KRAS/NF-κB/YY1/miR-489 Signaling Axis Controls Pancreatic Cancer Metastasis

Pancreatic cancer (PDAC) is one of the most lethal malignant tumors, with a 5-year survival rate less than 8% from 2005 to 2011 in United States (1). To date, lack of effective screening tool to detect asymptomatic premalignant or early-stage cancer results in the diagnosis of majority patients at their advanced stages. More than 90% of the patients with PDAC show distal metastasis at advanced stage, which is the major cause of mortality (2). Thus, effective therapeutic interventions to stop PDAC metastasis are urgently needed. However, our ability to design effective therapeutic interventions aiming to stop PDAC metastasis is limited because of our poor understanding of the molecular mechanisms underlying pancreatic cancer metastasis (3).

One prominent feature of PDAC is the high frequency (>90%) of KRAS mutations that generate oncogenic forms of KRAS in pancreatic tumors, the most common of which is KRAS^G12D (4). Mounting evidence indicates that such oncogenic mutations play critical roles in both the initiation and the progression of pancreatic cancer via persistent activation of KRAS signaling pathways (5). Sustained KRAS signaling leads to the activation of inflammatory signaling pathways that play critical roles in regulating the initiation of pancreatic intraepithelial neoplasia (PanIN) and the progression of PDAC (6, 7). Of note, NF-κB signaling, a major pathway that connects inflammation with cancers, is activated by KRAS signaling and has been shown to promote pancreatic cancer progression in animal models (8–10). However, the downstream targets of NF-κB signaling that are directly involved in pancreatic cancer progression and metastasis still need to be defined.

microRNAs (miRNA) are a class of small, noncoding RNAs that negatively regulate protein-coding genes at the posttranscriptional level and are involved in virtually all types of carcinogenesis (11). In particular, a number of miRNAs have been found to be dysregulated in PDAC and involved in PDAC carcinogenesis (12). Nevertheless, how these miRNAs are dysregulated in PDAC and how they interact with key regulatory signaling molecules in pancreatic cancer initiation and progression are questions that still remain largely unanswered. Most importantly, despite the well-established role of oncogenic KRAS in PDAC, how KRAS signaling engages miRNAs to drive pancreatic cancer metastasis remains unknown.

In our attempt to identify important miRNAs involved in pancreatic carcinogenesis, we identified KRAS signaling–repressed miR-489 as part of an important mechanism underpinning oncogenic KRAS-induced PDAC migration and metastasis. Oncogenic KRAS signaling activates NF-κB, leading to enhanced expression of YY1, the transcription factor that directly suppresses MIR489 transcription. Our results showed that miR-489 decreases the migration of PDAC cells in cell cultures and inhibits lung and liver metastatic colonization of PDAC cells in mice but contributes little to cell proliferation and anchorage-independent growth. Mechanistically, we identified 2 metalloproteinase genes, ADAM9 and MMP7, as novel targets of miR-489 that mediate its antimetastatic effect in these cells. Collectively, our findings not only provide new mechanistic insights into how oncogenic KRAS-induced inflammatory signaling promotes PDAC metastasis but also indicate that miR-489 is a robust inhibitor of metastasis and a potential therapeutic target for treating PDAC.

The human PDAC cell lines BxPC-3 and PANC-1 cells were obtained from ATCC (May, 2011) and cultured according to their guidelines. All the cell lines were mycoplasma-free and recently authenticated by cellular morphology and the STR analysis at Ji-Ying Inc. (January 2014 and May 2016) according to the guidelines from ATCC (13). The 2 cell lines stably expressing luciferase (PANC-1-luc and BxPC-3-luc) were generated from their parental cell lines as previously reported (14).

The antibodies used in this study included: anti-KRAS (60309-1; Proteintech), anti-PCNA (#2586; Cell Signaling Technology), anti-YY1 (ab38422; Abcam), anti-ETS-1 (sc-350; Santa Cruz Biotechnology), anti-p65 (10745-1-AP, Proteintech), anti-ADAM9 (ab186833, Abcam), anti-MMP7 (10374-2-AP, Proteintech), anti-β-actin (A3854; Sigma-Aldrich), anti-mouse secondary antibody (A9044, Sigma-Aldrich), and anti-rabbit secondary antibody (A9169, Sigma-Aldrich). These antibodies were diluted according to the manufacturers' instructions. The NF-κB signaling inhibitor BAY 11-7082 (S2913; Selleck) was first diluted in cancer cell culture medium and then used as in previous studies with some modifications (i.e., incubation at 5 μmol/L for 24 hours; ref. 15). The sequences of chemically synthesized DNA and RNA oligonucleotides are listed in Supplementary Table S1.

