MARCH5 RNA promotes autophagy, migration, and invasion of ovarian cancer cells

ABSTRACT

MARCH5 is a crucial regulator of mitochondrial fission. However, the expression and function of MARCH5 in ovarian cancer have not been determined. This study investigated the expression and function of MARCH5 in ovarian cancer with respect to its potential role in the tumorigenesis of the disease as well as its usefulness as an early diagnostic marker. We found that the expression of MARCH5 was substantially upregulated in ovarian cancer tissue in comparison with the normal control. Silencing MARCH5 in SKOV3 cells decreased TGFB1-induced cell macroautophagy/autophagy, migration, and invasion in vitro and in vivo, whereas the ectopic expression of MARCH5 in A2780 cells had the opposite effect. Mechanistic investigations revealed that MARCH5 RNA may function as a competing endogenous RNA (ceRNA) to regulate the expression of SMAD2 and ATG5 by competing for MIR30A. Knocking down SMAD2 or ATG5 can block the effect of MARCH5 in A2780 cells. Also, silencing the expression of MARCH5 in SKOV3 cells can inhibit the TGFB1-SMAD2/3 pathway. In contrast, the ectopic expression of MARCH5 in A2780 cells can activate the TGFB1-SMAD2/3 pathway. In turn, the TGFB1-SMAD2/3 pathway can regulate MARCH5 and ATG5 through MIR30A. Overall, the results of this study identified MARCH5 as a candidate oncogene in ovarian cancer and a potential target for ovarian cancer therapy.

Introduction

The membrane-associated RING-CH (MARCH) proteins belong to the RING finger protein family of E3 ubiquitin ligases, consisting of 11 members in mammals. MARCH proteins have numerous cellular functions, which include immune regulation, protein quality control, membrane trafficking, mitochondrial fusion and division, and spermatogenesis.¹ MARCH5 is an integral mitochondrial outer membrane protein, with 4 membrane-spanning segments, that belongs to the MARCH family.² MARCH5 is involved in the control of mitochondrial morphology. The overexpression of MARCH5 causes mitochondrial elongation. MARCH5 binds with MFN2 (mitofusin 2) and DNM1L/Drp1, which are involved in mitochondrial fusion and division respectively.^3,4 When MARCH5 is knocked down, most of the mouse embryonic fibroblasts become enlarged and their cell cycle is terminated,⁵ suggesting its crucial role in cell and tissue growth. However, little is known about the cellular localization and function of MARCH5 in epithelial ovarian carcinoma.

In this study, we observed an aberrant expression and localization of MARCH5 in ovarian cancer tissue through immunohistochemical staining. Our results indicate that MARCH5 RNA regulates cellular migration, invasion and autophagy in ovarian cancer cells, and that the effect of MARCH5 is mediated through the TGFB1-SMAD2/3 pathway as well as the competition with MIR30A.

Results

Aberrant expression of MARCH5 in ovarian carcinoma

We examined the expression pattern of MARCH5 in tissue samples of normal ovary and ovarian cancer using immunohistochemistry (IHC) (Fig. 1A–F). MARCH5 immunostaining was significantly higher in epithelial ovarian cancer samples than in normal ovary tissues (P < 0.05; Table 1). Within the cancer tissue samples, MARCH5 staining was significantly higher in tumor samples of advanced stages (stage III/IV) compared with those of early stages (stage I/II) of the disease (P < 0.01). Further, the staining intensity correlated with the tumor grade (grades 2–3 versus 1, P < 0.05). However, the associations between MARCH5 expression and age were not significant (P > 0.05; Table 1). In order to investigate the subcellular location of MARCH5, immunofluorescence was used to detect the expression of MARCH5 in the SKOV3 cell lines. We labeled the mitochondria using MitoTracker Green and then detected MARCH5 using immunohistochemical staining. Our data showed that the location of MARCH5 strongly overlapped the MitoTracker Green. This indicates that MARCH5 was localized predominantly in the mitochondria (Fig. 1G).

Figure 1.

Immunohistochemical analysis of MARCH5 expression in ovarian cancer. MARCH5 was predominantly localized in the plasma membrane (A and B) and the cytoplasm (D, E, and F). MARCH5 expression was not detectable in ovarian surface epithelium and stromal cells in normal ovary (C). The expression of MARCH7 in different types of ovarian cancer samples: (A) Serous papillary adenocarcinoma; (B) serous adenocarcinoma; (C) normal ovarian tissue; (D) mucinous adenocarcinoma; (E) endometrioid adenocarcinoma; (F) adult granulosa cell tumor. (G) Immunohistochemical staining was used to detect the expression of MARCH5 in ovarian cancer SKOV3 cell lines: localization of MARCH5 in SKOV3 cells shows red fluorescent staining in the mitochondria, combined staining with MitoTracker Greenand nuclei staining using DAPI (blue). (H). The mRNA expression of MARCH5 in ovarian cancer cell lines. Scale bar: 100 µm.

Table 1.

Association of MARCH5 expression with clinicopathological characteristics in 77 patients of EOC.

