Introduction
Lung cancer is the leading cause of cancer-related mortality worldwide (1) and 85% of the cases are represented by non-small cell lung cancer (NSCLC), which is classified into three different histological subtypes: adenocarcinoma (ADC), squamous cell carcinoma (SCC) and large cell carcinoma (LCC) (2–4). Despite a better understanding of the NSCLC pathogenesis and significant improvement in early diagnosis and treatment, the overall 5-year survival is extremely low (~15%) and the patients show high recurrence rates even at the early disease stages (1–7), highlighting the necessity of a deeper knowledge of NSCLC biology and the identification of more effective biomarkers.
MicroRNAs (miRNAs) are a highly conserved family of small (17–22 nucleotides), non-coding, endogenous, single-stranded RNA molecules that negatively regulate gene expression by binding to complementary sequences on target messenger RNA (mRNA) (8,9). Recently, miRNAs have been shown to regulate essential cell processes, such as cell proliferation, differentiation, apoptosis, development and metabolism (9–11), and to play a key role in cancer pathogenesis (12–16). Moreover, a miRNA prognostic and diagnostic value has been reported in several malignancies, including lung cancer (14–16).
The first reported miRNA aberrantly expressed in lung cancer was the Let-7 family (17). A reduced Let-7 expression has been significantly correlated with a short post-operative survival in the NSCLC patients (17). Moreover, the ectopic Let-7 expression inhibits cell proliferation in human NSCLC cell lines (18) and reduces tumor burden in mouse NSCLC xenografts (19). Let-7 family members have been demonstrated to behave as tumor suppressor genes and to functionally inhibit several cell cycle regulators and oncogenes, such as Ras family, c-Myc and HMGA2 genes, whose 3′UTRs show multiple Let-7 binding sites (13,20,21).
Conversely, a role as oncogene has been suggested for miR-21 that is deregulated in glioblastoma and lung cancer (22–25). A miR-21 overexpression has been suggested to be an independent negative prognostic factor for the overall survival in NSCLC patients (23) and to be related to the lung carcinogenesis in never smokers (26). Several mRNAs have been identified as miR-21 targets, including PDCD4, PTEN, TGF-β and MMP9 genes (22,27).
The aim of this study was to evaluate Let-7g and miR-21 expression in a series of 80 NSCLC patients to establish their involvement in the NSCLC pathogenesis and their potential diagnostic, prognostic and predictive value.
Materials and methods
Patients
Eighty NSCLC patients were retrospectively selected from patients who had undergone surgery at the Unit of Thoracic Surgery of the A.O.U.P. between 2005 and 2012. Histological diagnoses were independently formulated by two pathologists (G.F. and G.A.) according to the World Health Organization classification (2–4) and discrepant diagnoses were re-evaluated and discussed until an agreement was reached. Clinicopathological characteristics were collected whenever available for all the patients, while detailed clinical data were obtained only for 55 patients. The study was approved by the local Ethics Committee and all the patients gave their informed consent to the molecular analyses.
DNA and RNA isolation
DNA and RNA were isolated from 10-μm sections of formalin-fixed and paraffin-embedded (FFPE) tissues or cytological specimens after manual tumor macrodissection using the QIAamp DNA Mini kit (Qiagen) and miRNeasy FFPE kit (Qiagen), respectively, according to the manufacturer's instructions.
MiRNA expression
Quantification of Let-7g, miR-21 and RNU6B expression was carried out in triplicate into 80 NSCLC and 27 non-cancer lung tissues using specific TaqMan® MicroRNA assays (Applied Biosystems) according to the manufacturer's instructions. Briefly, 10 ng of total RNA were retro-transcribed by the TaqMan MicroRNA Reverse Transcription (RT) kit (Applied Biosystems) and 1.3 μl of RT product were analysed by quantitative real-time PCR (qRT-PCR) on the Rotor-Gene 6000 (Corbett Research). Threshold cycle (Ct) and baselines were determined by manual settings. MiRNA expression was calculated by relative quantification and fold expression changes were determined by the 2−ΔΔCt method using the DataAssist™ software (Applied Biosystems).
Target prediction and pathway analysis
Let-7g and miR-21 target genes were predicted by four different miRNA target prediction algorithms: miRanda (http://www.microrna.org/microrna), TargetScan (http://www.targetscan.org), Pictar (http://www.pictar.org) and miRDB (http://www.mirdb.org). Gene ontology classification and pathway analysis were performed using the PANTHER software (http://www.pantherdb.org).
Mutational analysis
K-Ras gene (Reference sequence: ENSG00000133703) status in codons 12 and 13 was analyzed by pyrosequencing using the Anti-EGFR MoAb response® kit (K-Ras status) (Diatech Pharmacogenetics) according to the manufacturer's instructions.
