Lung Endothelial MicroRNA-1 Regulates Tumor Growth and Angiogenesis

Abstract

Rationale: Vascular endothelial growth factor down-regulates microRNA-1 (miR-1) in the lung endothelium, and endothelial cells play a critical role in tumor progression and angiogenesis.

Objectives: To examine the clinical significance of miR-1 in non–small cell lung cancer (NSCLC) and its specific role in tumor endothelium.

Methods: miR-1 levels were measured by Taqman assay. Endothelial cells were isolated by magnetic sorting. We used vascular endothelial cadherin promoter to create a vascular-specific miR-1 lentiviral vector and an inducible transgenic mouse. KRAS^{G12D mut}/Trp^53−/− (KP) mice, lung-specific vascular endothelial growth factor transgenic mice, Lewis lung carcinoma xenografts, and primary endothelial cells were used to test the effects of miR-1.

Measurements and Main Results: In two cohorts of patients with NSCLC, miR-1 levels were lower in tumors than the cancer-free tissue. Tumor miR-1 levels correlated with the overall survival of patients with NSCLC. miR-1 levels were also lower in endothelial cells isolated from NSCLC tumors and tumor-bearing lungs of KP mouse model. We examined the significance of lower miR-1 levels by testing the effects of vascular-specific miR-1 overexpression. Vector-mediated delivery or transgenic overexpression of miR-1 in endothelial cells decreased tumor burden in KP mice, reduced the growth and vascularity of Lewis lung carcinoma xenografts, and decreased tracheal angiogenesis in vascular endothelial growth factor transgenic mice. In endothelial cells, miR-1 level was regulated through phosphoinositide 3-kinase and specifically controlled proliferation, de novo DNA synthesis, and ERK1/2 activation. Myeloproliferative leukemia oncogene was targeted by miR-1 in the lung endothelium and regulated tumor growth and angiogenesis.

Conclusions: Endothelial miR-1 is down-regulated in NSCLC tumors and controls tumor progression and angiogenesis.

At a Glance Summary

Scientific Knowledge on the Subject

Tumor endothelial cells are integral components of the tumor microenvironment, and angiogenesis is a hallmark of tumor progression. However, current antiangiogenic therapies have limited clinical efficacy. MicroRNAs regulate various cellular responses and are prime targets for antitumor therapies. The specific role of microRNAs in the tumor endothelium has not been adequately studied.

What This Study Adds to the Field

MicroRNA-1 is down-regulated in the non–small cell cancer tumor endothelium, and tumor microRNA-1 levels correlate with overall survival of patients with lung adenocarcinoma. MicroRNA-1 is also down-regulated in the endothelium of lung adenocarcinoma models and regulates tumor progression, endothelial proliferation, and extracellular signal–regulated protein kinases 1/2 signaling.

Endothelial cells are critical components of the tumor microenvironment, and tumor vascularity closely correlates with tumor progression and invasiveness (1–3). Vascular endothelial growth factor (VEGF) is one of the main orchestrators of tumor angiogenesis, and anti-VEGF agents are currently approved for the treatment of advanced lung cancer (4).

MicroRNAs (miRNAs) are small (21–24 nt) RNAs that regulate gene expression by base pairing with specific sequences within messenger RNAs (5–10). miRNAs are altered in malignancies and in many instances impart a significant effect on tumor behavior (11, 12). We had previously shown that VEGF regulates miRNA-1 (miR-1) levels in the lung endothelium of asthma models (13). Regarding the importance of VEGF in tumor progression and angiogenesis, we sought to determine the role of miR-1 in non–small cell lung cancer (NSCLC) tumors.

We found that miR-1 levels in NSCLC tumors are lower than the adjacent cancer-free tissues and correlate inversely with overall survival. miR-1 was also lower in the endothelial cells isolated from these tumors and was down-regulated in the lung endothelium of the Kirsten rat sarcoma viral oncogene homolog (KRAS) mutant/transformation-related protein 53 (P53 or Trp53) knockout (KP) mouse model of NSCLC on tumor formation. Vascular-specific overexpression of miR-1 significantly decreased tumor burden in NSCLC models and reduced tumor and lung angiogenesis.

Methods

NSCLC Patient Sample Preparation

The early–late NSCLC (all-NSCLC) cohort consisted of 68 patients with NSCLC who underwent surgical resection (n = 34) or bronchoscopy (n = 34) (see Table E1 in the online supplement) at the Yale Cancer Center. The early lung adenocarcinoma (early LA) cohort consisted of 41 patients who underwent surgical resection (n = 41) (see Table E2) at the University of Pittsburgh Medical Center. Research design, procedures, and human subject approvals are described in the online supplement.

miRNA and Messenger RNA Analysis by Quantitative Real-Time Polymerase Chain Reaction

miRNAs and messenger RNAs were measured using Taqman assays (Life Technologies, Carlsbad, CA) and SYBR green quantitative real-time polymerase chain reaction (Biorad, Hercules, CA) and according to the manufacturer’s instructions (13).

