Site-specific fibroblasts regulate site-specific inflammatory niche formation in gastric cancer
Fibroblasts are the commonest type of cancer stromal cells. Inflammation occurs in cancer tissue, and the inflammatory process has been suggested to be caused by interactions between immune cells and cancer cells. In this study, we clarified that site-specific fibroblasts regulate the formation of a site-specific inflammatory niche according to the depth of gastric cancer cell invasion.
Immunohistochemistry was performed with paraffin-embedded tissues. The numbers of immune cells and the fibroblast area were calculated according to the cancer depth. The gene expression patterns of submucosal fibroblasts and subperitoneal fibroblasts stimulated with HSC44PE-conditioned medium were analyzed with a microarray. To examine the effects on the cancer microenvironment of differences in gene expressions between HSC44PE-stimulated submucosal fibroblasts and subperitoneal fibroblasts, assays of HSC44PE proliferation, T cell migration, and M2-like macrophage differentiation were performed.
The distributions of immune cells differed between the submucosal layer and the subserosal layer. The number of M2 macrophages was significantly higher and the fibroblast area was significantly larger in the subserosal layer compared with the submucosal layer. High expression levels of IL1B, TNFSF15, and CCL13 were observed in HSC44PE-stimulated submucosal fibroblasts, and higher expression levels of TGFB2, CSF1, CCL8, and CXCL5 were found in HSC44PE-stimulated subperitoneal fibroblasts. HSC44PE-stimulated subperitoneal fibroblast medium promoted the differentiation of monocytes into M2-like macrophages, whereas HSC44PE-stimulated submucosal fibroblasts significantly induced the migration of Jurkat cells and the growth of HSC44PE cells.
The dynamic states of immune cells differ between the submucosal and subserosal layers in cancer tissues. Site-specific fibroblasts regulate site-specific inflammatory niche formation according to the depth of cancer cell invasion.
KeywordsInflammation Gastric cancer Immune cells Fibroblasts Inflammatory niche
Fibroblasts are among the commonest types of stromal cells in connective tissues and are essential for tissue homeostasis and wound healing in the process of inflammation . Fibroblasts comprise a very heterogeneous population of cells, with each subpopulation exhibiting different pathophysiological functions . We previously demonstrated in our laboratory that colon fibroblasts obtained from submucosal and subperitoneal tissues showed biological differences and differential gene expression . In cancer tissues, whereas some cancer-associated fibroblasts promote cancer cell proliferation and migration, others suppress cancer cell growth.
Rudolf Virchow hypothesized that cancer arises from inflammatory sites, so-called sites of “lymphoreticular infiltration”. Abundant evidence has been collected to suggest that infection and inflammatory disease serve as triggers for the development of various cancers [4, 5]. In addition to fibroblasts, cancer stroma also contains new blood vessels and connective tissue. In 1986, Dvorak showed that wound healing and tumor stroma formation share many important properties [5, 6] and defined cancers as wounds that do not heal . Therefore, inflammation is an inherent pathophysiological process in cancer tissues. Many reports have suggested that cancer-associated inflammation is caused by interactions between immune cells and cancer cells.
Inflammation has been reported to be closely involved in the initiation, progression, and metastasis of gastric cancer . For example, persistent infection with Helicobacter pylori leads to the stimulation of immune cells by bacterial toxins, resulting in the production of a wide variety of inflammatory factors, including tumor necrosis factor (TNF)-α, interleukin (IL)-1, IL-6, and IL-8. These factors cause inflammation and promote the development of gastric cancer and the infiltration of lymphocytes [9, 10]. Moreover, IL-6 is produced mainly by macrophages and lymphocytes under the regulation of the nuclear factor κB signaling pathway. IL-6 is also linked to signal transducer and activator of transcription 3 and exerts multiple actions, including growth-promoting and antiapoptotic actions on cancer cells. Therefore, the elevation of the serum level of IL-6 has been proposed as a poor prognostic factor for recurrence and overall survival in gastric cancer patients [11, 12]. Consistent with these reports, a number of other reports have also suggested that the key features of cancer-associated inflammation are the production of cytokines by immune cells and the interactions between cancer cells and immune cells [13, 14].
