Pan-cancer analysis of Ccl5 was conducted by Tumor Immune Estimation Resource (TIMER) 2.0 (28). Ccl5 was entered into the TIMER 2.0 web interface (http://timer.cistrome.org/), 'Gene_DE module' was applied to study the differential expression between tumor and adjacent normal tissues across all TCGA tumors. Distribution of gene expression are displayed using box plots. The gene expression profiling of Ccl5 influence was obtained from GEO datasets GSE105042 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE105042). Then two wild type macrophages and two Ccl5 knock out macrophages was retrieved from a mouse dataset using the GPL21273 HiSeq X Ten platform (Illumina, Inc.). The original gene expression profiles were analyzed to identify the upregulated or downregulated DEGs in Ccl5 knock out samples, respectively. The criteria for a DEG were |log2FC|>1 and adjusted P-value <0.05. Gene Ontology (GO) enrichment analysis and Kyoto Encyclopedia or Genes and Genomes (KEGG) pathway analysis were performed using the R package cluster Profiler (v. 4.0.5; https://www.R-project.org) with identified DEGs (29). GO enrichment analysis and KEGG pathway analysis were performed with the thresholds of a P-value <0.05.

The Search Tool for the Retrieval of Interacting Genes/Proteins (STRING) database (https://string-db.org/) was used to construct the PPI network (30). DEGs were mapped to a STRING list to perform a search for multiple proteins and obtain a PPI network with interaction scores >0.4. Cytoscape (v. 3.9.0) was used to visualize the results from the PPI network and perform module analysis (31). Module analysis was performed using the molecular complex detection (MCODE) plugin on the Cytoscape (v. 3.9.0) platform. The parameters set to identify enriched functional modules were as follows: Degree Cutoff=2, Node Score Cutoff=0.2, K-Core=2 and Maximum. Depth=100. Modules with the MCODE score ≥4 were identified as significant modules.

Compound 91b1 (Fig. 1B) was prepared by addition of Br₂ into commercially available 8-hydroxy-2-methylquinolin-8-ol (MilliporeSigma) in MeOH, followed by asymmetric hydrogenation under optimal reaction condition as previously described (23). The structure of 91b1 has been physically characterized using 1H- and 13C-NMR and liquid chromatography mass spectrometry (LC-MS) as previously reported (22). Dimethyl sulfoxide (DMSO) was used to dissolve 91b1 for the in vitro biological assays as described below.

Esophageal squamous cell carcinoma (ESCC) cell lines of Japanese origin KYSE150, KYSE450, KYSE30 and KYSE510 (32) were purchased from Leibniz-Institut DSMZ-Deutsche Sammlung von Mikroorganismen und Zellkulturen GmbH. ESCC cell lines of Hong Kong origin SLMT-1 (33) and HKESC-4 (34) were kindly provided by Professor Gopesh Srivastava in the Department of Pathology, The University of Hong Kong, China. Non-neoplastic esophageal epithelial cell line NE-3 (35) (immortalized by the induction of genes E6 E7 of human papillomavirus type 18) was kindly provided by Professor George S.W. Tsao in the Department of Anatomy, The University of Hong Kong, China. Non-neoplastic embryonic cell line 293 (36) was purchased from the American Type Culture Collection (ATCC). The culture medium for KYSE30, KYSE150, KYSE450, and KYSE510 was 45% RPMI with 45% F-12 and 10% FBS; that for SLMT-1, HKESC4, and 293 was 90% MEMα with 10% FBS; and that for NE3 was KSFM with complementary supplements. All the cells were cultured in media supplemented with 100 units/ml penicillin G and 100 μg/ml streptomycin and all cells were maintained in a humidified atmosphere of 95% air and 5% CO₂ at 37°C. The cultures were passaged at preconfluent densities of ~80% using a solution of 0.25% trypsin (Invitrogen; Thermo Fisher Scientific, Inc.). Cells were washed briefly with phosphate-buffered saline (PBS), treated with 0.25% trypsin, and harvested by centrifugation (300 × g for 5 min) at room temperature for subculturing.

Paraffin-embedded specimens containing 26 cancer tissues and 15 non-neoplastic tissues were collected after esophagectomy, with the consent of patients, at the Department of Surgery, Queen Mary Hospital, Hong Kong between January 1990 and December 2001 and the ethics approval for working on the specimens was obtained from the Hong Kong Polytechnic University (approval no. HSEARS20171213007). All the specimens were collected from patients who had received no prior treatment directed to the primary ESCC. The age of the patients ranged from 45 to 77 years old, with a median of 65, comprising 23 males and 3 females. Oral informed consent was obtained in from all patients. The authors confirm that all methods were carried out in accordance with relevant guidelines and regulations. The clinical and histological information of ESCC patient specimens were reported by a pathologist (AKY Lam).