	No. of patients (n = 77)	MARCH5 expression	P value
		Low no. (%) High no. (%)
Characteristics
Age (years)			>0.05
<50	44	13 (29.54) 31 (70.46)
≥50	33	13 (39.40) 20 (60.60)
Normal ovarian	10	10 (100) 0 (0)	<0.05
Cancer tissues	67	41 (61.20) 26 (38.80)
FIGO stage
I/II	46	37 (80.43) 9 (19.57)	<0.01
III/IV	21	4 (19.05) 17 (80.95)
Grade
1	17	14 (82.36) 3 (17.64)
2	21	11 (52.38) 10 (46.62)
3	29	17 (58.62) 12 (41.38)
		Grade 2–3 vs. 1	<0.05

Note. FIGO stage: (stage I) Tumor confined to ovaries; (stage II) Tumor involves one or both ovaries with pelvic extension; (stage III) Tumor involves one or both ovaries, with cytologically or histologically confirmed spread to the peritoneum outside the pelvis and/or metastasis to the retroperitoneal lymph nodes; (stage IV) Distant metastasis excluding peritoneal metastases.

To select suitable cell lines for functional assays, the expression of MARCH5 was screened in 7 cell lines at the mRNA level by real-time quantitative PCR (qPCR). Of these, MARCH5 expression was higher in the SKOV3, CaOV-3, and Es-2 cell lines in comparison with the A2780 cell line (Fig. 1H). Therefore, the A2780 cell line was selected for exogenous expression; the SKOV3 cell line was selected for the downregulation of MARCH5 to determine its functions.

MARCH5 regulation of SMAD2 and ATG5 is MIR30A dependent

To study the function of MARCH5 in SKOV3 cells, we silenced the expression of the MARCH5 gene. The mRNA and protein level of MARCH5 was reduced in LV3–1 and LV3–2 infected SKOV3 cells compared with LV3-NC infected SKOV3 cells (Fig. 2A and B). Bioinformatics analyses (http://starbase.sysu.edu.cn/index.php) showed that ATG5 and SMAD2 are putative ceRNAs of MARCH5. We observed that ATG5 and SMAD2 mRNA and protein expression were significantly reduced in LV3–1- and LV3–2-infected SKOV3 cells relative to LV3-NC-infected SKOV3 cells (Fig. 2C; P < 0.05). However, the mRNA and protein expression of ATG5 and SMAD2 were upregulated when the MARCH5 3′ untranslated region (UTR) was ectopically expressed in A2780 cells (Fig. 2D; P < 0.05).

Figure 2.

MARCH5 regulates SMAD2 and ATG5 mRNA and protein levels. (A and B) MARCH5 mRNA and protein levels were downregulated by infection with LV3–1 and LV3–2. (C) Ovarian cancer SKOV3 cells were infected with LV3-NC and LV3–1. SMAD2 and ATG5 mRNA and protein levels were downregulated by infection with LV3–1 and LV3–2. (D) Ovarian cancer SKOV3 cells were transfected with MARCH5 3′ UTR. SMAD2 and ATG5 mRNA and protein levels were upregulated. (E) Putative binding sites targeted by MIR30A were predicted to be located in the 3′ UTR of MARCH5, ATG5, and SMAD2 mRNA. (F) SKOV3 cells were cotransfected with MIR30A mimics or control RNA (NC) with luciferase reporter plasmids containing either wild-type (pMIR-MARCH5–3UTR, pMIR-ATG5–3UTR, and pMIR-SMAD2–3UTR) or mutant 3′ UTR (pMIR-MARCH5–3UTRm, pMIR-ATG5–3UTRm, and pMIR-SMAD2–3UTRm) of MARCH5, ATG5, and SMAD2 genes. Luciferase expression was measured. The fold changes of the relative luciferase activity in MIR30A mimics with the indicated plasmids transfected cells were normalized to NC with the corresponding indicated plasmid-transfected cells. (G) Ovarian cancer SKOV3 cells were transfected with MIR30 mimics or control RNA (NC). (H) Ovarian cancer SKOV3 cells were transfected with MIR30 inhibitor or control RNA (NC). (I) Ovarian cancer SKOV3 cells were infected with LV3-NC and LV3–1. One group was infected with LV3–1-transfected MIR30A inhibitors. Error bars represent standard error. The symbols * and ** indicate p < 0.05 and 0.01, respectively.

We investigated whether the effect of MARCH5 on ATG5 and SMAD2 is miRNA dependent. Interestingly, bioinformatics analyses showed that ATG5, MARCH5, and SMAD2 are predictive targets of MIR30A (http://starbase.sysu.edu.cn/index.php). Thus, we attempted to experimentally verify whether MIR30A modulated MARCH5, ATG5, and SMAD2 expressions in SKOV3 cells. We predicted that a MIR30A-specific binding site was located within the 3′ UTR of MARCH5, ATG5, and SMAD2 mRNAs (Fig. 2E) and constructed a vector to investigate if MIR30A could directly target the 3′ UTR of these genes. We found that MIR30A markedly inhibited luciferase activity when the MARCH5, ATG5, or SMAD2 3′ UTRs were inserted downstream of the luciferase cDNA in our reporter vector (pMIR-MARCH5 3UTR, pMIR-ATG5 3UTR, pMIR-SMAD2 3UTR). In contrast, no significant suppression of luciferase activity was observed in cells transfected with a control vector with mutant MARCH5, ATG5, and SMAD2 3′ UTR (MIR- MARCH5, ATG5 and SMAD2 3UTRm) when MIR30A expression was elevated (Fig. 2F; P < 0.05). We also found that MIR30A mimics could downregulate the mRNA and protein levels of MARCH5, ATG5, and SMAD2 (Fig. 2G; P < 0.05), whereas MIR30A inhibitor could upregulate the mRNA and protein levels of these genes (Fig. 2H; P < 0.05). These data indicate that MARCH5, ATG5, and SMAD2 are direct targets of MIR30A.

