Molecular Pathogenesis of Colorectal Cancer: Impact of Oncogenic Targets Regulated by Tumor Suppressive miR-139-3p

We recently determined the RNA sequencing-based microRNA (miRNA) expression signature of colorectal cancer (CRC). Analysis of the signature showed that the expression of both strands of pre-miR-139 (miR-139-5p, the guide strand, and miR-139-3p, the passenger strand) was significantly reduced in CRC tissues. Transient transfection assays revealed that expression of miR-139-3p blocked cancer cell malignant transformation (e.g., cell proliferation, migration, and invasion). Notably, expression of miR-139-3p markedly blocked RAC-alpha serine/threonine-protein kinase (AKT) phosphorylation in CRC cells. A combination of in silico database and gene expression analyses of miR-139-3p-transfected cells revealed 29 putative targets regulated by miR-139-3p in CRC cells. RNA immunoprecipitation analysis using an Argonaute2 (AGO2) antibody revealed that KRT80 was efficiently incorporated into the RNA-induced silencing complex. Aberrant expression of Keratin 80 (KRT80) was detected in CRC clinical specimens by immunostaining. A knockdown assay using small interfering RNA (siRNA) targeting KRT80 showed that reducing KRT80 expression suppressed the malignant transformation (cancer cell migration and invasion) of CRC cells. Importantly, inhibiting KRT80 expression reduced AKT phosphorylation in CRC cells. Moreover, hexokinase-2 (HK2) expression was reduced in cells transfected with the KRT80 siRNAs or miR-139-3p. The involvement of miRNA passenger strands (e.g., miR-139-3p) in CRC cells is a new concept in miRNA studies. Our tumor-suppressive miRNA-based approach helps elucidate the molecular pathogenesis of CRC.


Introduction
According to the World Health Organization (Globocan 2020), colorectal cancer (CRC) is the third most common cancer (over 1,800,000 cases) worldwide and the second leading cause of cancer-related deaths (over 880,000 deaths) [1]. In clinical practice, the prognosis of CRC is relatively good if diagnosed early. However, the prognosis is consistently poor in advanced cases, with a 5-year survival rate of approximately 14% (stage III or stage IV metastatic disease) [2]. At the time of the initial diagnosis, approximately 14-18% of patients with CRC have metastases, and the treatment strategies for unresectable cases are limited [3].
The oncogenesis of CRC is illustrated by a well-known multistep model of cancer cells [4][5][6]. From previous studies, mutations in various genes involved in the oncogenesis of CRC (e.g., APC, TP53, SMAD4, KRAS, and PIK3CA) and activation of cancer signaling pathways (e.g., WNT, RAS/MAPK, PI3K, TGF-β, P53, and DNA mismatch-repair) caused by these gene mutations have been reported [4]. CRC cells have highly heterogeneous properties, requiring new therapeutic parameters for CRC from genetic and genomic points of view. As a result of this molecular heterogeneity, recent genome-wide transcriptome analyses have revealed that CRC cells can be molecularly classified into four consensus molecular subtypes (CMS1 to CMS4) [7]. The future treatment strategies for CRC patients will be based on these subtypes.
As a result of the Human Genome Project, it has become clear that a vast number of functional non-coding RNA molecules (ncRNAs) are present in the human genome [8]. Current studies have shown that numerous ncRNAs play important roles in various biological activities such as the stabilization of RNA molecules and regulation of gene expression and the cell cycle [9,10]. Extensive research to date has revealed that ncRNA dysregulation is deeply involved in the initiation and development of human diseases, including cancer [11].
Among ncRNAs, microRNAs (miRNAs), consisting of only 19-22 nucleotides, have been well studied in cancer research fields. They function as fine-tuners of gene expression control in a sequence-dependent manner [12]. A single miRNA controls numerous genes, and in turn, a single gene is controlled by numerous miRNAs [13]. As a result, miRNAs and their target genes form a very complex network within cells, and it is easy to imagine that aberrant expression of miRNAs disrupts this RNA network. Many studies have shown aberrant expression of miRNAs in CRC cells, and these miRNAs act as oncogenes and/or tumor suppressors by targeting cancer-related genes in CRC cells [14][15][16].
In this study, we focused on miR-139-3p (the passenger strand of pre-miR-139) and investigated its functional significance and target oncogenes in CRC cells. Notably, ectopic expression of miR-139-3p markedly blocked the phosphorylation of RAC-alpha serine/threonine-protein kinase (AKT) in CRC cells. Our search strategy for miRNA targets revealed a total of 29 genes as putative candidate targets of miR-139-3p in CRC cells. Of these, keratin 80 (KRT80) was found to be a direct target of miR-139-3p, and its aberrant expression enhanced the malignant transformation of CRC cells. Involvement of the passenger strand of miRNA and its gene targets in CRC pathogenesis is a new concept and provides novel insights into the molecular pathogenesis of CRC.