The assays were performed as described previously (16). In brief, cancer cells were subjected to transfection for 24 hours. For cell proliferation assay, 3,000 viable cells were seeded into each well of 96-well plates. Cell growth was determined by MTT assays and verified by counting the cells excluding trypan blue. For soft agar colony formation assay, 5,000 viable cancer cells were triply seeded in 1.5 mL of tissue culture medium with 1% glutamine and antibiotics, and 0.4% soft agar was layered onto 0.8% solidified agar in tissue culture medium in 6-well plates. After incubation for 15 days, the colony foci were stained with 0.005% crystal violet and counted using a dissecting microscope. Experiments were carried out in triplicate.

Wound-healing assays were performed as previously described (14). Cancer cells were transfected for 24 hours and starved in culture medium containing 0.2% FBS for 12 hours. The monolayer of confluent cells was scratched using a 10-μL Pipetman tip. Cells were then photographed at different time points with an Olympus IX81 microscope. The relative wound areas were then measured using ImageJ software (NIH, Bethesda, MD).

Cancer cells were transfected for 24 hours, following by starvation for 12 hours. PANC-1 cells (2 × 10⁴) or BxPC-3 cells (5 × 10⁴) were used in each 20% (in culture medium without FBS) Matrigel-precoated (356231BD; BD Biosciences) insert (8.0-μm pore size inserts, 3422; Corning) in the Transwell assays according to the manufacturer's instructions. The nonmigrating cells were detached, and the migrating cells were stained with 4′,6-diamidio-2-phenylindole (DAPI) and counted after 24 hours using a fluorescence microscope. All fields were selected in a blind manner.

The 3D cell culture assay was performed as recently described (17). Transfected and viable cancer cells (1,000) in culture medium containing 2% Matrigel were seeded on the solidified 100% Matrigel-precoated chamber (177402; Nunc). The sphere morphology was observed with a microscope after 19 days.

NOD/SCID mice were purchased from SLAC Corporation and were housed under standard housing conditions at the animal facilities in the Institute of Biochemistry and Cell Biology, Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences. The lung and liver metastasis assay was performed as described previously (14, 18). PANC-1-luc cells (4 × 10⁴) or BxPC-3-luc cells (1 × 10⁵) in 100 μL Dulbecco PBS (D-PBS) were injected into 6- to 8-week NOD/SCID mice through the tail vein (n = 5–6), and the luciferase activity in lungs was analyzed 50 days later. PANC-1-luc cells (1 × 10⁵) orBxPC-3-luc cells (5 × 10⁵) in 50 μL D-PBS were injected into the spleens of 6- to 8-week-old NOD/SCID mice (n = 4–5), and the tumor numbers in livers were analyzed 60 days later. For bioluminescence imaging, mice were given luciferin 5 minutes before imaging and were then anesthetized (3% isoflurane). Luminescence imaging was performed and analyzed using the Xenogen IVIS Imaging System (Xenogen, a subsidiary of Caliper Life Sciences).

All results are presented as the mean ± SD. The Student t test was performed to judge the significance of differences between treated groups and their paired controls. P values indicate the level of statistical significance (*, P < 0.05; **, P < 0.01; ***, P < 0.001). The fold changes of the relative miRNA or mRNA levels were performed using Mev-HCL software algorithm analyses (19).

PDAC specimens and paired normal adjacent tissues were collected during surgery from Zhong Shan Hospital, which is affiliated with Fudan University (Shanghai, China) with written informed consent from patients. Samples were immediately snap-frozen and stored at −80°C. The specimen collection was approved by the Medical Ethical Committee of the hospital. All animal experiments were performed under protocols approved by the Institute of Biochemistry and Cell Biology, Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences and in accordance with the Guide for the Care and Use of Laboratory Animals (NIH publication nos. 80–23, revised 1996).

