Loss of miR125a Expression in a Model of K-ras–Dependent Pulmonary Premalignancy

Lung cancer is the leading cause of cancer-related mortality in the United States, accounting for more deaths than prostate, breast, colon, and pancreatic cancers combined (1). The high mortality rate is largely due to the late stage at which lung cancer is diagnosed, underscoring the need for identification of biomarkers for early lung cancer development and more effective lung cancer chemoprevention options (2). Although early detection has been the focus of research efforts resulting in recent notable achievements (3), lung cancer chemoprevention has lagged behind (4). Despite extensive efforts, agents evaluated in the majority of lung cancer chemoprevention phase III clinical trials have been found to be ineffective or even harmful (5). Thus, these guidelines do not recommend agents for lung cancer chemoprevention (6). Major clinical breakthroughs in advanced-stage lung cancer have been facilitated by the recent advent of patient selection based upon tumor mutational profiles that have fostered a personalized medicine approach for patients with non–small cell lung cancer (NSCLC; ref. 7). This paradigm shift has not yet fully reached clinical lung cancer chemoprevention; however, there is increasing recognition that the step-wise molecular alterations precipitating lung cancer development must be better defined to facilitate development of: (i) risk assessment targets, (ii) biomarkers for patient selection, and (iii) outcome prediction for specific agents (6, 8, 9).

miRNA are small, noncoding RNAs that inhibit expression of specific genes at the posttranscriptional level by binding to the 3′ untranslated region of their target mRNAs, causing suppression of translation or promotion of mRNA degradation (10). They have emerged as key regulators of proliferation, differentiation, and apoptosis, processes commonly altered in carcinogenesis (11). Indeed, a rapidly developing literature supports an important role for miRNA in the regulation of lung carcinogenesis and tumor progression (12, 13). Importantly, miRNAs are frequently dysregulated in cancer and have differential expression in tumors compared with normal tissues (14). Differences in miRNA expression patterns can also indicate prognosis. For example, expression of let-7a and let-7f miRNA is prognostic for survival of patients with lung cancer (15). Furthermore, the stability of miRNAs renders them attractive candidates for robust, noninvasive biomarkers for the early detection of lung cancer. The following studies arose from our search for mutation-driven alterations in miRNAs that are potential biomarkers in the lung “at-risk” for the development of lung cancer.

miR125a-3p and miR125a-5p are two mature isoforms of miR125a, derived from the 3′ and 5′ ends of pre-miR125a, respectively. miR125a is located at chromosome 19q13, a chromosomal region frequently deleted in human cancer. Several studies support the importance of miR125a in various types of cancer. Overexpression of miR125a reduces migration, invasion, and AIG of breast cancer cells, and a germline mutation of miR125a is highly associated with breast cancer tumorigenesis (16, 17). In hepatocellular carcinoma (HCC) tissues, expression of miR125a is frequently downregulated compared with matched adjacent liver tissue, and lower expression of miR125a inversely correlates with aggressiveness and poor prognosis of patients with HCC (18). Furthermore, overexpression of miR125a inhibits proliferation, migration, and invasion of HCC cells both in vitro and in vivo (18). In patients with oral squamous cell carcinoma, a tobacco-related cancer, salivary levels of miR125a are lower compared with healthy controls, suggesting its utility as a biomarker for oral cancer (19). Importantly, miR125a is also downregulated in NSCLC tissues compared with corresponding adjacent normal lung tissue, and expression of miR125a negatively correlates with pathologic tumor stage and lymph node metastasis (20). In lung cancer cell lines, a gain-of-function of miR125a inhibits cell migration, invasion, and proliferation, and induces apoptosis in vitro (20, 21). Conversely, miR125a loss-of-function enhances lung cancer cell growth, angiogenesis, invasion, and migration. Despite this evidence supporting the importance of miR125a in late-stage cancer, its role in premalignancy, particularly pulmonary carcinogenesis, has yet to be defined.

