S100A7 as a potential diagnostic and prognostic biomarker of esophageal squamous cell carcinoma promotes M2 macrophage infiltration and angiogenesis

2.1 S100A7 is upregulated in ESCC and indicates poor prognosis

We acquired transcriptome data from The Cancer Genome Atlas (TCGA) database and compared the mRNA expression levels of S100 family members between normal esophageal and ESCC tissues. Several S100 family members had abnormal mRNA expression in the ESCC tissues, including the well-known S100A8, S100A9, and S100A2 proteins (Figure 1A). In addition, the less well-known S100A7 protein was clearly upregulated in the ESCC tissues in comparison with the normal esophageal tissues (Figure 1B). The upregulation of S100A7 mRNA was subsequently verified in three independent RNA microarrays (GSE43624, GSE53622, and GSE23400), which comprised 119, 60, and 53 pairs of human ESCC and adjacent normal tissues, respectively^{19, 20} (Figure 1C-E). Immunohistochemical analysis of 341 ESCC tissues and 233 adjacent normal tissue samples demonstrated that S100A7 protein levels were significantly increased in the ESCC tissues (Figure 1F,G, P < .001). The patients were assigned into two groups with high (immunohistochemical score ≥ 6) and low (immunohistochemical score < 6) S100A7 expression, respectively. There were no significant differences in age, gender, smoking status, or TNM staging between the two groups; however, the patients with high S100A7 expression had more advanced tumor differentiation than those with low S100A7 expression (P = .001, Table 1). Our study showed that the patients with high S100A7 expression also had shorter survival than those with low S100A7 expression by Kaplan-Meier survival analysis (P = .024, Figure 1H). Meanwhile, we justified that S100A7 was an independent prognostic factor associated with poor prognosis by Multivariate Cox analysis (P = .026; odds ratio [OR] = 1.513, 95% CI: 1.094-2.092, Table 2). Furthermore, prognosis prediction analysis showed that the addition of S100A7 expression as a prognostic factor increased the accuracy of prognosis assessment of patients with ESCC (Figure 1I). Collectively, the results showed that S100A7 was upregulated in ESCC tissues and was an independent prognostic factor.

2.2 Serum S100A7 is a potential diagnostic biomarker for ESCC

A previous study showed that S100A7 could be secreted outside the cell and detected in serum.²¹ ELISA analysis showed that the level of secreted S100A7 protein was significantly higher in serum samples from 234 patients with ESCC than in those from 127 healthy control individuals (P < .001, Figure 2A). And receiver operating characteristic (ROC) analysis demonstrated that the detection of S100A7 in serum had considerable diagnostic efficacy in patients with ESCC (Figure 2B). The area under the ROC curve (AUC) for diagnosis based on serum S100A7 was as high as 0.790 (95% CI: 0.748-0.833, P < .001) with specificity up to 90.67% (95% CI: 84.94-94.36%) when the sensitivity was 53.65% (95% CI: 47.74-59.46%). Logistic regression analysis was used to integrate S100A7 with the widely used auxiliary diagnostic markers SCC and CRFRA21-1. The inclusion of S100A7 as a diagnostic factor for ESCC increased the AUC from 0.764 (95% CI: 0.686-0.842) for diagnosis based on SCC and crfra21-1 to 0.863 (95% CI: 0.802-0.925) for diagnosis based on S100A7, SCC, and crfra21-1 (Figure 2C, D). Taken together, the results indicated that the serum S100A7 protein level has great potential as a noninvasive diagnostic marker for ESCC.

