Hsa_circ_0128846 promotes tumorigenesis of colorectal cancer by sponging hsa‐miR‐1184 and releasing AJUBA and inactivating Hippo/YAP signalling

Abstract Hsa_circ_0128846 was found to be the most significantly up‐regulated circRNA in our bioinformatics analysis. However, the role of hsa_circ_0128846 in colorectal cancer has not been explored. We thus aim to explore the influence and mechanism of hsa_circ_0128846 in colorectal cancer by sponging its downstream miRNA target miR‐1184. We collected 40 colorectal cancer patients’ tumour tissues to analyse the expression of hsa_circ_0128846, miR‐1184 and AJUBA using qRT‐PCR and Western blot where needed. Then, we constructed stably transfected SW480 and HCT116 cells to study the influence of hsa_circ_0128846, miR‐1184 and AJUBA on colorectal cancer cell phenotypes. To obtain reliable results, a plethora of experiments including RNA immunoprecipitation assay, flow cytometry, EdU incorporation assay, wound healing migration assay, transwell invasion assay and live imaging of nude mice xenograft assay were performed. The binding relationship between hsa_circ_0128846, miR‐1184 and AJUBA mRNA in colorectal cancer was validated by reported gene assay. In colorectal cancer tissues, circ_0128846 and AJUBA were both significantly up‐regulated, while miR‐1184 was significantly down‐regulated compared with healthy tissues. Meanwhile, hsa_circ_0128846 can absorb miR‐1184 to promote the progression of CRC in vivo and SW480 and HCT116 cell phenotypes in vitro. The knockdown of AJUBA, a downstream target of miR‐1184, reversed the effect of miR‐1184 in CRC cells via enhancing the phosphorylation of the Hippo/YAP signalling pathway proteins MST1, LATS1 and YAP. This study revealed that hsa_circ_0128846 contributed to the development of CRC by decreasing the expression of miR‐1184, thereby increasing AJUBA expression and inactivating Hippo/YAP signalling.


| INTRODUC TI ON
Colorectal cancer (CRC) is not only the world's third most usual disease, but also the second most death-leading cancer-related disease. 1 Clinically, distant metastasis is a preponderant justification of death in CRC that can affect the 5-year survival time and the quality of life of the patients. 2 Given the poor survival time and life quality of patients, it is meaningful to discover new treatments for CRC. Many researches have been reported that molecular targeted therapy has become a new treatment for CRC besides conventional surgery and chemoradiotherapy. [3][4][5] At present, some molecular targeted therapies have come out such as anti-VEG monoclonal antibody, VEGF tyrosine kinase inhibitor and anti-EGFR monoclonal antibody which can improve patient's quality of life. [6][7][8][9][10] Nevertheless, little therapeutic effects happen in CRC patients taking the molecular targeted therapy drugs. Hence, it is imperative to find more effective therapeutic targets to treat CRC. Circular RNA (circRNA) is a newly discovered non-coding RNA with a circular structure, which is different from other RNAs. 11,12 CircRNAs are stable in the cytoplasm, which means that they are not easy to be degraded. [11][12][13][14] In recent researches, circRNAs have been found to have affluent binding sites for microRNAs. 11,12,15,16 By binding with circRNAs, miRNAs thus release their target mRNAs. 11,12,15,16 This mechanism of action was known as the competitive endogenous RNA (ceRNA) network. 11,12,15,16 It has been reported that cir-cRNAs could affect the development of colorectal cancer through the ceRNA network mechanism. [17][18][19] However, there has been no research reporting circ_0128846 as a CRC development regulator.
MiR-1184 has been reported to be down-regulated in various tumours such as bladder cancer, breast cancer and colorectal cancer, and up-regulated in non-small-cell lung cancer and prostate cancer. [20][21][22][23][24][25] It was once reported by Dengke Yang et al that miR-1184 got absorbed by circVANGL1 thus enhancing the bladder cancer phenotypes. 20 In this study, a novel ceRNA network involving miR-1184 and its upstream regulator, hsa_circ_0128846, is to be unravelled in CRC.
AJUBA protein is a member of the LIM protein subfamily with three tandem LIM domains at the C-terminus. 26 AJUBA can be transferred between the cytoplasm and the nucleus due to its nuclear importation and nuclear exportation sequences. [27][28][29] AJUBA has been proved to regulate the transmission of signals from the cytoplasm to the nucleus, and to participate in many signal transducer interactions such as JAK/SATA, Hippo/YAP, Smad/Snail and Wnt/β-catenin. [29][30][31][32][33] AJUBA was once reported to be up-regulated in CRC, 34 and to promote CRC cell survival, 30 suggesting that it is a possible regulator in CRC. Also, it has been reported that AJUBA could be regulated by miRNAs. 35,36 Nonetheless, how AJUBA being regulated by miRNAs in CRC has not been studied.
In our study, we first determined the stimulating effects of hsa_circ_0128846 on the development of CRC via in vivo and in vitro experiments. We also found that circ_0128846 could sponge miR-1184 to elevate the expression level of AJUBA for accelerating the progression of CRC. Besides, the regulative mechanism of hsa_circ_0128846/miR-1184/AJUBA ceRNA network on CRC might be related to the Hippo/YAP signalling pathway. Our findings may exhibit a new target for the treatment of CRC.