Expression of miR-139-5p and miR-139-3p in CRC Specimens
Recently, we determined the miRNA expression signature of CRC by miRNA sequencing using CRC clinical specimens (GEO accession number: GSE183437). Analysis of the signature showed that 84 miRNAs were upregulated, and 70 were downregulated in CRC tissues ( Figure 1A). Among downregulated miRNAs in CRC tissues, we focused on miR-139-5p (the guide strand) and miR-139-3p (the passenger strand), because both strands of miRNAs derived from pre-miR-139 were significantly downregulated in CRC tissues. Our interest is to clarify how the passenger strand of miRNA is involved in the malignant transformation of CRC cells. The mature sequences of the two microRNAs are shown in Figure 1B. mation of these specimens is shown in Table S1. The expression levels of miR-139-5p (p < 0.001) and miR-139-3p (p < 0.001) were significantly lower in CRC tissues than normal tissues ( Figure 1C). Next, we examined the expression levels of miR-139-5p and miR-139-3p in two CRC cell lines, HCT116 and DLD-1. In these cell lines, the expression levels of miR-139-5p and miR-139-3p were lower than those in normal epithelial tissues ( Figure 1C). Furthermore, a positive correlation was detected between miR-139-5p and miR-139-3p expression levels by Spearman's rank analysis (r = 0.559, p < 0.001; Figure 1D).  The log 2 fold-change (FC) is plotted on the x-axis and the log 10 p-value on the y-axis. The blue points represent the downregulated miRNAs with log 2 FC < −2.0 and p < 0.05. The red points represent the upregulated miRNAs with log 2 FC > 2.0 and p < 0.05. Downregulated expressions of miR-139-5p and miR-139-3p are plotted. (B) Chromosomal location of pre-miR-139 in the human genome. The mature sequences of miR-139-5p (the guide sequence) and miR-139-3p (the passenger strand) are shown. (C) Expression levels of miR-139-5p and miR-139-3p validated in CRC clinical specimens and CRC cell lines (HCT116 and DLD-1). The expression of both miRNAs was significantly downregulated in cancer tissues (p < 0.001). (D) Spearman's rank test showed positive correlations between miR-139-5p and miR-139-3p expression levels in clinical specimens (r = 0.559, p < 0.001).
CRC tissues and noncancerous tumor-adjacent tissues (27 paired) were used to verify the expression status of miR-139-5p, miR-139-3p, and their target genes. Clinical information of these specimens is shown in Table S1. The expression levels of miR-139-5p (p < 0.001) and miR-139-3p (p < 0.001) were significantly lower in CRC tissues than normal tissues ( Figure 1C). Next, we examined the expression levels of miR-139-5p and miR-139-3p in two CRC cell lines, HCT116 and DLD-1. In these cell lines, the expression levels of miR-139-5p and miR-139-3p were lower than those in normal epithelial tissues ( Figure 1C).

Ectopic Expression Assays of miR-139-5p and miR-139-3p in CRC Cell Lines
To investigate the tumor-suppressive functions of miR-139-5p and miR-139-3p, we ectopically expressed mature miR-139-5p and miR-139-3p in two CRC cell lines, HCT116 and DLD-1, and performed functional assays, e.g., cancer cell proliferation, migration, and invasion. After miR-139-5p transfection, cancer cell migration in both cell lines was significantly inhibited (Figure 2A-C). In contrast, the malignant phenotypes of cancer cells, e.g., proliferation, migration, and invasion, were significantly reduced by miR-139-3p transfection in both cell lines (Figure 2A-C). Representative images from the migration and invasion assays are shown in Figure S1.

