circ-EIF6 encodes EIF6-224aa to promote TNBC progression via stabilizing MYH9 and activating the Wnt/beta-catenin pathway

Abstract

The protein-coding ability of circular RNAs (circRNAs) has recently been a hot topic, but the expression and roles of protein-coding circRNAs in triple-negative breast cancer (TNBC) remain uncertain. By intersecting circRNA sequencing data from clinical samples and cell lines, we identified a circRNA, termed circ-EIF6, which predicted a poorer prognosis and correlated with clinicopathological characteristics in a cohort of TNBC patients. Functionally, we showed that circ-EIF6 promoted the proliferation and metastasis of TNBC cells in vitro and in vivo. Mechanistically, we found that circ-EIF6 contains a 675-nucleotide (nt) open reading frame (ORF) and that the −150-bp sequence from ATG functioned as an internal ribosome entry site (IRES), which is required for translation initiation in 5′ cap-independent coding RNAs. circ-EIF6 encodes a novel peptide, termed EIF6-224 amino acid (aa), which is responsible for the oncogenic effects of circ-EIF6. The endogenous expression of EIF6-224aa was further examined in TNBC cells and tissues by specific antibody. Moreover, EIF6-224aa directly interacted with MYH9, an oncogene in breast cancer, and decreased MYH9 degradation by inhibiting the ubiquitin-proteasome pathway and subsequently activating the Wnt/beta-catenin pathway. Our study provided novel insights into the roles of protein-coding circRNAs and supported circ-EIF6/EIF6-224aa as a novel promising prognostic and therapeutic target for tailored therapy in TNBC patients.

Graphical abstract

The article by Yaming Li et al. reported circ-EIF6 promoted the proliferation and metastasis of TNBC cells by encoding a novel peptide termed EIF6-224aa. They found EIF6-224aa could interact with MYH9 protein and inhibit its degradation, further leading to the activation of the Wnt/beta-catenin pathway.

Introduction

Breast cancer (BC) is a serious disease that threatens women’s health, and many women die from BC every year worldwide. In the United States, approximately 268,600 women were diagnosed, and 41,760 women died of BC in 2018.¹ BC is a heterogeneous disease, and approximately 15%–20% of patients with BC are diagnosed with the triple-negative subtype (triple-negative BC [TNBC]),² which is defined as a subtype that lacks estrogen receptor (ER), progesterone receptor (PR), and human epidermal growth factor receptor type 2 (Her-2) gene expression.³ We have previously reported a relatively poor prognosis for patients with TNBC due to rapid proliferation, early metastasis, a high recurrence rate, and a lack of molecular targets for treatment.⁴ The high speed of proliferation and early metastasis are the key factors that contribute to the high mortality of TNBC, and an understanding of the molecular mechanisms associated with TNBC tumorigenesis is vital.

With the rapid development of next-generation sequencing technologies, an increasing number of previously unknown transcripts have been identified. Noncoding RNA (ncRNA) is one of the above-mentioned transcripts that is not translated into proteins due to the lack of an open reading frame (ORF) and has been proven to play crucial roles in regulating the tumorigenesis and malignant behaviors of tumors.⁵ Circular RNA (circRNA) is characterized by the nonsequential backsplicing of pre-mRNA transcripts, resulting in the formation of covalently closed continuous loops, and this structure was first misinterpreted as a product of splicing errors.6, 7, 8 Endogenous circRNAs are widely expressed throughout the eukaryotic transcriptome and are generally considered ncRNAs with various biological functions, although synthetic circRNAs have been reported to be translatable since 1995.^9,10 Certain circRNAs can function as microRNA (miRNA) sponges through a ceRNA (competing endogenous RNAs) mechanism. For instance, circBMPR2 (bone morphogenetic protein type II receptor) inhibits the progression and tamoxifen resistance of BC while functioning as a miRNA (miR)-553 sponge and then relieves the suppression of USP4 (ubiquitin specific protease 4).¹¹ circAHNAK1 (neuroblast differentiation-associated protein) inhibits the proliferation and metastasis of TNBC by modulating miR-421 and RASA1 (RAS p21 protein activator 1).¹²

However, most circRNAs are generated from coding exons of genes; they are the most studied circRNAs and typically reside in the cytoplasm.¹³ Exonic circRNAs potentially contain specific ORFs or similar ORFs that overlap with related mRNAs, and certain so-called noncoding circRNAs might be translated into proteins.¹⁴ For instance, circFBXW7 (F-box and WD-40 domain protein 7) is translated and encodes a novel protein termed FBXW-185 amino acid (aa), which suppresses glioma tumorigenesis and serves as a potential biomarker.¹⁵ Protein-coding circRNAs have also been reported in colon cancer, liver cancer, BC, and other tumors.16, 17, 18 Protein-coding circRNAs play important roles in the mechanism of tumorigenesis, but researchers have not clearly determined whether protein-coding circRNAs are involved in TNBC, and their functional products remain elusive.

In the present study, we identified an upregulated circRNA, hsa_circ_0060055 (also termed circ-EIF6), in TNBC, which was correlated with the characteristics and prognoses of patients with TNBC. Silencing of circ-EIF6 with small interfering (si)RNAs suppressed the proliferation, migration, and invasion of TNBC cells. Moreover, a 224-aa peptide encoded by circ-EIF6 was identified. Further assays proved that EIF6-224aa was responsible for the oncogenic functions of circ-EIF6 by interacting with the MYH9 protein and activating the Wnt/beta-catenin pathway. In conclusion, our study revealed that the protein-coding circ-EIF6 promoted the progression of TNBC cells by encoding the novel EIF6-224aa peptide.

Results

Identification and characteristics of circ-EIF6 in BC

The expression of circRNAs in 2 pairs of BC tissues with or without distant metastasis was first explored using RNA sequencing to identify functional circRNAs involved in BC progression. As shown in Figure 1A, 99 circRNAs were upregulated, and 17 circRNAs were downregulated in metastatic BC tissues. The expression levels, backspliced reads, and length of detected circRNAs are also presented. RNA sequencing (RNA-seq) of the TNBC cell line MDA-MB-231 and its subcellular line 231_M (MDA-MB-231 with higher metastatic ability) was performed to further screen candidate circRNAs. As shown in Figure 1B, 141 circRNAs were significantly altered in the 231_M subcell line, and the expression, backspliced reads, and length of the detected circRNAs are also presented. By intersecting the differentially expressed circRNAs between the results described above, we identified 6 circRNAs that were both upregulated in metastatic tissues and the 231_M subcell line, together with 3 circRNAs that were downregulated (Figure 1C), and basic information on these circRNAs is displayed in Table S1. We then examined the expression of the common differentially expressed circRNAs in BC cell lines. As shown in Figure 1D, the expression of circ-EIF6 (hsa_circ_0060055) was increased in TNBC cells and was positively correlated with the metastatic ability, suggesting that circ-EIF6 might play a role in the metastasis of TNBC. Then ninety-eight patients with TNBC were selected and equally divided into two groups based on circ-EIF6 expression to further evaluate the potential function of circ-EIF6 in TNBC. High expression of circ-EIF6 predicted a poorer prognosis for overall survival (Figure 1E). The association between the clinicopathological characteristics of the 98 patients with TNBC and circ-EIF6 expression is shown in Table 1, and the results suggested that circ-EIF6 expression correlated with the histological grade and distant metastasis of patients. Moreover, univariate and multivariate analyses were performed to evaluate prognostic factors for patients with TNBC and indicated that circ-EIF6 and the distant metastasis status were independent prognostic factors for overall survival (Table 2). Based on the aforementioned results, circ-EIF6 was chosen as the potential functional circRNA in TNBC cells.

