Research ArticleCancer

Lung cancer–associated pulmonary hypertension: Role of microenvironmental inflammation based on tumor cell–immune cell cross-talk

See allHide authors and affiliations

Science Translational Medicine  15 Nov 2017:
Vol. 9, Issue 416, eaai9048
DOI: 10.1126/scitranslmed.aai9048

A tense environment

Lung cancer remains the most common cause of death worldwide, but some of the biology underlying the symptoms of this disease is still not well understood. In particular, the frequent occurrence of dyspnea in patients with advanced lung cancer remained to be explained. Pullamsetti et al. demonstrated that lung cancer drives vascular remodeling in its microenvironment, resulting in the development of pulmonary hypertension, helping to explain the patients’ symptoms. The authors showed evidence of this phenomenon in human patients, as well as in three different mouse models, where they also examined the mechanism underlying this condition.

Abstract

Dyspnea is a frequent, devastating, and poorly understood symptom of advanced lung cancer. In our cohort, among 519 patients who underwent a computed tomography scan for the diagnosis of lung cancer, 250 had a mean pulmonary artery diameter of >28 mm, indicating pulmonary hypertension (PH). In human lung cancer tissue, we consistently observed increased vascular remodeling and perivascular inflammatory cell accumulation (macrophages/lymphocytes). Vascular remodeling, PH, and perivascular inflammatory cell accumulation were mimicked in three mouse models of lung cancer (LLC1, KRasLA2, and cRaf-BxB). In contrast, NOD.Cg-Prkdcscid Il2rgtm1Wjl/SzJ immunodeficient xenograft and dominant-negative IKK2 mutant triple transgenic (Sftpc-rtTA/Tet-O-Ikk2DN) mice did not develop PH. Coculturing human lung cancer cells with macrophages and lymphocytes strongly up-regulated cytokine release, provoking enhanced migration, apoptosis resistance, and phosphodiesterase 5 (PDE5)–mediated up-regulation of human lung vascular cells, which are typical features of PH. The PDE5 inhibitor sildenafil largely suppressed PH in the LLC1 model. We conclude that lung cancer–associated PH represents a distinct PH category; targeting inflammation in the microenvironment and PDE5 offers a potential therapeutic option.

INTRODUCTION

Lung cancer is the leading cause of cancer death for both men and women and accounts for almost 28% of all cancer deaths worldwide, with global incidence increasing by ~0.5% per year (1). The overall 5-year survival rate is poor, but detection of lung cancer at earlier stages and development of individualized treatment strategies are expected to improve survival in the forthcoming years (2, 3). Lung cancer patients usually present with exertional dyspnea, which is associated with poor prognosis and reduced health-related quality of life (46). In addition to the loss of lung parenchyma owing to the tumor mass itself, the increased prevalence of cardiovascular and pulmonary comorbidities in lung cancer patients, which is about twice as high as in the general population (7), is thought to contribute to the dyspnea and to affect the survival rate (8, 9).

Pulmonary hypertension (PH) is characterized by increased pulmonary artery (PA) pressure and resistance, which negatively affects right ventricular (RV) function because of enhanced afterload, and it has also been linked with high mortality (10). It may appear as idiopathic pulmonary arterial hypertension but may also be associated with a variety of diseases, including those of the left heart and the lungs. Emerging evidence from many research groups indicates that the hyperproliferative processes of lung vascular cells driving remodeling and loss of the lumen of the lung vasculature exhibit “cancer-like” characteristics (1113). Nevertheless, the onset of PH in the context of lung cancer has hitherto largely been neglected, and only a few case reports have described the development of PH in cancer patients owing to tumor cell embolization of the lung vessels or to pulmonary thromboembolism caused by cancer-related venous thrombosis (14, 15).

Here, we present evidence from clinical diagnostics and histological evaluation of cancer-bearing human lung tissue that pulmonary vascular remodeling and PH frequently occur in lung cancer patients. Moreover, PH development was reproduced in three mouse models of lung cancer with different kinetics of tumor formation, and tumor cell chemokine production linked with perivascular accumulation of inflammatory cells was suggested to play a major pathomechanistic role. The recognition of this PH category may contribute to understanding of exertional dyspnea encountered in lung cancer patients, which will become even more relevant as individualized treatment results in longer survival of patients.

RESULTS

PA enlargement, indicating PH, and histopathological evidence of lung vascular remodeling in patients with lung cancer

Analysis of images acquired with high-resolution computed tomography (CT) of the chest revealed that 250 of 519 patients who presented with lung cancer in 2011–2013 had a mean PA diameter of ≥28 mm, which was suggestive of PH (Fig. 1A and Table 1). In addition, we calculated the mean pulmonary artery systolic pressure (PASP) of 70 patients in whom echocardiographic data were available. Both the mean and median PASP values were 33 mmHg, with individual values reaching up to 70 mmHg. Among these patients, 46% had PASP above the normal cutoff value of 34 mmHg (Fig. 1B and Table 2) (16).

Fig. 1. PA enlargement and pulmonary vascular remodeling detected in patients with lung cancer.

(A) Distribution of PA size in 519 patients with non–small cell and small cell lung cancer. The blue bars indicate PA diameter ≥ 28 mm. (B) Distribution of PASP in 70 patients with non–small cell and small cell lung cancer. The orange bars indicate PASP ≥ 34 mmHg. (C) Representative photomicrographs of elastica van Gieson–stained SCC and AC lung cancer tissues, followed by (D) quantification of medial wall thickness of pulmonary vessels in nontumor (N) and tumor (T) regions of SCC and AC lung cancer tissues, shown as a percentage of vessels that fall into the range specified on the y axis. Scale bars, 20 μm; n = 14. (E) Representative photomicrographs of von Willebrand factor (vWF) (brown)–stained and α-smooth muscle actin (SMA) (violet)–stained SCC and AC lung cancer tissues, followed by (F) quantification of pulmonary vascular muscularization (given in percentage). N, nonmuscularized pulmonary vessels; P, partially muscularized pulmonary vessels; M, fully muscularized pulmonary vessels. Scale bars, 20 μm; n = 14. (G) Medial wall thickness of pulmonary vessels from biopsies obtained from lung cancer patients with COPD (n = 10) grouped by stage of COPD. *P < 0.05, **P < 0.01, ***P < 0.001 in (D), (F), and (G) compared with nontumor regions of SCC or AC.

Table 1. Baseline characteristics of all lung cancer patients with available CT data.

SCLC, small cell lung cancer; NSCLC, non–small cell lung cancer.