Isolation of Endothelial Cells from Murine Lung by Magnetic-activated Cell Sorting

Endothelial (CD31⁺, CD45⁻, or CD105⁺, CD45⁻) and hematopoietic (CD45⁺) cells were isolated using Miltenyi Biotech (Auburn, CA) magnetic columns as described previously (13).

Mice Used in the Study

Animal protocols were approved by the Yale University Institutional Animal Care and Use Committee. CC10-reverse tetracycline-controlled transactivator (rtTA)-VEGF transgenic mice and measurement of tracheal angiogenesis have been described previously (14) and in the online supplement. Generation of miR-1 transgenic mice is described in the online supplement. In the Lewis lung carcinoma (LLC) model, 2 × 10⁶ tumor cells were implanted subcutaneously, and the size of the implants was monitored. When the tumor sizes reached 5–10 mm in diameter, they were injected with vectors and controls, and their sizes were monitored every other day over the next 10 days. The maintenance, use, and cross-breeding of Kras^LSL-G12D/Trp53^flx/flx mouse (KP mouse) are described in the online supplement.

Cell Culture

Mouse lung endothelial cells (MLECs) were isolated from murine lungs as described previously (13). Human umbilical vein endothelial cells (HUVECs) were purchased from the Department of Vascular Biology and Therapeutics Program at Yale. Detailed descriptions of cell culture assays are described in the online supplement.

Short Hairpin RNA and Lentiviral Vectors

Vascular-specific miR-1 (V-miR-1) was generated by cloning vascular endothelial cadherin (VE-cadherin) promoter into pSIH-H1-copGFP Lentiviral vector (SBI, Palo Alto, CA) as described in the online supplement.

Statistical Analysis and Data Availability

A detailed description of statistical analysis is included in the online supplement. All relevant data are available from authors with restrictions for the human data.

Results

miR-1 Level Is Lower in NSCLC Tumors and Correlates with Overall Survival

We examined the levels of mature miR-1 in tumor and cancer-free (normal) biopsy samples from a 68-patient cohort with early and late-stage NSCLC (all-NSCLC; see Table E1) and found that miR-1 levels were significantly lower in the tumor samples in the whole group (Figure 1A) and within both adenocarcinoma and squamous cell carcinoma subtypes (see Figures E1A and E1B). miR-1 levels were also lower than cancer-free tissue in a 41-patient cohort of early-stage lung adenocarcinoma (early LA; Figure 1B; see Table E2).

Figure 1.

The clinical significance of microRNA-1 (miR-1) levels in non–small cell lung cancer. (A) Mature miR-1/18s (control gene) levels were measured in tumors and cancer-free lung tissue samples from the all–non–small cell lung cancer cohort. Values were normalized to the median of the normal samples and expressed as 2^−∆∆Ct (n_Tumor = 61, n_Normal = 57; *P = 0.0033). (B) miR-1 values in samples from the early lung adenocarcinoma cohort were measured and graphed as in A (n_Tumor = 41, n_Normal = 41; *P < 0.0001). In A and B, red bars show medians in each group. (C) Patients were divided into T-high and T-low groups based on the median of all tumor miR-1 levels in the cohort. Kaplan-Meier graph shows survival (in days) for T-high and T-low patients (n = 61; P = 0.0082). (D) Patients with lung adenocarcinoma (Adenoca) were divided into T-high and T-low groups and their survival compared as described in C (n = 36; P = 0.0189). In C and D, gray tick marks indicate censored patients.

We next tested the clinical correlations of miR-1. In the all-NSCLC cohort, tumor miR-1 levels had a significant correlation with overall survival (P = 0.0022) and histologic subtype (P = 0.0008) and showed a strong trend toward lower levels in more advanced clinical stages (P = 0.0708), but no significant associations with age, sex, or tumor size (see Table E3A). In a subgroup analysis, miR-1 correlation with survival was only observed in the adenocarcinoma group (see Table E3B). Because of inadequate data and the short duration of follow-up, overall survival could not be assessed for the early-LA cohort.

High miR-1 Levels Predicted Longer Overall Survival

We next examined the value of tumor miR-1 levels as a predictor of tumor progression. The all-NSCLC cohort was divided into high miR-1 level (T-high) and low miR-1 level (T-low) groups based on the median of miR-1 levels within the cohort. T-high patients survived significantly longer than T-low patients (P = 0.031; see Table E3C) but showed no difference in sex, age, size of the tumor, or clinical stage. There was also a significant difference in the distribution of tumor histologic subtypes between the T-high and T-low patients (P = 0.0018). In Kaplan-Meier analysis, high miR-1 levels (T-high status) was predictive of longer survival (Figure 1C). Among tumor subtypes, miR-1 maintained this predictive ability only in the adenocarcinoma subgroup (Figure 1D). High miR-1 levels were also predictive of longer survival in a publicly available 542-patient lung adenocarcinoma cohort (The Cancer Genome Atlas Lung Adenocarcinoma; P = 0.000487; hazard ratio, 0.91 [0.86–0.96]) (see Figure E2).