Figure 1.

Identification and characterization of circular (circ)-EIF6 as a novel potential oncogene in breast cancer

(A) Heatmap of the significantly differentially expressed circRNAs between 2 metastatic and 2 nonmetastatic breast tumors (red represents upregulated circRNAs, and blue represents downregulated circRNAs). The backspliced reads and length are also presented. (B) Heatmap of the significantly differentially expressed circRNAs between the 231 and 231_M cell lines (red represents upregulated circRNAs, and blue represents downregulated circRNAs). The backspliced reads and length are also presented. (C) Screening strategy for candidate circRNAs. (D) The relative expression of circ-EIF6 was measured in different breast cancer cell lines using quantitative real-time polymerase chain reaction (PCR) (n = 3). (E) Tissues were collected from 98 patients with TNBC, and a Kaplan-Meier analysis was performed to evaluate the association between circ-EIF6 expression and the prognosis of patients with TNBC (n = 49 patients in each group). (F) Upper panel: the schematic diagram indicates the genomic loci of circ-EIF6. Lower panel: divergent and convergent primers for circ-EIF6 used in this study and Sanger sequencing conducted following PCR using the indicated divergent flanking primers confirmed the “head-to-tail” splicing of circ-EIF6 in 293T cells. (G) Total RNA extracted from TNBC cells with or without RNase R was subjected to PCR with divergent or convergent primers. Relative expression was detected using real-time PCR (n = 3). (H) RT-PCR assay with divergent or convergent primers indicated the existence of circ-EIF6 in cDNA, but not in gDNA. Actin was used as a negative control (n = 3). (I) Quantitative real-time PCR indicated the abundance of circ-EIF6 and the EIF6 mRNA in TNBC cells after treatment with actinomycin D at the indicated time points (n = 3). (J) The relative expression levels of circ-EIF6 and EIF6 in TNBC cells were analyzed using qRT-PCR after normalization to random primers and oligo dT primers (n = 3). ns, nonsignificant; ∗∗∗p < 0.001.

Table 1.

Association between clinicopathological variables and circ-EIF6 expression in patients with TNBC

Variable	Cases (n = 98)	circ-EIF6 expression		p value
		Low	High
Age
≤50	48	23	25	0.686
>50	50	26	24
Histologic grade
I + II	48	30	18	0.015
III	50	19	31
Tumor size
≤2 cm	36	21	15	0.209
>2 cm	62	28	34
N status
Negative	61	27	34	0.145
Positive	37	22	15
Distant metastasis
No + unknown	84	46	38	0.021
Yes	14	3	11

Table 2.

Univariate and multivariate analyses of prognostic factors for patients with TNBC

Variable	Univariate analysis		Multivariate analysis
	HR (95% CI)	p	HR (95% CI)	p
Age
≤50	Reference	−
>50	0.5 (0.125−2.002)	0.328
Histologic grade
I + II	Reference	−
III	1.34 (0.359−4.997)	0.663
Tumor size
≤2 cm	Reference	−
>2 cm	46.562 (0.185−11,702.05)	0.173
N status
Negative	Reference	−
Positive	3.509 (0.877−14.037)	0.076
Distant metastasis
No + unknown	Reference	−	reference	−
Yes	4.925 (1.32−18.381)	0.018	4.79 (1.285−17.585)	0.02
circ-EIF6 expression
Low	Reference	−	reference	−
High	8.828 (1.103−70.646)	0.04	8.661 (1.082−69.341)	0.042

HR, hazard ratio; CI, confidence interval.

As shown in the upper panel of Figure 1F, circ-EIF6 (hsa_circ_0060055) arose from exons 3−7 of the EIF6 gene on chr20:33866724-33872064 (20q11.22) and ultimately formed the mature sequence with a length of 906 nt¹⁹ (Figure 1F, lower panel). We designed specific convergent (blue) and divergent (red) primers for circ-EIF6, and PCR assays were used to verify the head-to-tail splicing of circ-EIF6 by Sanger sequencing of the circ-EIF6 junction sequence (Figure 1F, lower panel). The resistance of circ-EIF6 to digestion by the RNase R treatment (RNase R) exonuclease was first examined to confirm the circular form of circ-EIF6, and circ-EIF6 was resistant to digestion by RNase R (Figure 1G). Furthermore, cDNA and genomic DNA (gDNA) templates of both TNBC cell lines were used for PCR, and the convergent primers amplified both circ-EIF6 and actin, but the divergent primers amplified only circ-EIF6 using cDNA as templates, and no amplification was observed using gDNA templates (Figure 1H). In addition, an actinomycin D assay was used to investigate the stability of circ-EIF6. The half-life of this circRNA transcript exceeded 24 h, whereas that of the linear form was less than 8 h, indicating that circ-EIF6 was much more stable than the linear form EIF6 (Figure 1I). Moreover, we used random primers or oligo dT primers to perform quantitative reverse transcriptase (qRT)-PCR and further confirm the circular characteristics of circ-EIF6. Compared with random primers, the relative expression of circ-EIF6 was markedly decreased when using the oligo dT primers, whereas the expression of the EIF6 mRNA was not altered (Figure 1J), indicating that circ-EIF6 had no poly(A) tail.

Inhibition of circ-EIF6 suppressed the proliferation, migration, and invasion of TNBC cells

We first knocked down circ-EIF6 expression using two specific siRNAs to evaluate its functions in TNBC, and the position, sequence, and efficiency of the siRNAs are shown in Figure 2A. The specific fluorescence in situ hybridization (FISH) probe for the circ-EIF6 junction site further confirmed the specificity and efficiency of both siRNAs (Figure S1A). By further examining the mRNA expression levels of its host gene EIF6, we found that knockdown of circ-EIF6 did not alter EIF6 mRNA expression, indicating that the results from subsequent experiments did not result from nonspecific knockdown of the EIF6 mRNA (Figure S1B). MTT (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide) and flat plate colony formation assays were then performed to evaluate proliferation, and knockdown of circ-EIF6 suppressed the cell proliferation rate (Figures 2B and 2C). Further EdU (5-ethynyl-2'-deoxyuridine) and cell cycle assays also indicated that the proliferative activities of cells were inhibited after circ-EIF6 knockdown by inducing cell cycle arrest (Figures 2D and 2E). We subsequently examined the effects of circ-EIF6 knockdown on TNBC cell migration and invasion, and Transwell assays showed significantly reduced numbers of migrated or invaded cells after circ-EIF6 knockdown (Figures 2F and 2G). A wound-healing assay also revealed that the migration capacities were suppressed after circ-EIF6 knockdown (Figure 2H). Based on the results described above, the levels of proteins related to the cell cycle, migration, and invasion were further verified, and knockdown of circ-EIF6 contributed to cell cycle arrest and the inhibition of migration and invasion (Figure 2I). Taken together, our results proved that knockdown of circ-EIF6 suppressed the proliferation, migration, and invasion of TNBC cells in vitro.