On the basis of the observation that fibroblasts are among the commonest cell types within the stroma, it has been suggested that they may play important roles in cancer-associated inflammation. In addition to stimulating cell adhesion and inducing leucocyte activation during the process of inflammation, inflammatory fibroblasts from the rheumatoid synovium have also been shown to promote the onward migration of T cells through the endothelial cell layer in co-culture models . In pancreatic cancer, transgenic mice with a deficiency of alpha smooth muscle actin (α-SMA) myofibroblasts showed immune suppression as a result of an increase in the number of regulatory T cells . We also reported that HSC44PE-conditioned medium induced the expression of mainly IL-8 in fibroblasts . Fibroblasts are very heterogeneous and show biological differences between the submucosal layer and the subserosal layer in cancer tissue. Additionally, fibroblasts have the ability to induce the migration of immune cells to an inflammatory niche in cancer tissue.
These reports suggest that in gastric cancer, site-specific fibroblasts regulate site-specific inflammatory niche formation, according to the depth of the invasion of the gastric cancer cells. In this study, we focused on inflammatory niche formation according to the depth of the invasion of gastric cancer cells. We demonstrated that each type of fibroblast plays its own role in inflammatory niche formation in gastric cancer by comparing the character of stromal cells, such as immune cells and fibroblasts, with that of the submucosal layer and that of the subserosal layer.
Materials and methods
Two human gastric cancer cell lines were used in the study. MKN7 was obtained from Immuno-Biological Laboratories (Gunma, Japan) and HSC44PE was established previously . MKN7 was maintained in Dulbecco’s modified Eagle’s medium (DMEM; Life Technologies, NM, USA) containing 10 % heat-immobilized fetal bovine serum (FBS; Life Technologies) and supplemented with penicillin and streptomycin (Life Technologies). HSC44PE, Jurkat, and THP1 cells were maintained in RPMI 1640 medium (Life Technologies) containing 10 % heat-immobilized FBS and supplemented with penicillin and streptomycin.
Human primary fibroblast cells
Noncancerous submucosal or subperitoneal tissues (at least 5 cm from the cancer lesion) were obtained from surgically resected specimens of human gastric cancer patients. The procedure for isolation of human primary fibroblast cells (submucosal gastric fibroblasts, SMFs; subperitoneal gastric fibroblasts, SPFs) has been reported previously . The SMFs and SPFs were maintained in minimum essential medium α (Life Technologies) containing 10 % heat-immobilized FBS and supplemented with penicillin and streptomycin. The study protocol was approved by the Institutional Review Board of the National Cancer Center Hospital East.
Mouse monoclonal antibodies against CD8, CD15, CD56, CD68, Ki67, and α-SMA were obtained from Dako (Glostrup, Denmark). Mouse monoclonal antibodies against CD4 and CD20 were purchased from Abcam (Cambridge, UK). The mouse monoclonal antibody against CD204 was purchased from Trans Genic (Kumamoto, Japan). Horseradish peroxidase (HRP)-labeled species-specific whole antibody was also obtained from Dako.
Immunohistochemistry was performed with paraffin-embedded tissues obtained from 20 surgically resected specimens of human gastric cancer. The cancer had invaded the serosa in 20 specimens. Ten specimens exhibited up to moderate differentiation and were classified as differentiated subtype, and ten specimens were classified as poorly differentiated subtype. Most of the cancers had heterogeneous components, especially in terms of the differentiation status. Seven of 20 cases which consisted of an admixture of differentiated and poorly differentiated subtypes were classified into four cases with differentiated subtype and three cases with poorly differentiated subtype, respectively, according to their predominancy on the basis of the rules of the Japanese Classification of Gastric Carcinoma (14th edition, 2010).
