FAM64A: A Novel Oncogenic Target of Lung Adenocarcinoma Regulated by Both Strands of miR-99a (miR-99a-5p and miR-99a-3p)

Lung adenocarcinoma (LUAD) is the most aggressive cancer and the prognosis of these patients is unfavorable. We revealed that the expression levels of both strands of miR-99a (miR-99a-5p and miR-99a-3p) were significantly suppressed in several cancer tissues. Analyses of large The Cancer Genome Atlas (TCGA) datasets showed that reduced miR-99a-5p or miR-99a-3p expression is associated with worse prognoses in LUAD patients (disease-free survival (DFS): p = 0.1264 and 0.0316; overall survival (OS): p = 0.0176 and 0.0756, respectively). Ectopic expression of these miRNAs attenuated LUAD cell proliferation, suggesting their tumor-suppressive roles. Our in silico analysis revealed 23 putative target genes of pre-miR-99a in LUAD cells. Among these targets, high expressions of 19 genes were associated with worse prognoses in LUAD patients (OS: p < 0.05). Notably, FAM64A was regulated by both miR-99a-5p and miR-99a-3p in LUAD cells, and its aberrant expression was significantly associated with poor prognosis in LUAD patients (OS: p = 0.0175; DFS: p = 0.0276). FAM64A knockdown using siRNAs suggested that elevated FAM64A expression contributes to cancer progression. Aberrant FAM64A expression was detected in LUAD tissues by immunostaining. Taken together, our miRNA-based analysis might be effective for identifying prognostic and therapeutic molecules in LUAD.


Introduction
Lung cancer is one of the most common and lethal cancers. In 2018, approximately 2.1 million people were diagnosed with this disease, and 1.8 million patients died from it [1]. Lung cancers are divided into two pathological types: small-cell lung cancer and non-small-cell lung cancer (NSCLC). NSCLC includes squamous cell carcinoma, adenocarcinoma and large-cell carcinoma [2]. Among NSCLCs, lung adenocarcinoma (LUAD) is the most common, and it is often at an advanced  Kaplan-Meier plot and log-rank test using survival data from TCGA-LUAD revealed that low expression of miR-99a-5p was associated with a worse prognosis compared with high expression (DFS: p = 0.1264; OS: p = 0.00176) (Figure 2A). Similarly, low expression of miR-99a-3p was associated with a worse prognosis compared with high expression (DFS: p = 0.0316; OS: p = 0.0756) (Figure 2A).  Similarly, the patients were divided into two groups according to the expression levels of miR-99a-5p and miR-99a-3p (top 25%: red lines and low 25%: blue lines) and analyzed. Kaplan-Meier plot and log-rank test showed that low expression of miR-99a-5p was associated with a worse prognosis compared with high expression (DFS: p = 0.0035; OS: p = 0.0005) ( Figure 2B). Low expression of miR-99a-3p was associated with a worse prognosis compared with high expression (DFS: p = 0.0517; OS: p = 0.0139) ( Figure 2B).

Tumor-Suppressive Functions of miR-99a-5p and miR-99a-3p Assessed by Ectopic Expression Assays
We assessed changes in cell proliferation and cell cycle after ectopic expression of these miRNAs into A549 and H1299 cells. Cell proliferation (XTT assay) was significantly inhibited by miR-99a-5p or miR-99a-3p expression in A549 and H1299 LUAD cell lines ( Figure 3A). To investigate the synergistic effects of miR-99a-5p and miR-99a-3p, we performed proliferation assays with co-transfection of miR-99a-5p and miR-99a-3p in LUAD cells (A549 and H1299), but they did not show synergistic effects of these miRNAs transfection (Supplementary Figure S2). In the cell cycle analysis by flow cytometry, the number of LUAD cells in the G0/G1 phase was increased after ectopic expression of these miRNAs compared with control miRNA ( Figure 3B). Our data suggest that ectopic expression of miR-99a-5p and miR-99a-3p induces G1 arrest in LUAD cells.