Figure 2.

Knockdown of circ-EIF6 suppressed the proliferation, migration, and invasion of TNBC cells

(A) Schematic illustration showing the target sequences of the siRNAs specific to the backsplicing junction of circ-EIF6. The interference efficacies of circ-EIF6-targeting siRNAs were measured using qRT-PCR (n = 3). MTT (n = 6) (B) and flat plate colony formation assays (n = 3) (C) of MDA-MB-231 and MDA-MB-468 cells transfected with control sequences or siRNAs were performed to evaluate the breast cancer cell proliferation rate. EdU (n = 4) (D) and flow cytometry assays (n = 3) (E) were performed to evaluate the effects of circ-EIF6 knockdown on the proliferation of breast cancer cells. Scale bars, 100 μm. The migration (F) and invasion (G) abilities of MDA-MB-231 and MDA-MB-468 breast cancer cells with circ-EIF6 knockdown were evaluated using Transwell assays (n = 5). Scale bars, 200 μm. (H) A wound-healing assay was performed to examine the metastatic abilities of cells after circ-EIF6 knockdown (n = 3). Scale bars, 200 μm. (I) Important proteins involved in the cell cycle, migration, and invasion were examined after circ-EIF6 knockdown (n = 3). ∗p < 0.05, ∗∗p < 0.01, and ∗∗∗p < 0.001.

circ-EIF6 encoded a novel 224-aa peptide termed EIF6-224aa

Based on the results described above, we proved that circ-EIF6 knockdown suppressed the proliferation, migration, and invasion of TNBC cells. We further explored whether circ-EIF6 functioned by encoding a peptide, as emerging evidence indicates that some circRNAs have a protein-coding capacity.²⁰ We first examined the subcellular location of circ-EIF6. As shown in Figures 3A and 3B, circ-EIF6 was mainly located in the cytoplasm, which indicated the potential molecular mechanisms of circ-EIF6. We then performed a sucrose density gradient centrifugation-based polysome analysis to evaluate the coding ability of circ-EIF6, 293T cells were transfected with circ-EIF6 or circ-EIF6 ATG mutant vectors (circ-EIF6-ATG-mut), and cell extracts were subjected to 5%–50% sucrose gradient centrifugation. Absorbance was then measured at 254 nm, and fractions were divided into nonribosome (N), monosome (M), light polysome (L), and heavy polysome (H) fractions in both cells (Figure 3C, upper panel; left, circ-EIF6 overexpression vector [circ-EIF6-OV]; right, circ-EIF6-ATG-mut). Then, qRT-PCR was performed to detect the distribution of circ-EIF6 or circ-EIF6-ATG-mut among fractions. Notably, circ-EIF6 was mainly detected in the M and L fractions instead of the H fractions, which was consistent with a previous report.²¹ In contrast, overexpression of circ-EIF6-ATG-mut showed a significant reduction in the ribosome distribution, which showed a shift from M and L to N fraction, indicating that circ-EIF6 but not circ-EIF6-ATG-mut was translated (Figure 3C, lower left panel). As a control, overexpression of circ-EIF6 or circ-EIF6-ATG-mut in 293T cells showed no significant change on EIF6 mRNA distribution (Figure 3C, lower right panel). By further analyzing the sequence of circ-EIF6, we found that circ-EIF6 contains a 675-nt ORF (green), which might encode a 224-aa peptide (EIF6-224aa). The sequences of mature circ-EIF6 and EIF6-224aa are shown in Figure S1C. The bioinformatics analysis also identified a putative internal ribosome entry site (IRES) sequence (−150 bp from the ORF, red), which is required for translation initiation in 5′ cap-independent coding RNAs.^22,23 Full-length or truncated IRES together with CMV (CMV promoter) sequences were then cloned into the pGL3-Basic vector, and the activity of the predicted IRES was evaluated using a dual-luciferase assay (Figure 3D, upper panel), with the pRL-TK vector expressing the RLuc (Renilla luciferase ) reporter gene serving as a negative control. As shown in the lower panel of Figure 3D, the putative IRES sequence had a strong ability to initiate protein translation, whereas the truncated vectors showed no significant effects. Moreover, RFP (red fluorecent protein) and GFP were cloned into a dual-cistron reporter construct with the putative or truncated IRES sequence between them (Figure 3E, upper left panel). Under normal conditions, both RFP and GFP were detected in cells transfected with the putative IRES vector, whereas the putative IRES vector only induced the expression of GFP when eIF4E (eukaryotic translation initiation factor 4E) was inhibited (Figure 3E, lower left panel). Furthermore, the truncated IRES sequence induced lower GFP expression than the full-length IRES sequence, suggesting that the natural IRES in circ-EIF6 induced ribosome entry and initiated translation (Figure 3E, right panel).

Figure 3.

circ-EIF6 encoded a novel 224-aa peptide

(A) Fluorescence in situ hybridization (FISH) with junction-specific probes was used to determine the localization of circ-EIF6 in TNBC cells (n = 3). Scale bars, 40 μm. (B) Subcellular fractionation to determine nuclear and cytoplasmic circ-EIF6 expression levels in TNBC cells. U6 was used as a nuclear marker, and GAPDH (glyceraldehyde-3-phosphate dehydrogenase) and β-actin were used as cytosolic markers (n = 3). (C) Polysome fractions (N, nonribosome; M, monosome; L, light polysome; and H, heavy polysome) were extracted from circ-EIF6- or circ-EIF6-ATG-mut-overexpressing 293T cells by performing 5% to 50% sucrose gradient ultracentrifugation. qRT-PCR, using junction-specific primers, was then performed to analyze the translation potential of circ-EIF6. qRT-PCR, using linear-specific primers to determine linear EIF6 translation potential, served as a positive control (n = 3). (D) Detection of the putative IRES activity of circ-EIF6. Upper panel: IRES sequences and different truncation mutants were cloned into the pGL3-Basic vector, with the pRL-TK vector serving as a negative control. Lower panel: the relative luciferase activity of Luc/RLuc was tested in 293T cells (n = 3). (E) Upper left panel: full-length or truncated circ-EIF6 IRES sequences were cloned between RFP and GFP as indicated to construct reporter plasmids. Lower left panel: plasmids were transfected into 293T cells with or without eIF4E inhibitor treatment. RFP and GFP signals were detected. Right panel: the empty vector or full-length or truncated IRES vector was transfected into 293T cells, and RFP and GFP signals were detected (n = 3). Scale bars, 100 μm. (F) Upper panel: circ-EIF6-FLAG was designed to detect the circ-EIF6-encoded peptide, and the FLAG tag sequence was divided and cloned on both sides of the circRNA sequence; the circular junction was moved to the inside of the FLAG sequence. Circularization of this vector produced the same circRNA as endogenous circ-EIF6, except for the addition of a FLAG tag after the ORF (circ-EIF6-FLAG). A similar circ-EIF6 expression vector lacking the IRES was also constructed (circ-EIF6-FLAG-IRES-Del). Lower panel: a FLAG-tagged antibody was used to detect the EIF6-224aa-FLAG protein (n = 3). (G) Total protein was extracted from 293T cells transfected with pLCDH or circ-EIF6-FLAG. LC-MS analysis was performed to identify specific peptide sequences of EIF6-224aa. A specific antibody against EIF6-224aa, which targets the PDAGREVAESSLGLR sequence, was synthesized, and the expression of EIF6-224aa was detected after overexpression (H) or knockdown (I) (n = 3). (J). Total protein was extracted from 4 metastatic and 4 nonmetastatic TNBC tissues, and the expression levels of EIF6-224aa were detected using western blotting. ∗∗∗p < 0.001.

