Quiescent cancer cells are believed to cause cancer progression after chemotherapy through unknown mechanisms. We show here that human non–small cell lung cancer (NSCLC) cell line-derived, quiescent-like, slow-cycling cancer cells (SCC) and residual patient-derived xenograft (PDX) tumors after chemotherapy experience activating transcription factor 6 (ATF6)-mediated upregulation of various cytokines, which acts in a paracrine manner to recruit fibroblasts. Cancer-associated fibroblasts (CAF) underwent transcriptional upregulation of COX2 and type I collagen (Col-I), which subsequently triggered a slow-to-active cycling switch in SCC through prostaglandin E2 (PGE2)- and integrin/Src-mediated signaling pathways, leading to cancer progression. Both antagonism of ATF6 and cotargeting of Src/COX2 effectively suppressed cytokine production and slow-to-active cell cycling transition in SCC, withholding cancer progression. Expression of COX2 and Col-I and activation of Src were observed in patients with NSCLC who progressed while receiving chemotherapy. Public data analysis revealed significant association between COL1A1 and SRC expression and NSCLC relapse. Overall, these findings indicate that a proinflammatory niche created by the interplay between SCC and CAF triggers tumor progression.

Significance:

Cotargeting COX2 and Src may be an effective strategy to prevent cancer progression after chemotherapy.

Non–small cell lung cancer (NSCLC) is the leading cause of cancer-related death worldwide (1, 2). Despite significant advances in anticancer therapeutics, chemotherapy is currently the standard treatment option for patients with NSCLC (3). Although chemotherapy effectively debulks some tumors (4), nearly all patients ultimately experience progressive disease (5). Because conventional chemotherapy targets actively cycling cancer cells (ACC), tumor progression may stem from a small population of latent slow-cycling or quiescent cancer cells (6). Accordingly, cellular quiescence has been proposed as a principal yet little studied cause of tumor progression (7). However, precisely how latent cancer cells induce tumor progression is largely unknown.

The tumor microenvironment (TME) has been implicated in fostering tumor development, recurrence, and progression (6). Chronic inflammation is generally ascribed a role in creating a proinflammatory niche in the TME (8), wherein growth factors and cytokines induce COX2 expression and subsequent prostaglandin E2 (PGE2) production (9). In addition to its pleiotropic effects on the proliferation, survival, angiogenesis, and invasiveness of neoplastic cells (10), PGE2 has also been found to promote the proliferation of latent tumor cells (11). In addition, latent tumor cells support their own proliferation by remodeling the surrounding TME. A previous study demonstrated that fibrotic lungs resulting from TGFβ1-induced type I collagen (Col-I) deposition support the dormant-to-proliferative switch of dormant breast tumor cells (12). Col-I–mediated activation of integrins, particularly integrin β1, and activated Src, a key downstream effector of integrin signaling, were found to stimulate proliferation in dormant cancer cells (7, 12). Furthermore, the pharmacologic inhibition of Src induced dormancy in slow-cycling breast cancer cells (13). These results collectively suggest that the PGE2- and collagen/integrin/Src-mediated signaling pathways play important roles in tumor development, recurrence, and progression. An additional mechanism is the unfolded protein response (UPR), which is mediated by three sensor proteins, protein kinase R–like endoplasmic reticulum kinase (PERK), activating transcription factor 6 alpha (ATF6α), and inositol requiring enzyme 1 alpha (IRE1α; ref. 14). Numerous hazardous stresses on the microenvironment, including chemotherapy, trigger the UPR to restore endoplasmic reticulum (ER) homeostasis by suppressing protein synthesis and inducing transcription of several genes, including those involved in ER protein-folding or degradation (15). However, it is unclear whether and how the PGE2- and collagen/integrin/Src-mediated signaling pathways and UPR are implicated in the tumor progression caused by dormant-to-proliferative switch of a small subpopulation of latent NSCLC cells.

In this study, we sought to explore the critical triggers at the molecular and cellular levels that stimulate the transition of latent NSCLC cells into a proliferative state, leading to tumor progression. Here, we report that slow-cycling cancer cells (SCC) carrying quiescence-like phenotypes in residual NSCLC after chemotherapy form a proinflammatory niche through the ATF6-mediated transcriptional upregulation of various proinflammatory cytokines, which cooperatively act in a paracrine manner to recruit fibroblasts. Activated fibroblasts experience transcriptional upregulation of COX2 and Col-I, which in turn trigger the PGE2 and integrin/Src signaling-mediated transition of SCCs to ACCs. siRNA-mediated abrogation of ATF6 expression or treatment with celecoxib and dasatinib in combination significantly suppressed the dormant-to-proliferative switch in SCCs in vitro and SCC-mediated tumor recurrence/progression in vivo. The clinical relevance of these findings was also confirmed by using NSCLC patient-derived xenograft (PDX) tumors that survived chemotherapy. We also present the clinical utility of Col-I and Src as biomarkers for recurrence after chemotherapy in patients with NSCLC. These results collectively suggest that the proinflammatory niche established by the ATF6-initiated communication between SCCs and fibroblasts in the TME is a potential therapeutic target to prevent tumor progression.

Additional or detailed methods are described in the Supplementary Materials and Methods.

Cell culture

Human NSCLC cell lines (H460 and H1299 cells), Wi38 and MRC5 normal human lung fibroblasts, and Lewis lung carcinoma (LLC) cells were purchased from the ATCC. The cells were cultured in DMEM (LLC cells) or RPMI1640 medium (all other cells) supplemented with 10% FBS and antibiotics (Welgene, Kyeongsan-si, Republic of Korea) and maintained at 37°C with 5% CO2 in a humidified atmosphere. Chemotherapy-resistant SCC models were generated by continuously exposing the cells to increasing concentrations of the corresponding chemotherapeutic drugs for more than 6 months. The NSCLC cells and their chemotherapy-resistant sublines were authenticated and validated using the AmpFLSTR identifiler PCR Amplification Kit (Applied Biosystems; catalog no. 4322288) in 2013 and 2017. We used NSCLC cells and their chemotherapy-resistant SCC models, which were authenticated and passed for fewer than 6 months after receipt or resuscitation of validated cells.

Evaluation of the quiescence-like phenotypes of SCCs

To determine the quiescence-like phenotypes of SCCs compared with their corresponding parental cells (ACC), the cell proliferation and cell cycle status were analyzed by cell counting assay and flow cytometry, respectively. The accumulation of cultured cells in the G0–G1 phase of the cell cycle was determined by propidium iodide staining of fixed cells. Before harvest and fixation, serum-starved cells were stimulated with serum-containing media for 4, 8, and 12 hours. The Hoechst 33342/pyronin Y double staining of primary cells obtained from tumor xenografts on tumor onset was performed to distinguish resting cells in the G0 phase from proliferating cells based on the difference in DNA and RNA contents.

Establishment of the PDX #4 and #5

Tumor specimens were directly obtained from two patients with NSCLC who progressed after prior platinum-based chemotherapy. This study was approved by the institutional review board of Severance Hospital (No. 4-2013-0526). Tumor tissues were obtained from the patient who provided written informed consent. PDXs were created by using 6- to 8-week-old female severe combined immunodeficient (NOG) and nude (nu/nu) mice obtained from Orient Bio (Seongnam-si, Gyeonggi-do, Republic of Korea). All methods complied with the guidelines of our institutional animal research committee (Yonsei University College of Medicine, Seoul, Korea) and were approved by the Association for Assessment and Accreditation of Laboratory Animal Care. We carefully remove the necrotic and supporting tissues from core biopsy specimens by surgical blade. Subsequently, small specimens of the tumor tissue (3 mm × 3 mm × 3 mm) were implanted subcutaneously in 1–2 mice. After the tumor reached a diameter of approximately 1.5 cm, it was excised, dissected into small specimens (3 mm × 3 mm × 3 mm), and implanted into another set of mice. The original patient-derived specimens were defined as the F0 generation, whereas subsequent generations were numbered consecutively based on the number of reimplantations (e.g., F1, F2, or F3). Finally, the third mouse generation (F3) was expanded for in vivo drug efficacy testing. When the tumor measured >1 cm3 or the animal reached an endpoint described in the Dutch Code of Practice for animal experiments in cancer research (16), the tumor was harvested and placed in media for either storage or propagation into a subsequent generation. The tumors and related PDXs were assigned Yonsei Human In Mouse (YHIM) identifiers that corresponded to the original patient-derived tumors.

Animal studies

All animal procedures were performed using protocols approved by the Institutional Animal Care and Use Committee of Seoul National University (Seoul, Korea) or Yonsei University (Seoul, Korea). For xenograft or allograft experiments, cells (0.2–1 × 106 cells/mouse) were subcutaneously injected into the right flank of 6-week-old immunodeficient [nude or nonobese diabetic (NOD)/severe combined immune-deficient (SCID)] or C57BL/6 mice. For PDX tumor experiments, tumors were minced into 2 mm3 pieces and subcutaneously inoculated into NOD/SCID mice. After the tumor volume reached 50–250 mm3, the mice were randomly grouped and treated with vehicle or test materials. Tumor growth was determined by measuring the short and long diameters of the tumor with a caliper, and body weight was measured twice per week to monitor toxicity. Tumor volume was calculated using the following formula: tumor volume (mm3) = (short diameter)2 × (long diameter) × 0.5.

In silico analysis

We used publicly available datasets deposited in the Gene Expression Omnibus database [GEO, National Center for Biotechnology Information; GSE37745 (17) or GSE41271 (18)] for analyzing the association of SRC, COL1A1, or PTGS2 expression with histologic status and recurrence in patients with NSCLC and the correlation of gene expression with cell proliferation. The raw data containing the gene expression levels and clinical information, such as histology, survival status, and duration of survival, for each patient sample were manually downloaded and analyzed using GraphPad Prism 7. The association of SRC, COL1A1, or PTGS2 expression with histology or patient prognosis was analyzed by two-sided Student t test, Mann–Whitney test, or one-way ANOVA followed by Dunnett post hoc test according to the number of groups and the distribution of the data. Probes used to obtain gene expression values in each dataset were listed in Supplementary Table S1. The normality of the data was determined by using the D'Agostino–Pearson omnibus test. The Pearson correlation coefficient was calculated when using data with a Gaussian distribution, whereas the Spearman correlation coefficient was calculated when using data that did not pass the normality test.