View this table:
Table 2. Baseline characteristics of lung cancer patients with available echocardiography data.
View this table:

When comparing the demographic data between the PA diameter ≥28 mm group and the PA diameter <28 mm group, as well as the echocardiographic data between the PASP ≥34 mmHg group and the PASP <34 mmHg group, the baseline characteristics were not significantly different, although there were minor differences in body mass index and a relatively low number of women in the PA diameter ≥28 mm group (Tables 1 and 2). The groups did not differ with respect to histological classification, presence and Global Initiative for Chronic Obstructive Lung Disease (GOLD) stage of chronic obstructive pulmonary disease (COPD), and nasal oxygen dependency. The mean arterial partial pressure of oxygen (PO2) was lower in the PA diameter ≥28 mm group than in the PA diameter <28 mm group (65 mmHg versus 70 mmHg). To investigate whether patients with lung cancer exhibit lung vascular remodeling, which is the pathological hallmark underlying PH, we stained and morphometrically analyzed the lung vasculature in 14 different human lung cancer tissue sections (Fig. 1, C to F); patient characteristics, tumor type, and tumor-node-metastasis stage are provided in table S1. Medial wall thickness and the degree of muscularization were quantified in three different vascular categories (small, 20 to 50 μm; medium, 51 to 150 μm; and large, >151 μm). For both squamous cell carcinoma (SCC)–bearing and adenocarcinoma (AC)–bearing lung tissues, the medial wall thickness of pulmonary vessels of all sizes was consistently increased when comparing tumor regions with nontumor regions (Fig. 1, C and D). In addition, most small vessels (20 to 50 μm in diameter) were partially muscularized or nonmuscularized in the nontumor areas, whereas in the AC and SCC tumor areas, the percentage of fully muscularized precapillary vessels was increased at the expense of nonmuscularized vessels (Fig. 1, E and F). We did not find an association between the severity of lung vessel remodeling, as assessed by medial wall thickness, and the stage of COPD (Fig. 1G).

PH in Sftpc-cRaf-BxB transgenic mice

To explore whether mouse lungs display features of PH during cancer progression, we used three different mouse models of lung cancer, namely, the KRas knock-in line KRasLA2, the cRaf-BxB transgenic line Sftpc-cRaf-BxB (Sftpc; surfactant protein C), and Lewis lung carcinoma line 1 (LLC1), to characterize pulmonary hemodynamics and lung vascular morphology in detail (Figs. 2 to 4). Sftpc-cRaf-BxB transgenic mice displayed constitutively activated cRaf in alveolar type II (AT II) cells and developed ACs of the lung over a period of several months. By age 9 months, these mice demonstrated tumor burden in ~80% of the lung, as assessed by histopathology (Fig. 2, A and B). Echocardiographic follow-up of the Sftpc-cRaf-BxB mice between ages 7 and 9 months demonstrated increased RV internal diameter (RVID) and decreased cardiac output (CO) compared with wild-type (WT) littermates, indicating increased RV afterload owing to PH (Fig. 2, C and D). In addition, the lung cancer–bearing mice exhibited impairment of RV function, as shown by a marked reduction of tricuspid annular plane systolic excursion (TAPSE), increased RV myocardial performance index (MPI), and increased isovolumic relaxation time (IVRT/RR) (Fig. 2, C and D). Catheter measurements undertaken after the 9-month period corroborated these findings by showing increased RV systolic pressure (RVSP) and pulmonary vascular resistance (PVR) and a decrease in arterial oxygenation in Sftpc-cRaf-BxB mice, without any alteration in systemic arterial pressure (SAP) (Fig. 2E). Postmortem analysis demonstrated increased weight of the RV myocardium relative to body weight (RV/body weight) and to the left ventricle plus septum [RV/(LV + S)] (Fig. 2F).

Fig. 2. Features of PH in Sftpc-cRaf-BxB transgenic mice.

(A) Schematic of experimental design. Measurements were taken at age 7, 8, and 9 months (mo) for WT and Sftpc-cRaf-BxB lung tumor–bearing mice. (B) Representative hematoxylin and eosin–stained sections of mouse lungs. Scale bar, 500 μm. (C) Echocardiographic measurements and (D) representative echocardiographic images. (E) Physiological measurements of RVSP, PVR, PO2, SAP, and (F) RV hypertrophy [RV/(LV + S)] and RV weight (RV/body weight) of the mice. For WT, n = 5; for Sftpc-cRaf-BxB, n = 5 to 8. (G) Representative photomicrographs of vWF (brown)–stained and α-SMA (violet)–stained tissues, followed by (H) quantification of pulmonary vascular muscularization (given as a percentage for each range of vessel sizes). Scale bars, 20 μm. (I) Representative photomicrographs of elastica van Gieson–stained tissues, followed by (J) quantification of medial wall thickness of pulmonary vessels. Scale bars, 20 μm. (K) Representative photomicrographs of Sirius red staining (right). Scale bars, 20 μm. Quantitative image analysis of Sirius red staining (collagen deposition) in the RV (left). For WT, n = 5; for Sftpc-cRaf-BxB, n = 9. *P < 0.05, **P < 0.01, ***P < 0.001 compared with WT; §P < 0.05, §§P < 0.01, §§§P < 0.001 compared with the nontumor part of Sftpc-cRaf-BxB.

Fig. 3. Features of PH in KRasLA2 transgenic mice.

(A) Schematic of the experimental design. Measurements of 3-, 4-, and 5-month-old WT and KRasLA2 lung tumor–bearing mice. (B) Representative hematoxylin and eosin–stained sections of mouse lungs. Scale bar, 500 μm. (C) Echocardiographic measurements. (D) Physiological measurements of RVSP, SAP, PO2, and PVR of the mice. For WT, n = 7; for KRasLA2, n = 10. (E) Representative photomicrographs of vWF (brown)–stained and α-SMA (violet)–stained tissues, followed by (F) quantification of pulmonary vascular muscularization (given as a percentage for each range of vessel sizes). Scale bars, 20 μm. (G) Representative photomicrographs of elastica van Gieson–stained tissues, followed by (H) quantification of medial wall thickness of pulmonary vessels. Scale bars, 20 μm. (I) Representative photomicrographs of Sirius red staining (right). Scale bars, 20 μm. Quantitative image analysis of Sirius red staining (collagen deposition) in the RV (left). For WT, n = 7; for KRasLA2, n = 10. *P < 0.05, **P < 0.01, ***P < 0.001 compared with WT; §P < 0.05, §§P < 0.01, §§§P < 0.001 compared with nontumor regions of KRasLA2.

Fig. 4. Features of PH in LLC1 lung tumor mice.

(A) Schematic of the experimental design. C57BL/6 mice were intravenously injected with saline (WT) or LLC1 cells. After 10, 14, and 18 days from the initial injection, WT and LLC1 lung tumor–bearing mice underwent echocardiographic and physiological measurements. (B) Representative hematoxylin and eosin–stained sections of mouse lungs. Scale bar, 500 μm. (C) Echocardiographic measurements. (D) Physiological measurements of RVSP, SAP, PO2, and PVR of the mice. For WT, n = 9; for LLC1, n = 10. (E) Representative photomicrographs of vWF (brown)–stained and α-SMA (violet)–stained tissues, followed by (F) quantification of pulmonary vascular muscularization (given as a percentage for each range of vessel sizes). Scale bars, 20 μm. (G) Representative photomicrographs of elastica van Gieson–stained tissues, followed by (H) quantification of medial wall thickness of pulmonary vessels. Scale bars, 20 μm. (I) Representative photomicrographs of Sirius red staining (right). Scale bars, 20 μm. Quantitative image analysis of Sirius red staining (collagen deposition) in RV (left). For WT, n = 9; for LLC1, n = 10. *P < 0.05, ***P < 0.001 compared with WT; §P < 0.05, §§P < 0.01, §§§P < 0.001 compared with nontumor tissue of LLC1.