| Patients and cell lines
CRC tissues (n = 40) and adjacent healthy colon tissues (n = 24) collected from CRC patients from the First Hospital of Jilin University were used in this study. The collection and the use of tissues followed by the ethical standards in the Helsinki Declaration. The informed consent was signed by all patients. The clinical characteristics are shown in Table 1. The study protocol was approved by the ethics committee of the First Hospital of Jilin University.

| Constructs of circ_0128846 shRNA lentiviral vectors and circ_0128846 overexpression vectors
The pLV-CMV-puro-U6-shRNA lentiviral construct was purchased from OBiO Technology Shanghai Corp., Ltd. (China) and was used to establish the construct that could stably silence circ_0128846. The construction of pLV-CMV-puro-U6-sh-circ_0128846 lentiviral vector (sh-circ_0128846) was based on a previous study. 37 Briefly, a total of three pairs of shRNA sequences for silencing hsa_circ_0128846 were designed against the joint site of hsa_circ_0128846, and the corresponding sequences of these shRNAs are shown in Table 3. Shanghai Corp., Ltd. (China). Briefly, a total of three shRNAs of AJUBA were designed and inserted to the pLV-CMV-puro-U6-shRNA lentiviral vector in the same way as sh-circ_0128846. The sequences of sh-AJUBA are shown in Table 3. Then, the constructs mentioned above were transfected into the SW480 and HCT116 cells for 48 hours using Lipofectamine 2000 reagent (Thermo Fisher, 11668027, USA) on the basis of the protocol given in the manual.
The cells in the blank group were cultured without any transfection.
The transfection efficiency was assessed by qRT-PCR.

| Cell wound healing migration and cell transwell invasion assay
The 4 × 10 5 cells were seeded into the 12-well plate before we

| Animal studies
The animal experiment was carried out following the guidelines of the Institutional Animal Care and Use Committee (IACUC).
Six BALB/c nude mice were purchased from Beijing Vital River Laboratory Animal Technology Co., Ltd. The nude mice were divided into two groups (NC and sh-circ_0128846). Then, the SW480 cells transfected with sh-circ_0128846 or NC were injected into the nude mice in the corresponding groups. The tumour volume was measured every six days. After 30 days, the IVIS 200 bioluminescence imaging system and Living Image software (Caliper Life Sciences, Hopkinton, MA) were used to monitor and analyse tumour growth.
The nude mice were killed after 30 days to collect the xenografted tumours for the following Ki67 and H&E staining.

| Ki67 and H&E staining
First, tumours collected from the animal experiments were fixed in 4% formalin for 48 hours. Then, the tissues were sequentially placed in 75%, 80%, 95% ethanol and absolute ethanol to dehydrate. After dehydration, the tissues were dipped in melted paraffin, and then embedded in an embedder. Next, we used a tissue slicer to slice tumours into sections in 5 μm thickness. Before we started staining, we dewaxed the sections. Then, we treated the sections with peroxide blocker to remove endogenous peroxidases. Finally, the sections were incubated with the anti-Ki67 primaries (Cat#: ab16667, Abcam, UK) and the HRP-conjugated rabbit secondaries (Cat#: 6728, Abcam, UK) after being washed in PBS. In H&E staining, after dewaxing the sections, we stained the sections with haematoxylin for 5 minutes and eosin for 20 seconds.

| Western blot assay
The cell protein was obtained using RIPA buffer that included 5 mM EDTA and PMSF. All protein samples were quantified using a spectrophotometer to ensure that the protein samples were the GAPDH was used as an internal loading control. Then, the hybrid membranes were incubated with secondaries for 2 hour. The membrane went through exposure, and the protein band intensity was read using FluorChem FC2 software. with miR-1184 mimic or NC were lysed in RIP lysis buffer containing magnetic beads conjugated with human anti-AGO2 antibodies or IgGs. Then, Proteinase K was added to the mixture to remove the proteins. Lastly, the immunoprecipitated RNA was collected, and the abundance of hsa_circ_0128846 was detected by qRT-PCR.

