Nature Medicine
8, 702 - 710 (2002)
Published online: 10 June 2002; | doi:10.1038/nm721
There is a Corrigendum (November 2002) associated with this Article.
Loss of HIF-2 and inhibition of VEGF impair fetal lung maturation, whereas treatment with VEGF prevents fatal respiratory distress in premature miceVeerle Compernolle1, Koen Brusselmans1, Till Acker2, Peter Hoet3, Marc Tjwa1, Heike Beck2, Stéphane Plaisance1, Yuval Dor4, Eli Keshet4, Florea Lupu5, Benoit Nemery3, Mieke Dewerchin1, Paul Van Veldhoven6, Karl Plate2, Lieve Moons1, Désiré Collen1
& Peter Carmeliet11 The Center for Transgene Technology and Gene Therapy, Flanders Interuniversity Institute for Biotechnology, Leuven, Belgium 2 Neurological Institute, JWG Frankfurt University, Frankfurt, Germany 3 Laboratory of Pneumology, Unit of Lung Toxicology, KU Leuven, Leuven, Belgium 4 Department of Molecular Biology, Hebrew University-Hadassah Medical School, Jerusalem, Israel 5 Cardiovascular Biology Research Program, Oklahoma Medical Research Foundation, Oklahoma City, Oklahoma, USA 6 Department of Pharmacology, KU Leuven, Leuven, Belgium
Correspondence should be addressed to Peter Carmeliet peter.carmeliet@med.kuleuven.ac.beRespiratory distress syndrome (RDS) due to insufficient production of surfactant is a common and severe complication of preterm delivery. Here, we report that loss of the hypoxia-inducible transcription factor-2 (HIF-2) caused fatal RDS in neonatal mice due to insufficient surfactant production by alveolar type 2 cells. VEGF, a target of HIF-2, regulates fetal lung maturation: because VEGF levels in alveolar cells were reduced in HIF-2-deficient fetuses; mice with a deficiency of the VEGF164 and VEGF188 isoforms or of the HIF-binding site in the VEGF promotor died of RDS; intrauterine delivery of anti-VEGF-receptor-2 antibodies caused RDS and VEGF stimulated production of surfactant proteins by cultured type 2 pneumocytes. Intrauterine delivery or postnatal intratracheal instillation of VEGF stimulated conversion of glycogen to surfactant and protected preterm mice against RDS. The pneumotrophic effect of VEGF may have therapeutic potential for lung maturation in preterm infants.Preterm delivery is the chief problem in obstetrics, affecting 10% of all births and accounting for more than 70% of perinatal mortality1. 60% of the infants born at less than 32 weeks of gestation and weighing less than 1,000 g develop respiratory distress syndrome (RDS) with a mortality of 50% (ref. 2). RDS results from insufficient production of surfactant by immature alveolar type 2 pneumocytes. Surfactant is a mixture of phospholipids and surfactant-associated proteins (SP-A to SP-D), which lowers surface tension at the air-water interface and prevents alveolar collapse. Surfactant phospholipids are synthesized from metabolic substrates, provided by glycogen stores in fetal immature pneumocytes3. Neonatal intensive care and treatment with oxygen and steroids have improved the survival of infants with RDS, but often at the expense of the development of bronchopulmonary dysplasia or chronic lung disease of prematurity and other side-effects4. The pathogenesis of RDS remains incompletely understood.
Interactions between airways and blood vessels are critical for normal lung development5. A major factor in lung vascular development is vascular endothelial growth factor (VEGF), which binds the receptors Flk-1 (also known as VEGF receptor-2) and Flt-1 (also known as VEGF receptor-1)6. There are three VEGF isoforms: a diffusable VEGF120, a matrix-bound VEGF188 and VEGF164, which can bind matrix and is also diffusable. VEGF is deposited at the leading edge of branching airways, where it stimulates vascularization7. However, indirect evidence suggests that VEGF also affects epithelial growth and differentiation. Type 2 pneumocytes and bronchiolar epithelial cells produce VEGF and express VEGF receptors8,
9. VEGF levels are also considerably higher in the bronchoalveolar fluid than in the blood9, suggesting that epithelial cells affect their own function by releasing VEGF into the airway lumen. Furthermore, VEGF levels in tracheal aspirate were lower in infants with lung immaturity developing bronchopulmonary dysplasia than in those surviving without pulmonary complications in some10,
11,
12 but not in other studies13. Exogenous VEGF stimulates growth of lung epithelial cells in vitro14, but the relevance of endogenous VEGF for lung maturation in vivo and the possible therapeutic potential of VEGF in preventing RDS in preterm infants remain unknown. Hypoxia upregulates VEGF gene transcription by activating the hypoxia-inducible transcription factors HIF-1 and HIF-2 (refs. 15,
16,
17), which bind the hypoxia-response element in the VEGF promotor. HIF-2 is expressed in fetal type 2 pneumocytes16, but its relevance for RDS remains unknown. In studying the role of HIF-2 and VEGF in fetal-lung maturation, here we reveal a potential use of VEGF for treatment of RDS.