Ectopic Expression Assays of miR-139-5p and miR-139-3p in CRC Cell Lines
To investigate the tumor-suppressive functions of miR-139-5p and miR-139-3p, we ectopically expressed mature miR-139-5p and miR-139-3p in two CRC cell lines, HCT116 and DLD-1, and performed functional assays, e.g., cancer cell proliferation, migration, and invasion. After miR-139-5p transfection, cancer cell migration in both cell lines was significantly inhibited (Figure 2A-C). In contrast, the malignant phenotypes of cancer cells, e.g., proliferation, migration, and invasion, were significantly reduced by miR-139-3p transfection in both cell lines (Figure 2A-C). Representative images from the migration and invasion assays are shown in Figure S1.

Identification of Oncogenes Regulated by miR-139-3p in CRC Cells
Based on these expressions and functional analysis, aberrant expression of miR-139-3p and disruption of its gene regulation mechanisms were considered to be more deeply involved in the malignant pathogenesis of CRC. We focused on miR-139-3p (passenger strand) for further validation.
We investigated whether epigenetic modifications affect the downregulation of miR-139-3p in CRC cells. After treatment of Trichostatin A (TSA) in CRC cells, the expression level of miR-139 was increased compared to TSA untreated cells ( Figure S2A). In addition, miR-139-3p expression level was elevated by 5-aza-2-deoxycytidine (5-aza-dC) treatment

Identification of Oncogenes Regulated by miR-139-3p in CRC Cells
Based on these expressions and functional analysis, aberrant expression of miR-139-3p and disruption of its gene regulation mechanisms were considered to be more deeply involved in the malignant pathogenesis of CRC. We focused on miR-139-3p (passenger strand) for further validation.
We investigated whether epigenetic modifications affect the downregulation of miR-139-3p in CRC cells. After treatment of Trichostatin A (TSA) in CRC cells, the expression level of miR-139 was increased compared to TSA untreated cells ( Figure S2A). In addition, miR-139-3p expression level was elevated by 5-aza-2-deoxycytidine (5-aza-dC) treatment in CRC cells ( Figure S2B). These results suggest that histone deacetylation and DNA methylation are closely involved in the downregulation of miR-139-3p in CRC cells.
The following hypotheses regarding miR-139-3p target genes in CRC cells were made: the target genes of miR-139-3p have one or more binding site(s), are downregulated after miR-139-3p transfection in CRC cells, and are upregulated in CRC tissues. We combined the gene expression data from two databases (TargetScan and GEPIA2) with gene expression data from miR-139-3p-transfected CRC cells (GSE155659) to search for genes that meet these three criteria. A flowchart of the search strategy is shown in Figure 3. A total of 95 putative targets of miR-139-3p in CRC cells were identified. the gene expression data from two databases (TargetScan and GEPIA2) with gene expression data from miR-139-3p-transfected CRC cells (GSE155659) to search for genes that meet these three criteria. A flowchart of the search strategy is shown in Figure 3. A total of 95 putative targets of miR-139-3p in CRC cells were identified.
We assessed the expression levels of putative miR-139-3p target genes in CRC clinical tissues using The Cancer Genome Atlas database via the GEPIA2 platform. A total of 29 genes were significantly upregulated in CRC clinical specimens (colon adenocarcinoma or rectal adenocarcinoma) in this database (p < 0.01: Table 1, Figure S3). GEPIA2 analysis revealed that the expression level of KRT80 was fairly low in normal tissues ( Figure S3). Genes expressed exclusively in cancer cells are appropriate therapeutic targets for CRC. We focused on KRT80 in the subsequent functional analyses in CRC cells.   We assessed the expression levels of putative miR-139-3p target genes in CRC clinical tissues using The Cancer Genome Atlas database via the GEPIA2 platform. A total of 29 genes were significantly upregulated in CRC clinical specimens (colon adenocarcinoma or rectal adenocarcinoma) in this database (p < 0.01: Table 1, Figure S3). GEPIA2 analysis revealed that the expression level of KRT80 was fairly low in normal tissues ( Figure S3). Genes expressed exclusively in cancer cells are appropriate therapeutic targets for CRC. We focused on KRT80 in the subsequent functional analyses in CRC cells.

