Main

Improvements in care have dramatically improved survival from respiratory diseases in the newborn period. Unfortunately CLD continues to be a major problem resulting in significant morbidity. The pathophysiologic mechanisms underlying the development of CLD are multifactorial including the severity of acute lung disease plus barotrauma, oxygen therapy, and inflammation(1). Inadequate lung repair may also contribute, but little is known about growth factors involved in the healing process(2). Infants with CLD have abnormalities of both airways and vasculature, with evidence of increased vascular permeability. Treatment is primarily supportive with supplemental oxygen, fluid restriction, and diuretics. In some infants, dramatic improvement is seen with dexamethasone, but the mechanism(s) for this response is not known(3). In a recent report of recovery from oxygen-induced lung injury in rabbits, Maniscalco et al.(4) reported increased abundance of VEGF mRNA in alveolar epithelial cells.

VEGF, also known as vascular permeability factor and vasculotropin, is a specific endothelial cell mitogen(5–7). VEGF activities are mediated through two tyrosine kinase receptors (KDR and flt), which are expressed on endothelial cells(8–10). VEGF is expressed in a variety of tissues and cell types and is abundantly present in lung tissue(11,12). There are four alternatively spliced forms of VEGF that appear to have similar activity, but strikingly different affinities for heparin(7, 13, 14). Hypoxia induces VEGF production in a number of cell types(15–22), and multiple studies have suggested a role for VEGF in tumor angiogenesis(7, 23). In malignant glioblastomas increased expression of VEGF is found surrounding necrotic foci of tumor, presumbaly in hypoxic regions(24). VEGF activity is markedly up-regulated in ocular fluid after experimental induction of hypoxia in a primate model(25) and is proposed to be an angiogenic factor under similar circumstances in neovascularization in diabetic retinopathy and other ischemic retinal diseases(26). VEGF has also been studied in the setting of osteogenesis, a process dependent on angiogenesis. Prostaglandins E1 and E2, which are potent stimulators of new bone formation, were shown to markedly increase levels of VEGF in osteoblastic cell lines, and further, dexamethasone suppressed this up-regulation(27). One of the poorly understood activities of VEGF is its ability to induce increased vascular permeability in the skin(7, 28). Direct evidence for VEGF causing increased permeability in the lung has not been reported.

The expression of VEGF by cultured systemic vascular smooth muscle cells has been examined(18, 29, 30), but pulmonary derived cells have not been studied in vitro. Tuder et al.(31) found that VEGF was present in pulmonary vascular smooth muscle cells after lung injury secondary to chronic hypoxia. CLD results in marked increases in the cellularity of the lung, including increased numbers of smooth muscle cells. Whether VEGF derived from smooth muscle cells is important in CLD is not known. The purpose of this study was to better understand the regulation of VEGF in pulmonary vascular smooth cells by examining factors that are important regulators of clinical aspects of CLD. Specifically the effects of hypoxia and dexamethasone on the expression of VEGF mRNA in pulmonary vascular smooth muscle cells were studied.

METHODS

Cloning of ovine VEGF cDNA. Primers were constructed for use in RT-PCR using homologous regions of known human (GenBank™ accession no. X62568) and rat (GenBank™ accession no. M32167) VEGF sequences(32, 33). Primers common to all four splice variants of VEGF were chosen (the primers flank the regions of insertions and deletions). Upper primer 5′-GAAGTGGTGAAGTTCATGGA-3′; lower primer 5′-TCGGCTTGTCACATCTGCAA-3′. RT-PCR was performed using sheep lung RNA as a template at 94°C for 4 min, then 45 cycles of 94°C, 60°C, 72°C, followed by 72°C for 10 min (Life Technologies, Inc., Grand Island, NY). The amplified fragment was cloned into pGEM-T vector (Promega, Madison, WI), and sequenced using cycle sequencing (Circumvent, New England Biolabs, Beverly, MA), with multiple primers. Sequencing results were compared with known human sequences using PC/GENE (Intelligenetics; Geneva, Switzerland).