PCR-single stranded conformation polymorphism (PCR-SSCP) and sequencing analysis were used for genotyping exons 18–21 of the EGFR gene (Reference sequence: ENSG00000146648). The primer sequences were as follows: exon 18, 5′-CTCTGTGTTCTTGTCCCCCC-3′ (forward) and 5′-GCCTGTGCCAGGGACCTTAC-3′ (reverse); exon 19, 5′-CATGTGGCACCATCTCACA-3′ (forward) and 5′-CCACACAGCAAAGCAGAAAC-3′ (reverse); exon 20, 5′-CACACTGACGTGCCTCTCC-3′ (forward) and 5′-TATCTCCCCTCCCCGTATCT-3′ (reverse); exon 21, 5′-CCTCACAGCAGGGTCTTCTC-3′ (forward) and 5′-CCTGGTGTCAGGAAAATGCT-3′ (reverse). Briefly, 100 ng of DNA were amplified by PCR using the FastStart Taq DNA Polymerase (Roche Diagnostics) on the T3000 Thermocycler 48 (Biometra), as follows: 4 min at 95°C, 40 cycles at 95°C for 30 sec, 58°C for 30 sec and 72°C for 45 sec and 10 min at 72°C. PCR products were mixed with an equivalent formamide volume, denatured at 95°C for 5 min and run onto a non-denaturing 12.5% polyacrylamide gel (GE Healthcare) at 18°C and constant 25 mA for 1 h and 40 min. Denaturated DNA was visualized by the PlusOne DNA silver staining kit (GE Healthcare) and samples with altered mobility patterns were sequenced as previously described (28).
Statistical analysis
One-way analysis of variance and χ2 test were used to determine the association between miRNA expression and the different parameters, while survival analysis was performed by the Kaplan-Meier method. Statistical analyses were performed using the JMP10 software (SAS) and a two-tailed p<0.05 was considered statistically significant.
Results
Patient characteristics
This study was conducted in 80 patients with NSCLC, including 55 ADCs, 21 SCCs, 2 LCCs and 2 undifferentiated NSCLCs. The median age at diagnosis was 67 years (range 46–85) and the median follow-up was 32 months (range 7–98). Disease progression with distant and/or loco-regional recurrence and death from lung cancer were observed in 34 (61.8%) and 14 (25.5%) of the 55 NSCLC patients, respectively. The median progression-free survival (PFS) and overall survival (OS) were 18 months (95% CI, 14–24) and 24 months (95% CI, 18–30), respectively.
Let-7g and miR-21 expression profile
We quantified the mature Let-7g and miR-21 expression normalized to the RNU6B endogenous control in 80 NSCLC and 27 non-cancer lung tissues. The unsupervised hierarchical clustering analysis of miRNA expression using the Euclidean distance as a similarity measure and average linkage algorithm revealed two major clusters based on similarities in miR-21 expression that clearly separated the tumor from non-cancer tissues. On the contrary, we did not observe a clear separation between tumor and normal samples based on Let-7g expression (Fig. 1A).
Let-7g was barely detectable in lung tissues and we did not observe any significant difference between the NSCLC and normal samples (−0.897±0.148 vs. −0.709±0.168, p=0.585, Fig. 1B). Conversely, a highly significant increase in miR-21 expression was observed in the NSCLC tissues compared to the non-cancer ones (4.842±0.163 vs. 2.509±0.182, p<0.0001, Fig. 1C).
MiRNA profile and clinicopathological characteristics
To determine whether miRNA profile was correlated with the main clinicopathological characteristics, the NSCLC patients were divided into Let-7g and miR-21 high and low expression groups based on the median fold-change values (1.315±0.175 for Let-7g and 6.964±0.759 for miR-21). Except for a significant association between the low Let-7g expression and metastatic lymph node presence at diagnosis (p=0.046), no other statistically significant associations were observed between the analysed miRNA and the main clinicopathological characteristics of the NSCLC patients (Table I). Interestingly, both Let-7g and miR-21 were upregulated in ADCs compared to SCCs, although these relationships were not statistically significant (Fig. 2A and B).
MiRNA profile and K-Ras and EGFR status
To investigate whether Let-7g and miR-21 expression was correlated to K-Ras and EGFR mutational status, we performed genotyping in the 80 NSCLC patients, except 2, due to insufficient tissue. K-Ras and EGFR mutations were observed in 16 (20.5%) and 23 (29.5%) of the 78 NSCLC patients, respectively (Table II).
K-Ras and EGFR mutations were mutually exclusive, observed only in the NSCLC patients with ADC and associated with gender (Table III). As is shown in Table III, EGFR status was also significantly associated with the smoking habit (p=0.0086), performance status (p=0.0008) and response to the treatment with EGFR tyrosine kinase inhibitors (TKIs) (p=0.0076).