Statistical analyses

The data are presented as the mean ± SD. All in vitro experiments were performed independently at least twice, and a representative result is shown. Statistical significance of difference between means of two experimental groups were determined by two-tailed Student t test using Microsoft Excel 2010 (Microsoft Corp.) or GraphPad Prism 7 (GraphPad Software Inc.). P values less than 0.05 were considered significant. All in vivo experiments were performed at least twice using at least 6 mice per group, and the combined results from two independent experiments were presented as a graph. The sample size of each group was predetermined using the G*power program (http://www.gpower.hhu.de/; ref. 19) assuming α = 0.05 and β = 0.05, tailored to detect 40-50% reduction in tumor size with 20%–30% SD spans, yielding n = 6/group. Statistical power was also calculated using the G*power program; the statistical power of results from all in vivo experiments were over 95%. Statistical significance of difference between means of two experimental groups were determined by two-tailed Student t test using Graphpad Prism 7. The normal distribution of the data in each group was confirmed by Kolmogorov–Smirnov test. To determine statistical significance of difference among means of more than three experimental groups, we performed one-way ANOVA followed by the Holm–Bonferroni post hoc test. The difference was considered statistically significant when P < 0.05. Sample sizes and statistical analysis results of individual in vivo experiments are described in figure legends.

Chemoresistant subpopulations of NSCLC cells exhibit quiescence-like phenotypes in vitro that were highly reversible in vivo

A small subpopulation of SCCs, which presumably cause chemoresistance and tumor progression (20), have been observed even in rapidly growing tumors (21) and cancer cell lines (22). To investigate mechanisms underlying tumor progression after chemotherapy, we attempted to establish in vitro and in vivo SCC models. First, we tried to obtain SCCs by exposing H460 and H1299 human NSCLC cell lines to gradually increasing doses of paclitaxel, cisplatin, or pemetrexed (Fig. 1A), chemotherapeutic drugs frequently used in the clinic (4). We noted subpopulations of H460 cells with paclitaxel resistance (H460/PcR) and H1299 cells with cisplatin (H1299/CsR) or pemetrexed resistance (H1299/PmR) exhibited significantly increased chemoresistance, as determined by minimal changes in viability (Fig. 1B), colony formation capacity (Fig. 1C), and proapoptotic activity (Supplementary Fig. S1) in the presence of chemotherapeutic drugs, along with reduced proliferation (Fig. 1D). Furthermore, unlike their respective parental cells, most of the serum-starved cells in the subpopulations remained in the G0–G1 phase upon stimulation with serum (Fig. 1E).

Figure 1.

Generation of chemoresistant sublines, in vitro and in vivo characterization, and increased recruitment of fibroblasts in tumors derived from SCCs. A, Schematic diagram showing the generation of SCCs (H460/PcR, H1299/CsR, and H1299/PmR cells). B and C, Acquisition of resistance to the corresponding chemotherapy in H460/PcR, H1299/CsR, and H1299/PmR cells compared with that in the corresponding parental cells (H460 and H1299 cells), as determined by a cell counting assay (B) and anchorage-dependent colony formation assay (C). Quiescence-like phenotype of H460/PcR, H1299/CsR, and H1299/PmR cells compared with that of the parental cells (H460 and H1299 cells) in vitro and in vivo, as determined by a cell counting assay (D) and flow cytometric analyses of the accumulation in the G0–G1 phase using propidium iodide staining (E), and Hoechst 33342/pyronin Y-double staining using tumor cells isolated from xenograft tumors of ACCs and SCCs (G). G, Right, the quantification of primary cells from xenograft tumors of ACCs and SCCs (n = 3) in each phase of the cell cycle is depicted as graphs. F, Schematic diagram for analysis of xenograft tumors by flow cytometry and IHC analyses. H, Acquisition of resistance to the corresponding chemotherapy in H460/PcR (n = 12, each group), H1299/CsR (n = 12, each group), and H1299/PmR (n = 12, each group) cells compared with that in the corresponding parental cells [H460 (n = 12, each group) and H1299 (n = 12, each group) cells] in vivo, as determined by tumor xenograft experiments using nude mice. For all panels, the bars represent the mean ± SD. ***, P < 0.001, as determined by a two-tailed Student t test.

Figure 1.

Generation of chemoresistant sublines, in vitro and in vivo characterization, and increased recruitment of fibroblasts in tumors derived from SCCs. A, Schematic diagram showing the generation of SCCs (H460/PcR, H1299/CsR, and H1299/PmR cells). B and C, Acquisition of resistance to the corresponding chemotherapy in H460/PcR, H1299/CsR, and H1299/PmR cells compared with that in the corresponding parental cells (H460 and H1299 cells), as determined by a cell counting assay (B) and anchorage-dependent colony formation assay (C). Quiescence-like phenotype of H460/PcR, H1299/CsR, and H1299/PmR cells compared with that of the parental cells (H460 and H1299 cells) in vitro and in vivo, as determined by a cell counting assay (D) and flow cytometric analyses of the accumulation in the G0–G1 phase using propidium iodide staining (E), and Hoechst 33342/pyronin Y-double staining using tumor cells isolated from xenograft tumors of ACCs and SCCs (G). G, Right, the quantification of primary cells from xenograft tumors of ACCs and SCCs (n = 3) in each phase of the cell cycle is depicted as graphs. F, Schematic diagram for analysis of xenograft tumors by flow cytometry and IHC analyses. H, Acquisition of resistance to the corresponding chemotherapy in H460/PcR (n = 12, each group), H1299/CsR (n = 12, each group), and H1299/PmR (n = 12, each group) cells compared with that in the corresponding parental cells [H460 (n = 12, each group) and H1299 (n = 12, each group) cells] in vivo, as determined by tumor xenograft experiments using nude mice. For all panels, the bars represent the mean ± SD. ***, P < 0.001, as determined by a two-tailed Student t test.

Close modal

We then analyzed the cell-cycle profile of these three subpopulations in vivo. To this end, we established xenograft tumors of the subpopulations and their parental cells by their subcutaneous injection into NOD-SCID mice (Fig. 1F). On tumor onset, primary tumor cells were harvested and stained with Hoechst 33342 and Pyronin Y (23) to identify cancer cells in G0-like status. Xenograft tumors of H460/PcR, H1299/CsR, and H1299/PmR had significantly more cells in the G0 phase of the cell cycle compared with those of their corresponding parental cells (Fig. 1G). We then assessed the response of H460 and H460Pc/R xenograft tumors to chemotherapy. To this end, mice carrying xenograft tumors with 100 to 150 mm3 in volume were subjected to treatment with chemotherapeutics. We found that xenograft tumors of these subpopulations consistently showed chemoresistance (Fig. 1H). Therefore, these subpopulations were likely under a quiescence-like, slow-cycling status in vitro but recovered an active-cycling status in in vivo microenvironment. Hence, we chose these subpopulations as clinically relevant in vitro experimental models of SCCs with chemoresistance to study the biology of tumor progression.

Accumulation of fibroblasts in NSCLC cell lines and patient-derived xenograft tumors during chemotherapy treatment

We explored the mechanisms that caused the SCCs to resume growth in vivo. The proposed roles of diverse stromal cells in TME, including fibroblasts, macrophages, and vascular endothelial cells, in chemoresistance and tumor progression (7, 24), prompted us to identify the main player that contributed to tumor progression after the completion of chemotherapy. IHC analyses of xenograft tumors derived from the three paired parental lines and their corresponding SCCs using an antibody against FSP1, a molecular marker for cancer-associated fibroblast (CAF) facilitating malignant progression in tumor microenvironment (25) revealed consistent increases in FSP1 staining in the SCC-derived xenograft (Fig. 2A). Two additional (CAF) markers, including an intermediate-filament-associated protein alpha-smooth muscle actin (αSMA) and a serine protease fibroblast activation protein 1 (FAP1; ref. 26) were also consistently increased in the SCC-derived xenograft. VEGFR2+ vascular endothelial cells were also increased in the SCC-derived xenograft tumors but were not consistently significant. In contrast, the paired tumors showed no significant differences in the number of F4/80+ macrophages. These findings indicated the potential of the SCCs to recruit stromal cells, most notably fibroblasts.

Figure 2.

Increased recruitment of fibroblasts in tumors derived from those progressed after chemotherapy in vivo. A, IHC analyses to determine the accumulation of fibroblasts in xenograft tumors derived from H460/PcR, H1299/CsR, or H1299/PmR cells compared with that in the xenograft tumors derived from the corresponding parental cells (H460 and H1299 cells) as shown in Fig. 1H. Right, the quantification of the positive cells for each stromal-specific marker per field of view (FOV) is depicted as a graph. B, C, and F, Mice bearing H460 xenograft tumors (n = 12, each group; B), LLC allograft tumors (n = 12, each group; C), or lung PDX tumors derived from three different NSCLC patients (n = 12, each group; F) were subjected to three cycles of a one-week chemotherapy treatment [paclitaxel (20 mg/kg) and cisplatin (3 mg/kg) in combination]. D and E, IHC analysis showing the recruitment of fibroblasts in the tumors that escaped quiescence after chemotherapy. G, The quantification of the positive cells for each marker per field of view is depicted as a graph. Representative IHC images are shown in Supplementary Fig. S3. For quantification of IHC analyses, at least three tumors per group were analyzed. Scale bar, 50 μm. Scale bar (insets), 10 μm. For all panels, the bars represent the mean ± SD. *, P < 0.05; **, P < 0.01; ***, P < 0.001, as determined by a two-tailed Student t test. H&E, hematoxylin and eosin.

Figure 2.

Increased recruitment of fibroblasts in tumors derived from those progressed after chemotherapy in vivo. A, IHC analyses to determine the accumulation of fibroblasts in xenograft tumors derived from H460/PcR, H1299/CsR, or H1299/PmR cells compared with that in the xenograft tumors derived from the corresponding parental cells (H460 and H1299 cells) as shown in Fig. 1H. Right, the quantification of the positive cells for each stromal-specific marker per field of view (FOV) is depicted as a graph. B, C, and F, Mice bearing H460 xenograft tumors (n = 12, each group; B), LLC allograft tumors (n = 12, each group; C), or lung PDX tumors derived from three different NSCLC patients (n = 12, each group; F) were subjected to three cycles of a one-week chemotherapy treatment [paclitaxel (20 mg/kg) and cisplatin (3 mg/kg) in combination]. D and E, IHC analysis showing the recruitment of fibroblasts in the tumors that escaped quiescence after chemotherapy. G, The quantification of the positive cells for each marker per field of view is depicted as a graph. Representative IHC images are shown in Supplementary Fig. S3. For quantification of IHC analyses, at least three tumors per group were analyzed. Scale bar, 50 μm. Scale bar (insets), 10 μm. For all panels, the bars represent the mean ± SD. *, P < 0.05; **, P < 0.01; ***, P < 0.001, as determined by a two-tailed Student t test. H&E, hematoxylin and eosin.