To model the early pathogenesis of lung cancer, we used the in vitro HBEC model system (22, 23). These cells were immortalized in the absence of viral oncoproteins via ectopic expression of human telomerase (hTERT) and cyclin-dependent kinase 4 (CDK4). In contrast with cancer cell lines with a mixed genetic background, HBECs are a unique in vitro model of lung carcinogenesis that allows evaluation of pathway intermediaries in isolation. These immortalized HBECs are partially progressed lung epithelial cells that can differentiate into mature airway cells in organotypic cultures, but do not have a fully transformed phenotype (i.e., do not form tumors in immunodeficient mice), making them a physiologically appropriate preclinical model of lung premalignancy.

Here, we report that loss of miR125a-3p (hereafter referred to as miR125a) expression contributes to the malignant potential of HBECs harboring an activating point-mutation of the K-ras protooncogene, one of the most clinically challenging genetic changes commonly found in current and former smokers. Compared with the vector control, reduced expression of miR125a in K-ras– and/or K-ras/P53–mutated HBECs was associated with increased levels of VEGF and CXCL1, increased proliferation, and enhanced AIG. Furthermore, we demonstrate pharmacologic augmentation of miR125a expression can abrogate several of the deleterious downstream events associated with K-ras mutation. Taken together, our findings suggest that loss of miR125a expression participates in the pathogenesis of pulmonary premalignancy.

Immortalized HBEC lines (obtained from Dr. John D. Minna, University of Texas Southwestern Medical Center, Dallas, TX) were established by introducing CDK4 and hTERT into normal HBECs isolated from large airways of patients (22, 23). Three parental cell lines derived from 3 patients, HBEC3, HBEC7, and HBEC11, were used. These HBECs were subsequently manipulated to stably express the vector control (HBEC Vector) or an activating point-mutation of the K-ras protooncogene (K-ras^V12; HBEC K-ras), alone or in combination with stable knockdown of the P53 tumor-suppressor gene (HBEC K-ras/P53), as previously described (22); each was generated in the UCLA Vector Core. A restriction-fragment-length polymorphism analysis of K-ras cDNA validated that mutant K-ras^V12 transcripts were predominately expressed in HBEC K-ras cells compared with the vector control. All cell lines were routinely tested for the presence of Mycoplasma using the MycoAlert Mycoplasma Detection Kit (Lonza). Cell lines were authenticated in the UCLA Genotyping and Sequencing Core using Promega's DNA IQ System and Powerplex 1.2 system. All cells were used within 10 passages of genotyping. HBECs were cultured in Keratinocyte serum-free media (Life Technologies) supplemented with 30 μg/mL bovine pituitary extract and 200 ng/mL recombinant epidermal growth factor 1–53 (Life Technologies). Cell cultures were grown in a 5% CO₂ atmosphere at 37°C. Pioglitazone (Enzo Life Sciences) was prepared in DMSO at a concentration of 10 mmol/L, and cells were treated at a final concentration of 10 μmol/L in supplement-free media. CXCL1 (PeproTech) was prepared in 0.1% bovine serum albumin (BSA) at a concentration of 10 ng/mL, and cells were treated at a final concentration of 10 pg/mL in complete media.

Total RNA (including miRNA) was isolated from HBEC K-ras and Vector cell lines using the miRNeasy Mini Kit (Qiagen). Expression of 667 miRNAs was evaluated using the TaqMan Human miRNA Array v.2 (Life Technologies). Briefly, 1 μg of total RNA was reverse transcribed using Megaplex Primer Pools A and B and the resulting cDNA for each sample was then loaded on two 384-well microfluidic array cards containing TaqMan primers and probes for miRNA and normalization controls followed by real-time PCR. The PCR data were normalized and analyzed with SDS v.2.3 software (Life Technologies) using the ΔΔC_t method. To accurately compare miRNA expression across the study samples, the same C_t threshold was used for all samples. Wells with C_t values ≥35 were counted as negative.