Probes. The isolated ovine VEGF cDNA was used as a template to prepare riboprobes for analysis of VEGF mRNA expression. Riboprobes were synthesized using T7 RNA polymerase (Promega) and 32P-labeled UTP to a specific activity of 2 × 106 cpm/μg ([32P]UTP, Ci/mM; DuPont NEN)(34). For normalization, antisense riboprobes or cDNA probes were made to cyclophilin(35).

Cell culture. Sheep pulmonary artery smooth muscle cells were cultured in RPMI (Life Technologies, Inc.) supplemented with 10% fetal bovine serum (Hyclone Labs, Logan, UT) by explant outgrowth technique(36). Cells were identified as smooth muscle cells by their hill and valley growth pattern and positive staining for smooth muscle cell α-actin. Early passage cells (less than 6) were seeded into P-100s plates, grown to confluence, and made quiescent in serum-free medium for 24-48 h before use. Cells were exposed to room air or hypoxia, with and without dexamethasone (Sigma Chemical Co., St. Louis, MO). Dexamethasone was studied at concentrations of 1.3, 4.0, 13, 25, and 40 mg/L. Hypoxia was achieved by infusion of a gas mixture of 95% N2, 5% CO2 into a modular incubator chamber (Billups-Rothenberg, Del Mar, CA). In seven experiments, pH, Po2, and PCo2 were measured in media. Additional plates were treated with PMA (Sigma Chemical Co.), a known inducer of VEGF expression(37). Exposure time for all experimental conditions was 6 h, except for the hypoxia time course and the actinomycin D studies where exposure times varied and are noted in the results.

Actinomycin D transcriptional inhibition. Confluent cells were made quiescent in serum-free medium for 48 h and then pretreated for 60 min with actinomycin D (Sigma Chemical Co.) 10 mg/L followed by continued actinomycin D exposure ± hypoxia. RNA was isolated for northern analysis before addition of actinomycin D, and at 30, 60, 90, 120, and 240 min after exposure to actinomycin D ± hypoxia. Preliminary studies determined that actinomycin D at 10 mg/L during the time course of this experiment was not toxic to the cells in culture.

RNA isolation and analysis. At the end of the experimental period, the cells were examined by phase contrast microscopy for evidence of cell damage or death. Total RNA was isolated from the cells using standard techniques, and poly(A)+ RNA was isolated with an oligo(dT)(Collaborative Research; Bedford, MA) cellulose affinity column and quantified by measuring absorbance at 260 and 280 nm(34, 38). Poly(A)+ RNA was loaded at 2 or 3 μg/lane, total RNA loaded at 10μg/lane, and run overnight into a 1% agarose/2.2 M formaldehyde gel at 20 V. RNA was then transferred to nitrocellulose filter (Nitro-Plus, Micron Separations, Inc.; Westboro, MA) in 10 × SSC (1 × SSC = 150 mM NaCl, 15 mM sodium citrate) and immobilized on the filter with UV cross-linking at 1200 J (UV Stratalinker, Stratagene, La Jolla, CA). The blots were prehybridized for a minimum of 2 h at 42°C in 50% formamide, 250 ng/mL sheared salmon sperm DNA, 1 × Denhardt's solution, 50 ng/mL poly(A), 0.1% SDS, and 5 × SSC. Labeled riboprobe was added to blots at a concentration of 2 × 106 cpm/mL of hybridization solution, and hybridized for 18 h at 63°C. Blots were washed in 0.1 × SSC + 0.1% SDS, and then exposed to PhosphorImager plates for 24-48 h. Results were quantified using a Bio-Rad PhosphorImager with ImageQuant software. To normalize for loading, blots were also probed for cyclophilin expression(34,35).

RT-PCR. RT-PCR was performed to specifically detect the presence of splice variants using the primers described above in the cloning experiments. These primers flank the regions containing all the exons and should identify all splice variants of VEGF. RNA isolated from cell culture experiments was used as the template at 94°C for 4 min, then 45 cycles of 94°C, 60°C, 72°C, followed by 72°C for 10 min. The amplified product was run on a 1% agarose gel and the number and sizes of fragment(s) were identified using standard size markers. Using these primers, the VEGF splice variants would yield PCR products of 332, 444, 516, and 570 bp(VEGF121, VEGF165, VEGF189, and VEGF201, respectively).