| Statistical analyses
We did the statistical analysis using GraphPad Prism v7.0. All results came from at least three independent experiments. Data were expressed as mean ± SD. Two-tailed t test analysis was conducted for analysing differences between two groups, while one-way ANOVA with Dunnett's multiple comparison method was conducted for analysing differences among multiple groups. P-values that was less than 0.05 were deemed as statistically significant.

| Expression, characterization and subcellular distribution of circ_0128846 in CRC tissue samples and cell lines
The hsa_circ_0128846 was generated from ZFR gene. It was identified to be the most significant up-regulated circRNA by our bioinformatics analysis (Fig. S1A). Its structure is shown in Figure 1A.
Firstly, we employed qRT-PCR to explore the expression of hsa_ circ_0128846 in CRC tissues and adjacent normal tissues, and found that the expression of hsa_circ_0128846 in CRC tissues was 2.2-fold higher than that in the normal tissues ( Figure 1B). In addition, qRT-PCR was also conducted to measure the expression of hsa_circ_0128846 in CRC cells and normal colorectal cells, and the data showed that the expression of hsa_circ_0128846 was significantly up-regulated in CRC cells in comparison with the normal colorectal cells, especially in SW480 and HCT116 cells ( Figure 1C). Therefore, SW480 and HCT116 cell lines were selected as our cell line models. Then, RNase R degradation assay was used to confirm the stable existence of hsa_circ_0128846 in CRC cells. The data demonstrated that the hsa_circ_0128846 resisted to the RNase R treatment, while the linear RNA (ZFR mRNA) was degraded in both SW480 and HCT116 cells ( Figure 1D). Lastly, qRT-PCR was performed to, respectively, measure the expression of hsa_circ_0128846 in cytoplasm and nuclear, and it was confirmed that hsa_circ_0128846 majorly existed in the cytoplasm of SW480 and HCT116 cells ( Figure 1E).

| Hsa_circ_0128846 knockdown impeded the CRC cell phenotype, whereas hsa_circ_0128846 overexpression enhanced it in vitro
We conducted loss-of-function and gain-of-function experiments to determine the effect of hsa_circ_0128846 on CRC. Before the experiments, we detected the transfection efficiency of the sh-circ_0128846 and hsa_circ_0128846 overexpression vectors in SW480 and HCT116 cells. By qRT-PCR, shRNA1-circ_0128846 most efficiently reduced the expression of hsa_circ_0128846, by 70% (Fig. S2A), but did not affect the expression of linear RNA of ZFR (Fig. S2B). Therefore, shRNA1-circ_0128846 was selected to knock down hsa_circ_0128846. In addition, the hsa_circ_0128846 overexpression vectors were successfully transfected into the SW480 and HCT116 cells as well, seen by the approximately 2-fold increase of hsa_circ_0128846 expression (Fig. S2C).

| Hsa_circ_0128846 promoted the CRC tumour development in vivo
To further ascertain the CRC promoting role of hsa_circ_0128846, we conducted in vivo animal experiments. By implanting SW480 cells stably transfected with sh-circ_0128846 or sh-NC vectors into the abdominal cavity of the nude mice, we established the CRC animal models. We found that mice in the sh-circ_0128846 group showed stronger luminescence signals than those in the sh-NC group ( Figure 3A). The tumour volume of mice in sh-circ_0128846 group was significantly smaller than that of mice in the sh-NC group at days 18, 24 and 30 ( Figure 3B). The tumour in mice of sh-circ_0128846 was also found significantly lighter than in mice of the sh-NC group at day 30 when the mice went through euthanasia ( Figure 3C).
Moreover, Ki67 staining results showed that the knockdown of circ_0128846 caused repression of Ki67 protein expression in tumour tissues ( Figure 3D), suggesting that circ_0128846 silence inhibited the tumour growth. H&E staining results showed that the knockdown of circ_0128846 sabotaged the pathological structure of the tumour in nude mice ( Figure 3E). These data suggested that hsa_circ_0128846 could enhance CRC tumour growth in vivo.

| Hsa-miR-1184 was a downstream target of hsa_circ_0128846
Among the identified three miRNAs (from the intersection of differentially expressed miRNAs of GSE12 6095 data, the targets of hsa_circ_0128846 from circular RNA interactome database  Fig. S1D), hsa-miR-1184 was the most significantly down-regulated one, therefore being selected as our candidate miRNA of the ceRNA network. The complementary sequences of miR-1184 and hsa_circ_0128846 were obtained from circular RNA interactome database ( Figure 4A). Then, we conducted the luciferase activity assay to confirm the binding relationship between hsa_circ_0128846 and miR-1184, finding that miR-1184 obviously reduced the luciferase activity in cells transfected with wild-type circ_0128846 but not those transfected with mutated circ_0128846 ( Figure 4B). In addition, RIP analysis revealed that hsa_circ_0128846 was utterly pulled down when transfected with miR-1184 mimics in SW480 and HCT116 cells, compared to the miR-1184 NC group ( Figure 4C). Interestingly, miR-1184 was found down-regulated in CRC tissues compared with adjacent healthy tissues ( Figure 4D), and there was a negative relationship between miR-1184 and hsa_circ_0128846 expression ( Figure 4E) Figure 4F).