Close modal

To investigate whether the SCCs established in vitro accurately model the SCCs in residual tumors after chemotherapy in vivo, we determined whether fibroblasts emerge within recurrent or progressive lung tumors after a clinically relevant, short-term chemotherapy in mice. To this end, NOD/SCID mice bearing H460 xenograft tumors (Fig. 2B) and C57BL/6 mice bearing LLC allograft tumors (Fig. 2C; Supplementary Fig. S2A) were subjected to combinatorial chemotherapy. The chemotherapy-treated tumors showed significantly decreased growth compared with control tumors. However, the residual tumors resumed rapid growth after chemotherapy completion. The progressed tumors showed significantly increased numbers of fibroblasts compared with control tumors (Fig. 2D and E; Supplementary Fig. S2B). To link these experimental studies to the clinical phenomenon of tumor progression in patients, we further administered NSCLC PDX tumors established in NOD/SCID mice with a clinically relevant combinatorial chemotherapy. Among several PDX tumors, three PDX tumors (PDX #1–PDX #3) rapidly shrunk and remained impalpable during the combinatorial chemotherapy, but eventually regrew, resulting in tumor progression (Fig. 2F). The progressed PDX tumor tissues revealed consistent increases in the number of fibroblasts compared with their untreated control tumors (Fig. 2G; Supplementary Fig. S3), indicating CAFs accumulation in the tumors that progressed while receiving chemotherapy.

Soluble factor secretion mediated by communication between SCCs and CAFs stimulates the dormant-to-proliferative transition and tumor progression

On the basis of the role of CAFs in tumor development and vascular formation (25), we monitored the possible interplay between SCCs and CAFs. We first determined whether SCCs recruited fibroblasts. We found that the human fibroblasts (Wi38 or MRC5 cells) cocultured with the established SCCs (Fig. 3A) or incubated with the conditioned medium (CM) derived from the SCCs (Fig. 3B; Supplementary Fig. S4) displayed significantly increased migration compared with those cocultured with the corresponding ACCs or incubated with the CM from the corresponding ACCs, respectively. These findings indicated the capacity of the SCCs to chemoattract fibroblasts.

Figure 3.

Awakening SCCs from quiescence by soluble factor-mediated communication with fibroblasts in vitro and in vivo. A and B, Increases in fibroblast migration caused by coculture with SCCs (A) or incubation with CM from H460/PcR, H1299/CsR, and H1299/PmR cells (B) compared with those caused by coculture with the corresponding parental cells or incubation with CM from the parental cells. C–E, Increases in the proliferation (C), Ki67-positive proliferating cell population size (D), and anchorage-independent colony formation (E) of H460/PcR, H1299/CsR, and H1299/PmR cells cocultured with fibroblasts, as determined by a cell counting assay (C), immunofluorescence staining (D), and a soft agar colony formation assay (E). In the soft agar assay, cancer cells (1 × 103 cells) were mixed with Wi38 fibroblasts at ratios of 1:0 to 1:1 (depicted in the figure in detail) and then incubated in soft agar. In this experiment, the Wi38 cells did not form visible colonies in the soft agar. F, Modulation of the expression of cell-cycle–related proteins in H460/PcR, H1299/CsR, and H1299/PmR cells cocultured with Wi38, as determined by Western blot analysis. G, Increased anchorage-dependent colony formation (left) and Ki67+-positive proliferating cell population size (right) of SCCs caused by incubation with the CM obtained from the coincubation of Wi38 fibroblasts with H460/PcR, H1299/CsR, or H1299/PmR cells. Cancer cells (parental cells or SCCs, 3.3 or 6 × 105 cells in 60 mm dishes) were mixed with Wi38 fibroblasts at a ratio of 1:0 to 1:1 (depicted in the figure in detail) and incubated for 1–2 days. The culture medium was replaced with fresh serum-free medium, and the cells were further incubated for 1 day. The CM was harvested and normalized by adding additional serum-free medium according to the number of cells attached to the dish. H, i, The growth of xenografted tumors composed of H460 or H460/PcR cells alone or in combination with fibroblasts at a ratio of 1:0.1 or 1:1 (cancer cells: fibroblasts). Each bar in the graph represents one tumor. ii, Fold changes of tumor growth at 46 days after the tumor inoculation were depicted. I, i and J, ii, Parental cells (H460) or SCCs (H460/PcR cells) alone or in combination with an equal number of fibroblasts were inoculated into the left flank of NOD/SCID mice (n = 12, each group), and H460/PcR cells were inoculated into the right flank of the same NOD/SCID mice at the same time. Tumor growth was monitored every day and measured at the intervals depicted in the scheme. The growth of the tumors in the left flank [feeder tumors (I, i and J, i); and right flank (I, ii and J, ii)] of the NOD/SCID mice at 21, 28, and 35 days after the tumor inoculation was determined. For all panels, the bars represent the mean ± SD. *, P < 0.05; **, P < 0.01; ***, P < 0.001, as determined by a two-tailed Student t test.

Figure 3.

Awakening SCCs from quiescence by soluble factor-mediated communication with fibroblasts in vitro and in vivo. A and B, Increases in fibroblast migration caused by coculture with SCCs (A) or incubation with CM from H460/PcR, H1299/CsR, and H1299/PmR cells (B) compared with those caused by coculture with the corresponding parental cells or incubation with CM from the parental cells. C–E, Increases in the proliferation (C), Ki67-positive proliferating cell population size (D), and anchorage-independent colony formation (E) of H460/PcR, H1299/CsR, and H1299/PmR cells cocultured with fibroblasts, as determined by a cell counting assay (C), immunofluorescence staining (D), and a soft agar colony formation assay (E). In the soft agar assay, cancer cells (1 × 103 cells) were mixed with Wi38 fibroblasts at ratios of 1:0 to 1:1 (depicted in the figure in detail) and then incubated in soft agar. In this experiment, the Wi38 cells did not form visible colonies in the soft agar. F, Modulation of the expression of cell-cycle–related proteins in H460/PcR, H1299/CsR, and H1299/PmR cells cocultured with Wi38, as determined by Western blot analysis. G, Increased anchorage-dependent colony formation (left) and Ki67+-positive proliferating cell population size (right) of SCCs caused by incubation with the CM obtained from the coincubation of Wi38 fibroblasts with H460/PcR, H1299/CsR, or H1299/PmR cells. Cancer cells (parental cells or SCCs, 3.3 or 6 × 105 cells in 60 mm dishes) were mixed with Wi38 fibroblasts at a ratio of 1:0 to 1:1 (depicted in the figure in detail) and incubated for 1–2 days. The culture medium was replaced with fresh serum-free medium, and the cells were further incubated for 1 day. The CM was harvested and normalized by adding additional serum-free medium according to the number of cells attached to the dish. H, i, The growth of xenografted tumors composed of H460 or H460/PcR cells alone or in combination with fibroblasts at a ratio of 1:0.1 or 1:1 (cancer cells: fibroblasts). Each bar in the graph represents one tumor. ii, Fold changes of tumor growth at 46 days after the tumor inoculation were depicted. I, i and J, ii, Parental cells (H460) or SCCs (H460/PcR cells) alone or in combination with an equal number of fibroblasts were inoculated into the left flank of NOD/SCID mice (n = 12, each group), and H460/PcR cells were inoculated into the right flank of the same NOD/SCID mice at the same time. Tumor growth was monitored every day and measured at the intervals depicted in the scheme. The growth of the tumors in the left flank [feeder tumors (I, i and J, i); and right flank (I, ii and J, ii)] of the NOD/SCID mice at 21, 28, and 35 days after the tumor inoculation was determined. For all panels, the bars represent the mean ± SD. *, P < 0.05; **, P < 0.01; ***, P < 0.001, as determined by a two-tailed Student t test.

Close modal

We next assessed phenotypic changes in the SCCs upon the interaction with CAFs. Compared with the SCCs cultured alone or the corresponding ACCs cocultured with Wi38 cells under the same conditions, the SCCs cocultured with Wi38 cells displayed significantly increased proliferation (Fig. 3C), Ki67 index scores (Fig. 3D), a marker of G0 exit and cell proliferation (27), and anchorage-independent colony-forming capacity (Fig. 3E), an in vitro marker of tumorigenic potential (28). The SCCs cocultured with Wi38 cells revealed greater expression of cyclin D1, CDK4, and PCNA and lower expression of CDK inhibitors, including p21 and p27, than the SCCs cultured alone (Fig. 3F). Similarly, the SCCs cocultured with MRC5 cells also displayed significantly greater proliferation and Ki67 index scores with increased PCNA and reduced p21 and p27 expression than the SCCs cultured alone (Supplementary Fig. S5). These findings suggested that recruited fibroblasts stimulated the transition of SCCs into ACCs.

Soluble factors are crucial for indirect cell-to-cell communication (29). As the transwell coculture system permits cell-to-cell interactions only via soluble factors, the ability of the fibroblasts to stimulate the proliferation and migration of the SCCs seemed to be mediated by secreted soluble factors. Notably, compared with the naïve SCCs cultured alone, the SCCs incubated with the CM derived from cocultures of H460/PcR cells and Wi38 fibroblasts revealed significantly increased colony formation and Ki67 index scores (Fig. 3G). Hence, soluble factors derived from the communication between the SCCs and CAFs appeared to enable the SCCs to escape growth arrest and recover their capacity for proliferation.

To validate the in vitro findings in an in vivo system, mice were subcutaneously coinoculated with H460 or H460/PcR cells along with Wi38 fibroblasts in increasing ratios. The mice injected with H460/PcR cells in combination with Wi38 cells revealed earlier tumor development than those injected with H460/PcR cells alone (Fig. 3H, i). While all mice (n = 11) injected with H460 cells developed tumors, only 38% (3/8) of the mice that received H460/PcR cells alone developed tumors on day 24 after the injection. In contrast, 87.5% (7/8) and 92% (11/12) of the mice coinjected with H460/PcR and Wi38 cells in a 1:0.1 or 1:1 ratio, respectively, showed tumor formation. When analyzed on day 38, 100% of mice that received H460/PcR cells successfully developed tumors with 100% of incidence. Moreover, on day 46 after injection, the mice with [H460/PcR][Wi38] xenograft tumors showed greater tumor growth than the mice with [H460/PcR] xenograft tumors (Fig. 3H, ii).

