Main

Alveolar development in late fetal and early postnatal life requires endothelial cell proliferation, angiogenesis, and alignment of capillaries in close proximity to the alveolar epithelium. Two processes for blood vessel development and differentiation, vasculogenesis and angiogenesis, have been identified (1, 2). Vasculogenesis consists of in situ differentiation of mesenchymal cells into hemangioblasts, which coalesce, develop a lumen, and form primitive vessels. Angiogenesis is the formation of new vessels by sprouting from a preexisting endothelium and differentiation of the embryonic vascular tree during organogenesis. Both processes probably occur in lung development (3). Marked pulmonary microvascular development results in a 10-fold expansion of lung capillary surface area during the first 6 wk of postnatal life in the rat (4). The correct temporal and spatial development of alveolar capillaries is critical to lung development, but little is known about factors that regulate alveolar capillary formation.

Endothelial cell proliferation is regulated in vitro by several angiogenic growth factors including VEGF. VEGF, which is also known as VEGF-A, is a member of the VEGF family (VEGF A–E). VEGF is a 42-kD homodimeric heparin-binding glycoprotein that is a specific mitogen for vascular endothelial cells (5, 6). In addition to being mitogenic, it is also chemotactic for these cells (5, 6) and increases endothelial cell permeability (7). VEGF expression is critical for developmental vasculogenesis as demonstrated by embryonic lethality of heterozygous VEGF null mutants (8, 9). In highly vascular organs such as kidneys and lungs, VEGF can be expressed by several cell types, particularly epithelial cells that are in close proximity to the microvasculature (10). In early embryonic lung, VEGF is expressed by cells in the bronchial epithelium (11). VEGF expression during late fetal mouse lung development was noted mainly in alveolar epithelial cells (12). In studies of newborn rabbit, we noted VEGF expression in a subset of alveolar type II cells (13).

In addition to normal microvascular development, VEGF may also have a role in injured and healing tissue (14). In previous work, we found that acute oxidant injury in adult and newborn lung resulted in decreased epithelial VEGF expression (13). Acute hyperoxia also results in loss of pulmonary endothelial cells (15). Because VEGF may be a survival factor for endothelial cells (16), decreased VEGF may contribute to endothelial cell loss.

VEGF binds to two partially homologous tyrosine kinase receptors, Flk-1 and Flt-1 (17, 18). In vivo, Flk-1 is expressed exclusively on endothelial cells and their precursor cells. Ligand binding to Flk-1 induces endothelial cell proliferation. In contrast, Flt-1 activation does not induce endothelial cell proliferation either in vitro or in vivo (19). VEGF may increase Flk-1 and Flt-1 abundance in vitro (20).

Fetal and postnatal lung development are strongly modulated by glucocorticoids (21, 22), which are also used clinically to treat chronic lung disease in premature infants (23). In postnatal animals, dexamethasone, a synthetic glucocorticoid, decreased formation of secondary alveolar septa (21). Several in vitro studies have shown that dexamethasone strongly down-regulates and/or inhibits induction of VEGF expression in different cell types including cells derived from alveolar epithelial cells (24, 25). Dexamethasone treatment of postnatal lung may result in precocious maturation of the alveolar capillary network (26). In midtrimester fetal human lung explants, dexamethasone increased VEGF mRNA abundance (27). Although glucocorticoids affect VEGF expression in vitro, the effects of these agents on lung VEGF expression in vivo are not known.

Transcription factor regulation of VEGF gene transcription has been investigated, particularly in hypoxia. Hypoxia is a strong inducer of VEGF, increasing both VEGF gene transcription and mRNA stability (28). Hypoxic induction of VEGF gene transcription is probably mediated by the transcription factor HIF-1 (29). A related transcription factor, HLF, can induce VEGF gene transcription in normoxic condition (30). HLF mRNA was expressed most abundantly in lung, followed by heart and liver of adult mice.