After deparaffinization, the paraffin sections were immersed in the target retrieval solution at 95 °C for 20 min. To block endogenous peroxidase activity, the specimens were immersed in methanol containing 3 % hydrogen peroxide. The samples were washed and incubated overnight at 4 °C with each of the antibodies. The sections were washed and incubated with EnVision+ System-HRP labeled polymer anti-mouse (Dako). The tissues were then incubated with diaminobenzidine peroxidase substrate for the indicated time. Nuclear counterstaining was performed with Mayer’s hematoxylin solution (Muto Pure Chemicals, Tokyo, Japan).
Immunohistochemical scoring of the immune cells, fibroblasts, and cancer cells
Differential counts were performed for the submucosal layer, the muscularis propria, and the subserosal layer. The numbers of each type of immune cell [helper T cells (CD4+ cells), killer T cells (CD8+ cells), B cells (CD20+ cells), naïve T cells (CD56+ cells), neutrophils (CD15+ cells), macrophages (CD68+ cells), and M2 macrophages (CD204+ cells)] in the cancer tissues and noncancer tissues (at least 5 cm from the cancer lesion) were counted in five selected fields from each layer. The α-SMA-positive fibroblast area per field was measured with AxioVision release 4.7 (Zeiss, Oberkochen, Germany) and was expressed as the percentage of the total area.
Random high-power fields were selected to count the number of nuclei showing positive staining for Ki67. One thousand nuclei were counted, and the Ki67 immunoreactivity was expressed as the proportion of Ki67-positive cancer cells among 1000 cancer cells .
Collection of gastric cancer cell conditioned medium
On day 0, gastric cancer cells (5.0 × 106 per dish) were incubated in DMEM containing 10 % FBS. The culture-conditioned medium was replaced with DMEM not containing FBS on day 2 and was collected and filtered with a vacuum filtration system (Millipore, Billerica, MA, USA). The filtrates were dispensed in 1-mL aliquots into each tube on day 3. The collected specimens were stored at −80 °C.
Collection of HSC44PE-stimulated SMF medium and HSC44PE-stimulated SPF medium
On day 0, SMFs and SPFs (5.0 × 105 per dish) were incubated in DMEM containing 10 % FBS. HSC44PE-conditioned medium was added to the SMFs or SPFs on day 2. On day 3, the culture-conditioned medium were replaced with DMEM not containing FBS, and the conditioned medium (HSC44PE-stimulated SMF medium, HSC44PE-stimulated SPF medium) was collected and filtered with a vacuum filtration system (Millipore, Billerica, MA, USA); the filtrates were dispensed in 1-mL aliquots into each tube on day 4.
Stimulation with gastric-cancer-cell-conditioned medium
On day 0, the SMFs and SPFs (5.0 × 105 per dish) were incubated in DMEM containing 10 % FBS. Gastric-cancer-cell-conditioned medium or DMEM not containing FBS was added to the SMFs or SPFs on day 2, followed by incubation for the indicated periods.
Quantitative real-time reverse transcriptase polymerase chain reaction
Each of the fibroblasts was incubated with DMEM or HSC44PE-conditioned medium, and the THP1 cells were incubated with HSC44PE–SMF/SPF medium for the designated times. The cells were washed with phosphate-buffered saline and suspended in 1-mL of TRIzol. Total RNA was purified by the TRIzol/RNeasy minicolumn protocol (QIAGEN), and complement ray DNA was synthesized with a PrimeScript® RT reagent kit (TaKaKa, Shiga, Japan). Quantitative real-time reverse transcriptase polymerase chain reaction (qRT-PCR) was performed with a LightCycler 48 (Roche, Basel, Switzerland). A quality check of all the total RNA samples was performed with an Agilent Bioanalyzer with an RNA 6000 Nano assay kit (Agilent Technology), and the RNA integrity of all the RNA samples was confirmed to be greater than 9.0.
To analyze the messenger RNA (mRNA) expression levels of about 47,000 transcripts and variants from 38,500 well-characterized human genes, GeneChip Human Genome U133 Plus 2 arrays (Affymetrix) containing 54,675 probe sets were used. Target complementary RNA was generated from 100 ng of total RNA from each sample with a 3′ IVT Express kit (Affymetrix). The procedures for target hybridization, washing, and staining with signal amplification were undertaken according to the supplier’s instructions. The arrays were scanned with a GeneChip Scanner 3000 (Affymetrix).