To further assess the role of soluble factors secreted via the interaction between the SCCs and CAFs in tumor progression, we performed a dual-flank study in which “feeder” tumors composed of [H460] or [H460/PcR] or those composed of [H460][Wi38] or [H460/PcR][Wi38] could accelerate the growth of a “secondary” xenograft tumors of H460/PcR cells in the contralateral flank. [H460/PcR] (Fig. 3I, i) and [H460/PcR][Wi38] (Fig. 3J, i) feeder tumors grew at slower rates than the [H460] and [H460][Wi38] tumors, respectively. The feeder tumor sizes in the [H460] and [H460/PcR] groups were 109.97 ± 109.20 cm3 and 19.43 ± 30.61 cm3 on day 21, 429.44 ± 334.17 cm3 and 81.48 ± 81.90 cm3 on day 28, and reached 772.82 ± 542.09 and 296.77 ± 242.64 cm3 on day 35, respectively. The feeder tumor sizes in the [H460][Wi38] and [H460/PcR][Wi38] groups were 117.46 ± 71.47 cm3 and 53.58 ± 39.37 cm3 on day 21, 394.06 ± 210.62 cm3 and 226.31 ± 120.13 cm3 on day 28, and reached 966.53 ± 471.84 and 768.47 ± 265.92 cm3 on day 35, respectively. Despite this “disadvantage,” the secondary tumors composed of H460/PcR cells in the experimental groups with [H460/PcR] (Fig. 3I, ii) or [H460/PcR][Wi38] feeder tumors (Fig. 3J, ii) on the contralateral flank underwent more rapid tumor growth than those with [H460] or [H460][Wi38] feeder tumors, respectively. Hence, soluble factors secreted from even a limited SCC-derived tumor burden appeared to produce systemic effects that accelerated the growth of secondary tumors at a distant site. These findings clearly indicated that the SCCs in residual tumors after chemotherapy communicate with CAFs in the TME through soluble factor secretion and act as a source for tumor progression.

ATF6-mediated secretion of proinflammatory cytokines by SCCs and concomitant PGE2 and Col-I production by fibroblasts mediate the communication between the two cell types

We then investigated the SCC-derived proinflammatory cytokines implicated in the recruitment of fibroblasts and the subsequent interaction between SCCs and fibroblasts. By using a cytokine array kit, we identified markedly increased levels of various cytokines, most prominently IL1β, IL8/CXCL8, TGFβ1, and EGF, in H460/PcR cells compared with the control cells (Fig. 4A). Real-time PCR using gene-specific primers (Supplementary Table S2) validated transcriptional increases of these cytokines in all three SCC lines (Fig. 4B). Moreover, siRNA-mediated silencing of these cytokines significantly abrogated the ability of the SCCs to induce Wi38 and MRC5 cell migration (Fig. 4C; Supplementary Fig. S6). Hence, these soluble factors appeared to initiate the migration of the fibroblasts toward the SCCs.

Figure 4.

Identification of soluble factors mediating the communication between SCCs and fibroblasts and association of ATF6 with the secretion of soluble factors by SCCs. A, Top, schematic diagram of the experimental hypothesis. Bottom, the soluble factor levels that were significantly elevated in the CM from H460/PcR cells compared with the H460 cell-derived CM were determined by cytokine array. B, Real-time PCR analysis to determine the upregulation of IL1B, EGF, TGFB1, and CXCL8 expression in SCCs (H460/PcR, H1299/CsR, and H1299/PmR cells) compared with that in the corresponding parental cells. C, Abrogation of the migration of Wi38 fibroblasts mediated by knocking down the expression of IL1B, EGF, TGFB1, and CXCL8 in SCCs by single or combined siRNA transfections. D, Top, schematic diagram of the experimental hypothesis. Bottom, Wi38 cells were treated with CM obtained from parental cells or SCCs for 2 days. RNA was isolated, and real-time PCR was performed to analyze the expression of various factors associated with CAF phenotypes. E, ELISA to determine the PGE2 level in the CM derived from parental cells or SCCs CM-treated Wi38. F, Western blot analysis showing the upregulation of the signal transduction activated by PGE2 or Col-I in the SCCs cocultured with Wi38 fibroblasts. G, Upregulation of ATF6-mediated transcriptional activity in SCCs compared with the activity in ACCs was determined by a luciferase reporter assay using a reporter vector containing five repeats of the ATF6 DNA binding site (p5xATF6-GL3). H, Downregulation of IL1B, EGF, TGFB1, and CXCL8 expression in the H460/PcR cells transfected with the ATF6 siRNAs compared with that in the cells transfected with scrambled siRNAs (Scr) was determined by real-time PCR. I, Abrogation of the migration of Wi38 and MRC5 fibroblasts caused by coculture with the H460/PcR cells transfected with the ATF6 siRNAs. J, Attenuation of PTGS2, PTGES, PTGES2, and COL1A1 expression in WI38 cells by treatment with CMs from the H460/PcR cells transfected with the ATF6 siRNAs was determined by real-time PCR. K, Attenuation of tumor growth caused by intratumoral injection of liposome-encapsulated ATF6 siRNAs (n = 12, each group). L, The quantification of the positive cells for each marker shown as the number of positive cells per field of view (FOV) is depicted as a graph. Representative IHC images are shown in Supplementary Fig. S15. At least three tumors per group were analyzed for quantification of IHC analyses. For all panels, the bars represent the mean ± SD. *, P < 0.05; **, P < 0.01; ***, P < 0.001, as determined by a two-tailed Student t test.

Figure 4.

Identification of soluble factors mediating the communication between SCCs and fibroblasts and association of ATF6 with the secretion of soluble factors by SCCs. A, Top, schematic diagram of the experimental hypothesis. Bottom, the soluble factor levels that were significantly elevated in the CM from H460/PcR cells compared with the H460 cell-derived CM were determined by cytokine array. B, Real-time PCR analysis to determine the upregulation of IL1B, EGF, TGFB1, and CXCL8 expression in SCCs (H460/PcR, H1299/CsR, and H1299/PmR cells) compared with that in the corresponding parental cells. C, Abrogation of the migration of Wi38 fibroblasts mediated by knocking down the expression of IL1B, EGF, TGFB1, and CXCL8 in SCCs by single or combined siRNA transfections. D, Top, schematic diagram of the experimental hypothesis. Bottom, Wi38 cells were treated with CM obtained from parental cells or SCCs for 2 days. RNA was isolated, and real-time PCR was performed to analyze the expression of various factors associated with CAF phenotypes. E, ELISA to determine the PGE2 level in the CM derived from parental cells or SCCs CM-treated Wi38. F, Western blot analysis showing the upregulation of the signal transduction activated by PGE2 or Col-I in the SCCs cocultured with Wi38 fibroblasts. G, Upregulation of ATF6-mediated transcriptional activity in SCCs compared with the activity in ACCs was determined by a luciferase reporter assay using a reporter vector containing five repeats of the ATF6 DNA binding site (p5xATF6-GL3). H, Downregulation of IL1B, EGF, TGFB1, and CXCL8 expression in the H460/PcR cells transfected with the ATF6 siRNAs compared with that in the cells transfected with scrambled siRNAs (Scr) was determined by real-time PCR. I, Abrogation of the migration of Wi38 and MRC5 fibroblasts caused by coculture with the H460/PcR cells transfected with the ATF6 siRNAs. J, Attenuation of PTGS2, PTGES, PTGES2, and COL1A1 expression in WI38 cells by treatment with CMs from the H460/PcR cells transfected with the ATF6 siRNAs was determined by real-time PCR. K, Attenuation of tumor growth caused by intratumoral injection of liposome-encapsulated ATF6 siRNAs (n = 12, each group). L, The quantification of the positive cells for each marker shown as the number of positive cells per field of view (FOV) is depicted as a graph. Representative IHC images are shown in Supplementary Fig. S15. At least three tumors per group were analyzed for quantification of IHC analyses. For all panels, the bars represent the mean ± SD. *, P < 0.05; **, P < 0.01; ***, P < 0.001, as determined by a two-tailed Student t test.

Close modal

We next determined whether the communication with SCCs or ACCs allowed fibroblasts to obtain the phenotypes of CAFs. To this end, we analyzed the expression of several CAF-associated factors (25, 26, 30–32) in Wi38 fibroblasts exposed to the CM from ACCs or SCCs. Real-time PCR revealed that compared with those exposed to the CM derived from ACCs, the Wi38 fibroblasts exposed to the CM derived from three different SCCs consistently displayed elevated expression of COL1A1, PTGS2, PTGES, and PTGES2 (25, 30–32; Fig. 4D). Consistently, Wi38 cells incubated with the SCC-derived CM also revealed increased expression of COX2 and Col-I proteins (Supplementary Fig. S7). An ELISA analysis showed significant increases in the levels of PGE2 in the CM from the Wi38 cells that had been preincubated with the SCC (H460/PcR)-derived CM (Fig. 4E).

Because PGE2 and collagen transmit signals involved in cell proliferation, survival, and motility through EP receptors and integrins (11, 33, 34), we assessed the effects of communication between SCCs and CAFs on the downstream effectors of the Col-I- and PGE2-mediated signaling pathway, such as Src, FAK, and Akt (11, 33, 34). Wi38 cells incubated with the SCC-derived CM revealed increased levels of pSrc and pAkt (Supplementary Fig. S7). Notably, phosphorylation of Src, FAK, and Akt was also increased in the three SCCs upon coculture with Wi38 or MRC5 cells (Fig. 4F; Supplementary Fig. S8). IHC staining further revealed substantially elevated levels of COX2, Col-I, and pSrc in the xenograft tumors derived from the SCCs compared with those derived from the parental cells (Supplementary Fig. S9). Moreover, when the mRNA expression of surface receptors for PGE2 and Col-I (PTGER1, PTGER2, PTGER3, and PTGER4 for PGE2 and ITGB1 for Col-I, respectively; 11, 35) was ablated by siRNA transfection (Supplementary Fig. S10A), loss of PTGER1, PTGER3, and ITGB1 markedly suppressed the activation of FAK and Src (Supplementary Fig. S10B) and the Ki67-positive cell population (Supplementary Fig. S10C) in H460/PcR cells cocultured with Wi38 fibroblasts. Therefore, EP1 and EP3 receptors appeared to be implicated in the CAF-derived PGE2 signaling while β1 integrin was involved in the CAF-derived Col-I signaling. Collectively, these results suggested that communication between the SCCs and CAFs initiated by SCC-derived proinflammatory cytokines appeared to cause Col-I and PGE2 production in the TME, leading to activation of the integrin/Src- and PGE2-mediated signaling pathways in the SCCs and fibroblasts.