Tumor-Suppressive Functions of miR-99a-5p and miR-99a-3p Assessed by Ectopic Expression Assays
We assessed changes in cell proliferation and cell cycle after ectopic expression of these miRNAs into A549 and H1299 cells. Cell proliferation (XTT assay) was significantly inhibited by miR-99a-5p or miR-99a-3p expression in A549 and H1299 LUAD cell lines ( Figure 3A). To investigate the synergistic effects of miR-99a-5p and miR-99a-3p, we performed proliferation assays with cotransfection of miR-99a-5p and miR-99a-3p in LUAD cells (A549 and H1299), but they did not show synergistic effects of these miRNAs transfection (Supplementary Figure S2). In the cell cycle analysis by flow cytometry, the number of LUAD cells in the G0/G1 phase was increased after ectopic expression of these miRNAs compared with control miRNA ( Figure 3B). Our data suggest that ectopic expression of miR-99a-5p and miR-99a-3p induces G1 arrest in LUAD cells.

Clinical Significance of miR-99a-5p and miR-99a-3p Target Genes in LUAD Pathogenesis
We evaluated the associations between expression levels of the genes and survival using TCGA and GEO datasets. Among 22 of the 23 target genes (excluding FAM64A), high expression of 18 genes (CKS1B, KCMF1, CENPF, CASC5, MKI67, ESCO2, FANCI, SGOL1, MCM4, KIF11, NEK2, MTHFD2, NCAPG, RRM2, FAM136A, ZWINT, CDK1 and CDKN3) significantly predicted worse survival (OS: 5-year survival rate) in patients with LUAD ( Figure 5). We also adjusted the multiplicity by using Benjamini-Hochberg analysis and confirmed that 19 out of these 23 target genes were significant (Supplementary Table S3). All genes were upregulated in cancer tissues compared with normal tissues ( Figure 6). We also classified these target genes according to Gene Ontology (GO: Biological Process) criteria by using the GeneCodis database. The GO classification of the genes controlled by each miRNA was shown in Supplementary Table S4. Each miRNA has regulated genes associated with "cell cycle (GO: 0007049)" and "cell division (GO: 0051301)".

Clinical Significance of miR-99a-5p and miR-99a-3p Target Genes in LUAD Pathogenesis
We evaluated the associations between expression levels of the genes and survival using TCGA and GEO datasets. Among 22 of the 23 target genes (excluding FAM64A), high expression of 18 genes (CKS1B, KCMF1, CENPF, CASC5, MKI67, ESCO2, FANCI, SGOL1, MCM4, KIF11, NEK2, MTHFD2, NCAPG, RRM2, FAM136A, ZWINT, CDK1 and CDKN3) significantly predicted worse survival (OS: 5-year survival rate) in patients with LUAD ( Figure 5). We also adjusted the multiplicity by using Benjamini-Hochberg analysis and confirmed that 19 out of these 23 target genes were significant (Supplementary Table S3). All genes were upregulated in cancer tissues compared with normal tissues ( Figure 6). We also classified these target genes according to Gene Ontology (GO: Biological Process) criteria by using the GeneCodis database. The GO classification of the genes controlled by each miRNA was shown in Supplementary Table S4. Each miRNA has regulated genes associated with "cell cycle (GO: 0007049)" and "cell division (GO: 0051301)".

Clinical Significance of FAM64A in LUAD Pathogenesis
Overexpression of FAM64A in LUAD tissues was confirmed by RNA-seq data from TCGA-LUAD ( Figure 7A). Spearman's rank test indicated negative correlations of FAM64A expression with Figure 6. Expression levels of pre-miR-99a target genes in LUAD clinical specimens. Using TCGA datasets, the expression levels of all 22 pre-miR-99a target genes evaluated were upregulated in LUAD clinical specimens (n = 475) compared with normal lung tissues (n = 54). The expression data were downloaded from http://firebrowse.org/.

