Background

Pulmonary hypertension is a common and serious complication of chronic obstructive pulmonary disease (COPD). It is associated with increased mortality and morbidity [1] and can eventually lead to right ventricular failure and death unless diagnosed early and treated appropriately [2,3,4]. In recent years, various forms of pulmonary vasoactive agents, including agents targeting prostacyclin pathway (e.g., inhaled prostacyclin analogues and prostacyclin receptor agonist), endothelin receptor antagonists, and nitric oxide pathway enhancers have been approved for the treatment of pulmonary arterial hypertension (group 1 pulmonary hypertension). An oral stimulator of the nitric oxide receptor soluble guanylate cyclase riociguat was recently approved for group 1 and 4 pulmonary hypertension [5,6,7]. However, these agents are not recommended for pulmonary hypertension due to chronic lung disease and/or hypoxia (group 3) due to lack of studies and no proven benefit in terms of survival [4, 8,9,10,11]. It is also thought that pulmonary vasodilator treatment may worsen ventilation/perfusion mismatching in COPD due to the inhibition of hypoxic pulmonary vasoconstriction [12], but there is no direct evidence to support this concept.

Pulmonary hypertension in COPD is characterized by increased pulmonary artery pressure and pulmonary vascular resistance owing to extensive pulmonary vasoconstriction and vascular remodelling, including dysfunction and proliferation of pulmonary artery cells. Studies suggest that cigarette smoking, the major environmental risk factor for COPD, can cause a series of changes in the pulmonary vasculature, including pulmonary artery cell proliferation, which may be critical for pulmonary artery remodelling in COPD- associated pulmonary hypertension [13,14,15]. Although in vitro studies have shown that cigarette smoke extract (CSE) can induce proliferation of both human and rat pulmonary artery smooth muscle cells (PASMCs) [16, 17], which may eventually lead to vascular remodelling, the effect of CSE on proliferation of human pulmonary artery endothelial cells (PAECs) and the contribution of PAECs and PASMCs, the two key cell types that play a major role in the pathophysiology of pulmonary hypertension, to cigarette smoke-induced pulmonary vascular remodelling in COPD remain unclear.

Prostanoids are a group of vasoactive lipid mediators, including prostaglandins (PGs) and thromboxane (TX), that are synthesized from membrane-derived arachidonic acid (AA). AA is converted to unstable PGH2 by the rate-limiting PGH synthase (cyclooxygenase (COX)), which exists in two isoforms, the constitutive COX-1 and the inducible COX-2 (coded by prostaglandin-endoperoxide synthase 2 (PTGS2) gene). PGH2 is subsequently converted to main biologically functional prostanoids PGI2 (prostacyclin), TXA2 and PGE2 via their respective synthases PGI-synthase (PGIS, coded by PTGIS gene), TXA synthase (TXAS, coded by TBXAS1 gene), PGE synthases (PGES). Among the three different subtypes of PGES (microsomal PGE synthase-1 (mPGES-1), mPGES-2 and cytosolic PGE synthase (cPGES)), mPGES-1 is considered the key enzyme for regulating PGE2 [18,19,20]. PGI2 and TXA2 have opposing effects (anti-proliferative/vasodilatory vs proliferative/vasoconstrictive on pulmonary vasculature). Reduced PGIS expression has been observed in the lung tissue of patients with group 1 pulmonary hypertension. Importantly, this deficiency of PGIS expression is associated with pulmonary vascular remodelling [21], suggesting that reduced PGIS expression and PGI2 release may serve as a potential marker for dysfunction of pulmonary artery cells and vascular remodelling in patients with group 1 pulmonary hypertension. Although the role of PGE2 in pulmonary hypertension is unclear, previous studies showed that circulating PGE2 levels are reduced in patients with group 1 pulmonary hypertension [22] and that the PGE2 receptor 2 (EP2) is an important negative modulator of PASMC proliferation [23], suggesting a possible role of PGE2 in the development of group 1 pulmonary hypertension. Although an imbalance between the excretion of TXA2 and PGI2 metabolites [24] and reduction of circulating PGE2 levels [22] have previously been reported in patients with group 1 pulmonary hypertension, the imbalance and its functional consequence in the development of pulmonary hypertension in COPD have not been studied.

CSE has been reported to induce COX-2 expression in different cell types including human small airway epithelial cells [25], endothelial cells such as human umbilical vein endothelial cells [26], and human pulmonary microvascular endothelial cells [27]. Although effect of CSE on TXA2 and PGE2 levels has not been previously investigated in any cell types, including PASMCs and PAECs, reduced PGIS/PGI2 has been reported in the lungs of smokers with COPD, as well as in CSE-treated human umbilical vein endothelial cells [28]. The study also showed that umbilical vein endothelial cell stimulation with CSE can induce apoptosis, whilst the use of the PGI2 analogue beraprost sodium prevents CSE-induced cell death [28]. The findings of the same group also demonstrated decreased PGI2 production and induced apoptosis in the lungs of CSE-treated emphysematous rats [29], suggesting that cigarette smoke can cause cell apoptosis and eventually vascular remodelling possibly by reducing PGIS-derived PGI2. However, the effect of CSE on the release of vasoactive prostanoids PGI2, TXA2, and PGE2 and the expression of their respective synthases PGIS, TXAS, and mPGES-1 in PAECs, particularly in vascular smooth muscle cells including PASMCs, is unknown. In addition, the involvement of these vasoactive prostanoid meditators in CSE-induced pulmonary artery cell proliferation has not been previously investigated.