Cells were plated in 6-well plates at 1.5 × 10⁴ cells per well. For transient overexpression of miR125a, HBEC3 Vector, K-ras, and K-ras/P53 cells were transfected with the miRVana hsa-miR125a-3p mimic (Cat# 4464066; ID# MC12378) or negative control #1 (Cat# 4464058; both Life Technologies) at a final concentration of 10 nm, unless otherwise stated. For PEA3 knockdown, two different siRNA oligonucleotides specific for PEA3 were used: siRNA ID# s4816 (10 nmol/L) and siRNA ID# 115237 (50 nmol/L; both Life Technologies). HBEC3 K-ras cells were transfected with PEA3 siRNA or a nontargeting siRNA control. Transfections were carried out using the Lipofectamine RNAiMAX Transfection Reagent (Life Technologies) for 4 hours before replacement with complete media.

Total RNA was isolated using the miRNeasy Mini Kit, and cDNA for mRNA or miRNA analysis was prepared using the High Capacity RNA-to-cDNA Kit or the TaqMan miRNA Reverse Transcription Kit (both Life Technologies), respectively. Transcript levels of miR125a and PEA3 were measured by quantitative reverse transcription PCR (qRT-PCR) using the TaqMan Probe-based Gene Expression System (Life Technologies) in a MyiQ Cycler (Bio-Rad). Amplification was carried out for 40 cycles of 15 seconds at 95°C and 60 seconds at 60°C. All samples were run in triplicate, and relative gene/miRNA expression levels were determined by normalizing their expression to RNU6B (mir125a) or GUSB (PEA3). Expression data are presented as fold-change values relative to normalized expression levels in a reference sample using the following equation: $RQ\, = \,2^{ - \Delta \Delta C_{\rm t} }$⁠.

HBEC Vector and K-ras cells were plated in 6-well plates at 1.5 × 10⁴ cells per well. Cells were washed with ice-cold phosphate-buffered saline (PBS), and total cell lysates were collected using a modified radioimmunoprecipitation assay buffer. Protein concentrations were measured using the bicinchoninic acid (BCA) assay (Pierce Biotechnology). Proteins were resolved by 10% SDS–PAGE and transferred to polyvinylidene difluoride membranes. For detection of PEA3, membranes were blocked with 5% BSA in TBS with 0.05% Tween 20 and probed overnight with a PEA3 monoclonal antibody (1:100 dilution; Cat# sc-113; Santa Cruz Biotechnology). GAPDH was used as a loading control, and protein levels were detected using enhanced chemiluminescence reagent (Pierce Biotechnology). Densitometry was performed using ImageJ software (24).

As an indication of cell viability and proliferation, cellular ATP levels were measured using the ATPlite 1step Luminescence Assay Kit (Perkin Elmer). Cells were plated in 96-well plates at 1 × 10³ cells per well. Six replicates for each condition were plated for each independent experiment. For miR125a overexpression studies, HBEC3 Vector and K-ras cells were transfected with miR125a mimic or negative control (10 nmol/L) 24 hours before plating into 96-well plates. For pioglitazone studies, cells were treated with pioglitazone (10 μmol/L) or DMSO control diluted in 50% complete media + 50% supplement-free media. Treatment started following overnight attachment and continued for 48 hours. ATP luminescence was assessed at 0 and 48 hours; 48 hours readings were normalized to the 0 hour readings to control for plating differences.

We performed a fluorescence-based cytokine screen using the Luminex-based multiplex system from Bio-Rad. HBEC Vector and K-ras cells were cultured and grown to 30% to 40% confluency, then washed twice with PBS before fresh medium was added. Cells were incubated for an additional 48 hours, after which the supernatants were collected and centrifuged to remove contaminating cells. Lysates were prepared from the adherent cells, and the BCA assay was performed to determine protein concentration. A customized panel of seven human cytokine/growth factors (Bio-Rad) was used, and all samples were run in triplicate. Results were normalized to protein concentration from the matched lysates and then expressed as fold-change over unselected cell cytokine secretion (after normalization). We used ANOVA to compare cell populations for each analyte, and adjusted for multiple testing by computing the false-discovery rate using the q value method implemented in R. Cytokines differentially produced by the cells were verified by enzyme-linked immunosorbent assay (ELISA).