Furthermore, we found that MARCH5s regulation of ATG5 and SMAD2 was eliminated when MIR30A was inhibited (Fig. 2I; P < 0.05). These data indicate that MARCH5s regulation of ATG5 and SMAD2 is MIR30A dependent.

ATG5 and SMAD2 modulation of MARCH5 levels are MIR30A dependent

Since ATG5 and SMAD2 are ceRNAs of MARCH5, we predicted that ATG5 and SMAD2 can regulate MARCH5. Indeed, we observed that MARCH5 was decreased when ATG5 or SMAD2 was silenced (Fig. 3A and B; P < 0.05). The decrease in MARCH5 was reversed when MIR30A was absent (Fig. 3A and B; P < 0.05). We also found that the silencing of ATG5 and SMAD2 can inhibit SMAD2 and ATG5, respectively. These outcomes were also dependent on the presence of MIR30A (Fig. 3C and D; P < 0.05). Based on the above results, we infer that MARCH5, ATG5, and SMAD2 are reciprocal ceRNAs by competing for MIR30A.

Figure 3.

ATG5 and SMAD2 modulation of MARCH5 levels are MIR30A dependent. (A) Ovarian cancer SKOV3 cells were transfected with control RNA (NC), ATG5 siRNA, and ATG5 siRNA + MIR30A inhibitor. (B) Ovarian cancer SKOV3 cells were transfected with control RNA (NC), SMAD2 siRNA, and SMAD2 siRNA⁺ MIR30A inhibitor. (C) Ovarian cancer SKOV3 cells were transfected with control RNA (NC), ATG5 siRNA, and ATG5 siRNA + MIR30A inhibitor. (D) Ovarian cancer SKOV3 cells were transfected with control RNA (NC), SMAD2 siRNA, and SMAD2 siRNA + MIR30A inhibitor. Error bars represent standard error. The symbols * and **indicate p < 0.05 and 0.01, respectively.

MARCH5 regulates TGFB1-induced cell autophagy, migration and invasion

ATG5 and SMAD2 are involved in TGFB1-induced autophagy. Hence, we determined whether MARCH5 regulated TGFB1-induced autophagy. Ovarian cancer SKOV3 cells and A2780 cells were treated with TGFB1 (0, 10, 20, and 50 ng/mL) for 24 h. TGFB1 increased the protein level of MAP1LC3/LC3 (microtubule associated protein 1 light chain 3) in a dose-dependent manner (Fig. 4A). We observed that the expression of LC3-II was decreased in LV3–1- and LV3–2-infected SKOV3 cells in the presence of TGFB1 (10ng/mL) relative to LV3-NC-infected SKOV3 cells, whereas LC3-II was significantly increased when TGFB1 was added to LV5-MARCH5-infected A2780 cells (Fig. 4B). We also found that cell autophagy, migration and invasion were decreased in TGFB1-induced SKOV3 cells after MARCH5 knockdown. However, the ectopic expression of MARCH5 in TGFB1-induced A2780 cells had the opposite effect. These data indicate that MARCH5 regulated TGFB1-induced autophagy, migration, and invasion (Fig. 4C–F).

Figure 4.

MARCH5 regulates TGFB1-induced cell autophagy, migration and invasion. (A) The expression of LC3 protein level in SKOV3 and A2780 cells was regulated by TGFB1. (B) Ovarian cancer SKOV3 cells were infected with LV3-NC, LV3–1, and LV3–2. Ovarian cancer A2780 cells were infected with LV5-GFP and LV5-MARCH5. Forty-eight h after infection, puromycin was added at a concentration of 2.5 µg/ml. After 48 h, TGFB1 was added at a concentration of 10 ng/ml. The expression of LC3-II was detected by western blot after TGFB1 treatment for 24 h. (C) SKOV3 cells transfected with a plasmid encoding mRFP-GFP-LC3 and MARCH5 or control siRNAs. After 24 h, TGFB1 was added at a concentration of 10 ng/ml. mRFP-GFP-LC3 distribution in SKOV3 was analyzed by confocal microscopy after TGFB1 treatment for 24 h. The LC3 dots were quantified using image pro-plus 6.0 software. All experiments were repeated 3 times and the representative results are shown. The right panel indicates the quantification of LC3 puncta numbers. (D) A2780 cells transfected with mRFP-GFP-LC3 and pCMV5-MARCH5 or pCMV5 empty vector. After 24 h, TGFB1 was added at a concentration of 10 ng/ml. mRFP-GFP-LC3 distribution was analyzed by confocal microscopy after TGFB1 treatment for 24 h. The LC3 dots were quantified using image pro-plus 6.0 software. All experiments were repeated 3 times and the representative results are shown. The right panel indicates the quantification of LC3 puncta numbers. (E) Ovarian cancer SKOV3 and A2780 cells migration ability was detected by a wound healing assay. The SKOV3 cells were infected with LV3-NC, LV3–1 and LV3–2, and the A2780 cells were infected with LV5-GFP and LV5-MARCH5. After 48 h, puromycin was added at a concentration of 2.5 µg/ml. After 72 h, the migration assays were performed in the presence of TGFB1 (10 ng/ml). (F) Ovarian cancer SKOV3 and A2780 cells invasion ability was detected by a Matrigel invasion assay. The SKOV3 cells were infected with LV3-NC, LV3–1 and LV3–2, and the A2780 cells were infected LV5-GFP and LV5-MARCH5. After 48 h, puromycin was added at a concentration of 2.5 µg/ml. After 72 h, the invasion assays were performed in the presence of TGFB1 (10 ng/ml). Error bars represent standard error. The symbols * and ** indicated p < 0.05 and 0.01, respectively. Scale bar: 100 µm.

