Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

Vascular endothelial growth factor (VEGF) is a 34- to 46-kD disulfide-linked dimeric protein (1) that was initially identified as a potent mediator of enhanced permeability (2), simultaneous with its discovery as an endothelial cell mitogen (1, 3). Northern blot analysis and in situ hybridization have demonstrated expression of VEGF in most normal tissues (4, 5), with abundant VEGF transcript present in alveolar cells (5, 6). Pulmonary expression of VEGF is crucial for normal lung development (7, 8), and induction of VEGF during chronic hypoxia (6, 9) or recovery from hyperoxia (10) may be important in the process of pulmonary vascular remodeling. VEGF is also a potential mediator of pulmonary endothelial barrier dysfunction, although the role of VEGF in enhanced permeability with acute lung injury is largely unexplored.

Because we have previously shown that pulmonary vascular permeability increased markedly during ventilated ischemia (11), and several studies have demonstrated that VEGF is upregulated in response to myocardial (12) and cerebral (15) ischemia, we wondered whether VEGF was a potential mediator of ischemic lung injury. Pulmonary ischemia differs from ischemia of other organs in that pulmonary ischemia is not synonymous with hypoxia if ventilation with oxygen is maintained when blood flow is impaired. The stimulus for VEGF induction during ischemia of organs of the systemic circulation is thought to be tissue hypoxia (12, 13, 15). However, in vitro studies demonstrated upregulation of VEGF in response to nitric oxide inhibition (9), reactive oxygen species (16, 17), and glucose deprivation (18), all stimuli which may be relevant to the ventilated ischemic lung. In addition, VEGF is released by monocytes (19), eosinophils (20), and aggregating platelets (21), which may be pertinent with an inflammatory stimulus.

We therefore hypothesized that VEGF would increase during ventilated ischemia, independent of oxygen tension, and that increased expression of VEGF would be linked to endothelial barrier dysfunction. As a first test of this hypothesis, we measured the expression and distribution of VEGF in lungs exposed to ventilated ischemia and compared VEGF induction with increased vascular permeability to protein, measured in the same lungs. In addition, we evaluated the oxygen dependence of VEGF expression and injury during pulmonary ischemia. Because hypoxic upregulation of VEGF in ischemic organs is thought to be mediated by hypoxia-inducible factor (HIF)-1 (22), we also analyzed expression of HIF-1 as a function of ischemic duration and oxygen tension.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Preparation

Adult male ferrets were anesthetized with pentobarbital sodium (50 mg/kg intraperitoneally). After tracheostomy, ventilation was maintained with a warmed, humidified gas containing 28% O₂ at a frequency of 20 breaths/min and a tidal volume of 12 ml/kg. An abdominal aortic cannula was placed via midline incision, heparin was administered (1 mg/kg), and the animals were rapidly exsanguinated. After exsanguination, ventilation was continued with a frequency of 10 breaths/min and an end-expiratory pressure of 3 mm Hg, and the ventilatory gas mixture contained 5% CO₂ and oxygen concentrations as specified in the protocols described later. Cannulas were inserted into the left atrium via the left ventricle and into the pulmonary artery via the right ventricle, then the lungs were excised. Pulmonary arterial, left atrial, and airway pressures were continuously monitored (Grass model 7) with Statham P50 transducers referenced to the top of the lung. Residual blood was flushed from the lungs with physiologic salt solution containing 3 g/dl fatty acid-free porcine albumin and 2 g/dl Ficoll (PSS), as previously described (11).

After ventilated ischemia, the lungs were suspended from a force transducer for measurement of lung weight gain. Ventilation was transiently interrupted and freeze-clamp biopsies were taken from the right lung for assay of VEGF expression. The right hilum was then clamped, ventilation was restarted with a 50% reduction of tidal volume, and osmotic reflection coefficient for albumin (_alb) was measured in the left lung as described later. Alternatively, separate groups of lungs were fixed for immunohistochemistry after ventilated ischemia.