The objectives for the current study were to measure VEGF mRNA and its mitogenic receptor Flk-1 mRNA in developing mouse lung and to measure the effects of dexamethasone treatment in vivo on VEGF and VEGF receptor in newborn mouse lung. Our results show that the VEGF and Flk-1 messages increase in parallel during normal lung development and that the distal epithelium expresses VEGF mRNA at all ages examined. Dexamethasone treatment resulted in significantly increased VEGF, HLF, and Flk-1 mRNA in postnatal lung. Levels of VEGF protein were not significantly increased. These data suggest that the developing alveolar epithelium has an important role in regulating alveolar capillary development. Furthermore, unlike effects on cell culture, dexamethasone did not adversely affect VEGF expression in vivo.

METHODS

Animals and tissue preparation.

Use of animals was approved by the University Committee on Animal Resources. Swiss Webster timed pregnant mice (embryonic d 13, 15, and 18), newborn mouse pups, and adult mice were killed by injection of pentobarbital (5 mg). The lungs were removed and dissected free from the trachea and major bronchi. For fetal ages, lungs from a single litter were pooled for analysis. Pooled fetal lungs from at least four litters were analyzed separately. For postnatal animals, lungs from four to five animals were analyzed separately. From 18-d embryonic and older animals, one lung was homogenized for RNA isolation and another lung was fixed in 4% paraformaldehyde for 2 h and paraffin embedded for tissue sections.

For analysis of dexamethasone effects, litters of Swiss Webster newborn mice were culled to 10 newborns and divided into two groups on d 6 of life. Dexamethasone (0.1–5.0 mg/kg) was injected intraperitoneally daily from day of life 6 to day of life 9 (DEX group). Intraperitoneal sterile water (equivalent volume) was injected in controls (CON group). Seven litters were studied, and each litter had a CON and DEX group. Each pup was weighed on day of life 6 before administration of dexamethasone (5 mg·kg−1·d−1) or sterile water and again 24 h after the last dose. The animals were killed by injection of pentobarbital 24 h after the last dose of dexamethasone or water. The lungs were removed as described above, and the heart and kidney from each animal were also removed. Some specimens were immediately frozen in liquid nitrogen. Other lungs were weighed and stored at −80°C. Remaining specimens were fixed as above, and tissue sections were prepared.

cRNA probes.

The cDNA for VEGF was obtained by RT-PCR amplification of adult rabbit lung mRNA (31). The rabbit probe has 94% homology with mouse VEGF. This probe hybridizes to all known VEGF mRNA splice variants. The length of the VEGF probe was 590 bp. Flk-1 riboprobe was made from plasmid p4–21L1, a subclone of the mouse Flk-1 cDNA (a gift from Terry Yamaguchi). The length of the Flk-1 probe was 720 bp. A mouse cDNA for HLF-α was obtained by RT- PCR of adult mouse lung mRNA. Primers at 2269 and 3367 were used to PCR amplify cDNA. Primer sequences were ATG GGA GGC AGA TCC AAC ACGC at 2269 site and C ACC AGC CAC CAT GCT GCT TGT TAG at 3367 site. The length of HLF-α probe was 1064 bp. A cRNA for L-32, a housekeeper ribosomal protein with a length of 506 bp, was made from a mouse cDNA subcloned into pGEM 7Z (ft+) (32).

Transcription in the presence of 35S or 33P-UTP yielded probes with a specific activity of 1.8–3.3 × 109 dpm/μg for in situ hybridization. For Northern blots, the specific activity for 32P-labeled VEGF, Flk-1, HLF, and L-32 probes was ∼1.33 × 109 dpm/μg.

Northern hybridization.

Northern hybridization was conducted as described previously (31). Hybridization was carried out with 1.77 × 106 dpm/μL of VEGF probe, 1.38 × 106 dpm/μL of Flk-1 probe or 2.0 × 106 dpm/μL of HLF probe, and 1.67 × 106 dpm/μL of L-32 probe. Hybridization for VEGF or HLF and L-32 was overnight at 62°C and for Flk-1 and L-32 was overnight at 65°C. Signal was detected with either autoradiography with Kodak X-OMAT AR film exposed for 2 d without an intensifying screen or by phosphorimaging. Signal was quantified by computer image analysis (ImageQuant, Molecular Dynamics).

In situ hybridization.