2.3 S100A7 promotes oncogenic signaling activation via binding with JAB1 and paracrine interaction

Previous studies reported that intracellular S100A7 can interact with C-Jun activation domain-binding protein-1 (JAB1), whereas extracellular S100A7 might exert functions through the RAGE receptor.^{14, 22, 23} JAB1 was initially identified as a c-Jun coactivator, and subsequently revealed to be the fifth component of the constitutive photomorphogenic-9 signalosome (COPS5). Dysregulation of JAB1 is widespread in cancer and involved in affecting a series of cancer associated pathways by activating oncogenes and deactivating several tumor suppressors.²⁴ To determine the role of S100A7 in the activation of ESCC signaling pathways, we constructed KYSE-30 and KYSE-150 cell lines that stably overexpressed S100A7 (Figure 3A). We then established additional KYSE-30 and KYSE-150 cell lines in which the S100A7 overexpression was restoratively silenced by siRNA (Figure 3B). The S100A7 overexpression increased the amount of secreted S100A7 protein in the culture supernatant to more than 60 ng/mL, which was much higher than the level in cultures of control cells without S100A7 overexpression (Figure 3C). Anti-S100A7 coimmunoprecipitation assays and mass spectrometry analysis revealed that S100A7 was able to bind to the JAB1 protein, which was verified by western blot (Figure 3D,E). Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway and gene ontology enrichment analysis of the transcriptome data of S100A7-overexpressed cells and the TCGA ESCC cohort showed that genes that were differentially expressed between tumor cells with high and low S100A7 expression were enriched with functions related to tumor development and progression (Figures 3F and S1), which suggested that S100A7 plays a role in the activation of ESCC-related signal pathways. Western blot analysis showed that S100A7 overexpression or exogenous treatment with S100A7 protein could promote activation of the AKT, ERK, and NFκB signaling pathways (Figure 3G,H). Those effects could be partly abrogated by S100A7 silencing or treatment with RAGE receptor inhibitors (Figure 3G,H). Taken together, the results suggested that S100A7 promotes activation of oncogenic signaling in ESCC cells via binding with JAB1 and autocrine interaction.

2.4 Upregulated S100A7 accelerates the proliferation of ESCC cells via apoptosis repression

Because S100A7 appeared to be involved in cell proliferation-related signaling pathways (Figures 3F and S1), we conducted CCK8 assays to determine its effect on the proliferation of ESCC cells. Consistent with the changes in signaling pathways, S100A7 overexpression or stimulation with exogenous S100A7 protein increased cell proliferation activity, whereas S100A7 knockdown or treatment with RAGE receptor inhibitor reduced proliferation activity (Figures 4A and S2A). Although the regulation of cell proliferation is complicated, cell cycle regulation and apoptosis are the most important and widely recognized pathways involved. Flow cytometry was applied to detect the cell cycle distribution of clonal populations. The results showed that S100A7 had no effect on the cell cycle distribution of KYSE-30 or KYSE-150 cells (Figures 4B and S2B); however, S100A7 overexpression inhibited cisplatin-induced apoptosis of ESCC cells (Figures 4C,D and S2C, S2D). Furthermore, in subcutaneous xenograft experiments, S100A7-overexpressing ESCC cells had increased tumor-formation ability compared with ESCC cells expressing empty vector (Figure 4E-G). Immunohistochemical analysis showed that the S100A7-overexpressing xenograft tumors had higher Ki67 and proliferating cell nuclear antigen (PCNA) expression and lower P21 expression than the control tumors (Figure 4H). Overall, the results indicated thatS100A7 accelerates the proliferation of ESCC cells via apoptosis repression.

2.5 S100A7 promotes the metastatic capability of ESCC cells

The signaling pathways regulated by S100A7 are widely involved in local invasion and distant metastasis, which are one of the 10 hallmarks of cancer and a major cause of cancer-related death.²⁵ To further investigate the effect of S100A7 on ESCC metastasis, we performed Transwell assays to characterize cancer cell migration and invasion abilities in vitro. KYSE-30 and KYSE-150 cells overexpressing S100A7 or treated with exogenous S100A7 protein had significantly increased invasion and migration abilities, which were at least partly abrogated by S100A7 silencing or treatment with RAGE receptor inhibitor (Figure 5A,B). Consistent with the in vitro cellular experiments, mouse lung colonization assays further confirmed the prometastasis effect of S100A7 on ESCC cells. Tail vein injection with S100A7-overexpressing ESCC cells in mice resulted in significantly more lung metastases than injection with control ESCC cells (Figure 5C,F). Those results suggested that S100A7 plays a prometastasis role in ESCC.