HIF-2-/- neonates succumb to RDS
Deficiency of HIF-2 was accomplished by deletion of the second exon of the gene, which encodes the DNA-binding domain18. HIF-2-/- embryos represented approximately 25% of the littermates until embryonic day (E) 13.5, when half of the embryos died of cardiac failure. At birth, wild-type (WT) neonates breathed regularly, were well oxygenated and actively moved their limbs, whereas HIF-2-/- neonates breathed irregularly with gasping and signs of retraction, had a cyanotic skin color and succumbed within 2 to 3 hours in severe respiratory failure due to extensive lung collapse (Fig. 1a−c). In WT newborns, lung aeration (percentage of the total surface filled with air) doubled after birth and almost achieved adult levels (68  2%), but failed to increase at all in HIF-2-/- newborns (Table 1). RDS was not attributable to growth retardation, respiratory muscle dysfunction, lung hypoplasia, impaired fluid clearance, hypoxic stress or other organ defects.
 | | Figure 1. Impaired lung maturation and RDS in HIF-2-/- mice. | | |  | a, Skin oxygenation is normal in WT (upper) neonates but cyanotic in HIF-2-/- (lower) littermates. b and c, Normal inflation of WT lungs (b), but lung collapse in HIF-2-/- littermates (c). d and e, PAS+ (glycogen-rich) cells are minimal in WT neonates (d), but abundant (arrowhead in inset) in HIF-2-/- lungs (e). f, PAS+ cells progressively disappear in WT ( ), but not in mutant lungs ( ). *, P < 0.05 versus WT. n = 3−5. g and h, Semi-thin lung sections (toluidine blue), revealing thinning of the alveolar septa in WT (g) but not in mutant (h) neonates. i and j, SP-D+ alveolar type 2 cells are more numerous (arrow) in WT (i) than in HIF-2-/- (j) neonates. k and l, Transmission electromicrographs revealing an alveolar type 2 cell containing surfactant lamellar bodies (k; arrow) in WT lungs. In HIF-2-/- lungs (l), lamellar bodies (arrow) persist in the alveolar lumen ('L', lined with dashed line) and fail to form myelin structures and a surfactant layer. Scale bars, 100  m (b and c), 25 m (d and e), 20 m (g−j).
 Full Figure and legend (111K) |
| |
| | Table 1. Pulmonary maturity evidenced by the degree of aeration and thickness of alveolar septa | | |  |
Full Table |
| |
Defective surfactant production in HIF-2-/- mice
Loss of HIF-2 did not affect lung development during the pseudoglandular or canalicular phase. In the saccular phase (E17.5 through postnatal day (P) 0), both genotypes had a comparable density of terminal sacs, septa per terminal sac, amount of elastin per alveolar septa and airspace. However, thinning of the alveolar septa at birth, a prerequisite for blood-gas exchange, was impaired in HIF-2-/- mice (Fig. 1d and e; Table 1). This was not attributable to abnormal epithelial proliferation or apoptosis, but to impaired differentiation. Immature epithelial cells contain abundant PAS+ glycogen stores, which they convert to surfactant phospholipids. Beyond E18.5, PAS+ cells disappeared in WT mice but persisted in HIF-2-/- mice (Fig. 1d−f). Semi-thin sections confirmed the abundance of immature pneumocytes in HIF-2-/- mice (Fig. 1g and h). HIF-2-/- mice produced less surfactant phospholipids (nmol/lung phosphatidylcholine and phospholipids, 190 22 and 980 38 in WT versus 110 1 and 680 45 in HIF-2-/- lungs, n = 5, P < 0.05; phosphatidylcholine:sphingomyelin ratio, 3.5 0.2 in WT versus 2.8 0.2 in HIF-2-/- lungs; n = 5, P < 0.05) HIF-2-/- mice also produced less SP-A, SP-B and SP-D (copies per 100 copies  -actin for SP-A, SP-B and SP-D: 330 14, 160 11 and 5 0.4 in WT versus 230 34, 110 40 and 3 1 in HIF-2-/- lungs, respectively, n = 6, P < 0.05). They also had fewer SP-B and SP-D positive type 2 pneumocytes (positive cells per mm alveolus for SP-B and SP-D, 15 1.3 and 8 1.3 in WT versus 10 0.75 and 3 0.75 in HIF-2-/-, n = 5, P < 0.05, Fig. 1i and j). In contrast to the presence of surfactant lamellar bodies inside WT type 2 pneumocytes, abundant alveolar secretions of abnormal lamellar surfactant structures were often present in HIF-2-/- mice (Fig. 1k and l), as occurs in mice lacking SP-D (ref. 19). Alveolar epithelial defects were specific, as similar numbers of PGP9.5+ neuroepithelial cell bodies and CC10+ Clara cells were present in both genotypes.
