Introduction
Colorectal cancer (CRC) is the most common and second most deadliest malignant tumour of the digestive system worldwide [
1]. Every year, patients newly diagnosed with CRC account for 10% of newly diagnosed cancer patients worldwide. Similarly, CRC accounts for 10% of cancer-related deaths [
1]. In the past 20 years, with the progress of imaging technology and treatment strategies, the five-year survival rate of CRC has risen to 65%. However, the situations of patients with advanced CRC remain dismal. The two-year survival rate of patients with advanced CRC is only approximately 20–30%. Tumour metastasis is still the main cause of death for patients with this condition [
2,
3]. Therefore, reducing the proliferation and development of CRC cells has always been a difficult problem for medical workers.
In the past, research on the mechanism of tumour invasion and metastasis was mainly performed at the tissue and cellular levels. However, thanks to the rapid development of microarray and RNA sequencing (RNA-seq) technology, we began to carry out cancer research at the RNA level. In recent years, RNA helicases have attracted increasing attention from scientists. RNA helicases are enzymes that hydrolyse ATP to dissociate it from RNA and are mainly involved in RNA maturation, splicing and nuclear export processes [
4]. Depending on their function and mechanism, RNA helicases are divided into different families. The RNA helicase DEAD-box (DDX) family belongs to superfamily 2, which is named because of its highly conserved amino acid sequence (Asp-Glu-Ala-Asp/His) [
5]. There is a conserved core domain among DDX family members. This domain contains nine conserved motifs, including the most typical DEAD motif. These motifs endow DDX family members with ATP metabolism and RNA unwinding activities, which makes them RNA helicases. Many studies have proven that DDX family members play a crucial role in almost all steps of mRNA formation and translation [
6]. Moreover, an increasing number of studies have proven that members of the DDX family are closely related to the proliferation and metastasis of tumour cells. For instance, DDX1 can affect the occurrence and metastasis of breast cancer, cervical cancer and colorectal cancer [
7‐
11]. DDX5 is related to the progression of different tumours, such as breast cancer, colon cancer and multiple myeloma [
12‐
20]. DDX10 promotes the proliferation of breast cancer, osteosarcoma and ovarian cancer cells [
21‐
23]. Interestingly, we found that many members of the DDX family affect the invasion and metastasis of CRC [
11,
24‐
28], but no study has focused on the effect of DDX10 on the prognosis of CRC. Because the structure and function of DDX family members are similar, we speculated that DDX10 is a key factor affecting the prognosis of CRC.
In our study, we analysed the expression data of CRC samples from the Gene Expression Omnibus (GEO) and The Cancer Genome Atlas (TCGA) databases to determine whether the mRNA expression of DDX10 in CRC tissues is higher than that in normal tissues. Moreover, the clinical samples we collected from The Second Affiliated Hospital of Soochow University were used for immunohistochemistry (IHC), which could detect the differences in DDX10 expression at the protein level. Subsequently, we performed cytological experiments and animal experiments to explore the role of DDX10 in CRC cells. Furthermore, we performed Gene Ontology (GO)/Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis and protein–protein interaction (PPI) network analyses, which enabled us to better understand how DDX10 affects the invasion and metastasis of CRC. Finally, we predicted the interacting protein of DDX10 by LC–MS/MS and verified it by Co-IP and qPCR.
Materials and methods
Patients and specimens
Between January and December 2014, we collected 35 pairs of cancer and paracarcinoma tissues from patients with CRC at the Gastrointestinal Surgery Department of the Second Affiliated Hospital of Soochow University (Jiangsu Province, China). Our study was approved by the ethics committee of Soochow University. We obtained informed consent from every patient. All specimens were stored at − 80 °C.
Immunohistochemistry
Colorectal cancer and paracarcinoma tissues were dewaxed in xylene and rehydrated through different gradients of alcohol. Then, we placed paraffin in 3% H2O2 for 15 min at 22 °C. Next, the slide was heated in citrate buffer. After washing several times with PBS (pH = 7.2), the slide was placed into the solution with a primary antibody against DDX10 (dilution 1:50, 17857–1-AP, Proteintech, USA) at 4 °C for more than 12 h. Then, we added secondary antibody at 22 °C for 10 min. Finally, the slide was stained after the addition of 3,3′-diaminobenzidine (DAB) solution.