Statistical analysis did not show any significant association between EGFR mutations and Let-7g or miR-21 expression, while we found a highly significant association between K-Ras status and miR-21 expression (p=0.0003, Table III). Noteworthy, a significantly higher miR-21 expression was observed in the NSCLC patients with K-Ras-mutated tumors (14.237±1.638, p<0.001) compared to the patients with K-Ras-wild-type tumors (7.316±0.792, Fig. 2C).
miRNA target prediction and pathway analysis
Let-7g and miR-21 target gene analysis by miRanda, TargetScan, Pictar and miRDB prediction algorithms led to the identification of a plethora of putative target genes for these miRNAs. In order to minimize the number of false positives, we recorded a gene as a putative target gene of the analysed miRNAs only if it was predicted by at least two prediction algorithms with a high confidence score. According to these stringent criteria, we identified 24 putative target genes for Let-7g, including HMGA2, ERCC6 and MAP3K3 genes and 26 putative target genes for miR-21, including PDCD4, MSH2 and SPRY1/SPRY2 genes (Table IV).
We further investigated the biological consequences of Let-7g and miR-21 aberrant expression grouping the predicted target genes by gene ontology terms. This analysis revealed that most of cell processes regulated by these miRNAs play a key role in cancer pathogenesis and are mainly involved in cell proliferation, apoptosis, DNA repair, cell adhesion and signal transduction pathways (Table IV).
Survival analysis
To evaluate the relationships of Let-7g and miR-21 expression with the prognosis of the NSCLC patients, we performed a survival analysis by the Kaplan-Meier method using the disease recurrence and the overall post-operative survival as end-points. We did not observe any significant difference in PFS and OS of the NSCLC patients with a high Let-7g or miR-21 expression compared to the patients with a low expression of these miRNAs (Fig. 3). However, we further investigated Let-7g and miR-21 as prognostic indicators by restricting our analysis to the first 30 months of the follow-up to verify a possible short-term prognostic value of Let-7g and miR-21 evaluation. Interestingly, we found that the NSCLC patients with a high Let-7g or miR-21 expression showed a significantly shorter mean PFS and OS compared to the patients with a low expression of these miRNAs (Table V).
Discussion
Lung cancer is the first cause of death for cancer worldwide and >80% of the cases are NSCLC (1–4). Although early diagnosis and patient care have greatly improved in recent years, most of the NSCLC patients show locally advanced or metastatic disease at diagnosis and their prognosis remains extremely poor (1–7). Currently, no appropriate diagnostic biomarker exists for NSCLC, highlighting the need of a better knowledge of its biology to improve prevention, diagnosis and treatment.
MiRNAs are a highly conserved family of small non-coding RNA molecules that negatively regulate gene expression (8,9) and their aberrant expression has been found to play a key role in pathogenesis of several malignancies, including NSCLC (13–16). This study was aimed to evaluate Let-7g and miR-21 expression profile in the NSCLC patients in order to establish their role in NSCLC pathogenesis and their potential diagnostic, prognostic and predictive significance.
We demonstrated that miR-21 expression strongly differentiates the NSCLC from non-cancer lung tissues, while we did not observe any Let-7g discriminative value. In our study, a highly significant increase was found in miR-21 expression in NSCLC tissues compared to the non-cancer ones, in agreement with previous results that demonstrated a miR-21 overexpression in tumor tissues from several human malignancies (23,29–31). Conversely, we observed a reduced Let-7g expression that was expressed at comparable levels in NSCLC and non-cancer lung tissues. Let-7g downregulation in NSCLC tissues has been previously reported by several authors, who have also demonstrated that the aberrant expression of Let-7 family represents an early event during NSCLC carcinogenesis and is more common in SCCs compared to ADCs (12,19,32,33). In our study, Let-7g and miR-21 are downregulated in SCCs compared to ADCs, but their evaluation has not been shown to have a significant diagnostic value in discriminating between these two different histotypes, as previously reported in larger miRNA profiling studies (29,30). Landi et al(30) reported that Let-7g and miR-21 differential expression allows to discriminate between ADC and SCC in the early-stage tumors (stage I), but not in the advanced stage (stage II–IV), suggesting that miRNA expression loses its histology-specificity in the more advanced and less differentiated tumors. Therefore, the lack of a statistical significance we observed between the altered Let-7g and miR-21 expression and NSCLC histology could be explained by the fact that most of the enrolled patients were diagnosed in an advanced stage, where miRNA histology-related expression may be not specific.