miR-1 Is Down-regulated in the Tumor Endothelium

miR-1 levels were significantly lower in the endothelial (CD45⁻, CD31⁺) cells isolated from NSCLC tumors compared with their adjacent cancer-free tissues (Figure 2A), suggesting that the lower levels of miR-1 in the whole tumors are at least partly caused by its down-regulation in the endothelial compartment. We tested this hypothesis in the KP mouse model. Activation of floxed Kras^LSL-G12D and deletion of Trp53^flx/flx alleles in this model lead to tumor formation in 4–6 months (15). Mature miR-1 levels were more than 80% lower in the tumor-bearing lungs and endothelial cells isolated from these lungs, compared with the control subjects (Figures 2B and 2C).

Figure 2.

MicroRNA-1 (miR-1) is down-regulated in tumor endothelium and regulates angiogenesis. (A) Endothelial cells (CD31⁺, CD45⁻) were isolated from non–small cell lung cancer tumors (Tumor) and adjacent cancer-free tissues (Normal). miR-1/18s levels were normalized to the median of the control (Normal) group and expressed as 2^−∆∆Ct (n = 28; *P = 0.0351). Red horizontal lines indicate medians of the groups. (B) Kirsten rat sarcoma viral oncogene homolog (KRAS) mutant/transformation-related protein 53 (P53) (KP) knockout mice received intranasal Cre recombinase (or control [ctrl]) vector and their lungs harvested after 6 months. miR-1/18s levels were measured in the whole lungs, normalized to the mean of the control group, and expressed as 2^−∆∆Ct (n = 12 in each group from two experiments; *P = 0.01533). (C) Endothelial (CD31⁺, CD45⁻) cells were isolated from the lungs of KP and ctrl mice and miR-1 levels were measured and normalized as described in B (n ≥ 5 in each group from two experiments; *P = 0.016). (D) Vascular-specific miR-1 expression vector (V-miR-1). Vascular endothelial cadherin (VECad) promoter was cloned upstream of primary (pri)-miR-1 sequence in a lentiviral transfer vector containing enhanced green fluorescent protein as a marker gene driven by elongation factor 1 α subunit (EF1α) promoter. (E–G) Effect of vascular-specific miR-1 overexpression on tumor growth and angiogenesis. Lewis lung carcinoma implants were injected with V-miR-1 or scrambled control vector (V-scr) on Day 10, and harvested on Day 19 after implantation. The sizes of the tumors were measured and volumes calculated according to the following formula: volume = 0.52 × width² × length. (E) Tumor volumes are graphed versus the time of measurement (number of days on the x-axis; data points are means of tumor volume) (n ≥ 13 in each group from three experiments; *P = 0.036013, **P = 0.016034). (F) Representative tumors from the two groups on Day 19 after implantation. (G) Tumor sections were stained with anti-CD31 antibody and DAPI (4',6-diamidino-2-phenylindole). (Left) Representative images. (Right) Quantification of vessel density based on the percentage of CD31-positive areas/whole area examined (n ≥ 7 in each group from two experiments; *P = 0.002645). Error bars represent SEM. ψ = packaging signal; EGFP = enhanced green fluorescent protein; LTR = long terminal repeat.

Endothelial miR-1 Controls Tumor Growth and Vascularity

Lentiviral delivery model

To examine the importance of miR-1 down-regulation, we tested the effects of endothelial-specific miR-1 overexpression on tumor formation. We first constructed a vascular-specific lentiviral vector (V-miR-1) that contained VE-cadherin promoter (vascular-specific promoter) upstream from primary miR-1 (pri-miR-1) sequence, and eukaryotic elongation factor alpha (EF1α) promoter (a universal promoter driving expression in all cell types) (17) upstream from green fluorescent protein (Figure 2D). As expected (16, 17), V-miR-1 delivery led to green fluorescent protein expression in all lung structural cells (see Figure E3A), whereas miR-1 was only overexpressed in the lung endothelium (see Figures E3B–E3E). Delivery of V-miR-1 to fully formed LLC xenografts (250–500 mm³ tumor volume) significantly decreased their growth rate and vascularity, showing the inhibitory effect of endothelial miR-1 in advanced tumors (Figures 2E–2G).

Transgenic model

We created an inducible vascular-specific miR-1 transgenic mouse model (miR-1 TG) by cloning miR-1 primary transcript (pri-miR-1) sequence downstream from a tetracycline-inducible cytomegalovirus promoter (tet-O-pCMV-pri-miR-1-hGH) and coexpressing this vector with a second construct containing VE-cadherin–driven transactivator (Figure 3A). Addition of doxycycline to the drinking water of these mice led to overexpression of miR-1 specifically in endothelium (Figure 3B). As with our lentiviral method, induction of miR-1 in this model also reduced the growth and vascularity of the fully formed (500 mm³ tumor volume) LLC xenografts (Figures 3C and 3D).