In this study we explored the effect of CSE on vasoactive prostanoid synthase gene expression and mediator release and the role of imbalanced prostanoid mediator release (by the use of agents that target the imbalance prostanoid pathway in CSE-induced human PASMC and PAEC proliferation). We report for the first time that CSE induces imbalanced vasoactive prostanoid mediator release characterized by the reduced PGI2/TXA2 ratio, likely as a result of an imbalanced expression of PGIS and TXAS in both PASMCs and PAECs, and that the imbalance plays a key role in mediating CSE-induced cell proliferation in both cell types and may thereby contribute to vascular remodelling. Our results strongly suggest that the imbalanced prostanoid release, TXA2 in particular, has the potential to be a novel therapeutic target in COPD-associated pulmonary hypertension. We have previously reported some of the results of this study in the form of abstracts [30,31,32].

Materials and methods

Cigarette smoke extract (CSE) preparation

Aqueous CSE from 3R4F research-grade cigarettes (Kentucky Tobacco Research and Development Center, University of Kentucky, USA) was prepared under standardised conditions by bubbling the smoke from two cigarettes at 1 cigarette/min rate into 20 ml low-serum (0.5% foetal bovine serum (FBS)) Dulbecco’s Modified Eagle Medium (DMEM) for PASMCs or endothelial cell growth medium (ECGM) for PAECs. The smoke was drawn through the medium using a vacuum pump (Charles Austen Pumps, UK) with a constant pressure of 0.2 bars to achieve a static burning rate. The resulting CSE was filtered through a 0.22 μm pore size filter and the absorbance at 320 nm was used to measure CSE strength (1.5 = 100%) [33] prior to preparation of the desired concentrations.

Cell culture

Human PASMCs from three different healthy donors were obtained from Thermo Fisher Scientific (Waltham, MA, USA), Cell Applications (San Diego, CA, USA), and Lonza Group (Basel, Switzerland) and human PAECs, isolated from the main pulmonary artery of a healthy donor, were purchased from PromoCell (Heidelberg, Germany). PASMCs at passage 6 were cultured in growth DMEM (supplemented with 20% FBS) as previously described [34] until 90% confluence, treated with CSE in growth DMEM for 48 h and then with CSE in low serum DMEM (0.5% FBS) for 24 h. PAECs at passage 8 were cultured in ECGM (containing 2% FBS, 0.4% endothelial cell growth supplement, 0.1 ng/ml epidermal growth factor, 1 ng/ml basic fibroblast growth factor, 90 ng/ml heparin and 1 ng/ml hydrocortisone) until 90% confluence and then treated with CSE for 24 h. For cell proliferation study, cells were pre-treated with or without celecoxib, beraprost sodium and daltroban (all from Merck Life Science UK, Dorset, UK) for 1 h before being treated with CSE in growth DMEM and ECGM, respectively, for 24 h. The concentrations of CSE and drugs and treatment time used in this study were determined based on the cytotoxicity assesed by 3-(4,5-dimethylthiazol-2-yl)-2,5- diphenyltetrazolium bromide (MTT) assay (Merck Life Science UK).

Quantitative real-time RT-PCR (qRT-PCR) analysis

Confluent human PASMCs and PAECs were treated with CSE for 72 h and 24 h, respectively. Total RNA was isolated from the cells using the NucleoSpin® RNA kit following the instructions of the manufacturer (Macherey Nagel, Germany). After total RNA quantification using NanoDrop spectrophotometer (Thermo Fisher Scientific), 1 μg total RNA of each sample was reverse transcribed to cDNA with the high-capacity cDNA reverse transcription kit (Applied Biosystems, USA) [35]. Reversed transcribed cDNA was diluted (1:10 ratio) with nuclease free water. 5 μl of the cDNA sample were amplified using 1 μM primers and the KAPA SYBR® FAST qPCR Kit (Kapa Biosystems, USA). The relative quantification was calculated with the 2−ΔΔCT method using the internal control β2-microglobulin (β2M) as a reference gene. Data were expressed as fold change over untreated (0% CSE) cells. The primer sequences used are: PTGS2 (FW): 5'-AAGCAGGCTAATACTGATAGG-3' (RV): 5'-TGTTGAAAAGTAGTTCTGGG-3'; PTGIS (FW): 5'-GAAAGACTTTTACAAGGATGGG-3' (RV): 5'-ATTGTTTGATGCTGTTGACC-3'; TBXAS1 (FW): 5'-CTACTGCAATTACACCACAG-3' (RV): 5'-AAGAGTAAAACCAGGATAGGTC-3'; β2M (FW): 5'-AAGGACTGGTCTTTCTATCTC-3' (RV): 5'-GATCCCACTTAACTATCTTGG-3'.