HBEC3 Vector, K-ras, or K-ras/P53 cells were plated in 6-well plates at 1.5 × 10⁴ cells per well. Following overnight attachment, complete media were removed, and the cells were cultured in supplement-free media. After 48 hours, supernatants were collected and protein levels of VEGF and CXCL1 were quantified using a DuoSet ELISA Development System (R&D Systems). For miR125 overexpression studies, cells were transfected with miR125a mimic or negative control (10 nmol/L) following overnight attachment. Supernatants were collected 48 hours post-transfection and assayed as above. For pioglitazone studies, treatment with pioglitazone (10 μmol/L) or the DMSO control started following overnight attachment. Supernatants were collected 48 hours after treatment initiation. Lysates were prepared from the adherent cells, and the BCA assay was performed to determine protein concentration. Results were normalized to protein concentration from the matched lysates.

We used a modified high-throughput cell transformation assay for evaluation of soft agar colony growth. Briefly, the cells are suspended in 0.4% agar and plated atop a thick layer of solidified 0.6% agar. HBEC Vector, K-ras, and K-ras/P53 cells were plated in 96-well plates at 1 × 10⁴ cells per well. For miR125a overexpression studies, HBEC3 K-ras/P53 cells were transfected with miR125a mimic or negative control (1 nmol/L) 24 hours before plating into 96-well plates. Twenty-four hours after plating, cells were treated with CXCL1 or vehicle (0.1% BSA) at a final concentration of 10 pg/mL in complete media for a total of 5 days. Media containing CXCL1 or vehicle were replaced on day 2, and colony formation was evaluated after 5 days in culture. Three-dimensional projection images (×100 total magnification) of the colonies in each sample were obtained by assembling Z-stacks comprised of 60 optical sections, at an interval of approximately 30 μm, using the Nikon Eclipse Ti microscope. The 60 images were then focused into a single representative image per well using the extended depth of focus function of the Nikon Elements AR software, and the total number of colonies in each focused image was manually counted. Twelve replicates for each condition were plated for each independent experiment.

Samples were plated/run in triplicate, unless otherwise indicated, and all experiments were performed at least three times. Data represent mean and SEM of one representative experiment, and one representative experiment/image is shown. Statistical significance was determined by the two-tailed, nonpaired Student t test, and data are reported significant as follows: *, P ≤ 0.05; **, P ≤ 0.01, and ***, P ≤ 0.001.

To determine the contribution of miRNAs to the malignancy-promoting effects of K-ras mutation, we characterized the miRNA expression profile in the K-ras–mutated HBEC3 cell line (HBEC3 K-ras) compared with its vector control (HBEC3 Vector) using the TaqMan Human miRNA Array v.2 that included 667 miRNAs. Expression levels of miRNAs that have known implications in cancer progression, including the let-7 family, the miR200 family, and miR34, are shown in Fig. 1A. Among these miRNAs, miR125a showed the most striking reduction in expression in K-ras–mutated HBEC3 cells compared with the vector control. We then validated the microarray data by TaqMan-based qRT-PCR in three sets of K-ras–mutated and Vector control HBEC lines, each of which was derived from an individual patient—HBEC3, HBEC7, and HBEC11. Compared with their respective vector controls, the basal expression of miR125a is significantly reduced in all three sets of premalignant K-ras–mutated HBECs, with a 3- to 4-fold reduction seen in HBEC3 and HBEC7 K-ras lines (Fig. 1B). Importantly, we observed a trend in reduction of miR125a expression in at least one premalignant atypical adenomatous hyperplasia lesion compared with normal tissue in 5 of 8 patient samples (Supplementary Table S1).