In line with these hemodynamic changes, an increase in the number of fully muscularized vessels and a decrease in the number of nonmuscularized vessels were noted in the small vessels (<70 μm) of the tumor tissues of the Sftpc-cRaf-BxB mice (Fig. 2, G and H). Moreover, Sftpc-cRaf-BxB mice exhibited increased medial wall thickness of vessels of all sizes in the tumor regions compared with the nontumor regions and with WT mice (Fig. 2, I and J). We also noted enhanced collagen deposition in the RV (Fig. 2K).

PH in KRasLA2 transgenic and LLC1 mice

In KRasLA2 mice, considerable tumor load developed over a 5-month period (Fig. 3, A and B). Echocardiographic follow-up of these mice between ages 3 and 5 months demonstrated decreased TAPSE and CO and increased RVID, RV IVRT/RR, and RV MPI compared with WT littermates (Fig. 3C). Moreover, RVSP and PVR were increased and PO2 decreased after the 5-month period, in the absence of SAP changes (Fig. 3D). In line with these echocardiographic and hemodynamic measurements, KRasLA2 mice exhibited a shift from nonmuscularized to partially and fully muscularized small vessels in the tumor-bearing lung areas (Fig. 3, E and F), as well as increased medial wall thickness in vessels of all sizes (Fig. 3, G and H), and enhanced RV fibrosis (Fig. 3I).

Intravenous injection of LLC1 tumor cells into mice resulted in substantial lung cancer formation within 18 days (Fig. 4, A and B). Echocardiographic analysis of LLC1 tumor cell–injected mice from day 10 to day 18 and catheter measurements at day 18 demonstrated the same profile of PH and compromised RV function as observed in the Sftpc-cRaf-BxB and the KRasLA2 transgenic mice, with some additional decrease in SAP (Fig. 4, C and D). Similarly, a shift from nonmuscularized to partially and fully muscularized small vessels in the tumor-bearing lung areas (Fig. 4, E and F), increased medial wall thickness in vessels of all sizes (Fig. 4, G and H), and enhanced RV fibrosis (Fig. 4I) were again noted.

Accumulation of macrophages and T lymphocytes in lung cancer–associated PH and its role in lung vascular remodeling

To study the putative contribution of immune and inflammatory cells to the development of PH in lung cancer, we evaluated the number of macrophages (CD68+), dendritic cells (CD209+), and T lymphocytes (CD3+) in explanted human lung cancer tissues. Immunohistochemistry revealed that perivascular T lymphocytes were scant in the nontumor sections of AC- or SCC-bearing lung tissue (Fig. 5A). In contrast, markedly increased numbers of T lymphocytes were found in pulmonary vessels of all sizes in the tumor areas of these lung tissues (Fig. 5A). Similarly, the total number of CD68+ cells was increased in tumors compared with nontumor areas of the lung cancer tissues (Fig. 5B). In contrast, the numbers of CD209+ cells were not altered in small, medium, or large pulmonary vessels in AC and SCC lung tissues (fig. S1, A and B).

Fig. 5. Role of immune and inflammatory cells in lung cancer–associated PH.

Representative photomicrographs and quantification of (A) CD3+ (lymphocytes) and (B) CD68+ (macrophages) cells in pulmonary vessels from nontumor (N) and tumor (T) SCC and AC tissues. Scale bars, 20 μm; n = 7. *P < 0.05, **P < 0.01, compared with nontumor tissues of SCC or AC. NSG mice were injected with saline (NSG-control), LLC1 (NSG-LLC1) cells, or A549 (NSG-A549) cells. Twelve and 20 days after injection, the NSG-control (n = 7), NSG-LLC1 (n = 9), and NSG-A549 (n = 9) lung tumor–bearing mice underwent echocardiographic and physiological measurements. (C) Representative photographs of mouse lungs. Scale bars, 2 mm. (D) Representative echocardiographic images and echocardiographic measurements of (E) NSG-A549 and (F) NSG-LLC1 lung tumor–bearing mice. Physiological measurements (RVSP and PO2), RV hypertrophy [RV/(LV + S)], and PVR of (G) NSG-A549 and (H) NSG-LLC1 lung tumor–bearing mice.

To investigate the role of immune and inflammatory cells in cancer-driven PH, A549 and LLC1 cells were intravenously injected into the severely immunodeficient NOD.Cg-Prkdcscid Il2rgtm1Wjl/SzJ (NSG) mice. Both approaches resulted in substantial lung cancer formation (Fig. 5C), but the development of PH was virtually fully blocked and no deterioration of RV function was detected by echocardiographic and hemodynamic measurements in the NSG-LLC1 and NSG-A549 mice (Fig. 5, D to H). In addition, there was no change in the medial wall thickness of vessels of all sizes (fig. S2, A and B).

To investigate whether macrophages or lymphocytes alone, or cocultures of tumor cells with either macrophages or lymphocytes, could directly affect the pulmonary vascular cell phenotype, we established corresponding in vitro models. We exposed pulmonary arterial smooth muscle cells (PASMCs) or pulmonary arterial adventitial fibroblasts (PAAFs) isolated from normal human donor lung tissue to conditioned medium (CM) derived from AC tumor cell/macrophage cocultures, which resulted in increase in the proliferation and migration of these cells (Fig. 6A). CM from AC tumor cell/lymphocyte cocultures resulted in enhanced PASMC and PAAF proliferation but not migration (fig. S3A). To more broadly assess the relevance of tumor cell–immune cell interactions to human PASMC and PAAF behavior, we performed corresponding studies in two additional human AC lines (H1299 and A427) and one SCC line (H226). Notably, the CM of all lung cancer cell lines increased PASMC proliferation upon coculture with macrophages and lymphocytes (fig. S3, B and C).

Fig. 6. Tumor cell and immune cell cross-talk driving pulmonary vascular remodeling.

(A) Proliferation and migration of human donor PASMCs and PAAFs subjected to CM and derived from AC tumor cells (A549), macrophages, or AC tumor cells cocultured with macrophages as indicated. (B) Cytokine array of different CM as indicated. Strongly regulated cytokines are highlighted. 1, IL-8; 2, CCL2; 3, GM-CSF; 4, IL-1RA; 5, CCL3; 6, CCL5. (C) Relative amounts of protein for each cytokine detected; n = 2. (D) Schematic diagram of the triple transgenic mice used to down-regulate IKK2 (Sftpc-cRaf-BxB/Ikk2DN) in epithelial cells. (E) Representative FACS images and (F) histological quantification of macrophage (F4/80 and CD68 antibodies for FACS and immunostaining, respectively) and T lymphocyte (CD3 antibody for both FACS and immunostaining) composition in mouse lung tissues. (G) Representative photomicrographs of elastica van Gieson–stained tissues (right), followed by quantification of medial wall thickness of pulmonary vessels (left). Scale bars, 20 μm; n = 3. **P < 0.01 compared with Sftpc-cRaf-BxB.

