A regulatory BMI1/let‐7i/ERK3 pathway controls the motility of head and neck cancer cells

Extracellular signal‐regulated kinase 3 (ERK3) is an atypical mitogen‐activated protein kinase (MAPK), whose biological activity is tightly regulated by its cellular abundance. Recent studies have revealed that ERK3 is upregulated in multiple cancers and promotes cancer cell migration/invasion and drug resistance. Little is known, however, about how ERK3 expression level is upregulated in cancers. Here, we have identified the oncogenic polycomb group protein BMI1 as a positive regulator of ERK3 level in head and neck cancer cells. Mechanistically, BMI1 upregulates ERK3 expression by suppressing the tumor suppressive microRNA (miRNA) let‐7i, which directly targets ERK3 mRNA. ERK3 then acts as an important downstream mediator of BMI1 in promoting cancer cell migration. Importantly, ERK3 protein level is positively correlated with BMI1 level in head and neck tumor specimens of human patients. Taken together, our study revealed a molecular pathway consisting of BMI1, miRNA let‐7i, and ERK3, which controls the migration of head and neck cancer cells, and suggests that ERK3 kinase is a potential new therapeutic target in head and neck cancers, particularly those with BMI1 overexpression.


Introduction
Dysregulation of signal transduction pathways is a hallmark of many cancers (Cargnello and Roux, 2012;Lei et al., 2014). While the implication of several conventional mitogen-activated protein kinase (MAPK) pathways in cancers is well studied, the involvement of the atypical MAPKs in tumorigenesis is poorly understood (Kostenko et al., 2012). Extracellular signalregulated kinase 3 (ERK3), also known as MAPK6, is an atypical member of the MAPK family (Coulombe and Meloche, 2007;Kostenko et al., 2012). The importance of ERK3 signaling in cancers has been recently recognized following our previous finding that ERK3 promotes cancer cell invasiveness by phosphorylating steroid receptor coactivator 3 (SRC-3) oncoprotein and upregulating SRC-3-mediated transcription of matrix metalloproteinase (MMP) genes (Long et al., 2012). In addition, ERK3 was shown to promote breast cancer cell migration by regulating cell morphology and spreading (Al-Mahdi et al., 2015). Furthermore, ERK3 enhances the activity of tyrosyl DNA phosphodiesterase 2 (TDP2) in DNA damage response and increases the chemoresistance of lung cancer cells to topoisomerase-2 inhibitors (Bian et al., 2016). In line with its important roles in cancer cell migration, invasion, and DNA damage response, ERK3 is upregulated in multiple cancers, including non-small-cell lung cancer (Long et al., 2012), gastric cancer (Liang et al., 2005), and oral squamous cell carcinoma (Rai et al., 2004). Little is known, however, about the molecular mechanisms of ERK3 upregulation in cancers. The level of ERK3 protein in cells is thought to be a critical regulator for ERK3 activity, as unlike other MAPK family members, ERK3 is a highly unstable protein with a half-life of 30-45 minutes in exponentially proliferating cells (Coulombe et al., 2003(Coulombe et al., , 2004. BMI1 is a key regulatory component of the transcription suppressor complex, the polycomb repressive complex-1 (PRC1) (Cao et al., 2011;Siddique and Saleem, 2012). It plays important roles in the maintenance and self-renewal of normal and cancer stem cells (Lessard and Sauvageau, 2003;Park et al., 2003;Rizo et al., 2009;Schuringa and Vellenga, 2010) and promotes tumor cell growth, migration, and invasion, thereby promoting tumor growth and progression (Cao et al., 2011;Jiang et al., 2009;Siddique and Saleem, 2012;Wu et al., 2011). BMI1 functions as an oncoprotein by silencing various tumor suppressor genes, such as p16Ink4a, p14Arf, PTEN (Cao et al., 2011;Jacobs et al., 1999;Song et al., 2009), and microRNAs (miRNAs) including let-7i (Chou et al., 2013;Yang et al., 2012). miRNAs act as post-transcriptional regulators of gene expression by repressing mRNA translation and/or facilitating mRNA degradation (Lee, 2014;Ranganathan and Sivasankar, 2014). Recent studies have shown that let-7i plays tumor suppressive roles by inhibiting tumor cells' growth and migration (Fawzy et al., 2016;Subramanian et al., 2015;Tian et al., 2015;Wu et al., 2015Wu et al., , 2016Yang et al., 2012;Zhang et al., 2015). let-7i is shown to be downregulated in several cancers including head and neck squamous cell carcinomas (HNSCCs; Liu et al., 2012;Roush and Slack, 2008;Subramanian et al., 2015;Yang et al., 2012). HNSCC patients with lower levels of let-7i had increased local invasion of tumor cells to adjacent tissues .
In this study, we revealed a molecular mechanism for the regulation of ERK3 expression in head and neck cancer cells: BMI1 upregulates ERK3 by suppressing let-7i miRNA that directly targets ERK3 mRNA. Importantly, our study reveals a regulatory pathway consisting of BMI1, let-7i, and ERK3 that is important for controlling cancer cell migration.
2.3. Generation of stable cell pools expressing shRNA or cDNA OECM1 cell line with stable knockdown of BMI1 (OECM1-shBMI1) was generated by lentiviral expression of a short hairpin RNA (shRNA) specifically targeting BMI1 (shBMI1) as described previously . OECM1 cells with stable expression of a scrambled nontargeting shRNA (OECM1-shCtrl) served as a control. Fadu cells with stable overexpression of BMI1 (Fadu-BMI1) or the control empty vector pCDH-CMV-MCS-EF1-Puro (Fadu-CDH) were generated as follows. First, pseudotyped lentiviral particles were produced in 293T cells by cotransfecting pCDH-BMI1 or pCDH-CMV-MCS-EF1-Puro with Trans-Lentiviral Packaging Plasmid Mix (Open Biosystems, Lafayette, CO, USA). Pseudoviral particles were harvested 48 h after transfection and concentrated using PEG-it Virus Precipitation Solution (System Biosciences), following the manufacturer's instructions. Next, Fadu cells were transduced with the prepared virus in the presence of polybrene (4 lgÁmL À1 ). Two days post-transduction, cells were split and selected by puromycin (1 lgÁmL À1 ) for 10 days. The stable overexpression of BMI1 was verified by western blotting analysis.