MARCH5 promotion of SKOV3 cell autophagy, migration, and invasion involves MIR30A, ATG5, SMAD2 and the TGFB1-SMAD2/3 pathway

The role of MARCH5 in tumor migration and invasion of ovarian cancer SKOV3 cells was investigated in an animal model. The tumor burden was significantly lower in the LV3–1-infected group than in the LV3-NC-infected group (P < 0.01) (Fig. 5A). IHC revealed that the expression of ATG5 and SMAD2 in tumors of the LV3–1-infected group was lower than in the LV3-NC-infected group (P < 0.01) (Fig. 5B). These data show that silencing MARCH5 inhibited tumor migration and invasion in vivo.

Figure 5.

Silencing MARCH5 inhibited pelvic peritoneal metastasis in a nude mice model. (A) Mean tumor weight on d 35 after tumor cell injection. LV3–1- and LV3-NC-infected SKOV3 cells were implanted by IP injection. The arrow indicates the lesion in the abdominal cavity. (B) Immunohistochemical analysis of MARCH5, ATG5, SMAD2, and p-SMAD2 expression was performed on tumor xenografts. Representative images are shown (original magnification ×200). Error bars represent standard error. The symbols * and **indicate p < 0.05 and 0.01, respectively. Scale bar: 100 µm.

We have demonstrated that ATG5 and SMAD2 can regulate the expression of MARCH5. Hence, we presumed that MARCH5 regulation of TGFB1-induced cell autophagy, migration and invasion requires ATG5 and SMAD2. When ATG5 or SMAD2 was silenced, the cell autophagy, migration and invasion were inhibited in A2780 cells ectopically expressing MARCH5 in the presence of TGFB1. We also observed that cell autophagy, migration and invasion were decreased in TGFB1-induced MARCH5 ectopically expressing A2780 cells when the TGFB1-SMAD2/3 pathway was inhibited using LY2109761 (10 µM) (Fig. 6A and B).

Figure 6.

MARCH5 promotion of SKOV3 cell autophagy, migration, and invasion involves ATG5, SMAD2 and the TGFB1-SMAD2/3 pathway. (A) A2780 cells infected with LV5-GFP and LV5-MARCH5. After 48 h, puromycin was added at a concentration of 2.5 µg/ml. The cells were transfected with ATG5 siRNA, SMAD2 siRNA or LY2109761 was added (10 µM) for 48 h. The migration and invasion assays were performed in the presence of TGFB1 (10 ng/ml). (B) SKOV3 cells transfected with a plasmid encoding mRFP-GFP-LC3 and pCMV5 empty vector, pCMV5-MARCH5, pCMV5-MARCH5 + siATG5, pCMV5-MARCH5 + siSMAD2, and pCMV5-MARCH5 + LY2109761. After 48 h, TGFB1 was added at a concentration of 10 ng/ml. mRFP-GF-LC3 distribution in SKOV3 was analyzed by confocal microscopy after TGFB1 treatment for 24 h. The LC3 dots were quantified using image pro-plus 6.0 software. All experiments were repeated 3 times and the representative results are shown. The right panel indicates the quantification of LC3 puncta numbers. Error bars represented standard error. The symbols * and ** indicate p < 0.05 and 0.01, respectively. Scale bar: 100 µm.

MARCH5 is a target of MIR30A. Thus, we investigated the role of MIR30A in cellular autophagy, migration and invasion. Our results show that the expression of LC3-II and the ability of migration and invasion were decreased when MIR30A mimics were transfected into TGFB1-induced SKOV3 cells (Fig. 7A–C). These phenotypes were restored by the ectopic expression of MARCH5.

Figure 7.

MIR30A regulates ovarian cancer SKOV3 cell migration, invasion and autophagy in the presence of TGFB1. (A) Ovarian cancer SKOV3 cells were transfected with LV5-GFP + MIR30A NC, LV5-GFP + MIR30 mimics or LV5-MARCH5 + MIR30 mimics. After 48 h, the invasion ability of ovarian cancer SKOV3 cells was detected by a Matrigel invasion assay in the presence of TGFB1 (10 ng/ml). (B) Ovarian cancer SKOV3 cells were transfected with LV5-GFP + MIR30A NC, LV5-GFP + MIR30 mimics or LV5-MARCH5 + MIR30 mimics. After 48 h, the migration ability of ovarian cancer SKOV3 cells was detected by a wound healing assay in the presence of TGFB1 (10 ng/ml). (C) SKOV3 cells were transfected with a plasmid encoding mRFP-GFP-LC3 and pCMV5 empty vector (PCMV5 NC) + MIR30A NC, pCMV5 NC + MIR30A mimics or pCMV5-MARCH5 + MIR30A mimics. After 48 h, TGFB1 was added at a concentration of 10 ng/ml. mRFP-GFP-LC3 distribution in SKOV3 was analyzed by confocal microscopy after TGFB1 treatment for 24 h. The LC3 dots were quantified using image pro-plus 6.0 software. All experiments were repeated 3 times and the representative results are shown. The right panel indicates the quantification of LC3 puncta numbers. Error bars represented standard error. The symbols * and ** indicate p < 0.05 and 0.01, respectively. Scale bar: 100 µm.