2.6 Extracellular secreted S100A7 chemotactically recruits macrophages and promotes M2 polarization

In addition to classical tumor-related pathways, many immune-related signaling pathways were also enriched among the genes that were differentially expressed between wild-type ESCC cells and S100A7-overexpressing ESCC cells (Figures 3F and S1). Those results suggested that S100A7 participates in the regulation of the immune microenvironment of ESCC. We used the xCell online tool²⁶ to analyze microarray data of 119 pairs of ESCC tissues (GSE53625) and found that the expression of S100A7 was correlated with the infiltration of various immune cells. In particular, the infiltration of M2 macrophages in the tumor microenvironment was positively correlated with the expression level of S100A7 (Figures 6A,B and S3A, S3B, S3C). Immunohistochemical analysis of 306 ESCC tissue microarrays using anti-CD163 antibodies, a molecular marker of M2 macrophages, revealed that the infiltration of M2 macrophages was increased in ESCC tissues with high S100A7 expression compared with that ESCC tissues with low S100A7 expression (Figure 6C,D). Furthermore, the patients with higher M2 macrophage infiltration had shorter overall survival than those with less M2 macrophage infiltration (Figure 6E). We performed a chemotaxis assay using the Transwell chamber system and found that treatment with exogenous S100A7 protein or S100A7 overexpression in ESCC cells had chemotactic effects on PMA-activated THP-1 cells (Figure 6F). Similarly, mouse S100a7a protein was capable of stimulating chemotaxis in the murine macrophage cell line RAW264.7 (Figure S3D). In addition, S100A7 upregulated the expression of M2 macrophage-associated markers induced by IL4 (Figure 6G), suggesting that S100A7 promotes M2 macrophage polarization.

To further investigate the function of the interaction between tumor cells and macrophages mediated by secreted S100A7, we mixed S100A7-overexpressing or control ESCC cells with PMA-activated THP-1 cells and injected them subcutaneously into mice. We observed that KYSE-30 and KYSE-150 cells formed larger tumor nodules when mixed with macrophages than when they were injected without macrophages (Figures 6H,I and S3E, S3F). Moreover, interaction analysis revealed significant synergy between S100A7 upregulation and mixing of the ESCC cells with macrophages (Figure 6J). Those results suggested that, in addition to directly affecting tumor cells, secreted S100A7 promotes the progress of tumor cells indirectly by interacting with macrophages.

2.7 Extracellular secreted S100A7 promotes angiogenesis

In addition to the inherent invasion and migration abilities of tumor cells, tumor metastasis is the result of interaction between cancer cells and the surrounding tumor microenvironment, in which angiogenesis is one of the most important factors.²⁷ Moreover, increased tumor-associated macrophages in the tumor microenvironment can promote cancer angiogenesis.²⁸ Therefore, we explored the effect of exocrine S100A7 protein on tube formation by vascular endothelial cells and found that treatment with S100A7 protein or supernatant from cultures of S100A7-overexpressing cells enhanced the tube-forming ability of endothelial cells (Figures 7A and S4A). In addition, the migration ability of vascular endothelial cells, another important factor in angiogenesis, was facilitated by treatment with S100A7 protein or supernatant from S100A7-overexpressing cells (Figures 7B and S4B). To further investigate the effect of S100A7 on angiogenesis, we performed aortic arch angiogenesis assays. The results showed that S100a7a, the mouse homolog of human S100A7, significantly promoted neovascularization in mouse aortic rings (Figure 7C). Treatment with human recombinant S100A7 protein also promoted microvessel formation in mouse aortic rings (Figure 7C), suggesting that the human S100A7 protein and the homologous mouse protein S100a7a are sometimes functionally interchangeable. Similar results were obtained in Matrigel plug assays in vivo; compared with injection of Matrigel alone as a control, subcutaneous injection of Matrigel containing recombinant murine S100a7a protein resulted in increased formation of new blood vessels (Figure 7D-F). Previous studies confirmed that secreted S100A7 can bind to the RAGE receptor on the membrane of macrophages, which is also expressed on vascular endothelial cells. To determine if the proangiogenesis effect of S100A7 depends on the RAGE receptor, we incubated S100A7 protein with RAGE-overexpressing human umbilical vein endothelial cells (HUVECs) for coimmunoprecipitation assays. As shown in Figure 7G, S100A7 could bind with RAGE receptor on the HUVEC outer membrane, which suggested that S100A7 secreted by cancer cells might promote angiogenesis by activating the RAGE pathway. We then detected the downstream signaling and found that both S100A7 protein and supernatant of S100A7-overexpressing ESCC cells could promote the phosphorylation of ERK and FAK (Figure 7H), which are vital signals for angiogenesis. Those findings suggested that extracellular secreted S100A7 promotes angiogenesis by interacting with the tumor microenvironment.