Direct Regulation of KRT80 by miR-139-3p in CRC Cells
In CRC cells transfected with miR-139-3p, both the mRNA and protein levels of KRT80 were significantly downregulated ( Figure 4A).

Knockdown Assays by Small Interfering RNAs (siRNAs) Targeting KRT80 in CRC Cell Lines
To assess the functional significance of KRT80 in CRC cells, we performed knockdown assays using siRNAs corresponding to KRT80 mRNA. First, the inhibitory effects of two different siRNAs (siKRT80-1 and siKRT80-2) targeting KRT80 in two cell lines were Next, RNA immunoprecipitation (RIP) analysis was performed to confirm that KRT80 mRNA was incorporated into the RNA-induced silencing complex (RISC) after miR-139-3p transfection. The RIP assay concept is illustrated in a schematic in Figure 4B. In samples subjected to immunoprecipitation using an Argonaute2 (AGO2) antibody, quantitative real-time reverse-transcription PCR (qRT-PCR) showed that the KRT80 mRNA level was significantly higher than that in mock and miRNA control-transfected cells (p < 0.001; Figure 4B). Ago2-bound miR-139-3p and KRT80 mRNA were isolated by immunoprecipitation using the AGO2 antibody, suggesting that the RISC plays a central role in miRNA biogenesis ( Figure 4B).
Finally, a dual-luciferase reporter assay was performed to confirm that miR-139-3p binds directly to the 3 untranslated regions (UTR) of KRT80. Luciferase activity was significantly reduced following co-transfection with miR-139-3p and a vector containing the miR-139-3p-binding site within the 3'-UTR of KRT80 ( Figure 4C). In contrast, co-transfection with a vector containing the KRT80 3'-UTR in which the miR-139-3p-binding site was deleted resulted in no change in luciferase activity ( Figure 4C).

Knockdown Assays by Small Interfering RNAs (siRNAs) Targeting KRT80 in CRC Cell Lines
To assess the functional significance of KRT80 in CRC cells, we performed knockdown assays using siRNAs corresponding to KRT80 mRNA. First, the inhibitory effects of two different siRNAs (siKRT80-1 and siKRT80-2) targeting KRT80 in two cell lines were examined. Both KRT80 mRNA and protein levels were effectively suppressed after transfection of each siRNA into HCT116 and DLD-1 cells ( Figure S4).
Knockdown of KRT80 slightly inhibited cell proliferation ( Figure 5A) and markedly inhibited migration and invasion in both HCT116 and DLD-1 cells ( Figure 5B,C). Representative photographs from the migration and invasion assays are shown in Figure S5.

Aberrant Expression of KRT80 Protein in CRC Clinical Specimens
Protein expression of KRT80 was assessed by immunohistochemistry in CRC clinical specimens. Overexpression of KRT80 protein was detected in cancer lesions ( Figure 6). Based on the previous report that overexpression of KRT80 induced epithelial-mesenchymal transition (EMT)-related genes and activated AKT signaling via phosphorylation of AKT (Ser 473) [22], Western blotting for phosphorylation of AKT was performed on KRT80 and miR-139-3p.
Notably, transfection of the KRT80 siRNAs suppressed the phosphorylation of AKT ( Figure 5D).
In addition, expression of miR-139-3p markedly inhibited the phosphorylation of AKT in CRC cells, according to Western blot analysis.

Aberrant Expression of KRT80 Protein in CRC Clinical Specimens
Protein expression of KRT80 was assessed by immunohistochemistry in CRC clinical specimens. Overexpression of KRT80 protein was detected in cancer lesions ( Figure 6).