Western blotting analysis

At the end of the experiment, the cell culture medium was collected for mediator release analysis, and human PASMCs and PAECs were lysed with RIPA buffer. After being collected and quantified for protein concentrations using a bicinchoninic acid (BCA) assay kit (Thermo Fisher Scientific), protein samples were diluted in 4 × Laemmli buffer and boiled for 10 min. 20 μg of total protein were separated with SDS-PAGE and transferred into polyvinylidene fluoride (PVDF) membrane (BioRad, USA). Next, the membrane was blocked with 5% non-fat dry milk (Santa Cruz, USA) in TBS-T (Tris-Buffered Saline with 0.1% Tween 20). The membrane was then probed with specific antibodies recognizing human forms of COX-2 (Cayman Chemical, USA), PGIS (R&D Systems, UK), TXAS (Novus Biologicals, USA), mPGES-1(Cambridge bioscience, UK), and GAPDH (Santa Cruz Biotechnology, USA). ImageJ software (National Institutes of Health, USA) was used to quantify protein band density from Western blot. The values were first normalised to the corresponding GAPDH and then to the control sample as described before [35]. Data were presented as fold changes over control.

Prostanoid analysis

PGE2, 6-keto-PGF1α (stable metabolite of PGI2) and TXB2 (stable metabolite of TXA2) released in the culture medium were quantified using respective ELISA kits (Cayman Chemical), following the manufacturer’s instructions. Data were normalized with total protein and expressed as pg/mg protein.

Cell proliferation analysis

Cell proliferation of human PASMCs (in growth DMEM) and PAECs (in ECGM) in response to CSE treatment for up to 72 h and 24 h, respectively, was assessed using the cell proliferation reagent WST-1 (water solubale tetrazolium salt 1) following the instructions of the manufacturer (Sigma-Aldrich) as described before [36]. Data were expressed as relative proliferation (% change over control).

Statistical analysis

Data were expressed as mean ± SEM. GraphPad Prism 8 was used for statistical analysis. An unpaired Student's t test was carried out to compare two data sets. ANOVA followed by a t-test was performed to compare the differences between control and specific treated groups. P < 0.05 was regarded as statistically significant.

Results

CSE differentially regulates the mRNA and protein expression of PTGS2, PTGIS, and TBXAS1

COX-2, PGIS and TXAS are vasoactive enzymes responsible for the production of their downstream prostanoid products that have potent anti-proliferative (e.g. PGE2 and PGI2) and proliferative (e.g. TXA2) properties in pulmonary vasculature. However, the effect of CSE on the mRNA expression of their respective genes in both human PASMCs and PAECs is largely unknown. As shown in Fig. 1A, B, PTGS2 mRNA was basally expressed under unstimulated condition in both human PASMCs and PAECs. In PASMCs (Fig. 1A), treatment with 1% and 2.5% CSE for 72 h had no effect, but treatment with 5% CSE significantly upregulated PTGS2 mRNA expression (fold change 5.6 ± 1.3) compared with untreated cells. In PAECs (Fig. 1B), treatment with 1, 2, and 3% CSE for 24 h all significantly upregulated PTGS2 mRNA expression (fold change 2.0 ± 0.28, 2.0 ± 0.17, and 2.7 ± 0.40, respectively, p < 0.01 for all). PTGIS and TBXAS1 mRNA was also detectable in both cell types under unstimulated condition (Fig. 1C–F). In PASMCs (Fig. 1C), treatment with 2.5% and 5% (but not 1%) CSE significantly reduced PTGIS mRNA expression compared with untreated cells (fold change 0.7 ± 0.05, p < 0.05 and fold change 0.5 ± 0.03, p < 0.01, respectively). Similarly, stimulation with 3% CSE caused a significant reduction of PTGIS mRNA expression (fold changed 0.7 ± 0.05, p < 0.01) in PAECs, whereas 1% and 2% CSE had no effect (Fig. 1D). Similar concentration-dependent increase of TBXAS1 mRNA expression by CSE treatment was also observed in both PASMCs (Fig. 1E) and PAECs (Fig. 1F). The mRNA data suggest that CSE induces an imbalanced expression of the PTGIS gene (downregulated) and the TBXAS1 gene (upregulated) in both cell types.