Statistical analysis of the microarrays
The gene expression data were analyzed with GX12.6 (Agilent Technologies). Raw data were summarized with the MAS5 algorithm (Affymetrix) and were normalized to log-transformed and median-centered data for the numerical analysis to permit gene selection. The differentially expressed probe sets used in the supervised hierarchical clustering were selected on the basis of P < 0.05 and fold changes greater than 2.0. P values were calculated by a one-way ANOVA with Benjamini and Hochberg multiple correction. For the hierarchical clustering, average linkage clustering with the Pearson correlation distance was performed.
Functional analysis of the microarrays
The predicted target genes were imported for functional analysis into DAVID version 6.7, a Web-based tool for annotation, visualization, and integrated discovery. We performed the biological process of Gene Ontology (GO) and pathway analysis on the mRNAs of genes showing differential expression between fibroblasts stimulated with cancer cells (MKN7 and HSC44PE) and those stimulated with DMEM. The statistical significance of the results of the GO and pathway analysis was determined on the basis of P < 0.01.
Jurkat cell migration was assessed with a modified Boyden’s chamber assay—that is, in Transwell cell culture chambers (Corning, Corning, NY, USA). Polycarbonate filters with 5-μm pores were used to separate the upper and lower chambers. On day 0, Jurkat cells (1.0 × 106 cells per dish) were incubated in DMEM containing 10 % FBS. On day 1, DMEM not containing FBS was added to the Jurkat cells. Then, the Jurkat cells were added to the upper compartment of the chamber at a density of 1.0 × 107 cells per milliliter of DMEM not containing FBS and were incubated for 4 h. The Jurkat cells were allowed to migrate toward the reagents (HSC44PE–SMF medium, HSC44PE–SPF medium) in the lower chamber. After the reaction, the cells that had migrated were collected from the bottom chamber and were counted with a hemocytometer . The number of Jurkat cells in the positive control (DMEM containing FBS) was determined, and the rate of Jurkat cell migration was determined as the number of cells in the HSC44PE-stimulated fibroblast medium relative to that in DMEM.
Cell viability assay
Cell viability was measured with the 2-(2-methoxy-4-nitrophenyl)-3-(4-nitrophenyl)-5-(2,4-disulfophenyl)-2H-tetrazolium (WST-8) assay (Dojindo Laboratories) according to the manufacturer’s instructions. HSC44PE cells (5.0 × 106 per dish) were incubated in DMEM containing 10 % FBS, penicillin, and streptomycin. After 2 days, HSC44PE cells were seeded in triplicate at 4.0 × 103 cells per well in a 96-well flat-bottomed culture plate containing HSC44PE–SPF medium, HSC44PE–SMF medium, DMEM–SPF medium, or DMEM–SMF medium. The WST-8 cell counting kit was added to each well, followed by incubation for 4 h. Subsequently, the optical density was measured at 450 nm.
Differentiation of M2-like macrophages
On day 0, THP1 cells (1.0 × 106 cells per 10-cm dish) were incubated in DMEM containing 10 % FBS. On day 2, DMEM not containing FBS was added to the THP1 cells. On day 3, the THP1 cells were treated for 48 h with HSC44PE–SMF medium or HSC44PE–SPF medium. Cells were collected and washed with phosphate-buffered saline and then suspended in 1 mL of TRIzol.
The statistical significance of the difference between any two groups of interest was evaluated with the Student t test. Differences were considered significant at P < 0.05. For the results of the immunostainings, the error bars show the mean ± standard deviation. For other results, the error bars show the mean ± standard error. For the Ki67 labeling index, the statistical analysis was performed with the Wilcoxon signed-rank test.
Inflammatory niche formation in cancer tissue
To investigate the differences in the inflammation niches according to the depth of invasion of gastric cancer, immunohistochemistry was performed with paraffin-embedded tissues obtained from 20 surgically resected specimens of human gastric cancer.