Transient lentiviral transduction
OECM1 stable cell pools were transiently transduced with lentiviruses expressing either an empty vector pCDH or pCDH-Myc6-ERK3 in the presence of 4 lgÁml À1 polybrene for 2 days.

RNA extraction and RT-qPCR
For gene expression analysis, the total RNA was extracted from cells using Trizol reagent (Ambion, Waltham, MA, USA) and reverse transcription (RT) was carried out using SuperScript VILO Master Mix (Invitrogen, Carlsbad, CA, USA) according to the manufacturer's protocol. Quantitative polymerase chain reaction (qPCR) was performed using TaqMan Probe system (Roche Diagnostics) on the Applied Biosystems 7500 (Applied Biosystems/ThermoFisher Scientific, Foster City, CA, USA) with GAPDH as the internal control. Relative expression to normalizer sample was calculated using the DDCT method.
To measure miRNA expression, total RNA was extracted using mirVana miRNA Isolation Kit (Ambion) and reverse-transcribed to cDNA using Taq-Man Advanced MicroRNA cDNA Synthesis Kit (Applied Biosystems/ThermoFisher Scientific, Foster City, CA, USA). qPCR was performed using TaqMan Advanced MicroRNA Assay kit (ThermoFisher Scientific, Waltham, MA, USA) following the manufacturer's instructions. The relative expression of hsalet-7i-5p was normalized to that of hsa-miR-191-5p.