Fig. 1
figure 1

CSE treatment differentially modulates the mRNA expression of PTGS2, PTGIS, and TBXAS1 in human PASMCs and PAECs. Confluent human PASMCs (A, C, and E) and PAECs (B, D, and F) were treated with different concentrations of CSE for 72 h and 24 h, respectively. Total RNA was isolated, and mRNA levels of PTGS2 (A, B), PTGIS (C, D), TBXAS1 (E, F), and the internal control β2M were determined by real-time RT- PCR. Results are calculated as the ratio of target gene mRNA and β2M mRNA and are expressed as fold change over untreated (0% CSE) cells. Each data point represents mean ± SEM from three independent experiments using PASMCs (A, C, and E) from three different donors and PAECs (B, D, and F) from one donor. *P < 0.05, **P < 0.01, ****P < 0.0001 compared with corresponding untreated cells

Full size image

We next questioned whether the differentially regulated mRNA expression of PTGS2, PTGIS, and TBXAS1 by CSE could be translated into similar changes of protein expression of the respective genes. Although PTGS2 mRNA was detectable, COX-2 protein was not detected by Western blotting in unstimulated PASMCs and PAECs (Fig. 2A, B). Consistent with upregulated PTGS2 mRNA expression, CSE treatment also induced a marked upregulation of COX-2 protein in a concentration-dependent manner in both cell types (Fig. 2A, B). In contrast to COX-2, PGIS protein was abundantly expressed in unstimulated PASMCs, but the expression was reduced in a concentration-dependent manner (significant at 5%) by CSE treatment in PASMCs (Fig. 2C). Unexpectedly, PGIS protein was not detected under both unstimulated and CSE-stimulated conditions in PAECs (Fig. 2D); and TXAS protein was also not detected with or without CSE treatment in both PASMCs and PAECs (Fig. 2E). The protein data suggest that CSE may induce an imbalanced expression of PGIS and TXAS at least in PASMCs by reducing PGIS expression.

Fig. 2
figure 2

CSE treatment differentially modulates the protein expression of key enzymes of prostanoid synthesis in human PASMCs and PAECs. Confluent human PASMCs (A, C, and E) and PAECs (B, D, and E) were treated with different concentrations of CSE for 72 h and 24 h, respectively. Total cell lysates were collected, and protein levels of COX-2 (A, B), PGIS (C, D), TXAS (E), and the internal control GAPDH were analyzed by Western blot. Total cell lysates from immortalized human bronchial epithelial (BEAS-2B) cells treated with or without CSE were used as a positive control (E). Irrelavant parts of the Western blotting images were cropped. Optical densitometry analysis of Western blotting bands was then conducted. Results are calculated as the ratio of target protein and GAPDH and are expressed as fold change over untreated (0% CSE) cells. Each data point represents mean ± SEM from three independent experiments using PASMCs (A, C, and E) from three different donors and PAECs (B, D, and E) from one donor. *P < 0.05, **P < 0.01, ***P < 0.001 compared with corresponding untreated cells

Full size image

CSE induces opposing effects on PGE2 production and an imbalanced PGI2/TXA2 release

We next went on to explore whether the altered mRNA and protein expression of the vasoactive enzymes could lead to imbalanced prostanoid mediator release. COX-2 is the rate-limiting enzyme in the synthesis of its downstream prostanoid products, such as PGE2, PGI2, and TXA2. PGE2, a major prostanoid product, is commonly used as an indication for COX-2 activity. As shown in Fig. 3A, PGE2 was produced in unstimulated PASMCs (863.2 ± 79.6 pg/mg protein). Treatment with 1% CSE had no effect on PGE2 production; but surprisingly, treatment with 2.5% and 5% CSE resulted in a significant reduction in PGE2 production (422.7 ± 63.4 pg/mg protein and 195.1 ± 30.0 pg/mg protein, respectively) compared with unstimulated cells. PGE2 was also produced at basal levels in human PAECs (263.5 ± 32.0 pg/mg protein) (Fig. 3B). In contrast to the findings from PASMCs, CSE treatment at all concentrations significantly increased PGE2 production in a concentration-dependent manner (Fig. 3B).

Fig. 3
figure 3

CSE treatment induces 6-keto PGF1α/TXB2 imbalance and elicits opposing effects on PGE2 production in human PASMCs and PAECs. Confluent human PASMCs (A, C, E, and G) and PAECs (B, D, F, and H) were treated with different concentrations of CSE for 72 h and 24 h, respectively. Medium was collected, and levels of PGE2 (A, B), 6-keto PGF1α (C, D) and TXB2 (E, F) were determined by ELISA and 6-keto PGF1α/TXB2 ratio (G and H) was calculated. Results were normalised with total cell protein and are expressed as pg/mg protein (A–F) or 6-keto PGF1α/TXB2 ratio (G, H). Each data point represents mean ± SEM from three independent experiments using PASMCs (A, C, E and G) from three different donors and PAECs (B, D, F, and H) from one donor. *P < 0.05, **P < 0.01, ***P < 0.001 compared with corresponding untreated cells

