Microarray dataset GDS4382, GDS4515, GDS3756 and GDS2947 were downloaded from the Gene Expression Omnibus database (GEO DataSets). GDS4382 (16), based on GPL570 platform, consisted of 17 CRC tumor tissues and 17 normal tissues; GDS4515 (17), based on GPL96 platform, included 34 microsatellite-unstable colorectal tumor samples and 15 normal colonic mucosa samples; 22 rectal tumor samples and 20 normal rectal tissue samples were selected from GDS3756 (18) based on GPL2986 platform; the array data of GDS2947 (19), based on GPL570 platform, included 32 adenoma samples and 32 normal mucosa samples. Under the same experimental conditions, tumor samples and normal samples were divided into two groups for screening. Data Analysis Tools in GEO DataSets was used for identifying differentially expressed genes (DEGs). The cut-off criterion was set as P<0.05 and value means difference >2+ fold.

The prediction of regulator for CXCL11 was performed by miRWalk 1.0 (http://zmf.umm.uni-heidelberg.de/apps/zmf/mirwalk/index.html), a database that not only documents miRNA binding sites within the complete sequence of a gene, but also combines this information with a comparison of binding sites resulting from other existing miRNA-target prediction programs (20). A total of 10 established miRNA-target prediction programs (Diana-microT, miRanda, miRDB, miRWalk, RNAhybrid, PICTAR4, PICTAR5, PITA, RNA22 and TargetScan) were available in miRWalk. The potential targets of one miRNA are considered to be the genes that have been predicted by at least 5 programs (21).

To determine the expression of genes or miRNAs in CRC tissues, tumor and surrounding normal tissue samples were collected from 39 patients with CRC (23 men and 16 women; age range, 25–65 years; median age, 45.3 years). All the patients had undergone radical surgery at the Department of General Surgery, West China Hospital, Sichuan University, between May 2010 and September 2011. Prior to surgery, none of the patients had received chemotherapy or radiotherapy. The excised tumors and adjacent tissues were stored in liquid nitrogen for molecular biology experiments. In addition, peripheral blood was collected from all 39 patients and 26 healthy control subjects (16 men and 10 women; age range, 23–66 years; median age, 44.8 years). The separation of serum was performed by centrifugation at 1,000 × g for 10 min at 4°C. All procedures performed in studies involving human participants were in accordance with the ethical standards of the Institutional and/or National Research Committee and with the 1964 Helsinki Declaration and its later amendments or comparable ethical standards. Tissue and blood samples were collected from consenting individuals according to the protocols approved by the Ethics Review Board at Sichuan University.

Total RNA was extracted from tissue samples using E.Z.N.A^® Total DNA/RNA/Protein kit (Omega Bio-Tek, Inc., Norcross, GA, USA). Total RNA was extracted from serum samples using the GenElute™ Plasma/Serum RNA Purification Mini kit (Sigma-Aldrich; Merck KGaA, Darmstadt, Germany). RT for RT-qPCR was performed with a PrimeScript™ RT reagent kit (Takara Biotechnology Co., Ltd., Dalian, China) according to the manufacturer's instructions. The real-time PCR reaction system was prepared using SYBR^® Premix Ex Taq™ II (RR420A; Takara Biotechnology Co., Ltd.) and the qPCR assay was performed with the CFX96TM Real-Time PCR Detection System (Bio-Rad Laboratories, Inc., Hercules, CA, USA). The reaction system (20 µl) consisted of 10 µl SYBR Premix Ex Taq, 0.5 µl forward primer, 0.5 µl reverse primer, 2 µl cDNA and 7 µl double-distilled water (ddH₂O). The reaction protocol was as follows: Initial denaturation at 95°C for 5 min, followed by 40 cycles at 95°C for 15 sec, 60°C for 30 sec and 72°C for 10 sec. All the reactions were performed in accordance with the manufacturer's instructions. The primer sequences (miR-144, CXCL11, GAPDH, U6) for qPCR are listed in Table I. The relative expression of CXCL11 (reference gene GAPDH) and miR-144 (reference gene U6) was analyzed using the 2^−ΔΔCq method (22).

