MSC cultures (Fig. 4I). These results suggest that lung cancer stem-like cells enhance the lung cancer organoid construction ability by inducing the conversion of MSCs into αSMA-positive cells.
The identification of IL-6 mRNA-positive lung epithelial cells in the clinical samples. To evaluate whether IL-6-expressing lung cancer cells exist in human lung cancer tissues, we performed in situ hybridization for IL-6 using 10 lung adenocarcinoma tissue specimens from 8 patients of various backgrounds (Table 1) Figure S5D). Moreover, IL-6 mRNA-positive cells were also found in non-cancer lung epithelial cells near a primary lung cancer lesion (Figs 5D, S5E; arrow). In order to exclude the possibility that the IL-6-expressing cells were intraepithelial lymphocytes or alveolar macrophages, we performed double staining of IL-6 mRNA and CAM5.2 in the same tissue sections and confirmed that double-positive cells definitely existed in human lung cancer tissues (Fig. 5E). In our evaluation of the 10 lesions in the 8 cases, IL-6 mRNA positivity seemed to have no obvious correlation with the background of any of the patients or any known driver gene mutations (Table 1).

Discussion
Lung CSCs are considered to be responsible for the poor prognosis of patients with lung cancer due to therapeutic resistance and the possible development of recurrence or metastasis 1,16,17 . Thus, lung CSCs have been the subject of great interest. However, difficulty in obtaining proper research materials has hindered the elucidation of lung CSC biology and the development of novel therapies that target lung CSCs. In order to isolate lung CSCs from clinical specimens, several previous studies 17- 21 have attempted (but failed) to identify specific markers for lung CSCs via immunostaining. In the present study, we successfully generated lung CSC-like-cells by introducing defined factors into a KRAS-mutated lung cancer cell line (A549) and made lung cancer organoids that resembled and that the proportion of αSMA-positive cells in the organoids was decreased. (H) The proportion of αSMApositive cells in the organoids was decreased by the addition of IL-6 antibodies (n = 3, **P < 0.01; two-tailed paired t-test). (I) Immunocytochemistry for αSMA. Seven days treatment of IL-6 (10 ng/ml) could induce αSMA-positive cells from MSCs culture.
Scientific RepoRts | 7: 12317 | DOI:10.1038/s41598-017-12017-y bona fide human lung cancer tissues. Although patient-derived cancer organoid technologies in various types of cancers 22 including lung cancer 23 were recently reported, they do not provide isolated stem cell populations. We considered that the triad consisting of the lung CSC-like cells, the lung cancer non-stem cells differentiated from these cells and the organoids could be useful novel materials for lung CSC research. Indeed, we obtained new knowledge about lung CSCs by utilizing the triad in the current study.
Co-culturing with HUVECs and MSCs led the induced lung CSC-like cells-but not the parental A549 cellsto form organoids that were similar to human lung cancer tissues. Previously, Takebe et al. succeeded in making various organoids, which they termed, "organ buds", by co-culturing the parenchymal cells of various tissues with HUVECs and MSCs 24 . It should be noted that in their report, fetal parenchymal cells could form organoids via self-organization, whereas adult parenchymal cells could not. These findings indicate that the type of parenchymal cell is a determinant of self-organization. Our current data suggest that only cancer stem cells, at least in lung cancer, can assemble parenchymal and non-parenchymal cells into cancer tissues. Sasai classified the basic processes of tissue self-organization into three categories: self-assembly, self-patterning and self-morphogenesis 14 .
As far as we know, the present study is the first to demonstrate that stem cells proactively take the initiative in self-assembly.
We then revealed that IL-6, which was differentially expressed between the induced lung CSC-like cells and their derivatives, plays an important role in the formation of lung cancer organoids via the conversion of MSCs into αSMA-positive cells, which have been considered to exert large contractile forces in cancer tissues, thereby promoting cancer growth 25 . We also found that IL-6 mRNA-positive cancer cells actually exist in clinical lung cancer tissue samples. IL-6 is a cytokine with a wide range of physiological functions. It has been reported to be associated with chemoresistance 26-31 , progression 26,31-33 , the expansion of cancer stem cell-like cells and a poor prognosis in various types of cancer, including lung cancer 26,28,29,[34][35][36][37] . These previous reports demonstrated the effects of extrinsic IL-6 on cancer cells, and showed that non-cancer cells (such as cancer-associated fibroblasts) in various cancers [38][39][40] , as well as MSC 41 , indeed secreted IL-6. However, in our present experiments, the parental A549 cells could not form lung cancer organoids-even when they were co-cultured with MSCs and HUVECs. This suggests that the production of IL-6 by lung CSCs and the consequent high IL-6 concentration just around the lung CSCs is important for the construction of lung cancer tissue. A few previous reports 30,31,33 that showed the production of IL-6 by cultured cancer cells support our present findings which demonstrated the expression of IL-6 in the induced lung CSCs and a small number of the cancer cells in bona fide human lung cancer tissues.
The current results raise the hope that IL-6 signal blockade and conventional chemotherapy may be used as a combination therapy targeting both lung cancer stem cells and bulk cancer cells. Previously, the inhibition of IL-6 signaling has been reported to attenuate tumorigenicity-especially in KRAS-driven cancers 28,37,42,43 -and IL-6 knockdown in A549 cells was shown to strengthen the chemosensitivity to cisplatin in vivo 29 , and our current data illuminate the potential of targeting IL-6 in the context of lung cancer stem cells. We should be concerned with both the survival of the individual lung cancer cells and with the interaction among the lung cancer stem and non-stem cells and non-cancer cells (such as cancer-associated fibroblasts) that constitute cancer tissues. Although xenografts have been the standard method 6 for evaluating the effects of drugs on cancer tissues, there have been concerns about differences in the microenvironments of humans and recipient animals, both in terms of species differences in non-parenchymal cells and immunodeficiency 6 . The lung cancer organoids that were established in the present study provide complementary evidence to support the efficacy of IL-6-targeted therapy and strengthen the rationale for a clinical trial of the therapy. Moreover, that we found IL-6 mRNA-positive cancer cells in KRAS mutation-negative cases suggests the potential for IL-6-targeted therapy to be applied to various types of lung cancer. A monoclonal IL-6 receptor blocker (tocilizmab) has already been used in clinical practices for rheumatoid arthritis [44][45][46] and Castleman disease 45,47 . Thus, it will be much easier to apply in the clinical treatment of lung cancer patients than a newly discovered drug.
In addition to IL-6, we might be able to uncover other factors that are important for the tissue reconstruction ability of lung cancer stem cells by utilizing our currently established methods. Interestingly, we observed that the HUVECs in the lung cancer organoids survived for longer than the HUVECs in the co-cultures of parental A549, MSCs and HUVECs. This suggests some encouragement from the lung CSC-like cells. A GO term analysis revealed that the genes associated with "blood vessel development" were upregulated in the lung CSC-like cells (Table S1). Further evaluation will be necessary to clarify the functions of these genes in the reconstruction of lung cancer tissue. Multicellular systems like the lung cancer organoids are considered to involve huge numbers of regulatory components; however, it is inferred that there should be robust interactions among a limited number of core control modules that play predominant roles in the critical dynamics of the complex system 48 . A strategy that includes the comparison of the molecular signature between lung CSC-like cells and non-stem cells and phenotype analyses in in vitro lung cancer organoid assays may help to elucidate the complex mechanisms involved in lung cancer tissue construction and may lead to novel therapies for lung cancer patients.