In situ hybridization for VEGF mRNA was performed as described previously (13). Briefly, tissue sections were treated with 1 μg/mL proteinase K for 30 min, equilibrated with 100 mM triethanolamine, and treated with acetic anhydride. The slides were then washed and dehydrated in graded ethanols. Hybridization was performed overnight with 35S-labeled probe for VEGF (3 × 104 dpm/μL) at 53°C. Slides were then treated with RNAses A and T1, dehydrated, dipped in 1:1 dilution of NTB-2 emulsion, and counterstained.

RT-PCR amplification of VEGF splice variants.

cDNA was synthesized using murine leukemia virus RT and oligo d(T)16 primers (GeneAmp RNA PCR Kit; Perkin-Elmer Cetus, Norwalk, CT) according to the manufacturer's instructions. To determine the relative proportions of each VEGF splice variant in different samples, amplification was performed as described previously (33) using human VEGF-specific primers from the fourth and eighth exons: sense primer 5′ GAGATGAGCTTCCTACAGCAC 3′ and antisense primer 5′TCA CCGCCTCGGCTTGTCAC 3′ (including the underlined translation stop site). Amplification was performed through 35 cycles, and the PCR products were separated by electrophoresis and quantified using phosphorimaging.

Immunohistochemistry.

As described previously (13), immunohistochemistry was performed on lung tissue sections by using 1:100 dilution of rabbit anti-human VEGF in TBS (Santa Cruz Biotech, Santa Cruz, CA). Slides were treated with 0.5% H2O2 in methanol to remove endogenous peroxidase activity, blocked with 3% normal goat serum, and incubated with a 1:100 dilution of rabbit anti-human VEGF. Nonimmune rabbit IgG was the negative control. Slides were rinsed and incubated with biotinylated goat anti-rabbit IgG. Vectastain ABC elite reagent (Vector Laboratories, Burlingame, CA) was added for 30 min. Slides were then incubated with diaminobenzidine tetrahydrochloride solution and counterstained with hematoxylin/eosin.

ELISA.

Frozen lung specimens were homogenized in 1% PBS containing 50 μg/mL heparin. The homogenates were centrifuged in a microcentrifuge for 10 min, and supernatant fractions were collected. ELISA for VEGF protein was performed using Quantikine M-Mouse VEGF Immunoassay kit (R&D Systems, Minneapolis, MN). Fifty microliters of supernatant was diluted with 200 μL of calibrator diluent RD5T available in kit. The standard protocol was followed as detailed by the manufacturer. Assay for total protein in the supernatant was performed using BCA protein assay reagent (Pierce, Rockford, IL).

RESULTS

Northern hybridization of mouse lung total RNA using highly homologous cRNA rabbit probe for VEGF and mouse probes for Flk-1, HLF, and L-32 demonstrated a major VEGF mRNA species of 3.5 kb, Flk-1 species of approximately 5.5 kb, HLF species of approximately 4.5 kb, and L-32 of approximately 0.5 kb , consistent with sizes reported for these messages (not shown).

Abundance of VEGF and Flk-1 mRNA increases in parallel indeveloping mouse lung.

Using phosphorimaging and image analysis, we compared VEGF mRNA abundance and Flk-1 mRNA abundance (relative to L-32) in developing lung. The ratio of VEGF or Flk-1 to L-32 at fetal age 13 was assigned a value of 1 (Fig. 1). Relative to L-32, the abundance of VEGF mRNA increased 3-fold between d 13 and 18 of fetal life (term = 20 d). The VEGF mRNA abundance increased an additional 3-fold during the first 2 postnatal weeks, when alveolarization occurs and expansion of the microvasculature is active. Relative abundance of VEGF message also increased from 2 wk of age to the adult, giving a 12-fold increase between 13-d fetal and adult animals. The relative abundance of Flk-1 mRNA increased in parallel to the increase in VEGF mRNA (Fig. 1). The message also remained relatively abundant in the adult lung. These data demonstrate that VEGF mRNA and Flk-1 mRNA abundance increase substantially and in parallel during late fetal and early postnatal life. Furthermore, the adult lung, which has low endothelial proliferation, has relatively abundant VEGF and Flk-1 mRNA.