To examine whether the depth of cancer cell invasion influences the formation of the inflammatory niche, the numbers of each of the immune cell types and the area of the α-SMA-positive fibroblasts were examined according to the depth of cancer cell invasion.
The distributions of the immune cells were different between the submucosal layer and the subserosal layer. These results indicate that the inflammatory environment surrounding the cancer cells varies according to the depth of the invasion of the cancer cells, with the interaction between cancer cells and stromal cells regulating the formation of a specific inflammatory niche in each layer.
Differential gene expression between SMFs and SPFs stimulated with HSC44PE-conditioned medium
We examined whether differences in the stromal fibroblasts might underlie the difference in the inflammatory niches formed in the submucosal and subserosal layers in gastric cancer tissue, because fibroblasts are the most abundant stromal cells in cancer tissue .
Biological process of Gene Ontology (GO) and pathway analysis of the messenger RNA expression of genes showing differential expression between fibroblasts stimulated by cancer cells and Dulbecco’s modified Eagle’s medium
GO:0005125 cytokine activity
9.58 × 10−21
GO:0008009 chemokine activity
9.63 × 10−13
GO:0042379 chemokine receptor binding
2.20 × 10−12
GO:0008083 growth factor activity
5.37 × 10−8
GO:0005507 copper ion binding
4.95 × 10−6
GO:0046870 cadmium ion binding
7.19 × 10−6
GO:0005539 glycosaminoglycan binding
3.458 × 10−3
GO:0030246 carbohydrate binding
4.041 × 10−3
We previously demonstrated that SMFs and SPFs stimulated with a colon-cancer-conditioned medium exhibit different gene expression patterns .
To examine whether the gene expression patterns of SMFs and SPFs stimulated with HSC44PE-conditioned medium differed, a microarray analysis was performed. Moreover, to confirm the mRNA expression of signature genes in the SMFs and SPFs, qRT-PCR was used. Genes encoding hormonal factors were extracted in a data analysis.
Effects of fibroblasts stimulated with cancer cells on the cancer microenvironment
To examine the effect of differences in the gene expression between SMFs and SPFs stimulated with HSC44PE-conditioned medium on the cancer microenvironment, assays for HSC44PE proliferation, T cell migration, and M2-like macrophage differentiation were performed.
The number of lymphocytes was higher in the submucosal layer than in the subserosal layer (Fig. 1). To examine whether SMFs more strongly than SPFs induced the migration of T cells, we examined the migration of Jurkat cells. Compared with the HSC44PE–SPF medium, the HSC44PE–SMF medium induced significantly more pronounced migration of Jurkat cells (Fig. 5b). This finding indicates that SMFs also have the ability to promote the migration of T lymphocytes, as observed by histopathology examinations.
Our present study showed that the inflammatory environment varied from the early stage of inflammation to the late stage of inflammation according to the depth of cancer cell invasion. Site-specific fibroblasts might be involved in the regulation of site-specific inflammatory niche formation. The activation of site-specific fibroblasts by gastric-cancer-cell-conditioned medium showed significantly different inflammatory pathways between SMFs and SPFs.
In this study, we used two types of fibroblasts: SMFs SPFs. Histopathology examinations and analyses of gene expression indicated that the cancer microenvironment differed between the submucosal layer and the subserosal layer (Figs. 1, 2).
Fibroblasts are the most abundant stromal cells in cancer tissues. From the perspective of the kinetics of immune cells, chemokine (C–C motif) ligand 13, TNF superfamily member 15, and IL-1β promote infiltration and the accumulation of T lymphocytes at sites of inflammation . HSC44PE-stimulated SMFs induced the migration of Jurkat cells and showed increased expression of CCL13, TNFSF15, and IL1B (Figs. 3, 5b). These results indicate that cancer-stimulated SMFs promote the migration of lymphocytes. On the other hand, colony stimulating factor 1, chemokine (C–C motif) ligand 8, chemokine (C–X–C motif) ligand 5, and transforming growth factor β2 are involved in the differentiation of M2 macrophages and promotion of fibrosis [25, 26, 27, 28]. HSC44PE-stimulated SPFs induced the differentiation of THP1 cells into M2-like macrophages and showed increased expression of CSF1, CCL8, CXCL5, and TGFB2 (Figs. 3, 5a). These results suggest that SPFs promote the differentiation of monocytes into M2-like macrophages. Tumors with myofibroblast depletion have been reported to show a significant decrease in the overall peritumoral infiltration by CD45+ cells, CD3+ cells, and CD19+ B cells in pancreatic cancer . The existence of fibroblasts would be necessary for the formation of an inflammatory niche, and each type of fibroblast has its own role in promoting the differentiation or migration of immune cells.