Luciferase reporter assay
HeLa cells were cotransfected with a luciferase reporter plasmid and a miRNA mimic or inhibitor. The luciferase activity was measured 30 h or 44 h, respectively, post-transfection using LightSwitch Luciferase Assay Kit (SwitchGear Genomics).

Two-chamber transwell cell migration assay
Cell migration was analyzed using a modified twochamber transwell system (BD Biosciences, San Jose, CA, USA), following the manufacturer's instructions. Cells were detached by trypsin/EDTA, washed once with serum-free medium, and then resuspended in medium with 2% FBS (for OECM1 cells) or serumfree medium (for Fadu cells). Complete culture medium with 10% FBS was added to each bottom well. Cells were added in transwell inserts and allowed to migrate for 18 h (for OECM1 cells) or 30 h (for Fadu cells) in a 37°C cell incubator. The cells in the upper surface of the transwell were removed using cotton swabs and migrated cells attached on the undersurface were fixed with 4% paraformaldehyde for 10 min and stained with crystal violet solution (0.5% in water) for 10 min. Migrated cells were then photographed and counted under a microscope at 50 9 magnification.

Wound healing assay
Wound healing was assayed in confluent OECM1 cell monolayers. Cells were scraped using a standard 200-lL pipette tip and the wounded monolayers were washed twice to remove nonadherent cells. Cells were photographed every 5 h, and the wound width was measured. Wound closure was calculated by subtracting the wound width at 20-h time point from the initial wound width (mm).

Immunohistochemical analysis of ERK3 and BMI1 in tumor tissue microarrays
To determine the expression of ERK3 and BMI1 proteins in head and neck cancer tissues, we purchased two identical head and neck cancer tissue microarrays (HN803c; US Biomax, Derwood, MD, USA). Each microarray was comprised of the following formalinfixed, paraffin-embedded (FFPE) tissues: 60 cases of squamous cell carcinoma of larynx, palate, lower lip, upper jaw, tongue, gingiva, larynx, throat, mandible, or cheek and eight cases of metastatic carcinoma. ERK3 and BMI1 proteins were immunostained following the procedures described previously (Cai et al., 2010) with the following modifications. Antigen retrieval was performed by treating the slides in citrate-based antigen unmasking solution (pH 6.0; Vector Laboratories, Burlingame, CA, USA) at 90°C for 12 min using a pressure cooker. One slide was incubated with an antibody against ERK3 (1:50 dilution; Abcam) and the other slide was incubated with an antibody against BMI1 (1 : 400; Cell Signaling) overnight at 4°C, followed by the fluorescent labeling with Alexa Fluor 488 anti-rabbit secondary antibody (Invitrogen) and DAPI staining. The tissues were visualized and imaged using a Leica CTR 6000 Microscope (Leica Microsystems, Wetzlar, Germany) and IMAGEPRO 6.2 software (Media Cybernetics, Rockville, MD, USA). Tumor lesions in each sample were identified by examining an H&Estained slide with the same tumor tissue samples. Multiple measurements of the fluorescent staining intensity of ERK3 and BMI1 in tumor regions (at least three fields) for each sample were made using IMAGEPRO 6.2 software and were normalized by subtracting background fluorescence intensity.

Statistics
Data are expressed as mean AE standard deviation (SD) or standard error (SE) as specified in the figure legends. Statistical significance was determined by a two-sided Student's t-test, where a P-value of less than 0.05 was considered statistically significant. The correlation between ERK3 and BMI1 mean fluorescence intensities was determined using the Pearson's correlation analysis and P = 0.001.