Materials and Methods
Cell lines and cell culture. The human lung adenocarcinoma cell line (A549) was purchased from the Riken Bioresource center (RCB0098). The cells were cultured in Dulbecco's modified Eagle's medium (DMEM) (Nacalai Tesque, Kyoto, Japan) containing 10% fetal bovine serum (FBS) (Life Technologies), penicillin (100 Units/ml) and streptomycin (100 mg/ml) (Life Technologies) at 37 °C in 5% CO 2 ; cells from an early passage were used for the experiments. HUVECs and Human MSCs were purchased from Lonza. HUVECs were maintained in endothelial growth medium (Lonza) at 37 °C in 5% CO 2 . Human MSCs were maintained in MSC growth medium (Lonza) at 37 °C in 5% CO 2 and were used before Passage 5. We confirmed all of the cell lines were free of mycoplasma contamination by using a mycoplasma detection kit (Takara).

Retroviral transduction.
For retroviral transduction we modified previously described methods 4 , and used pMXs-OCT3/4, pMXs-SOX2 and pMXs-KLF4 vectors obtained from Addgene. PLAT-A packaging cells were plated at 1.2 × 10 6 cells per 6 cm dish, and were incubated overnight. On the following day, the cells were transfected with pMXs vectors using the Fugene HD transfection reagent (Promega). pMXs-EGFP was used as a transfection control. At 24 h after transfection, the medium was replaced with new medium, which was collected (as the virus-containing supernatant) after another 24 h. The virus-containing supernatants were filtered through a 0.45-μm pore filter and supplemented with 4 μg/ml polybrene (Nacalai Tesque). Equal amounts of supernatants containing each of the retroviruses were mixed, transferred to the cancer cell line or HUVEC dish that was prepared the previous day, and incubated overnight. At another 24 h after infection, the virus-containing medium was replaced with fresh medium without retroviruses.