Figure 3.

The effects of vascular-specific microRNA-1 (miR-1) in the transgenic model. (A) Constructs used for the generation of inducible endothelial-specific transgenic mouse. VE-Cad-rtTA-hGH: contains vascular endothelial cadherin (VE-Cad) promoter, reverse tetracycline transactivator (rtTA), and human growth hormone (hGH) intronic, nuclear localization, and polyadenylation sequences. tet-O-pCMV-pri-miR-1-hGH: contains a polymeric tetracycline operator (tet-O), minimal cytomegalovirus (CMV) promoter, and hGH intronic, polyadenylation, and nuclear localization signals flanking its multiple cloning site. Primary (pri)-miR-1 sequence was cloned between the hGH intronic sequence and tet-O-CMV promoter. (B) miR-1 levels were measured in endothelial (CD31⁺, CD45⁻) and immune (CD45⁺) cell fractions isolated from wild-type and miR-1 transgenic (miR-1 TG) mice. Bar graphs represent mean level of miR-1 expressed as 2^−∆∆Ct value (see supplementary methods for description of normalization). The asterisk indicates significant increase in CD31⁺ CD45⁻ fraction as compared to CD45⁺ fraction in the miR-1 TG group (n = 4 per group; *P = 0.001768). (C and D) The effects of miR-1 transgene expression on tumor growth and angiogenesis in Lewis lung carcinoma model. Tumor implantation and volume measurements performed as described in Figure 2E (n ≥ 4 from three experiments; *P = 0.033645; **P = 0.007521). (D) Tumors sections were stained with anti-CD31 antibody and analyzed for vascular density as described in Figure 2G. (Left) Representative images of tumors. (Right) Quantification of CD31-positive area (vascular density) expressed as percentage of the whole area examined (n ≥ 12 from three experiments; *P = 0.017903). (E and F) The effect of endothelial miR-1 in Kirsten rat sarcoma viral oncogene homolog (KRAS)/transformation-related protein 53 (P53) knockout (KP) model. KP and KP + M mice (KP cross with miR-1-TG mice) received Cre recombinase at 1 month of age. miR-1 overexpression was induced by adding doxycycline to the drinking water 5 months after Cre delivery and lungs were harvested 6 months after Cre delivery. (E) Relative expression of miR-1 (miR-1/18s in KP + M and KP mice normalized to the levels in KP mice) in endothelial cells isolated from these lungs (n = 9 from 2 experiments; *P = 0.001211). (F) Lungs were sectioned (5 μm) and stained with hematoxylin and eosin. Tumor burden was determined by measuring tumor area/whole lung area. (Left) Representative images of the lungs from two mice in each group (each image was assembled from multiple smaller images at ×400 magnification). (Right) Quantification of tumor burden in the two groups (n = 9 from two experiments; *P = 0.02887). Error bars represent SEM. WT = wild-type.

Endothelial miR-1 Inhibits Tumor Growth in the KP Model

We next tested the role of endothelial miR-1 in the KP model. We created a quadruple-transgenic mouse model by crossing KP and miR-1 vascular-transgenic mice. The progeny of this breeding carried the two miR-1 overexpression constructs in addition to the floxed KRAS and P53 cassettes. miR-1 overexpression in these mice reduced the tumor burden by more than 85% (Figures 3E and 3F), confirming the critical role of miR-1 in the lung endothelium.

miR-1 Regulates Angiogenesis in the Lung

We next asked if miR-1 exerts a specific antiangiogenic effect or decreases tumor vascularity because of its antitumoral effects. We thus tested the effect of miR-1 in lung-specific VEGF transgenic mouse, which is a tumor-free model of lung angiogenesis (18). Intranasal treatment of these mice with a double-stranded miR-1 mimic decreased tracheal vascularity by 50% (Figure 4A). As described previously (13), this intranasal treatment increased miR-1 levels in the lung endothelium by more than 10 times without altering the VEGF levels (see Figures E4A and E4B). Vascular-specific overexpression of miR-1 through delivery of V-miR-1 vector had a similar effect and significantly reduced tracheal angiogenesis (Figure 4B), confirming the endothelial-specificity of the miR-1 effect.

Figure 4.

