Cigarette smoke induces angiogenic activation in the cancer field through dysregulation of an endothelial microRNA

Abstract

Cigarette smoke (CS) creates a "cancer field" in the lung that promotes malignant transformation. The molecular changes within this field are not fully characterized. We examined the significance of microRNA-1 (miR-1) downregulation as one of these changes. We found that tumor miR-1 levels in three non-small cell lung cancer cohorts show inverse correlations with the smoking burden. Lung MiR-1 levels follow a spatial gradient, have prognostic significance, and correlate inversely with the molecular markers of injury. In CS-exposed lungs, miR-1 is specifically downregulated in the endothelium. Exposure to CS induces angiogenesis by selectively degrading mature miR-1 via a vascular endothelial growth factor-driven pathway. Applying a multi-step molecular screen, we identified angiogenic genes regulated by miR-1 in the lungs of smokers. Knockdown of one of these genes, Notch homolog protein 3, simulates the anti-angiogenic effects of miR-1. These findings suggest that miR-1 can be used as an indicator of malignant transformation.

Fig. 1. Cigarette smoke and miR-1 levels in NSCLC and non-cancerous lung.

A, B MiR-1 expression (mature miR-1/reference gene, expressed as 2^−∆Ct) were measured in tumor samples from NSCLC patients (A) MiR-1 in smokers vs non-smokers (n = 9 for non-smokers and n = 63 for smokers, p = 0.0186). B MiR-1 in smokers with ≤ 30 vs >30 pack-year smoking history (n = 27 for ≤ 30 and n = 34 for >30, p = 0.012). C, D MiR-1 expression in surgically-resected NSCLC tumor samples from patients in the PROSPECT cohort was determined by array analysis (Illumina v3) and expressed as Log2-transformed values. C Comparison in all NSCLC patients (n = 246, fold = −1.47, *p = 0.004). D Comparison in adenocarcinoma (LUAD) patients (n = 170, fold change = −1.44, *p = 0.007). E MiR-1 expression in lung samples from LGRC cohort. Mature miR-1/18S levels were normalized to the mean value of the non-smokers group (control) and expressed as 2^−∆∆Ct (n = 6 and 15, *p = 0.0151). F MiR-1 expression in AT tissues from former and current smokers (n = 38 and 9 respectively, *p = 0.0066). G Lungs were harvested from CC-LR (LSL-KrasG12D) mice after exposure to cigarette smoke for 2 months (smoked) or room air (control). MiR-1/reference gene levels were normalized to the mean value from the control group and expressed as 2^−∆∆Ct. (n = 3 per group, p = 0.02). H Lungs were harvested from mice after 6 months of smoking (smoked) and non-exposed littermates (Control). MiR-1/reference gene levels were measured and expressed as described in (G), (n = 5 for control and 4 for smoked, *p = 0.02). I Human lung tissues were cultured ex vivo and exposed to various concentrations of CSE. MiR-1/reference gene levels were measured, normalized to the mean value of the control group(0), and expressed as 2^−∆∆Ct (n = 4 patients, 3 replicates from each, *p = 0.0021 **p = 0.0014).

Fig. 2. MiR-1 levels in the cancer field.

A MiR-1/ reference gene levels were measured in the tumor (T), adjacent tissue (AT), and distant tissue (DT) samples from smokers and expressed as 2^−∆Ct.(n = 12 patients, p-value = 0.0048, one-way ANOVA nonparametric Friedman test). B, C T and AT samples were divided into low miR-1( < median) and high miR-1( > median) and the corresponding survivals of the patients in each group were compared using Kaplan Meier analysis (B) T-low vs T-high patients (n = 72 subjects, p = 0.0403, Log-Rank Mantel Cox test) (C) AT-low vs AT-high patients (n = 58 subjects, p = 0.0357, Log-Rank Mantel Cox test). D–H The levels of miR-1 and PI3K pathway genes/reference gene were measured in AT samples (described in table S2) by qRT-PCR and expressed as Log 2^−∆Ct. D Phosphatidylinositol-4,5-Bisphosphate 3-Kinase Catalytic Subunit Alpha (PI3KCA), (Spearman r = −0.3491, P = 0.0235, n = 42), (E) Growth Arrest And DNA Damage Inducible Beta (GADD45B), (Spearman r = −0.513, P = 0.0023, n = 33), (F) Forkhead box protein O1 (FOXO1), (Spearman r = −0.3232, P = 0.0419, n = 40) (G) Mechanistic Target Of Rapamycin Kinase (mTOR1), (Spearman r = −0.3499, P = 0.0394, n = 35), and (H) AKT Serine/Threonine Kinase 2 (AKT2), (Spearman r = −0.6081, P < 0.0001, n = 40).