To gain new insight into the role of miRNAs in oncogenic KRAS-initiated pancreatic tumorigenesis, we compared the expression of 53 miRNAs that are dysregulated during PDAC progression in the Kras^G12Dtransgenic animal model and in KRAS^G12D-transfected human pancreatic ductal epithelial cells (20, 21) between 2 human PDAC cell lines, namely, wild-type KRAS-containing BxPC-3 and KRAS^G12D-containing PANC-1 cells (Supplementary Fig. S1A, left). Our qRT-PCR assays showed that 17 miRNAs were differentially expressed between the 2 PDAC cell lines (Supplementary Fig. S1B, left column), including 3 miRNAs (miR-489, miR-889, and miR-574) enriched in BxPC-3 cells and 4 miRNAs (miR-125b, miR-491, miR-21, and miR-23b) enriched in PANC-1 cells (by >3.0-fold). To examine the effect of oncogenic KRAS signaling on miRNA expression in these PDAC cells, we knocked down KRAS in PANC-1 cells and ectopically expressed KRAS^G12D in BxPC-3 cells, respectively (Supplementary Fig. S1C). miR-21 was the most markedly upregulated by KRAS^G12D overexpression in BxPC-3 cells (Supplementary Fig. S1B, middle column) and downregulated by KRAS knockdown in PANC-1 cells (right column), consistent with previous studies reporting an important role for miR-21 in oncogenic KRAS–induced cell proliferation (22, 23). Interestingly, miR-489, among all the tested miRNAs, showed the greatest reduction in KRAS^G12D-transfected BxPC-3 cells (Supplementary Fig. S1B, middle column) and the greatest increase in KRAS siRNA–transfected PANC-1 cells (right column).

The strong negative correlation between KRAS signaling and miR-489 prompted us to examine whether downregulation of miR-489 is involved in KRAS signaling–mediated pancreatic tumorigenesis. As expected, ectopic expression of KRAS^G12D in BxPC-3 cells substantially increased cell proliferation (Fig. 1A), anchorage-independent growth (Fig. 1B), and cell migration in vitro (Fig. 1C and D). Intriguingly, restoration of miR-489 expression in these cells by transfection of the miR-489 expression vector pSIF-miR-489 barely affected cell proliferation and soft agar colony formation (Fig. 1A and B; Supplementary Fig. S1D) but rescued the KRAS^G12Dexpression–induced cell migration in a dosage-dependent manner (Fig. 1C and D). Conversely, knockdown of KRAS in PANC-1 cells markedly decreased cell proliferation (Fig. 1E), anchorage-independent growth (Fig. 1F), and cell migration in vitro (Fig. 1G and H), whereas concurrent suppression of miR-489 by anti-miR-489 showed no significant effect on cell proliferation (Fig. 1E and F; Supplementary Fig. S1E) but significantly reversed the impact of KRAS knockdown on cell migration in these KRAS-mutant cells (Fig. 1G and H). These results together suggest that downregulation of miR-489 contributes little to KRAS signaling–driven cell proliferation but is an important mechanism underlying KRAS signaling–driven cell migration of PDAC cells.

We next asked how KRAS signaling downregulates miR-489 in PDAC cells. To this end, we first searched for potential transcription factor–binding sites in MIR489 promoter using the PROMO and TransFac programs (24, 25). Interestingly, ETS1 and YY1, 2 transcription factors that have been shown to control miRNA expression in cancer cells (26, 27), were predicted as candidate transcription factors for MIR489 (Fig. 2A, top). To explore whether ETS1 or YY1 directly regulates MIR489, we first performed a chromatin immunoprecipitation (ChIP) assay in PANC-1 cells. We found that the P_MIR489(promoter of MIR489) fragment was effectively enriched by anti-YY1, but not by anti-ETS1, compared with the IgG control (Fig. 2B). Moreover, RNAi knockdown of YY1 strongly enhanced miR-489 expression in these cells, whereas knockdown of ETS1 had little effect (Fig. 2C). We further found that overexpression of YY1, but not ETS1, potently decreased miR-489 expression in PANC-1 cells (Fig. 2D). These results strongly suggest that YY1 is a repressor for MIR489. To further substantiate this conclusion, we constructed 2 luciferase reporters under control of either the wild-type human MIR489 promoter or a mutant version with 2 putative YY1-binding sites deleted (termed the WT P_MIR489 reporter and the mut P_MIR489 reporter, respectively; Fig. 2A, bottom). As expected, the activity of the WT P_MIR489 reporter was potently enhanced by knockdown of YY1 but reduced by YY1 overexpression in PANC-1 cells (Fig. 2E). The mut P_MIR489 reporter displayed about 4-fold activity in PANC-1 cells compared with WT, and it was not affected by either YY1 knockdown or overexpression (Fig. 2E). Collectively, these results support the notion that YY1 is a direct transcriptional repressor for MIR489.