BMI1 upregulates the expression of ERK3 in head and neck cancer cells
To elucidate the regulation of ERK3 expression in cancers, we attempted to search databases and publications for potential regulators of ERK3. Interestingly, a recent study of transcriptional profiling in a multiple myeloma cell line with stable knockdown of BMI1 showed that ERK3 is among the downstream target genes of BMI1 (Jagani et al., 2010). As BMI1 is an oncogenic transcriptional factor for cancer cell growth and migration (Cao et al., 2011) and both BMI1 and ERK3 are overexpressed in head and neck cancers (Rai et al., 2004;Song et al., 2006), we were interested in investigating the molecular regulation of ERK3 by BMI1 and the functional significance of this axis in   (Chou et al., 2013). Hence, we examined ERK3 levels in OECM1 cells after stable knockdown of BMI1 and in Fadu cells with stable BMI1 overexpression. Stable knockdown of BMI1 in OECM1 cells (OECM1-shBMI1) greatly reduced ERK3 protein level (Fig. 1A), whereas BMI1 overexpression in Fadu cells increased ERK3 protein level (Fig. 1C). Interestingly, while ERK3 mRNA level was significantly decreased upon BMI1 knockdown in OECM1 cells (Fig. 1B), BMI1 overexpression in Fadu cells did not alter ERK3 mRNA expression (Fig. 1D). The positive regulation of ERK3 expression by BMI1 was confirmed in two other head and neck cancer cell lines, UMSCC1 and Detroit 562, in which transient knockdown of BMI1 by siRNA (siBMI1) led to a decrease in ERK3 protein level (Fig. 1E). Taken together, these results demonstrate that BMI1 positively regulates ERK3 expression in head and neck cancer cells.

BMI1 regulates ERK3 by an indirect mechanism involving let-7i
Next, we wanted to elucidate the molecular mechanism of regulation of ERK3 by BMI1. As BMI1 is a transcriptional repressor but positively regulates ERK3 expression, we postulated an indirect mechanism for this regulation, by which BMI1 suppresses a molecule that downregulates ERK3 expression. miRNAs, negative regulators of gene expression at post-transcriptional level, would be potential candidates. Interestingly, miRNA let-7i was shown to be repressed by BMI1 in head and neck cancer cells (Chou et al., 2013;Yang et al., 2012). In agreement with previous findings, stable knockdown of BMI1 led to increase in let-7i level in OECM1 cells ( Fig. 2A), suggesting that let-7i may target ERK3 mRNA because knockdown of BMI1 led to decrease in ERK3 expression (Fig. 1). Indeed, treatment of OECM1 cells with let-7i mimic greatly reduced ERK3 expression at both mRNA and protein levels (Fig. 2B,C). Conversely, suppression of let-7i by let-7i inhibitor in OECM1 cells led to increase in ERK3 expression at both mRNA (Bar 2 versus Bar 1, Fig. 2D) and protein levels (Lane 2 versus Lane 1 of ERK3 blot, Fig. 2E). Importantly, while stable knockdown of BMI1 led to decrease in ERK3 protein level (Lane 3 versus Lane 1, Fig. 2E) and increase in let-7i level ( Fig. 2A), suppression of let-7i by let-7i inhibitor restored ERK3 protein level under the condition of stable BMI1 knockdown in OECM1 cells (Lane 4 versus Lane 3, Fig. 2E). These results strongly suggest that let-7i negatively regulates ERK3 expression and that BMI1 upregulates ERK3 expression through suppressing let-7i expression. The regulation of ERK3 by BMI1/let-7i axis was confirmed in Fadu cells. Treatment with let-7i inhibitor in parental Fadu cells led to increase in ERK3 protein level (Fig. 2G), whereas let-7i mimic greatly decreased ERK3 protein level in Fadu cells stably overexpressing BMI1 (Fig. 2H). Of note, while treatment with let-7i inhibitor led to increase in ERK3 expression at both protein and mRNA levels in OECM1 cells (Fig. 2D,E), it increased ERK3 protein level, but had no effect on ERK3 mRNA expression in Fadu cells (Fig. 2F). These results suggest that in OECM1 cells let-7i facilitates ERK3 mRNA degradation and/or suppresses ERK3 mRNA translation, whereas in Fadu cells, let-7i may only affect ERK3 mRNA translation. In line with the differential effects of let-7i on ERK3 expression in different cell lines, BMI1 upregulates ERK3 at both protein and mRNA levels in OECM1 cells (Fig. 1A,B), but only at protein level in Fadu cells (Fig. 1C,D), further suggesting that BMI1 regulates ERK3 expression through targeting let-7i.
3.3. let-7i downregulates ERK3 expression by directly targeting the 3 0 UTR of ERK3 mRNA miRNAs regulate gene expression by binding to the 3ʹUTRs of the targeted mRNAs via partial complementarity (Lee, 2014;Ranganathan and Sivasankar, 2014). Thus, we wanted to determine whether let-7i regulates ERK3 expression by targeting the 3 0 UTR of ERK3 mRNA. Indeed, we identified a putative let-7i binding site in the 3 0 UTR of ERK3 mRNA (Fig. 3A). Next, we performed a luciferase reporter assay to validate the targeting of the 3 0 UTR of ERK3 mRNA by let-7i. As shown in Fig. 3B, while let-7i mimic did not change the luciferase activity when luciferase gene is followed by a random 3 0 -UTR, it greatly reduced the luciferase activity when ERK3 3 0 -UTR is present downstream of the luciferase gene, suggesting that ERK3 3 0 UTR is a target for let-7i. To confirm this, we generated a luciferase reporter construct in which the ERK3 3 0 UTR harboring mutations of the let-7i binding site (shown in Fig. 3A) is placed downstream of the luciferase gene. Importantly, mutation of the let-7i binding site inhibited the negative regulation of let-7i mimic on luciferase activity (Fig. 3B). On the contrary, let-7i inhibitor significantly increased luciferase activity in cells transfected with ERK3 3 0 UTR luciferase reporter, but had no significant effect on the random 3 0 UTR luciferase reporter nor the mutant ERK3 3 0 UTR luciferase reporter (Fig. 3C). These results clearly demonstrate that ERK3 mRNA 3 0 -UTR is a direct target of let-7i.