In addition, the percentage of Ki67-positive cancer cells and the proliferation of HSC44PE cells were significantly higher in the submucosal layer containing SMFs (Fig. 4). A previous report also clarified that the major pool of proliferating cancer cells resided in the superficial layers (mucosa and submucosa) of gastric cancer . Moreover, HSC44PE-stimulated SMFs showed enhanced expression of IL1B, TNFSF15, and CXCL13, which promote cancer cell growth (Fig. 3). On the other hand, HSC44PE-stimulated SPFs showed enhanced expressions of TGFB2 (Fig. 3). A previous report suggested that transforming growth factor β1 effectively diminished the expression of the epithelial marker E-cadherin and enhanced the expression of the mesenchymal marker vimentin in gastric cancer cells . These reports and results suggest that SMFs promote cancer cell growth and that SPFs promote cancer cell invasion.
In gastric cancer, cancer cell invasion occurs from the submucosal layer to the subserosal layer, progressing over time. Therefore, the submucosal layer would be expected to show late-stage inflammation and the subserosal layer would be expected to show early-stage inflammation. However, we obtained precisely the opposite result (Fig. 6). Thus, the responses of SPFs may differ according to the type of stimulation, whereas SMFs may respond similarly to both pathogens and cancers. In the case of sepsis, microorganisms react with inflammatory cells and fibroblasts in the peritoneal membrane, which results in the production of a large array of cytokines and chemokines . Furthermore, it has been shown that following the intraperitoneal administration of lipopolysaccharide to mice, the cells lining the peritoneal membrane produce IL-6, IL-1β, and TNF-α [31, 32]. These reports suggested that SPFs also produce the proinflammatory cytokines IL-1β, IL-6, and TNF-α. However, in our present experiment, stimulation with HSC44PE-conditioned medium did not induce the gene expression of proinflammatory cytokines in the SPFs (Fig. 3). Thus, the roles of SPFs might differ in infection and cancer.
Functional differentiation between SMFs and SPFs against cancer cells may depend on the external environment. The mucosae of the digestive tract provide an interface for interactions between multicellular hosts and their external environment . Because of continuous environmental exposure, the mucosal surface is a major route for the entry of numerous pathogens . The gastrointestinal tract contains gut-associated lymphoid tissues and mucosa-associated lymphoid tissues to protect against invasion by pathogens . Both cancer cells and pathogens share the ability to destroy tissue during the inflammatory stage. From the standpoint of fibroblasts during inflammation, SMFs may acquire the ability to defend against the establishment of cancer cells as a result of exposure to the threat of tissue destruction by pathogens. On the other hand, the serosa has a lower chance of being invaded by the numerous pathogens, resulting in the SPFs in the subserosal layer not having acquired the ability to attack such pathogens. Therefore, the responses of SMFs and SPFs to cancer cells may differ.
We conclude that the dynamic state of immune cells differs between the submucosal layer and the subserosal layer in cases of cancer. Site-specific fibroblasts regulate the formation of site-specific inflammatory niches in cancer tissues according to the depth of cancer cell invasion.
We thank Yoshiko Onuma and Motoko Suzaki for helpful advice.
Compliance with ethical standards
Conflicts of interest
The authors declare that they have no conflict of interest.
All procedures followed were in accordance with the ethical standards of the responsible committee on human experimentation (institutional and national) and with the Helsinki Declaration of 1964 and later versions.