Figure 1
figure 1

VEGF and Flk-1 mRNA abundance in developing mouse lung. Data points are the ratio of VEGF/L-32 (black bar) and Flk-1/L-32 (gray bar) for each time point as measured by Northern blot analysis and quantified by phosphorimaging and image analysis. The data are mean ± SEM (n = 4–5 samples for each point). The ratio of VEGF or Flk-1 to L-32 at fetal age 13 d was assigned a value of 1. Relative to L-32, the abundance of both VEGF and Flk-1 mRNA increased 3-fold between d 13 and 18 of fetal life (term = 20 d). VEGF mRNA abundance increased an additional 3-fold during the first 2 postnatal weeks. Relative abundance of VEGF message also increased from 2 wk of age to the adult. The Flk-1 message paralleled the VEGF message. Both messages remained relatively abundant in the adult lung.

Full size image

Effects of dexamethasone on VEGF, Flk-1, and HLF mRNA abundance inpostnatal mouse lung.

Litters of newborn mice divided into DEX and CON groups at d-6 postnatal age were injected with dexamethasone (0.1–5.0 mg·kg−1·d−1) or sterile water daily until d 9 and were killed 24 h after the last injection. The mean of ratio of VEGF/L-32 or Flk-1/L-32 mRNA abundance in the CON group was assigned a value of 1. The relative abundance of VEGF mRNA increased with increasing dexamethasone dose, reaching a statistically significant 2-fold increase (n = 7 litters, p < 0.005) at 5.0 mg·kg−1·d−1 (Fig. 2, A and B ). The relative abundance of Flk-1 mRNA in lung also increased 2-fold in the DEX group (n = 6 litters, p < 0.04, Fig. 2B). Dexamethasone treatment did not affect heart or kidney VEGF mRNA abundance (not shown).

Figure 2
figure 2

Effects of dexamethasone on VEGF, Flk-1, and HLF mRNA abundance in postnatal mouse lung. Neonatal mice were injected with dexamethasone (DEX) or sterile water (CON) as detailed in “Methods.” Northern hybridizations were quantified by phosphorimaging and the mean of ratio of VEGF, Flk-1, or HLF/L-32 mRNA abundance in the CON group was assigned a value of 1. Data are mean ± SEM. (A) A representative Northern blot showing VEGF mRNA and L-32 mRNA. (B) The relative abundance of VEGF mRNA increased in a dose-dependent manner, reaching a statistically significant 2-fold increase at 5.0 mg·kg−1·d−1 of DEX compared with CON animals (+p < 0.005 compared with CON). Each point represents data from three to seven litters of animals. The relative abundance of Flk-1 mRNA in lung also increased 2-fold in the DEX (5 mg·kg−1·d−1) group (n = 6 litters, ★p < 0.04). (C) A representative Northern blot showing HLF mRNA and L-32 mRNA. (D) The relative abundance of HLF mRNA increased significantly 3-fold in the DEX (5 mg·kg−1·d−1) group compared with CON animals (n = 7 litters, p < 0.02).

Full size image

Because dexamethasone increased lung VEGF mRNA abundance, we also analyzed dexamethasone effects on abundance of HLF-α mRNA. HLF-α is the unique monomer of HLF, a transcription factor related to HIF-1 that increases VEGF transcription in normoxia and hypoxia. Dexamethasone (5 mg·kg−1·d−1) treatment resulted in a 3-fold increase in HLF-α mRNA abundance in lung (n = 7 litters, p < 0.02, Fig. 2, C and D ).

Effects of dexamethasone on VEGF protein in postnatal mouselung.

To evaluate the dexamethasone effect on VEGF protein, we performed ELISA for VEGF on lungs from the DEX (5 mg·kg−1·d−1) and CON group (Fig. 3). We found 115.7 ± 9.5 pg/mg VEGF protein/total lung protein (mean ± SEM) of VEGF protein in the DEX and 102.1 ± 9.4 pg/mg (mean ± SEM) VEGF protein in the CON (n = 7 animals, p = 0.4). Unlike studies in vitro, these results suggest that dexamethasone did not decrease VEGF mRNA or protein in postnatal lung in vivo.