The TGFB1-SMAD2/3 pathway regulates MARCH5, ATG5, and SMAD2 through MIR30A

We examined whether MARCH5 regulates the TGFB1-SMAD2/3 pathway. We found that the ectopic expression of MARCH5 in ovarian cancer A2780 cells significantly increased the TGFB1-SMAD2/3 pathway, based on luciferase reporter activity (P < 0.01). The TGFB1-SMAD2/3 luciferase reporter activity was significantly decreased in LV3–1- and LV3–2-infected SKOV3 cells (Fig. 8A and B). We also found that phosphorylated (p)-SMAD2 was regulated by MARCH5. However, this regulation was decreased when ATG5 and SMAD2 were silenced (Fig. 8C). These data indicate that MARCH5 regulation of the TGFB1-SMAD2/3 pathway requires ATG5 ceRNAs (ATG5 and SMAD2).

Figure 8.

The TGFB1-SMAD2/3 pathway regulates MARCH5, ATG5, and SMAD2 through MIR30A. (A) The TGFB1–SMAD2/3 signal pathway luciferase reporter activity was detected in ectopically expressing MARCH5 ovarian cancer A2780 cells infected with LV5-MARCH5. (B) The TGFB1–SMAD2/3 signal pathway luciferase reporter activity was detected in LV3–1-, LV3–2-, and LV3-NC-infected SKOV3 cells. (C) Ovarian cancer SKOV3 cells were transfected with LV5-GFP, LV5-MARCH5, LV5-MARCH5 + ATG5 siRNA, or LV5-MARCH5 + SMAD2 siRNA. The protein expression of p-SAMD2 was detected by western blot. (Di) The expression of MARCH5, ATG5, and SMAD2 mRNA in SKOV3 cells was regulated by TGFB1 and MIR30A mimics. (Dii) The expression of MIR30A in SKOV3 cells was regulated by TGFB1. (Diii) The expression of MARCH5, ATG5, and SMAD2 mRNA in SKOV3 cells was regulated by LY2109761 and MIR30A inhibitor. (Div) The expression of MIR30A in SKOV3 cells was regulated by LY2109761. (E) The putative binding sites between MIR30A and SMAD2/3. (F) A graphic abstract for the significance of the MARCH5 pathway (MIR30A, SMAD2/3, MARCH5, ATG5, and SMAD2). Data are expressed as mean ± SD from 3 independent experiments. * p < 0.05, and **p < 0.01.

We also determined whether the TGFB1-SMAD2/3 pathway could regulate MARCH5, ATG5, SMAD2, and MIR30A. The activation of the TGFB1-SMAD2/3 pathway using TGFB1 increased the expressions of MARCH5, ATG5, and SMAD2 but inhibited MIR30A expression. MIR30A mimics inhibited the upregulation of MARCH5, ATG5, and SMAD2 induced by TGFB1. MIR30A mimics partly inhibited the upregulation of MARCH5, ATG5, and SMAD2 induced by TGFB1 (Fig. 8 Di and 8 Dii). LY2109761 inhibited MARCH5, ATG5, and SMAD2 expression but increased MIR30A expression. In contrast, MIR30A inhibitors partly reversed the downregulation of MARCH5, ATG5, and SMAD2 induced by LY2109761 (Fig. 8 Diii and 8 Div).

Bioinformatics analyses (http://deepbase.sysu.edu.cn/chipbase/index.php) showed that there is a potential SMAD2/3 binding site upstream of the MIR30A gene (Fig. 8E). By using a chromatin immunoprecipitation (ChIP) assay, we determined that SMAD2/3 interacted with the MIR30A promoter, indicating that MIR30A may be a transcriptional target of the TGFB1-SMAD2/3 signaling pathway (Fig. 8F).

Discussion

In this study, we observed that MARCH5 expression was higher in epithelial ovarian cancer tissues compared with normal ovarian tissues. The upregulation of MARCH5 in ovarian cancer correlated with the tumor stage and histological grades. Our results also show that the upregulation of MARCH5 expression in epithelial ovarian cancer cells promoted migration, invasion, and autophagy. These data suggest that an elevated level of MARCH5 aids the progression of ovarian cancer, and promotes an aggressive behavior. Based on these findings, MARCH5 can function as a novel tumor marker as well as a potential therapeutic target for ovarian cancer. Our data also demonstrate that MARCH5 can regulate ATG5 and SMAD2 in a MIR30A-dependent manner; it can also regulate the TGFB1-SMAD2/3 pathway and conversely, SMAD2/3 regulates MARCH5 through MIR30A.

As a cellular means of clearing damaged or superfluous proteins and organelles, autophagy has been considered to play an important role in maintaining the quality control of proteins and organelles as well as energy homeostasis.⁶ Although the assessment of autophagosome formation at a single time point may be used for evaluating the extent of autophagy, the results could be misleading because the extent of autophagosome formation often dissociates from the level of autophagic flux.⁷ Conversely, tandem fluorescent mRFP-GFP-LC3 allows one to evaluate the extent of autophagosome and autolysosome formation simultaneously, because LC3 puncta labeled with both GFP and mRFP represent autophagosomes, whereas those labeled with mRFP alone represent autolysosomes. Increased red spots appear in the merged section, indicating the formation of autolysosomes in these cells.⁸ Accordingly, we introduced the mRFP-GFP-LC3 reporter to determine the role of MARCH5 in autophagy flux. Our data showed that MARCH5 knockdown reduced the number of yellow and red spots. In contrast, ectopic expression of MARCH5 in TGFB1-induced A2780 cells increased the number of red spots. Our data suggest that MARCH5 knockdown blocked autophagy, whereas ectopic expression of MARCH5 activated autophagy.