KRT80-Mediated RNA Networks in CRC Cells
To explore KRT80-regulated RNA networks in CRC, we performed comprehensive gene expression analyses in KRT80-knockdown CRC cells. A total of 52 genes were identified as downregulated in both KRT80-knockdown CRC cell lines (log2 fold change <−1.0: Table 2). Our expression data were deposited in the GEO database (GEO accession number: GSE208785).
In this study, we focused on hexokinase 2 (HK2) because it was identified as a miR-139-3p target in CRC cells (Table 1). HK2 was commonly regulated by miR-139-3p and KRT80 in CRC cells ( Figure 7A). Moreover, HK2 was directly regulated by miR-139-3p in CRC cells, by RIP assay and dual luciferase reporter assay (Figures S6). In addition, HK2 expression was upregulated in CRC tissues (Figure S3), and a vast number of studies showed that aberrant expression of HK2 enhances cancer cell malignant transformation in various types of cancers. Our results showed that HK2 expression was reduced in cells transfected with siKRT80 ( Figure 7B) or miR-139-3p ( Figure 7C).

KRT80-Mediated RNA Networks in CRC Cells
To explore KRT80-regulated RNA networks in CRC, we performed comprehensive gene expression analyses in KRT80-knockdown CRC cells. A total of 52 genes were identified as downregulated in both KRT80-knockdown CRC cell lines (log 2 fold change < −1.0: Table 2). Our expression data were deposited in the GEO database (GEO accession number: GSE208785).
In this study, we focused on hexokinase 2 (HK2) because it was identified as a miR-139-3p target in CRC cells (Table 1). HK2 was commonly regulated by miR-139-3p and KRT80 in CRC cells ( Figure 7A). Moreover, HK2 was directly regulated by miR-139-3p in CRC cells, by RIP assay and dual luciferase reporter assay ( Figure S6). In addition, HK2 expression was upregulated in CRC tissues (Figure S3), and a vast number of studies showed that aberrant expression of HK2 enhances cancer cell malignant transformation in various types of cancers. Our results showed that HK2 expression was reduced in cells transfected with siKRT80 ( Figure 7B) or miR-139-3p ( Figure 7C).

Expression of Target Genes in Clinical Specimens and Correlation
In the analysis using surgical specimens (27 paired normal and cancerous tissues), we observed marked suppression of miR-139-3p and marked upregulation of KRT80 in cancer tissues (Figures 1C and S7A). In addition, a negative correlation was observed between the expression of miR-139-3p and KRT80 in CRC specimens ( Figure S7B). Contrary to the TCGA data analysis, we did not find any significant upregulation of HK2 in our cancerous samples.

Discussion
Because CRC is a heterogeneous disease, as indicated by our genome-wide transcriptome analysis, it is necessary to search for diagnostic markers and therapeutic target molecules in an individualized manner. Recently, we determined the miRNA expression signature of CRC using RNA sequencing [17]. In that study, we found that miR-490-3p acted as a tumor-suppressive miRNA in CRC cells, and expression of its gene targets (IRAK1, FUT1, and GPRIN2) was significantly predictive of 5-year overall survival in CRC patients

Expression of Target Genes in Clinical Specimens and Correlation
In the analysis using surgical specimens (27 paired normal and cancerous tissues), we observed marked suppression of miR-139-3p and marked upregulation of KRT80 in cancer tissues (Figures 1C and S7A). In addition, a negative correlation was observed between the expression of miR-139-3p and KRT80 in CRC specimens ( Figure S7B). Contrary to the TCGA data analysis, we did not find any significant upregulation of HK2 in our cancerous samples.