Pulmonary angiogenesis in HIF-2-/- mice
Vascular development during the pseudoglandular and canalicular stages was normal in both genotypes, presumably because pulmonary VEGF levels were comparable. However, one day before birth, a subtle deficit in vascularization of alveolar septa was detected in HIF-2-/- mice (vessels per alveolus, 17 1 in WT versus 13 1 in HIF-2-/-, n = 5, P < 0.05) (Fig. 2a and b). Onset of these vascular defects in HIF-2-/- mice coincided with the time when expression of HIF-2 and VEGF in alveolar epithelial cells was upregulated in WT but not in HIF-2-/- fetuses. In addition, alveolar capillaries failed to remodel properly in HIF-2-/- mice before birth. At E17.5, capillaries lie scattered amidst thick septa (Fig. 2c); however, during subsequent maturation at E19.0, capillaries become aligned in two layers, juxtaposed to the alveolar lumen (Fig. 2d). Upon ventilation after birth, alveolar expansion further stretches the septa, so that capillaries are aligned in a single layer and gas exchange is facilitated (Fig. 2e). In WT mice, capillaries lied juxtaposed to the lumen in 95 1% of terminal sacs, whereas in HIF-2-/- neonates, capillaries were separated from the lumen in 31 5% of terminal sacs (n = 5, P < 0.05) (Fig. 2a and b). However, alveolar vessels in HIF-2-/- fetuses were not leaky and had normal basement membranes (Fig. 2f and g). Muscularization of peripheral vessels (Fig. 2h and i) and branching of large vessels following the bronchiolar structures (Fig. 2j and k) were also normal.
| | Figure 2. Pulmonary vascular development in HIF-2-/- mice. | | |  | a and b, Thrombomodulin staining, revealing a double layer of capillaries, aligned in the immediate vicinity of the alveolar lumen ('L') in WT mice (arrows in a), whereas capillaries in  30% of the alveoli lay more distant from the alveolar lumen ('L') in HIF-2-/- mice (arrows in b) at E18.5. c−e, Schematic illustration of capillary remodeling in alveolar septa. At E17.5, capillaries lie randomly scattered in the thick alveolar septum (c). At E19.0, the alveolar septum becomes thinner, and capillaries lie in a double layer aligned with and immediately juxtaposed to the alveolar lumen (d). On ventilation at birth, further thinning of the septum results from inflation-induced stretching so that the alveolus expands and capillaries are aligned in a single layer (e). f and g, Laminin staining, revealing normal basement membrane formation of microvessels in alveolar septa in WT (f) and HIF-2-/- (g) fetuses at E18.5. Alveolar lumen is marked with L. h and i, Smooth muscle -actin staining, revealing normal muscularization of large pulmonary vessels in WT (h) and HIF-2-/- (i) mice at E18.5. j and k, Angiogram, revealing normal branching of large pulmonary vessels in WT (j) and HIF-2-/- (k) fetuses at E18.5. Scale bars, 10 m (a, b), 20 m (f and g), 50 m (h and i).
Full Figure and legend (76K) |
| |
Expression of HIF-2 during pulmonary maturation
HIF-2 transcript levels were comparable in the heart, lungs and kidneys in WT fetuses at E16.5 (Fig. 3a). Thereafter, expression of HIF-2 increased more than five-fold in the lungs, but less than two-fold in the heart and kidneys (Fig. 3a). By immunostaining, HIF-2 was abundantly expressed in the nucleus of alveolar pneumocytes (Fig. 3c and d), whereas HIF-1 was detectable in bronchiolar but not in alveolar epithelium (Fig. 3e), consistent with previous findings16,
20. Double-immunostaining for HIF and alkaline phosphatase (AP, a marker of type 2 pneumocytes) confirmed that type 2 alveolar pneumocytes expressed HIF-2 (Fig. 3f) but not HIF-1 (Fig. 3h). Other cells in alveolar septa, presumably endothelial and mesenchymal cells, also expressed HIF-2 (Fig. 3c and f). Alveolar type 2 cells stained positively for the hypoxia-marker pimonidazole hydrochloride (Fig. 3i), suggesting that hypoxia might have triggered activation of HIF-2 in these cells (as also shown by its nuclear localization).
| | Figure 3. Pulmonary expression of HIF-2, VEGF and its receptors. | | |  | a, HIF-2 mRNA levels increased more than 5-fold in WT lungs and less than 2-fold in hearts and kidneys during the final stage of fetal development. E16.5, ; E17.5, ; E18.5,  and P0,  , * P < 0.05 versus E16.5 b, VEGF protein levels in lung extracts increased beyond E16.5 in WT mice () but not in HIF-2-/- mice (). *, P < 0.05 versus WT. Panels c-k: sections from E18.5 WT lungs; the nucleus is marked with an asterisk; 'L' denotes alveolar lumen and arrowheads in c−i denote alveolar type 2 cells. c and d, HIF-2-immunostaining (c) and nuclear DAPI staining (d) on adjacent sections, revealing nuclear localization of HIF-2- in alveolar cells. e, HIF-1-immunostaining, revealing expression of HIF-1 in bronchiolar (arrows) but not in alveolar cells. f, Double-immunostaining, revealing co-expression of cytosolic alkaline phosphatase (AP; orange) and nuclear HIF-2 (green) in type 2 pneumocytes. g, Double-immunostaining, revealing co-expression in type 2 pneumocytes of AP (red) and Flk-1 (blue), resulting in a pink colorization. Flk-1 was also detectable in other cells in the alveolar septa, presumably in endothelial cells. h, Double-immunostaining, revealing that expression of HIF-1 (green) is undetectable in AP+ (red) type 2 pneumocytes. i, Immunostaining for pimonidazole, revealing hypoxic type 2 alveolar pneumocytes. j and k, VEGF expression in type 2 cells (arrow, in situ hybridization; j) and abundant epithelial localization of the protein at the alveolar side (arrow, immunostaining; k). l and m, Immunostaining for Flt-1 (l) and CD31 (m) at E17.5, revealing a similar expression pattern, suggesting that Flt-1 is primarily expressed on endothelial cells (arrow). Scale bars, 10 m (c,d,f,g,i,j,l and m); 20 m (e and h), 5 m (k).