To identify the mechanism driving the proliferation of PASMCs and PAAFs when exposed to coculture CM, we used a cytokine array to screen for secreted mediators in AC/SCC–tumor cell–macrophage cocultures and control CM. Six cytokines [chemokine ligand 2 (CCL2), CCL3, CCL5, interleukin-1 receptor antagonist (IL-1RA), IL-8, and granulocyte-macrophage colony-stimulating factor (GM-CSF)] were substantially up-regulated under coculture conditions (Fig. 6, B and C, and fig. S4A). In silico analysis identified nuclear factor κB (NF-κB) binding sites in all six regulated cytokines (fig. S4B). To investigate the role of tumor cell–derived NF-κB–dependent cytokine expression in the development of lung vascular remodeling and PH in lung cancer, we used a mouse model in which IKK2 (required for activation of the classical NF-κB pathway) (17) is constitutively down-regulated in Sftpc-expressing AT II cells (Sftpc-rtTA/Tet-O-Ikk2DN) and crossed these mice with Sftpc-cRaf-BxB mice, thus causing NF-κB blockade in lung tumor cells (Fig. 6D). In these mice, the numbers of total and perivascular CD3+ T lymphocytes and CD11c F4/80+ macrophages were markedly reduced, supporting an important role of NF-κB in attracting immune cells into the lung and pulmonary vasculature (Fig. 6, E and F). Moreover, we observed tumor reduction and a far-reaching blockade of PH development such that the increase in medial wall thickness was virtually suppressed to WT levels under these conditions (Figs. 2J and 6G and fig. S5).

Lung vascular PDE5 up-regulation in lung cancer–associated PH and therapeutic intervention by PDE5 inhibition

Investigating the link between lung cancer–induced up-regulation of inflammatory cytokines and the lung vascular cell hyperproliferative response, we found that CCL2, CCL5, and GM-CSF all up-regulated phosphodiesterase 5 (PDE5) expression in both human PASMCs and PAAFs (Fig. 7, A and B). In corroboration, PDE5 immunoreactivity was intensified in pulmonary vessels embedded in the tumor compared with nontumor areas of human lung cancer–bearing lungs (Fig. 7C). Moreover, these cytokines, as well as IL-8, all induced human PASMC or PAAF migration and resistance to apoptosis, each of which was attenuated by the PDE5 inhibitor sildenafil (Fig. 7, D and E).

Fig. 7. Role of PDE5 up-regulation in lung cancer–associated PH.

Expression of PDE5 mRNA in (A) PASMCs and (B) PAAFs that were stimulated with basal medium (BM) or indicated cytokines for 24 hours (n = 4). (C) Representative photomicrographs of PDE5 (green), α-SMA (red), and DAPI (nuclei, blue) in pulmonary vessels from human lung tissue without (Donor) or with lung cancer. Scale bars, 100 μm. (D) Migration and (E) apoptosis of human PASMCs and PAAFs that had been stimulated with BM or indicated cytokines in the absence or presence of the PDE5 inhibitor sildenafil; n = 3. *P < 0.05, **P < 0.01, ***P < 0.001 compared with BM; §P < 0.05, §§P < 0.01 compared with cytokine stimulation in the absence of sildenafil. C57BL/6 mice were treated with sildenafil or vehicle and intratracheally instilled with LLC1 cells. (F) Echocardiographic measurements. (G) Physiological measurements. For WT, n = 5; for LLC1 treated with vehicle or sildenafil, n = 8 each. (H) Representative photomicrographs of elastica van Gieson–stained tissues (bottom), followed by quantification of medial wall thickness of pulmonary vessels (top). Scale bars, 20 μm; n = 4. §P < 0.05, §§P < 0.01 compared with vehicle-treated LLC1 mice.

To evaluate the therapeutic efficacy of PDE5 inhibition in vivo, we injected LLC1 cells into mice, followed by treatment with sildenafil or vehicle alone. Notably, sildenafil treatment improved all echocardiographic, hemodynamic, and morphometric (medial wall thickness) variables related to lung vascular remodeling and PH associated with lung cancer development (Fig. 7, F to H).

DISCUSSION

The present study provides evidence that lung vascular remodeling and PH frequently accompany lung cancer with chemokine release due to tumor cell–immune cell cross-talk linked with perivascular accumulation of immune and inflammatory cells. This concept, summarized in the diagram in fig. S6, is based on the findings that (i) PA enlargement indicating PH was observed in 250 of 519 patients with lung cancer; (ii) 32 of 70 lung cancer patients had elevated PASP values on echocardiography; (iii) extensive pulmonary vascular remodeling was consistently observed on histology in human lung cancer tumor tissues; (iv) three mouse models of lung cancer reproduced PH features in vivo; (v) immune and inflammatory cell deficiency completely protected against PH development in NSG xenograft lung cancer models; (vi) tumor cell cross-talk with inflammatory cells and release of several cytokines promoted lung vascular cell proliferation and migration, which are typical phenotypic features of PH, in vitro; (vii) genetic inhibition of NF-κB–driven chemokine generation in lung tumor cells largely suppressed the accumulation of inflammatory cells, lung vascular remodeling, and the development of PH in the cRaf-BxB model in vivo; and (viii) lung vascular up-regulation of PDE5, a well-known downstream event in PH pathogenesis, was demonstrated in human and mouse lung cancer, enabling effective PH treatment by PDE5 inhibition in the LLC1 mouse lung cancer model.

Several studies have demonstrated that CT measurements of pulmonary arterial enlargement correlate with invasively documented PH (1820). At a cutoff PA diameter of 28 mm (21), 48% of the 519 lung cancer patients showed evidence of PH. For comparison, the normal reference value of mean PA diameter in the community-based Framingham heart study was 25.1 ± 2.8 mm (22). Of the 70 patients who had available echocardiographic data, the presence of PH was further confirmed in >45% when a cutoff PASP value of ≥34 mmHg (16) was used. In contrast, a recent large-scale population-based study on 3381 participants with a mean age of 76 years (23) demonstrated a mean PASP value of 26.3 ± 7.0 mmHg.

These findings were corroborated by the histological documentation of substantial lung vascular remodeling in all human lung cancer tissue samples (AC and SCC). Moreover, lung vascular remodeling with consistent media hypertrophy and loss of nonmuscularized small vessels, which are typical features of pulmonary arterial hypertension, was reproduced in the three mouse lung cancer models. Although the kinetics of development of an overt lung cancer burden varied widely (2 weeks to 9 months), both the echocardiographic and the hemodynamic findings were very similar for all models—near doubling of the PVR with impairment of RV function and reduction in CO in the presence of tumors. In all models, the increase in RV afterload resulted in RV hypertrophy and enhanced myocardial collagen deposition. The systemic vasomotor regulation apparently adapted to the reduction in CO, with SAP remaining constant in the longer-term KRasLA2 and cRaf models, whereas some SAP decrease was noted in the more rapidly developing LLC1 model. In our study on various mouse lung cancer models, echocardiographic evaluation of the function of RV demonstrated right ventricular dilatation (RVID) along with decreased CO; this profile corresponded to that induced by hypoxia in mice. This finding indicated that, in this species, even a moderate increase in PA pressure induced a probable progressive RV dysfunction (maladaptation) that limits further increase in PASP through pulmonary vascular remodeling. Along the same line, clinical studies on patients with PH suggested that the RV initially compensates for increased afterload through adaptive hypertrophy and remodeling; however, when a high afterload is sustained, the RV dilates and fails (24). Notably, the prognosis of PH patients was strongly correlated with the function of the RV (25, 26).