We next examined whether KRAS signaling represses MIR489 via YY1 in PDAC cells. The level of YY1 protein was significantly lower in wild-type KRAS-containing BxPC-3 cells than in KRAS^G12D-containing PANC-1 cells (Fig. 2F, left, lane 1 vs. lane 4). Interestingly, YY1 mRNA and protein were markedly increased by ectopic expression of KRAS^G12D in BxPC-3 cells (Fig. 2F, left, lane 6) and significantly reduced by KRAS knockdown in PANC-1 cells (Fig. 2F, left, lane 3), indicating that oncogenic KRAS activates YY1 expression in PDAC cells. Then, we searched and found that NF-κB family member was predicted to be transcription factor in YY1 promoter, which is consistent with the study in myoblasts (28). As NF-κB is a well-documented downstream effector of KRAS signaling (8, 10), we examined whether the YY1/miR-489 axis is regulated by KRAS/NF-κB signaling in human PDAC cells. Both KRAS knockdown and treatment with an NF-κB inhibitor, BAY 11-7082, significantly elevated miR-489 expression, with a concomitant reduction of YY1 expression in PANC-1 cells, whereas the elevation of miR-489 expression by KRAS knockdown or BAY 11-7082 treatment was completely suppressed by ectopic expression of Flag-YY1 in these cells (Fig. 2G). These results suggest that the YY1/miR-489 axis is controlled by KRAS/NF-κB signaling in PDAC cells. To further substantiate this conclusion, we examined the effect of KRAS knockdown or BAY 11-7082 treatment on the activity of the P_MIR489reporter in PANC-1 cells. In a similar manner, both KRAS knockdown and treatment with NF-κB inhibitor significantly enhanced the activity of the WT P_MIR489reporter in PANC-1 cells, and their effects were completely abrogated by ectopic expression of Flag-YY1 (Fig. 2H). In sharp contrast, KRAS knockdown or BAY 11-7082 treatment barely affected the activity of the mut P_MIR489reporter (Fig. 2H). Taken together, these results indicate that the MIR489 gene is repressed by the novel KRAS/NF-κB/YY1 axis in human PDAC cells.

We next examined the role of miR-489 in PDAC tumorigenesis. Modulation of miR-489 expression in PDAC cells, including transfection of pSIF-miR-489 in PANC-1 cells (which have lower endogenous miR-489 expression) or anti-miR-489 in BxPC-3 cells (which show a high level of endogenous miR-489 expression; Supplementary Fig. S1A, right), barely altered cell proliferation and anchorage-independent growth (Supplementary Fig. S2). Western blot analyses confirmed that the proliferation marker proliferating cell nuclear antigen (PCNA) was little altered in pSIF-miR-489–transfected PANC-1 or anti-miR-489–transfected BxPC-3 cells (Supplementary Fig. S2, middle). These results together demonstrate that miR-489 does not affect the proliferation of PDAC cells. We further examined the effect of miR-489 on PDAC cell migration and metastasis. In vitro studies showed that miR-489 overexpression in PANC-1 cells significantly reduced cell migration in both wound-healing (Fig. 3A, top) and Transwell migration assays (Fig. 3B, top) and markedly suppressed the ability of these cells to spread out and form colonies in a 3-dimensional (3D) cell culture assay (Fig. 3C, top). To determine the function of miR-489 on PDAC metastasis in vivo, we infected firefly luciferase–labeled PANC-1 (PANC-1-luc) cells with pSIF-miR-489 pseudovirus and performed tail vein xenografts. We observed strong luciferase foci in the lungs of mice injected with control pSIF-H1 pseudovirus–infected cells but a dramatic reduction in the luciferase signal in the lungs of mice injected with miR-489–overexpressing cells (Fig. 3D, top), suggesting that miR-489 inhibits the lung metastasis of PDAC cells. Hematoxylin and eosin (H&E) staining of the lung sections confirmed that miR-489–overexpressing cells colonized the lung much less efficiently than the control populations (Fig. 3E, top). We next injected pSIF-miR-489 pseudovirus–infected PANC-1-luc cells into mouse spleens and examined liver metastasis. Compared with controls, miR-489 overexpression substantially reduced the liver metastatic colonization of PDAC cells (Fig. 3F, top) and resulted in fewer metastatic liver nodules (Fig. 3G, top). In a reciprocal experiment, inhibition of miR-489 by anti-miR-489 in BxPC-3-luc (firefly luciferase–labeled BxPC-3) cells, which have a higher endogenous miR-489 expression (Supplementary Fig. S1A, right), led to a significant increase in cell migration in wound healing (Fig. 3A, bottom) and Transwell migration (Fig. 3B, bottom), cell invasion in 3D cell culture (Fig. 3C, bottom), and lung and liver metastasis (Fig. 3D–G, bottom). These results together indicate that miR-489 inhibits the migration and metastasis of PDAC cells, supporting that miR-489 is a critical regulator of PDAC metastasis.