Full size image

Both PGI2 (measured as the stable metabolite 6-keto PGF1α) and TXA2 (measured as the stable metabolite TXB2) were produced in control PASMCs and PAECs (Fig. 3C–F) and the ratio of PGI2/TXA2 (calculated as 6-keto PGF1α/TXB2) was around 12 and 1.75 for PASMCs and PAECs (Fig. 3G, H), respectively. Treatment of PASMCs with CSE significantly and concentration-dependently reduced PGI2 production, but increased TXA2 production (Fig. 3C and E). As a result, PGI2/TXA2 ratio was markedly reduced after treatment of cells with 1%, 2.5% and 5% CSE (4.0 ± 0.5, p < 0.05, 2.3 ± 1.2, p < 0.05 and 0.6 ± 0.2, p < 0.01, respectively) (Fig. 3G). Interestingly, treatment of PAECs with CSE had no effect on PGI2 production, but significantly increased TXA2 production in a concentration-dependent manner (Fig. 3D and F), thereby resulting in a significantly reduced PGI2/TXA2 ratio by 1%, 2% and 3% CSE (0.8 ± 0.2, p < 0.05, 0.6 ± 0.1 and 0.6 ± 0.1, p < 0.01, respectively) (Fig. 3H).

Although CSE induced COX-2 expression in both PASMCs and PAECs (Fig. 2A, B) and increased PGE2 production in PAECs (Fig. 3B), PGE2 production was unexpectedly reduced by CSE in PASMC (Fig. 3A), suggesting that CSE may regulate the expression of mPGES-1, the enzyme responsible for PGE2 production downstream of COX-2, differently in the two cell types. Therefore, the effect of CSE on mPGES-1 protein expression was analyzed. It was revealed that mPGES-1 was basally expressed in PASMCs and the expression was significantly reduced by treatment with 5% CSE (Fig. 4). In contrast, mPGES-1 was also basally expressed in PAECs and no change of the expression was observed after treatment with CSE (data not shown).

Fig. 4
figure 4

CSE treatment downregulates the protein expression of mPGES-1 in human PASMCs. Confluent human PASMCs were treated with CSE (5%) for 72 h. Total cell lysates were collected, and protein levels of mPGES-1 and the internal control GAPDH were analyzed by Western blot. Irrelavant parts of the Western blotting images were cropped. Optical densitometry analysis of Western blotting bands was then conducted. Results are calculated as the ratio of mPGES-1 and GAPDH and are expressed as fold change over untreated (0% CSE) cells. Each data point represents mean ± SEM from three independent experiments using PASMCs from three different donors. *P < 0.05 compared with untreated cells

Full size image

These data altogether suggest that CSE induces an imbalanced prostanoid release characterized by the reduced PGI2/TXA2 ratio, likely as a result of an imbalanced expression of PGIS and TXAS in PASMCs and PAECs, and that this imbalance may promote proliferation of both cell types, thereby contributing to pulmonary vascular remodelling in COPD.

CSE treatment stimulates proliferation of PASMCs and PAECs, and the proliferation is inhibited by celecoxib, beraprost sodium and daltroban

We then assessed whether CSE treatment stimulated proliferation of PASMCs. It was found that CSE at 5% significantly stimulated proliferation of PASMCs at all time points tested compared with the corresponding control (Fig. 5A). To test whether CSE-induced COX-2 expression plays a role in PASMC proliferation, the effect of celecoxib, a selective COX-2 inhibitor, on CSE-induced cell proliferation was assessed. The results showed that CSE treatment significantly increased cell proliferation compared with control, and this effect was inhibited by celecoxib in a concentration-dependent manner (Fig. 5B). Since mPGES-1 expression and PGE2 release were reduced by CSE, the results suggest the release of prostanoids other than PGE2 downstream of COX-2 plays a role in CSE-induced PASMC proliferation. As CSE reduced the release of the anti-proliferative prostanoid PGI2 and increased the release of the proliferative TXA2 downstream of COX-2, the effect of the stable PGI2 analogue beraprost sodium and the selective TXA2 receptor antagonist daltroban on CSE-induced PASMC proliferation was investigated. As shown in Fig. 5C, D, both beraprost sodium and daltroban significantly inhibited CSE-induced PASMC proliferation in a concentration-dependent manner, suggesting reduced PGI2 and increased TXA2 both contribute to CSE-induced PASMC proliferation. Similarly, CSE at 3% also significantly increased PAEC proliferation at 16 and 24 h compared with the corresponding control (Fig. 6A), and the increase at 24 h was significantly and concentration-dependently inhibited by celecoxib (Fig. 6B), beraprost sodium (Fig. 6C) and daltroban (Fig. 6D), but not by the mPGES-1 inhibitor PF-03549184 (data not shown). The results suggest that reduced PGI2 and increased TXA2, but not increased PGE2, contribute to CSE-induced PAEC proliferation.