Fig. 3. CS downregulates miR-1 specifically in the endothelial cells.

A Human ex-vivo cultured lungs were treated with 1% CSE or vehicle (PBS) for 24 h, and epithelial (CD 45-, EPCAM +), endothelial (CD 45-, CD31 +), immune (CD 45 +), and double-negative (CD45−, CD31−) cells were isolated using magnetic sorting. The graph represents miR-1/reference gene levels in CSE-treated cellular fractions normalized to the levels in PBS groups and expressed as 2^−∆∆Ct (n = 3 patients, *p = 0.0176). B–E Cells were treated with increasing concentrations of CSE for 24 h and miR-1/ reference gene levels were measured, normalized to the mean value of ‘0’ CSE concentration group (control), and expressed as 2^−∆∆Ct (B) EA.hy926 (n = 21, 4, 18, and 9 from 4 experiments, *p = 0.0035, **p = 0.00145) (C) A549 cells (n = 6/concentration, from two experiments, *p = 0.0043) (D) HPMECs (n = 7 for 0 and 10% and 4 for 5%, from 2 experiments *p = 0.0012) (E) HUVECs (n = 11,10 and 8, from 3 experiments, *p = 0.007) (F) Endothelial cells were isolated from non-cancerous lung tissue (HPMECs) and tumor (TEC) from a lung cancer patient and cultured in growth media. miR-1/ reference gene levels were measured in total RNA from the cells, normalized to the mean levels in HPMEC group, and expressed as 2^−∆∆Ct (n = 6, *p = 0.041).

Fig. 4. CS downregulates miR-1 through VEGF pathway.

A NSCLC patients were grouped into non-smokers and smokers based on their smoking status and VEGF/reference gene levels were measured in the tumor samples and expressed as 2^−∆Ct. (n = 9 for non-smokers and n = 57 for smokers, p = 0.0067). B Human lung tissues were cultured ex-vivo and exposed to various concentrations of CSE. VEGF/ reference gene levels were measured, normalized to the mean value of the control group(0), and expressed as 2^−∆∆Ct (n = 3 subjects, 4 replicates in each. *p < 0.0001). C Human ex-vivo cultured lungs were treated with 1% CSE or vehicle (PBS) for 24 h. Endothelial (CD45− CD31+), immune (CD45+), and double-negative (CD45−, CD31−) cells were isolated using magnetic sorting. The graph represents VEGF/reference gene levels in CSE-treated cellular fractions normalized to the levels in PBS groups and expressed as 2^−∆∆Ct (n = 3 patients, *p = 0.0366). D, E Endothelial cells were treated with various concentrations of CSE for 24 h and VEGF/reference gene levels were measured, normalized to the mean value of the control group(0), and expressed as 2^−∆∆Ct. D HUVECs (n = 8 in each group, from 2 experiments, *p = 0.0002). E EAhy926 (n = 10 or more from 2 experiments, *p = 0.0048, **p = 0.0062). F HUVECs were exposed to 10% CSE, collected at the time points shown on the X axis, and VEGF protein levels measured by ELISA (n = 3/time point, *p < 0.0) (G) EAhy926 were starved overnight, incubated with and without a VEGFR2 blocker (sUs, and treated with 2% CSE for 24 hours. MiR-1/reference gene levels were measured, normalized to the control group, (vehicle, DMSO) and expressed as 2^−∆∆Ct (n = 6 or more from 2 experiments, *p = 0.0221). H Murine lung endothelial cells (MLECs) were treated with blockers, exposed to 5% CSE, and miR-1 was measured and expressed as described in (G). (n = 4 per group, *p < 0.03).

Fig. 5. The effect of CS on miR-1 biogenesis.