To dissect the molecular mechanism of the inhibitory effect of miR-489 on PDAC metastasis, we used miRNA target prediction tools (29) to search for potential target genes of miR-489. We found that ADAM9 and MMP7, 2 metalloproteinase genes that have been implicated as being crucial to PDAC progression (30, 31), were predicted to be targets of miR-489 (Fig. 4A, top). To experimentally test whether miR-489 targets ADAM9 and MMP7, we constructed luciferase reporters by cloning the wild-type 3′-untranslated regions (UTR) of ADAM9 and MMP7 or their mutant versions (with deletion of the 7-bp sequence complementary to the 5′ sequence of miR-489) downstream of the firefly luciferase cDNA in the pmirGLO vector (Fig. 4A, bottom). We found that cotransfection of pSIF-miR-489 into 293T cells substantially decreased the luciferase activity of the wild-type reporters but barely affected that of the mutant reporters (Fig. 4B), suggesting that ADAM9 and MMP7 are targets of miR-489. To further corroborate this conclusion, we transfected pSIF-miR-489 into PANC-1 cells and found that both ADAM9 and MMP7 proteins were greatly reduced in pSIF-miR-489–transfected cells (Fig. 4C, left, lane 3). qRT-PCR analyses showed that their mRNA levels were also significantly reduced in PANC-1 cells transfected with pSIF-miR-489 (Fig. 4C, right). In contrast, inhibition of miR-489 by anti-miR-489 in BxPC-3 cells led to enhanced ADAM9 and MMP7 expression (Fig. 4D). These results together confirmed ADAM9 and MMP7 as authentic targets of miR-489. In addition, depleting YY1 or KRAS proteins in PANC-1 cells significantly reduced ADAM9 and MMP7 expression, whereas inhibition of miR-489 by anti-miR-489 effectively increased their expression in these cells (Supplementary Fig. S3). This result is consistent with our above finding that the KRAS/NF-κB/YY1 axis is required for repressing MIR489 (Fig. 2), further supporting the notion that ADAM9 and MMP7 are direct targets of miR-489 in human PDAC cells.