Differential expression analysis
The TCGA is an open public tumour information database. We selected 50 pieces of CRC data, including that for cancer and paracarcinoma tissues, from the TCGA database to compare the mRNA expression of DDX10. The GEO database (
http://www.ncbi.nlm.nih.gov/geo) [
29] is an international public database that stores abundant high-throughput functional genomics data. The gene expression dataset analysed in our study was GSE74604 downloaded from the GEO. GSE74604, based on the GPL6104 platform (Illumina humanRef-8 v2.0 expression beadchip), contained data for 30 paired normal and tumour colorectal samples. We analysed the differences in gene expression data for 60 samples. Then, we also used Gene Expression Profiling Interactive Analysis 2 (GEPIA2,
http://gepia.cancer-pku.cn/) [
30] and ONCOMINE (
https://www.oncomine.org/) to verify the results. GEPIA2 is a web tool that includes RNA-seq data for a large number of tumour and normal samples from the TCGA and Genotype-Tissue Expression (GTEx) projects. Gene expression data from ONCOMINE are freely available to the public. ONCOMINE is meant to facilitate related studies by providing genome-wide expression array data.
Real-time PCR
Total RNA was extracted by SuPerfecTRI™ (Cat#3101-100) (Shanghai Pufei Biotech Co., Ltd). This RNA was then reverse-transcribed to cDNA by M-MLV Reverse Transcriptase (M1705) (Promega, Beijing, China). SYBR Master Mixture (TAKARA, Tokyo, JP) was used for real-time PCR. We used the expression of endogenous GAPDH as a control to take standardized quantification. The DDX10 primer sequences were as follows: forward, 5′-TTGAGGTTCTCCGAAAAGTAGG-3′ and reverse, 5′-ACATTTGGAGGTCGGTAGCAT-3′; the GAPDH primer sequences were as follows: forward, 5′-TGACTTCAACAGCGACACCCA-3′ and reverse, 5′-CCATTGCCCGTGTTCTCACA-3′; the 60S ribosomal protein L35 (RPL35) primer sequences were as follows: forward, 5′-TGACTTCAACAGCGACACCCA-3′ and reverse, 5′-GCTCCTTCCGCTGCTGCTTC-3′.
Cell lines and lentiviral transduction
Our study used the human CRC cell lines RKO, HCT116, SW480 and LoVo, which were purchased from GeneChem (Shanghai, China). The cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM; HyClone, Thermo Fisher Scientific) containing 10% foetal bovine serum (FBS; Gibco‐Invitrogen Corp.) at 37 °C with 5% CO2 in strict accordance with the standard. All lentiviruses used in this research were purchased from GeneChem (Shanghai, China). We transfected short hairpin RNA (shRNA) lentivirus targeting DDX10 and empty vector (controls) into HCT116 cells and RKO cells in strict accordance with the official technical instructions. The DDX10 RNAi sequence was 5′-GATGTGAGCAAGTTACCTATA-3′.
Celigo cell growth assay
HCT116 cells and RKO cells were plated in 96-well plates (1500 cells/well). Next, we cultured the cells at 37 ℃ with 5% CO2. During the next five days, we counted the number of cells with a Celigo Imaging Cytometer (Nexcelom Bioscience, Lawrence, MA, USA) every day. All experiments were repeated at least three times.
First, we collected HCT116 cells and RKO cells. Next, those cells were added to complete medium (10% FBS) and placed into 6-well plates (800 cells/well). Then, we cultured the cells in standard conditions for 8 days and changed the medium every three days. Then, they were fixed for 30–60 min with 4% paraformaldehyde. Last, we added 1000 μl/well crystal violet at room temperature. Approximately 10–20 min later, we counted the number of cells under a fluorescence microscope (Olympus).