The method of culturing OSK-A549-Colony and OSK-A549-SN cells. OCT3/4-, SOX2-and
KLF4-transduced A549 cells were maintained in DMEM containing 10% FBS and 0.5% penicillin and streptomycin at 37 °C in 5% CO 2 . Colonies appeared in the OSK-A549 cells from 10 to 15 days. We then trypsinized them for 6 minutes with 0.25% trypsin and aspirated the trypsin. After removing the trypsin, the remaining cells were washed with PBS twice. The colony cells remained on the dish, and were picked up, gently pipetted 5 to 7 times and transferred to a new dish. Approximately 5 colonies in a 60 mm dish was sufficient for passaging. From 7 to 10 days after passaging, larger colonies appeared and spindle-shaped cells were found around the colonies. In order to passage colonies, they were trypsinized for 6 minutes with 0.25% trypsin. The dissociated cells in the supernatant were then collected and as OSK-A549-SN cells. After collecting the OSK-A549K-SN cells, the remaining cells were washed with PBS twice and the colonies were picked up, pipetted 5 to 7 times, and then transferred to a new dish.

Tumorigenicity in vivo.
A total of 1.0 × 10 6 cells in 100 μl of serum-free PBS were injected subcutaneously into both dorsal flanks of 8-week-old male BALB/c nude mice. The tumor volume was calculated by the formula 0.5 × L × W 2 (L: length, W: width). The experiments were reviewed, approved by the Animal Welfare Committee, Kitayama Labes Co., Ltd. (Nagano, Japan) (Permit Number: IBC53-016), and conducted in accordance with institutional guidelines. All efforts were made to minimize suffering.

Co-culturing with HUVECs and MSCs.
To generate spheres in vitro, we resuspended ~1.0 × 10 6 the parental A549 or ~1.0 × 10 6 OSK-A549-colony cells, ~3.0 × 10 5 HUVECs, and ~3.0 × 10 5 MSCs were resuspended in sphere forming medium and seeded on a low-attachment 24-well flat plate (Prime surface Sumitomo). We adopted the following specific ratios with reference to a previous report 24 . A549 or OSK-A549-Colony cells: HUVEC: MSC = 10:1:4 ~ 5:4:4. During 6 to 12 days of culture, the self-organized spheres were photographed using a BZ8000 microscope (Keyence) and were pathologically analyzed. To assess the chemosensitivity of the spheres and the IL-6 function of the forming spheres, the spheres were co-cultured with or without 5 μM of cisplatin, 1 μg/ml of anti-IL-6 antibody (R & D, MAB206) and 10 ng/ml of IL-6 protein from the first day of this assay. The photographed pictures were then analyzed to calculate fluorescence intensity using the ImageJ software program.  The microarray and data analyses. RNAs from the parental A549 cells, OSK-A549-Colony cells, and OSK-A549-SN cells were obtained from three independent retroviral transduction experiments. Gene expression profiling was carried out using the SuperPrint G3 Human Gene Expression 8 × 60K v2 Microarray (Agilent Technologies) according to the manufacturer's protocol. The data were analyzed using the GeneSpring GX 13.0 software program (Agilent Technologies). Control probes were removed and only the "detected" probes that were present in at least one of the analyzed samples were used for the further analysis. The number of probes used in the analysis was 30886 (the parental A549 vs A549-OSK-colony cells). Volcano plots were generated and a GO analysis was performed using the Gene Spring GX13.0 software program.