Using an immunohistochemical assay kit (BosterBio, Wuhan, China) according to the manufacturer's instructions, the tissue samples were fixed in 4% paraformaldehyde and embedded in paraffin, then cut into 5-µm sections. The slides were treated by a graded series of alcohol and finally rinsed with ddH₂O. After antigen retrieval, endogenous peroxidase activity was blocked with 3% hydrogen peroxide at room temperature for 10 min. Subsequently, the slides were blocked with 10% normal goat serum (C0265; Beyotime Institute of Biotechnology, Haimen, China) for 1 h, followed by incubation with 1:250 diluted CXCL11 primary antibody (ab9955, rabbit anti-human; Abcam, Cambridge, MA, USA) at 4°C overnight. After washing off the primary antibody, the second antibody (ab6721, goat anti-rabbit IgG; Abcam) was applied and incubated at room temperature for 1 h. After DAB staining (P0203; Beyotime Institute of Biotechnology) until the appearance of brown color, the nuclei were stained with hematoxylin (C0107; Beyotime Institute of Biotechnology) for 1 min. Subsequently, the slides were washed with ddH₂O, sealed with neutral gum and covered with coverslips. An optical microscope (TS100-F; Nikon Corporation, Tokyo, Japan) was used for cell observation: Brown staining was considered as positive; the slides were observed under a magnification of ×10 and ×40; five non-overlapping fields were randomly selected from each section and images were captured. The mean integrated optical density (IOD/area) of the images was measured by Image-Pro Plus v.6.0 software, which was used for semi-quantitative analysis.

Total protein was extracted with the E.Z.N.A^® Total DNA/RNA/Protein kit (Omega Bio-Tek Inc.). The BCA protein assay kit (Pierce Biotechnology, Rockford, IL, USA) was applied for protein concentration detection. The protein samples were mixed with loading buffer (P0015; Beyotime Institute of Biotechnology) and boiled for 10 min; then, 50 µg protein from each sample was subjected to 10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis and transferred onto a PVDF membrane (EMD Millipore, Billerica, MA, USA). The blot was blocked with 5% non-fat milk at room temperature for 2 h and then incubated with CXCL11 primary antibody (ab9955, rabbit anti-human, Abcam, USA) for 20 h at 4°C. After TBST washing, the membrane was incubated with secondary antibody (ab6721, goat anti-rabbit IgG; Abcam) at room temperature for 1 h. An ECL kit (P0018; Beyotime Institute of Biotechnology) was used for blot chemiluminescence and an exposing gel documentation system was applied for imaging the blots. Image Lab v.3.0 software was used to obtain and analyze the protein signal.

CXCLL11 in serum specimens was tested using a Human CXCL11 ELISA kit (ab187392; Abcam); 50 µl serum of each samples was added into appropriate wells for the detection. All the procedures were performed according to the manufacturer's instructions. The OD was measured at 450 nm with a microplate spectrophotometer (Multiskan™ FC Microplate Photometer; Thermo Fisher Scientific, Inc., Waltham, MA, USA).

293T and HCT116 cells in this study were provided by the Type Culture Collection of the Chinese Academy of Sciences (Shanghai, China), and cultured with DMEM high-glucose medium (HyClone; GE Healthcare Life Sciences, Logan, UT, USA) supplemented with 10% FBS (Shengong, Shanghai, China) at 37°C in 5% CO₂.

After growing to 50–70% confluence, HCT116 cells were transfected with agomiR-144 or agomiR-NC by using ExFect Transfection Reagent (Vazyme Biotech Co., Ltd., Nanjing, China) according to the manufacturer's instructions. AgomiR-144 and agomiR-NC were designed and chemically synthesized by Guangzhou RiboBio Co., Ltd., Guangzhou, China.

According to the bioinformatics analysis results, a fragment of the CXCL11 3′-untranslated region (UTR) containing the wild type (WT) or mutant (MT) seed regions of miR-144 was chemically synthesized in vitro. The products were inserted into Spe-1 and Hind III restriction sites of pMIR-REPORT luciferase reporter plasmids (Ambion; Thermo Fisher Scientific, Inc.). The constructs were co-transfected with agomiR-144 into 293T cells using ExFect^® Transfection Reagent (Vazyme Biotech Co., Ltd.); the empty plasmid was set as control. After 48 h of transfection, the cells were harvested. The luciferase activity was measured using a dual-luciferase reporter assay kit (Promega Corporation, Madison, WI, USA) and GloMax 20/20 luminometer (Promega Corporation). All the procedures were performed according to the manufacturer's instructions.