The effects of vascular-specific microRNA-1 (miR-1) on lung angiogenesis. (A) miR-1, scrambled control (scr-ctrl), or vehicle were delivered intranasally to vascular endothelial growth factor (VEGF) transgenic (VEGF TG) or wild-type (WT) mice 1 day after inducing the transgene. Lungs were harvested after 10 days. Tracheas were stained with anti-CD31 antibody. (Left) Representative images. (Right) Quantification of CD31-positive area (vascular density) expressed as a percentage of the whole area examined (n ≥ 8 from four experiments; *P < 0.00001). (B) Effect of vascular-specific miR-1 overexpression on tracheal vascular density. WT and VEGF TG mice received intranasal V-miR-1 or V-scr and VEGF transgene was induced by adding doxycycline to the drinking water. Tracheal vascular density was measured as described in A. (Left) Representative images. (Right) Results of quantification (n ≥ 6 from three experiments; *P = 0.00970). Error bars represent SEM. V-miR-1 = vascular-specific miR-1; V-scr = vascular-specific scrambled control.

miR-1 Regulates the Proliferation of Lung Endothelial Cells

To gain insight into the mechanism of the miR-1 antiangiogenic effect, we tested its effects on specific endothelial phenotypes of the MLECs. miR-1 transfection decreased VEGF-induced MLEC growth over a wide range of VEGF concentrations (Figures 5A and 5B) and reduced endothelial sprouting (Figure 5C), but did not affect survival or migration (see Figures E5A and E5B). We further examined this specific antiproliferative effect by testing the effect of miR-1 on de novo DNA synthesis (5-bromo-2'-deoxyuridine [BrdU], incorporation). As shown in Figure 5D, miR-1 transfection reduced BrdU incorporation by more than 30%.

Figure 5.

The effect of microRNA-1 on mouse lung endothelial cells (MLEC). (A–D) MLECs were transfected with double stranded miR-1 mimic (miR-1) or scrambled control RNA (Scr-ctrl) and stimulated with recombinant human vascular endothelial growth factor (VEGF)-A. (A) MLEC growth at 24 hours in response to various concentrations of VEGF was measured using WST-1 kit (Roche) (n = 6 from two experiments; *P < 0.05). (B) MLEC growth in response to 50 ng/ml of VEGF at the indicated time points was measured as in A (x-axis shows time) (n = 24 from four experiments; *P < 0.01, **P < 0.000001). (C) Endothelial sprouting (capillary tube formation in matrigel). (Left) Representative images. (Right) Quantitation of the relative tube length (n = 9 from three experiments; *P < 0.005). (D) Bromodeoxyuridine incorporation in response to VEGF was measured as an index of de novo DNA synthesis by ELISA (Roche) and values were normalized to the control reaction (n ≥ 24 from six experiments; *P = 1.4 × 10⁻⁶). Error bars represent SEM. BrdU = bromodeoxyuridine.

miR-1 Regulates Proliferation and Extracellular Signal–regulated Protein Kinase Activation in Human Endothelial Cells

We next examined the role and regulation of miR-1 in human primary endothelial cells. We used HUVECs that are sensitive to VEGF and commonly used as an in vitro model of angiogenesis (19, 20). VEGF stimulation reduced miR-1 levels in HUVECs in a dose-dependent manner. Similar to our approach in MLECs, we verified the importance of this regulation by testing the effects of miR-1 alteration on VEGF-induced cell proliferation. Overexpression of miR-1 reduced the endothelial growth rate, whereas inhibition of miR-1 reduced this rate (Figures 6A and 6B). In complementary experiments, miR-1 reduced de novo DNA synthesis but did not alter apoptosis (Figures 6D and 6E). In signaling experiments (21), miR-1 specifically inhibited extracellular signal–regulated protein kinases (ERK) 1/2 phosphorylation at its peak (5 and 10 min) without affecting phosphoinositide 3-kinase (PI3 kinase), or JNK pathways (Figures 6F and 6G). We also observed that the relative phospho-P38/total P38 levels were higher at baseline in the miR-1–transfected cells, but did not find a consistent inhibitory effect on this axis.

Figure 6.

The role and regulation of microRNA-1 in human primary endothelial cells. (A) miR-1/18s levels were measured in human umbilical vein endothelial cells (HUVECs) stimulated with 10 ng or 50 ng/ml of recombinant human vascular endothelial growth factor (VEGF). Values were normalized to the mean of the control group and expressed as 2^−∆∆Ct (n ≥ 15 from five experiments; *P < 0.0001). (B–E) HUVECs were transfected with miR-1 antagomir (miR-1-antag), mature miR-1 double-stranded mimic (miR-1), or their respective scrambled controls (scr-antag, or scr-ctrl). HUVECs were then starved overnight and stimulated with VEGF (10 ng/ml). (B) The effects of miR-1-antag and its control on proliferation were measured and presented as in Figure 5A (n = 12 from two experiments; *P = 0.043). (C) The effects of miR-1 and its control on proliferation were measured and presented as in Figure 5A (n = 16 from three experiments; *P = 0.00194; **P = 0.0307). (D) The effects of miR-1 and scr-ctrl on bromodeoxyuridine incorporation (de novo DNA synthesis) was measured and presented as in Figure 5D (n = 20 from four experiments; *P = 0.00123). (E) The effect of miR-1 and scr-ctrl on cell death was analyzed by fluorescence-activated cell sorter analysis for Annexin V and propidium iodide (PI). (Left) Results of a typical fluorescence-activated cell sorter analysis experiments. (Top) Results of staining with each reagent in a representative experiment. (Bottom left) Representative dot plots for each experimental group. (Bottom right) Quantification of apoptotic cell fraction defined as a percentage of Annexin V–positive cells (n ≥ 6 from two experiments; *P = 0.028; **P = 0.047). (F and G) HUVECs were transfected with miR-1 or scr-ctrl, starved overnight, and stimulated with VEGF. Reaction was stopped at 5, 10, or 30 minutes and cells were lysed. Phosphorylated and total extracellular signal–regulated protein kinase (ERK) (p-ERK and t-ERK, respectively) fractions were detected by Western blotting. (F) Representative immunoblot (top) and quantification of the activated ERK1/2 (p-ERK/t-ERK) (bottom) (x-axis shows time in minutes) (n = 8; *P < 0.04). (G) Representative immunoblots for activated (phosphorylated, p-) and total (t-) P38 mitogen-activated protein kinase (P38K), phosphoinositide 3-kinase (PI3K), and c-jun N-terminal kinase (JNK). (H) HUVECs were starved, incubated with blockers, and stimulated with VEGF for 24 hours. miR-1 was measured as described in Figure 2 (n ≥ 10 from five experiments; *P < 0.001; **P < 1 × 10⁻⁵). Error bars represent SEM. BrdU = bromodeoxyuridine; FSC = forward scatter; PBS = phosphate-buffered saline; SSC = side scatter.