A novel circ-EIF6 vector containing the FLAG sequence was constructed to further assess the protein-coding ability of circ-EIF6. As shown in the upper panel of Figure 3F, the junction of endogenous circ-EIF6 is located inside the ORF, and we moved the junction to the stop codon of the ORF and shifted the FLAG sequence to both sides (circ-EIF6-FLAG). A similar FLAG-labeled vector lacking the IRES sequence was also constructed (circ-EIF6-FLAG-IRES-Del). These vectors and their empty vectors were transfected into cells, and the expression of EIF6-224aa-FLAG was detected, which further proved the important roles of the IRES sequence in the translation of circ-EIF6 (Figure 3F, lower panel). We next used liquid chromatography-tandem mass spectrometry (LC-MS) to characterize the amino acid sequences of EIF6-224aa in circ-EIF6-FLAG-transfected 293T cells, and the specific protein fragments from EIF6-224aa were successfully identified (EVAESSLGLR), indicating that circ-EIF6 was translated into EIF6-224aa in cells (Figure 3G). Moreover, primary MDA-MB-468 and MDA-MB-231 cells were also sent for MS analysis, and the specific peptide sequence of EIF6-224aa was detected in both cell lines, which further verified the translation of endogenous circ-EIF6 (Figure S1D). A 15-aa unique sequence (PDAGREVAESSLGLR) of EIF6-224aa was then selected as the antigen and sent to Abcepta (San Diego, CA, USA) to synthesize a specific antibody against EIF6-224aa to determine the presence of endogenous EIF6-224aa in TNBC cells and tissues. The antibody was designed to recognize EIF6-224aa only, not the linear EIF6 protein, and detailed information about this antibody is supplied in Data S1. Vectors expressing circ-EIF6-ATG-mut and a linearized EIF6-224aa ORF plus FLAG tag (EIF6-224aa-FLAG) were then constructed and transfected into TNBC cells, and the expression of the EIF6-224aa peptide was detected after overexpression (Figure 3H) and knockdown (Figure 3I) using an EIF6-224aa antibody, which further confirmed the translation of circ-EIF6 and specificity of this antibody. Moreover, 4 samples of nonmetastatic TNBC tissues and 4 samples of metastatic TNBC tissues were selected, and the expression of the EIF6-224aa peptide together with the circ-EIF6 RNA was detected (Figures 3E and 3J). Taken together, our results proved that circ-EIF6 was translated into EIF6-224aa in an IRES-dependent manner and that EIF6-224aa was endogenously expressed in TNBC cell lines and tissues.

EIF6-224aa, but not circ-EIF6, promoted the proliferation and metastasis of TNBC cells

As we have proven that circ-EIF6 was translated into a novel peptide termed EIF6-224aa, we further explored whether circ-EIF6 or EIF6-224aa contributed to its functions. TNBC cells were stably transfected with the circ-EIF6-OV, circ-EIF6-ATG-mut, and its empty vector (pLCDH), together with EIF6-224aa-FLAG and pcDNA3.1 vectors (Figure 4A). The efficiency of circ-EIF6 overexpression and its effects on EIF6 mRNA expression were further verified by performing FISH and qRT-PCR (Figures S1E and S1F). As shown in Figure 4B, overexpression of putative circ-EIF6 promoted the proliferation of TNBC cells, whereas mutation of the ATG sequence reversed this effect; moreover, overexpression of EIF6-224aa-FLAG further increased proliferation. The flat plate colony formation assay produced similar results (Figure 4C). Similar results were also obtained in Transwell assays of the migration and invasion of TNBC cell lines (Figures 4D and 4E). The results of the EdU, cell cycle, and wound-healing assays examining the effects of circ-EIF6 and EIF6-224aa are presented in Figure S2.

Figure 4.

EIF6-224aa, but not circ-EIF6, promoted the proliferation and metastasis of breast cancer cells

(A) RNA expression levels were detected using specific primers in MDA-MB-468 and MDA-MB-231 cells stably transfected with pLCDH, circ-EIF6-OV, circ-EIF6-ATG-mut, pcDNA3.1, or EIF6-224aa-FLAG (n = 3). (B) The growth curve was determined by performing the MTT assay in breast cancer cells expressing the vectors listed above (n = 6). (C) The vectors listed above were transfected into both MDA-MB-468 and MDA-MB-231 cells, and the proliferation rates were evaluated using flat plate colony formation assay (n = 3). The migration (D) and invasion (E) abilities of TNBC cells transfected with the aforementioned vectors were further examined using Transwell assays (n = 5). (1) pLCDH, (2) circ-EIF6-OV, (3) circ-EIF6-ATG-mut, (4) pcDNA3.1, (5) EIF6-224aa-FLAG. Scale bars, 200 μm. (F) Images of xenograft tumors obtained from BALB/c nude mice at the endpoint (n = 5 mice in each group). (G) Growth curve of the tumor volume. The volumes were measured every 5 days after the initial 10 days (n = 5 mice in each group). (H) Weights of xenograft tumors at the endpoint (n = 5 mice in each group). (I) Representative images of lung metastatic nodules from BALB/c nude mice at the endpoint (n = 5 mice in each group). (J) H&E staining of lungs isolated from mice injected with circ-EIF6- or control vector-transfected cells (n = 5 mice in each group). (K) IHC (immunohistochemistry) staining for Ki67 and N-cadherin in the xenograft tumors (n = 5 mice in each group). ∗p < 0.05, ∗∗p < 0.01, and ∗∗∗p < 0.001.