To define the mechanism underlying loss of miR125a expression in HBEC K-ras cells compared with Vector, we investigated the role of PEA3, a member of the erythroblast transformation specific (ETS) transcription factor family. The miR125a promoter contains five binding sites for PEA3, and chromatin immunoprecipitation has demonstrated that PEA3 association with the miR125a promoter suppresses its activity, consequently decreasing levels of miR125a in ovarian cancer cells (25). Importantly, PEA3 is upregulated in lung cancer and induced and activated by Ras mutation (26–28). Our data indicate that PEA3 expression (mRNA and protein levels) is abundant and elevated in HBEC3 K-ras and HBEC7 K-ras cells relative to their respective vector controls (Fig. 2A). To evaluate whether genetic silencing of PEA3 could restore the reduced expression of miR125a in K-ras–mutated HBECs, we transfected HBEC3 K-ras cells with siRNA specific for PEA3 or a nontargeting siRNA control. Knockdown of PEA3 mRNA and protein was confirmed by qRT-PCR and Western blot analysis, respectively (Fig. 2B). As seen in Fig. 2C, knockdown of PEA3 expression using two different siRNA oligonucleotides both yielded significantly increased expression of miR125a by 2-fold, suggesting that miR125a expression is regulated by the PEA3 transcription factor in K-ras–mutated HBECs.

To evaluate the endogenous role of miR125a in regulating tumorigenic phenotypes of the K-ras–mutated airway, HBEC3 Vector and K-ras cells were transfected with the miRVana hsa-miR125a-3p mimic or negative control (10 nmol/L). Forced expression of miR125a was confirmed by qRT-PCR (Fig. 3A). Compared with the negative control, transfection with the miR125a mimic led to a significant increase in levels of miR125a in both Vector and K-ras–mutated HBECs. To assess a functional consequence of miR125a loss, we measured proliferation of HBEC3 Vector and K-ras cells. At baseline, HBEC K-ras cells display a higher proliferation rate as compared with Vector (Fig. 3B). Interestingly, miR125a overexpression reduced this increased proliferation rate of HBEC K-ras cells to the level of Vector (Fig. 3B). To assess the impact of miR125a loss on tumor-promoting factors in K-ras–mutated HBECs, we collected cell culture supernatants from HBEC3 Vector and K-ras cells 48 hours post-transfection with the miR125a mimic or negative control and evaluated a panel of seven inflammatory and angiogenic proteins that we selected for their demonstrated importance to lung carcinogenesis. All seven proteins were measured within the same sample using a high-throughput Luminex-based multiplex assay. Although a number of proteins were differentially expressed, we focused on VEGF and CXCL1 to define mechanisms of miR125a that limit tumor growth, because computational prediction tools (TargetScan and mirRanda) identified VEGF, as well as positive regulatory factors of CXCL1 (e.g., cyclic AMP response element binding protein and poly ADP-ribose polymerase 1), as putative miR125a targets. Furthermore, murine models that develop ras-induced tumors have been shown to be dependent upon expression of VEGF and CXCL1, which recruit inflammatory cells and endothelial cells to the tumor (29). Data from the Luminex assay indicated that the basal protein expression of VEGF and CXCL1 is significantly higher in HBEC3 K-ras cells compared with Vector (Fig. 3C). These data were then confirmed by ELISAs specific for VEGF or CXCL1 (data not shown). The increased levels of these tumorigenic proteins are in accord with the cytokine and growth factor expression profiles previously reported for NSCLC cells both in vitro and in vivo (30). Importantly, overexpression of miR125a significantly reduced protein levels of these tumor-promoting factors in HBEC K-ras cells, whereas no changes were observed in Vector (Fig. 3C). Further studies indicated that the reduction in protein levels of CXCL1 is regulated at the mRNA level, and in agreement with others, luciferase assays confirmed that VEGF is a direct target gene of miR125a (data not shown; ref. 18).