Several mechanisms can explain the PH development in both the lung cancer patients and the mouse models. The first possibility is that tumor cells occlude the vessel lumen, as has been described in case reports (14). However, such occlusion was not noted in any of the extensive histological examinations of both human and mouse lungs in the present study. Another possibility is that the tumors cause chronic thromboembolism. Again, the histological workup in the present study did not provide evidence for such events, and moreover, none of the charts of the cancer patients undergoing CT for lung cancer assessment provided evidence for major deep venous thrombosis and/or lung thromboembolism. It is also possible that the effects are due to concomitant underlying disease, such as COPD, which often accompanies lung cancer. Some COPD was noted in 62% of the lung cancer patients undergoing CT. However, there was no correlation between the presence or absence of COPD and/or the COPD stage and the PA diameter. Moreover, none of the lung cancer mouse models showed evidence of chronic bronchial abnormalities or bronchoconstriction. Another possibility is that the lung vascular remodeling is driven by hypoxia. Both in the human and the mouse lung sections, the vascular remodeling was largely restricted to the areas with tumor cell infiltration. Ventilation is expected to be markedly reduced in these areas, depending on the extent of tumor cell infiltration, probably resulting in local hypoxia, which drives initial hypoxic vasoconstriction and subsequent prolonged lung vascular remodeling (27, 28). Accordingly, a decrease in arterial oxygenation was noted in all mouse lung cancer models. In the patient samples, however, nasal oxygen dependency, which indicates severe hypoxia, was a very rare event. Moreover, the mean PO2 values did not differ between the PASP ≥34 mmHg and PASP <34 mmHg groups (P < 0.608). Although the mean PO2 value of 65 mmHg in the PA ≥28 mm group was less than that in the PA <28 mm group (P < 0.001), it was not expected to drive hypoxia-induced PH. Therefore, the extent of contribution of hypoxia-driven lung vascular remodeling to the development of PH in lung cancer could not be definitively determined and may depend on the mode of tumor cell infiltration into lung tissue. Finally, it is possible that the inflammatory processes in the tumor microenvironment drive PH development. Inflammation plays a critical role in tumorigenesis, and an inflammatory microenvironment promotes tumor growth (29, 30). Substantial lung parenchymal infiltration with immune and inflammatory cells has also been demonstrated for human lung cancer (31). Moreover, recruitment of inflammatory cells to the perivascular space in conjunction with inflammatory processes has generally been suggested as one major pathomechanism underlying PH of different origins (3236). We observed drastically increased numbers of macrophages and lymphocytes in the perivascular compartment in explanted human lung cancer tissues. PH development was virtually fully blocked in the tumor cell–injected NSG mice, which lacked functional T cells, B cells, natural killer cells, functional macrophages, and dendritic cells (37, 38). These data lend strong credibility to the idea that the inflammatory microenvironment in lung cancer is the major driving force for PH development.

Corroborating this conclusion, we demonstrated that coculture of macrophages with human AC cells resulted in strong release of cytokines such as chemokine ligands (CCL2, CCL3, and CCL5), interleukins (IL-1RA, and IL-8), and GM-CSF. Furthermore, the CM used in these cocultures caused marked increases in proliferation and migration of ex vivo isolated human PASMCs and PAAFs, typical features of the PH phenotype. Finally, in a triple transgenic mouse, blockage of NF-κB–dependent gene regulation in the lung cancer cells resulted in a substantial decrease of the accumulation of lung and perivascular lymphocytes and macrophages, concomitant with a reduction of pulmonary vascular remodeling and suppression of PH development, as well as some reduction in tumor growth. Together with the observations in the NSG mice, these findings provided further evidence that inflammatory processes in the tumor microenvironment, which are driven by the tumor cells themselves, are the main driving force of tumor-associated PH.

As shown for other variants of PH, medial wall up-regulation of PDE5 was noted in the human lung cancer samples. Moreover, several of the cytokines being released upon lung cancer cell–macrophage/lymphocyte cross-talk induced up-regulation of PDE5 in human PASMCs and PAAFs in vitro. Conversely, the PDE5 inhibitor sildenafil reduced promigratory and apoptosis-inhibitory effects of these cytokines in PASMCs and PAAFs. Furthermore, sildenafil greatly reduced vascular remodeling, the development of PH, and impairment of RV function in the LLC1 model, suggesting that therapeutic intervention at the level of the cyclic guanosine monophosphate (cGMP)–PDE5 axis may be a potential treatment strategy in lung cancer–associated PH.

There are some limitations of this study. (i) Echocardiographic diagnosis of PH in our lung cancer cohort. Although we analyzed the CT scans of all (519) lung cancer patients presenting at our center in 2011–2013, echocardiographic investigations for reliable assessment of PASP were only available for 70 of these patients. Nevertheless, among these 70 patients, 32 presented with elevated PASP values. Of note, in 2011–2013, echocardiography was not a routine investigation in lung cancer patients at our center, because we, and probably other clinicians, were not yet aware of the putative presence of PH in this population. (ii) Impact of anti-inflammatory therapies in lung cancer–associated PH. Our data strongly support the notion that immune and inflammatory cells in the tumor microenvironment represent the major driving force for PH development and can be therapeutically targeted with anti-inflammatory agents. The potential of such an approach in lung cancer–associated PH was, however, not yet explored in detail.

In conclusion, this study provided clinical, histopathological, and experimental evidence of a PH category that is associated with lung cancer. The inflammation-enhancing cross-talk between lung cancer and immune cells in the tumor microenvironment, recently shown to promote lung cancer growth (39), is suggested as the major underlying pathogenic pathway. Pulmonary vascular abnormalities may thus contribute to the symptoms presented by lung cancer patients. This constellation of symptoms will become even more relevant as recently developed therapeutics increase the life expectancy of lung cancer patients.

MATERIALS AND METHODS

Study design

The aim of this study was to demonstrate that pulmonary vascular remodeling and PH frequently occur in lung cancer patients and in mouse models of lung cancer. This objective was pursued by CT imaging and echocardiography from human lung cancer patients and pulmonary vascular morphometric analysis of human lung cancer tissues. Further, to explore whether mouse lungs display features of PH during cancer progression, we used three different mouse models of lung cancer, namely, the KRas knock-in line KRasLA2, the cRaf-BxB transgenic line Sftpc-cRaf-BxB, and LLC1, to characterize pulmonary hemodynamics and pulmonary vascular morphometric analysis in detail.