We used annexin V-APC (eBioscience) to stain these cells to analyse apoptosis in strict accordance with official technical instructions. The cells were plated in 6-well plates (5 × 105 cells/well). Then, we collected the cells and washed them with PBS. After the cells were resuspended, we added 10 μl annexin V-APC to the suspensions. Finally, we cultured the cells at 22 °C and protected them from sunlight. After 15 min, the cells were sorted by flow cytometry (C6 PLUS, BD).
MTT assay
MTT assays were used to observe the proliferation of cells. All operations were in strict accordance with the standard. First, we placed HCT116 or RKO cells with DDX10 knockdown and negative control cells into 96-well plates (1500 cells/well) in triplicate. Then, 100 μl medium was added to every well. Finally, a microplate reader (M2009PR, Tecan Infinite) was used to quantify the colour change (490 nm) at 24 h, 48 h, 72 h, 96 h and 120 h.
Wound-healing assay
First, we harvested HCT116 cells and RKO cells (negative control and DDX10 knockdown). Then, the cells were cultured in 96-well plates. After the confluence reached 90%, we aspirated the medium of every well and used PBS to wash the wells three times. Next, we used a pipette tip to gently draw a straight line down the wells of a 96-well plate. Finally, medium was added to every well after three washes with PBS. After 0 h, 24 h and 72 h, we used the Celigo platform to scan the plate and analyse the migration area.
Transwell migration and invasion assays
Transwell assays were used to assess the migration and invasion abilities of HCT116 cells and RKO cells. HCT116 cells or RKO cells (1 × 105) in 100 μl medium that had been cultured in serum-free medium were added to the upper chamber, and 600 μl culture medium (30% FBS) was added to the lower chamber. Then, the plate was incubated at 22 °C overnight. Finally, we dyed the cells on the lower surface of the membrane to analyse cell migration. The Transwell invasion assay was carried out as described above, but Matrigel was added to the bottom of the upper chamber.
Animal studies
The animal experiment of our study used 5-week-old female BALB/c nude mice (GemPharmatech, Co., Ltd., China) weighing between 16 and 19 g. Twenty nude mice were randomly divided into two groups with 10 mice in each group. Two groups (negative control and DDX10 knockdown) of HCT116 cells (stably expressing luciferase) (2 × 106 cells/100 μl) were injected into the mice via the tail vein. Then, we used an in vivo imaging system (Lumina LT, Perkin Elmer, USA) to observe tumour metastasis every week. We anaesthetized the mice before performing imaging monitoring. First, the mice were intraperitoneally injected with D-luciferin (15 mg/ml) at a concentration of 10 μl/g. After 15 min, the mice were anaesthetized by intraperitoneal injection of 0.7% pentobarbital sodium at a concentration of 10 μl/g. After a few minutes, the mice were anaesthetized and placed under bioluminescence imaging for imaging monitoring. All mice were cultured in a specific-pathogen-free (SPF) culture environment. At the same time, we also measured and recorded the tumour size.
Mass spectrometry and protein identification
Briefly, we used a scraper to harvest RKO cells that stably expressed Flag-DDX10 and lysed the cells with lysis buffer. The reference dosage for Co-IP was 400 µl FLAG beads to collect 15,000 μg of protein. First, we used weak lysis buffer to wash the beads four times. Then, we carried out SDS-PAGE and placed the gel into Coomassie brilliant blue R250 solution for staining. Next, we collected the protein bands and used LC–MS for analysis. The original graph file (.raw) exported from Q Exactive was changed into a “.mgf ” file by Proteome Discoverer 2.1 (Thermo Fisher Scientific), and then we delivered these data to the MASCOT2.6 server for database searching by built-in facilities. Then, we received the search files (.dat) that had been processed from the MASCOT server. Finally, we selected candidate peptides according to the standard of FDR < 0.01. The protein database used in this analysis was UniProt_HomoSapiens_20386_20180905.
LinkedOmics
LinkedOmics (
http://www.linkedomics.org/) [
31] is an unique platform for biologists and clinicians to access, analyze and compare cancer multi-omics data within and across tumor types. In our study, we used the “colorectal adenocarcinoma (TCGA_COADREAD)” dataset to analyse the correlation between E2F family and DDX10 or RPL35.