Discussion
Because CRC is a heterogeneous disease, as indicated by our genome-wide transcriptome analysis, it is necessary to search for diagnostic markers and therapeutic target molecules in an individualized manner. Recently, we determined the miRNA expression signature of CRC using RNA sequencing [17]. In that study, we found that miR-490-3p acted as a tumor-suppressive miRNA in CRC cells, and expression of its gene targets (IRAK1, FUT1, and GPRIN2) was significantly predictive of 5-year overall survival in CRC patients [17]. This new miRNA expression signature of CRC will be a useful tool for elucidating the molecular pathogenesis of this disease.
Aberrant expression of miRNAs is frequently observed in several types of cancers [14][15][16]. A vast number of studies showed that epigenetic modification (histone modifications and promoter DNA methylation) is closely involved in the silencing of miRNAs expression in cancer cells [23][24][25][26]. A recent study showed that miR-139 was epigenetically silenced by histone H3 lysine 27 trimethylation (H3K27me3) in lung cancer cells [25]. Our present data (TSA and 5-aza-dC treatment) showed that both events of histone deacetylation and DNA methylation were closely involved in the silencing of miR-139-3p on CRC cells. It has been shown that miR-139-3p silencing plays a pivotal role in human oncogenesis.
Our recent studies revealed that some passenger strands of miRNAs are closely involved in the molecular pathogenesis of a wide range of human cancers, e.g., miR-30c-2-3p, miR-101-5p, miR-143-5p, and miR-145-3p [19,21,27,28]. Based on our CRC signature, we focused on miR-139-3p (the passenger strand derived from pre-miR-139) in this study. We have analyzed the passenger stand miR-139-3p in several types of cancers and found that it acts as a tumor-suppressive miRNA in bladder cancer, renal cell carcinoma, and head and neck squamous cell carcinoma by targeting several genes closely linked to cancer pathogenesis [29][30][31]. Here, the function of miR-139-3p in CRC cells was clarified and found to be consistent with previous reports. As we have discussed, our in vitro assays showed that miR-139-3p acted as a tumor suppressive miRNA in CRC cells. However, the endogenous expression levels of passenger strands of miRNAs are little, and the full picture of the functions of passenger strands of miRNAs in vivo remains unknown. In order to investigate the in vivo functions of miRNAs, it is essential to generate and analyze cells that constitutively express miRNAs or cells in which miRNA expression is completely knocked out.
Next, we searched for target genes regulated by miR-139-3p in CRC cells, particularly those involved in AKT phosphorylation. A unique feature of miRNAs is that they regulate different sets of genes depending on the cancer cell type.
We identified 29 genes as tumor-suppressive targets of miR-139-3p in CRC cells. Of these, we focused on KRT80 because its expression was significantly different between cancer and normal tissues. Ideally, a therapeutic target molecule for cancer is not expressed in normal cells. Expression levels of KRT80 in normal tissues were assessed using previous large-scale transcriptional analysis data [41]. Expression of KRT80 was detected in skin, esophagus, and salivary glands. In contrast, KRT80 was hardly expressed in other tissues ( Figure S8).
We showed that aberrant expression of KRT80 enhanced the malignant phenotypes of cancer cells (i.e., proliferation, migration, and invasion). Interestingly, overexpression of KRT80 induced EMT-related genes and activated the AKT signaling through phosphorylation of AKT (Ser 473) [22]. Considering our present data and previous reports, it was strongly suggested that the miR-139-3p/KRT80/p-AKT axis influences the migration and invasive abilities of CRC cells. In ovarian cancer, overexpression of KRT80 induced the expression of genes related to epithelial-mesenchymal transition and activated both MEK and ERK [42]. In gastric cancer, overexpression of the circular RNA CircPIP5K1A induced expression of KRT80 and activated the PI3K/AKT pathway via miR-671-5p adsorption [43]. Moreover, KRT80 expression was significantly correlated with clinical parameters, such as lymph node metastasis and pathological stage, in CRC and ovarian cancer [22,42]. Together, these data suggest that KRT80 is a potential therapeutic target for CRC.
We also investigated genes affected by KRT80 in CRC cells. In CRC cells, the expression of several genes was suppressed after the knockdown of KRT80 expression. Among these genes, we focused on HK2. The four members of the HK family (HK1-4) in mammals catalyze the conversion of glucose to glucose-6-phosphate, and they are involved in the first and rate-limiting step of glycolysis [44][45][46]. Previous studies reported that Akt and HK2 are overexpressed in cancer cells and that there is a positive correlation between activation of the PI3K/Akt/mTORC1 pathway and HK2 expression [47][48][49]. These findings indicate that simultaneous inhibition of glycolysis and the AKT/mTOR signaling pathway is effective in suppressing the growth of cancer cells [50].