MDA-MB-231 cells were transfected with pLCDH and circ-EIF6-OV due to their highly tumorigenic and metastatic properties to further validate the role of circ-EIF6. The in vivo proliferation and metastasis of both pLCDH and circ-EIF6-OV cell lines were validated. As shown in Figures 4F−4H, both pLCDH and circ-EIF6-OV cell lines were seeded subcutaneously, and the tumor volume was detected at 10, 15, 20, 25, and 30 days after tumor injection. Tumors in the pLCDH group were significantly smaller than those in the circ-EIF6-OV group. According to previous studies, lung metastasis is an indicator of tumor aggressiveness. As shown in Figure 4I, the metastasis of the circ-EIF6-OV group was significantly increased compared with the pLCDH group, as confirmed by hematoxylin and eosin (H&E) staining (Figure 4J). In addition, IHC staining for the proliferation indicator Ki67 and metastasis indicator N-cadherin (N-cad) was also performed in the xenograft tumors, and the staining of Ki67 and N-cadherin was significantly increased after circ-EIF6 overexpression (Figure 4K).

EIF6-224aa directly interacted with MYH9 and protected it from degradation

We performed coimmunoprecipitation in EIF6-224aa-FLAG-overexpressing 293T cells to identify its potential target molecules and further explore the molecular mechanisms underlying the tumor-promoting functions of EIF6-224aa. As shown in the left panel of Figure 5A, 293T cells transfected with EIF6-224aa-FLAG and corresponding control cells were subjected to immunoprecipitation using an anti-FLAG antibody. The precipitates were subjected to LC-MS to identify potential EIF6-224aa-interacting proteins. Among various candidates, EIF6-224aa potentially bound to the MYH9 protein (Figure 5A, right panel). Furthermore, immunofluorescence staining also showed that EIF6-224aa and MYH9 were colocalized in the cytoplasm (Figure 5B), suggesting EIF6-224aa might interact with MYH9. EIF6-224aa conformations were modeled with PEP-FOLD and docked to MYH9 based on the ATTRACT2 force field through the PEP-SiteFinder pipeline to investigate the potential direct interaction between EIF6-224aa and MYH9.²⁴ We split EIF6-224aa into seven sections: 1–36 aa, 34–65 aa, 63–98 aa, 96–131 aa, 129–164 aa, 162–197 aa, and 190–224 aa. The top ten proteins complexing with MYH9 were visualized in different colors using PyMOL (the PyMOL Molecular Graphics System, version 1.8; Schrödinger), whereas protein–peptide interface residues were labeled (Figure 5C). Immunoprecipitation further confirmed a mutual EIF6-224aa/MYH9 interaction (Figure 5D). The MYH9 protein was separated into 4 fragments with hemagglutinin (HA) tags to further explore the interaction between EIF6-224aa and MYH9 (Figure 5E). The MYH9 truncated vectors and EIF6-224aa-FLAG vector were cotransfected into 293T cells, and immunoprecipitation assays were performed. EIF6-224aa mainly interacted with the first and fourth fragments of MYH9 (Figure 5F). Similar results were obtained from both MDA-MB-468 and MDA-MB-231 cell lines (Figure S2D).

Figure 5.

EIF6-224aa directly interacts with the MYH9 protein and inhibits its degradation

(A) Left panel: immunoprecipitation was performed using an immunoglobulin G (IgG) or anti-FLAG antibody in EIF6-224aa-FLAG-transfected 293T cells. The precipitates were then subjected to LC-MS to identify potential EIF6-224aa-interacting proteins. Right panel: the MYH9 protein was identified in EIF6-224aa precipitates. (B) Immunofluorescence (IF) was performed in breast cancer cells to determine EIF6-224aa/MYH9 colocalization using anti-FLAG and anti-MYH9 antibodies (n = 3). Scale bars, 20 μm. (C) EIF6-224aa conformations were modeled with PEP-FOLD and docked to MYH9 based on the ATTRACT2 force field using the PEP-SiteFinder pipeline. EIF6-224aa was split into multiple segments: 1–36 aa, 34–65 aa, 63–98 aa, 96–131 aa, 129–164 aa, 162–197 aa, and 190–224 aa. The top ten peptides forming a complex with MYH9 were visualized in different colors using PyMOL (the PyMOL Molecular Graphics System, version 1.8; Schrödinger). (D) EIF6-224aa-FLAG was transfected into cells, and the interaction between EIF6-224aa and MYH9 was evaluated using immunoprecipitation in breast cancer cells (n = 3). (E) Schematic illustration showing that the MYH9 protein was divided into four fragments and labeled with HA tags. (F) The direct mutual interactions of EIF6-224aa with different domains of HA-tagged MYH9 were tested using immunoprecipitation (n = 3). (G) TNBC cells were treated with 20 μg/mL CHX for 0, 4, 8, or 24 h, and the effect of EIF6-224aa on stabilizing the MYH9 protein was examined (n = 3). (H) MG132, a proteasome inhibitor, was added to TNBC cells at a concentration of 10 μM to verify the effects of EIF6-224aa on the proteasomal degradation of the MYH9 protein (n = 3). (I) HA-Ub, MYH9-His, and pcDNA3.1 or linear-EIF6-FLAG were cotransfected into 293T cells, and the effects of EIF6-224aa on the ubiquitination level of the MYH9 protein were verified in the presence of 10 μM MG132 (n = 3).

Cycloheximide (CHX), a protein synthesis inhibitor, was added to TNBC cells transfected with the EIF6-224aa-FLAG vector or pcDNA3.1, and the expression of the MYH9 protein was detected at 0, 4, 8, and 12 h to further examine the effects of EIF6-224aa on the MYH9 protein. As shown in Figure 5G, the overexpression of EIF6-224aa reduced the degradation speed of MYH9, indicating that the endogenous EIF6-224aa expressed from circ-EIF6 protected MYH9 from degradation. Moreover, MG132, a proteasome inhibitor, was added to TNBC cells transfected with the EIF6-224aa-FLAG vector or pcDNA3.1, and the expression of the MYH9 protein was detected after 6 h of treatment. As shown in Figure 5H, EIF6-224aa overexpression protected the MYH9 protein from proteasomal degradation. Based on the aforementioned results, the ubiquitination of MYH9 protein was further examined, and overexpression of EIF6-224aa decreased the ubiquitination level of the MYH9 protein, which protected MYH9 from proteasomal degradation (Figure 5I).

EIF6-224aa promoted TNBC progression by activating the MYH9/Wnt/beta-catenin pathway

MYH9 expression was knocked down with an siRNA, and the efficiency in both MDA-MB-468 and MDA-MB-231 was verified using qRT-PCR and western blotting to evaluate the effects of MYH9 on the oncogenic functions of circ-EIF6/EIF6-224aa (Figure 6A). Then, both TNBC cell lines were transfected with pcDNA3.1, EIF6-224aa-FLAG, or EIF6-224aa-FLAG + siMYH9, and the proliferation (Figure 6B), migration (Figure 6C), and invasion (Figure 6D) abilities were verified. As shown in Figures 6B−6D, MYH9 knockdown antagonized the oncogenic effects of EIF6-224aa, and the proliferation, migration, and invasion abilities of cells were significantly inhibited, indicating that MYH9 was a functional target of EIF6-224aa.