The cytotoxic effect of 91b1 and CDDP on the ESCC cell lines and non-neoplastic cell lines was examined by CellTiter-96-AQueous One Solution Cell Proliferation Assay (Promega Corporation) as previously described and expressed as the MTS₅₀ values (27). Each assay was performed in triplicate.

Total RNA was extracted from 2×10⁸ cells of KYSE150 treated with 91b1 at 9.5 μg/ml (based on the MTS₅₀ value of 91b1 on KYSE150 with the signal of MTS cytotoxicity decreased by 50%) and DMSO (0.05%; MilliporeSigma) for 48 h using RNeasy Mini kit (Qiagen, Inc.). The cDNA microarray analysis and the associated quality control using Human Genome U133 Plus 2.0 arrays (Affymetrix; Thermo Fisher Scientific, Inc.) were performed at the Centre for Genomic Sciences of the University of Hong Kong according to the manufacturer's protocol as previously described (4). The signals of the differentially expressed genes in 91b1-treated KYSE150 were compared with the DMSO-treated KYSE150 control.

The total RNA of non-neoplastic cells, ESCC parental cells and 91b1-treated ESCC cells was extracted using RNeasy Mini kit (Qiagen, Inc.) as previously described (4). cDNA was synthesized from total RNA using the GoScript Reverse Transcription System (Promega Corporation) according to manufacturer's instruction. The expression level of Ccl5 in the tested cells was determined by qPCR analysis using Go Taq qPCR Master Mix (Promega Corporation) and Thermo Scientific PikoReal Real-Time PCR System (Thermo Fisher Scientific, Inc.) according to manufacturer's protocol. The synthesized cDNA (4 μl) was mixed with 10 μl of 2X qPCR Mastermix (Promega Corporation), 2 μl of 2 μM forward and reverse primers of either target gene or reference gene (β-Actin was applied as reference gene), and 12 μl of RNase free water to get a total volume of 20 μl in a PCR tube. All 20 μl of sample mixtures were transferred into the wells of PikoReal 96-well strips (n=3). qPCR reactions were carried out by PikoReal Real-Time PCR System (Thermo Fisher Scientific, Inc.). The thermocycling conditions were: Polymerase activation at 95°C for 2 min, then followed by 40 cycles of denaturation at 95°C for 15 sec, annealing and primer extension at 60°C for 1 min, then melt curve data were identified by gradually increasing temperature from 60-95°C until the fluorescent signal dropped to zero. Cq (cycle of quantification) of each sample was determined and recorded by the program PikoReal Software 2.0 (Thermo Fisher Scientific, Inc.). cDNA (~2 μg) produced by reverse transcription from the RNA was amplified using a specific Ccl5 and β-Actin gene primer pairs (IDT). Primers for Ccl5 were 5′-CGTGCCCACATCAAGGAG-3′ (forward) and 5′-GGACAAGAGCAAGCAGAAA-3′ (reverse). Primers for β-Actin were 5′-ACCTTCTACAATGAGCTGCG-3′ (forward) and 5′-CCTGGATAGCAACGTACATGG-3′ (reverse). Relative Ccl5 expression was determined by comparing with vehicle DMSO (0.05%) control, after being normalized with expression of β-Actin. For all the qPCR reactions, the relative expression of target genes in different samples were calculated and compared by using the 2^−ΔΔCq method. The expression level of target genes was normalized by the reference gene β-actin. Each independent experiments is conducted three times.

The calculation of 2^−ΔΔCq method was (37):

ΔCq of target gene=Cq of target gene-Cq of reference gene

Paraffin-embedded cell-line blocks of KYSE150, KYSE510, KYSE450, KYSE520 KYSE30, HKESC-3 HKESC-4, SLMT-1, NE-3, DMSO-treated (0.05%, 48 h) and 91b1-treated (6.5, 9.5 and 21 μg/ml 91b1) for 48 h KYSE150 cells were prepared from ~5×10⁶ cells of each respective cell line with formalin-fixation (37% formalin and 15% methanol at room temperature for 48 h). Flowing fixation, tissues were rinsed under running water for 3 h. Then the tissues underwent a grade dehydration process starting with 50% ethanol for 2 h, followed by 70% ethanol for 2 h, 80% ethanol overnight, 90 and 95% ethanol for 2 h each, and concluding twice in 100% ethanol for 2 h each time. Subsequently, the samples were placed in a mixture of ethanol:xylene (1:1) for 30 min and cleared in xylene for 30 min twice before being prepared for embedding in paraffin wax at 60°C twice for 1 h each time. An automated embedding system was employed to encapsulate the tissues in paraffin. Dewaxed paraffin sections (8 μm) of cell-line blocks and samples from patients with ESCC were immunostained using the streptavidin-biotin-peroxidase complex method. As pretreatment, microwave-based antigen retrieval was be performed in 10 mM citrate buffer (pH 6.0). CCL5 mouse monoclonal antibody (1 mg/ml; Abnova) was applied at dilution of 1:100 for overnight incubation at 4°C. Images of stained samples were captured under an inverted optical microscope (CKX41; Olympus Corporation) at magnification, ×400 and four fields of images of stained sections were examined for the percentage of positively stained cells in cytoplasm by ImageJ (v 1.54; National Institutes of Health). Immunostaining results of ESCC cell line were compared with that of non-tumor cell line NE-3. Immunohistochemical staining images of stained sections were examined and graded according to the percentage of positively stained neoplastic cells as previously described (4).