Full Figure and legend (83K) |
| |
Role of VEGF in alveolar epithelial maturation
Because VEGF is a downstream target of HIF-2 and has been implicated in neonatal lung disease10,
11,
12, we analyzed a number of previously generated VEGF mutant mouse strains. A fraction of VEGF120/120 mice (selectively expressing VEGF120)21 died of RDS and had abundant PAS+ pneumocytes (cells per mm, 2 1 in WT neonates versus 51 20 in VEGF120/120 littermates, n = 3−5, P < 0.05) (Table 1). In contrast, pulmonary development was normal in VEGF164/164 or VEGF188/188 mice (expressing exclusively VEGF164 or VEGF188, respectively22), or in placental growth factor (PlGF)-/- mice lacking the VEGF homolog PlGF (Table 1)23. A small fraction of VEGF / neonates that died at birth and lacked the hypoxia-responsive HIF-binding site in their VEGF promotor also suffered lung prematurity (Table 1)24. Thus, absence of critical VEGF isoforms, or impaired HIF-dependent VEGF regulation caused RDS.
Pulmonary VEGF protein levels were comparable in both genotypes at E16.5 (Fig. 3b). During the final stage of fetal development, VEGF levels increased approximately four-fold in WT mice, but only minimally in HIF-2-/- mice (Fig. 3b). Compared with WT neonates, pulmonary transcript levels of VEGF120, VEGF164 and VEGF188 were reduced by 31%, 39% and 39%, respectively, in HIF-2-/- mice. By in situ hybridization, VEGF was detectable in type 2 pneumocytes but not in alveolar blood vessels (Fig. 3j). The co-expression of HIF-2 and VEGF in pneumocytes, but not in blood vessels, suggests that HIF-2 regulates VEGF expression primarily in pneumocytes. Abundant VEGF protein was detectable on the apical surface of type 2 cells (Fig. 3k) and in the broncho-alveolar lavage fluid of WT neonates (120 49 pg/mg protein), indicating that VEGF was secreted in the alveolar lumen. No genotypic differences were detected in pulmonary expression of PlGF. Double labeling revealed that Flk-1 was present in septal microvessels, but also in AP+ type 2 pneumocytes (Fig. 3g). By triple labeling, type 2 pneumocytes expressed both Flk-1 and HIF-2 (data not shown). Flt-1 was detectable in capillaries and colocalized with the endothelial marker CD31 on adjacent sections (Fig. 3l and m), whereas the VEGF165-isoform selective receptor neuropilin-1 was undetectable (data not shown).
Notably, freshly isolated type 2 pneumocytes also expressed Flk-1 transcripts (data not shown) and responded to VEGF by increasing their expression of SP-B and SP-C (copies per 100 copies -actin for SP-B and SP-C, 15 1 and 12 1 after saline versus 27 4 and 21 4 after VEGF, n = 4−6, P < 0.05). Thus, surfactant-producing alveolar type 2 cells produce VEGF in a HIF-2-dependent manner and are responsive to VEGF.
Mouse models of lung prematurity and RDS
To evaluate the role of VEGF in lung maturation in an intact animal in vivo, a mouse model of lung prematurity was established. After cesarean section (C-section) at E17.5, approximately 90% of 113 premature newborns exhibited severe signs of lung immaturity and succumbed due to respiratory failure within 10 hours after delivery; this model was used to evaluate the effect of intratracheal instillation of VEGF. After C-section of preterm fetuses at E18.5, more than 90% of 30 neonates suffered RDS during the first hours, but generally survived thereafter. The clinical condition of preterm pups was monitored within the first 20 minutes after delivery by determining an 'activity pulse grimace appearance respiration-like' (APGAR) score on a scale from 0 to 10 (see Methods). In general, saline-treated pups had an APGAR score of 5 0.3 at 5 and 10 minutes after birth (n = 10). This model was used to evaluate the effect of intra-amniotic injection, at E17.5, of VEGF or of antibodies against VEGF receptor. These compounds reliably reached the airways after intra-amniotic administration from E17.5 onwards.
Inhibition of Flk-1 impairs fetal lung maturation
To analyze which VEGF receptor mediated lung maturation, neutralizing anti-Flk-1 or anti-Flt-1 antibodies23,
24 were intra-amniotically injected in WT fetuses at E17.5, and pups were delivered by C-section at E18.5. Anti-Flt-1 antibodies were ineffective, but anti-Flk-1 antibodies prevented the thinning of the alveolar septa and the disappearance of PAS+ cells (Table 2). By immunostaining, intra-amniotically injected antibodies remained restricted to the alveolar compartment, suggesting that the observed effects on lung maturation were due to inhibition of alveolar VEGF. Taken together, Flk-1, not Flt-1, mediates the effect of endogenous VEGF on lung maturation in vivo.