Clinical Significance of FAM64A in LUAD Pathogenesis
Overexpression of FAM64A in LUAD tissues was confirmed by RNA-seq data from TCGA-LUAD ( Figure 7A). Spearman's rank test indicated negative correlations of FAM64A expression with both miR-99a-5p and miR-99a-3p expression ( Figure 7B). We investigated the clinical significance of FAM64A expression in LUAD patients using the TCGA database. High expression of FAM64A was associated with a significantly poor prognosis compared with low expression (DFS: p = 0.0276; OS: p = 0.0175; Figure 7C) and was identified as an independent prognostic factor of survival in the multivariate analysis (p < 0.01; Figure 7D). FAM64A protein expression was also evaluated in LUAD clinical specimens using immunohistochemistry. Overexpression of FAM64A protein was detected in cancer lesions in LUAD clinical specimens (Figure 8).
both miR-99a-5p and miR-99a-3p expression ( Figure 7B). We investigated the clinical significance of FAM64A expression in LUAD patients using the TCGA database. High expression of FAM64A was associated with a significantly poor prognosis compared with low expression (DFS: p = 0.0276; OS: p = 0.0175; Figure 7C) and was identified as an independent prognostic factor of survival in the multivariate analysis (p < 0.01; Figure 7D). FAM64A protein expression was also evaluated in LUAD clinical specimens using immunohistochemistry. Overexpression of FAM64A protein was detected in cancer lesions in LUAD clinical specimens (Figure 8). Spearman's rank test showed a negative correlation between FAM64A and miR-99a-5p or miR-99a-3p expression levels in clinical specimens. (C) Kaplan-Meier survival curves and log-rank comparisons of patients with LUAD using TCGA database. Patients were divided into two groups according to the median FAM64A expression level: high and low expression groups. The red and blue lines represent the high and low expression groups, respectively. (D) Forest plot of the multivariate analysis results assessing independent prognostic factors for disease-free and overall survival, including FAM64A expression (high vs. low) (* p < 0.05, ** p < 0.01, *** p < 0.001). Cells 2020, 9, x 14 of 23

Effects of FAM64A Knockdown on Cell Proliferation and Cell Cycle in LUAD Cells
There is one miRNA-binding site for each miRNA strand (miR-99a-5p and miR-99a-3p) in the 3 UTR region of FAM64A ( Figure 9B). In dual-luciferase reporter assays, luciferase activity was significantly decreased by co-transfection of miR-99a-5p or miR-99a-3p with the vector containing the wild-type 3 UTR of FAM64A in A549 cells. On the other hand, the transfection of the deletion vector (containing the deletion-type 3 UTR of FAM64A) prevented this decrease in luminescence ( Figure 9B), suggesting that miR-99a-5p and miR-99a-3p bind directly to the 3 UTR of FAM64A in LUAD cells.

Effects of FAM64A Knockdown on Cell Proliferation and Cell Cycle in LUAD Cells
To investigate the oncogenic function of FAM64A in LUAD cells, we performed knockdown assays using siRNAs. The expression level of FAM64A was successfully reduced by two different siRNAs (siFAM64A-1 and siFAM64A-2; Figure 10A). Cells 2020, 9, x 16 of 23 To investigate the oncogenic function of FAM64A in LUAD cells, we performed knockdown assays using siRNAs. The expression level of FAM64A was successfully reduced by two different siRNAs (siFAM64A-1 and siFAM64A-2; Figure 10A).
The proliferation of LUAD cells was attenuated by the transfection of each siFAM64A ( Figure  10B).
Cell cycle assays demonstrated that the number of LUAD cells in the G0/G1 phase was increased after knockdown of FAM64A ( Figure 10C). These data indicate that the expression of FAM64A enhances cell cycle progression. The proliferation of LUAD cells was attenuated by the transfection of each siFAM64A ( Figure 10B). Cell cycle assays demonstrated that the number of LUAD cells in the G0/G1 phase was increased after knockdown of FAM64A ( Figure 10C). These data indicate that the expression of FAM64A enhances cell cycle progression.

FAM64A Effects on Molecular Pathways in LUAD
We identified differentially expressed genes from TCGA-LUAD RNA-seq between FAM64A high expression group and low expression group.
GSEA showed that the top signaling pathways enriched in the high FAM64A expression group were cell cycle-associated terms, such as E2F targets, G2M checkpoints, MYC targets and mitotic spindle assembly (Figure 11).

FAM64A Effects on Molecular Pathways in LUAD
We identified differentially expressed genes from TCGA-LUAD RNA-seq between FAM64A high expression group and low expression group.
GSEA showed that the top signaling pathways enriched in the high FAM64A expression group were cell cycle-associated terms, such as E2F targets, G2M checkpoints, MYC targets and mitotic spindle assembly ( Figure 11).
Finally, we found that the proportion of genome alterations (percentage of chromosome regions with copy number alterations relative to all regions evaluated) and the mutation count (the number of mutational events per case) were significantly increased in the high FAM64A expression group (Figure 12), suggesting that FAM64A expression may be associated with genetic mutations and genomic instability in LUAD cells.   Finally, we found that the proportion of genome alterations (percentage of chromosome regions with copy number alterations relative to all regions evaluated) and the mutation count (the number of mutational events per case) were significantly increased in the high FAM64A expression group (Figure 12), suggesting that FAM64A expression may be associated with genetic mutations and genomic instability in LUAD cells.