The second objective investigated was to study the putative contribution of immune and inflammatory cells to the development of PH in lung cancer both in vitro and in vivo. The third objective was to investigate the role of PDE5 in lung cancer–associated PH and therapeutic intervention by PDE5 inhibition. To achieve these objectives, T lymphocyte and macrophage numbers [determined by fluorescence-activated cell sorting (FACS) and immunohistochemistry] were quantified in a blinded fashion. mRNA expression, cytokine arrays, proliferation, migration, and apoptosis assays upon coculture were investigated as predefined end points. To study the in vivo contribution of immune and inflammatory cells in cancer-driven PH, we used two different mouse models of lung cancer, namely, the NOD.Cg-Prkdcscid Il2rgtm1Wjl/SzJ (NSG) mice and Sftpc-rtTA/Tet-O-Ikk2DN, to characterize pulmonary hemodynamics and perform pulmonary vascular morphometric analysis in detail. To evaluate the therapeutic efficacy of PDE5 inhibition in vivo, we injected LLC1 cells into mice, followed by treatment with sildenafil or vehicle alone.

In vivo studies were carried out in two to three groups of mice on the basis of previous experience with comparable work (39). No randomization of animal groups was performed. The pulmonary vascular morphometric analysis throughout the study was performed in a blinded fashion. Details of sampling and experimental replicates are provided in each figure legend.

CT and echocardiography examinations in patients with lung cancer

Axial CT was used to examine the PA for enlargement. In accordance with previous studies (18, 40, 41), we measured the diameter of the main PA at the level of its bifurcation and that of the ascending aorta at its widest point using the same image for both. The measuring point on the main PA was at the origin of the bilateral PAs, which was visible on axial CT in most patients because the angles of the PA outline changed sharply at this point (20, 22). Estimation of PASP by echocardiography was performed as previously detailed (23).

Human lung specimens

Lung tumor tissues were obtained from the Universities of Giessen and Marburg Lung Center (UGMLC) tissue bank (table S1). The study protocol for tissue donation was approved by the Ethics Committee of the Department of Human Medicine of Justus Liebig University Hospital, Giessen, Germany, in accordance with national law and with the “Good Clinical Practice/International Conference on Harmonisation” guidelines. Written informed consent was obtained from each patient or the patient’s next of kin (AZ 31/93).

Animal experiments

Local authorities (Regierungspräsidium Darmstadt, Hessen, Germany) approved all animal studies. All mice were kept under specific pathogen–free conditions and handled in accordance with the guidelines of the European Union Commission on Laboratory Animals. C57BL/6 mice (7 to 8 weeks old), NSG mice (7 to 8 weeks old), and the transgenic lines KRasLA2 (3 to 5 months old) and Sftpc-cRaf-BxB (7 to 9 months old) were used for these experiments. C57BL/6 and KRasLA2 mice were purchased from Charles River Laboratories or Jackson Laboratory.

Injection of tumor cells

LLC1 cells (1 × 106) were injected into male C57BL/6 mice into the tail vein. At day 18, mice were sacrificed after hemodynamics analysis, and their lungs were harvested, immersed in formalin, and histologically examined. In other experiments, LLC1 (1 × 106) and A549 (1 × 106) cells were injected into NSG mice. After hemodynamics analysis on day 12 (LLC1) or day 20 (A549), the mice were sacrificed; their lungs were harvested, immersed in formalin, and histologically examined.

KRasLA2 mouse model

We used a mouse model derived from spontaneous somatic activation of the latent KRasLA2 allele, which develops alveolar adenomatous hyperplasia 1 week after birth (42).

Sftpc-cRaf-BxB and Sftpc-rtTA/Tet-O-Ikk2DN/Sftpc-cRaf-BxB transgenic mice

Sftpc-cRaf-BxB transgenic mice were maintained as hemizygotes in the C57BL/6 mouse strain background. They express a truncated cRaf lacking the regulatory domain under the control of the Sftpc promoter, resulting in lung-targeted expression of cRaf-BxB in AT II cells and formation of lung adenomas as early as 1 to 2 months of age (43).

For the generation of the compound triple transgenic mice (Sftpc-rtTA/Tet-O-Ikk2DN/Sftpc-cRaf-BxB), Sftpc-cRaf-BxB mice were crossbred with Sftpc-rtTA/Tet-O-Ikk2DN mice. Transgenes were expressed conditionally in AT II cells upon feeding the mice food supplemented with doxycycline. Expression was induced in the compound mice for 50 to 56 weeks.

Cell culture

Mouse LLC1 and human AC (A427, A549, and H1299) and SCC (H226) cell lines were obtained from the American Type Culture Collection and cultured in sterile tissue culture flasks in an incubator with 5% CO2 at 37°C and a water vapor–saturated atmosphere. For cell culture, RPMI 1640 or Dulbecco’s modified Eagle’s medium/F12 medium was supplemented with 10% fetal calf serum (FCS) and 1% penicillin (10,000 U/ml stock solution) and streptomycin (10 mg/ml stock solution).

Lymphocyte and macrophage isolation and cocultures

Human macrophages were generated from blood monocytes. Briefly, peripheral blood leukocytes were prepared from buffy coats obtained from the UGMLC blood bank. Cells were then separated by density gradient using Ficoll-Paque PLUS (GE Healthcare). Platelets present in the enriched monocyte fraction were removed by three washing steps in phosphate-buffered saline (PBS) with 2% FCS. Finally, monocytes were seeded in RPMI containing 10% FCS, 4 mM l-glutamine, and penicillin/streptomycin at a concentration of 5 × 105 cells/ml in six-well tissue culture plates for 10 days in the presence of 2.5% human serum. Human peripheral blood lymphocytes were isolated from buffy coats using the Ficoll-Paque reagent (Amersham) gradient centrifugation, followed by counterflow elutriation.

Direct coculture of tumor cell lines (A549, A427, and H226) and lymphocytes/macrophages was performed in six-well plates; in addition, tumor cells and lymphocytes were seeded separately as controls. All coculture was performed in RPMI 1640 supplemented with 1% FCS. After 24 hours, the supernatant (CM) was harvested, centrifuged, filtered, and stored at −80°C for further experiments, as described (39).

PASMC isolation and culture

Primary human PASMCs and PAAFs were isolated from human PAs (<2 mm in diameter) obtained from control donors (n = 4 each) (34). PASMCs and PAAFs were used between passages 4 and 8. A primary culture of human PASMCs was also obtained from Lonza (CC-2581) and grown in SmGM-2 Bulletkit medium (Lonza). Cells were maintained at 37°C in a 5% CO2 incubator. Experiments were performed with cells at passages 4 to 6.

Proliferation assays and in situ cell detection assay

Proliferation assays were performed with serum-starved primary human PASMCs and PAAFs exposed for 24 hours to tumor cell CM, lymphocyte CM, macrophage CM, cocultured CM, BM, IL-8 (10 ng/μl), CCL5 (10 ng/μl), CCL2 (25 ng/μl), GM-CSF (25 ng/μl), and sildenafil (50 μM) using a bromodeoxyuridine colorimetric cell proliferation enzyme-linked immunosorbent assay (ELISA) and In Situ Cell Death Detection kit (Roche).