Figure 6.

Stabilization of the MYH9 protein by EIF6-224aa activated the Wnt/beta-catenin pathway and accounted for the oncogenic roles of circ-EIF6

(A) MYH9 expression was knocked down, and the efficiency was verified using qRT-PCR and western blotting (n = 3). (B) MYH9 expression was knocked down after EIF6-224aa overexpression, and the proliferation rate of cells was measured using the MTT assay (n = 6). (C) The migration and (D) invasion abilities of TNBC cells with EIF6-224aa overexpression and/or MYH9 knockdown were examined (n = 5). Scale bars, 200 μm. (E) The effect of EIF6-224aa overexpression and/or MYH9 knockdown on the activation of the Wnt/beta-catenin pathway was verified using western blotting (n = 3). (F) Expression of representative downstream genes in the Wnt/beta-catenin pathway was verified by qRT-PCR after EIF6-224aa overexpression and/or MYH9 knockdown (n = 3). (G) Schematic diagram showing the mechanism by which circ-EIF6 modulates TNBC proliferation and metastasis. ∗p < 0.05, ∗∗p < 0.01, and ∗∗∗p < 0.001.

Previous studies have proved that MYH9 expression correlates with the activation of the Wnt/beta-catenin pathway,²⁵ and we verified effects of EIF6-224aa overexpression and MYH9 knockdown on the Wnt/beta-catenin pathway and downstream target genes in our study. As shown in Figure 6E, the expression of the beta-catenin protein was upregulated after EIF6-224aa overexpression, whereas it was downregulated when MYH9 was knocked down with an siRNA. The expression of downstream target proteins (cyclin D1, c-Myc, N-cadherin, E-cadherin, and vimentin) changed accordingly. Moreover, we also detected the changes in the mRNA expression of downstream target genes (CCND1, AXIN2, SURVIVIN, LEF1, NKD1, and TCF1), and we proved that the target genes in the Wnt/beta-catenin pathway were upregulated by EIF6-224aa, whereas knockdown of MYH9 antagonized the effects of EIF6-224aa (Figure 6F).

In conclusion, our study proved that circ-EIF6 (hsa_circ_0060055) interacted with the ribosome and was translated into a novel peptide termed EIF6-224aa. EIF6-224aa directly interacted with the MYH9 protein and inhibited its proteasomal degradation by decreasing the ubiquitination of MYH9, subsequently activating the Wnt/beta-catenin pathway to promote the oncogenic functions of circ-EIF6 (Figure 6G).

Discussion

In recent decades, numerous investigations have documented abnormal expression profiles of ncRNAs in a wide spectrum of cancer types, including miRNAs, longnoncoding RNAs (lncRNAs), and circRNAs. Recent studies have revealed vital roles for many miRNAs and lncRNAs in modulating tumor growth and cancer metastasis.^26,27 circRNAs are a widespread RNA species in the human transcriptome.²⁸ In addition to functioning as regulators of gene expression or development by adsorbing miRNAs, circRNAs have also been shown to be critical for human malignancies.^29,30 The most well-known mechanism of circRNAs is their functions as miRNA sponges, but only a few circRNAs contain perfect multiple miRNA trapping sites, raising the question of whether circRNAs exhibit functions beyond serving as miRNA sponges.^31,32 Recent evidence has confirmed the existence of functional peptides translated from sORFs (small open reading frames) in ncRNAs, including pri-miRNAs (hairpin-containing primary miRNA transcripts), lncRNAs, and circRNAs,^33,34 suggesting that the coding potential of these ncRNAs has been largely underestimated. Protein-coding circRNAs have recently been a hot topic and might express proteins in a more stable manner due to their specific circular character. A few articles have proven the crucial roles of these circRNAs and their protein products in cancers, including glioblastoma, liver cancer, and colon cancer.^16,17,35 For example, circFBXW7 suppresses glioma tumorigenesis by encoding FBXW7-185 aa, which reduces the half-life of c-Myc by antagonizing USP28 (ubiquitin specific peptidase 28)-induced c-Myc stabilization.¹⁵ However, the important roles of protein-coding circRNAs in TNBC have not been well studied.

We first filtered several circRNAs that were deregulated in metastatic BC tissues and cell lines to explore potential functional circRNAs, and 9 circRNAs were proven to be significantly differentially expressed. Next, circ-EIF6 was selected for further analyses because its expression correlated with the prognoses and clinicopathological characteristics of patients with TNBC. circ-EIF6 originated at chromosome 20q11 and correlated with BC metastasis and TNBC tumorigenesis,^36,37 and its host gene EIF6 was reported to be uniquely amplified in patients with highly proliferative breast tumors.³⁸ A further evaluation of the protein-coding ability of circ-EIF6 recently reported that artificial circRNAs are translated in a cap-independent manner using a cis-regulatory element termed IRES. The sequence of circ-EIF6 was assessed using circRNADB³⁹ and IRESfinder²³ to determine whether it contains an IRES structure, and the −150-bp sequence from the initiation codon was identified, which indicated the protein-coding ability. Further experiments proved the translation initiation ability of IRES, and circ-EIF6 encodes a novel 224-aa peptide.

In vitro and in vivo assays were performed to explore the functions of circ-EIF6 in regulating the malignant behaviors of TNBC cells. We first knocked down circ-EIF6 expression in both MDA-MB-468 and MDA-MB-231 cell lines and found that the proliferation, migration, and invasion abilities of cells were inhibited (Figure 2). Since we have proven that circ-EIF6 encodes EIF6-224aa, we further evaluated whether circ-EIF6 or EIF6-224aa was responsible for the oncogenic functions. Because IRES is a key promotor of circRNA translation, a circ-EIF6-OV lacking the IRES was constructed and compared with a putative circ-EIF6-OV to determine the roles of EIF6-224aa. As shown in Figure 4, EIF6-224aa, not circ-EIF6, promoted the proliferation and metastasis of TNBC cells in vitro and in vivo.