We next investigated the mechanism by which the chemoresistant sublines increased the transcription of the cytokines. Various cellular stresses, including chemotherapy, are known to activate the UPR signaling that has been implicated in the chemoresistance of solid tumors (15). During UPR, ATF6 translocates from the ER to the Golgi, where it is processed to its cleaved active form (cl-ATF6; ref. 36). cl-ATF6 translocates into the nucleus and induces UPR target genes via ER stress response element (36). Given the impact of ATF6 activity on cytokine production (37), we assessed the role of ATF6 activity in cytokine expression by the SCCs. Immunofluorescence staining and luciferase reporter assay revealed the consistently increased nuclear localization of ATF6 (Supplementary Fig. S11) and ATF6 transcriptional activity (Fig. 4G) in SCCs compared with corresponding parental cells. Pharmacologic inactivation of ATF6 similarly inhibited transcription of the cytokines and nuclear localization of ATF6 in H460/PcR cells (Supplementary Fig. S12). Consistently, H460/PcR cells with siRNA-mediated suppression of ATF6 displayed significantly decreased cytokine transcription (Fig. 4H) and reduced capacity to recruit Wi38 and MRC5 fibroblasts (Fig. 4I). In contrast, the siRNA-mediated ablation of ATF6 in H460 cells induced modest changes in their transcription of the SCC-associated cytokines and capacity for Wi38 recruitment (Supplementary Fig. S13). Wi38 and MRC5 fibroblasts exposed to the CM from the ATF6-silenced H460Pc/R cells displayed significantly lower levels of PTGS2, PTGES, PTGES2, and COL1A1 than those exposed to the CM from control H460Pc/R cells (Fig. 4J; Supplementary Fig. S14). We next determined the effects of ATF6 depletion on the growth of xenograft tumors of SCCs (H460/PcR). Notably, treatment with a liposome-encapsulated ATF6 siRNA effectively suppressed the outgrowth of these xenograft tumors (Fig. 4K). IHC analysis of xenograft tumors revealed ablation of ATF6 levels in siRNA-treated cells after 2 weeks growth in vivo. We also confirmed that the ATF6 siRNA treatment induced obvious decreases in fibroblast recruitment, expression of pSrc and PCNA in the tumors along with upregulation of cleaved caspase-3 levels (Fig. 4L; Supplementary Fig. S15). These results collectively suggest that the ATF6-mediated production of several proinflammatory cytokines in SCCs leads to the recruitment of fibroblasts in the TME, which in turn, triggers the growth of SCCs in vivo, ultimately leading to tumor progression.

Targeting Src and COX2 in combination prevents the slow-to-active proliferative switch in SCCs in vitro and recurrent SCC tumor growth in vivo

To link these findings to a clinical application, we attempted to identify clinically available therapeutic strategies to suppress the outgrowth of the SCCs. Both the collagen and PGE2 pathways have pleiotropic effects on cell survival, proliferation, migration, and invasion (38). Hence, we hypothesized that the Col-I and PGE2 produced via the interplay between CAFs and SCCs play a major role in the transition of SCCs into ACCs. Notably, the stimulation of SCCs with PGE2 or Col-I induced increases in proliferation, colony-forming capacity (Fig. 5A), and cyclin D1, CDK4, and PCNA expression (Fig. 5B). Costimulation with PGE2 and Col-I augmented these changes. We then determined whether suppression of the signaling pathways prevents the spontaneous proliferative outbreak of the SCCs. Treatment with dasatinib or celecoxib markedly attenuated cell proliferation and colony formation of H460/PcR cells under the coculture with Wi38 cells (Fig. 5C). Combined treatment with the two inhibitors exhibited more effective regulation of these events. Compared with untreated control or single treatments, the combined treatment resulted in the suppression of cyclin D1, CDK4, and PCNA expression H460/PcR cells under the coculture with Wi38 or MRC5 fibroblasts (Fig. 5D; Supplementary Fig. S16A). Furthermore, compared with the naïve SCCs incubated with the CM from cocultures of H460/PcR and Wi38 cells in the presence of vehicle or either agent alone, the naïve SCCs incubated with the CM derived from cocultures of H460/PcR and Wi38 cells in the presence of dasatinib and celecoxib revealed significantly decreased proliferation (Fig. 5E). Notably, the combined treatment induced apoptosis in the H460/PcR cells under the coculture with Wi38 cells or MRC5 fibroblasts as shown by increases in the cleavage of caspase-3 and PARP (Fig. 5F; Supplementary Fig. S16B). Together, soluble factors derived from the communication between SCCs and fibroblasts appeared to enable the SCCs to recover their proliferative capacity.

Figure 5.

Targeting Src or COX2 to inhibit the awakening of quiescent SCCs and recurrent tumor formation in vitro and in vivo. A, Increases in cell proliferation (left) and anchorage-independent colony formation in H460/PcR cells (right) caused by treatment with PGE2 or Col-I, alone or in combination, as determined by a cell counting assay (left) and anchorage-independent colony formation assay (right). B, Western blot analysis of cyclin D1 (CycD1), CDK4, and PCNA in H460/PcR cells treated with PGE2 or Col-I, alone or in combination. C, Suppression of cell proliferation (left) and colony formation (right) in the H460/PcR cells cocultured with Wi38 fibroblasts caused by treatment with dasatinib and celecoxib in combination, as determined by a cell counting assay (left) and anchorage-independent colony formation assay (right). D, Suppression of cell proliferation-related protein expression in the H460/PcR cells cocultured with Wi38 fibroblasts caused by treatment with dasatinib and celecoxib in combination, as determined by Western blot analysis. E, Changes in cell proliferation of SCCs (H460Pc/R) after incubation with the CM from the SCC (H460/PcR)-Wi38 coculture treated with dasatinib, celecoxib, or their combination, as determined by cell counting assay. F, Western blot analysis of cleaved PARP (Cl-PARP) and cleaved caspase-3 (Cl-Cas3) in H460/PcR cells treated with dasatinib, celecoxib, or their combination. G–J, Suppression of the growth of residual H460 xenograft tumors after chemotherapy (n = 12, each group; G and H) and chemoresistant NSCLC PDX tumors (n = 12, each group; I and J) caused by treatment with dasatinib (40 mg/kg), celecoxib (100 mg/kg), or their combination. G and I, Changes in tumor growth at each day after drug treatment are depicted as graphs. H and J, Top, IHC analyses showing a decrease in pSrc and PCNA expression and an elevation in cleaved caspase-3 expression in the residual H460 (H) or PDX tumors (J) treated with dasatinib, celecoxib, or their combination. Bottom, the quantification of the positive cells for each marker shown as the number of positive cells per field of view (FOV) is depicted as a graph. At least three tumors per group were analyzed for quantification of IHC analyses. Scale bar, 50 μm. Scale bar (insets), 10 μm. For all panels, the bars represent the mean ± SD. *, P < 0.05; ***, P < 0.001, as determined by a two-tailed Student t test or one-way ANOVA (G and I). H&E, hematoxylin and eosin.

Figure 5.

Targeting Src or COX2 to inhibit the awakening of quiescent SCCs and recurrent tumor formation in vitro and in vivo. A, Increases in cell proliferation (left) and anchorage-independent colony formation in H460/PcR cells (right) caused by treatment with PGE2 or Col-I, alone or in combination, as determined by a cell counting assay (left) and anchorage-independent colony formation assay (right). B, Western blot analysis of cyclin D1 (CycD1), CDK4, and PCNA in H460/PcR cells treated with PGE2 or Col-I, alone or in combination. C, Suppression of cell proliferation (left) and colony formation (right) in the H460/PcR cells cocultured with Wi38 fibroblasts caused by treatment with dasatinib and celecoxib in combination, as determined by a cell counting assay (left) and anchorage-independent colony formation assay (right). D, Suppression of cell proliferation-related protein expression in the H460/PcR cells cocultured with Wi38 fibroblasts caused by treatment with dasatinib and celecoxib in combination, as determined by Western blot analysis. E, Changes in cell proliferation of SCCs (H460Pc/R) after incubation with the CM from the SCC (H460/PcR)-Wi38 coculture treated with dasatinib, celecoxib, or their combination, as determined by cell counting assay. F, Western blot analysis of cleaved PARP (Cl-PARP) and cleaved caspase-3 (Cl-Cas3) in H460/PcR cells treated with dasatinib, celecoxib, or their combination. G–J, Suppression of the growth of residual H460 xenograft tumors after chemotherapy (n = 12, each group; G and H) and chemoresistant NSCLC PDX tumors (n = 12, each group; I and J) caused by treatment with dasatinib (40 mg/kg), celecoxib (100 mg/kg), or their combination. G and I, Changes in tumor growth at each day after drug treatment are depicted as graphs. H and J, Top, IHC analyses showing a decrease in pSrc and PCNA expression and an elevation in cleaved caspase-3 expression in the residual H460 (H) or PDX tumors (J) treated with dasatinib, celecoxib, or their combination. Bottom, the quantification of the positive cells for each marker shown as the number of positive cells per field of view (FOV) is depicted as a graph. At least three tumors per group were analyzed for quantification of IHC analyses. Scale bar, 50 μm. Scale bar (insets), 10 μm. For all panels, the bars represent the mean ± SD. *, P < 0.05; ***, P < 0.001, as determined by a two-tailed Student t test or one-way ANOVA (G and I). H&E, hematoxylin and eosin.

Close modal

This combination treatment was also evaluated in vivo in mice bearing residual H460 xenograft tumors after chemotherapy. To this end, H460 xenograft tumors were subjected to three cycles of a combinatorial chemotherapy, and the residual tumors were treated with dasatinib, celecoxib, or both for approximately 2 weeks after the completion of the chemotherapy regimen (Fig. 5G). IHC analysis revealed that dasatinib treatment alone was able to clear off pSrc expression in the tumors better than celecoxib and the combined treatment was less effective in Src blockade than desatinib alone (Fig. 5H). These findings indicated potential mechanisms that mediate Src phosphorylation in response to the combined treatment. Nevertheless, the growth of the residual H460 xenograft tumors markedly decreased after the combined treatment with dasatinib and celecoxib (Fig. 5G). Similarly, the combination of the two inhibitors significantly suppressed the growth of H460/PcR xenograft tumors (Supplementary Fig. S17).

To link these findings to a clinical application, we determined the efficacy of the combined therapy using two PDX tumors of patients with NSCLC who progressed after platinum-based chemotherapy. YHIM-1010 (PDX #4) and -2012 (PDX #5) were originated from adenocarcinoma resistant to prior paclitaxel/carboplatin chemotherapy and squamous carcinoma progressed after gemcitabine/carboplatin treatment, respectively. Notably, treatment with dasatinib and celecoxib in combination more prominently and synergistically suppressed the outgrowth of the PDX tumors than each dasatinib or celecoxib monotherapy (Fig. 5I; Supplementary Table S3; Supplementary Fig. S18). IHC analysis of the tumors revealed that compared with the other treatments, the combined treatment with the two inhibitors induced obvious decreases in pSrc and PCNA expression with a concomitant increase in cleaved caspase-3 expression (Fig. 5J; Supplementary Fig. S18 and S19). The combination therapy did not result in any more weight loss than that seen with the single inhibitors or vehicle (Supplementary Fig. S20). These findings suggest that PGE2- and collagen-mediated signaling pathways support the survival of SCCs in an in vivo microenvironment, ultimately leading to tumor progression. Furthermore, targeting the two pathways may reduce the viability of SCCs and thus prevent their spontaneous outgrowth in vivo.