Protein expression levels of CCL5 in cells were measured using RANTES (CCL5) Human SimpleStep ELISA kit (Abcam; cat. no. ab174446) according to the manufacturer's instructions. Cells receiving no treatment, DMSO-treatment (0.05%) or 91b1-treatment in the concentration of 6.5, 9.5 and 21 μg/ml (with reference to the MTS₅₀ value) were collected by cell scrapper after 48 h. Total protein concentration of each sample was determined using a Micro BCA Protein Assay kit (Thermo Fisher Scientific, Inc.) according to manufacturer's manual for normalization.

The invasion ability of ESCC cells was evaluated using chambers with Matrigel-coated membrane (8-μm pore size; BD Biocoat; Corning, Inc.) in 24-well plate. The lower chamber was filled with RPMI 1640 medium (Gibco; Thermo Fisher Scientific, Inc.) containing 10% fetal bovine serum (FBS, Biosera) with recombinant human CCL5 protein (rhCCL5; Abnova) at concentration of 0, 50, 100 and 500 ng/ml. KYSE30 cells were cultured in 200 μl serum free RPMI 1640 medium in the upper chamber at a density of 2.5×10⁵ cells/ml. The same number of cells were cultured on an uncoated membrane (8-μm pore size) chamber as control. After 24 h, the uninvaded cells on the upper chamber were scraped off with a cotton swab. The transmembrane cells which migrated to the opposite side of the membrane were fixed in 100% methanol for 10 min and stained with 0.5% crystal violet solution (0.5 g crystal violet in 75 ml methanol and 25 ml ddH₂O) for 30 sec at room temperature after washing twice with phosphate buffered saline (PBS). The transmembrane cells were counted under microscope in five random fields at magnification of ×100. The percentage of invasion was calculated as follows:

A wound-healing assay was performed to evaluate cell migration and growth. KYSE150 cells (~1×10⁶) were cultured in a 6-well plate at 37°C with 5% CO₂ overnight to let the cells adhere and grow to 70-80% confluent monolayers. On the second day, the monolayer was gently scratched with a new 1 ml pipette tip across the center of the well to generate a wound area without changing the medium. After scratching, the well was gently washed twice with warm PBS buffer to remove detached cells, and the well was replenished with serum-starved medium or different concentrations of compound 91b. The cells were incubated at 37°C with 5% CO₂ again and observed by microscope (Olympus CKX41; Olympus Corporation) at different time points (0, 12, and 24 h after scratching) for image capture. The number of cells invaded across the wound area was counted by ImageJ (v 1.54, National Institutes of Health).

The comparative ΔΔCq method was applied for relative quantification in qPCR analysis. Statistical significance of the differences among groups in MTS cytotoxicity assay, qPCR analysis, ELISA, Transwell Matrigel invasion assay data were compared by two-tailed t test or one-way analysis of variance (ANOVA) followed by Dunnett's Correction using GraphPad Prism 7 (Dotmatics). P<0.05 was considered to indicate a statistically significant difference.

According to the pan-cancer analysis in Fig. 2A, Ccl5 was differentially expressed in several types of cancer, in which Ccl5 was notably high expressed in breast invasive carcinoma, esophageal carcinoma, glioblastoma multiforme, head and neck squamous cell carcinoma, kidney renal clear cell carcinoma, kidney renal papillary cell carcinoma and metastasis skin cutaneous melanoma, suggesting the possible roles of Ccl5 in tumorigenesis and tumor metastasis.

GO enrichment analysis results are in Fig. 2B. The most enriched GO molecular functions were identified as 'small GTPase mediated signal transduction', 'regulation of GTPase activity', 'lymphocyte differentiation', 'regulation of protein serine/threonine kinase activity', 'regulation of MAP kinase activity', 'cell-substrate adhesion', 'protein folding', 'wound healing', 'actin filament organization' and 'positive regulation of kinase activity'. KEGG pathway analysis results are in Fig. 2C, indicating that the DEGs were enriched in nine pathways, 'coronavirus disease', 'prion disease', 'human T-cell leukemia virus 1 infection', 'regulation of actin cytoskeleton', 'ribosome', 'lipid and atheroselerosis', 'non-alcoholic fatty liver disease', 'growth hormone synthesis, secretion and action', 'antigen processing and presentation' and 'viral myocarditis'.