Protocol

Isolated lungs were exposed to either 45 (short ischemia, n = 12) or 180 (long ischemia, n = 12) minutes of ventilated ischemia (37°C). In each group, half of the lungs were ventilated with 95% O₂ (hyperoxic ischemia), and half with 95% N₂ (hypoxic ischemia), to determine whether upregulation of VEGF during ischemia was oxygen-dependent. Results were compared with control lungs (n = 6) ventilated with 16% O₂, and flushed with PSS containing 5 mM glucose. Control lungs were exposed to a minimal period of ischemia ( 20 min) during the isolation period. Five separate groups of lungs (n = 3) were subjected to the same duration of ventilated ischemia or control conditions, then were fixed for immunohistochemistry immediately after the specified ischemic period.

Measurements

Messenger RNA expression. Lung tissue was snap-frozen in liquid nitrogen, homogenized in STAT-60 RNAzol lysis buffer (Tel-Test, Inc., Friendswood, TX), then total RNA was isolated for Northern blot analysis. RNA was fractionated by 1% agarose gel electrophoresis, then transferred to Gene Screen Plus membrane filters (NEN Life Science Products, Boston, MA) using 10× saline sodium citrate. Membranes were then hybridized with [³²P]-labeled complementary DNA (cDNA) encoding either full-length human or a 363-base pair (bp) fragment of rat VEGF₁₂₁ (graciously provided by Marsha Merrill [National Institute of Neurologic Disorders and Stroke, National Institutes of Health]). This probe labels all isoforms of VEGF. Hybridization and wash conditions were as previously described (26). Autoradiogram signals were quantified by densitometric analysis (Molecular Dynamics, Sunnyvale, CA). After stripping of the VEGF probe in wash buffer (1% sodium dodecyl sulfate [SDS], 40 mM phosphate buffer [pH 8], and 1 mM ethylenediaminetetraacetic acid), the same membranes were probed with [³²P]-labeled cDNA encoding a 900-bp fragment of human HIF-1 cDNA. To control for variation in either the amount of RNA in different samples or loading errors, blots were stripped of the original probe in wash buffer, then hybridized with a 24-bp oligonucleotide (5' ACG GTA TCT GAT CGT CTT CGA ACC 3') corresponding to 18s RNA, graciously provided by Augustine Choi (Yale University School of Medicine) (data not shown). As a positive control for VEGF labeling, RNA was extracted from a fibroblast cell line exposed to hypoxia, known to express VEGF.

Protein expression. Tissue samples from the same lungs were analyzed for VEGF and HIF protein expression by Western blot analysis. Samples were solubilized by homogenization in Tris buffer containing phenylmethylsulfonyl fluoride and aprotinin, boiled for 5 min, then resolved by SDS-polyacrylamide (12% or 8%, for VEGF and HIF Western blots, respectively) gel electrophoresis, as described by Laemmli (27). Proteins were then transferred onto nitrocellulose membranes and stained using rabbit polyclonal antihuman antibodies against the C-terminus of VEGF (Santa Cruz Biotechnology, Santa Cruz, CA), or antihuman HIF-1 antibodies. After washing away the primary antibody, peroxidase-conjugated avidin secondary antibody was used for visualization. Autoradiograms were quantitated by densitometric analysis as previously described. To control for differences in protein concentration between samples and for loading errors, gels were stained with Coomassie blue and membranes were also labeled with monoclonal rabbit antihuman antibodies against the constitutive protein, HIF-1.

Immunohistochemistry. Lungs were fixed for immunohistochemical evaluation after the specified ischemic periods by intratracheal instillation (25 cm H₂O) of freshly made 4% paraformaldehyde in 0.1 M phosphate buffer (pH 7.2), followed by submersion in the same fixative (4°C) for 36 to 48 h. After fixation, tissue blocks were paraffin-imbedded, then 5-µm sections were stained using the streptavidin-peroxidase method (28, 29). Briefly, slides were deparaffinized in three changes of xylene, with rinses in absolute, 95%, and 80% ethanol. After blocking endogenous peroxidase activity with 3% hydrogen peroxide in phosphate-buffered saline (PBS), slides were incubated overnight (4°C) with primary antibody, diluted 1:500 in PBS containing 1% bovine serum albumin (BSA). The secondary biotinylated antibody was used at a 1:500 dilution in PBS with 1% BSA and 5% nonimmune serum used as a blocker of nonspecific attachment of the antibody. Streptavidin horseradish peroxidase (Jackson ImmunoResearch, West Grove, PA) was used at a 1:1,000 dilution in PBS containing 1% BSA. Aminoethylcarbazole (Sigma Chemical Co., St. Louis, MO) was used as the chromagen, and slides were counterstained with Lerner-2 hematoxylin (Lerner Laboratories, Pittsburgh, PA) and mounted with Crystal Mount (Biomeda Corp., Foster City, CA).