miR-1 Is Regulated via the VEGFR2-PI3 Kinase Pathway

We next sought to determine the mechanism of miR-1 down-regulation. We have previously shown that lung miR-1 level is regulated through VEGF receptor 2 (VEGFR2) (13). We tested the role of four major VEGF signaling axes on miR-1 regulation by measuring the effects of their corresponding blockers on miR-1 levels. As shown in Figure 6H, among the ERK, P38, JNK, and PI3 kinase blockers, only PI3 kinase blockade inhibited VEGF-driven miR-1 down-regulation. Furthermore, blockade of Akt, a downstream effector of the PI3 kinase pathway, had a similar effect, confirming the role of the PI3 kinase in miR-1 down-regulation.

miR-1 Targets Thrombopoietin Receptor (Mpl) in the Lung Endothelium

Mpl is one of the miR-1 targets in the endothelium (13). We asked if miR-1 targets Mpl in the context of angiogenesis. VEGF overexpression increased Mpl levels in the lung and intranasal miR-1 delivery lowered Mpl levels in that context (Figures 7A and 7B). This inhibition corresponded to a 70% decrease of Mpl in the lung endothelial cells, suggesting that miR-1 is targeting Mpl in the endothelium. In accord with this finding, endothelial-specific overexpression of miR-1 also decreased lung endothelial Mpl levels (Figure 7C).

Figure 7.

The role of myeloproliferative leukemia virus oncogene (Mpl) in angiogenesis and tumor growth. (A) Mpl expression was detected by Western blot analysis in the lungs of vascular endothelial growth factor (VEGF) transgenic (TG) mice (VEGF TG⁺) and their wild-type (WT) littermates (WT/VEGF TG⁻) that received double-stranded microRNA-1 (miR-1) or scrambled control (Scr-ctrl) RNA. (Top) Representative Western blot. (Bottom) Results of densitometric analysis (TG+M1 = VEGF TG + miR-1) (n = 3; *P and **P < 0.05). (B) Mpl expression was measured by Taqman quantitative real-time polymerase chain reaction in lung endothelial cells isolated from mice that received intranasal miR-1 or Scr-ctrl-RNA. The graph shows Mpl/glyceraldehyde phosphate dehydrogenase levels normalized to the control group and expressed as 2^−ΔΔCt (n = 11 from four experiments; *P < 0.03). (C) A similar measurement as shown in B on lung endothelial cells from mice that received vascular-specific lenti-miR-1 (V-miR-1) or control (V-Scr-ctrl). The graph shows Mpl/glyceraldehyde phosphate dehydrogenase levels normalized to the control group and expressed as 2^−ΔΔCt (n = 8, from two experiments; *P < 0.002). (D) VEGF-TG and -WT mice received small interfering RNA (siRNA) against Mpl (Mplsi) or Scr-ctrl. Tracheas were isolated and angiogenesis was assessed as described in Figure 4A. (Left) Representative images of tracheal vessels from WT and TG mice treated with vehicle (buffer), Mpl siRNA, or scrambled control RNA (n = 9 from three experiments; P < 0.0005). (E) Lewis lung carcinoma implants were injected with lentiviral vector containing Mpl short hairpin RNA (shRNA) (Mplsh) or the control vector (Ctrl). Tumors were harvested on Day 21 and vascularity assessed after staining with anti-CD31 antibody (n = 14 from two experiments; *P = 0.004923). (F) Tumor sizes were measured at the indicated time points and volumes calculated as in Figure 2E (n = 17 from two experiments; *P = 0.01606, **P = 0.00302, ***P = 4.07 × 10⁻⁵). (G and H) Mouse lung endothelial cells were transfected with Mplsi or Scr-ctrl and stimulated with VEGF (50 ng/ml). (G) Cell proliferation was measured as described in Figure 1D (n = 24 from four experiments; *P < 0.01, **P < 0.00001). (H) BrdU incorporation was compared between the two groups (values were normalized to the control reaction, n = 20 from four experiments; *P = 0.02). Error bars represent SEM. BrdU = bromodeoxyuridine.