To assess the effect of miR125a loss on malignant transformation in vitro, we evaluated AIG in soft agar assays as an index of transformed cell growth and tumorigenicity. Because loss of P53 function has been shown to be a critical co-oncogenic step in the malignant transformation of HBECs (22, 23), we used HBECs harboring an oncogenic K-ras mutation in combination with P53 knockdown (HBEC3 K-ras/P53) for the soft agar assays. As seen in Fig. 4A, basal expression of miR125a is significantly reduced in HBEC3 K-ras/P53 cells compared with the Vector control. In agreement with previous studies, HBEC3 K-ras/P53 cells display robust AIG, whereas Vector cells fail to form colonies in soft agar (data not shown; refs. 22, 23). To determine the contribution of miR125a loss to AIG of HBEC3 K-ras/P53 cells, we transfected HBEC3 K-ras/P53 cells with the miR125a mimic or negative control (1 nmol/L) 24 hours before plating into soft agar. Compared with the negative control, miR125a overexpression markedly decreased colony formation of the HBEC3 K-ras/P53 cells by 50% (Fig. 4C and D). We then assessed a potential role of CXCL1 in the miR125a-mediated reduction in AIG of HBEC3 K-ras/P53 cells, as CXCL1 has been reported to contribute to epithelial-to-mesenchymal transition, a process that has been implicated in transformation, as well as regulation of proteins involved in AIG (31). Furthermore, miR125a overexpression reduces protein levels of CXCL1 by 31% ± 3% in HBEC K-ras/P53 cells (Fig. 4B). As quantified in Fig. 4C and seen in Fig. 4D, exogenous CXCL1 treatment (10 pg/mL) restored colony formation of miR125a-transfected HBEC3 K-ras/P53 cells to the level of negative control-transfected cells, whereas no changes were observed with the 0.1% BSA vehicle control.

To illustrate the utility of miR125a as a potential pharma-cologic target, we determined the capacity of pioglitazone, a member of the thiazolidinedione (TZD) class of agents, to augment miR125a and limit premalignancy within the “at-risk” K-ras–mutated pulmonary microenvironment. Pioglitazone was selected as a model agent for our studies based on its ability to regulate miRNAs and its capacity to modulate mediators of malignancy, such as VEGF and CXCL1 (32–34). Moreover, a 33% reduction in lung cancer risk among TZD users compared with nonusers was observed, suggesting that TZDs may be chemopreventive for lung cancer in select patient populations (35). Expression of miR125a was determined in HBEC3 Vector and K-ras cells after 48 hours of treatment with pioglitazone (10 μmol/L) or the DMSO control. Interestingly, pioglitazone exposure increased miR125a expression by 3-fold in HBEC3 K-ras cells compared with the DMSO control, but did not alter levels of miR125a in Vector cells (Fig. 5A). Further studies determined that upregulation of miR125a expression by pioglitazone in HBEC3 K-ras cells is accompanied by a significant reduction in expression of the PEA3 transcription factor (Fig. 5B).

Because pioglitazone upregulates miR125a expression in HBEC3 K-ras cells, we determined whether pioglitazone could also limit tumorigenic phenotypes. Cell culture supernatants were collected from HBEC3 K-ras and Vector cells treated for 48 hours with pioglitazone (10 μmol/L) or the DMSO control, and VEGF protein levels were measured via ELISA. Similar to forced miR125a expression, pioglitazone significantly reduced VEGF protein levels by 38% ± 6% in K-ras–mutated HBECs, whereas no changes were observed in Vector cells (Fig. 6A). In addition, the increased proliferation rate of HBEC3 K-ras cells was suppressed to the same level as Vector cells by pioglitazone (Fig. 6B). Interestingly, 72 hours of pioglitazone treatment induced a phenotypic change indicative of mesenchymal-to-epithelial transition (MET) in K-ras–mutated HBECs, whereas no changes were observed in Vector; HBEC K-ras cells displayed changes from a spindle-like to a more cobblestone-like morphology, typical of epithelial cells (Fig. 6C).