As miR-489 inhibits the migration and metastasis of PDAC cells (Fig. 3), we then examined whether miR-489 exerts its inhibitory effects on PDAC metastasis by targeting ADAM9 and MMP7. Intriguingly, either ADAM9 or MMP7 knockdown by shRNA in PANC-1 cells barely altered cell proliferation and anchorage-independent growth (Supplementary Fig. S4A and S4B) but robustly decreased cell migration in wound-healing, Transwell migration, and 3D cell culture migration assays (Supplementary Fig. S4C–S4E) and significantly suppressed both lung and liver metastases in NOD/SCID mice (Supplementary Fig. S4F–S4I). Interestingly, we found that either ADAM9 inhibitor SI-27 (32, 33) or MMP7 inhibitor sulfur-2-(4-chlorine-3-trifluoromethyl phenyl)-sulfonamido-4-phenylbutyric acid (SCTPSPA; ref. 34) substantially suppressed the lung metastasis of PANC-1-luc cells (Supplementary Fig. S4J and S4K). These results indicate that inhibition of ADAM9 and MMP7 recapitulates the inhibitory effect of miR-489 on the migration and metastasis of PDAC cells. To verify the functional role of the miR-489:ADAM9/MMP7 regulatory axis in PDAC migration and metastasis, we constructed miR-489–resistant expression vectors (pCMV-myc-ADAM9 and pCMV-myc-MMP7, without their 3′UTRs) for ectopic expression of ADAM9 and MMP7 proteins. Cotransfection of these vectors in PANC-1 cells barely altered cell proliferation and anchorage-independent growth (Supplementary Fig. S5) but largely rescued the impact of miR-489 on cell migration in vitro as well as lung and liver metastases in NOD/SCID mice (Fig. 5), supporting the notion that targeting ADAM9 and MMP7 is an important mechanism of miR-489–mediated suppression of cell migration and metastasis of PDAC cells. Collectively, these results indicate that the miR-489:ADAM9/MMP7 axis is of functional importance in regulating PDAC metastasis.

To test whether our above findings in PDAC cells are clinically relevant, we examined the miR-489 level as well as the levels of YY1, ADAM9, and MMP7 mRNAs in 30 human primary PDAC tumors from nonmetastatic and metastatic patients, respectively. We found that compared with paired adjacent normal tissue, the miR-489 level was reduced in primary tumors from nonmetastatic patients and further reduced in those from metastatic patients, whereas YY1, ADAM9, and MMP7 mRNAs were all significantly upregulated in nonmetastatic primary tumors and further upregulated in metastatic primary tumors (Fig. 6A). Importantly, we found a significant inverse correlation between miR-489 and YY1 mRNA levels (Pearson R = −0.797, P < 0.001), ADAM9 mRNA levels (Pearson R = −0.804, P < 0.001), or MMP7 mRNA levels (Pearson R = −0.818, P < 0.001; Fig. 6B). We further found that PDAC tumors with wild-type KRAS showed greater miR-489 expression and diminished YY1, ADAM9, and MMP7 immunohistochemical staining (Fig. 6C, left; n = 10), whereas tumors with mutant KRAS showed a significantly lower level of miR-489 expression and more intense YY1, ADAM9, and MMP7 staining (Fig. 6C, right, n = 10; Supplementary Fig. S6). Collectively, these results strongly suggest that the newly discovered YY1/miR-489:ADAM9/MMP7 regulatory axis is clinically relevant in human PDAC.

KRAS mutations occur with a frequency of up to 90% in pancreatic cancer and as early as the PanIN stage (4). Mutant KRAS, alone or in conjunction with other mutant genes such as p53 or p16, drives initiation of pancreatic cancer by inducing acinar-to-ductal transformation and maintaining progression to metastasis (5, 35). Moreover, both basic and clinical studies have pointed out that many risk factors influence PDAC progression by inducing inflammation, which accompanies cancer from initiation to metastasis (36).NF-κB signaling has been recognized as one of the most important pathways that link inflammation to cancer, including PDAC, which confers a reciprocal interaction between inflammation and tumor progression (37, 38). The upregulation of p65, the most common NF-κB family member, greatly induces the constitutive activation of NF-κB signaling and promotes PDAC progression into metastasis, but the key downstream molecular events still remain largely elusive (8, 10, 39). In the present study, we established miR-489 as a key regulator of PDAC metastasis, acting downstream of KRAS/NF-κB signaling (Fig. 6D). The importance of miR-489 in mediating PDAC metastasis was supported by multiple lines of evidence, including in vitro cell migration and invasion assays, in vivo xenograft experiments, and human PDAC specimens.