Concerning the other analysed clinicopathological characteristics, we did not observe any significant correlation between the Let-7g and miR-21 dysregulated expression and the clinicopathological features, including age, tumor stage, therapy response and smoking habit. In particular, our results concerning the relationship between the miR-21 expression and the smoking habit are in disagreement with data reported by Seike et al(26), who demonstrated that miR-21 expression is significantly higher in tumors from smokers than from never smokers; however, this discrepancy could be due to the small number of patients for whom we had smoking data. Interestingly, we found that Let-7g expression is significantly associated with lymph nodal status. We showed that most of the NSCLC patients with a low Let-7g expression present metastatic lymph nodes at diagnosis, while no substantial differences were observed for the patients with a high Let-7g expression. This result suggests an important role of Let-7g in NSCLC tumor progression and acquisition of metastatic potential and is supported by in vivo studies showing that ectopic Let-7g expression in NSCLC xenografts induce a significant decrease in tumor growth and spread (17,18).
Since the importance of EGFR and K-Ras mutation detection in current management of the NSCLC patients, we explored the relationship between their mutational status and Let-7g and miR-21 expression profile. In our study, the frequency of K-Ras mutation (20.5%) was in agreement with previously reported data (34,35), whereas EGFR mutation incidence was slightly higher (29.5%) than that reported in the literature for lung cancer (15–20%) (6,36–38), because many of the recruited patients belonged to a larger study designed to evaluate the EGFR TKI response. According to previously reported data (20,21,30,39), Let-7g expression was not correlated with EGFR or K-Ras mutational status. However, several studies have demonstrated that Let-7g acts as a K-Ras negative regulator by binding to multiple sites of their 3′UTRs (40) and that lung cancer tissues with reduced Let-7g levels have significantly higher K-Ras levels compared to their corresponding normal tissues (18,20). Therefore, it is possible that the aberrant expression of Let-7g and K-Ras mutations are mutually exclusive in NSCLC carcinogenesis with a more predominant effect of Let-7g dysregulation in SCCs, which show a low expression of this miRNA compared to the other NSCLC histotypes and a more prominent role of K-Ras mutations in ADC carcinogenesis (32,33).
Furthermore, we first demonstrated a strong and highly significant correlation between the high miR-21 expression and the presence of mutations in the codons 12 and 13 of K-Ras gene, suggesting a synergistic interplay between miR-21 and K-Ras oncogenes that supports neoplastic phenotype in NSCLC. Based on miR-21 expression and target gene prediction results, we might hypothesize an auto-regulatory loop between oncogenic K-Ras and miR-21 mediated by the MAPK/ERK signalling pathway, SPRY1/SPRY2 and PDCD4 (Fig. 4). K-Ras mutations determine a constitutive protein activation with a consequent activation of the MAPK/ERK signalling pathway, which plays an important role in lung carcinogenesis by inhibiting apoptosis and promoting cell proliferation, cell growth, angiogenesis, invasion and metastasis (41). On the other hand, miR-21 modulates several components critical to the NSCLC pathogenesis by targeting apoptotic effectors and antagonists of the MAPK/ERK signalling pathway (22,25,27). The high miR-21 expression observed in our NSCLC patients might cause a decrease in SPRY1/SPRY2 expression that has been demonstrated to negatively regulate the MAPK/ERK signalling pathway and to enhance cell migration (42). In addition, the negative regulation of PDCD4 and Apaf-1 genes by miR-21 leads to apoptosis inhibition (22,27,43), as well as to the removal of the PDCD4 inhibitory effect on AP-1, which is downstream the MAPK/ERK signalling pathway and promotes miR-21 expression (44,45). This complex and auto-regulatory circuit might justify the high levels of miR-21 expression observed in our study in the NSCLC patients harbouring K-Ras mutations and might have a final stimulation effect on the processes that promote tumor progression and therapy resistance (Fig. 4).
We investigated the relationship between the differential Let-7g and miR-21 expression and prognosis of the NSCLC patients without observing any statistically significant correlation. These results are in disagreement with data reported by other authors that support a negative prognostic role for Let-7g, whose downregulation has been associated with a reduced overall post-operative survival in NSCLC patients (18–21), and miR-21, whose overexpression has been associated with a poor prognosis irrespective of the TNM stage and lymph nodal status (23,26,46). However, these discrepant results could be due to the small number of patients with available clinical data. Interestingly, by restricting our analysis to the first 30 months of the follow-up observation, we demonstrated that the NSCLC patients with a high expression of either Let-7g or miR-21 show a highly significant shorter PFS and OS compared to the patients with a low expression of both these miRNAs, suggesting a possible negative short-term prognostic value of the evaluation of Let-7g and miR-21 expression.
In conclusion, our data show that Let-7g and miR-21 are aberrantly expressed in the NSCLC patients and that there is a close interplay among K-Ras, miR-21 and Let-7g in NSCLC, suggesting that their systematic evaluation could represent a useful biomarker in the molecular characterization and management of NSCLC patients.