Fig. 5
figure 5

CSE treatment stimulates proliferation of human PASMCs, and celecoxib, beraprost sodium and daltroban inhibit the proliferation. (A) confluent human PASMCs were treated with or without CSE for up to 72 h, and then WST-1 assay was conducted. (B–D) confluent human PASMCs were pre-treated with or without different concentrations of celecoxib (B), beraprost sodium (C) and daltroban (D) for 1 h before being treated with CSE for 24 h. WST-1 was then conducted. Data are expressed as relative proliferation (% change over control). Each data point represents mean ± SEM from three independent experiments using PASMCs from three different donors. *P < 0.05, ***P < 0.001 compared with control; +P < 0.05, ++P < 0.01, +++P < 0.001 compared with CSE alone

Full size image
Fig. 6
figure 6

CSE treatment stimulates proliferation of human PAECs, and celecoxib, beraprost sodium, and daltroban potently inhibit the proliferation. (A) confluent human PAECs were treated with or without CSE for up to 24 h, and then WST-1 assay was conducted. (B–D) confluent human PAECs were pre-treated with or without different concentrations of celecoxib (B), beraprost sodium (C), and daltroban (D) for 1 h before being treated with CSE for 24 h. WST-1 was then conducted. Data are expressed as relative proliferation (% change over control). Each data point represents mean ± SEM from three independent experiments using PAECs from one donor **P < 0.01, ****P < 0.0001 compared with control; +P < 0.05, ++P < 0.01, +++P < 0.001, ++++P < 0.0001 compared with CSE alone

Full size image

The COX-2 inhibitor celecoxib could inhibit the release of all prostanoids downstream of CSE-induced COX-2, including PGE2, PGI2 and TXA2. But in PASMCs, the release of PGE2,and PGI2 was reduced by CSE, and in PAECs, PGI2 release was not affected by CSE and the mPGES-1 inhibitor PF-03549184 had no effected on CSE-induced cell proliferation, it is plausible that the anti-proliferative effect of celecoxib could be mediated by its inhibition of TXA2 release in both cell types. To test this possibility, the effect of celecoxib on TXA2 release (measured as TXB2) was assessed. As shown in Fig. 7A, basal level of TXB2 was very high (about 6000 pg/mg protein) due to high levels of TXB2 in the medium with 20% FBS (for cell proliferation), and no further increase by CSE and no inhibition by celecoxib were observed in PASMCs. In contrast, TXB2 was also detected basally (about 700 pg/mg protein) in the culture medium for PAECs with 2% FBS and a significant increase was observed after CSE treatment, which was markedly reduced by celecoxib (Fig. 7B). The results suggest that celecoxib can inhibit CSE-induced TXA2 release from both PASMCs and PAECs, thereby contributing to its anti-proliferative effect and providing further evidence that TXA2 is a key mediator in CSE-induced pulmonary cell proliferation.

Fig. 7
figure 7

Selective COX-2 inhibition reduces CSE-induced TXB2 production in human PAECs, but not in PASMCs. Confluent human PASMCs (A) and PAECs (B) were pre-treated with one concentration of celecoxib for 1 h before being treated with CSE for 24 h. Medium was collected, and TXB2 concentration was determined by ELISA. Results were normalised with total cell protein and are expressed as pg/mg protein. Each data point represents mean ± SEM from three independent experiments using PASMCs (A) from three different donors and PAECs (B) from one donor. ***P < 0.001 compared with corresponding untreated cells; +P < 0.05 compared with CSE alone

Full size image

Discussion

The main novel findings of the current study are that CSE induced imbalanced vasoactive gene expression and mediator release, particularly the reduced PGI2/TXA2 ratio in both PASMCs and PAECs mainly due to increased TXA2 production and that the restoration of PGI2 and TXA2 imbalance by beraprost sodium and celecoxib (via the inhibition of COX-2-derived TXA2 production) and the antagonism of TXA2 receptor by daltroban all inhibited CSE-induced cell proliferation in both PASMCs and PAECs. To the best of our knowledge, these findings are the first to suggest that PGI2 and TXA2 imbalance plays a key role in cigarette smoke-induced pulmonary vascular cell proliferation and consequently PGI2 analogues and inhibitors of TXA2 production and effect may prevent pulmonary vascular remodelling and ultimately have therapeutical potential in pulmonary hypertension associated with COPD.

COX-2 is regarded as a vasoactive gene, and its induction is important in mediating the production of its downstream prostanoid products that have potent anti-proliferative (e.g. PGE2 and PGI2) and proliferative (e.g. TXA2) properties [37]. The fact that CSE, in the present study, upregulated PTGS2 mRNA and COX-2 protein expression in both PASMCs and PAECs strongly suggests that CSE regulates COX-2 expression via transcriptional regulation. Although in vitro studies have reported COX-2 upregulation by CSE in non-pulmonary artery cell types, such as human small airway epithelial cells [25], in endothelial cells, such as human umbilical vein endothelial cells [26] and human pulmonary microvascular endothelial cells [27], and in vivo studies have demonstrated COX-2 induction in the lung tissue of both patients with COPD and smokers without COPD compared with non-smokers without COPD [26], our study is the first to show that CSE induces COX-2 expression in PASMCs and PAECs, the two cell main types of pulmonary artery vasculature.