Clinical Specimens Used to Evaluate miR-139-5p and miR-139-3p Expression
Fifty-four clinical specimens (27 CRC tissues and 27 normal colon tissues) were used to evaluate the expression status of miR-139-5p/3p. All specimens used in this study were obtained by surgical resection at Kagoshima University Hospital between 2014 and 2017. Normal colon tissue was collected from adjacent sites to the specimen from which each CRC tissue sample was taken. All patients provided written informed consent for the use of their specimens. This study was conducted in accordance with the guidelines of the Declaration of Helsinki and was approved by the Ethics Committee of Kagoshima University (approval number 160038 (28-65); date of 19 March 2021). The clinical information was described in our previous study [17].

CRC Cell Lines and Cell Culture
Two CRC cell lines, HCT116 and DLD-1, were used in this study. HCT116 cells were obtained from the RIKEN Cell Bank (Tsukuba, Ibaraki, Japan), and DLD-1 cells were obtained from the Cell Resource Center for Biomedical Research Bank (Sendai, Miyagi, Japan). HCT116 was cultured in DMEM medium supplemented with 10% concentration of fetal bovine serum (FBS), and DLD-1 was cultured in RPMI-1640 medium, also supplemented with 10% concentration of fetal bovine serum (FBS).

RNA Extraction and Quantitative Real-Time Reverse-Transcription PCR (qRT-PCR)
The protocols used for RNA extraction and qRT-PCR were described in our previous studies [51,52]. In brief, Total RNA was isolated from cell lines using TRIzol reagent according to the manufacturer's protocol. RNA samples were reverse transcribed using the High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems, Waltham, MA, USA). qPCR was performed using PCR Master Mix (Applied Biosystems, Waltham, MA, USA) and Fast SYBR Gren Master Mix (Applied Biosystems, Waltham, MA, USA), StepOnePlus real-time PCR system (Applied Biosystems, Waltham, MA, USA). Gene expressions were quantified relatively by the delta-delta Ct method (used GUSB as internal control). TaqMan assays used in this study are summarized in Table S2. The sequences of primers for SYBR green assays are summarized in Table S3.

Regulation of miR-139-3p Expression by DNA Demethylation
Cells were treated with 5-Aza-2 -deoxycytidine (5-aza-dC; Wako, Osaka, Japan) at concentrations of 0.5, 1,2,5, and 10 µmol/L for 96 h. Cells were first cultured in growth medium; after 24 h of incubation, the medium was replaced with fresh medium containing 5-aza-dC or Dimethyl sulfoxide (DMSO, negative control), and cells were incubated for another 48 h; after 48 h of treatment, the medium was again replaced with fresh medium containing 5-aza-dC or DMSO and cells were cultured for additional 48 h. After 120 h treatment, total RNA was isolated. The expression levels of miR-139-3p were measured by qRT-PCR.

Regulation of miR-139-3p Expression by Histone Deacetylation
Cells were treated with Tricostatin A (TSA; Wako, Osaka, Japan) at 0.1 or 0.5 µmol/L concentration for 24 h. Cells were first grown in growth medium. After 24 h of incubation, the medium was replaced with fresh medium containing TSA or DMSO and the cells were incubated for an additional 24 h. After 48 h treatment, total RNA was isolated. Expression levels of miR-139-3p were measured by qRT-PCR.

Transfection of miRNAs and siRNAs into CRC Cells
The protocols used for transient transfection of miRNAs and siRNAs were described in our previous studies [51,52]. The miRNA precursors and siRNAs used in this report were detailed in Table S2. Opti-MEM (Gibco, Carlsbad, CA, USA) and Lipofectamine TM RNAiMax Transfection Reagent (Invitrogen, Waltham, MA, USA) were used for miRNA and siRNA transfection of miRNAs and siRNAs into CRC cell lines. All miRNA precursors and siRNAs were transfected into the CRC cell line at 10 nM. Mock transfection consisted of cells without precursors or siRNAs. Control groups were transfected with the negative control precursor.