Intra-amniotic VEGF administration prevents RDS
After intra-amniotic VEGF delivery, the APGAR score improved to 7.5 0.7 and 8 0.4 after 5 and 10 minutes, respectively (n = 8), which is significantly better than after saline (P < 0.005). In contrast to the 24 saline-treated pups, of which 75% remained completely immobile after 20 minutes, 60% of the 24 VEGF-treated pups breathed spontaneously and regularly, had a pink skin color after 10 minutes, and actively moved their limbs after 20 minutes (P < 0.02). As a result of the improved aeration after VEGF (Table 2), inflated VEGF-treated lungs floated, while atelectatic control lungs sank to the bottom in a water-filled recipient (Fig. 4d). After VEGF treatment, alveolar septa were thinner, PAS+ glycogen stores were mobilized (Table 2; Fig. 4a and b) and surfactant production was increased (phosphatidylcholine per lung, 180 14 nmol in control versus 220 10 nmol after VEGF, n = 5, P < 0.05). The therapeutic effect of VEGF was specific, as intra-amniotic injection of PlGF, a specific ligand of Flt-1 but not Flk-1, was ineffective (Fig. 4c; Table 2). Notably, VEGF was comparably effective with the glucocorticoid dexamethasone (0.8 mg/kg), administered to pregnant mice at gestational day 15.5 and 16.5. Dexamethasone improved the APGAR score to 6.7 0.7 and 8.4 0.4 at 5 and 10 minutes, respectively (n = 71; P = n.s. versus VEGF; P < 0.005 versus saline for 5 and 10 min, respectively) and stimulated lung aeration (57 3% after dexamethasone versus 58 1% after VEGF; n = 3−9; P = n.s.; as compared with 39 2% after saline) (Table 2). VEGF was slightly more efficient in thinning of the septa than dexamethasone (10 0.1 m after dexamethasone versus 8 0.1 m after VEGF, n = 3−9, P < 0.05). VEGF and glucocorticoids may interact in pulmonary maturation, as pulmonary VEGF levels were increased by a low dose, but suppressed by a high dose of dexamethasone (pg/mg protein at E18.5, 270 12 after saline versus 330 9 and 210 18 after 0.8 or 2.4 mg/kg dexamethasone, respectively, n = 5−11, P < 0.05 versus saline). Thus, intrauterine VEGF improved fetal lung maturation and prevented RDS.
| | Figure 4. VEGF treatment improves lung maturation and protects against respiratory distress. | | |  | a−c, Intra-amniotic treatment of WT E17.5 fetuses with VEGF (b), but not with saline (a) or PlGF (c), reduced the number of PAS+ pneumocytes in fetal lungs. d, Lungs from VEGF-treated fetuses were inflated and floated (right), while lungs from saline-treated fetuses sank to the bottom (left). e and f, Thrombomodulin-staining, revealing a similar number of alveolar capillaries after intra-amniotic VEGF-treatment (f) of WT at E17.5 as compared with saline-treatment (e). g, Western blotting of IgG in plasma and perfused lung extracts, illustrating that intra-amniotic VEGF treatment did not increase extravasation of plasma IgG into the lung parenchyma. h, Premature pups survived longer after intratracheal treatment with VEGF () than with saline () (P < 0.05 versus saline starting from 6 hours). i, Model illustrating the proposed pneumotropic effect of HIF-2 and VEGF on fetal lung maturation: PAS+ glycogen in immature pneumocytes is converted to glucose, and used as substrate for synthesis of surfactant phospholipids. HIF-2 and VEGF stimulate this conversion of glycogen to surfactant and thereby improve fetal lung maturation and protect against RDS. Scale bars, 20 m (a−c), 10 m (e−f).
Full Figure and legend (67K) |
| |
VEGF administration improves lung function and survival
After C-section at E17.5, approximately 60% of preterm pups had an aerated lung area of less than 25% and died immediately after birth (category A), whereas another approximately 10% of preterm pups ventilated well and were normally oxygenated (category B). The remaining approximately 30% of preterm pups had an aerated lung area of 39 2% and suffered severe RDS. These pups were able to live for at least six hours, although they ultimately succumbed to fatal exhaustion (category C). Only mice of class C were intratracheally injected with VEGF (500 ng/pup). To allocate pups to categories A, B and C, newborns were monitored for 30 minutes after C-section, when their lung function and clinical condition was easily scored. Intratracheal VEGF administration prevented RDS in preterm pups of category C. Within 4−6 hours after VEGF administration, breathing became easier and more regular, skin color turned pink and pups moved more actively. As a result, a third of the VEGF-treated pups (n = 22), but none of the controls (n = 14), survived for up to 20 hours when they were killed (P < 0.05) (Fig. 4h). The average survival time was significantly longer after VEGF treatment than in control mice (8.5 1.3 h after saline versus 13 1.3 h after VEGF, n = 14−22, P < 0.05). Histological analysis after six hours revealed that VEGF-treatment improved lung aeration, accelerated alveolar septal thinning and stimulated conversion of glycogen stores, as evidenced by the disappearance of PAS+ cells (Table 2). No differences were found in the number of cells expressing SP-B (positive cells per mm alveolus, 2.6 0.8 after saline versus 3.5 0.9 after VEGF, n = 4, P = n.s.). Considering that intratracheally delivered VEGF could only reach ventilated lung areas (50% of the lung), and taking the short duration of VEGF exposure (6 h) and the young fetal age (E17.5) into consideration, the observed improvement of the clinical condition and lung maturation is remarkable.