The lack of a systematic approach for chemoprevention agent selection and the disregard for individual differences when randomizing to intervention groups have led to clinical prevention trials that have failed or, in some cases, increased the risk for lung cancer (2). In this regard, defining molecular abnormalities in the pulmonary epithelium of a given individual at-risk will facilitate rational application of chemoprevention agents with the greatest capacity to interrupt specific mutation-driven events in cancer initiation and progression. This personalized, targeted approach has been the recent focus in advanced-stage lung cancer, and is now expected to yield more effective cancer prevention. To this end, we have demonstrated that loss of miR125a expression in K-ras–mutated and/or K-ras/P53–mutated HBECs contributes to elevated levels of tumorigenic proteins, including VEGF and CXCL1, and increases proliferation as well as AIG (Fig. 6D). Furthermore, treatment of K-ras–mutated HBECs with a pharmacologic agent increases levels of miR125a by downregulating expression of the PEA3 transcription factor and abrogates several of the deleterious downstream events associated with the mutation. Importantly, recent studies have demonstrated that expression of miR125a is reduced in NSCLC tissues compared with adjacent normal lung tissues, and expression of miR125a negatively correlates with pathologic tumor stage and lymph node metastasis (20). These findings, in addition to our findings in the HBEC model of pulmonary premalignancy demonstrated here, suggest that loss of miR125a expression participates in the pathogenesis of pulmonary premalignancy.

The potential of miRNAs to serve as noninvasive biomarkers for the early diagnosis of cancer has been an area of active investigation. In a seminal study by Schembri and colleagues, the authors' findings indicate that miRNAs are regulators of smoking-induced gene expression changes in human bronchial airway epithelial cells, suggesting that miRNA may play a role in the pathogenesis of smoking-related lung diseases (13). Although a number of studies have since emerged identifying various miRNAs in lung carcinogenesis, ours is the first examination of miR125a in the context of pulmonary premalignancy. Moreover, recent findings suggest that miRNA expression measured in readily collected samples can be used for early cancer detection (12, 36, 37). For example, in a blood-based microarray profiling of miRNA expression in preoperative serum of patients with NSCLC and healthy individuals, miR361-3p and miR625* were identified as the two most downregulated miRNAs (38). It was further determined that both miRNAs could differentiate NSCLC from benign lesions and healthy controls. In next-generation sequencing of small RNA from human bronchial airway epithelium, miR4423 was identified as a regulator of airway epithelium differentiation and a repressor of lung carcinogenesis (12). Interestingly, expression of miR4423 is downregulated in the cytologically normal bronchial airway epithelium of smokers with lung cancer. This suggests that expression of miR4423 and/or other miRNAs might be influenced by a field cancerization effect and could be useful for the early detection of lung cancer in the relatively accessible proximal airway. Importantly, studies documenting the regulation of miRNAs by diverse classes of cancer chemoprevention agents highlight their utility in the clinical setting. In a miRNA expression profiling study of rat lungs exposed to environmental cigarette smoke (ECS), ECS induced downregulation of several families of miRNAs that regulate tumorigenic phenotypes, including apoptosis, angiogenesis, and proliferation (39). In a subsequent study, the authors demonstrated by microarray that orally administered chemopreventive agents could modulate ECS-induced alterations in miRNA expression, as well as tumorigenic phenotypes (40). As these studies highlight the potential use of miRNAs as biomarkers for NSCLC early detection and targets for NSCLC chemoprevention, our findings suggest that miR125a may have similar potential that is worthy of further investigation.