By further exploring the mechanism of EIF6-224aa, we confirmed that EIF6-224aa interacted with the MYH9 protein. MYH9 is a member of the myosin II subfamily^40,41 that plays important roles in cell adhesion, cytokinesis, and the maintenance of cell morphology.^42,43 MYH9, which encodes a nonmuscle myosin IIA (NMIIA) protein, has been proven to play crucial roles in cancer cell proliferation, survival, invasion, and metastasis. According to previous reports, MYH9 functions as a tumor suppressor in skin,⁴⁴ head and neck squamous cell,⁴⁵ and tongue squamous cell⁴⁶ cancers. Conversely, the most recent reports show that MYH9 is an oncogene in gastric,⁴⁷ colorectal,⁴⁸ esophageal squamous cell,⁴⁹ non-small cell lung,⁵⁰ cancers, and nasopharyngeal carcinoma.⁵¹ In BC, several studies have also investigated the roles of MYH9 in tumor metastasis. For instance, Derycke et al.⁵² reported that NMIIA was present in invasive BC cells but rarely detected in noninvasive cells, indicating that NMIIA is a decisive protein regulating BC cell invasion. In another study, researchers reported that MYH9 was required for the invasion and lung colonization of MDA-MB-231 cells.⁵³ In addition, Wang et al.⁵⁴ also showed that TFPI-2 (tissue factor pathway inhibitor-2) regulates BC cell proliferation and invasion through ERK1/2 (extracellular-regulated kinase 1/2) signaling via interactions with myosin-9 and actinin-4. In our study, the direct interaction between EIF6-224aa and the MYH9 protein inhibited the degradation of MYH9 protein by the ubiquitin-mediated degradation pathway, suggesting that MYH9 is a promising target of EIF6-224aa. Recent studies have proven that MYH9 interacts with GSK3β (glycogen synthase kinase 3 beta) and reduces its protein expression through ubiquitin-mediated degradation, which subsequently activates the Wnt/beta-catenin pathway and induces downstream oncogenic tumor phenotypes.^25,55,56 In the present study, the interaction between EIF6-224aa and MYH9 activated the Wnt/beta-catenin pathway to further promote the proliferation and metastasis of TNBC cells (Figure 6).

In conclusion, circ-EIF6 significantly enhanced the proliferation and metastasis of TNBC cells by encoding the novel peptide EIF6-224aa and further activating the MYH9/Wnt/beta-catenin pathway (Figure 6G). In addition, circ-EIF6 represents a potential prognostic biomarker in patients with TNBC. The newly identified encoded circ-EIF6 broadens our insights into the underlying mechanisms of TNBC and represents a potential prognostic and therapeutic target for the treatment of patients with TNBC.

Materials and methods

Ethics statement and human tissue samples

All experimental procedures were approved by the Ethical Committee of Shandong University. Tissues were obtained at the time of surgery from patients admitted to Qilu Hospital and immediately stored at −80°C. All patients provided written, informed consent for the use of these clinical materials in research.

circRNA sequencing

Two metastatic BC tissues and paired non-metastatic tissues collected at Qilu Hospital of Shandong University in 2017 were used for sequencing analysis. The sample preparation and circRNA sequencing were performed using Cloud-Seq Biotech (Shanghai, China). Significantly differentially expressed circRNAs were retained by screening for fold change R 2.0 and p < 0.05.

Cell lines and vectors

All cell lines were obtained from the American Type Culture Collection (Manassas, VA, USA) and were maintained using standard media and conditions. Briefly, MDA-MB-231 and MDA-MB-468 were cultured with Dulbecco’s modified Eagle’s medium (DMEM; Invitrogen, Carlsbad, CA, USA) containing 10% fetal bovine serum (FBS; HyClone). All cell lines were grown at 37°C in a 5% CO2 cell culture incubator. The primers used for vectors’ construction in this study were shown in Table S2. Vectors, including circ-EIF6-FLAG-IRES-Del, circ-EIF6-ATG-mut, pcDNA3.1-RFP-GFP, MYH9-OV, MYH9-truncated-1, MYH9-truncated-2, MYH9-truncated-3, and MYH9-truncated-4, were synthesized from Vigene Biosciences (Rockville, MD, USA).

Actinomycin D and RNase R

Transcription was prevented by the addition of 2 μg/mL actinomycin D (Sigma-Aldrich, USA) or DMSO (no treatment [Mock]) (Sigma-Aldrich, USA) as the negative control for the indicated times. RNase R was used to identify and confirm the character of the circRNA. The RNAs extracted from MDAMB-231 or MDA-MB-468 cells were divided into two equal parts, respectively, one for RNase R and the other for Mock. Total RNA (2 μg) was incubated with 3 U/μg of RNase R for 30 min at 37°C. After treatment with actinomycin D and RNase R, the RNA expression levels of circ-EIF6 and EIF6 were detected using qRT-PCR. The internal reference (β-actin) in the mock group was used as the calculation standard.

Cell proliferation assay

Cells were seeded into 96-well plates. At the indicated time points, the cells were incubated with 20 μL of sterile MTT for 4 h at 37°C, after which the medium was removed and replaced with 100 μL of DMSO. The absorbance was measured at 570 nm.

EdU incorporation assay

The EdU Proliferation Kit (RiboBio Guangzhou, China) was used to assess cell proliferation viability. Briefly, 48 h after transfection, 1 × 10⁴ cells were seeded in 96-well plate. After incubation with 50 mM EdU for 2.5 h, the cells were fixed with 4% paraformaldehyde (PFA) and stained with Apollo Dye Solution. Hoechst was used to stain the nucleic acid. Images were obtained with an Olympus microscope (Olympus, Tokyo, Japan).

Flat plate colony formation assay

The transfected cells were seeded into six-well plates at a density of 500 cells per well. After 2 weeks, the cells were washed with cold PBS twice, fixed with methanol, and stained with 0.1% crystal violet solution. Pictures were imaged and counted.

Cell cycle assays

Cells were harvested via trypsinization, washed in ice-cold PBS, fixed in ice-cold 75% ethanol in PBS, centrifuged at 4°C, and suspended in PBS. After, 20 mg/mL propidium iodide (Beyotime, Shanghai, China) was added, and the samples were incubated for 20 min at room temperature. The cells were analyzed via flow cytometry.

Migration and invasion assays

Migration and invasion assays were performed as described previously using the Transwell system (Corning Costar, Lowell, MA, USA). In the migration assay, 700 μL of medium with 20% FBS was added to the lower well of each chamber, and 1 × 10⁵ cells suspended in serum-free medium were added to the upper inserts. After incubation for the indicated time, the total number of cells adhering to the lower surface of the membrane was quantified in six representative fields. The invasion assay was performed in the same way as the migration assay except that the membrane was coated with Matrigel (BD Biosciences, Bedford, MA, USA).

Wound-healing assay

Cells were seeded in a 24-well plate and incubated in DMEM containing 10% FBS until a confluent monolayer had formed. Then a sterile 200-μL plastic pipette tip was used to make scratches on the single cell layer, and PBS was used to remove the detached cells. Cells were grown in serum-free DMEM to inhibit cell proliferation, and images were captured at the indicated times (0 and 48 h) using an Olympus light microscope.

RNA FISH

Cells were incubated at 37°C in a solution containing 50% formamide, 2 × saline sodium citrate (SSC), 0.25 mg/mL Escherichia coli transfer RNA, 0.25 mg/mL salmon sperm DNA (Life Technologies, Carlsbad, CA, USA), 2.5 mg/mL BSA (Roche, Indianapolis, IN, USA), and a fluorescently labeled junction probe at 125 nM (Generay, Shanghai, China). After 12 h, the cells were washed and mounted in ProLong Gold (Life Technologies, Carlsbad, CA, USA) and left overnight at room temperature.