Discussion
Active genomic research has led to the discovery of driver genes/mutations critical to lung cancer [31]. Molecularly targeted drugs were developed based on these driver genes, and the prognosis of advanced LUAD has greatly improved due to the emergence of molecularly targeted therapeutic agents [32]. However, even with these therapeutic agents, it is difficult to eliminate cancer cells from patients. Continued exploration of molecular networks in LUAD cells provides useful information for developing novel therapeutics.
To identify novel therapeutic targets and pathways, we have previously identified tumor-suppressive miRNAs and their oncogenic targets in LUAD [22][23][24]. A feature of our study is that we analyzed both strands of pre-miRNAs: the guide and passenger strands. The general theory regarding miRNA biogenesis so far is that the passenger strand of a miRNA derived from a pre-miRNA is decomposed in the cytoplasm and has no function [9,10]. Contrary to this belief, recent reports have shown that some passenger strands of miRNAs regulate oncogenes in cancer cells and exert tumor-suppressive functions [33]. Our recent studies demonstrated that some passenger strands of miRNAs, e.g., miR-143-5p, miR-145-3p and miR-150-3p, behave as tumor-suppressive miRNAs in LUAD cells by targeting oncogenes, e.g., LMNB2, MCM4 and TNS4, respectively [22][23][24].
In lung cancer, several studies have shown that miR-99a-5p acts as a tumor-suppressive miRNA by targeting critical oncogenic pathways, including AKT1 and mTOR signaling [34,35]. In contrast, there are few reports on miR-99a-3p function in lung cancer cells. Based on our miRNA signatures, we showed that miR-99a-3p also acts as a tumor-suppressive miRNA in prostate cancer and head and neck squamous cell carcinoma [36,37]. Of particular interest in those papers is that many of the genes identified as targets of miR-99a-3p contribute to malignant phenotypes of cancer cells and significantly predict the worse prognosis of the patients [37]. The search for genes regulated by the passenger strands of miRNAs will provide new information for exploring the molecular mechanisms of LUAD.
In this study, a total of 19 genes (CKS1B, KCMF1, CENPF, CASC5, MKI67, ESCO2, FANCI, SGOL1, MCM4, KIF11, NEK2, MTHFD2, NCAPG, RRM2, FAM136A, ZWINT, CDK1, CDKN3 and FAM64A) identified as pre-miR-99a targets appear to be intimately involved in LUAD pathogenesis. Interestingly, many of these genes are involved in the cell cycle, cell division and chromosome segregation. These molecules are essential for cell division and may be potential targets for cancer drug development. For example, KIF11 is a kinesin, a microtubule-based motor protein that mediates diverse intracellular functions, such as its critical roles in cell division and intercellular vesicle and organelle transport [38,39]. Several inhibitors of KIF11 have entered phase I and II clinical trials [39]. Functional analyses of the genes regulated by pre-miR-99a are useful for exploring molecular networks in LUAD.
In this study, we focused on FAM64A because its expression is regulated by both strands of pre-miR-99a (miR-99a-5p and miR-99a-3p) in LUAD cells. FAM64A (also known as PIMREG, CAKM, CATS and RCS1) was initially identified as a CALM/PICALM-interacting protein using a yeast two-hybrid system [40]. The fusion protein CALM/AF10, t(10;11)(p13;q14), plays a crucial role in acute myeloid leukemia, acute lymphoblastic leukemia and malignant lymphoma [41,42]. Previous studies demonstrated that FAM64A contributes to cell cycle progression [43][44][45]. Overexpression of FAM64A was reported in leukemia, lymphoma and several types of solid cancer [46]. In breast cancer, overexpression of FAM64A enhanced the transactivation of NF-κB by disrupting the NF-κB/IκBα negative feedback loop [47]. Another study demonstrated that FAM64A regulates STAT3 activation and is involved in Th17 differentiation, colitis and colorectal cancer development [48]. These findings indicate that FAM64A behaves as a transcriptional regulator contributing to cell cycle progression. FAM64A might be a potential prognostic factor and therapeutic target in LUAD.