Data analysis. Results for each experiment were corrected for loading and normalized to control for each experiment. Mean and SEM were calculated for each experimental group. Data from experimental groups were compared using nonparametric statistics (Mann-Whitney) with a p value of less than 0.05 considered significant.

RESULTS

Cloning of ovine VEGF cDNA fragment. A 453-bp fragment was isolated and sequencing revealed that it corresponded to VEGF165. This fragment includes 72% of the coding sequence, and is highly homologous to both human and rat sequences (94 and 88%, respectively). The predicted amino acid sequence is shown in Table 1, aligned with known human and rat sequences. The ovine cDNA fragment includes exons common to all four forms of VEGF; thus, probes made from this cDNA can detect all four species of VEGF.

Table 1 Alignment of predicted amino acid sequence of ovine VEGF cDNA fragment with human and rat sequences
Full size table

Expression of VEGF mRNA by cultured pulmonary vascular smooth muscle cells. Controls. In cells exposed to room air, VEGF mRNA was expressed predominantly as a single band at approximately 3.9 kb, as seen in Figure 1. This size would be consistent with VEGF165 mRNA as reported in other cells(29). Exposure to PMA resulted in a 9-fold increase in the abundance of this band, consistent with the reported induction by PMA(37). In some experiments a second band was seen, at about 4.1 kb. Results from RT-PCR revealed the predominant band was VEGF165, and when two bands were present, the second band was VEGF189 (data not shown).

Figure 1
figure 1

Representative Northern blot showing expression of VEGF mRNA at baseline and after PMA treatment. Three micrograms of poly(A+) RNA from pulmonary artery smooth muscle cells were loaded in each lane. The blots were hybridized to antisense riboprobes for VEGF and cyclophilin (for loading). At baseline a single band of VEGF at about 3.9 kb could be detected. After exposure to PMA, there was a marked increase in the abundance of this band. Cyclophilin expression did not change.

Full size image

Hypoxia. Time course experiments revealed that exposure to hypoxia resulted in 2-fold increase in the abundance of VEGF mRNA by 4 h, which further increased at 6 h, and remained elevated at 16 h (data not shown). Subsequent data are reported for the 6-h time point. InFigure 2, the first lane shows control expression with a single predominant band about 3.9 kb. After exposure to 6 h of hypoxia, there was a striking increase in the abundance of the 3.9-kb band, as well as the appearance of bands at 3.7 and 4.1 kb. Not all experiments showed multiple bands after exposure to hypoxia, but all had increases in VEGF mRNA. The data for all experiments are summarized in Figure 3, where results were corrected for loading (cyclophilin expression) and normalized to control expression in each experiment. Exposure to hypoxia resulted in a significant increase over control expression (3.2 ± 0.4-fold increase; M ± SEM; n = 7). When multiple bands were present only the abundance of the 3.9-kb band was used in summarizing the data; however, inclusion of other bands gave similar results (data not shown). There was no visible evidence of cell damage from exposure to hypoxia.

Figure 2
figure 2

Representative Northern blot showing the expression of VEGF mRNA in pulmonary artery smooth muscle cells that have been exposed to hypoxia and/or dexamethasone for 6 h. Three micrograms of poly(A+) RNA were run in each lane. In control cells, there was a strong band of hybridization at 3.9 kb. The addition of dexamethasone did not alter the baseline expression of VEGF in this experiment. Hypoxia resulted in a striking increase in the abundance of VEGF mRNA, with several size transcripts apparent. With the addition of dexamethasone at 1.3 mg/L, there was a slight attenuation of the response to hypoxia, but at a dose of 4.0 mg/L the response was clearly attenuated. Results for all studies are summarized inFigure 3. C, room air; H, hypoxia; DEX, dose of dexamethasone, mg/L.

Full size image
Figure 3
figure 3

Data from all experiments. VEGF expression was corrected for loading and then normalized to 1 for control. After exposure to hypoxia, VEGF mRNA abundance increased over 3-fold. The response to hypoxia was attenuated in the presence of dexamethasone. *p < 0.05 when compared with control; C, room air; H, Hypoxia; Dex, dose of dexamethasone, mg/L.