Vascular permeability. We measured _alb by methods previously described in detail (11). Briefly, the pulmonary vasculature was filled with the same PSS mixture that was present during ischemia, to which washed ferret red blood cells (RBCs) (mean hematocrit [Hct] 21.9 ± 1.5%) were added. The pulmonary arterial and left atrial cannulas were connected to a common reservoir containing the PSS/RBC mixture, which was pre-equilibrated with the ventilatory gas. Intravascular pressure was raised to 30 mm Hg in 5-mm Hg increments over 15 min, then held at 30 mm Hg for 20 min, until the rate of lung weight gain was constant. The left atrial cannula was then connected to a peristaltic pump (Gilson Minipuls, Gilson, Middleton, WI) in series with a fraction collector (Gilson 203), adjusted to pump samples from the left atrial cannula in 1 ml aliquots. Hct and albumin concentration (30) were measured in duplicate for each sample, then _alb was estimated iteratively from the equation: C/C_i = {1  Hct_i  [(1  Hct)/(1  Hct_i)]^x}/(1  Hct_i  ), where x = (1  Hct_i  )/Hct, Hct represents RBC concentration, C represents protein concentration, and i represents initial value in the reservoir (11).

Statistical Analysis

Differences between groups in _alb and in fold-induction of messenger RNA (mRNA) and protein were compared using one-way analysis of variance. When significant variance ratios were obtained, least significant differences were calculated to allow comparison of individual group means (31). Differences were considered significant at P  0.05.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

VEGF mRNA Expression

Figure 1 demonstrates VEGF mRNA expression in a representative experiment after 180 min of hyperoxic ischemia (Figure 1, lane 2) in an isolated ferret lung. The predominant band labeled after hybridization with human VEGF cDNA is at approximately 3.9 kb, consistent with transcript for VEGF₁₆₅. Shown as a positive control is RNA extracted from a rat fibroblast cell line after exposure to hypoxia, which is known to induce VEGF mRNA expression. Representative VEGF mRNA expression from lungs exposed to short (45 min) or long (180 min) ischemia with hyperoxic ventilation is seen in the upper panel of Figure 2A (lanes 2 and 3, respectively). VEGF mRNA increased at both time points when compared with control lung. Interestingly, a similar degree of VEGF mRNA induction was seen after short or long ischemia with hypoxic ventilation (Figure 2A, upper panel, lanes 4 and 5, respectively). Mean fold-induction for all experiments is shown in the upper panel of Figure 2B. VEGF mRNA doubled after short ischemia, with 3-fold induction after long ischemia, regardless of oxygen tension in the ventilatory gas (P < 0.05 for hyperoxic and hypoxic long ischemia versus control).

View larger version (28K): [in this window] [in a new window]   Figure 1.   Northern blot analysis of ischemic ferret lung RNA (lane 2) hybridized with full-length human VEGF cDNA. Shown for comparison as a positive control is RNA from a rat fibroblast cell line exposed to hypoxia, known to express VEGF (lane 1). Equal loading of samples was confirmed by ethidium bromide staining of the gel.

