Background
Colorectal cancer (CRC) is one of the most malignant cancer types around the world due to its high morbidity and mortality [
1,
2]. Many risk factors for CRC have been identified, including smoking, obesity, unhealthy diet,
Helicobacter pylori infection, physical inactivity and precancerous lesions [
3,
4]. Among these various causes of CRC, the aberrant activation or upregulation of oncogenes (e.g., KRAS [
5] or EGFR [
6]) and loss of function or downregulation of tumour suppressors (e.g., PDCD4 [
7] or TIA1 [
8]) are central. A better understanding of the underlying mechanism for the abnormal development of CRC is vital to the diagnosis, treatment, prognosis and prevention of this disease.
RNA binding proteins (RBPs) are heterogeneous sets of proteins with essential roles in RNA metabolism and post-transcriptional gene regulation [
9]. Scientists have identified more than 500 RBPs, some of which are tightly linked to the initiation and progression of human cancers [
9,
10]. Hu antigen R (HuR), also known as embryonic lethal abnormal vision 1 (ELAVL1), is one of the most famous cancer-related RBPs [
11‐
13]. HuR mainly localises to the nucleus, until the cell receives one of various stimulations that promote the translocation of HuR to the cytoplasm [
14], where HuR uses three RRMs (RNA recognition motifs) to bind UTRs (untranslated regions) of downstream mRNAs at their AREs (AU-rich elements). Through this mechanism of interaction, HuR stabilizes target mRNAs or promotes their translation, yet HuR occasionally represses the translation of some targets [
12,
14]. A considerable number of target mRNAs of HuR encode proteins essential for cell survival and proliferation (e.g., CCNA [
15], HIF1A [
16], COX-2 [
17] and VEGF [
18]), and therefore HuR plays an oncogenic role in the development and progression of various cancers [
11‐
14]. For example, HuR promotes tumour cell growth by stabilizing Bcl-2 in glioblastoma [
19]. In breast cancer, the binding of HuR to CCNE1’ 3’-UTR significantly increases the mRNA stability and protein half-life of CCNE1, thus promoting breast cancer proliferation [
20]. In addition, HuR affects the metastasis of oral cancer cells [
21]. For CRC, the roles of HuR have also been intensively investigated. HuR was reported to be upregulated in CRC [
22‐
24] and stabilizes many oncogenes (e.g., COX-2 [
24], VEGF [
25] and IL-8 [
25]), leading to enhanced CRC cell growth and tumourigenicity. Another study found a robust correlation between increased cytoplasmic HuR levels with COX-2 expression and colon cancer stage [
26]. In a nude mouse model of CRC, HuR significantly promotes xenografted tumour growth [
22]. In summary, these studies support the oncogene role of HuR in CRC. However, the underlying mechanism for the aberrant expression of HuR in CRC is poorly understood.
Among various regulatory mechanisms for gene expression, microRNAs (miRNAs) are highlighted as a prominent and intriguing one due to their extensive expression and functions in widespread organisms and biological activities [
27], including tumourigenesis [
28]. During tumourigenesis, many miRNAs undergo changes in expression, thus negatively regulating their cancer-related target genes to affect tumour phenotypes. These miRNAs are referred to as oncomiRs or tumour-suppressive miRNAs [
28]. miR-22 is known as one of the most important tumour-suppressive miRNAs in many different cancer types [
29,
30]. In hepatocellular carcinoma, miR-22 suppresses cell proliferation and tumourigenicity and is correlated with patient prognosis [
31]. In breast cancer, miR-22 inhibits cell invasion and migration by targeting Sp1, CD147 and GLUT1 [
32,
33]. In gastric cancer, miR-22 inhibits both tumour proliferation and metastasis by targeting MMP14 and Snail [
34]. For CRC, miR-22 even has a more profound tumour-suppressive effect. miR-22 is significantly downregulated in CRC tissue compared with that in normal adjacent mucosa [
35] and improves 5-FU and paclitaxel sensitivity in chemotherapy [
36,
37]. Overexpression of miR-22 inhibits HIF-1α and VEGF, thus suppressing CRC cell angiogenesis [
38]. miR-22 is also activated by vitamin D and exerts anti-proliferative and anti-migratory roles in CRC cells by targeting TIAM1, MMP-2 and MMP-9 [
39,
40]. Although miR-22 shows vital significance in CRC, the exact mechanism through which miR-22 influences CRC progression is far from understood.