Full size image

pH, Po2, and Pco2 measurements of media. Media from cells exposed to room air had a mean Po2 of 139 ± 1.5 kPa. Exposure to hypoxia resulted in a gradual decrease in the Po2 which reached steady state after 4 h (2 h, Po2 = 91.5 ± 4.1; 4 h = 51.5 ± 3.5; 6 h = 61.7 ± 3.1; 16 h = 62.5 ± 3.5). pH and Pco2 did not differ with varying conditions (data not shown).

Dexamethasone and room air. Exposure of cells in room air to dexamethasone 1.3 mg/L did not alter RNA abundance. Similarly, at concentrations of 4 mg/L or greater there was no statistically significant change in VEGF mRNA abundance (48-82% of control). Dexamethasone did not result in any apparent cell damage.

Dexamethasone and hypoxia. Exposure of cells to dexamethasone concurrently with hypoxia reduced the hypoxic induction of VEGF mRNA in a dose-dependent fashion. In Figure 2, the presence of low dose dexamethasone (1.3 mg/L) combined with hypoxia resulted in decreased abundance of all bands, compared with hypoxia alone; however, the expression of VEGF was still increased over control. At 4 mg/L, the abundance of VEGF was no longer significantly increased over control. Higher doses of dexamethasone(13, 25, and 40 mg/L) did not result in further suppression (data not shown). The data are summarized in Figure 3 after normalization to control. Hypoxic exposure combined with low dose dexamethasone (1.3 mg/L) resulted in a partial attenuation of the hypoxic induction (1.8 ± 0.21;n = 3), whereas hypoxia combined with higher dose dexamethasone (4.0 mg/L) almost completely blocked the hypoxic response (1.4 ± 0.27;n = 3).

RT-PCR for splice variant evaluation. RT-PCR was performed to determine whether multiple species of VEGF mRNA were present. VEGF165 was found in all samples. When two species were seen by Northern, RT-PCR identified VEGF165 and VEGF189. In some experiments, Northern blot experiments identified multiple bands only after induction by hypoxia; however, by RT-PCR both VEGF165 and VEGF189 were present in both control samples and hypoxic samples even when Northern blot identified only one band (data not shown). There was no pattern in the appearance of VEGF189 to suggest differential regulation. VEGF121 and VEGF201 were not detected.

Actinomycin D transcriptional inhibition. The rates of decline in VEGF mRNA in the presence of actinomycin D are shown inFigure 4. In cells exposed to room air the half-life was 171 ± 12 min. After exposure to hypoxia, there was no significant change in the stability of the mRNA, with a half-life of 175 ± 22 min.

Figure 4
figure 4

VEGF mRNA levels after exposure to actinomycin D. The decline in VEGF mRNA was similar in cells exposed to room air (open circles) and in those exposed to hypoxia (closed circles).

Full size image

DISCUSSION

VEGF is a highly conserved growth factor whose localization to epithelial cells in normal lung places it in a strategic position to stimulate the adjacent growth of vessels which is critical for normal respiration to occur. The increased expression of VEGF in epithelial cells during recovery from hyperoxia in rabbits supports the idea that VEGF may be participating in the repair process and recovery of injured vessels(4). Although VEGF mRNA expression in smooth muscle cells is not found in normal lung, it is present in lung injury(31). VEGF has been associated with repair of other tissues and is thought to support angiogenesis and possibly to be related to increased permeability in wounds(39,40). In the lung, VEGF derived from nonepithelial cells may have a different role than epithelial derived VEGF. Additional cellular sources of VEGF could result in higher local concentrations of VEGF and function not only as a mitogen but also contribute to increased vascular permeability.

