IRF4-activated TEX41 promotes the malignant behaviors of melanoma cells by targeting miR-103a-3p/C1QB axis

To investigate the role of TEX41, we first used GEPIA database to figure out the expression of TEX41 in tumor tissues and normal tissues. The results showed that TEX41 was expressed at a higher level in skin cutaneous melanoma (SKCM) tissues than in normal skin tissues (Fig. 1 A-B). Moreover, survival analysis based on GEPIA showed that the overall survival of melanoma patients with high TEX41 expression was poorer than that of patients with low TEX41 expression (Fig. 1 C). These findings indicated that highly expressed TEX41 might be a factor associated with the survival of SKCM patients. Accordingly, we detected the expression of TEX41 in melanoma cell lines (A375, WM35, A2058, SK-MEL-2) and normal epidermal melanin cell line, HEMa-LP. The results of qRT-PCR analysis showed that TEX41 expression was higher in melanoma cell lines (especially in A375 and SK-MEL-2 cell lines) than in HEMa-LP cell line (Fig. 1D). In summary, TEX41 was over-expressed in SKCM tissues and melanoma cell lines. As the statistics illustrated that high TEX41 expression was associated with unfavorable prognosis of melanoma patients, the specific roles and mechanism of TEX41 in melanoma cells deserved further exploration.

TFs have been reported to work as crucial mediators of gene expression and their dysregulation may impact cancer prognosis [18]. Herein, we utilized bioinformatics tools to predict TFs that might bind to TEX41 and were up-regulated in SKCM tissues, and the searching outcomes were displayed as Venn diagram. Specifically, human TFDB was employed to screen out potential TFs that might combine with regulatory sequences of TEX41 under the indicated conditions: predicted score >= 15, upstream by 2000 bases and downstream by 100 bases. Then, we searched on GEPIA database to find the candidate genes at a high expression level in SKCM with the indicated conditions: log2 fold change > 2 and p < 0.05. Based on the overlap of two outcomes, three potential TFs of TEX41 stood out, including IRF4, FOXM1 and FOXD3 (Fig. 2 A). Next, the expression levels of three TFs were detected by qRT-PCR. The results showed that the three TFs all exhibited a higher expression in melanoma cell lines, compared with HEMa-LP. And IRF4 was more differentially expressed than the other two TFs (Fig. 2B). By means of ChIP assay, it was shown that IRF4 had a stronger affinity with the TEX41 promoter in A375 and SK-MEL-2 cells, compared with FOXM1 and FOXD3 (Fig. 2 C). The additional results of luciferase reporter assays demonstrated that there were no visible changes observed in TEX41-Wt group after the overexpression of FOXM1 and FOXD3, which excluded the possibility that FOXM1 or FOXD3 might take part in regulating the expression of TEX41 (Fig. S1A). After IRF4 was chosen, the overexpression efficiency of pcDNA3.1-IRF4 and knockdown efficiency of sh-IRF4#1/2 in A375 and SK-MEL-2 cells were detected by means of qRT-PCR. It turned out IRF4 could be remarkably overexpressed or reduced by pcDNA3.1-IRF4 or sh-IRF4#1/2 (Fig. 2D). Subsequently, JASPAR was applied to predict the possible binding sites between IRF4 and TEX41 promoter (Fig. 2E). The results of luciferase reporter assays displayed that IRF4 interacted with TEX41 promoter only when the binding sites were not mutated (Fig. 2 F). Moreover, qRT-PCR was conducted to detect the expression of TEX41 after melanoma cells were transfected with pcDNA3.1-IRF4 or sh-IRF4#1/2. The results demonstrated TEX41 could be elevated or diminished by IRF4 overexpression or knockdown, showing that IRF4 could positively regulate TEX41 expression (Fig. 2G). All the above results confirmed that IRF4 acted as a TF to activate the transcription of TEX41 in melanoma cells.