Association of Src or type I collagen with poor clinical outcomes in patients with NSCLC

To validate the clinical relevance of our findings, we analyzed the PDX tumors that were progressed upon completion of chemotherapy (Fig. 2). IHC analyses of consecutive slides revealed the upregulation of pSrc and COX2 expression in the progressed PDX tumors compared with the untreated control tumors (Fig. 6A). We also confirmed the concordant elevations in pSrc, COX2, and FSP1 expression in FFPE tumor tissue samples collected from 9 advanced patients with NSCLC who failed to prior platinum-based chemotherapy compared with those in baseline tumor tissue without prior chemotherapy (Fig. 6B). We further determined whether the expression of Src, COX2, and Col-I was correlated with histologic subtype or recurrence in patients with NSCLC using publicly available datasets deposited in the GEO database. In two independent GEO datasets (GSE37745 and GSE41271), neither SRC nor COL1A1 expression exhibited significant differences between histologic subtypes of NSCLC, whereas the level of PTGS2 expression was significantly lower in lung squamous cell carcinoma (SCC) than in lung adenocarcinoma (ADC; Fig. 6C, Supplementary Fig. S21A). Interestingly, tumors from patients with NSCLC who experienced recurrence were likely to express significantly higher levels of SRC or COL1A1 mRNA, and SRC or COL1A1 expression was significantly elevated in the tumors from patients with recurrent lung ADC compared with those from patients without recurrence (Fig. 6D). However, there was no significant association between PTGS2 expression and recurrence in these datasets (Fig. 6D; Supplementary Fig. S21B). Notably, SRC or COL1A1 expression was positively correlated with the expression of cell proliferation–related genes, such as CCNE1, CDK2, or MIK67 (Fig. 6E). These results suggest the potential of SRC and COL1A1 as biomarkers for the prediction of tumor progression in patients with NSCLC, especially those with ADC.

Figure 6.

The association of SRC, COL1A1, or PTGS2 expression with recurrence in patients with NSCLC. A, Left, IHC analyses showing the upregulation of COX2 and pSrc expression in the PDX tumors (Fig. 2F) exposed to chemotherapy. Right, the quantification of the positive cells for each marker per field of view (FOV) is depicted as a graph. Scale bar, 50 μm. Scale bar (insets), 10 μm. B, Top, representative IHC images showing the increase in the accumulation of FSP1+ fibroblasts and expression of pSrc and COX2 in NSCLC patient tumor samples collected after chemotherapy compared with those in the paired tumor samples collected prior to chemotherapy. Bottom, quantification of IHC analyses. Scale bar, 50 μm. Scale bar (insets), 10 μm. C–E, The in silico analyses to determine the associations of SRC, COL1A1, or PTGS2 expression with NSCLC histology (C), recurrence in patients with NSCLC (D), and the correlation between SRC or COL1A1 expression and the gene expression related to cell proliferation (E) using publicly available datasets deposited in the Gene Expression Omnibus database (GSE37745). In the correlation analysis, Pearson or Spearman correlation coefficients were determined on the basis of Gaussian distribution of the data. F, The proposed model for awakening chemoresistant cells from quiescence by soluble factor-mediated communication with fibroblasts. For all panels, the bars represent the mean ± SD. *, P < 0.05; **, P < 0.01; ***, P < 0.001, as determined by a Wilcoxon signed-rank test (B), one-way ANOVA (C), a two-tailed Student t test or a two-tailed Mann–Whitney test based on Gaussian distribution of the data (D) H&E, hematoxylin and eosin.

Figure 6.

The association of SRC, COL1A1, or PTGS2 expression with recurrence in patients with NSCLC. A, Left, IHC analyses showing the upregulation of COX2 and pSrc expression in the PDX tumors (Fig. 2F) exposed to chemotherapy. Right, the quantification of the positive cells for each marker per field of view (FOV) is depicted as a graph. Scale bar, 50 μm. Scale bar (insets), 10 μm. B, Top, representative IHC images showing the increase in the accumulation of FSP1+ fibroblasts and expression of pSrc and COX2 in NSCLC patient tumor samples collected after chemotherapy compared with those in the paired tumor samples collected prior to chemotherapy. Bottom, quantification of IHC analyses. Scale bar, 50 μm. Scale bar (insets), 10 μm. C–E, The in silico analyses to determine the associations of SRC, COL1A1, or PTGS2 expression with NSCLC histology (C), recurrence in patients with NSCLC (D), and the correlation between SRC or COL1A1 expression and the gene expression related to cell proliferation (E) using publicly available datasets deposited in the Gene Expression Omnibus database (GSE37745). In the correlation analysis, Pearson or Spearman correlation coefficients were determined on the basis of Gaussian distribution of the data. F, The proposed model for awakening chemoresistant cells from quiescence by soluble factor-mediated communication with fibroblasts. For all panels, the bars represent the mean ± SD. *, P < 0.05; **, P < 0.01; ***, P < 0.001, as determined by a Wilcoxon signed-rank test (B), one-way ANOVA (C), a two-tailed Student t test or a two-tailed Mann–Whitney test based on Gaussian distribution of the data (D) H&E, hematoxylin and eosin.

Close modal

It has been long-lasting questions how quiescent cancer cells in residual tumors after chemotherapy are triggered to induce tumor progression. This study demonstrated that SCCs in residual tumors after chemotherapy secrete various proinflammatory cytokines, including IL1β, TGFβ1, CXCL8, and EGF, which chemoattract stromal cells, most notably fibroblasts, in the TME (Fig. 6FI). The proinflammatory cytokines cooperatively acted on CAFs in a paracrine manner to elevate COX2 and collagen I expression in the fibroblasts, leading to collagen deposition and PGE2 production in the TME (Fig. 6F, II), which activate integrin/Src- and EP receptor-mediated signaling pathways, resulting in the outgrowth of SCCs to initiate tumor progression (Fig. 6F, III). We have further demonstrated that a combination therapy using SFK and COX2 inhibitors reduces the viability of SCCs induced by their bidirectional, reciprocal interaction with fibroblasts, causing apoptotic cell death in the SCCs and thus preventing tumor progression (Fig. 6F, IV). Collectively, our results highlight the potential of the PGE2- and Src-mediated signaling pathways as effective targets to prevent the transition of SCCs into ACCs and tumor progression after chemotherapy.

Despite the obvious efficacy of chemotherapy in shrinking some primary tumors, tumor progression is a major hurdle for chemotherapy treatments (6, 7). A small subpopulation of SCCs with quiescent phenotypes in residual tumor tissue is likely to survive initial chemotherapy and repropagate disease at local or metastatic locations (7, 39, 40). However, the contribution of the SCCs in NSCLC to chemoresistance and tumor progression has not been clearly demonstrated, mainly due to the lack of clinically relevant experimental model systems. In this study, we established SCCs from human NSCLC or mouse lung cancer cell lines as preclinical experimental models to gain insights into the key signaling mechanisms that trigger the slow-to-active proliferative switch in SCCs and tumor relapse.

Our analyses of the established SCCs yielded surprisingly robust results. While these SCCs revealed a slowly proliferating state in vitro, but consistently generated tumors in mice. Moreover, once the SCC xenograft tumors developed, they showed rapid growth and chemoresistance. Considering the increasingly recognized role of the TME in cancer progression (12, 41), we hypothesized that the SCCs hold quiescence-like phenotypes, characterized by a nonproliferative or possibly slowly proliferating state with reversible growth arrest (27), but induce specific changes within the TME to establish a growth-promoting niche. In support of this idea, the SCCs displayed the ability to recruit fibroblasts in the TME. Given the ability of CAFs to change the extracellular matrix topology as a main component of tumor progression and anticancer drug resistance (42, 43), it was highly likely that the SCCs recruited fibroblasts in the TME to evade tumor dormancy and promote chemoresistance. Indeed, the SCCs appeared to enter the active cell cycle after communicating with fibroblasts. Consistent with the in vitro results, a small number of fibroblasts significantly accelerated the growth of SCC xenograft tumors in an in vivo study of mixed tumors in mice. Essential data was acquired from the dual-flank experiment, wherein the interaction between the SCCs and CAFs accelerated the growth of secondary tumors at a distant site on the contralateral flank of the mice. In this experiment, despite the “disadvantage” that [H460/PcR] or [H460/PcR][Wi38] feeder tumors grew at a slower rate than the tumors derived from H460 cells or H460 and Wi38 cells, respectively, the secondary tumors in the H460/PcR and H460/PcR plus Wi38 feeder tumor groups eventually grew larger than those in the corresponding control groups. Although the growth-promoting effects of SCCs may not be mediated by a simple one-way process, our results support the concept that soluble factors generated by the interaction between SCCs and CAFs, even in a limited tumor burden context, exert both local and systemic effects on tumor growth.

Our subsequent experiments revealed that the established SCCs generated via ATF6-dependent transcriptional upregulation of several cytokines chemoattract fibroblasts, and ATF6 antagonism significantly delayed the SCC-derived recurrent xenograft tumors. The UPR, an adaptive cellular reaction (44) to various types of ER stress, stimulates specific ER membrane proteins, including ATF6α, IRE1α, and PERK (15). It has been proposed that tumor cells coopt ER stress signaling to adapt to and survive environmental stresses, including chemotherapy (15). Others have reported that persistent ATF6 activation promotes survival in solid tumor cells, predominantly those in a quiescent phase, by circumventing the need for growth factors and Akt signaling (45). These findings and our current data suggest that SCCs rely on ATF6 activity in two ways: one mechanism that will protect them from microenvironment stressors, including chemotherapy, and another that allows them to resume growth later by mediating the communication with CAFs in the TME.