In this study we show that cultured pulmonary vascular smooth muscle cells produce VEGF mRNA and thus are potential sources for VEGF production in the lung in vivo. The induction of VEGF mRNA in smooth muscle cells by decreased oxygen tension, suggests that these cells could be important producers of VEGF during hypoxia associated with lung injury. One of the proposed mechanisms for sensing oxygen tension is through heme-containing proteins, acting in a fashion similar to the regulation of erythropoietin(16). In human retinal epithelial cells and rat astrocytoma cells, the increase in VEGF mRNA after hypoxia was found to be secondary to a dramatic increase in mRNA stability(15). However, in human umbilical vein endothelial cells, expression of VEGF mRNA was not present at baseline, but a significant increase occurred after hypoxia, presumably through transcriptional up-regulation(21). In rat aortic smooth muscle cells, IL-1β increases VEGF abundance, through both an increase in mRNA stability and increased transcription(30). The mechanism(s) of hypoxic induction of VEGF has not previously been reported in vascular smooth muscle cells. In the presence of actinomycin D, we found no evidence of increased mRNA stability in cells exposed to hypoxia, suggesting that increased abundance of VEGF mRNA seen after hypoxic exposure was via a transcriptional mechanism. We cannot exclude the possibility that actinomycin D was inhibiting a factor mediating stability; however, we think this is unlikely because in the presence of actinomycin, hypoxia has been shown to increase stability in other cells, e.g. retinal epithelial cells(15). Interestingly, the half-life of VEGF mRNA in the pulmonary vascular smooth muscle cells was 3-fold longer than the half-life reported for VEGF mRNA from other types of cells(15, 22). The mechanism for the increased stability of the mRNA in smooth muscle cells is not known but such cell specific regulation of VEGF mRNA could allow for cell specific effects of exogenous agents to block VEGF expression.

The decrease in VEGF mRNA abundance found when dexamethasone was present during hypoxic exposure of the smooth muscle cells is likely through blocking transcription. Dexamethasone is known to block AP-1 mediated transcription, and the VEGF promoter has potential binding sites for AP-1 transcription factors(13). In National Institutes of Health 3T3 cells, dexamethasone has no effect on either baseline or hypoxia-induced VEGF expression but dexamethasone did block phorbol ester induction of VEGF(19). Similar results have also been reported in human mesangial cells with dexamethasone blocking induction by phorbol esters, but not altering baseline or hypoxia effects(41). In contrast, dexamethasone did suppress baseline expression of VEGF in osteoblasts(27). In our study, dexamethasone blocked the increase in VEGF mRNA during exposure of the pulmonary vascular smooth muscle cells to hypoxia, suggesting that transcription was via the AP-1 site. Recent studies have implicated hypoxia-inducible factor 1 in transcriptional activation of VEGF in Hep3B cells(42). The role of hypoxia-inducible factor 1 in the hypoxic induction of VEGF in pulmonary vascular smooth muscle cells has not been examined.

In the pulmonary vascular smooth muscle cells VEGF165 was the predominant splice variant, although VEGF189 was also detected in some samples. Because the isoforms have varying affinities for heparin, differential expression could result in alterations in the bioavailability of VEGF(43, 44). However, although we could identify multiple isoforms by RT-PCR, there was no evidence that hypoxia resulted in differential induction. This is similar to the results of Levy et al.(45), who examined hypoxic induction of VEGF in rat cardiac myocytes, including RT-PCR to determine the specific isoforms expressed and found that the isoforms were coordinately regulated. The significance of the expression of the alternative forms needs further study.

VEGF has been shown to cause increased permeability in the skin after s.c. injection in rabbits(28). Many studies have implicated VEGF as a permeability enhancing factor primarily because of its co-localization in regions of edema, e.g. tumors and delayed hypersensitivity(7, 46). An early study of malignant human astroglial tumor cells described a factor present in conditioned media (likely VEGF) that induced a marked increase in capillary vascular permeability which could be blocked with dexamethasone pretreatment(47). The high levels of expression in normal lung, where there is no evidence of vascular leak suggests that VEGF does not always induce increased permeability. The factors that determine the ability of VEGF to induce vascular leak are not known but possibly higher concentrations (as might be present in injured lung) or receptor expression are involved.

We speculate that expression of VEGF mRNA by pulmonary vascular smooth muscle cells is involved in the pathophysiology of chronic lung disease. In chronic lung disease, hypoxia may induce VEGF expression and dexamethasone therapy may block this augmented VEGF expression. Further study of the role and regulation of VEGF in lung diseases is needed, especially to pinpoint factors that may modulate the permeability enhancing effects of this growth factor.