To explore the effects of TEX41 on melanoma cells, we knocked down TEX41 in A375 and SK-MEL-2 cells (Fig. 3 A). In CCK-8 assays, the proliferation ability of melanoma cells with the transfection of sh-TEX41#1/2 was significantly impaired (Fig. 3B). The results of colony formation assay manifested that the colony formation ability of melanoma cells was weakened when the expression level of TEX41 was decreased (Fig. 3 C). Similarly, EdU assays also demonstrated that melanoma cell proliferation capacity was inhibited in response to TEX41 depletion (Fig. 3D). In addition, transwell assays were implemented to detect the migration and invasion ability of melanoma cells, and the results showed that the number of migrated and invaded cells was significantly reduced when the expression of TEX41 was lessened (Fig. 3E-F). Finally, TUNEL and flow cytometry assays were conducted to detect melanoma cell apoptosis, and it was found that the down-regulation of TEX41 accelerated the apoptosis of melanoma cells (Fig. 3G and S1B). Taken together, TEX41 served as an oncogene to facilitate the proliferation, migration, invasion, but hampered the apoptosis of melanoma cells.

Knowing that the TEX41 had an oncogenic effect on melanoma cells, we further explored its regulatory mechanism. At the very beginning, the fluorescent images captured in FISH assay showed that TEX41 was mainly distributed in the cytoplasm. After TEX41 knockdown, cytoplasmic TEX41 was significantly reduced in melanoma cells (Fig. 4 A). Nucleo-cytoplasmic separation experiments, with U6 as the nuclear reference and GAPDH as the cytoplasmic reference, further verified the cytoplasmic distribution of TEX41 (Fig. 4B). Many reports have explained that cytoplasmic lncRNAs can act as competing endogenous RNAs (ceRNAs) by binding to miRNAs to modulate the targets of miRNAs at the post-transcriptional level [19, 20]. Thus we hypothesized that TEX41 might function through ceRNA network in melanoma cells. StarBase (http://​starbase.​sysu.​edu.​cn/​) was used to predict potential miRNAs that might interact with TEX41 with the condition of Pan-Cancer >= 6, and four miRNAs (miR-15a-5p, miR-15b-5p, miR-103a-3p and miR-885-5p) were obtained. As shown in Fig. 4 C, the results of RNA pull down assay demonstrated that substantial miR-103a-3p could be pulled down by biotin-labeled TEX41 probe, for which the other three candidate miRNAs were excluded. Ago2 RIP assays further validated the interaction between miR-103a-3p and TEX41 (Fig. 4D). According to the above predictive outcomes of candidate miRNAs from starBase, we also obtained the potential binding sites between miR-103a-3p and TEX41 in Alignment column. We mutated binding sites in TEX41 for RNA pull down and luciferase reporter assays (Fig. 4E). RNA pull down results implied that miR-103a-3p was enriched in the biotin-labeled TEX41-Wt group rather than in the biotin-labeled TEX41-Mut group (Fig. 4 F). Then, the data from luciferase reporter assays proved that overexpressed miR-103a-3p significantly reduced the luciferase activity of TEX41-Wt in A375 and SK-MEL-2 cells, rather than that of TEX41-Mut (Fig. 4G). All these results above manifested that miR-103a-3p could bind to TEX41 in melanoma cells.

Since miRNAs could exert their biological functions by mediating downstream mRNA, we tried to identify the potential downstream mRNAs of miR-103a-3p. GEPIA database was employed to output most differential survival genes (p < 0.05) and up-regulated genes (log2 fold change < 2 and p < 0.05) in SKCM. In addition, with no specific conditions, all candidate mRNAs targeted by miR-103a-3p were obtained from starBase. Considering the overlapped data from three predictions, we determined three candidate mRNAs (FOXM1, BST2 and C1QB) as the candidates (Fig. 5 A). To test if FOXM1, BST2 and C1QB were target genes of miR-103a-3p, qRT-PCR was conducted, and the results showed that when miR-103a-3p was up-regulated, C1QB was the most significantly down-regulated one in melanoma cells (Fig. 5B). In Ago2-RIP experiment, we verified the interaction between C1QB and miR-103a-3p as they were overtly enriched in anti-Ago2 (Fig. 5 C). StarBase website was employed to predict the potential binding sites between C1QB and miR-103a-3p, and the binding site in C1QB was mutated (Fig. 5D). The consequence of luciferase reporter assays showed that overexpressing miR-103a-3p reduced the luciferase activity of C1QB-Wt, but had little effects on that of C1QB-Mut (Fig. 5E). To sum up, C1QB was validated to work as the target gene of miR-103a-3p in melanoma cells.