Sucrose gradient fractionation assay

Sucrose gradient fractionation assays were performed as described previously.⁵⁷ Briefly, 293T cells were transfected with circ-EIF6 or circ-EIF6-ATG-mut vectors for 48 h in 10 cm plates, treated with 100 μg/mL CHX for 15 min, and washed with cold PBS twice. The cells were then harvested and lysed with 500 μL of lysis buffer (5 mM Tris-HCl, pH 7.5, 2.5 mM MgCl₂, 1.5 mM KCl, 1 × protease inhibitor cocktail [EDTA free], 0.5% Triton X-100, 2 mM DTT, 0.5% sodium deoxycholate, 100 U of RNase inhibitor, and 100 μg/mL CHX) for 15 min. Then, the polysome lysate was centrifuged at 16,000 × g for 7 min, and the supernatant was collected. The 5%–50% sucrose gradient solution was prepared in an ultracentrifuge tube and stored at 4°C overnight to obtain a linearized gradient. The supernatant was added on top of a sucrose gradient and centrifuged at 35,000 rpm for 2 h. After centrifugation, the solution was collected from top to bottom with 150 μL per tube, and the absorbance was determined at 254 nm with a UV spectrophotometer. In addition, RNA was further isolated using TRIzol LS Reagent, and the curve graphs showing the distributions of circ-EIF6 or circ-EIF6-ATG-mut among fractions were drawn as previously reported.⁵⁸

Luciferase reporter assay

Cells were co-transfected with pRL-TK and pGL3-Basic plasmids containing full-length or deleted IRES using Lipofectamine 2000 (Invitrogen). After 48 h, luciferase activities were determined by the Dual-Luciferase Reporter Assay Kit (Promega). Results are presented as ratios of luminescence from firefly to Renilla luciferase.

LC-MS analysis

Proteins were separated on SDS-PAGE gels, and bands were manually excised from the gel and digested with sequencing-grade trypsin (Promega, Madison, WI, USA). The digested peptides were analyzed using a QExactive mass spectrometer (Thermo Fisher Scientific, Carlsbad, CA, USA). Fragment spectra were analyzed using the National Center for Biotechnology Information nonredundant protein database with Mascot (Matrix Science).

Co-immunoprecipitation

The cells were lysed by cold lysis buffer (1% Triton X-100, 50 mM Tris-7.5, 1 mM EDTA, 150 mM NaCl, and protease inhibitors). The supernatant from cell lysates was incubated with indicated antibodies. Then, the protein A/G beads (protein A/G agarose beads) were incubated with the lysates. The beads were washed by cold lysis buffer, and the protein loading buffer was added to the beads, followed by western blotting analysis.

CHX-chase assay

CHX-chase assay was performed using CHX (Selleck Chemicals), an inhibitor of protein synthesis. The cells in each group were treated with 20 μg/mL of CHX, and the expression of MYH9 protein was determined by western blot analysis at 0, 4, 8, and 12 h.

Ubiquitination assay

Cells were transfected with HA-labeled ubiquitin vector (HA-Ub) and His-labeled MYH9 vector (MYH9-His) with or without the EIF6-224aa overexpression vector. 24 h after transfection, 10 μM MG132 (Selleck Chemicals) was added to the cells for 6 h. Cell lysates were then obtained using lysis buffer, and immunoprecipitation was performed by incubating lysates with a His antibody and Protein A/G beads (Life Technologies) overnight at 4°C. Finally, the proteins were analyzed using western blotting.

RNA extraction and qRT-PCR

Total RNA was isolated from tissues or cells using the TRIzol Reagent (Invitrogen, USA). Then cDNA was synthesized using the PrimeScript RT Reagent Kit (Takara, Shiga, Japan). qRT-PCR was carried out using the SYBR Green PCR mix (Takara). Actin was used as the endogenous control. The primers used in this work were supplied in Table S2.

Protein isolation and western blot

Cells were lysed with RIPA (Radio-Immunoprecipitation Assay) buffer containing PMSF (Biocolors, Shanghai, China), separated on 10% SDS-PAGE gels, and electroblotted onto a polyvinylidene fluoride (PVDF) membrane (Millipore). After blocking with 5% nonfat milk, the membrane was incubated with various specific primary antibodies overnight at 4°C. Blots were washed and incubated with horseradish peroxidase-coupled secondary antibodies (Millipore) for 2 h. The protein bands were detected using the Pro-lighting horseradish peroxidase (HRP) agent. β-actin served as a loading control. The antibodies used in this study are shown in Table S3, and the density of the western blot bands was analyzed and is shown in Figure S3.

Production of the EIF6-224aa antibody

The 48- to 62-aa sequence of EIF6-224aa (PDAGREVAESSLGLR) was selected as the template for the antigen, and a cysteine (C) was added to the 5′ end of the 15-aa sequence (CPDAGREVAESSLGLR), which is necessary to obtain the BSA-coupled antigen, to generate an antibody specific to EIF6-224aa. The specific polyclonal antibody against the antigen sequence was produced by inoculating rabbits, which were obtained from Abcepta (San Diego, CA, USA). This antibody only recognizes EIF6-224aa and not the linear EIF6 protein, due to the unique target sequence of the antigen. The construction procedures and inspection reports of antibodies and antigens are described in Data S1.

In vivo proliferation and metastasis assay

The in vivo proliferation and metastasis assay was performed as described previously.⁵⁹ Briefly, for MDA-MB-231 expressing pLCDH and circ-EIF6 cells (1 × 10⁷cells) in 200 μL of PBS, Matrigel (1:3, v/v) was injected subcutaneously into the left flank of 4- to 6-week-old BALB/c nu/nu female mice (five mice per group). Tumor growth rate was monitored by measuring tumor diameters every 5 days after the initial 10 days. Both maximum (S) and minimum (W) length of the tumor were measured using a slide caliper, and the tumor volume was calculated as ½SW². When mice were killed, tumors were collected. To produce experimental lung metastasis, 5 × 10⁵ cells were injected into the lateral tail veins of 4- to 6-week-old BALB/c nu/nu female mice (five mice per group). After 2 weeks, all of the mice were killed under anesthesia. The lungs were collected and fixed in 10% formalin. For tissue morphology evaluation, H&E staining was performed on sections from embedded samples as previously described.⁶⁰ All animal experiments were performed with the approval of the Shandong University Animal Care and Use Committee.

Statistical analysis

Student’s t test was performed to analyze whether two experimental groups had significant differences, and one-way analysis of variance (ANOVA) was used for comparisons among multiple groups. The data in the statistical tests conformed to a normal distribution, and the variances were similar. A two-tailed value of p < 0.05 was considered statistically significant. Data are presented as the mean ± SD from at least three independent experiments. The survival analysis was performed by generating Kaplan-Meier curves, and the log-rank test was used to determine significance.