In silico analysis of NF-κB binding site in cytokine gene promoters

The putative promoter in the 5′-flanking regions of six human cytokine genes (CCL2, CCL3, CCL5, IL1RA, IL8, and GMCSF) was analyzed using ConSite tool (http://consite.genereg.net) to identify the sites of transcription factor binding. The NF-κB binding sites were predicted/screened in the region located 3000 bp upstream of the first exon of the corresponding gene. ConSite profile was extracted from the JASPAR database, which is an open-access, nonredundant collection of curated profiles. The location of the NF-κB binding sites is depicted in a schematic diagram in fig. S4B.

Cytokine array and ELISA

The Human XL Cytokine Array Kit (R&D Systems) was used to assess cytokine secretion. Briefly, nitrocellulose membranes with spotted capture antibodies were blocked and then incubated with 250 μl of CM from tumor cells, macrophages, or coculture mixed with a cocktail of biotinylated detection antibodies, as described (39).

Complementary DNA synthesis and quantitative PCR

Primary human PASMCs and PAAFs in culture were exposed for 24 hours to BM, IL-8 (10 ng/μl), CCL5 (10 ng/μl), CCL2 (25 ng/μl), or GM-CSF (25 ng/μl), and total cellular RNA was extracted using Tri Reagent (Biozol). Next, 800 ng of total RNA was reverse-transcribed using the ImProm-II Kit (Promega). Real-time polymerase chain reaction (PCR) was performed on the Stratagene Mx3005P qPCR system with the MxPro software and the program “SYBR Green with Dissociation curve” to confirm the specificity of each primer. PCR products were separated on a 3% agarose gel to confirm that every primer yielded one specific product. All primer pairs were designed with the NCBI (National Center for Biotechnology Information) primer blast tool (www.ncbi.nlm.nih.gov/tools/primer-blast/) and purchased from Metabion GmbH. Real-time PCR was performed with the Platinum SYBR Green qPCR SuperMix-UDG Kit (Invitrogen, Life Technologies) and the following conditions: 10 min at 95°C, followed by 40 cycles of 30 s at 95°C, 1 min at 58° to 60°C, and 30 s at 72°C. Analysis was done using the MxPro software and GraphPad Prism. The following primers were used: PDE5A (NM_001083), 5′-GAAAAGGACTTTGCTGCTTA-3′ (forward) and 5′-TGATTTTGTTTGCATCATGT-3′ (reverse); HPRT1 (NM_000194.1), 5′-TGACACTGGCAAAACAATGCA-3′ (forward) and 5′-GGTCCTTTTCACCAGCAAGCT-3′ (reverse).

Echocardiography

Mice were anesthetized using isoflurane (3%) delivered using a vaporizer (VisualSonics). After induction, animals were maintained on isoflurane anesthesia (1.0 to 1.5%). Body temperature was monitored using a rectal thermometer (Indus Instruments). The chest of each mouse was shaved and treated with a chemical hair remover to enhance ultrasound quality. To provide a coupling medium for the transducer, a prewarmed ultrasound gel was spread over the chest wall. Transthoracic echocardiography was performed with VEVO2100 and VEVO1100 systems (VisualSonics) to measure RVID, TAPSE, and CO, as described (44). The RV MPI or Tei index was measured in mice by tissue Doppler imaging (TDI) at the lateral part of the tricuspid annulus in the four-chamber view and calculated as the sum of isovolumic contraction time and IVRT divided by ejection time (45). IVRT was measured as the time interval between end of PA outflow and the onset of tricuspid inflow by TDI. IVRT was normalized to echocardiography-derived RR interval by using the following formula: IVRT/RR = (IVRT/RR interval)*100. PVR was calculated using the following formula: PVR = (80/3) × RVSP/CO (46).

Hemodynamic and RV hypertrophy measurements

Hemodynamic parameters and RV hypertrophy were assessed as described (44). Briefly, 20 min after an intraperitoneal injection of heparin (2500 IU/kg), anesthesia was induced intraperitoneally with ketamine hydrochloride (60 mg/kg) and xylazine hydrochloride (5 mg/kg). In deep anesthesia, mice were ventilated through a custom-made tracheal tube connected to a ventilation pump, positioned via tracheotomy (MiniVent Type 845, Hugo Sachs Elektronik).

Core body temperature was maintained at constant 37°C using a water bath pad (Keutz Labortechnik). SAP was monitored by cannulating the left carotid artery with a polyethylene cannula (Micro Cannulation System, FST GmbH) connected to a fluid-filled force transducer (Braun). The RV was catheterized via the right jugular vein with a silicone catheter (Hyman Mouse Pressure, Catheter 14 cm, Nu MED Inc.). The transducers were calibrated before every measurement. After completion of hemodynamic measurements, the murine lungs were first flushed with a mixture of 0.9% sterile saline solution and 3.5 to 3.7% formalin (1:1) at a constant pressure of 22 cmH2O above the pulmonary hilum to remove blood. Right lobes were separated and used together with the left lobe for histologic purposes. The RV was dissected from the LV + S, and the samples were weighed to obtain the RV/(LV + S) ratio as a measure of RV hypertrophy.

Hematoxylin and eosin staining

Tissue sections were deparaffinized in xylene and then rehydrated in 100, 90, and 70% ethanol in distilled water. The slides were then incubated in fresh hematoxylin (Invitrogen Corporation) for 20 min, washed in distilled water, incubated in acidified eosin solution (Richard-Allan Scientific) for 4 min, and then washed in distilled water. Finally, the slides were dehydrated in 90 and 100% ethanol, air-dried, and mounted.

Histology and pulmonary vascular morphometry

Lung lobes were perfused with a mixture of paraformaldehyde (3.5 to 3.7% in PBS) and saline (0.9%) through the PA with constant pressure of 22 cmH2O, as described (34). The lung lobes were stored in paraformaldehyde for the next 24 hours and then in PBS before being dehydrated. The lungs were then embedded in paraffin, and 3-μm-thick sections were generated. The degree of muscularization of small peripheral PAs was assessed by double staining the sections with an α-SMA antibody (1:900 dilution; clone 1A4, Sigma) and a human vWF antibody (1:1200 dilution; IS527, Dako). Sections were counterstained with methyl green and examined by light microscopy, as described (44). The color along the perimeter of the vessel was analyzed by light microscopy using a computerized morphometric system (Q Win V3, Leica) that differentiates the purple staining of the smooth muscle and the brown staining of the endothelial layer. In murine lungs, we counted a total of 100 vessels per lung; 85 vessels with a diameter of 20 to 70 μm, 10 vessels with a diameter of 71 to 150 μm, and 5 vessels with a diameter of >151 μm were analyzed at ×40 magnification by an observer blinded to the treatment groups. This counting procedure was followed in the tumor area and in the nontumor part of the lung section. In the human lungs, we counted a total of 220 vessels per lung: 85 vessels with a diameter of 20 to 50 μm, 20 vessels with a diameter of 51 to 150 μm, and 5 vessels with a diameter of >151 μm.