Importantly, transcriptional rewiring appeared to occur not only in the SCCs but also in the CAFs during their communication. After a coculture with SCCs or incubation with CM from SCCs, Wi38 fibroblasts revealed activation of integrin/Src and PGE2 signaling through transcriptional upregulation of COX2 and Col-I. Cancer-derived soluble factors were shown to stimulate PGE2 production in fibroblasts and mesenchymal stem cells (46, 47). Previous studies have suggested that a Col-I-enriched fibrotic microenvironment may provide a fertile ‘soil’ for the dormant-to-proliferative switch in cancer cells (12). Several clinical studies have shown that enriched stromal Col-I expression is correlated with cancer recurrence or reduced survival in various cancers (12, 48). Patients with enriched stromal Col-I expression display a higher risk of developing local recurrence after mastectomy or radiotherapy (49). Furthermore, patients with fibrotic foci (FF) have an increased risk for disease recurrence and lymph node and bone metastasis (50). The Col-I–mediated activation of integrin, binding of PGE2 to its receptor, and subsequent activation of the downstream effectors, such as Src, Akt, and ERK1/2, are known to stimulate proliferation and migration in various types of human cancer cells, including NSCLC cells (51, 52). Consistently, inhibiting Src and COX2 with clinically available pharmacologic inhibitors significantly prevented the recovery of proliferative activities in SCCs in vitro and tumor progression in vivo.

Our findings also have important clinical implications. Increased Col-I, COX2, and pSrc expression was observed in residual PDX and patient lung tumor tissue samples that progressed despite chemotherapy. Analysis of publicly available datasets demonstrated the positive correlation between SRC or COL1A1 expression and cell cycle–related gene expression and the potential of Src or Col-I as predictive biomarkers of tumor relapse in patients with NSCLC, especially those with lung ADC, which is in line with previous reports studying patients with cervical cancer, oral cancer, or malignant astrocytoma (53–55). In addition, our in vitro and in vivo results suggest that utilizing SFK and COX2 inhibitors in combination could be a novel strategy to achieve better outcome in patients with NSCLC after chemotherapy, especially those treated with platinum-based chemotherapy. Taken together, these results support the concept that the interplay between SCCs and CAFs produces a proinflammatory niche, which ignites the slow-to-active proliferative switch in SCCs, leading to the outgrowth of SCCs as recurrent tumors.

Our results provide new insights into the role of tumor interactions with CAFs in tumor relapse and chemoresistance and identify the Col-I- and PGE2-mediated signaling pathways as potential therapeutic targets to suppress tumor progression. However, some tumors showed statistically significant, but modest anti-tumor responses to the proposed targeted therapies. Moreover, correlations of fibroblast marker expression in patients with and without prior chemotherapy were statistically significant but quantitatively modest. These results may be mostly due to the heterogeneity of cancer cells and their microenvironment. In the cases, immune-checkpoint inhibitors (ICI), such as those targeting the programmed death receptor 1 (PD-1), programmed death-ligand 1 (PD-L1), and CTL-associated protein 4 (CTLA-4), could be an excellent therapeutic option. Indeed, ICIs have shown extraordinary efficacy in preclinical studies and clinical trials (56). On the basis of the promising results, the FDA has approved ICIs for melanoma, kidney cancer, NSCLC, and other malignancies (57). However, it has turned out that only a minority of patients show durable responses, and drug resistance and fatal adverse effects have been recently reported (58), indicating the needs to develop comprehensive ICI-based treatment strategies. The cytokine-mediated tumor-TME interaction is closely associated with resistance to ICIs (59). Therefore, combined therapeutic strategies with ICIs and anti-cytokine drugs could provide a better way of blocking tumor progression. Indeed, several preclinical studies and clinical trials combining anti-cytokine drugs with ICIs, such as cotargeting immune checkpoint protein and TGFβ or IL2 variant, are ongoing to overcome the limitations (60, 61). We are planning to investigate the efficacy of combining ICIs with ATF6-targeting drugs.

In summary, we have shown for the first time, to our knowledge, how residual chemoresistant SCCs make changes in stromal fibroblasts to establish a growth-promoting TME for reactivating SCCs or the proliferative capacity of quiescent tumor cells, thus triggering disease progression. Our results also show that abrogating the growth-promoting activities in the TME caused by the interplay between latent cancer cells and CAFs through PGE2- and collagen-mediated signaling pathways may provide an innovative opportunity to prevent disease progression after chemotherapy.

No potential conflicts of interest were disclosed.

Conception and design: H.R. Kim, H.-Y. Lee

Development of methodology: H.-Y. Lee

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): J. Cho, H.-J. Lee, S.J. Hwang, H.-Y. Min, H.N. Kang, S.Y. Hyun, J.Y. Sim, Y.-A. Suh, S. Hong, Y.K. Shin, H.R. Kim, H.-Y. Lee

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): J. Cho, H.-Y. Min, S.Y. Hyun, H.-J. Jang, Y.-A. Suh, H.R. Kim, H.-Y. Lee

Writing, review, and/or revision of the manuscript: J. Cho, H.-J. Lee, H.-Y. Min, H.R. Kim, H.-Y. Lee

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): H.N. Kang, A.Y. Park, H.J. Lee, Y.-A. Suh, Y.K. Shin, H.-Y. Lee

Study supervision: H.-Y. Lee

This work was supported by a grant from the National Research Foundation of Korea (NRF), the Ministry of Science and ICT (MSIT), Republic of Korea (no. NRF-2016R1A3B1908631).

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.