Figure 3
figure 3

Effects of dexamethasone on VEGF protein in postnatal mouse lung. The protocol for the two treatment groups, DEX and sterile water (CON), are given in “Methods.” VEGF protein was quantified by ELISA on lungs from the DEX (5 mg· kg−1·d−1) and CON group. Data are mean ± SEM. There was no significant difference between the two groups (n = 7 animals, p = 0.4).

Full size image

VEGF mRNA is located mainly in epithelial cells in alveolar septa,and this pattern was not altered by dexamethasone treatment.

We identified VEGF-expressing cells by in situ hybridization on mouse lung tissue from d 18 fetal, d 2, 8, and 14 postnatal, and adult mice. We also performed in situ hybridization on lung, heart, and kidney from the DEX and CON group. In the fetal d-18 lung, VEGF mRNA is localized mainly in the cuboidal epithelial cells of distal airspaces (Fig. 4A). Little or no message is seen in epithelial cells of large conducting airways, large vascular endothelial cells, or smooth muscle cells. A relatively low but significant level of VEGF message was noted in the large amount of mesenchymal tissue between potential airspaces.

Figure 4
figure 4

VEGF in situ hybridization on mouse lung. (A) In the fetal d-18 lung, VEGF mRNA (black grains) is localized mainly in the cuboidal epithelial cells of distal airspaces. A relatively low but discernable level of VEGF message was noted in mesenchymal tissue between airspaces. (B) Second postnatal day lung and (C) eighth postnatal day lung. VEGF-expressing cells (white grains) were mainly in the alveolar epithelium, and little message was detected in mesenchymal cells, vascular smooth muscle cells, or airway cells. (D) On higher magnification, in eighth postnatal day lung, the VEGF-expressing cells (black grains) were often rounded alveolar epithelial cells located at alveolar corners, suggestive of type II alveolar epithelial cells. No appreciable message was seen in flattened alveolar epithelial cells. The images in panel A and D are light field, and the other images are dark field.

Full size image

Similar to embryonic mouse lung, VEGF-expressing cells in postnatal lung (Fig. 4, B and C ) were mainly in the alveolar epithelium, less message was detected in mesenchymal cells, and little message in vascular smooth muscle cells or airway cells. On higher magnification (Fig. 4D), the VEGF-expressing cells were often rounded alveolar epithelial cells located in alveolar corners, suggestive of type II alveolar epithelial cells. No appreciable message was seen in flattened alveolar epithelial cells. In situ hybridization with sense probe gave minimal signal. To investigate the distribution of L-32 mRNA, we performed in situ hybridization using 35S riboprobe. L-32 message was detected uniformly in all embryonic and postnatal lung cells (not shown).

Similar to control lung, VEGF-expressing cells in lung from DEX animals were primarily in the distal alveolar epithelium. Dexamethasone treatment did not alter the pattern of cells that express VEGF (Fig. 5, A and B ). Dexamethasone treatment also did not alter the pattern of VEGF expression in heart or kidney tissue (not shown).

Figure 5
figure 5

VEGF in situ hybridization on postnatal mouse lungs treated with dexamethasone. The protocol for the two treatment groups, DEX and sterile water (CON), are given in “Methods.” Similar to control lung (A), VEGF-expressing cells (white grains) in lung from DEX (5 mg·kg−1·d−1) animals (B) were primarily in the distal alveolar epithelium. Dexamethasone treatment did not alter the pattern of cells that express VEGF.

Full size image

Dexamethasone did not alter the proportions of VEGF mRNA splicevariants in postnatal lung.

VEGF protein has at least four isoforms that can be produced by alternative splicing of the pre-mRNA. The isoforms have differing mitogenicity, extracellular matrix (ECM) binding, and receptor affinity. To evaluate potential dexamethasone effects on VEGF splice variant expression, we performed RT-PCR on RNA isolated from the CON and DEX (5 mg·kg−1·d−1) animals. The CON group showed 44% VEGF189, 32% VEGF165, and 24% VEGF121. These proportions were similar to those in the DEX group (Fig. 6, A and B ).