The peroxisome proliferator–activated receptor γ (PPARγ) has been identified as a potential target for lung cancer chemoprevention. Considerable evidence suggests that the TZD group of drugs, also known as PPARγ ligands, modulate cancer progression by inhibiting tumor initiation, progression, and metastasis. Using in vitro and in vivo xenograft models of lung cancer, TZDs have been shown to induce growth arrest, differentiation, and apoptotic cell death, thereby inhibiting tumor growth (41–48). Currently, the TZD pioglitazone is being evaluated in an early-phase chemoprevention clinical trial for subjects at elevated risk for lung cancer. To help define the role of miR125a in pulmonary premalignancy and demonstrate its potential as a pharmacologic target, we used the TZD pioglitazone as a model agent for our studies. Interestingly, pioglitazone selectively upregulated miR125a expression in HBEC K-ras cells through downregulation of the PEA3 transcription factor. In agreement with previous studies, pioglitazone exhibited antitumorigenic activity, including the reduction of proliferation and tumorigenic proteins. At baseline, HBECs are preneoplastic epithelial cells, but their epithelial/mesenchymal phenotype is highly plastic and shifts along the continuum as the HBECs are genetically manipulated and acquire full malignant potential. In accord with reports of TZD-induced morphologic and molecular changes associated with differentiation of NSCLC cells (41), pioglitazone induced morphologic changes indicative of MET in K-ras–mutated HBECs. It should be noted that pioglitazone is pleiotropic in its activities; therefore, the antitumorigenic effects we observed are likely not solely miR125a-dependent.

Here, we report findings supporting a role for the loss of miR125a expression in the pathogenesis of K-ras–dependent pulmonary premalignancy. Further studies will be required to determine whether this miRNA can function in evaluation and selection of chemoprevention agents. In addition, our finding that miR125a expression is reduced in at least one premalignant lesion compared with normal tissue in 5 of 8 patient samples surveyed demonstrates the marked heterogeneity inherent to lung carcinogenesis within a given individual. Further studies will be required to confirm our data in larger numbers of premalignant lesions and to determine whether miR125a expression levels in premalignant lesions correlate with K-ras status. Likewise, we have focused our studies on the K-ras–mutated airway epithelium, but future studies will be required to determine whether miR125a functions in the pathogenesis of premalignancy involving other mutations considered to be drivers of lung cancer pathogenesis. Understanding the contribution of specific miRNA in the pathogenesis of pulmonary premalignancy and their potential pharmacologic manipulation will be important in the future application of personalized chemopreventive agents targeted to the underlying molecular and epigenetic drivers of early lung cancer development.

No potential conflicts of interest were disclosed.

Conception and design: E.L. Liclican, T.C. Walser, S. Hazra, K. Krysan, S.J. Park, P.C. Pagano, S.M. Dubinett

Development of methodology: E.L. Liclican, T.C. Walser, S. Hazra, K. Krysan, S.J. Park, P.C. Pagano, B.K. Gardner, J.D. Minna, S.M. Dubinett

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): E.L. Liclican, K. Krysan, B.K. Gardner, J.E. Larsen, S.M. Dubinett

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): E.L. Liclican, S. Hazra, K. Krysan, B.K. Gardner, S.M. Dubinett

Writing, review, and/or revision of the manuscript: E.L. Liclican, T.C. Walser, S. Hazra, K. Krysan, J.D. Minna, S.M. Dubinett

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): E.L. Liclican

Study supervision: T.C. Walser, S.M. Dubinett

Grant support: J.D. Minna, S.M. Dubinett

The authors thank Ying Lin for technical assistance and the following UCLA Core Facilities for technical services and advice: The UCLA Vector Core, the UCLA Genotyping and Sequencing Core, and the CNSI Advanced Light Microscopy/Spectroscopy Shared Resource Facility at UCLA.

These studies were supported by funding from the following sources: NIH/NHLBI #T32HL072752 (to S.M. Dubinett, E.L. Liclican, T.C. Walser, S.J. Park), NCI #U01CA152751 (to S.M. Dubinett, T.C. Walser, K. Krysan, B.K. Gardner), NIH/NCATS UCLA CTSI #UL1TR000124 (to S.M. Dubinett), Department of Veteran Affairs #5I01BX000359 (S.M. Dubinett), NCI Lung Cancer SPORE #P50CA70907 (to J.D. Minna, J.E. Larsen), and Cancer Prevention & Research Institute of Texas #RP120732 (to J.D. Minna).

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.