Mpl Regulates Angiogenesis in Mouse and Human Models

To determine the role of Mpl in angiogenesis, we first tested the effect of Mpl knockdown through siRNA delivery (13). Mpl knockdown reduced tracheal vascularity by approximately 50% (Figure 7D) and tumor angiogenesis by 30% (Figures 7E and 7F), closely resembling the inhibitory effects of miR-1 overexpression. Mpl knockdown in MLECs also mimicked miR-1 effects and significantly inhibited cellular growth and differentiation (Figure 7G; see Figure E6), and de novo DNA synthesis (Figure 7H). We next tested the role of Mpl in human primary endothelial cells. Consistent with its targeting by miR-1, Mpl knockdown decreased de novo DNA synthesis (Figure 8A) and ERK 1/2 activation in HUVECs (Figure 8B), and Mpl overexpression had the opposite effect in both assays (Figures 8C and 8D).

Figure 8.

The role of myeloproliferative leukemia virus oncogene (Mpl) in human endothelial cells. (A and B) Human umbilical vein endothelial cells (HUVECs) were transfected with Mpl small interfering RNA (siRNA) (Mplsi) or scramble control RNA (scr-ctrl) and stimulated with vascular endothelial growth factor. (A) Bromodeoxyuridine (BrdU) incorporation in HUVEC was measured by ELISA and normalized to Scr-ctrl (n ≥ 5 from four experiments; *P < 0.0005). (B) Extracellular signal–regulated protein kinase (Erk) activation (phospho-Erk protein [pErk]/total Erk protein [tErk]) was measured as described in Figure 6F. (Top) Representative blot. (Bottom) Results of densitometric analysis (x-axis shows time in minutes) (n = 6 for each data point from three experiments; *P = 0.05). (C and D) HUVECs were transfected with Mpl overexpression vector (MplOE) or control vector (Ctrl) and stimulated with vascular endothelial growth factor. (C) BrdU incorporation was measured and depicted as described in C (n = 12 from two experiments; *P < 0.0001). (D) Erk activation was measured as described in B (n = 6 from six experiments; *P = 0.0026; **P = 0.0344). Error bars represent SEM.

Discussion

Tumor progression is the result of a complex and multifaceted interaction between malignant cells and their microenvironment (2). However, recent efforts to “personalize” anticancer interventions have mostly focused on the tumor cells. We had previously shown that VEGF down-regulates miR-1 and that miR-1 regulates type 2 inflammation (13). In this manuscript we show that miR-1, a lung endothelial miRNA, is altered in NSCLC, and controls tumor progression and angiogenesis.

We found that miR-1 was significantly lower in NSCLC tumor samples and showed for the first time that tumor miR-1 levels have a direct correlation with overall survival of patients with NSCLC. We also observed a strong trend toward an inverse correlation between tumor miR-1 levels and the clinical stage (P = 0.07), which is consistent with association of miR-1 levels and survival. Statistical association between miR-1 levels and tumor progression has been previously reported for gastrointestinal and urinary cancers (22–25) and a recent study on 55 patients with NSCLC reported a correlation between miR-1 levels and venous metastasis (26). Another study on 33 squamous cell carcinoma samples reported lower miR-1 levels in tumors but no association with tumor progression (27). In our cohort, tumor miR-1 levels were lower than the cancer-free tissue in both adenocarcinoma and squamous cell carcinoma patients, but the correlation with survival was only observed in the adenocarcinoma subgroup, and later confirmed in a publicly available 542-patient cohort. We also observed lower miR-1 levels in an early lung adenocarcinoma cohort but could not confirm the correlation with survival in that cohort because of inadequate data and short duration of follow-up. Even though these observations suggest a specificity for the miR-1 role in adenocarcinoma tumors, considering the smaller size of the squamous cell carcinoma subgroup within our cohort, we cannot confirm this specificity at this time.

We have found that PI3 kinase/Akt signaling pathway mediates miR-1 regulation in the endothelium. PI3 kinase pathway plays a significant role in initiation and propagation of tumors (28–32). The role of this pathway in the NSCLC microenvironment and its contribution to tumor resistance, and specifically tumor angiogenesis, have been demonstrated by various groups (33–37). Interestingly, a recent clinical study showed an inverse correlation between miR-1 and PI3 kinase in NSCLC tumors (26). Inhibition of miR-1 down-regulation through PI3 kinase blockade strongly suggests that PI3 kinase activation is the main mechanism regulating miR-1 levels in tumors.