Figure 6
figure 6

Effects of dexamethasone on VEGF mRNA splice variant expression in postnatal lung. (A) RT-PCR on RNA isolated from the lungs of CON (C) and DEX (D) animals. Each lane represents a separate CON or 5 mg·kg−1·d−1 DEX animal. (B) Quantification of RT-PCR. There were no major differences in the proportions of VEGF splice variant between the two groups.

Full size image

Immunostaining for VEGF peptide.

VEGF protein was immunostained using a polyclonal rabbit anti-human VEGF antibody. Lung tissue sections from d-18 embryo (Fig. 7B) had VEGF protein mainly surrounding distal airspaces with some staining of intervening mesenchymal tissue and conducting airway epithelial cells. Similarly, at 2 and 8 postnatal days, VEGF staining was prominent in alveolar septa (Fig. 7, C and D ). However, these stages also had staining in vascular smooth muscle cells but not airway smooth muscle cells or endothelial cells (Fig. 7F, arrowhead). VEGF protein was also localized in some bronchial epithelial cells. In adult lung, prominent staining was observed in rounded alveolar epithelial cells (Fig. 7E, arrows). Minimal background was observed when tissue sections were incubated with nonimmune rabbit IgG (Fig. 7A).

Figure 7
figure 7

Immunostaining for VEGF peptide in developing mouse lung. VEGF protein was immunostained using a polyclonal rabbit anti-human VEGF antibody. (A) Minimal background was observed when tissue sections were incubated with nonimmune rabbit IgG. (B) D-18 embryo. VEGF protein mainly surrounding distal airspaces with some staining of intervening mesenchymal tissue and conducting airway epithelial cells. (C) Postnatal d 2 and (D) postnatal d 8. VEGF staining was prominent in alveolar septa. However, these stages also had staining in vascular smooth muscle cells but not airway smooth muscle cells. VEGF protein was also localized in some bronchial epithelial cells. (E) Adult. Prominent staining observed in rounded alveolar epithelial cells (arrows). (F) Vascular smooth muscles but not endothelial cells (arrowhead) stain for VEGF peptides.

Full size image

Postnatal dexamethasone inhibits body growth and disproportionatelyreduces lung weight.

Because glucocorticoids affect growth, we evaluated the body and lung weight in the DEX and CON groups as a measure of dexamethasone effect in our experiments. Mean weight gain between d 6 and the day the animals were killed in the CON group was 1.26 ± 0.11 g. Mean weight gain in the DEX group was 0.09 ± 0.13 g, which is significantly lower than the CON group (n = 24 animals, p ≤ 0.002).

The lung weight/body weight for each pup was also calculated. The ratio of mean lung weight/mean body weight in the CON group was 23.8 ± 1.1. This ratio in the DEX group was 19.8 ± 1.3, which is significantly lower than the CON group (n = 24 animals, p ≤ 0.005). Although there was no major effect on VEGF expression, compared with L-32 or total lung protein, dexamethasone-treated animals had significantly impaired body and lung growth, indicating a dexamethasone effect in the treated animals.

DISCUSSION

In lung, microvascular development in late fetal and early postnatal life is characterized by endothelial cell proliferation, capillary alignment in close proximity to the alveolar epithelium, and capillary remodeling (3, 4). The endothelial cell-specific mitogen VEGF is a major mediator of angiogenesis. Flk-1 is a tyrosine kinase receptor for VEGF that is expressed mainly on endothelial cells. Transcription of the VEGF gene can be induced by the transcription factor HLF. Glucocorticoids, which modulate lung development and alveolarization, strongly inhibit VEGF expression by several cell types in vitro but increase VEGF mRNA in human lung explants (27).

A major finding of this study is that VEGF and Flk-1 mRNA abundance increases in parallel during late fetal and postnatal mouse lung development. Adult lung retains relatively high levels of VEGF mRNA, although little endothelial cell proliferation occurs at this time. The major cells expressing VEGF mRNA at all ages examined are in the alveolar epithelium, although mesenchymal cells in embryonic lung had detectable message. We detected a small but statistically significant increase in VEGF mRNA in the DEX animals, but dexamethasone did not significantly alter VEGF protein in vivo. Unlike studies in vitro, our experiment found no inhibitory effect of dexamethasone, even in relatively large doses, on lung VEGF expression.