The BMI1/let-7i/ERK3 pathway regulates head and neck cancer cell migration
To determine the functional significance of the BMI1/ let-7i/ERK3 pathway in cancer cells, we decided to study the effects of their interplay on cancer cell migration, a cellular function shared by all of them. As expected, stable depletion of BMI1 reduced ERK3 protein level (Lane 2 versus Lane 1, Fig. 4A) and also decreased the migration of OECM1 cells in a transwell migration assay (shBMI1/CDH versus shCtrl/CDH, Fig. 4B). Importantly, restoring ERK3 level by lentiviral expression of ERK3 cDNA (Lane 3, Fig. 4A) rescued the migration ability of OECM1 cells with stable depletion of BMI1 (shBMI1/CDH-ERK3 versus shBMI1/CDH, Fig. 4B). Similar results were observed in wound healing cell migration assay (Fig. 4C). These results suggest that ERK3 mediates the role of BMI1 in promoting cancer cell migration.
As let-7i level was shown to be increased concomitant with decrease in ERK3 expression when BMI1 was stably knocked down in OECM1 cells ( Fig. 1A and Fig. 2A), we anticipated that treatment with let-7i inhibitor would restore ERK3 protein level and cell migration ability of OECM1-shBMI1 cells. Indeed, both ERK3 protein levels (Fig. 5A) and cell migration ability (Fig. 5B) were greatly increased when OECM1-shBMI1 cells were treated with let-7i inhibitor (compare let-7i inh./siCtrl with ctrl inh./siCtrl). Notably, knockdown of ERK3 (Fig. 5A) inhibited the increase in cell migration induced by the treatment with let-7i inhibitor (compare let-7i inh./siERK3 with ctrl inh./siCtrl, Fig. 5B), suggesting that let-7i suppresses OECM1 cell migration by downregulating ERK3. Taken together, the results in Figs 4 and 5 clearly demonstrate the importance of the BMI1/let-7i/ERK3 pathway in regulating head and neck cancer cell migration: Both BMI1 and ERK3 promote migration and let-7i, as an intermediator for BMI1 and ERK3, inhibits cell migration.