We found for the first time that the transcription factor YY1 acts downstream of KRAS/NF-κB signaling to suppress the expression of miR-489. YY1 is a ubiquitously distributed transcription factor that directly binds to the promoter region of its target genes (40). Consistently, our results showed that YY1 inhibits miR-489 expression by binding to 2 closely spaced sites that are located near the MIR489 transcription start site and that this negative regulation greatly attenuates the expression of miR-489, thus promoting PDAC metastasis. These findings enriched our understanding of the mechanism of KRAS signaling and YY1 function in cancer development, especially in PDAC metastasis progression. Previous studies have shown that YY1 acts as an oncogene to promote cancer development by regulating the cell cycle, cell apoptosis, chemoresistance, inflammation, and cell invasion (40, 41), which is in keeping with our finding of upregulation of YY1 and YY1 downregulation of miR-489 to promote PDAC cell metastasis. In addition, dysregulation of YY1 in PDAC and several other types of cancers has been correlated with the poor prognosis of patients (41–44).

The important role of miRNAs in tumorigenesis is well established. An increasing body of evidence has demonstrated that miRNAs control nearly all aspects of tumorigenesis, including cancer cell proliferation, apoptosis, and metastasis (11). Here, we uncovered the role of miR-489 as a critical suppressor of PDAC metastasis, which is consistent with the global miRNAs profiling data from a previous study showing a strong negative correlation of miR-489 expression in human PDAC tumors with liver metastasis (45). In addition, miR-489 has been found to be downregulated in breast, ovarian, and non–small cell lung cancers (46–48). Of note, upregulation of miR-21 by oncogenic KRAS have been shown to promote pancreatic tumor growth (22). Interestingly, our findings show that oncogenic KRAS signaling employs miR-489 to specifically promote metastasis. More importantly, we found that miR-489 expression was well correlated with the progression of PDAC, which is consistent with the report that miR-489 may be a useful biomarker for PDAC clinicopathologic parameters (45). Moreover, in light of the role of miR-489 in PDAC metastasis, this miRNA is also a potential candidate for designing novel therapeutics targeting the metastasis of PDAC.

We went further to show that miR-489 negatively regulates PDAC metastasis by targeting 2 key metastasis mediators, ADAM9 and MMP7. These genes encode metalloproteinases that remodel the extracellular matrix, thereby facilitating cancer cells in forming a local or distant metastasis and are highly expressed in many kinds of cancers, including PDAC (49, 50). Given the important role of ADAM9 and MMP7 in regulating metastasis, the regulation of these 2 important metalloproteinases by the KRAS/NF-κB/YY1/miR-489 axis may be critical to PDAC metastasis.

In summary, our study defined a regulatory axis centered on miR-489 that plays an important role in oncogenic KRAS-driven PDAC metastasis. Mechanistically, miR-489 targets 2 metalloproteinase genes, ADAM9 and MMP7, to inhibit metastasis. Oncogenic KRAS signaling activates inflammatory NF-κB signaling to upregulate YY1, the transcriptional repressor of MIR489, thus attenuating the level of miR-489 and thereby promoting tumor progression and metastasis. This study not only identified an miRNA-based mechanism underlying oncogenic KRAS–associated inflammation that promotes PDAC metastasis but also provided a potential candidate molecule with which to design oligonucleotide drugs targeting PDAC metastasis.

No potential conflicts of interest were disclosed.

Conception and design: P. Yuan, Y. Li, W. Lou, M.-F. Liu

Development of methodology: P. Yuan, X.-H. He, Y.-F. Rong, W. Lou, M.-F. Liu

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): P. Yuan, X.-H. He, Y.-F. Rong, J. Cao, Y. Hu

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): P. Yuan, X.-H. He, Y.-F. Rong, Y.-P. Hu, M.-F. Liu

Writing, review, and/or revision of the manuscript: P. Yuan, Y. Li, Y. Liu, D. Li, W. Lou, M.-F. Liu

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): Y. Liu

Study supervision: W. Lou, M.-F. Liu

We thank Prof. Xinyuan Liu for pGL3-basic and pmirGLO vectors and Prof. Ying Yu for help in animal luminescence imaging. The authors are grateful to Dr. Yonggang Zheng for his critical reading of this article.

This work was supported by grants from the National Natural Science Foundation of China (31325008, 91640201, 91419307, 31300656, and 31270840), Ministry of Science and Technology of China (2014CB943103, 2014CB964802, and 2012CB910803), Science and Technology Commission of Shanghai Municipality (13ZR1464300 and 16XD1404900), and Chinese Academy of Sciences (KJZD-EW-L01-2, 2013KIP202, and "Strategic Priority Research Program" Grant XDB19010202).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.