In this study, we showed that upregulated HuR functions as a potent oncogene in promoting CRC proliferation and migration and is a target gene of miR-22. We also found that miR-22 inhibits CRC cell proliferation and migration in vitro and decelerates xenografted tumour growth in vivo by targeting HuR. Moreover, the onco-transcription factor Jun was found to suppress miR-22 expression at the transcriptional level. Thus, the Jun/miR-22/HuR regulatory axis may contribute to tumourigenesis of colorectal cancer.
Methods
Tissue samples
CRC tissues were collected from patients who underwent surgical resection at the Affiliated Drum Tower Hospital of Nanjing University Medical School (Nanjing, China). All patients signed consent letters and all manipulation of the tissues was approved by the Ethics Committee of Nanjing University. After surgery, the tissue samples were immediately frozen in liquid nitrogen and stored at −80 °C. All experiments were performed in accordance with The Code of Ethics of the World Medical Association (Declaration of Helsinki) and the guidelines of the Nanjing University. Clinical features of the patients are listed in Additional file
1: Table S1.
Cell culture and transfection
All cell lines used in this study were purchased from the Shanghai Institute of Biochemistry and Cell Biology, Chinese Academy of Sciences (Shanghai, China) and verified by short-tandem repeat (STR) profiling. All cells were cultured in the appropriate medium (RPMI-1640 for NCM460, SW480, HT29, HCT15 and HCT116; L-15 for SW620; DMEM for Caco2; and F-12 K for LOVO) supplemented with 10% FBS (Gibco, Carlsbad, CA, USA) in a humidified atmosphere with 5% CO
2 at 37 °C. Lipofectamine 2000 (Invitrogen, Carlsbad, CA, USA) was used for transient transfection of small RNA oligos and plasmids. For miRNA overexpression or knockdown, miRNA mimics or inhibitors (GenePharma, Shanghai, China) were used, respectively. For protein overexpression or knockdown, gene-specific overexpression vectors (FulenGen, Guangzhou, China) or siRNAs (GenePharma) were used, respectively. The siRNA sequences are listed in Additional file
2: Table S2.
Online database analysis
Targetscan [
41] (
http://www.targetscan.org/vert_71/) was used to predict potential miRNAs that could target HuR. Oncomine database (
https://www.oncomine.org/resource/login.html) [
42] was utilised to analyse the HuR expression level in CRC patients from a TCGA cohort. We adopted a Cancer vs. Normal Analysis to compare the expression levels of HuR in normal colon and rectum with those of colon adenocarcinoma and rectal adenocarcinoma. To explore the association between HuR or miR-22/miR-129 expression levels and the life expectancy of CRC patients, we downloaded RNA-Seq raw data and survival data of CRC patients from the TCGA data portal (
http://cancergenome.nih.gov/). We utilised Kaplan-Meier curves to compare overall survival differences between “high” and “low” expression groups and calculated
p values using the log-rank test in the survival package in R. To predict transcriptional factors that could affect miR-22, JASPAR (
http://jaspar.binf.ku.dk/) [
43] and SABiosciences [
44] (
http://www.sabiosciences.com/chipqpcrsearch.php) were used.
Protein isolation and western blot
Total protein was extracted with RIPA lysis buffer (Beyotime, China) supplemented with the proteinase inhibitors PMSF (Roche, USA) and PI (Thermo, USA). Proteins were separated by 10% SDS-PAGE (Bio-Rad, USA). GAPDH was used as an internal control. Antibodies against HuR and GAPDH were purchased from Santa Cruz Biotechnology (sc-5261 and sc-365,062, respectively) and the antibody against Jun was purchased from CST (#9165).
RNA isolation and qRT-PCR
TRIzol reagent (Sigma, St. Louis, USA) was used for total RNA extraction. TaqMan miRNA Assay Primers (Applied Biosystems, USA) or oligo d(T)18 primers (TaKaRa, Japan) were used for reverse transcription of miRNAs and protein-coding genes, respectively. To generate fluorescence signal in qRT-PCR, TaqMan miRNA Assay Probes (Applied Biosystems, USA) and SYBR Green dye (Ambion, Carlsbad, CA, USA), combined with gene-specific primer pairs, were used for miRNA and protein-coding gene quantification, respectively. After the qRT-PCR procedure, we set a fixed threshold for the cycle threshold (C
T) data, and the mean C
T was determined from triplicate reaction wells. U6 snRNA or GAPDH was used as an internal control (IC) for miRNAs or protein-coding genes, respectively, and the relative change in the level of target genes (TGs) normalised to IC between experimental groups (EGs) and the control group (CG) was calculated with the eq. 2
-ΔΔCT, in which ΔΔC
T = (C
T TG− C
T IC)
EG − (C
T TG − C
T IC)
CG. All sequences of the primers used are listed in Additional file
2: Table S2.