RNA isolation and the quantitative reverse-transcriptase polymerase chain reaction.
Immunocytochemistry. For immunocytochemistry, the cultured cells, which were fixed with 4% para- A cell cycle analysis using Ki67/Hoechst co-staining was performed according to a previously reported protocol 49 . Briefly, paraformaldehyde-fixed cells were permeabilized with 0.2% Triton-X (Sigma), and the cells were stained with PE Mouse Anti-Human Ki67 (1:50, BD, 556027) overnight in Permwash buffer (BD). PE mouse IgG1, k Isotype Control (1:50, BD, 556027) staining was used as a control. The cells were then stained for 5 min at RT with Hoechst 33342 (1:10,000; Life Technologies). The cells were analyzed by flow cytometry (FACS Aria III).
The cell-cycle assay. A total of 2.0 × 10 4 parental A549 and OSK-A549-Colony cells were plated on a 24-well plate and incubated for 24 hours. The cell-cycle was analyzed using a Cell-Clock Mammalian Cell Cycle Assay Kit (Biocolor life science assays) according to the manufacturer's protocol.
The wound healing assays. The parental A549 and OSK-A549-SN cells were seeded in 35 mm cell culture dishes. At 100% confluence, the cell layers were scratched using a plastic pipette tip. The wound healing of each cell was photographed at 30 hours.
The double-layered collagen gel hemisphere (DL-CGH) assay. As previously reported 12 , acid-soluble collagen 1 (Nitta Gelatin Inc., Osaka, Japan), 10-fold-concentrated DMEM medium, and reconstruction buffer (2.2 g NaHCO3 + 4.77 g HEPES in 100 ml of 0.05-N NaOH) were mixed at a volume ratio of 8:1:1. Five microliters of mixture containing 3.0 × 10 4 cells was then dropped onto a plastic dish. After 15 min of incubation at room temperature, a second 30-μl drop of collagen was placed directly on top of the first gel drop, encapsulating it completely. After 1 h of incubation at room temperature, the gel hemisphere was then submerged in medium and incubated at 37 °C for 2 weeks in a 5% CO 2 incubator. The gel hemisphere was observed using a BZ8000 microscope (Keyence), and the invasion area was calculated using the ImageJ software program.
The cisplatin sensitivity test. A total of 1.0 × 10 5 of the parental A549, OSK-A549-Colony, and OSK-A549-SN cells were seeded in each well of 6-well plates and co-cultured with 2 μM or 10 μM of cisplatin (WAKO) for 3 days. The number of surviving cells was counted and the survival rate was calculated.
Digital holographic microscopy. Twenty 3D holographic cell images were captured using a HoloMonitor M4 (Phase Holographic Imaging) with a 20x magnification objective and a low-power 635nm diode laser. Images were converted from wavelength interactions to cellular representations by a computer algorithm (HstudioM4, Phase Holographic Imaging) and the positional information, including the X and Y positions were recorded each 5 minutes. For the motility and migration analysis, we randomly selected 8 cells per HPF.
RNA in situ hybridization. RNA in situ hybridization for human IL-6 mRNA was performed using an RNA scope Multiplex Fluorescent Reagent Kit (Advanced Cell Diagnostics, Inc. 320850) in accordance with the manufacturer's instructions. In brief, 5-μm-thick formalin-fixed, paraffin-embedded tissue sections were deparaffinized and pre-treated with heat and protease before hybridization with the IL-6 oligonucleotide probes (425161-C1). Preamplifier, amplifier and alkaline-phosphatase-labeled oligonucleotides were then hybridized sequentially, coupled with a fluorescent conjugate. The RNA integrity quality was controlled with an RNAscope ® Scientific RepoRts | 7: 12317 | DOI:10.1038/s41598-017-12017-y LS-positive Control Probe -Hs-PPIB (Advanced Cell Diagnostics, Inc. 313907), and for the background with an RNAscope ® -negative Control Probe-DapB (Advanced Cell Diagnostics, Inc. 310043). The specific RNA staining signal for IL-6 was identified as red.
For double staining of IL-6 mRNA and cytokeratin protein, after in situ hybridization for IL-6 mRNA, the sections were autoclaved in Tris buffer (pH 8.0; 2.5 min) for antigen retrieval and incubated for 60 min with a monoclonal antibody against human cytokeratins (BD, clone CAM5.2, dilution 1:10. AF 349205). The reaction product was visualized with Alexa Fluor ® fluorescent goat anti-mouse IgG antibodies (1:500, Abcam, Cambridge, UK).
Lung adenocarcinoma clinical samples. Patients with lung adenocarcinoma who underwent surgical resection in the department of thoracic surgery in Kobe University Hospital were randomly selected. The medical records were reviewed. The mutation status had already been clinically confirmed by a PCR (for EGFR mutations), fluorescence in situ hybridization (FISH) and immunohistochemistry (for EML4-ALK rearrangement), and the Scorpion-ARMS method (for KRAS mutations). The institutional review board of Kobe University Hospital approved the use of the tumor specimens (No. 160073). Written informed consent was obtained from all the patients. Statistical analysis. All data were analyzed using the JMP software program, ver. 9.0.0 (SAS Institute, Cary, NC, USA). The data values were presented as the mean ± standard error of the mean (SEM) of three independent experiments. Bonferroni's test was used for the analyses of the data in Figs 1A and 3B. A repeated measures analysis of variance was used for the analysis of the data in Fig. 1F, 1J and S3B. Dunnett's test was used for multiple comparisons in Figs 1G, 3E, S1H and S4. The differences in the mean values between two groups were analyzed using the two-tailed paired t-test in Fig. 4B, D, F, H and Fig. S3E. The differences were considered to be statistically significant for P-values < 0.05 (*) and <0.01 (**).
All methods were performed in accordance with the relevant guidelines and regulations including Declaration of Helsinki and Ethical Guidelines for Medical and Health Research Involving Human Subjects. Data Availability. The microarray data were deposited in the NCBI Gene Expression Omnibus under accession number GSE89229.