Since COX-2 is the rate-limiting enzyme in prostanoid synthesis, its upregulation is expected to lead to the increase in the production of its downstream prostanoid products, such as PGE2. The observation that CSE induced COX-2 expression as well as PGE2 production in PAECs supports this concept. This is further supported by our previous findings demonstrating that inflammatory mediators such as IL-1ß can induce the protein expression of COX-2 and its product PGE2 in human PASMCs [38]. Similar to our findings from PAECs, it has been reported that CSE can induce the expression of COX-2 and mPGES-1 and production of PGE2 in normal human lung fibroblasts [39]. Unexpectedly, there was a dissociation between increase of COX-2 expression and decrease of its downstream product PGE2 in response to CSE in PASMCs in our study. This finding, together with the increased PGE2 levels by CSE in PAECs, suggests that CSE may regulate differently in PASMCs and PAECs the expression of mPGES-1, the enzyme that mediates PGE2 conversion from the COX-2 product PGH2. The fact that mPGES-1 expression was reduced in PASMCs, but unchanged in PAECs, strongly supports the concept that mPGES-1 downregulation plays a crucial role in mediating the reduced PGE2 production in PASMCs by CSE, despite the increased COX-2 expression. Although PGE2 is poorly characterized as a key factor of pulmonary artery cell dysfunction in all forms of pulmonary hypertension, there is evidence to suggest a potential role of dysregulated PGE2 in the development of group 1 pulmonary hypertension. For instance, it has been reported that circulating PGE2 levels are reduced in patients with group 1 pulmonary hypertension [22] and that highly selective EP2 receptor agonist butaprost is able to inhibit proliferation of human PASMCs derived from patients with group 1 pulmonary hypertension [23]. However, since CSE reduced PGE2 production in PASMCs but increased PGE2 production in PAECs, similar inhibition on CSE-induced cell proliferation in both cell types by celecoxib, which inhibits production of all prostanoids downstream of COX-2, including PGE2, suggests that PGE2 is not critically involved in CSE-induced PASMC and PAEC proliferation. The fact that mPGES-1 inhibitor PF-03549184 did not have any effect on CSE-induced PASMC and PAEC proliferation (unpublished observation) provides further evidence that other prostanoids, such as PGI2 and TXA2, may play a more important role than PGE2 in CSE-induced cell proliferation. However, the exact role of CSE-mediated reduced PGE2 in PASMCs in cigarette smoke-induced pulmonary vascular remodelling requires further investigation.

PGI2 plays a key role in maintaining local vascular tone. It has been reported that urinary metabolites of PGI2 are decreased in group 1 pulmonary hypertension [24]. A reduction of PGIS expression and PGI2 production has also been demonstrated in lung tissue of patients with COPD [27]. Importantly, in vitro studies showed that the use of PGI2 analogue can prevent CSE-induced cell apoptosis in human pulmonary microvascular endothelial cells [27] and umbilical vein endothelial cells [28], indicating a potential role of decreased PGI2 levels in cigarette smoke-induced pulmonary endothelial cell dysfunction and apoptosis in patients with COPD. Our study is the first to explore the effect of CSE on PGIS-derived PGI2 and contribution of PAECs, particularly PASMCs in this process. We showed that PTGIS mRNA expression, PGIS protein expression and PGI2 production were all reduced by CSE in PASMCs. This finding, together with the inhibition of CSE-induced PASMC proliferation by the PGI2 analogue beraprost sodium, strongly suggests that reduced PGIS/PGI2 plays a key part in CSE-induced imbalanced vasoactive prostanoid production in PASMCs and may ultimately contribute to pulmonary vascular remodelling in COPD. PTGIS mRNA expression was also reduced by CSE in PAECs. Surprisingly, PGIS protein expression was undetectable by Western blotting in both untreated and CSE-treated PAECs, suggesting that PAECs constitutively express less PGIS than PASMCs. This is supported by the fact that PAECs constitutively produced less PGI2 than PASMCs (792.0 ± 94.7 pg/mg protein compared with 2987.2 ± 403.5 pg/mg protein). It is also reasonable to speculate that CSE had no effect on PGIS protein expression since PGI2 release was unchanged. These findings in PAECs are in disagreement with the previous in vitro study showing reduced PGI2 release in response to CSE (24 h, up to 10%) in human umbilical vein endothelial cells [28]. The discrepancies may be due to the differences in cell type and cigarette brand (Marlboro) used in the other study. In addition, our findings in PAECs also differ from those in PASMCs demonstrating reduced PGIS expression and PGI2 production by CSE. This may be explained by the differences in cell types and in CSE treatment time (24 h for PAECs and 72 h for PASMCs). Although exogenous PGI2 exerted anti-proliferative effect on CSE-induced PAEC proliferation, the unchanged PGI2 production following CSE treatment in PAECs suggests that PGI2 may not play a key role in CSE-induced imbalanced vasoactive prostanoid production and cell proliferation in PAECs.