Pull-down assay
The pull-down assay was carried out according to a previously described protocol [
8]. Briefly, a DNA probe complementary to HuR mRNA and labelled with biotin at the 3′ terminal, was synthesised to pull down HuR mRNA. A scrambled biotinylated probe was used as a negative control (Genescript, Nanjing, China). The probes were incubated with streptavidin-coated magnetic beads (New England BioLabs, USA) and then with SW480 lysate. After incubation, beads were washed and treated with Trizol reagent to extract RNA. The sequence of the probe is listed in Additional file
2: Table S2.
Luciferase assay
Luciferase vectors used in this study were purchased from Genescript (China). Briefly, for miRNA binding site tests, pMIR-report luciferase vectors containing binding sites for miR-22 or miR-129 on HuR’s 3’-UTR were constructed. We also purchased mutant plasmid to test binding specificity. The miR-22 binding site was mutated from GGCAGCT to CCGTCGA, and the binding site of miR-129 was mutated from CAAAAA to GTTTTT. For the miR-22 promoter assay, miR-22 promoter regions containing different Jun binding sites were inserted into pGL3 basic reporter vectors (Promega, USA). When transfecting SW480 with luciferase vectors and small RNA oligos, we also co-transfected the cells with a β-galactosidase (β-gal) expression vector (Ambion) as a control. Luciferase activity was tested using a luciferase assay kit (Promega, USA).
Cell proliferation assay
To measure the proliferation rate of SW480, we conducted CCK-8 and EdU assays according to protocols described elsewhere [
8]. Briefly, SW480 was seeded in 6-well plates and transfected with small RNA oligos or plasmids. At 24 h after transfection, cells were harvested and reseeded in 96-well plates for CCK-8 or 48-well plates for EdU assays, respectively. For the CCK-8 assay, Cell Counting Kit-8 (Dojindo, Japan) was added into cells at the following time points: 12, 24, 36, 48 and 60 h after reseeding. After incubation for 2 h, the absorbance was measured at a wavelength of 450 nm. For the EdU assays, an EdU assay kit (RiBoBio, China) was used to determine the proliferation rate of cells according to the manufacturer’s instructions.
Cell migration assay
SW480 was transfected with small RNA oligos or plasmids. After 24 h, cells were resuspended in FBS-free RPMI-1640 medium and reseeded on the upper surface of 24-well Millicell plates (Millipore, USA). Cells were allowed to migrate across the 8-μm membrane toward medium with 20% FBS for 24 h. Then, the cells were fixed with 4% paraformaldehyde and dyed with 0.5% crystal violet. Nonmigrating cells were removed using a cotton swab. The migrant cells were blindly counted under a light microscope (BX51 Olympus, Japan).
Chromatin immunoprecipitation (ChIP) assay
The ChIP assay was performed using a commercial kit (Beyotime, Shanghai, China) according to the manufacturer’s instructions. An antibody against Jun was used to immunoprecipitate Jun-chromatin complexes. Anti-IgG (Santa Cruz, USA) served as a negative control. The ChIP products were amplified by PCR and then separated on 1.5% agarose gels. The primers for amplification are listed in Additional file
2: Table S2.
Mouse experiments
To explore tumour growth in vivo, SW480 overexpressing or knocked-down for the corresponding small RNA or protein and control cells were injected into nude mice (purchased from the Model Animal Research Center of Nanjing University) in their armpits. Mice were sacrificed after 25 days or 30 days, and tumours were removed for RNA and protein extraction, haematoxylin and eosin (H&E) staining or immunohistochemical (IHC) staining. All animal experiments complied with the ARRIVE guidelines and were carried out in accordance with the National Institutes of Health guide for the care and use of Laboratory animals (NIH Publications No. 8023, revised 1978) and the guidelines of the Nanjing University.
Statistical analysis
All experiments were performed at least in triplicate, and each individual experiment was repeated several times. Student’s t-test was used to analyse differences between two groups. P values less than 0.05 were considered statistically significant.