Competing endogenous RNAs refer to RNA transcripts, including mRNAs, noncoding RNAs, pseudogene transcripts and circular RNAs, that can regulate each other by competing for the same pool of miRNAs.⁹ To explore the function of MARCH5 in epithelial ovarian cancer, we predicted the candidate ceRNAs of MARCH5 by bioinformatics analyses.¹⁰ Our results indicate that MARCH5, ATG5, and SMAD2 regulate each other by competition for MIR30A. In cancer, autophagy may promote carcinoma cell invasion and this process correlates with poor prognosis.^11,12 In our study, we found that the blockage of autophagy flux by the downregulation of ATG5 or MARCH5 expression led to the inhibition of TGFB1-induced cellular migration and invasion. Our results are inconsistent with previous reports, which indicate that the inhibition of autophagy impairs tumor cell invasion.¹³ The MIR30 family impairs autophagy in human carcinoma by targeting BECN1, ATG5, and ATG2B.^14,15 MIR30B regulates the migration and invasion of human colorectal cancer via SIX1.¹⁶ MIR30A suppresses cell migration and invasion by downregulating PIK3CD in colorectal carcinoma.¹⁷ In this study, MIR30A inhibited autophagy, migration, and invasion in ovarian cancer. However, the phenotypes can be partially restored by the expression of a MIR30A-resistant MARCH5. These data confirm the role of MARCH5 in regulating cellular autophagy, migration, and invasion.

TGFB (ransforming growth factor β) has broad impacts on an array of diverse cellular functions including cell growth, differentiation, adhesion, migration, and apoptosis.^18,19 The TGFB1 signaling pathway induces autophagy in certain human cancer cells, and the induction of autophagy is a novel aspect of the biological functions of TGFB1.¹⁹ In this study, we found that TGFB1 induced autophagy in ovarian cancer cells. The overexpression of MARCH5 promoted TGFB1-induced autophagy whereas silencing MARCH5 inhibited TGFB1-induced autophagy. These results demonstrate that MARCH5 not only regulates cellular autophagy but is also involved in the TGFB1-SMAD2/3 pathway. TGFB1 can promote tumor invasion and metastasis in ovarian cancer.²⁰ We also observed that MARCH5 modulated TGFB1-induced ovarian cell migration and invasion. Indeed, MARCH5 and SMAD2 regulate each other by competing for MIR30A. Our results confirm that MARCH5 can activate the TGFB1-SMAD2/3 pathway.

Previous studies have shown that TGFB1 upregulates MARCH5 in cancer cells.¹⁹ In our study, we found that the TGFB1-activated TGFB1-SMAD2/3 pathway increased the expression of MARCH5 and ATG5 but inhibited MIR30A expression. In contrast, MIR30A inhibitors reversed the downregulation of MARCH5 and ATG5 induced by LY2109761. Moreover, MIR30A mimics inhibited the upregulation of MARCH5, ATG5, and SMAD2 induced by TGFB1. These data indicated that the TGFB1-SMAD2/3 pathway modulated MARCH5 and ATG5 through MIR30A.

In conclusion, this study confirms that MARCH5 is a tumor-promoting gene in ovarian cancer, which is involved in the TGFB1-SMAD2/3 pathway. These results suggest that MARCH5 may be a potential therapeutic target in epithelial ovarian cancer.

Materials and methods

Tissue specimens

The tissue microarray slides containing malignant and normal ovarian tissues (n = 77) were obtained from US Biomax Inc. cancer tissue bank collection (US Biomax, ov2085). The Ethics Committee of the Chongqing Medical University approved the study documents and the use of archived cancer tissues.

Cell culture, transfection procedure, and reagents

Human ovarian cancer cells were cultured in RPMI 1640 medium (Sigma-Aldrich, R8758) containing 10% fetal bovine serum and antibiotics. The cells were incubated under 5% carbon dioxide at 37°C. Double-strand oligonucleotides corresponding to the target sequences were synthesized by Genepharma (Shanghai, China). The following sequences were targeted for human MARCH5, ATG5, and SMAD3 small interfering RNA (siRNA): MARCH5–1:

5′-GCACUUGGGAGUAAUUUGA-3′; MARCH5–2: 5′-GCACACGUGUCCGAUUUAU-3′;

SMAD2: 5′-GUCCCAUGAAAAGACUUAA-3′;

SMAD3: 5′-GAGCUUGGUGAAGAAGCUCUU-3′;

ATG5: 5′-CAAUCCCAUCCAGAGUUGCUUGUGA-3′; and NC (negative control) siRNA: 5′-UUCUUCGAAGGUGUCACGUTT-3′. Lentiviral vectors expressing shRNA targeting MARCH5 (named LV3–1 and LV3–2) and the MARCH5-lentiviral expression vector (named LV5-MARCH5) were provided by Genepharma. MIR30A mimics (sense: 5′-UGUAAACAUCCUCGACUGGAAG-3′) were synthesized at Ruibo Biotechnology (Guangzhou, China). Also, we constructed pCMV5-March5 for the overexpression of MARCH5.