Safety of pulmonary VEGF treatment
When 1 g hVEGF was intratracheally administered (resulting in an estimated concentration of 10 g/ml alveolar fluid), less than 0.1% of the hVEGF was recovered in the fetal plasma after 1 hour (500 60 pg/ml hVEGF as compared with 50 4 pg/ml murine VEGF in plasma of uninjected pups). After 3 and 5 hours, hVEGF plasma levels were undetectable (<2 pg/ml), confirming previous findings that VEGF remains restricted to the alveolar compartment with minor spill-over to the interstitium and circulation9. Similar findings were obtained after intra-amniotic injection of hVEGF. Neither intra-amniotic nor intratracheal VEGF stimulated angiogenesis in alveolar septa (Table 2; Fig. 4e and f), vascular leakage (amount of extravasated immunoglobulin G (IgG) in perfused lungs, analyzed by western blotting) (Fig. 4g) or bronchial edema. There were also no microscopic abnormalities, leakiness or neovascular growth in the gastrointestinal tract, placenta or fetal membrane after intra-amniotic delivery.
Discussion
Here we show that HIF-2 and its downstream target VEGF are critical for fetal lung maturation. Loss of HIF-2, absence of critical VEGF isoforms or inhibition of VEGF in utero impaired lung maturation and caused RDS at birth due to insufficient surfactant production. When administered intra-amniotically to unborn fetuses or intratracheally after birth, VEGF increased conversion of glycogen stores to surfactant, improved lung function, protected severely preterm mice against RDS and prolonged their survival, with a comparable efficiency as prenatal steroid treatment but without acute adverse effects.
Our findings indicate that HIF-2 has a critical role in controlling the conversion of glycogen to surfactant. Glycogenolysis, the first step in this pathway, is critical for lung maturation, as defective glycogen breakdown causes insufficient surfactant production in rats with a deficiency of glycogen phosphorylase-B kinase25 and in fetuses from diabetic mothers26. Moreover, glucocorticoids stimulate surfactant production, in part by increasing glycogenolysis27. Hypoxic activation of HIF-2 has a role in this process, given that hypoxia is known to enhance glycogenolysis28, fetal type 2 cells were hypoxic, upregulation of HIF-2 in type 2 pneumocytes coincided with the onset of surfactant production, and HIF-binding sites are present in several genes implicated in glucose metabolism29. HIF-2 could also upregulate the expression of genes encoding surfactant-associated proteins or of genes involved in the conversion of glycogen to surfactant phospholipids (including VEGF). Although we did not detect HIF-1 by immunostaining in hypoxic fetal alveolar cells, the protein has been immunolocalized in adult alveolar cells under extreme hypoxic conditions30, raising the question whether HIF-1 might contribute to fetal lung maturation under more severe conditions.
We also uncovered a novel role of VEGF in lung maturation and surfactant production in vivo. Indeed, pulmonary VEGF levels were reduced in HIF-2-/- neonates with RDS, loss of the long VEGF-isoforms, impaired HIF-dependent upregulation of VEGF, and intrauterine delivery of anti-Flk-1 antibodies caused lung prematurity. HIF-2 developmentally upregulated VEGF expression in alveolar pneumocytes, probably by binding the hypoxia-response element in the VEGF promoter31. VEGF seems to affect alveolar type 2 pneumocytes directly, as these cells expressed Flk-1 and synthesized more SP-B and SP-C in response to VEGF, and anti-Flk-1 antibodies caused RDS without crossing the epithelial barrier. VEGF enhanced surfactant synthesis and improved lung function in vivo rapidly, for example within 4−6 hours, suggesting that its beneficial effect may not require extensive epithelial differentiation but could rely on switching on the metabolic conversion of glycogen to surfactant and/or on surfactant release. Although the precise mechanism remains to be determined, VEGF is known to upregulate synthesis of platelet-activating factor32, a potent inducer of glycogenolysis in fetal lung33 and to activate protein kinase C (ref. 6), a central regulator of surfactant secretion34 and glycogen metabolism35.
Loss of HIF-2 was reported to cause reduced catecholamine levels36 and vascular remodeling defects in embryos37. Catecholamine production was indeed lower in HIF-2-/- neonates, but the RDS was not rescued by treatment of pregnant HIF-2+/- females with D,L-threo-3,4-dihydrophenylserine (DOPS; a substrate that is converted to noradrenaline) (data not shown), indicating that RDS was not attributable to insufficient catecholamine production. We did not detect any defects in pulmonary vascular development in HIF-2-/- fetuses until the last phase of fetal-lung maturation, precisely when HIF-2 and VEGF expression were upregulated in WT but not in HIF-2-/- lungs. Thus, early pulmonary vascular development (including branching and muscularization of proximal lung vessels and initial formation of distal alveolar capillaries) proceeded normally as long as VEGF was not upregulated by the elevated HIF-2 levels. Another reason why pulmonary vascular defects were subtle may relate to the finding that HIF-2 seems to be more important for upregulation of VEGF in alveolar epithelial than in vascular cells (as suggested by our co-expression studies). Pulmonary vascular defects were more severe in VEGF120/120 neonates (data not shown), which suggests a critical role of the longer and most abundant VEGF164 and VEGF188 isoforms in the lung38,
39. As these longer isoforms bind to the subepithelial heparin sulfate-rich extracellular matrix at the branching tips of airways in the distal lung, matrix-associated VEGF may provide critical spatial guidance cues for growing vessels and link airway branching with blood vessel formation7. Although these vascular defects could contribute to impaired lung maturation in HIF-2-/- mice, the findings that RDS develops in the presence (VEGF120/120) or absence (HIF-2-/-) of severe vascular defects suggest that the latter are not a prerequisite for RDS. In addition, VEGF prevented RDS via an epithelial effect without increasing alveolar angiogenesis.