View larger version (27K): [in this window] [in a new window]   Figure 2.   (A) Upper panel: Representative Northern blot analysis of VEGF expression during ventilated ischemia. Even loading of RNA samples was confirmed by ethidium bromide staining of the gel and labeling the membranes with 18s. Lower panel: Representative VEGF protein expression after ventilated (95% or 0% O2) ischemia, assessed by Western blot. Even loading of protein samples was confirmed by Coomassie blue staining of the gel. Lane 1, control; lane 2, short ischemia (95% O2); lane 3, long ischemia (95% O2); lane 4, short ischemia (0% O2); lane 5, long ischemia (0% O2). (B) Upper panel: Mean fold-induction of VEGF mRNA, measured by Northern blot analysis, during ventilated (95% or 0% O2) ischemia. Data are expressed relative to VEGF mRNA in control lungs. Middle panel: Mean fold-induction of VEGF protein during ventilated ischemia, measured by Western blot analysis, is expressed relative to VEGF protein in control lungs. Lower panel: Mean osmotic reflection coefficient for albumin, measured in the contralateral lung for the experiments shown in upper and middle panels. Values represent mean values (± SE) from six experiments in each group; *P < 0.05 versus control; P < 0.05 versus control, short ischemia.

Next, VEGF protein was measured as a function of ischemic duration and oxygen tension in the same lungs. The lower panel of Figure 2A shows representative results of a Western blot analysis for VEGF protein. A 22- to 23-kD protein, consistent with VEGF, was present in control lungs (Figure 2A, lane 1). After short hyperoxic ischemia (Figure 2A, lanes 2 and 3, respectively), increased amounts of this 22- to 23-kD protein were seen, and there was labeling of a 16-kD protein, which may also represent VEGF. Results were comparable after the same durations of hypoxic ischemia (Figure 2A, lanes 4 and 5). Mean fold-expression of protein, measured by densitometric analysis of the 23-kD band, is shown in the middle panel of Figure 2B. VEGF protein levels substantially increased in all ischemic lungs as compared with control.

To determine where VEGF was localized in the lung parenchyma, and whether the distribution of VEGF was altered by ischemia, immunohistochemistry was performed on separate groups of lungs exposed to short or long hyperoxic or hypoxic ischemia. The pattern seen in control lungs, shown at low power in Figure 3, was typified by staining of bronchiolar epithelium and smooth muscle. This pattern of airway staining was not altered by ischemia. Higher-power examination, shown in Figure 4A, demonstrated staining of cuboidal alveolar cells in corners, typical of type II pneumocytes. A similar staining pattern was seen in lungs exposed to short hyperoxic ischemia (Figure 4B). In contrast, after long hyperoxic ischemia (Figure 4C), there was the appearance of diffuse immunohistochemical staining for VEGF along alveolar septae.

View larger version (73K): [in this window] [in a new window]   Figure 3.   (Left) Immunohistochemical staining for VEGF in a control lung under low-power examination (×85) demonstrating staining of bronchiolar epithelium and smooth muscle. This staining pattern of airways was not altered by ischemia.

View larger version (159K): [in this window] [in a new window]   Figure 4.   Immunohistochemical staining for VEGF. Results are representative of three separate experiments. (A) Control lung: High-power field (×285) demonstrating staining of cuboidal cells in alveolar corners, marked with arrows, morphologically consistent with type II pneumocytes. (B) Short ischemia-hyperoxia: A staining pattern is seen similar to that in control lungs. (C) Long ischemia-hyperoxia: Increased staining is noted along alveolar septae, marked by arrowheads, compared with control lungs. (D) Short ischemia-hypoxia: Staining along alveolar septae is seen on high-power view, as compared with short ischemia-hyperoxia. (E) Long ischemia-hypoxia: Intense staining is seen along alveolar septae on high-power view, as is seen after long ischemia-hyperoxia. (F ) Control antibody: Specificity of VEGF antibody staining was confirmed substituting nonimmune rabbit immunoglobulin G as the primary antibody after long ischemia-hyperoxia.

In contrast to short hyperoxic ischemia, immunostaining after short hypoxic ischemia (Figure 4D) revealed patchy staining for VEGF along alveolar septae. While less intense than staining seen after long hyperoxic ischemia, the appearance of alveolar staining in this group of lungs clearly differed from lungs exposed to the same duration of hyperoxic ischemia, despite similar increases in VEGF protein expression in both groups measured by Western blot analysis. Marked VEGF immunostaining along alveolar septae occurred after long hypoxic ischemia (Figure 4E), similar to the pattern seen after long hyperoxic ischemia.