Previous studies have shown that miR-1 regulates multiple aspects of the tumor cell behavior, such as proliferation (38, 39), motility (27, 40), apoptosis (41, 42), angiogenesis (43, 44), and glucose metabolism (45). Two recent studies suggested that miR-1 modulates the paracrine function of cancer-associated fibroblasts (46, 47), and one recent report showed that ectopic miR-1 modulates the extracellular vesicles released from glioblastoma tumor cells (48). However, demonstration of the endothelial-specific functions of miR-1 and mechanistic aspects of these effects are unique aspects of our study.

In vivo elucidation of the role of miRNAs in the tumor microenvironment has only been attempted in a few instances through delivery of nanocomplexes or engraftment of miRNA-transfected cells (49–52). These approaches are informative and therapeutically relevant but do not provide the clarity and stability of promoter-driven expression from a genomic locus. We, for the first time, used VE-cadherin promoter to overexpress miR-1 specifically in the tumor endothelium, confirmed the efficacy and specificity of miRNA expression with this method, and showed miR-1 antiangiogenic properties in various contexts.

We used both in LLC xenografts and KRAS/P53 model in our experiments. KRAS/P53 mouse is one of the most faithful models of NSCLC (53–56) and significant inhibition (>85%) of tumor growth in this model by endothelial-specific miR-1 demonstrates the critical role of miR-1 in the tumor endothelium. As a candidate mechanism for this role, we tested the effect of miR-1 on angiogenesis. miRNAs have differential effects on endothelial behavior (57, 58). miR-126 and miR-132 stimulate angiogenesis (49, 59), whereas miR-92a or miR-519c inhibit this process (60, 61). A previous study on zebrafish reported an antiangiogenic role for miR-1 and miR-206 (62). In line with this report, a few studies indicated that miR-1 targets VEGF in malignant cells (43, 44). Our findings in VEGF transgenic mice and in the in vitro VEGF supplementation experiments strongly suggest that the antiangiogenic effects of miR-1 can occur independent of its anti-VEGF properties. The VEGF transgene in our mouse model does not carry a 3′UTR, and its expression was not affected by miR-1 delivery (see Figure E3B). Furthermore, we could not find any statistical association between miR-1 and VEGF levels within our clinical cohorts (personal communication).

In mechanistic studies, alterations of the miR-1 levels inhibited endothelial cell growth and sprouting, without affecting cell death or migration. Cellular proliferation and differentiation require de novo DNA synthesis (3, 63), and miR-1 reduced this activity in a BrdU incorporation assay. Furthermore, miR-1 inhibited ERK1/2, which is one of the main signaling pathways downstream of VEGF mediating endothelial proliferation, sprouting, and de novo DNA synthesis (64–66). Inhibition of ERK by miR-1 is thus consistent with its antiangiogenic properties. However, it should be noted that direct proof for the functionality of this mechanism requires in vivo measurement of ERK activity in the tumor endothelium. We also observed that miR-1–transfected endothelial cells show a higher level of P-38 activity at baseline (Figure 6G). P-38 activation leads to endothelial cell apoptosis (67, 68,) migration (69, 70), and inhibition of cell differentiation and ERK activation (67, 71). Even though activation of P-38 does not completely explain our observations, it is possible that this pathway contributes to the miR-1 antiangiogenic properties. It should also be noted that miR-1 antiproliferative effect is most probably only one of the endothelial phenotypes affected by miR-1, and other effects of miR-1 on endothelium await further inquiry.

We showed that miR-1 targets endothelial Mpl in the context of angiogenic activation and that Mpl knockdown mimicked miR-1 effects in the lung, tumor, and endothelial cells. Mpl is a type I cytokine receptor that mediates platelet production and hematopoietic stem cell maintenance (72) through activation of a number of downstream pathways, including STAT3 and 5 (73, 74), Shc-Grb2-SOS (75, 76), and Raf-1/MAP kinase (77, 78). Mpl is expressed in endothelial cells (79–81) and has been implicated in the angiogenic response within certain contexts (79, 82).

Our findings show for the first time that Mpl is one of the genes targeted by miR-1 in the lung microenvironment and suggest that Mpl mediates a significant portion of miR-1 inhibitory effects. Consistent with this idea, in preliminary experiments, overexpression of Mpl in the endothelium promoted tumor growth and angiogenesis, and reversed the inhibitory effects of miR-1 (personal communication). However, it should be noted that miRNAs target multiple genes and regulate cellular phenotypes through robust inhibition of multiple mediators (83), and that Mpl is most probably only one of the miR-1 mediators in endothelium. Further in vivo, such as Argonaute HITS-CLIP, will yield other novel miR-1 targets within the endothelium (84).

In this manuscript we used a variety of strategies to delineate the specific role and regulation of miR-1 in NSCLC tumors and the lung microenvironment. Our findings showed that miR-1 is a predictor of survival in patients with NSCLC, is regulated within the endothelium, and controls tumor growth and angiogenesis. These findings suggest that miR-1 may have clinical utility in the management and monitoring of patients on antiangiogenic therapy.