It is worth noting that although it is commonly considered that PAECs are the main source of PGI2, the facts that PASMCs produce higher basal levels of PGI2 than PAECs and that CSE-induced imbalanced vasoactive prostanoid mediator release is more prominent in PASMCs than in PAECs suggest that PASMC dysfunction may be more important than the traditionally recognized PAEC dysfunction in contributing to pulmonary artery remodelling and pulmonary hypertension in COPD. Although reports have shown other cell types (e.g. adventitial fibroblasts) in the pulmonary vasculature can also contribute to vascular remodelling [13], their role in COPD-associated pulmonary hypertension in response to cigarette smoke has not been explored.

Our findings that CSE induced COX-2 expression but reduced both PGI2 and PGE2 production in PASMCs as a result of PGIS and mPGES-1 downregulation, respectively, strongly suggest that CSE can selectively reduce vasodilatory and anti-proliferative gene expression and mediator release in prostanoid biosynthesis and that the downregulated PGIS and mPGES-1 expression may divert the COX-2 product PGH2 away from being converted to PGI2 and PGE2 towards the systhesis of the vasoconstrictive and proliferative mediator TXA2 via TXAS. To our knowledge, this study is the first to assess CSE effect on TXAS-derived TXA2 in human PASMCs and PAECs. Although TXAS protein expression was undetectable in both untreated and CSE-treated PASMCs and PAECs despite increased TBXAS1 mRNA expression, it is highly likely that TXAS protein was below the detection of Western blotting under both conditions. This is supported by our findings that hypoxia (1% O2) induced TXAS expression using the same human PASMCs (unpublished observation) and that TXAS was detected at basal levels and was further induced in CSE-treated normal human bronchial epithelial cells (positive control). Since TBXAS1 mRNA and TXAS product TXA2 were detected in unstimulated PASMCs and PAECs and were increased by CSE in both cell types, it is likely that TXAS protein expression was increased in response to CSE, although the possibility that TXAS protein expression was unchanged or decreased could not be excluded.

Despite the undetected TXAS protein expression, the increase of TXA2 by CSE suggests it is the result of the conversion by the existing TXAS of the increased PGH2, due to the combined effect of CSE on COX-2 upregulation (both cell types), PGIS and mPGES-1 downregulation (PASMCs) and unchanged PGIS expression (PAECs). It is also possible that CSE may increase TXAS activity (e.g. by post-translational modifications) rather than its expression, although there is no direct evidence. The increase of vasoconstrictive and proliferative mediator TXA2 by CSE in both cell types suggest that TXA2 may act as a key mediator of pulmonary artery cell dysfunction in COPD. This is supported by the ability of the TXA2 receptor antagonist daltroban to inhibit CSE-induced PASMC and PAEC proliferation, which also suggests that blocking TXA2 effect may reduce pulmonary vascular remodelling induced by cigarette smoking. The fact that selective inhibition of COX-2 by celecoxib significantly reduced CSE-induced TXA2 production in PAECs, but not in PASMCs due to the interfernce of the assay by existing high levels of TXB2 in FBS (20%) in the medium used for cell proliferation, strongly suggests that the anti-proliferative effect of celecoxib is the result of the inhibition of COX-2-derived TXA2 production and provides further evidence that CSE-induced COX-2 expression may have a detrimental effect on pulmonary vascular remodelling by mediating the induction of imbalanced prostanoid production, particularly increased TXA2 levels.

Together with TXA2, PGI2 plays an important role in maintaining homeostatic balance in pulmonary circulation. In the present study, we assessed CSE effect on PGI2/TXA2 ratio and found, for the first time, that CSE reduced the vasodilatory and anti-proliferative mediator PGI2 and increased the vasoconstrictive and proliferative mediator TXA2 in PASMCs, resulting in a reduced PGI2/TXA2 ratio. Although CSE had no effect on PGI2 production in PAECs, CSE also reduced the PGI2/TXA2 ratio in PAECs due to increased TXA2 production. These observations suggest that CSE can cause PASMC and PAEC dysfunction by inducing an imbalance between PGI2 and TXA2, thereby contributing to cigarette smoke-induced vascular remodelling in COPD and that agents that target the imbalance by either compensating for PGI2 reduction or blocking TXA2 effects may have therapeutic potential in COPD-associated pulmonary hypertension.

Conclusions

In conclusion, our findings that CSE induced an imbalanced prostanoid release characterised by the reduced PGI2/TXA2 ratio, likely as a result of an imbalanced expression of PGIS and TXAS, and that agents targeting the imbalance inhibited CSE-induced pulmonary artery cell proliferation point to TXA2 as a novel mediator of pulmonary artery cell dysfunction and proliferation in pulmonary hypertension in COPD and strongly suggest that targeting the imbalance may have therapeutic potentials for this fatal disease and that smoking cessation may prevent further vascular remodelling.