Discussion
HuR is a representative RNA binding protein that plays vital roles in CRC tumourigenesis [
22‐
24]. In this study, significant upregulation of HuR was observed in CRC tissues compared with that in adjacent normal tissues, and high HuR levels predicted a lower survival rate among CRC patients. The oncogenic roles of HuR in CRC were also investigated, and the results indicated that HuR promoted CRC cell proliferation and migration in vitro and accelerated CRC tumour growth in vivo. Interestingly, the HuR mRNA and protein levels changed inconsistently in CRC samples, raising the possibility that HuR was regulated by aberrantly expressed miRNAs at the post-transcriptional level in CRC. Combining bioinformatics predictions and in vitro validation, miR-22 and miR-129 were demonstrated to be upstream repressors of HuR by directly binding to its 3’-UTR.
miRNAs are closely involved in CRC tumourigenesis. In every stage of CRC, there are many miRNAs that have been shown to have altered expression and are thus involved in the regulation CRC cancer hallmarks [
49]. Among these myriad CRC-related miRNAs, miR-22 is one of the most important. miR-22 can affect various CRC phenotypes, including proliferation, migration, chemoresistance, apoptosis and angiogenesis [
37‐
40]. Here, miR-22 was found to be markedly reduced in CRC, and lower miR-22 expression predicted a shorter life expectancy. miR-22 functioned as a tumour-suppressive miRNA in CRC to inhibit CRC proliferation and migration and tumour growth by targeting HuR. Our results highlighted the importance of miR-22 and HuR in CRC, and noted the possibility that targeting miR-22 or HuR might be a practical way to treat CRC in clinical environments. Kota et al. systemically delivered tumour-suppressive miR-26a in a mouse model by adeno-associated virus (AAV) to successfully treat hepatocellular carcinoma [
50]. This strategy is also applicable to miR-22. For HuR intervention, besides administration of miR-22, a specific small molecule inhibitor or siRNA [
51,
52] should be a practical and efficient treatment. More attention should be payed to the miR-22/HuR regulatory axis in CRC treatment.
Research on miRNAs has mainly focused on identifying the target genes of miRNAs. Indeed, these studies have been critical and necessary. However, as gene expression regulators, miRNAs themselves undergo complicated regulations in cancers [
47,
53]. Cancer-associated transcription factors are key players in orchestrating gene expression networks in cancers, including miRNAs [
45,
46]. Many TFs/miRNA regulatory pairs have been discovered, and their vital roles in cancer progression have been explored, such as P53/miR-34 and CMYC/miR-17-92 [
54,
55]. Various downstream target genes of miR-22 have been elucidated, including HIF1α, VEGF, TIAM1, MMP-2, COX-2 [
38‐
40], and HuR here. However, the causes for its downregulation in CRC are unknown. In this study, we demonstrated that miR-22 was directly repressed at the transcriptional level by the onco-TF Jun, which is a core member of transcription factor complex AP-1 involved in the oncogenesis of various cancers [
56,
57]. The MAPK pathway (including MEK, ERK and p38) lies upstream of Jun, and can activate Jun expression [
57]. One study reported that ERK can repress the expression of miR-22 [
58]. Considering our results, this inhibition might be explained by ERK-activated Jun, which could then inhibit miR-22. Yang et al. reported that in ischaemia-reperfusion (I/R)-induced myocardial injury, miR-22 could repress the level of c-Jun-AP-1 and p-c-Jun-AP-1 by reducing p38 MAPK [
59]. In another paper, miR-22 could significantly inhibit the DNA-binding ability of AP-1 [
60]. These data suggested a double-negative regulatory relationship between miR-22 and Jun. It should be noted that Jun might also regulate miR-22 via an indirect mechanism. Some studies have reported that Jun can inhibit p53 expression and activity [
61,
62], whereas p53 can transcriptionally activate miR-22 [
63,
64]. There is the possibility that Jun/p53/miR-22 axis exists in CRC also.
AREs are widely distributed in the 3’-UTRs of protein-coding genes, including Jun [
65]. Thus, Jun may be stabilised post-transcriptionally by HuR. This possibility was partially validated by a recent study, which revealed that HuR can increase Jun expression by binding one ARE in its 3’-UTR, and this effect can be reinforced by miR-200a [
66]. This finding, combined with our results, suggests that Jun, miR-22 and HuR participate in a double-negative feedback loop in CRC cells. Because a double-negative feedback is equal to positive feedback and is known for its ability to amplify a response into a self-sustained mode that is independent of the original stimuli, the feedback loop composed of Jun, miR-22 and HuR may minimize miR-22 expression and amplify HuR expression in CRC cells, thus allowing CRC cells to become more autonomous, for example, to reproduce more rapidly and to metastasize to new microenvironments. Thus, this feedback regulation may explain the widespread downregulation of miR-22 and the overexpression of HuR in CRC.