IBM SPSS software (v.20.0; IBM Corp., Armonk, NY, USA) was used for statistical analysis. Data are expressed as mean ± standard deviation. Comparisons between two groups were performed using the Student's t-test, and multi-group measurement data were analyzed using one-way analysis of variance. The post hoc test was performed using Student-Newman-Keuls and Least Significant Difference-t test. P<0.05 was considered to indicate a statistically significant difference.

To measure the expression of CXCL11 in CRC tissues, microarray datasets GDS4382, GDS4515, GDS3756 and GDS2947 were downloaded from GEO DataSets. The mean values of CXCL11 in CRC tissues were 3+fold higher compared with normal tissues (Fig. 1A); in microsatellite-unstable colorectal tumor tissues they were 2+fold higher compared with normal colonic mucosa tissues (Fig. 1B); in rectal tumor tissues they were 4+fold higher compared with normal rectal tissues (Fig. 1C); and in adenoma tissues they were 4+fold higher compared with normal mucosa tissues (Fig. 1D). All the above mentioned data supported the upregulation of CXCL11 in CRC tissues.

RT-qPCR was used to determine the levels of CXCL11 mRNA in tumor tissue and serum samples; western blotting and immunohistochemistry were used to detect the expression and localization of the CXCL11 protein in tumor tissue samples; ELISA was conducted to measure the concentration of the CXCL11 protein in serum samples. The data revealed that the levels of CXCL11 mRNA were significantly upregulated in CRC tissues (2+fold higher, P<0.05; Fig. 2A) and serum samples (3+fold higher, P<0.05; Fig. 2D) collected from CRC patients. Upregulation of the CXCL11 protein was detected in CRC tissues (Fig. 2B and C) and the localization of CXCL11 was in the cytoplasm and on the cell membrane (Fig. 2C). Increased expression of the CXCL11 protein was also found in serum samples from patients with CRC (Fig. 2E). The expression trend of CXCL11 in clinical specimens was similar to our bioinformatics analysis, indicating that increased CXCL11 expression may be implicated in CRC.

To search for the potential regulator of CXCL11, miRWalk was used for the bioinformatics predictions. Of the 10 the established miRNA-target prediction programs, 6 predicted that CXCL11 was a potential target of miR-144 (Fig. 3A). The miR-144 target site in the CXCL11 mRNA 3′UTR was shown in some programs (Fig. 3B-D).

The levels of miR-144 in the serum and tumor tissues of patients with CRC were analyzed by RT-qPCR. Compared with adjacent tissues, miR-144 in CRC tissues was significantly reduced (P<0.05; Fig. 4A). The expression of miR-144 was also downregulated in the serum of CRC patients compared with the control subjects (P<0.05; Fig. 4B). Taking the expression of CXCL11 into account, these results indicated that miR-144 may be a negative regulator of CXCL11.

To investigate the predicted interaction of miR-144 with CXCL11 mRNA, cell transfection and dual-luciferase reporter assay were used. HCT116 cells were selected for transfection with agomiR-144 or agomiR-NC in vitro. Compared with the agomiR-NC group, agomiR-144 significantly upregulated the expression of miR-144 (P<0.05; Fig. 5A), while it significantly downregulated the expression of CXCL11 (P<0.05; Fig. 5B and C) in HCT116 cells. WT and MT miR-144 binding sites were cloned into the pMIR-REPORT luciferase reporter plasmids (Fig. 5D) and co-transfected with agomiR-144 into 293T cells; the empty plasmid was used as control. Compared with the control group, transfection with agomiR-144 resulted in a significantly reduced fluorescence intensity in the WT group, but not in the MT group (Fig. 5E). These observations suggest that miR-144 may directly target the 3′-UTR of CXCL11 mRNA.