A total of 1,579 interactions were obtained with interaction scores >0.4 using the STRING database. The PPI network was then constructed and presented with the Cytoscape (v. 3.9.0) platform (Fig. 2D). The top 10 hub genes included Itgb2, Pik3cg, Pik3r1, Cbl, Pten, Syk, Pik3cb, Ptprc, Met and Inpp5d. A total of four modules were obtained through MCODE analysis. The function of key proteins, such as PIK3, PTEN, and ATM were associated with metastases, IFNG, CD74, CD40, and CD3g were associated with immune response.

Immunostaining was employed to detect the cytoplasmic protein expression level of CCL5 in eight ESCC cell lines and non-tumor esophageal cell line NE-3. Upregulation of CCL5 was detected in 7/8 (87.5%) ESCC cell lines (KYSE510, KYSE450, KYSE520, KYSE150, HKESC-3, HKESC-4 and SLMT-1) when compared with NE-3 (Fig. 3A and B). In addition, high expression level of CCL5 was also observed in 20/26 (76.9%) of ESCC tissues and only 6/15 (40%) of non-neoplastic esophageal tissues. ESCC expressing CCL5 protein in high level was more frequently observed than non-neoplastic esophageal mucosa (P=0.018). Immunohistochemical staining showed high expression and low expression of CCL5 in ESCC (Fig. 3C).

The effect of CCL5 on the invasion ability of KYSE30 was examined by Transwell Matrigel invasion assay. As shown in Fig. 3E, no transmembrane cells were detected in the invasion assay after 24 h without the addition of recombinant human CCL5 protein (rhCCL5). Increasing numbers of transmembrane cells were detected with increasing rhCCL5 concentration. The percentage of invasion increased with the concentration of rhCCL5 (Fig. 3G), suggesting that CCL5 enhanced the invasion ability of ESCC cells. Moreover, to evaluate the cell migration and growth properties affected by compound 91b1, wound healing analysis was performed on KYSE510 cell line. A wound gap was created by scratching, and healing progress was captured at different time points. Cancer cells were treated with low (5 μg/ml) or high (10 μg/ml) dose of compound 91b1 (Fig. 3D and F). After 12-h, or 24-h incubation, fewer cells of compound 91b1 treatment groups migrated into the scratched area than the control groups, indicating the reduction of cell migration of the cancer cells following 91b1 treatment.

cDNA microarray analysis was performed to study the changes of gene expression caused by compound 91b1 in cancer cells. Density plots (Fig. 4A) showed the different expression profile of KYSE150 cells treated with compound 91b1 compared with a blank control. The fold changes of normalized signal intensity of each gene obtained from the microarray analysis were evaluated. The five most downregulated genes were Ccl5, Lumican, Ston1, Igfb5 and Cp while the five most upregulated genes were C7orf57, Zbed2, Clgn, Csf2 and Slc16a6 (Table I and Fig. 4B). Ccl5 was found to be downregulated in 91b1-treated KYSE150 cells with the highest fold change (-2.12 times), thus 91b1 was a promising anti-cancer compound to inhibit metastasis by targeting Ccl5.

The effect of 91b1 on mRNA expression and protein expression of CCL5 in ESCC cells were examined by qPCR analysis and ELISA. All the tested ESCC cell lines showed the reduction in CCL5 mRNA (Fig. 5A-D) and protein expression (Fig. 5E-H) after the treatment with 91b1, which is in agreement with the microarray results using KYSE150. In general, the suppressing effect of 91b1 on CCL5 expression was dose dependent. The results thus confirmed that CCL5 is one of the affected downstream candidates for the cytotoxicity induced by 91b1. The suppression effect of 91b1 on CCL5 protein expression was also demonstrated by IHC staining with CCL5 antibody on KYSE150 cells. The number of cells with cytoplasmic staining signals revealed the relative protein expression level of CCL5. Dark-brownish stained cells in the vehicle-control revealed the high CCL5 protein expression (Fig. 5I) with quantitative analysis (Fig. 5J). Fewer CCL5 positive cells were evident after being treated with increasing concentration of 91b1 for 48 h, suggesting that the protein expression of CCL5 was reduced with 91b1 treatment in a dose-dependent manner.

The anticancer effect of 91b1 on the four ESCC cell lines (KYSE30, KYSE150, KYSE450 and HKESC-4) and two non-tumor cell lines (NE-3 and 293) was evaluated using MTS cytotoxicity assay using CDDP as the positive control (Fig. 6). The MTS₅₀ values (concentration of tested compounds that had 50% inhibition on MTS activity; Table II). 91b1 showed stronger cytotoxic effect in ESCC cell lines and lesser cytotoxic effect in non-neoplastic cells (NE-3 and 293) than CDDP.