ERK3 and BMI1 protein levels positively correlate in head and neck cancer tissues
To investigate the clinic relevance of our finding on the regulation of ERK3 expression by BMI1, we  Fig. 3. The 3 0 UTR of ERK3 mRNA is a direct target for let-7i miRNA. (A) Schematic illustration of the putative let-7i binding site in ERK3 3 0 UTR. The let-7i seed region, the putative let-7i binding site of ERK3 3 0 UTR, and the mutated let-7i binding site were underlined and bold. (B,C) Luciferase reporter assay of HeLa cells cotransfected with a luciferase reporter plasmid (harboring either a random 3 0 UTR sequence, ERK3 3 0 UTR, or the mutant ERK3 3 0 UTR) and a miRNA mimic (a nontargeting control mimic or let-7i mimic) in (B) or miRNA inhibitor (a nontargeting control inhibitor or let-7i inhibitor) in (C). Values in bar graphs are the relative luciferase activity (percentage of control mimic or inhibitor) and represent mean AE SE of three independent experiments. *P < 0.05; NS, not significant (Student's t-test). examined the levels of both ERK3 and BMI1 proteins by immunofluorescent staining in a FFPE head and neck cancer tissue microarray which contains tumor specimens with stages I to IV from totally 68 patients. ERK3 staining was primarily localized in the cytoplasm (Fig. 6A), whereas BMI1 was mainly observed in the nucleus (Fig. 6B). In agreement with our finding in cultured HNSCC cells that ERK3 is upregulated by BMI1, there is a positive correlation between ERK3 level and BMI1 level in the head and neck cancer tissues (correlation coefficient r = 0.432, P = 0.001) (Fig. 6C).
Based on the results of this study, we propose a novel molecular mechanism for the upregulation of ERK3 in cancer: BMI1 suppresses the transcription of the miRNA let-7i, a negative regulator of ERK3 expression, which leads to elevation in ERK3 protein level and increase in cancer cell migration (Fig. 7).