A–C HUVECs were exposed to increasing concentrations of CSE for 24 h. Cells were collected at the time points illustrated on the X axis and the levels of miR-1/reference gene, (A), Pri-miR-1/reference gene (B), and pre-miR-1-1 and -2/reference gene (C) were measured, normalized to the corresponding mean value at ‘0’ CSE concentration (control), and expressed as 2^−∆∆Ct. (n = 5 /time point, *p < 0.01). D HUVECs were transfected with a mature miR-1 mimic (22nt). Twenty-four hours after transfection cells were exposed to CSE and miR-1 levels were measured at various time points as described in (A). (n = 5 /time point, *p < 0.05).

Fig. 6. The role of miR-1 in CS-induced angiogenesis.

A EA.hy 926 were transfected with miR-1 or scrambled control RNA (ctrl) and exposed to 1% CSE for the indicated times. Cell numbers were determined and normalized to values at time ‘0’ (n = 6 from 2 experiments, *p = 0.039). B HUVECs were transfected with miR-1 (or ctrl) treated with CSE and phospho- and total- ERK and beta-actin visualized by Western blot. The top and bottom lines in the marker lane represent 50 and 37 KD bands. C, D Cells were transfected with antagomiRs (amiR-1 or control RNA, actrl), treated with CSE, and relative cell numbers measured as described in (A). C EA.hy 926 (n = 6, *p = 0.0022) (D) HUVECs (n = 9 from 2 experiments, *p = 0.011).

Fig. 7. Comparative analysis for miR-1 targets in NSCLC smokers.

We compared gene expression levels between tumor and AT samples from 6 smokers and 3 non-smokers in our cohort. We selected the genes that were (A) expressed at a higher level in tumors (vs AT samples) and (B) in tumors from smokers vs non-smokers. (C) Had a known miR-1 binding site in their 3′UTR (based on three different prediction algorithms) and (D) were regulated by VEGF in ECs. E We compared the above genes with the list of the genes recruited by miR-1 to the RISC.

Fig. 8. Validation of miR-1 targets in endothelial cells.

A HUVECs were transduced with v-miR-1 (or control vector). Ago-RIP was performed on cell lysates and the levels of each gene in the Ago pull immunoprecipitate/input lysate was measured and expressed as 2^−∆∆Ct. (n = 4, *p < 0.03. B–E HPMEC were transfected with miR-1 mimic (miR-1) or control RNA (Ctrl) and exposed to 10% CSE for 24 h. mRNA/ reference gene levels were measured, normalized to Ctrl in PBS group, and expressed as 2^−∆∆Ct. B NOTCH3 (n = 6, *p < 0.05), (C) HS3ST1 (n = 6, *p < 0.05), (D) SEMA4B (n = 6, *p < 0.025,**p = 0.0026 (E) TFAP4 (n = 6, p = NS). F, G HUVECs were transfected with NOTCH3 siRNA or scrambled control RNA (ctrl), exposed to 10% CSE for the indicated times. F Cell numbers were determined and normalized to values at time ‘0’ (n = 3 per group and time, *p < 0.03, **p < 0.02). G De novo DNA synthesis was determined using a BRDU ELISA colorimetric assay. Values were normalized to the baseline (time ‘0’) and presented as relative BRDU incorporation. (n = 8 per group and time, *p < 0.05, **p < 0.0002).

Fig. 9. MiR-1 target genes in human lung and cancer field.

A–D Human lung tissues were cultured ex-vivo and exposed to 1% CSE. mRNA/ reference gene levels were measured, normalized to the mean value of the control group(0), and expressed as 2^−∆∆Ct. A NOTCH3 (n > 10,*p = 0.02) (B) HS3ST1 (n > 10, *p = 0.013) (C) SEMA4B (n > 10, *p = 0.0071) (D) TFAP4 (n > 10, *p = 0.024). E–H mRNA/ reference gene levels were measured in the tumor (T), adjacent tissue, (AT), and distant tissue (DT) samples from 12 smokers and expressed as 2^−∆Ct. Analysis was done by applying the one-way ANOVA nonparametric Friedman test. E NOTCH3 (n = 12, p = 0.0001), (F) HS3ST1 (n = 12, p = 0.0005). G SEMA4B (n = 12, p = 0.0003), (H) TFAP4 (n = 12, p = 0.04).