4.1 Human tissues and serum samples

All human tissues and serum samples were collected from the Cancer Hospital, Chinese Academy of Medical Sciences. ESCC tissues and adjacent nontumor esophageal tissues for tissue arrays were obtained from 341 patients that underwent radical resection from December 2005 to December 2008. Additional serum samples from 274 patients with ESCC and 151 healthy controls were obtained from Laboratory Department during May 2018 to December 2019. All cases of ESCC were pathologically confirmed. None of the patients with ESCC had received antitumor therapy or had a history of other malignancies within the three years prior to their ESCC diagnosis. The clinicopathological characteristics of the patients were collected through telephone interviews and routine laboratory studies conducted prior to surgery. Tumor and adjacent normal tissues obtained during surgery were snap frozen in liquid nitrogen within 30 minutes of resection and stored at –80°C until the tissue arrays were manufactured. Serum samples were aliquoted into 200 μL portions, which were stored in separate tubes at –80°C until use. Exclusion criteria for the study were incomplete medical records, hemolysis, and exposure of a given sample to repeated freeze-thaw cycles. The Ethics Review Committee of Research Involving Human Subjects of the Cancer Hospital, Chinese Academy of Medical Sciences granted our ethics approval (20/287-2483).

4.2 Aortic arch angiogenesis assay

Aortic arch angiogenesis assays were performed according to a previous report.⁴³ Briefly, 5-week-old C57/BL6 mice were sacrificed and dissected to remove the thoracic aorta. The dissected aortae were immersed in Opti-MEM and cut into rings ∼0.1-0.5 mm in width and transferred to a 10 cm dish containing Opti-MEM. The rings were serum-starved overnight in a cell incubator and then embedded separately in a Slide-chamber well (BIOLOGIX) with Collagen type I (Millipore, #08-115). Add 150 μL of Opti-MEM medium supplemented with penicillin-streptomycin, 2.5% (vol/vol) FBS and VEGF (30 ng/mL) to the embedded rings. Recombinant human S100A7 protein or murine S100a7a protein was then added to the medium. Next, the arterial rings were divided into three groups for treatment with S100A7 (100 ng/mL, R&D), S100a7a (100 ng/mL, Origene), and PBS, respectively. After the treatments were added to the cultures, the growth medium was changed first on day 3 after the start of the treatments and then every other day.

Microvessel growth during the experiment was quantified by live phase-contrast microscopy. On day 7, the culture medium was removed, and the whole plates were washed with wash buffer A (PBS, CaCl₂, MgCl₂). The cells in each well were fixed with 4% formalin for 30 minutes at room temperature, permeabilized by with buffer (PBS, CaCl₂, MgCl₂, 0.25%(vol/vol) Triton X-100) for 15-minute incubations at room temperature (twice), and then blocked with 2% fetal bovine serum (FBS) for 30 minutes at 37°C. The following primary antibodies were prepared in PBLEC solution: BS1 lectin-FITC (Sigma, #L9381) to label endothelial cells and antiactin α-smooth muscle Cy3 (Sigma, #C6198)) for supporting cells. The permeabilized and blocked cells were incubated overnight with antibody solution at 4°C, washed by buffer C (PBS, 0.1% (vol/vol) Triton X-100), and then sealed with antifade mounting medium with DAPI (Thermo #P36966). The stained neo-microvessels were then analyzed with a fluorescence microscope.

4.3 In vivo animal experiments

All mice were purchased from Huafukang Bioscience (Beijing, China) and housed at the SPF Animal Experiment Center, Cancer Hospital, Chinese Academy of Medical Sciences. All animal experimental procedures were approved by the Animal Care and Use Committee.

4.4 Statistics

GraphPad Prism 8 (GraphPad Software, Inc.) and SPSS 23 (IBM, USA) were used for data analysis. Student's t-test was used for comparisons between two groups. Mann-Whitney U test was used to compare nonnormally distributed data. Pearson's chi-square test was used for clinicopathological correlation analysis. Kaplan-Meier analysis and Cox regression were performed for univariate and multivariate prognosis analysis, respectively. Correlation coefficients were obtained by Pearson correlation analysis. Logistic regression analysis was carried out to integrate multiple diagnostic markers. Two-sided test was used for all statistical analysis, and P -values < .05 were considered statistically significant. The graphical abstract was created with BioRender.com.

The details of other experimental materials and methods can be found in the Supplementary Materials