Immunohistochemistry

IHC was performed according to the SP kit instructions (ZSGB-BIO ORIGENE, SP-9000). After dewaxing and hydration, the sections were heated in citrate buffer (Sigma-Aldrich, PBS1) in a microwave oven for 20 min for antigen retrieval. Next, the sections were cooled naturally to room temperature and washed thrice for 3 min per cycle. Subsequently, the sections were incubated in 3% hydrogen peroxide (ZSGB-BIO ORIGENE, SP-9000) for 15 min at room temperature and washed thrice with phosphate-buffered saline (PBS; Sigma-Aldrich, P5368) for 3 min per cycle. The sections were then blocked with 5% goat serum (Bioss Biotechnology Company, C-0005) for 30 min at 37°C. Anti-MARCH5 rabbit polyclonal antibody (Bioss Biotechnology Company, bs-9339R) was incubated with the sections overnight at 4°C. Negative controls included the omission of primary antibodies and the use of irrelevant primary antibodies. The corresponding secondary antibodies that were conjugated to horseradish peroxidase (Bioss Biotechnology Company, bs-0295D) were incubated with the sections for 1 h at room temperature. The sections were washed thrice in PBS for 3 min per cycle. The sections were then incubated in horseradish enzyme-labeled chain avidin solution ZSGB-BIO (ORIGENE, SP-9000) for 30 min at 37°C and washed in PBS for 3 min × 3 cycles. The proteins were visualized by diaminobenzidine (ZSGB-BIO ORIGENE, ZLI-9017). The staining data were obtained from manually recorded reports. Staining intensity was graded on a 0–3 scale as follows: 0, absence of staining; 1, weakly stained; 2, moderately stained; and 3, strongly stained. The percentage of positive tumor cells was scored as follows: 0, absence of tumor cells; 1, <33% of tumor cells; 2, 33–66% of tumor cells; and 3, >66% of tumor cells. The immunohistochemical score (ranging from 0 to 9) was calculated by multiplying the intensity score and the percentage score.²¹

Immunohistochemical staining

Ovarian cancer SKOV3 cells were seeded on sterile glass coverslips at 37°C for 48 h. Mitochondrial staining using MitoTracker Green (Invitrogen, M7514) was performed as described.²² Briefly, cells were incubated in serum-free medium with 150 nM MitoTracker Green FM for 30 min in the dark. After staining, cells were washed twice with cold PBS. Then, the cells were fixed in methanol at room temperature for 15 min. Subsequently, cells were blocked with 10% goat serum to eliminate nonspecific binding at 37°C for 30 min, then incubated with the MARCH5 antibody (Bioss Biotechnology Company, bs-9339R) overnight at 4°C. Following washes, cells were incubated with RBITC-labeled goat anti-rabbit (Bioss Biotechnology Company, bs-0295G-RBITC) secondary antibodies at 37°C for 30 min. Following incubation, the slides were washed 3 times with PBS and cell nuclei were stained with DAPI (Sigma-Aldrich, D9542). Fluorescence was detected using confocal laser scanning microscopy.

Tandem mRFP-GFP-LC3 fluorescence

A tandem mRFP-GFP-tagged LC3 was used to monitor autophagy flux as previous reported.⁸ SKOV3 cells transfected with mRFP-GFP-LC3 and MARCH5 or control siRNAs. A2780 cells transfected with a plasmid encoding mRFP-GFP-LC3 and pCMV5-MARCH5 or pCMV5 empty vector. After 24 h, TGFB1 was added at a concentration of 10 ng/ml. mRFP-GFP-LC3 distribution in SKOV3 or A2780 cells was analyzed by confocal microscopy after TGFB1 treatment for 24 h. The LC3 dots were quantified using the image pro-plus 6.0 software. First, we labeled the dots using image pro-plus 6.0 software. Next, the software performed automatic counting of the LC3 puncta. Finally, we calculated the average dots per cell. Twenty cells were quantified (done in triplicate) as previously reported.¹⁹ All experiments were repeated 3 times.

Quantitative real-time polymerase chain reaction (RT-qPCR)

Total RNA was isolated using a pure high-purity Total RNA Rapid Extraction Kit (Bioteke Corporation, RP1201) according to the manufacturer's instruction. cDNA was synthesized using the iSCRIPT cDNA synthesis kit (Bio-Rad Laboratories, 4106228). The primers used for amplifying MARCH5, ATG5, SMAD2, and GAPDH were synthesized by GeneCopoeia. The real-time PCR kit was purchased from GeneCopoeia. PCR conditions were 95°C for 10 sec, 60°C for 20 sec, and 72°C for 10 sec. Each sample was analyzed in triplicate. Relative quantification of mRNA was performed using the comparative threshold cycle (CT) method. This value was used to plot the gene expression using the formula 2^{−Δ ΔCT}.

Detection of protein expression by western blotting

The expressions of SMAD2, p-SMAD2, ATG5, MARCH5, and GAPDH proteins were analyzed by western blot.²³ The primary antibodies used include polyclonal rabbit anti-MARCH5 (Bioss Biotechnology Company, bs-9339R); rabbit monoclonal to SMAD2 (Abcam, ab40855); rabbit monoclonal to p-SMAD2 (Abcam, ab188334); and polyclonal rabbit anti-GAPDH (Santa Cruz Biotechnology, AB10016). The band density was analyzed using a gel imaging system and compared with an internal control.