Each vessel was categorized as nonmuscularized (<5% SMA around the vessel), partially muscularized (5 to 75% SMA around the vessel), or fully muscularized (>75% SMA around the vessel) (47). The percentage of pulmonary vessels in each muscularization category was determined by dividing the number of vessels in that category by the total number counted in the same experimental group.

To assess the medial wall thickness, 3-μm sections were stained with Elastin Nuclear Fast Red. Medial wall thickness was defined as the distance between the lamina elastica interna and lamina elastica externa and was expressed as a percentage of the total external diameter of the vessel. The analysis was performed by light microscopy using a computerized morphometric system (QWin, Leica) and was calculated by the following formula: %MWT = (2 × medial wall thickness/external diameter) × 100.

Analysis of collagen content of the RV

Freshly dissected RV tissues were fixed in 3.5 to 3.7% formalin overnight, then stored in 1× PBS before getting dehydrated, embedded in paraffin, and sectioned (3 μm). To detect collagen fibers, all sections were stained with 0.1% Sirius red F3B (Niepoetter) in picric acid (AppliChem GmbH). Photomicrographs were quantified to determine the interstitial collagen fraction using Leica QWin V3 computer-assisted image analysis software (Leica Microsystems).

Immunohistochemistry of T lymphocytes, dendritic cells, and macrophages

Formalin-fixed, paraffin-embedded blocks of lung tissue from patients were mounted on positively charged glass slides (R. Langenbrinck). From each tissue block, sections of 3-μm thickness were obtained. Antibodies against the following proteins were used: macrophage CD68 [1:100 dilution; Calbiochem (EMD Millipore)], dendritic cell CD209 (DC-SIGN) (1:10 dilution; BD Biosciences), and T lymphocyte CD3 (human: 1:150, mouse: 1:400; Biocare Medical LLC). Incubation with each antibody was carried out overnight at 4°C. Afterward, antibody-stained sections were washed in PBS, and antibody binding was determined using an ImmPRESS reagent kit (Vector Laboratories). After extensive washing, sections were incubated with the NovaRED substrates (Vector Laboratories). Positive cells were counted as described (31, 34).

Immunofluorescence staining

Double immunofluorescence staining was performed with primary antibodies to PDE5A (1:100 dilution; ab14672, Abcam Ltd.) and α-SMA (1:400 dilution; A2547, Sigma). After overnight incubation, slides were washed and incubated with the secondary antibodies Alexa 488– and Alexa 555–conjugated goat anti-rabbit immunoglobulin G (1:1000; Molecular Probes), respectively, for 1 hour. All sections were counterstained with nuclear 4′,6-diamidino-2-phenylindole (DAPI; 1:1000) and mounted with fluorescence mounting medium (Dako) (48).

Lung tumor digestion and FACS analysis

To prepare lung tumors for flow cytometric analysis of infiltrating macrophages and T lymphocytes, extracted lung tumors were placed in 35 mm × 10 mm Petri dishes (Greiner Bio-One) and cut into small cubes (<1 mm3) with a scalpel. A digestion solution (2 ml) of collagenase D (0.2 mg/ml) (Roche Diagnostics), pronase (1 mg/ml) (Roche Diagnostics), and 2 μl of deoxyribonuclease I (400 U/ml, Promega) was added, and samples were incubated for 40 min at 37°C on a shaker. The digested tissue was resuspended in 10 ml of PBS with 10% FCS to stop the digestion reaction and centrifuged for 5 min at 1600 rpm. The pellet was washed with PBS and passed through a 40-μm cell strainer (BD Biosciences), and cells were counted. The cell suspension was then centrifuged, resuspended in 1 ml ice-cold 1% paraformaldehyde per 106 cells, and fixed for 15 min on ice. After centrifugation, cells were resuspended in FACS buffer [PBS, 0.15% EDTA, 10% FCS (pH 7.2)] with a macrophage marker (F4/80-APC, BD Biosciences) and a T lymphocyte marker (CD3-FITC, BD Biosciences) and analyzed by flow cytometry. EDTA solution was purchased from Biochrom AG.

Sildenafil treatment

LLC1 cells (1 × 106) were injected into male C57BL/6 mice intravenously by the tail vein. At 4 days after injection, mice that had developed tumors were treated with oral sildenafil (100 mg/kg body weight) daily. At day 18, mice were scanned with echocardiography, and hemodynamic measurements were performed.

Statistics

Statistical analyses were performed with the GraphPad Prism 7 software following the manufacturer’s guidelines. One-way analysis of variance (ANOVA) followed by Student-Newman-Keuls multiple comparison test was used to compare the means of more than two independent groups. For echocardiographic analysis, we used two-way ANOVA. Two independent groups were compared with Student’s t test. Data are expressed as means ± SEM.

SUPPLEMENTARY MATERIALS

www.sciencetranslationalmedicine.org/cgi/content/full/9/416/eaai9048/DC1

Fig. S1. Dendritic cell recruitment in the pulmonary vasculature of human lung cancer tissues.

Fig. S2. Role of immune/inflammatory cells in lung cancer–associated vascular remodeling.

Fig. S3. Induction of pulmonary vascular cell proliferation by CM from cocultures of macrophages/T lymphocytes with lung cancer cells.

Fig. S4. Secreted factors in the CM of macrophage/SCC cocultures.

Fig. S5. Reduction of tumor growth by IKK2 down-regulation in epithelial cells.

Fig. S6. Proposed mechanism of PH development in lung cancer.

Table S1. Demographic data, tumor type, and tumor stage for samples undergoing histological examination.

REFERENCES AND NOTES

  1. Acknowledgments: We thank V. Golchert, Y. Knepper, U. Eule, E. Bieniek, and N. Wilker for excellent technical assistance; Pullamsetti laboratory members S. Dabral, P. Chelladurai, and E. Gamen for providing vascular cells; and C. Samakovlis for helpful comments and discussions. We thank T. Wirth for providing Tet-O-Ikk2DN mice. Funding: This work was supported by the Max Planck Society, Excellence Cluster Cardio-Pulmonary System, Verein zur Förderung der Krebsforschung in Gießen e.V., a Von-Behring-Röntgen-Stiftung grant, a Rhön Klinikum AG grant, a LOEWE UGMLC grant, and the German Center for Lung Research (DZL). Author contributions: S.S.P., W.S., and R.S. planned and initiated the project, designed experiments, wrote the manuscript, and supervised the entire project; B.K. and S.S. conducted the experiments and analyzed the data; Y.S. performed cell culture experiments; J.W., H.G., H.A.G., S.G., W.S., F.G., and G.A.K. performed the human CT analysis; U.R.R. provided Sftpc-cRaf-BxB mice; N.E.-N. provided Sftpc-cRaf-BxB/Ikk2DN tissues, which were analyzed by A.W.; L.F. provided human lung cancer tissues and supervised human histopathological data; R.T.S. supervised animal experiments. Competing interests: The authors declare that they have no competing interests.
View Abstract

Navigate This Article