Our findings have potential medical implications. First, reduced levels of HIF-2 and VEGF may identify infants at risk for RDS. Second, intra-amniotic or intratracheal delivery of VEGF improved surfactant production and protected preterm newborns against RDS. The rapidity with which VEGF stimulates conversion of preformed glycogen to surfactant phospholipids (<5 hours) makes VEGF an attractive therapeutic target. VEGF did not cause adverse effects on vascular leakage or bleeding in the lung, possibly because it barely crossed the alveolar epithelium. Third, steroids are often used to induce lung maturation but may cause adverse effects40. Our findings indicate that dexamethasone upregulated pulmonary VEGF expression in fetuses at low doses41, but suppressed VEGF production at a high dose. Thus, excessive amounts of glucocorticoids might counteract the beneficial pneumotrophic effect of VEGF. Fourth, oxygen improves oxygenation of preterm infants with RDS but, as it also suppresses VEGF expression in alveolar type 2 pneumocytes8, it would deprive alveolar cells from pneumotrophic effects. VEGF supplements might also lower the toxicity of high oxygen concentrations in the neonate. In concert, our findings suggest that the pneumotrophic effect of VEGF might have a therapeutic potential for lung maturation in preterm infants at risk for RDS.
Methods
Generation of HIF-2-/- transgenic mice.
HIF-2-/- ES cells18 were used to generate HIF-2-/- mice (129/Sv  Swiss). VEGF120/120 mice21, VEGF164/164 and VEGF188/188 mice22, PlGF-/- mice23 and VEGF/ mice24 were previously generated. Quantitative real-time PCR was performed as described and transcripts were normalized per 100 -actin transcripts21,
23. ELISA-kits (R&D; Systems, Abingdon, U.K) were used to measure murine and human VEGF. Antibodies used for western blotting23: rabbit anti-human Flt-1 (clone Flt-11; Sigma, Bornem, Belgium), rabbit anti-mouse Flk1 (Santa Cruz, Sanvertech, Boechout, Belgium), rat anti-mouse endoglin (Pharmingen, BD Biosciences, Erembodegem, Belgium) and rat anti-mouse VE-Cadherin (gift from E. Dejana). Housing and procedures involving experimental animals were approved by the Institutional Animal Care and Research Advisory Committee of the University in Leuven.
Histologic analysis.
Lungs were paraformaldehyde-fixed and 7 m paraffin sections were stained for H&E;, Hart's elastin, PAS, thrombomodulin (gift from Dr. R. Jackman, Boston, USA), smooth muscle cell -actin (Sigma, Bornem, Belgium), SP-D and CC10 (Santa Cruz, Sanvertech, Boechout, Belgium), or PGP9.5 (UltraClone, Cambridge, UK). Apoptosis was evaluated by TUNEL staining (Roche Diagnostics, Mannheim, Germany), while proliferating cells were detected using an anti-BrdU antibody (Seralab Ltd, Sussex, UK) after i.p. injection of BrdU (Sigma, 50 mg/kg body weight; at 3 intervals of 90 minutes prior to c-section) in pregnant females. Immunostaining for Flt-1 (rat anti-Flt-1; MF1) and Flk-1 (goat anti-Flk-1; #457-683; gift from ImClone Systems Inc, New York, USA) was performed on acetone-fixed 7 m frozen sections, while immunofluorescent staining for laminin (Sigma), HIF-2 (gift from I. Flamme) and Flk-1 (rat anti-Flk-1, gift from H. Kataoka) were performed on cryofrozen lung sections. No nuclear HIF-2 staining was observed in HIF-2-/- lungs. In situ hybridization and ultrastructural analysis were performed as described42. For angiograms, ink was injected into the right fetal ventricle. Subsequently the lungs were paraformaldehyde-fixed (pH 7.4), embedded in gelatin (30%) and sectioned (100 m).
Premature mouse models.
For intra-amniotic injections, pregnant WT Swiss mice were anesthesized using isoflurane and, after laparatomy, 10 l of Evans blue (final concentration 0.5%), saline, hVEGF165 (R&D; Systems; 0.5 g/10 l saline) or hPlGF-2 (Reliatech, Braunschweig, Germany; 0.5 g/10 l saline) were injected in the amniotic cavity of E17.5 fetuses through the uterine wall, taking care not to injure the fetuses, placenta or fetal membranes. Pups were prematurely delivered by C-section at E18.5 (one day before the end of gestation), and scored for respiration and skin oxygenation using an APGAR-like score considering the skin color (cyanosis, 0; pink, 4) and lung function: no respiration (0); no spontaneous respiration but only respiration after pain stimulus (2); spontaneous but irregular respiration with RDS (4); and regular breathing without RDS (6). For intratracheal injections, WT E17.5 fetuses were delivered by C-section, surviving pups were anesthesized on ice and intratracheally injected with hVEGF165 (0.5 g/5 l) or saline (5 l). Survival of the premature pups was followed during 20 hours, while in other pups, lungs were analyzed histologically 6 hours after intratracheal injection.