Dual-luciferase reporter gene assay

Luciferase reporter gene assay was performed using the Dual-Luciferase Reporter Assay System (Promega Corporation, E1910) according to the manufacturer's instruction. For 3′ UTR luciferase reporter assays for MARCH5, SMAD2, and ATG5, wild-type or mutant reporter constructs (termed WT or Mut; purchased from Genepharma) were cotransfected into SKOV3 cells in 24-well plates with 100 nM MIR30A or 100 nM miR-NC and Renilla plasmid by using Endofectin-Plus (GeneCopoeia, Z01010A). SMAD2/3 reporter plasmids were purchased from Shanghai Qcbio Science & Technologies (GM021043). A reporter gene assay was performed 48 h post-transfection using the Dual Luciferase Assay System. Firefly luciferase activity was normalized for transfection efficiency using the corresponding Renilla luciferase activity. All experiments were performed at least 3 times.

Chromatin immunoprecipitation (ChIP) assay

ChIP assay was performed using the ChIP Assay Kit (Cell Signaling Technology, 9003) according to the manufacturer's instructions. Briefly, cross-linked chromatin was sonicated into 200- to 1,000-bp fragments. Then, the chromatin was immunoprecipitated using anti-SMAD2/3 antibody. Quantitative PCR was conducted according to the method described earlier.

Migration and Matrigel invasion assays

The migration of SKOV3 and A2780 cells were analyzed using the wound-healing assay in vitro. The SKOV3 cells were infected with LV3-NC, LV3–1 and LV3–2, while the A2780 cells were infected with LV5-GFP or LV5-MARCH5. After 48 h, puromycin was added at a concentration of 2.5 µg/ml. After 72 h, the migration assays were performed as follows: Cells were seeded into 6-well plates and cultivated until 90% growth confluence was achieved. TGFB1 was added at a concentration of 10 µg/ml. Adherent cell gaps were inflicted by scraping the monolayer cells with a sterile pipette tip. At 0 and 48 h after the scraping, the cells were observed under the low power of an Olympus light microscope. The distance across the gap was measured at each time point and expressed as the average percentage of the gap closure compared with that at zero time.

The invasion of SKOV3 and A2780 cells were evaluated by Matrigel invasion assay. The SKOV3 cells were infected with LV3-NC, LV3–1 and LV3–2, while the A2780 cells were infected with LV5-GFP or LV5-MARCH5. After 48 h, puromycin was added at a concentration of 2.5 µg/ml. After 72 h, the invasion assays were performed as follows: The upper side of the 8-µm pore and the 6.5-mm polycarbonate transwell filter (Corning Inc., CLS3422) chamber were uniformly coated with Matrigel basement membrane matrix (BD Biosciences, 356234) for 2 h at 37°C before the cells were added. A total of 5 × 10⁴ infected cells (SKOV3 cells infected with LV3-NC, LV3–1 and LV3–2, and A2780 cells infected with LV5-GFP or LV5-MARCH5) were seeded into the top chamber of a transwell filter (in triplicate) and incubated in a serum-free medium containing TGFB1 at a concentration of 10 µg/ml for 48 h. The invasive cells on the lower side of the filter were fixed with 4% paraformaldehyde, stained with 0.5% crystal violet (Beyotime Institute of Biotechnology, C0121), and counted using a microscope. A total of 5 fields were counted for each transwell filter. Each field was counted and photographed at 200× magnification.

Mouse orthotopic xenograft model of ovarian cancer

The xenograft model was described previously.²⁴ All procedures for animal experiments were approved by the Committee on the Use and Care of Animals (Chongqing Medical University, Chongqing, China) and performed in accordance with the institution's guidelines. Ovarian cancer SKOV3 cells were infected with LV3–1 and LV3-NC and injected intraperitoneally (IP) into 6-wk-old BALB/c nude mice (5 × 106 cells per mouse in 200 ul). Five wk after the IP injections, the animals were sacrificed to confirm the presence of tumors and weigh the established tumors. After the sacrifice, the degree of ascites was quantified; the number of metastases was counted and carefully dissected, and the removed tumors were weighed.

Statistical analysis

All statistical analyses were performed using SPSS software, version 17.0 (Chicago, IL). Each experiment was performed in triplicate. Statistical analysis was performed by Student t test or analysis of variance (ANOVA). The chi-square test was used to compare the associations between MARCH5 overexpression and the clinicopathological variables of ovarian cancer samples. Data were presented as mean ± standard deviation. Statistical significance was defined as a p-value less than 0.05.

Abbreviations

3′ UTR

3′-untranslated region

ATG5

autophagy related 5

ChIP

chromatin immunoprecipitation

ceRNA

competing endogenous RNA

GFP

green fluorescent protein

IHC

Immunohistochemistry

IP

intraperitoneally

MAP1LC3/LC3

microtubule associated protein 1 light chain 3

MIR30A

imicroRNA 30a

LV3-NC

LV3-shMARCH5-negative control

LV3–1

LV3-shMARCH5–1

LV3–2

LV3-shMARCH5–2

RFP

red fluorescent protein

TGFB1

transforming growth factor β 1

Disclosure of potential conflicts of interest

No potential conflicts of interest were disclosed.

Acknowledgments

We thank Dr. LiFei for editing our manuscript.

Funding

This work was supported by grants from the Natural Science Foundation of Chongqing (CSTC 2012JJB10030), Chongqing Municipal Health Bureau (2011–1–056), and the National Science Foundation of China (81172492).