Discussion
ERK3, as a member of the atypical MAPK subfamily, was cloned 25 years ago together with ERK2 by homology screening of a rat brain cDNA library using an ERK1-derived probe (Boulton et al., 1991). While the conventional MAPKs, such as ERK1/2, have been extensively studied, much less is known about the molecular regulation of ERK3 activity and expression. ERK3 has gained attention in recent years due to the discovery of its important physiological and pathological functions. Mice with homologous deletion of ERK3 gene show neonatal lethal phenotype due to severe defect in lung differentiation and maturation (Klinger et al., 2009). ERK3 is also an important regulator for thymocyte activation and survival (Marquis et al., 2014a,b;Sirois et al., 2015), neuronal morphogenesis (Brand et al., 2012), and endothelial cell migration . As described in the Introduction, the importance of ERK3 in cancers has also been revealed. Although ERK3 is known to be upregulated in multiple cancers and promotes cancer cell migration and invasion, the regulation of ERK3 expression in cancer cells is largely unknown except for one study showing the positive regulation of ERK3 by BRAF in melanoma cells (Hoeflich et al., 2006).
Here, we reveal a new molecular mechanism for regulating ERK3 expression in head and neck cancer cells. Our finding about the tight regulation of ERK3 expression by the BMI1/let-7i axis is important given that BMI1 is a well-known oncoprotein and that let-7i is a tumor suppressor in multiple cancers including head and neck cancers. BMI1 protein is a key regulatory component of PRC1 transcription suppressor complex (Cao et al., 2011). Overexpression of BMI1 has been detected in several human cancers including head and neck cancer, colorectal carcinoma, nonsmall-cell lung cancer, and prostate cancer (Cao et al., 2011). Numerous studies have shown that BMI1 promotes cancer cell growth and migration/invasion, but inhibits apoptosis (Cao et al., 2011;Chou et al., 2013;Siddique and Saleem, 2012;Song et al., 2009;Yang et al., 2012). It acts as an oncoprotein through suppressing tumor suppressor genes including p16Ink4a, p14Arf, PTEN, and miRNAs (Cao et al., 2011;Jacobs et al., 1999;Song et al., 2009). In addition, BMI1, through indirect mechanisms, can upregulate genes that promote cancer progression, such as the kinase No. of migrated cells/ field * * A Fig. 5. Suppression of let-7i by let-7i inhibitor greatly increases migration of OECM1 cells with stable BMI1 knockdown and ERK3 depletion inhibits this effect. OECM1-shBMI1 cells were cotransfected with a control inhibitor (Ctrl inh.) or let-7i inhibitor (let-7i inh.) and with either a nontargeting control siRNA (siCtrl) or siRNA against ERK3 (siERK3), followed by western blot analysis of ERK3 (A) or a transwell migration assay (B). Values represent mean AE SD. *P < 0.001 (Student's t-test).
Aurora A (Chou et al., 2013), and several genes involved in epithelial-to-mesenchymal transition and transforming growth factor-b and epidermal growth factor/platelet-derived growth factor pathways (Ferretti et al., 2016). Our study identified ERK3 as a new target of BMI1 that is critical for cancer cell motility. Restoring ERK3 level by exogenous expression in cells with BMI1 depletion rescued cancer cell migration, while depletion of ERK3 in BMI1-overexpressing cells greatly diminished the effect of BMI1 promoting cancer cell migration, demonstrating that ERK3 functions downstream of BMI1 in promoting head and neck cancer cell motility. More importantly, we show that the levels of these two proteins positively correlate in HNSCC tumor specimens, highlighting the clinical significance of our study. Let-7i is a member of the let-7 family of tumor suppressor miRNAs that inhibits cancer cell proliferation, survival, migration, and invasion by downregulating a variety of oncogenes such as myc, Ras, HMGA2, and Lin28B (Boyerinas et al., 2010). Downregulation of let-7 miRNAs leads to upregulation of oncogenes and promotes tumor initiation and progression. As such, they can serve as diagnostic and/or prognostic tumor markers (Monroig-Bosque Pdel et al., 2015). More importantly, let-7 miRNAs have been utilized for suppressing tumor growth and progression in animal tumor studies, underscoring their therapeutic use for treating cancers. For example, delivery of let-7b or let-7g either by adenoviral infection or by nanoparticles successfully inhibited lung tumor progression in both xenograft and transgenic tumor mouse models (Kasinski et al., 2015;Kumar et al., 2008;Trang et al., 2010). In comparison with let-7b and let-7g, the therapeutic potential of let-7i in cancers is largely unexplored. Our study identified ERK3 as a new target of let-7i and reveals that the let-7i/ERK3 axis plays an important role in controlling head and neck cancer cell motility. Interestingly, the let-7i/ERK3 axis is tightly controlled by BMI1, an oncoprotein well known to be important for tumor growth and recurrence, as well as metastasis. Hence, our study not only provides better understanding of let-7i's tumor suppressing roles and the underlying mechanisms but also substantiates the importance of targeting BMI1 and let-7i for treating cancers. In addition, our findings suggest that ERK3 kinase is a potential new therapeutic target of cancers, particularly those with BMI1 overexpression.

BMI1 let-7i
Cancer cell migration Fig. 7. A schematic model for the BMI1/let-7i/ERK3 pathway in controlling cancer cell migration. This model presents the regulation of ERK3 expression by BMI1 and let-7i: ERK3 mRNA is a direct target of let-7i miRNA; elevation of BMI1 (indicated by the upward arrow) suppresses let-7i (Chou et al., 2013;Yang et al., 2012) and upregulates ERK3, leading to increase in head and neck cancer cell migration.