Co-immunoprecipitation assay
First, we collected RKO cells that overexpressed DDX10 after washing them three times with PBS. Then, these cells were added to precooled cell lysis buffer. Next, the final cell lysates were sonicated and centrifuged at 14,000 rpm for 10 min. The lysates were mixed with preblocked Flag beads (A2220, Sigma) at 4 °C overnight. After Co-IP, the Flag beads were washed as follows. First, we washed with lysis buffer three times. Then, we added 6 × loading buffer and left the samples at room temperature for 10 min. Finally, the cell lysates were mixed with 5 × sodium dodecyl sulfate (SDS) loading buffer and then boiled for 10 min. Proteins were eluted from the beads and separated by western blotting.
Western blot analysis
We used a 15% SDS-PAGE gel to isolate proteins, and the proteins were transferred to a PVDF membrane. The primary antibodies used in our study were mouse anti-GAPDH (1:5000; Santa Cruz, USA), rabbit anti-RPL35 (1:100; ABclonal, China), rabbit anti-DDX10 (1:300; Proteintech, USA), mouse anti-FLAG (1:2,000; Sigma, USA), rabbit anti-RPL18 (1:100; ABclonal, China), rabbit anti-HNRNPU (1:100; ABclonal, China), rabbit anti-NCL (1:100; ABclonal, China), rabbit anti-PRMT5 (1:100; ABclonal, China), rabbit anti-E2F1 (1:100; ABclonal, China), rabbit anti-E2F4 (1:100; ABclonal, China), rabbit anti-E2F8 (1:100; ABclonal, China), goat anti-rabbit IgG (1:2,000; CST, USA), and goat anti-mouse IgG (1:2,000; CST, USA).
Enrichment analyses and PPI network construction
GO and KEGG analyses are important bioinformatics tools for annotating genes and researching gene functions, biological processes and signalling pathways [
32‐
34]. In our study, we analysed the genes that were downregulated by DDX10 antibody using R version 4.0.5. The PPI network for these genes was predicted using the Search Tool for the Retrieval of Interacting Genes (STRING;
https://string-db.org/) online database [
35].
Gene set enrichment analysis (GSEA)
The RNA-seq data from the COAD and READ datasets were obtained from the TCGA portal (
https://portal.gdc.cancer.gov/). Then, we divided the TCGA samples into two groups according to the expression of DDX10 and analysed these two groups of data by GSEA. Finally, we screened the results with the criterion of normalized enrichment score (NES) ≥ 1.0 and adjusted P value < 0.25.
Tumour immune estimation resource (TIMER) database analysis
TIMER (
https://cistrome.shinyapps.io/timer/) [
36,
37] is an interactive website offering comprehensive analysis of immune infiltration among different types of cancer. TIMER uses various immune deconvolution methods to calculate the immune infiltrate abundances. Moreover, the subscriber can obtain high-quality figures. In this work, we searched “RPL35; COAD” in the “gene module” and performed an analysis of immune infiltration.
Statistical analysis
We used Prism version 8.0 software (GraphPad Software Inc.) to analyse all the data in this study. Unpaired t tests were used to compare two groups, and one-way ANOVA was used to compare multiple groups. The screening criteria for all data was P value < 0.05. All of the trials were repeated at least three times.
Discussion
In recent years, CRC has become the second leading cause of death in cancer patients. The ultimate cause of death in CRC patients is usually tumour metastasis. Therefore, it is very important to explore the mechanism underlying the metastasis of CRC cells. Many DDX family members, including DDX1, DDX3, DDX5, DDX17, DDX27 and DDX56, have been confirmed to be closely related to CRC [
11,
24,
28,
38‐
40]. Moreover, through the research of Patrick Linder et al. [
5] we know that the members of the DDX family are similar in structure and function. Hence, we reasonably speculated that DDX10 plays a key role in the invasion and metastasis of CRC cells.