Interestingly, pulmonary vascular permeability to albumin increased under the same conditions in which redistribution of VEGF protein to alveolar septae was detected. Shown in the lower panel of Figure 2B is mean _alb after short or long ventilated ischemia. _alb averaged 0.69 ± 0.09 in control lungs, and was not significantly decreased after short hyperoxic ischemia (0.50 ± 0.12). However, permeability increased significantly after long hyperoxic ischemia, as evidenced by a decrease in _alb to 0.20 ± 0.03 (P < 0.05 compared with control and short ischemia). In contrast, after hypoxic ischemia, pulmonary vascular protein permeability increased after both short and long ischemia (_alb 0.26 ± 0.08 and 0.23 ± 0.03, respectively, as compared with control lungs).

To determine whether increased VEGF expression during pulmonary ischemia might be mediated by HIF-1, expression of HIF-1 mRNA and protein was evaluated as a function of ischemic duration and oxygen concentration. Representative results of Northern blot analysis for HIF-1 mRNA are shown in the upper panel of Figure 5A. Expression of HIF-1 increased with short ischemia (Figure 5A, lanes 2 and 4), with further induction after long ischemia (Figure 5A, lanes 3 and 5), as compared with control (Figure 5A, lane 1). Mean fold-induction for all experiments is shown in the upper panel of Figure 5B. HIF-1 mRNA increased 4-fold during 180 min of ischemia, and, interestingly, this upregulation of HIF-1 mRNA was independent of the oxygen tension during ischemia.

View larger version (26K): [in this window] [in a new window]   Figure 5.   (A) Representative Northern (upper panel) and Western (lower panel) blot analysis of HIF-1 expression during ventilated ischemia. Even loading of RNA samples was confirmed by ethidium bromide staining of the gels, and even loading of protein was confirmed by Coomassie blue staining of the gels and labeling of membranes with HIF-1 antibodies. Lane 1, control; lane 2, short ischemia (95% O2); lane 3, long ischemia (95% O2); lane 4, short ischemia (0% O2); lane 5, long ischemia (0% O2). (B) Mean fold-induction of HIF-1 mRNA (upper panel) and protein (lower panel) during ventilated (95% or 0% O2) ischemia. Values represent the mean (± SE) of six experiments in each condition; *P < 0.05 versus control.

Representative expression of HIF-1 protein is shown in the lower panel of Figure 5A. HIF-1 protein expression increased only minimally during hyperoxic ischemia (Figure 5A, lanes 2 and 3), but increased robustly during hypoxic ischemia (Figure 5A, lanes 4 and 5), as compared with control lungs (Figure 5A, lane 1). Mean data for all experiments, shown in the lower panel of Figure 5B, demonstrated a 6-fold increase in HIF-1 protein level with long hypoxic ischemia (P < 0.05). In contrast, HIF-1 protein expression did not significantly differ from control during hyperoxic ischemia.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

Prior studies have demonstrated that abundant VEGF mRNA is present under normal circumstances in the lungs of adult animals (5, 6), yet the physiologic and pathophysiologic functions of VEGF in the adult lung are largely unexplored. The current study was undertaken to investigate the potential role of VEGF as a mediator of enhanced pulmonary vascular permeability during acute lung injury, using a well-characterized isolated lung model of pulmonary ischemic injury. With this model we have previously demonstrated that exposure of isolated ferret lungs to 180 min of ventilated ischemia significantly increased pulmonary vascular permeability to both water and protein (11). Our current data demonstrate increased expression of both VEGF mRNA and protein during the same duration of ventilated ischemia. The effects of pulmonary ischemia on VEGF expression have not been previously reported. VEGF induction has been described in vivo and in vitro in response to several biochemical stimuli which are potentially relevant to the ischemic lung, including hypoxia (6, 9, 13), glucose deprivation (18), inhibition of nitric oxide (9), reactive oxygen species (16, 17), and various inflammatory cytokines (32). Our earlier studies demonstrated that several of these factors, including hypoglycemia (33), oxidant generation (34), and NO inhibition (35) during ischemia modulated increased vascular permeability in our isolated ferret lung model. In addition, because upregulation of VEGF in the developing lung has been demonstrated in association with type II cell differention (36), it is possible that VEGF expression in the ischemic lung is due to type II cell proliferation, as a reparative response to injury (37).