Isolation and culture of type 2 pneumocytes and measurement of surfactant phospholipids.
Alveolar type 2 cells were isolated from Wistar rat lungs according to the previously described methods43. Human VEGF (200 ng/ml, R&D; Systems) was added to the culture medium. SP-B and SP-C transcripts were measured after 30 hours. Phospholipids were measured in lung homogenates (chloroform/methanol/water; 5/10/4; v/v) as described19.
Received 4 March 2002; Accepted 17 May 2002; Published online: 10 June 2002.
REFERENCES
- Goldenberg, R.L., Hauth, J.C. & Andrews, W.W. Intrauterine infection and preterm delivery. N. Engl. J. Med. 342, 1500–1507 (2000). | Article | PubMed | ISI | ChemPort |
- Bhakoo, O.N., Narang, A., Karthikeyan, G. & Kumar, P. Spectrum of respiratory distress in very low birthweight neonates. Indian J. Pediatr. 67, 803–804 (2000). | PubMed | ChemPort |
- Bourbon, J.R., Rieutort, M., Engle, M.J. & Farrell, P.M. Utilization of glycogen for phospholipid synthesis in fetal rat lung. Biochim. Biophys. Acta 712, 382–389 (1982). | Article | PubMed | ISI | ChemPort |
- Kennedy, J.D. Lung function outcome in children of premature birth. J. Paediatr. Child Health 35, 516–521 (1999). | Article | PubMed | ISI | ChemPort |
- Jakkula, M. et al. Inhibition of angiogenesis decreases alveolarization in the developing rat lung. Am. J. Physiol. Lung Cell. Mol. Physiol. 279, L600–607 (2000). | PubMed | ISI | ChemPort |
- Ferrara, N. Role of vascular endothelial growth factor in regulation of physiological angiogenesis. Am. J. Physiol. Cell. Physiol. 280, C1358–1366 (2001). | PubMed | ISI | ChemPort |
- Healy, A.M., Morgenthau, L., Zhu, X., Farber, H.W. & Cardoso, W.V. VEGF is deposited in the subepithelial matrix at the leading edge of branching airways and stimulates neovascularization in the murine embryonic lung. Dev. Dyn. 219, 341–352 (2000). | Article | PubMed | ISI | ChemPort |
- Klekamp, J.G., Jarzecka, K. & Perkett, E.A. Exposure to hyperoxia decreases the expression of vascular endothelial growth factor and its receptors in adult rat lungs. Am. J. Pathol. 154, 823–831 (1999). | PubMed | ISI | ChemPort |
- Kaner, R.J. & Crystal, R.G. Compartmentalization of vascular endothelial growth factor to the epithelial surface of the human lung. Mol. Med. 7, 240–246 (2001). | PubMed | ISI | ChemPort |
- Lassus, P. et al. Pulmonary vascular endothelial growth factor and Flt-1 in fetuses, in acute and chronic lung disease, and in persistent pulmonary hypertension of the newborn. Am. J. Respir. Crit. Care Med. 164, 1981–1987 (2001). | PubMed | ISI | ChemPort |
- Bhatt, A.J. et al. Disrupted pulmonary vasculature and decreased vascular endothelial growth factor, Flt-1, and TIE-2 in human infants dying with bronchopulmonary dysplasia. Am. J. Respir. Crit. Care Med. 164, 1971–1980 (2001). | PubMed | ISI | ChemPort |
- Lassus, P., Ristimaki, A., Ylikorkala, O., Viinikka, L. & Andersson, S. Vascular endothelial growth factor in human preterm lung. Am. J. Respir. Crit. Care Med. 159, 1429–1433 (1999). | PubMed | ISI | ChemPort |
- D'Angio, C. et al. Vascular endothelial growth factor in pulmonary lavage fluid from premature infants: Effects of age and postnatal dexamethasone. Biol. Neonate 76, 266–273 (1999). | Article | PubMed | ChemPort |
- Brown, K.R., England, K.M., Goss, K.L., Snyder, J.M. & Acarregui, M.J. VEGF induces airway epithelial cell proliferation in human fetal lung in vitro. Am. J. Physiol. Lung Cell. Mol. Physiol. 281, L1001–1010 (2001). | PubMed | ISI | ChemPort |
- Tian, H., McKnight, S.L. & Russell, D.W. Endothelial PAS domain protein 1 (EPAS1), a transcription factor selectively expressed in endothelial cells. Genes Dev. 11, 72–82 (1997). | PubMed | ISI | ChemPort |
- Ema, M. et al. A novel bHLH-PAS factor with close sequence similarity to hypoxia-inducible factor 1 regulates VEGF expression and is potentially involved in lung and vascular development. Proc. Natl. Acad. Sci. USA 94, 4273–4278 (1997). | Article | PubMed | ChemPort |
- Flamme, I. et al. HRF, a putative basic helix-loop-helix-PAS-domain transcription factor is closely related to hypoxia-inducible factor-1 and developmentally expressed in blood vessels. Mech. Dev. 63, 51–63 (1997). | Article | PubMed | ISI | ChemPort |
- Brusselmans, K. et al. Hypoxia-inducible factor-2 (HIF-2) is involved in the apoptotic response to hypoglycemia but not to hypoxia. J. Biol. Chem. 276, 39192–39196 (2001). | Article |
|