Prior to the exploration on the function of C1QB in melanoma cells, we first searched on GEPIA database to find the expression of C1QB in SKCM tissues and normal skin tissues. The data demonstrated the expression of C1QB was remarkably higher in SKCM tissues than that in normal skin tissues (Fig. S1C). Then, we performed qRT-PCR to detect the expression of C1QB in HEMa-LP and melanoma cells. The collected data indicated that C1QB was expressed at a higher level in melanoma cells, in contrast to HEMa-LP cells (Fig. S1D). A375 and SK-MEL-2 cells held the highest expression level of C1QB. Next, we successfully knocked down C1QB in A375 and SK-MEL-2 cells (Fig. 6 A). A series of functional experiments, including CCK8, transwell and TUNEL were carried out to respectively evaluate the influence of C1QB on proliferation, migration/invasion and apoptosis of melanoma cells. The outcomes indicated that knockdown of C1QB prominently dampened the proliferative, migratory and invasive abilities of melanoma cells (Fig. 6B-D), but accelerated melanoma cell apoptosis (Fig. 6E). Moreover, WM35 was selected for the following gain-of-function assays due to its relatively lower expression of C1QB compared with other melanoma cells. CCK-8, transwell migration and invasion assays were carried out using WM35 and HEMa-LP cells. The results suggested that the proliferative, migratory and invasive abilities of WM35 cells were stronger than those of HEMa-LP cells, and C1QB augment enhanced these abilities (Fig. S1E-G). In terms of TUNEL assay, the apoptosis of WM35 cells was weaker than that of HEMa-LP cells. Moreover, the apoptosis of WM35 and M3Ma-LP cells was restricted by C1QB overexpression (Fig. S1H). To conclude, these findings demonstrated that C1QB promoted proliferation, migration and invasion of melanoma cells, while inhibiting the cell apoptosis.

To further confirm whether TEX41 exerted its functions in melanoma cells via targeting miR-103a-3p/C1QB, we conducted a series of rescue experiments. Prior to that, qRT-PCR was conducted and it was found C1QB expression declined due to TEX41 knockdown but was restored by miR-103a-3p inhibition (Fig. 7 A). Next, the overexpression efficiency of pcDNA3.1-C1QB/C1QB-Mut was detected via qRT-PCR. The results showed that the expression level of C1QB was increased by transfection with pcDNA3.1-C1QB/C1QB-Mut plasmid (Fig. 7B). The results of CCK-8 assay showed that inhibition of miR-103a-3p or overexpression of C1QB could restore the suppressive effect of sh-TEX41#1 on cell viability (Fig. 7 C). Similarly, colony formation and EdU assays disclosed that when miR-103a-3p was diminished or C1QB was overexpressed, weakened proliferative ability of melanoma cells in response to sh-TEX41#1 transfection was significantly restored (Fig. 7D-E). Results of transwell assays manifested inhibition of miR-103a-3p or overexpression of C1QB rescued the reduced cell migration and invasion caused by TEX41 depletion (Fig. 7 F-G). TUNEL and flow cytometry assays were also performed to examine the apoptotic rate, and the augmented cell apoptosis induced by decreased TEX41 was recovered by miR-103a-3p inhibitor or pcDNA3.1-C1QB/C1QB-Mut (Fig. 7 H and S2A). In summary, the abovementioned findings suggested that TEX41 facilitated proliferation, migration and invasion of melanoma cells, but restrained the cell apoptosis by targeting miR-103a-3p/C1QB axis.

Aiming to verify the relationship between IRF4 and C1QB, we conducted qRT-PCR first. It turned out IRF4 depletion reduced the expression of C1QB, and IRF4 overexpression enhanced C1QB expression (Fig. S3A). Rescue assays were then performed to study the underlying mechanisms. As shown in Fig. S3B, cell proliferation was dramatically impeded after IRF4 knockdown, whereas co-transfection of pcDNA3.1-C1QB recovered the weakened cell proliferative ability. Data from transwell assays demonstrated suppressive effect of sh-IRF4#1 on cell migration and invasion was offset by C1QB overexpression (Fig. S3C-D). Moreover, the results of TUNEL assay reflected that enhanced cell apoptosis caused by IRF4 reduction was restored by up-regulation of C1QB (Fig. S3E). In brief, IRF4 raised C1QB expression in melanoma cells to facilitate cell growth.