| Knockdown of hsa_circ_0128846 inhibited the proliferation, migration and invasion and promoted the apoptosis of CRC cells in vitro by sponging miR-1184
To further explore whether hsa_circ_0128846 could affect the phe- NC, the cells were transfected with negative control plasmids. sh-AJUBA, the cells were transfected with AJUBA shRNA plasmids. Inhibitor, the cells were transfected with miR-1184 inhibitor plasmids. sh-AJUBA + inhibitor, the cells were co-transfected with AJUBA shRNA and miR-1184 inhibitor plasmids. *P < .05, **P < .001 versus blank group. (G) The expression of AJUBA protein was regulated by hsa_circ_0128846. Blank, the cells were cultured without any treatments. NC, the cells were transfected with negative control plasmids. shcirc, the cells were transfected with circ_0128846 shRNA plasmids. Inhibitor, the cells were transfected with miR-1184 inhibitor plasmids. sh-circ + inhibitor, the cells were co-transfected with circ_0128846 shRNA and miR-1184 inhibitor plasmids. *P < .05, **P < .001 versus blank group hsa_circ_0128846 inhibited the phenotypes of CRC cells by freeing miR-1184 in vitro.

| AJUBA was a downstream target of miR-1184
AJUBA was identified from the intersection of differentially expressed mRNAs of GSE12 6095 data set and the Hippo signalling pathway genes. Among the nine common mRNAs, AJUBA and SCRIB were the upstream effectors of the Hippo signalling pathway, whereas the others were all the downstream effectors (Fig.   S1B). Then, we interrogated TCGA database and found that SCIB was not significantly up-regulated in colon adenocarcinoma, but AJUBA was (Fig. S1C). TargetScan Human 7.2 was used to predict the binding relationship between AJUBA mRNA and miR-1184 ( Figure 6A). The luciferase activity in the wild-type AJUBA + miR-1184 group was the lowest, and that in the mutated sites 1 and 2 of AJUBA 3′UTR groups was lower than that in the co-mutated AJUBA 3′UTR group ( Figure 6B). Interestingly, we found that AJUBA showed a 3-fold up-regulation in tumour tissues compared with adjacent healthy tissues ( Figure 6C) and a negative correlation with the expression of miR-1184 in tumour tissues ( Figure 6D).
These results proved that AJUBA was a downstream target of miR-  Figure 8B). These data suggested that miR-1184 inhibitor enhanced the phenotypes of CRC cells via releasing AJUBA to inactivate the Hippo/YAP signalling pathway (a schematic mechanism illustration is presented in Figure 8C).  Similarly, miR-1184 in our study was proved to inhibit the CRC cell phenotypes via suppressing AJUBA expression. Taking together, we concluded that miR-1184 was a significant tumour suppressor in human CRC.
AJUBA protein, consisting of 55 amino acids, is a member of the Zyxin/AJUBA family, which also includes Zyxin, Trip6, Wtip6, Limd1 and Lpp. 28,[45][46][47][48] AJUBA is composed of the carboxy-terminal LIM region and the amino-terminal PreLIM region, 27,48,49 and is a cytoplasmic linker protein that had the ability to shuttle between the cell junction and the nucleus, thereby playing an important part in signal transmission. 50 AJUBA was reported to be an upstream regulator of the Hippo signalling pathway, 51 and it had the function of an epithelial integrity sensor and was necessary for re-entering the cell cycle progression. 31,48 In addition, AJUBA promoted oncogenic activity of YES-associated protein (YAP) by suppressing large tumour suppressor type 1/2 (LATS 1/2) core kinase in the Hippo signalling pathway. 29,31,52,53 In other words, AJUBA could interact with LATS to prevent LATS from phosphorylating YAP, which was active in the un-phosphorylated state to increase cell proliferation. 29

ACK N OWLED G EM ENTS
None.

CO N FLI C T O F I NTE R E S T S
None declared by the authors. Writing-review & editing (equal).

AUTH O R CO NTR I B UTI O N
TL and DS designed the study. XW and YJC performed the experiments. WL interpreted the data and drafted the manuscript. All authors reviewed and edited the manuscript.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data used in this study are available from the corresponding author upon reasonable request.