Leptin and insulin up-regulate miR-4443 to suppress NCOA1 and TRAF4, and decrease the invasiveness of human colon cancer cells

To check the effects of leptin and insulin exposure on the expression levels of a wide cross-section of miRNAs, we profiled miRNA levels in the CRC-derived cell lines HCT-116, HT-29 and DLD-1 that had been treated with leptin or insulin (at 200 ng/ml) for 24 h, using the Nanostring nCounter probe array platform (clustered heat map for leptin-induced changes, Additional file 2: Figure S1; all expression data, Additional file 3: Table S2). A positive correlation between the effects of insulin and leptin on miRNA expression profiles was observed in HCT-116 and HT-29 cells, but not in DLD-1 (Fig. 1a–c). Additionally, the effects of insulin on miRNA expression profiles correlated in all three cell lines (Fig. 1d, e) but the effects of leptin only correlated between HCT-116 and HT-29, but not between them and DLD-1 (Fig. 1f, g). Of ~800 miRNAs profiled, miR-4443 stood out in its robust and similar response to leptin and insulin. Thus, miR-4443 was up-regulated by insulin in all three lines, and by leptin in HCT-116 and HT-29, but not in DLD-1 (miR-4443 marked in black, in Fig. 1d–g). These results suggest that HCT-116 and HT-29 express the functional leptin receptor (LEPR), which is necessary for leptin’s downstream effects on the miRNA profile, while DLD-1 may lack the receptor or express an inactive form of LEPR. RT-qPCR using primers targeting the domain common to all variants of LEPR yielded a specific product at the expected length for HCT-116 and HT-29 – derived DNA, but that of DLD-1 was shorter (Fig. 1h). Additionally, no PCR product was observed for DLD-1, when primers specifically targeting LEPR variant 1 (the long and active form) were employed (Fig. 1h). This lack of a full-length LEPR transcript was confirmed in 2 separate stocks of DLD-1 cells, from different sources (data not shown). Immunoblots for LEPR produced two bands in DLD-1, the upper one rather strong, and both differing in size from the single band in the other cell lines (Fig. 1i), suggesting the accumulation of altered or inactive LEPR variants in DLD-1. We chose HCT-116 for further study as this cell line was morphologically and phenotypically stable in culture and was confirmed to express LEPR.

The Nanostring profiling results of miRNA levels obtained from HCT-116 cells treated with leptin or insulin at 200 ng/ml were validated by RT-qPCR (Fig. 2a). miR-4443, which showed robust up-regulation by leptin in both the Nanostring profiling and the qRT-PCR validation, was chosen as a candidate for further study. In a dose-response experiment, leptin treatment at 100 ng/ml was found to up-regulate miR-4443 reproducibly as well as significantly (p < 0.05) (Fig. 2b). This dose of leptin also caused a significant (p < 0.05) decrease in cell invasion through a MatriGel-coated membrane (Fig. 2c). We therefore used this dosage of leptin for further experiments. We also established that 20 ng/nl insulin was sufficient to cause the up-regulation of miR-4443 (data not shown).

Transfection of HCT-116 with a miR-4443 mimic tended to decrease cell invasion as compared to transfection with a control oligo (Fig. 2d); this trend was reproducible but fell short of statistical significance due to large variation between individual experiments. Overexpression of miR-4443 also significantly decreased the proliferation of HCT-116 cells (Fig. 2e), implying that miR-4443 may act as a tumor suppressor via its downstream targets.

To identify possible cancer-relevant targets of miR-4443, the TargetScan algorithm [9, 24] was used. Among the top-scored predicted targets of miR-4443, NCOA1 and TRAF4 have known roles in cell migration and cancer metastasis [35‐39], and were chosen for validation. Exposure to leptin (at 100 ng/ml for 24 h), as well as overexpression of miR-4443 in HCT-116 cells (24 h post-transfection) resulted in significant (p < 0.05) down-regulation of endogenous NCOA1 and TRAF4 on the mRNA and protein levels (Fig. 3a–d). Insulin treatment (at 20 ng/ml) led to a similar down-regulation of NCOA1 and TRAF4 mRNA (Fig. 3a).

To check if miR-4443 directly targets the 3′UTR region of NCOA1 and TRAF4 mRNAs, HCT-116 cells were co-transfected with miR-4443 mimic (or control oligos) and a secreted dual-reporter encoding vector, with or without the relevant 3′UTR (SecretePair from Genecopoeia). The Gaussia Luciferase signal was significantly suppressed by the co-transfected miR-4443 mimic, compared with the negative control oligo, in both NCOA1 and TRAF4 3′UTR-containing constructs. but not in a control plasmid lacking the 3′UTRs (Fig. 3g), supporting the predicted direct regulation of these mRNAs by miR-4443.

Since miR-4443 was up-regulated following both leptin and insulin treatment, we hypothesized that its promoter or enhancer regions may contain the CCAAT motif, which binds transcription factors from the CCAAT-enhancer-binding proteins (C/EBP) family, that act downstream of MEK1/2 in leptin and insulin signaling [3, 40‐43] and were previously described as up-regulated by both insulin and leptin [44, 45]. To investigate this possibility, a 10 kb region of the human genomic sequence upstream of pre-miR-4443, which contains no other known genes on the same (plus) strand, was analyzed using the Cister algorithm [33]. Two high-score (0.99) matches for CCAAT motifs were found at positions 1903–18 and 1955–70 of the sequence, in addition to other adjacent cis-elements predicted to bind SP1 and E2F transcription factors that combine to form a predicted high probability enhancer region ~8 kb upstream of the miR-4443-encoding locus (Additional file 4: Figure S2). Additional binding motifs present on the opposite (minus) strand at the same locus, likely affect the expression of the adjacent CDC25A gene. This in-silico analysis suggests that miR-4443 is regulated by the MEK1/2 – C/EBP pathway downstream of both the insulin and leptin receptors, providing a possible explanation for the similar effects of exposure to insulin and leptin on miR-4443 and its downstream targets.

To assess if MEK1/2 indeed regulate miR-4443 downstream of leptin and insulin signaling, HCT-116 cells were pre-incubated for 45 min with or without a MEK inhibitor, PD-98059 (10 μM), before being exposed to leptin (100 ng/ml) or insulin (20 ng/ml) for 24 h. The inhibitor attenuated the leptin-induced up-regulation of miR-4443 (Fig. 4a). A similar trend was observed with insulin, although the up-regulation was modest (Fig. 4a). Co-treatment with PD-98059 also abolished the down-regulation of NCOA1 and TRAF4 mRNA levels by leptin in these cells (Fig. 4b). These results support the notion that miR-4443 is regulated by the MEK1/2 – C/EBP pathway downstream of both the insulin and leptin receptors, although it is impossible to discount other, miR-4443-independent signaling pathways that could affect the levels of NCOA1 and TRAF4 downstream of these receptors.