1.
Jung
KW
,
Won
YJ
,
Kong
HJ
,
Lee
ES
Community of population-based regional cancer R. cancer statistics in Korea: incidence, mortality, survival, and prevalence in 2015
.
Cancer Res Treat
2018
;
50
:
303
16
.
2.
Torre
LA
,
Bray
F
,
Siegel
RL
,
Ferlay
J
,
Lortet-Tieulent
J
,
Jemal
A
. 
Global cancer statistics, 2012
.
CA Cancer J Clin
2015
;
65
:
87
108
.
3.
Artal Cortes
A
,
Calera Urquizu
L
,
Hernando Cubero
J
. 
Adjuvant chemotherapy in non-small cell lung cancer: state-of-the-art
.
Transl Lung Cancer Res
2015
;
4
:
191
7
.
4.
Zarogoulidis
K
,
Zarogoulidis
P
,
Darwiche
K
,
Boutsikou
E
,
Machairiotis
N
,
Tsakiridis
K
, et al
Treatment of non-small cell lung cancer (NSCLC)
.
J Thorac Dis
2013
;
5 Suppl
4
:
S389
96
.
5.
Tsvetkova
E
,
Goss
GD
. 
Drug resistance and its significance for treatment decisions in non-small-cell lung cancer
.
Curr Oncol
2012
;
19
:
S45
51
.
6.
Aguirre-Ghiso
JA
. 
Models, mechanisms and clinical evidence for cancer dormancy
.
Nat Rev Cancer
2007
;
7
:
834
46
.
7.
Yeh
AC
,
Ramaswamy
S
. 
Mechanisms of cancer cell dormancy–another hallmark of cancer?
Cancer Res
2015
;
75
:
5014
22
.
8.
Cole
SW
. 
Chronic inflammation and breast cancer recurrence
.
J Clin Oncol
2009
;
27
:
3418
9
.
9.
Slice
LW
,
Chiu
T
,
Rozengurt
E
. 
Angiotensin II and epidermal growth factor induce cyclooxygenase-2 expression in intestinal epithelial cells through small GTPases using distinct signaling pathways
. 
2005
;
280
:
1582
93
.
10.
Greenhough
A
,
Smartt
HJM
,
Moore
AE
,
Roberts
HR
,
Williams
AC
,
Paraskeva
C
, et al
The COX-2/PGE 2 pathway: key roles in the hallmarks of cancer and adaptation to the tumour microenvironment
.
Carcinogenesis
2009
;
30
:
377
86
.
11.
O'Callaghan
G
,
Houston
A
. 
Prostaglandin E2 and the EP receptors in malignancy: possible therapeutic targets?
Br J Pharmacol
2015
;
172
:
5239
50
.
12.
Barkan
D
,
El Touny
LH
,
Michalowski
AM
,
Smith
JA
,
Chu
I
,
Davis
AS
, et al
Metastatic growth from dormant cells induced by a col-I-enriched fibrotic environment
.
Cancer Res
2010
;
70
:
5706
16
.
13.
El Touny
LH
,
Vieira
A
,
Mendoza
A
,
Khanna
C
,
Hoenerhoff
MJ
,
Green
JE
. 
Combined SFK/MEK inhibition prevents metastatic outgrowth of dormant tumor cells
.
J Clin Invest
2014
;
124
:
156
68
.
14.
Chevet
E
,
Hetz
C
,
Samali
A
. 
Endoplasmic reticulum stress-activated cell reprogramming in oncogenesis
.
Cancer Discov
2015
;
5
:
586
97
.
15.
Wang
M
,
Kaufman
RJ
. 
The impact of the endoplasmic reticulum protein-folding environment on cancer development
.
Nat Rev Cancer
2014
;
14
:
581
97
.
16.
Alkema
NG
,
Tomar
T
,
Duiker
EW
,
Jan Meersma
G
,
Klip
H
,
van der Zee
AG
, et al
Biobanking of patient and patient-derived xenograft ovarian tumour tissue: efficient preservation with low and high fetal calf serum based methods
.
Sci Rep
2015
;
5
:
14495
.
17.
Botling
J
,
Edlund
K
,
Lohr
M
,
Hellwig
B
,
Holmberg
L
,
Lambe
M
, et al
Biomarker discovery in non-small cell lung cancer: integrating gene expression profiling, meta-analysis, and tissue microarray validation
.
Clin Cancer Res
2013
;
19
:
194
204
.
18.
Riquelme
E
,
Suraokar
M
,
Behrens
C
,
Lin
HY
,
Girard
L
,
Nilsson
MB
, et al
VEGF/VEGFR-2 upregulates EZH2 expression in lung adenocarcinoma cells and EZH2 depletion enhances the response to platinum-based and VEGFR-2-targeted therapy
.
Clin Cancer Res
2014
;
20
:
3849
61
.
19.
Faul
F
,
Erdfelder
E
,
Buchner
A
,
Lang
AG
. 
Statistical power analyses using G*Power 3.1: tests for correlation and regression analyses
.
Behav Res Methods
2009
;
41
:
1149
60
.
20.
Lin
WC
,
Rajbhandari
N
,
Wagner
KU
. 
Cancer cell dormancy in novel mouse models for reversible pancreatic cancer: a lingering challenge in the development of targeted therapies
.
Cancer Res
2014
;
74
:
2138
43
.
21.
Roesch
A
,
Fukunaga-Kalabis
M
,
Schmidt
EC
,
Zabierowski
SE
,
Brafford
PA
,
Vultur
A
, et al
A temporarily distinct subpopulation of slow-cycling melanoma cells is required for continuous tumor growth
.
Cell
2010
;
141
:
583
94
.
22.
Moore
N
,
Houghton
J
,
Lyle
S
. 
Slow-cycling therapy-resistant cancer cells
.
Stem Cells Dev
2012
;
21
:
1822
30
.
23.
Eddaoudi
A
,
Canning
SL
,
Kato
I
. 
Flow cytometric detection of G0 in live cells by hoechst 33342 and pyronin Y staining
.
Methods Mol Biol
2018
;
1686
:
49
57
.
24.
Quail
DF
,
Joyce
JA
. 
Microenvironmental regulation of tumor progression and metastasis
.
Nat Med
2013
;
19
:
1423
37
.
25.
Gascard
P
,
Tlsty
TD
. 
Carcinoma-associated fibroblasts: orchestrating the composition of malignancy
.
Genes Dev
2016
;
30
:
1002
19
.
26.
Ohlund
D
,
Elyada
E
,
Tuveson
D
. 
Fibroblast heterogeneity in the cancer wound
.
J Exp Med
2014
;
211
:
1503
23
.
27.
Li
LT
,
Jiang
G
,
Chen
Q
,
Zheng
JN
. 
Ki67 is a promising molecular target in the diagnosis of cancer (review)
.
Mol Med Rep
2015
;
11
:
1566
72
.
28.
Borowicz
S
,
Van Scoyk
M
,
Avasarala
S
,
Karuppusamy Rathinam
MK
,
Tauler
J
,
Bikkavilli
RK
, et al
The soft agar colony formation assay
.
J Vis Exp
2014
:
e51998
.
29.
Gerdes
HH
,
Pepperkok
R
. 
Cell-to-cell communication: current views and future perspectives
.
Cell Tissue Res
2013
;
352
:
1
3
.
30.
Inoue
K-i
,
Kishimoto
S
,
Akimoto
K
,
Sakuma
M
,
Toyoda
S
,
Inoue
T
, et al
Cancer-associated fibroblasts show heterogeneous gene expression and induce vascular endothelial growth factor A (VEGFA) in response to environmental stimuli
.
Ann Gastroenterol Surg
2019
;
3
:
416
25
.
31.
Kock
A
,
Larsson
K
,
Bergqvist
F
,
Eissler
N
,
Elfman
LHM
,
Raouf
J
, et al
Inhibition of microsomal prostaglandin E synthase-1 in cancer-associated fibroblasts suppresses neuroblastoma tumor growth
.
EBioMedicine
2018
;
32
:
84
92
.
32.
Rozenchan
PB
,
Carraro
DM
,
Brentani
H
,
Mota
LDdC
,
Bastos
EP
,
Ferreira
ENe
, et al
Reciprocal changes in gene expression profiles of cocultured breast epithelial cells and primary fibroblasts
.
Int J Cancer
2009
;
125
:
2767
77
.
33.
Dajani
OF
,
Meisdalen
K
,
Guren
TK
,
Aasrum
M
,
Tveteraas
IH
,
Lilleby
P
, et al
Prostaglandin E2 upregulates EGF-stimulated signaling in mitogenic pathways involving Akt and ERK in hepatocytes
.
J Cell Physiol
2008
;
214
:
371
80
.
34.
Tsai
KS
,
Kao
SY
,
Wang
CY
,
Wang
YJ
,
Wang
JP
,
Hung
SC
. 
Type I collagen promotes proliferation and osteogenesis of human mesenchymal stem cells via activation of ERK and Akt pathways
.
J Biomed Mater Res A
2010
;
94
:
673
82
.
35.
Jokinen
J
,
Dadu
E
,
Nykvist
P
,
Kapyla
J
,
White
DJ
,
Ivaska
J
, et al
Integrin-mediated cell adhesion to type I collagen fibrils
.
J Biol Chem
2004
;
279
:
31956
63
.
36.
Wiseman
RL
,
Haynes
CM
,
Ron
D
. 
SnapShot: the unfolded protein response
.
Cell
2010
;
140
:
590
e2
.
37.
Rao
J
,
Yue
S
,
Fu
Y
,
Zhu
J
,
Wang
X
,
Busuttil
RW
, et al
ATF6 mediates a pro-inflammatory synergy between ER stress and TLR activation in the pathogenesis of liver ischemia-reperfusion injury
.
Am J Transplant
2014
;
14
:
1552
61
.
38.
Zhang
S
,
Yu
D
. 
Targeting Src family kinases in anti-cancer therapies: turning promise into triumph
.
Trends Pharmacol Sci
2012
;
33
:
122
8
.
39.
Froudarakis
ME
,
Briasoulis
E
. 
Advanced non-small cell lung cancer: on relapse rechallenge the tumor, not the patient
.
BMC Res Notes
2010
;
3
:
195
.
40.
Valdes
M
,
Nicholas
G
,
Goss
GD
,
Wheatley-Price
P
. 
Chemotherapy in recurrent advanced non-small-cell lung cancer after adjuvant chemotherapy
.
Curr Oncol
2016
;
23
:
386
90
.
41.
Hu
M
,
Polyak
K
. 
Microenvironmental regulation of cancer development
.
Curr Opin Genet Dev
2008
;
18
:
27
34
.
42.
Hazlehurst
LA
,
Landowski
TH
,
Dalton
WS
. 
Role of the tumor microenvironment in mediating de novo resistance to drugs and physiological mediators of cell death
.
Oncogene
2003
;
22
:
7396
402
.
43.
Kharaishvili
G
,
Simkova
D
,
Bouchalova
K
,
Gachechiladze
M
,
Narsia
N
,
Bouchal
J
. 
The role of cancer-associated fibroblasts, solid stress and other microenvironmental factors in tumor progression and therapy resistance
.
Cancer Cell Int
2014
;
14
:
41
.
44.
Schonthal
AH
. 
Endoplasmic reticulum stress: its role in disease and novel prospects for therapy
.
Scientifica (Cairo)
2012
;
2012
:
857516
.
45.
Schewe
DM
,
Aguirre-Ghiso
JA
. 
ATF6α-Rheb-mTOR signaling promotes survival of dormant tumor cells in vivo
.
Proc Natl Acad Sci
2008
;
105
:
10519
24
.
46.
Alcolea
S
,
Anton
R
,
Camacho
M
,
Soler
M
,
Alfranca
A
,
Aviles-Jurado
FX
, et al
Interaction between head and neck squamous cell carcinoma cells and fibroblasts in the biosynthesis of PGE2
.
J Lipid Res
2012
;
53
:
630
42
.
47.
Li
HJ
,
Reinhardt
F
,
Herschman
HR
,
Weinberg
RA
. 
Cancer-stimulated mesenchymal stem cells create a carcinoma stem cell niche via prostaglandin E2 signaling
.
Cancer Discov
2012
;
2
:
840
55
.
48.
Zhou
ZH
,
Ji
CD
,
Xiao
HL
,
Zhao
HB
,
Cui
YH
,
Bian
XW
. 
Reorganized collagen in the tumor microenvironment of gastric cancer and its association with prognosis
.
J Cancer
2017
;
8
:
1466
76
.
49.
Park
CC
,
Rembert
J
,
Chew
K
,
Moore
D
,
Kerlikowske
K
. 
High mammographic breast density is independent predictor of local but not distant recurrence after lumpectomy and radiotherapy for invasive breast cancer
.
Int J Radiat Oncol Biol Phys
2009
;
73
:
75
9
.
50.
Hasebe
T
,
Sasaki
S
,
Imoto
S
,
Mukai
K
,
Yokose
T
,
Ochiai
A
. 
Prognostic significance of fibrotic focus in invasive ductal carcinoma of the breast: a prospective observational study
.
Mod Pathol
2002
;
15
:
502
16
.
51.
Bazzani
L
,
Donnini
S
,
Finetti
F
,
Christofori
G
,
Ziche
M
. 
PGE2/EP3/SRC signaling induces EGFR nuclear translocation and growth through EGFR ligands release in lung adenocarcinoma cells
.
Oncotarget
2017
;
8
:
31270
87
.
52.
Kim
JI
,
Lakshmikanthan
V
,
Frilot
N
,
Daaka
Y
. 
Prostaglandin E2 promotes lung cancer cell migration via EP4-betaArrestin1-c-Src signalsome
.
Mol Cancer Res
2010
;
8
:
569
77
.
53.
Hou
T
,
Xiao
J
,
Zhang
H
,
Gu
H
,
Feng
Y
,
Li
J
. 
Phosphorylated c-Src is a novel predictor for recurrence in cervical squamous cell cancer patients
.
Int J Clin Exp Pathol
2013
;
6
:
1121
7
.
54.
Cheng
SJ
,
Kok
SH
,
Lee
JJ
,
Yen-Ping Kuo
M
,
Cheng
SL
,
Huang
YL
, et al
Significant association of SRC protein expression with the progression, recurrence, and prognosis of oral squamous cell carcinoma in Taiwan
.
Head Neck
2012
;
34
:
1340
5
.
55.
Sun
S
,
Wang
Y
,
Wu
Y
,
Gao
Y
,
Li
Q
,
Abdulrahman
AA
, et al
Identification of COL1A1 as an invasionrelated gene in malignant astrocytoma
.
Int J Oncol
2018
;
53
:
2542
54
.
56.
Hargadon
KM
,
Johnson
CE
,
Williams
CJ
. 
Immune checkpoint blockade therapy for cancer: an overview of FDA-approved immune checkpoint inhibitors
.
Int Immunopharmacol
2018
;
62
:
29
39
.
57.
Chang
X
,
Lu
X
,
Guo
J
,
Teng
GJ
. 
Interventional therapy combined with immune checkpoint inhibitors: Emerging opportunities for cancer treatment in the era of immunotherapy
.
Cancer Treat Rev
2019
;
74
:
49
60
.
58.
Zhao
B
,
Zhao
H
,
Zhao
J
. 
Serious adverse events and fatal adverse events associated with nivolumab treatment in cancer patients: Nivolumab-related serious/fatal adverse events
.
J Immunother Cancer
2018
;
6
:
101
.
59.
Fares
CM
,
Allen
EMV
,
Drake
CG
,
Allison
JP
,
Hu-Lieskovan
S
. 
Mechanisms of resistance to immune checkpoint blockade: why does checkpoint inhibitor immunotherapy not work for all patients?
American Society of Clinical Oncology Educational Book
2019
:
147
64
.
60.
Liu
Z
,
Ge
Y
,
Wang
H
,
Ma
C
,
Feist
M
,
Ju
S
, et al
Modifying the cancer-immune set point using vaccinia virus expressing re-designed interleukin-2
.
Nat Commun
2018
;
9
:
4682
.
61.
Joshi
S
,
Durden
DL
. 
Combinatorial approach to improve cancer immunotherapy: rational drug design strategy to simultaneously hit multiple targets to kill tumor cells and to activate the immune system
.
J Oncol
2019
;
2019
:
5245034
.