Functional Analyses (Tumor Suppression and Promotion Assays) in CRC Cell Lines
The tumor-suppressive functions of miRNAs were evaluated by transient transfection assays using mature miR-139-5p and miR-139-3p. The tumor-promoting functions of KRT80 (loss-of-function assays) were assessed by siRNA transfection assays using siRNAs targeting KRT80. Functional assays (proliferation, migration, and invasion assays) were performed according to procedures of previous studies [51,52]. Briefly, for proliferation assays, HCT116 or DLD-1 cells were transferred into 96-well plates at 3.0 × 10 3 cells/well. Cell proliferation was assessed using XTT assay kit II (Sigma-Aldrich, St. Louis, MO, USA) 72 h after the transfection procedure. For the migration and invasion assay, HCT116 and DLD-1 cells were transfected in 6-well plates at 3.0 × 10 5 cells/well; 48 h later, transfected HCT116 and DLD-1 cells were added to each chamber at 1.0 × 10 5 cells/well. Corning BioCoat TM cell culture chambers (Corning, Corning, NY, USA) were used for the migration assay and Corning BioCoat Matrigel Invasion Chambers were used for the invasion assay. cells on the underside of the chamber membrane were stained and counted for analysis. All experiments were performed in triplicate. The details of the reagents used in these analyses are listed in Table S2.

RIP Assay
The assay for RIP was performed according to previous studies [55]. Briefly, CRC cells were cultured in 6-well dish at 3.0 × 10 5 /well concentration. Negative control miRNA precursors and miR-139-3p precursors were transfected. After 12 h, immunoprecipitation was performed using the MagCapture TM microRNA Isolation Kit, Human Ago2, obtained from FUJIFILM Wako Pure Chemical Corporation (Wako, Osaka, Japan) according to the manufacturer's protocol. Expression levels of KRT80 and HK2 bound to Ago2 were measured by qRT-PCR. TaqMan assays used in this study are summarized in Table S2. The sequences of primers for SYBR green assays are summarized in Table S3.

Dual-Luciferase Reporter Assay
The dual-luciferase reporter assay was performed to determine whether miR-139-3p binds directly to the 3 -UTR of KRT80. A partial wild-type sequence, including the seed sequence, of the KRT80 3 -UTR, was inserted into the psiCHECK-2 vector (C8021; Promega, Madison, WI, USA). Alternatively, the same KRT80 3 -UTR sequence but with the miR-139-3p binding site deleted was also inserted into the same vector to create the deletion-type construct. The design of each vector cloning sequence into wild-type and deletion-type were shown in Figures S9 and S10 mRNA sequences of KRT80 and HK2 were cited from National Center for Biotechnology Information database [56]. The dual-luciferase reporter assay was performed according to previous studies [17,52]. The reagents used in the assay are listed in Table S2.

Western Blot and Immunohistochemical Analyses
The procedures for Western blot and immunohistochemical analyses were performed according to our previous studies [51,52]. In brief, 72 h after transfection, cells were collected, and lysates were prepared. Next, 18 µg/lane of protein lysate was separated on e-PAGEL (ATTO, Tokyo, Japan), transferred to PVDF membranes, and incubated with primary antibody overnight at 4 • C and with secondary antibody for 1 h at room temperature. GAPDH was used as an internal control. The antibodies used are listed in Table S2, and the clinical specimens evaluated by immunohistochemistry are shown in Table S4.

Statistical Analyses
JMP Pro 15 (SAS Institute Inc., Cary, NC, USA) was used for the statistical analyses. Differences between two groups were assessed using Welch's t-test and those among multiple groups using Dunnett's test. Spearman's test was used for the correlation analyses. A p-value less than 0.05 was considered statistically significant.

Conclusions
Based on the miRNA expression signature of CRC obtained by RNA sequencing, the expression of miR-139-3p (the passenger strand) was significantly reduced in CRC tissues. Functional assays revealed that expression of miR-139-3p attenuated cancer cell malignant phenotypes, indicating that miR-139-3p acts as a tumor suppressor in CRC cells. KRT80 was identified as a direct target of miR-139-3p, and aberrant expression of KRT80 was confirmed in CRC clinical specimens. Moreover, HK2 expression was regulated by both miR-139-3p and KRT80 in CRC cells. Exploration of miRNA-regulated molecular networks provides important information for identifying therapeutic targets for CRC.  Informed Consent Statement: Written prior informed consent and approval were obtained from all patients who were diagnosed with CRC and resected at Kagoshima University Hospital.