DDX10 is a member of the DDX family that is characterized by the presence of an Asp-Glu-Ala-Asp (DEAD) motif, and its main function is to mediate the biogenesis of ribosomes [
41]. However, it was not until the late 1990s that DDX10 attracted the attention of biologists [
41,
42]. Our pancancer analysis results showed that DDX10 is highly expressed in a variety of cancer tissues; therefore, DDX10 is likely to be a key gene in many cancers. Previous studies have shown that the fusion of NUP98 and DDX10 may lead to leukaemia [
43]. Through gene rearrangement studies, Dr. Jiao confirmed that DDX10 is a potential oncogene affecting the growth and proliferation of breast cancer cells [
21]. Recently, it was confirmed that the expression of DDX10 is significantly higher in osteosarcoma patients [
22]. Interestingly, DDX10 plays an antitumour role in the development of ovarian cancer. Studies have shown that decreased expression of DDX10 promotes the proliferation of ovarian cancer through the Akt/NF-kB pathway [
23]. In our study, we first reported that DDX10 was upregulated in CRC and was closely associated with CRC clinical stage and prognosis. To the best of our knowledge, this is the first study on the function of DDX10 as a carcinogenic driver and prognostic biomarker for CRC.
In our study, the data from multiple databases were used to verify that the expression of DDX10 in tumour tissues was evidently higher than that in normal tissues, and the difference was also obvious in tumour-node-metastasis (TNM) stage and Dukes’ stage. Then, we used a lentiviral transfection method to knockdown and overexpress DDX10 to carry out cytological experiments and to construct a tumour metastasis model in mice, which helped us explore the effect of DDX10 on the growth, invasion and metastasis of CRC cells. These results showed that the growth of CRC cells was significantly inhibited after DDX10 knockdown and that the invasion ability was also significantly reduced. The decrease in the number and total weight of metastatic tumours in mice injected with shDDX10-CRC cells also fully demonstrated that the expression of DDX10 promoted the migration of CRC. The above results strongly suggest that DDX10 is a key gene regulating the invasion and metastasis of CRC and that DDX10 could be used as a new target for CRC treatment.
To better comprehend the mechanism of DDX10 in CRC, we carried out LC–MS/MS analysis and employed Co-IP, q-PCR and western blotting to verify the interaction. Fortunately, we found that DDX10 and RPL35 were strongly coexpressed. RPL35 encodes a ribosomal protein that is a component of the 60 s subunit. The 60 s subunit and 40 s subunit make up ribosomes that catalyse protein synthesis. RPL35 belongs to the L29 family of ribosomal proteins. These proteins of this family are usually associated with the activation of oncogenes and anti-oncogenes. However, there are few studies on RPL35. At present, studies on the identified diseases related to RPL35 mainly focus on Diamond-Blackfan anaemia (DBA) [
44]. In addition, one study speculated that the expression of RPL35 could predict lymph node metastasis in patients with early-stage cervical carcinoma [
45]. A recent study reported that lncNB1 promotes neuroblastoma tumorigenesis by interacting with RPL35. This study also identified that RPL35 is a key factor for promoting E2F1 protein synthesis [
46]. E2Fs are a classic group that regulate the occurrence of cancer and the cell cycle. Moreover, the GSEA results showed that the high expression of DDX10 is closely related to the cell cycle and E2F targets. Therefore, we speculated whether the E2F pathway is a key pathway in regulating CRC after DDX10 alternatively splices the mRNA of RPL35. To confirm our conjecture, we observed the relationship between DDX10, RPL35 and E2Fs in LinkedOmics. As expected, both DDX10 and RPL35 were closely related to E2F1, E2F2, E2F3, E2F4 and E2F5. Consequently, we believe that the E2F pathway is very likely to play an important role in the DDX10-RPL35-CRC regulatory pathway. Alternative splicing of mRNA is an important way to regulate gene expression in cancer development. As an RNA helicase, DDX10 is likely to participate in the splicing of the mRNA of RPL35. As we expected, high expression of DDX10 increased the incidence of AT (Alternate terminator) in mRNA of RPL35. Interestingly, through the TIMER database, we found that there was a strong correlation between the expression of DDX10 and the number of infiltrating immune cells, such as CD4 + T cells, CD8 + T cells and macrophages, in COAD. Based on this discovery, we hypothesize that DDX10 may mediate the immune response in the development of CRC, which must be of great significance for the immunotherapy of CRC.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.