Although VEGF protein increased in lung homogenates after both 45 and 180 min of hyperoxic and hypoxic ischemia, immunohistochemical staining for VEGF demonstrated that distribution of VEGF protein differed among groups. This redistribution was manifest by the appearance of VEGF protein along alveolar septae in lungs exposed to long hyperoxic ischemia and short or long hypoxic ischemia, as compared with control lungs and lungs exposed to short hyperoxic ischemia. This altered staining pattern suggests increased VEGF expression in, or redistribution to, alveolar capillaries, the presumed site of endothelial barrier dysfunction, although we cannot exclude the possibility that this pattern reflects staining of type I pneumocytes. The three groups of ischemic lungs in which this altered immunohistochemical staining pattern was seen were the same three groups in which pulmonary vascular permeability during ischemia increased significantly. Others have reported increased vascular permeability in response to exogenous VEGF administration in skin (38), muscle (40), gastrointestinal tract (41), and airways (41), as well as increased permeability of large vessel (42, 43) and microvascular (44, 45) endothelial cells in culture. Increased VEGF expression in the lung in response to a physiologic stimulus has not previously been linked to pulmonary vascular barrier dysfunction.

Expression of VEGF has been shown in vivo in response to coronary (12) and cerebral (15) ischemia, in both animal and human studies, with a time course similar to that seen in the current study. In those studies, attention was focused on the role of VEGF in angiogenesis during ischemia, and the stimulus for VEGF induction was thought to be tissue hypoxia (12, 13, 15). Somewhat surprisingly, we found that the degree of VEGF expression during ventilated pulmonary ischemia was independent of oxygen concentration in the ventilatory gas mixture. Hypoxia is thought to upregulate VEGF primarily via activation of the transcription factor HIF-1 (22). There are conflicting published data regarding whether HIF-1 is involved in upregulation of VEGF mRNA in response to glucose deprivation in embryonic stem cells (46). We therefore wondered whether HIF-1 played a role in the oxygen- independent increased VEGF expression seen during ventilated pulmonary ischemia. HIF-1 mRNA increased to a similar degree during 180 min of ventilated ischemia, regardless of oxygen concentration. In contrast, HIF-1 protein did not increase significantly during hyperoxic ischemia but increased 6-fold in hypoxic ischemic lungs. The time course for HIF-1 protein expression in the hypoxic ischemic ferret lung was remarkably similar to that seen in isolated perfused ferret lungs exposed to oxygen concentrations of less than 4% (49). Although activation of HIF-1 during pulmonary ischemia has not yet been evaluated, these results suggest that the stimulus for VEGF induction may differ between pulmonary ischemia and ischemia in other organs. The lung is the sole organ in which the effects of hypoxia and ischemia can be evaluated independently, thus providing a unique model to further explore potential mechanisms of VEGF induction during ischemia.

In summary, we have demonstrated that VEGF increased during 180 min of pulmonary ischemia, in an oxygen-independent fashion. Induction of VEGF expression during hypoxic ischemia was associated with increased expression of HIF-1 protein, suggesting a possible causal relationship, whereas VEGF expression during hyperoxic ischemia was likely mediated by a different mechanism. In addition, redistribution of VEGF protein to alveolar septae, demonstrated by immunohistochemical staining, occurred under the same ischemic conditions that were associated with increased pulmonary vascular permeability to protein. These findings suggest that VEGF may be an important mediator of acute lung injury following pulmonary ischemia. Inhibition of VEGF may therefore represent a potential therapeutic strategy to attenuate clinically relevant ischemic lung injury, as occurs, for example, during lung preservation for autologous transplantation.