During the embryonic stage of human lung development (0–8 wk), scattered primitive blood vessels are formed from mesenchymal cells by vasculogenesis. During the pseudoglandular stage (9–16 wk), more primitive blood vessels are seen and, by 15 wk, their margins are demarcated by a thin layer of flattened cells. During the canalicular stage (17–26 wk), angiogenesis contributes to new capillary formation, and capillaries align with the overlying cuboidal alveolar epithelium. The saccular (27–37 wk) and alveolar stages (postnatal) are characterized by marked expansion of capillary surface area, active angiogenesis, remodeling of alveolar capillaries, and reduction in mesenchymal tissue. Remodeling of alveolar capillaries involves fusion and preferential growth of fused capillaries (4).

VEGF is a heparin-binding growth factor with major biologic functions that include stimulation of endothelial cell proliferation, angiogenesis, and endothelial cell permeability. VEGF may also be necessary for endothelial cell survival, particularly in hyperoxic conditions (16). VEGF expression is spatially and temporally related to the proliferation of blood vessels during early embryonic development and in the developing mouse brain (11).

In the current study, we found VEGF mRNA abundance increased in parallel to Flk-1 mRNA during late fetal and postnatal mouse lung development when active angiogenesis is occurring. Millauer et al. (34) demonstrated that spatial and temporal expression of Flk-1 coordinated with that of VEGF during early embryonic and early postnatal brain development. Our data support the hypothesis that the VEGF/Flk-1 transduction system plays a major role in regulating lung capillary development. Northern hybridization of whole lung may underestimate VEGF mRNA abundance in the alveolar epithelium, especially in fetal lung that has abundant mesenchyme. Part of the increase in the relative abundance of VEGF that we found on Northern hybridization may be due to changes in cell population, particularly a decrease in mesenchymal cells and a relative increase in alveolar epithelial cells during development.

The adult lung has increased abundance of VEGF mRNA compared with other organs. In this study, we noted higher VEGF mRNA levels in adult lung than in developing lung. Because little endothelial cell proliferation occurs in adult lung, high abundance of VEGF mRNA suggests a possible role in maintenance of the normal microvasculature. As a survival factor, VEGF may function to maintain the large pulmonary microvasculature. Alternatively, VEGF may regulate the baseline microvascular permeability in the lung.

In situ hybridization shows that VEGF mRNA is located mainly in distal airspace epithelial cells in the late fetal and postnatal lung. Similar findings were also noted in the human fetus (35), embryonic mouse (12), and postnatal rabbit lung (13). In postnatal rabbit lung, we found that a subpopulation of alveolar type II cells had VEGF mRNA. Together, these data suggest that VEGF acts as a paracrine angiogenic factor and that the alveolar epithelium may regulate alveolar capillary development. Our finding that discrete cells in the fetal alveolar epithelium express VEGF suggests strict cell-specific spatial regulation of VEGF expression. It is likely that VEGF secreted by fetal pulmonary epithelial cells acts as a specific mitogen and chemotactic factor causing the endothelial cells, which bear the Flk-1 receptor, to proliferate and migrate toward the alveolar epithelium. This may be the mechanism leading to capillary alignment along the alveolar epithelial lining.

VEGF immunostaining demonstrated the localization of the protein in distal airspace epithelium, in some conducting airway epithelial cells, in intervening mesenchymal tissue, and in vascular smooth muscle cells. The difference in the cell-specific localization of VEGF mRNA noted on in situ hybridization and VEGF peptide may result from differing sensitivity of these methods. In situ hybridization may lack sufficient sensitivity to detect relatively low abundance of VEGF mRNA in some lung cells. Alternatively, different lung cells may produce different protein isoforms. Cells that produce the cell-binding isoforms may accumulate the protein but have relatively low message levels. Four species of VEGF, which are generated by alternate splicing of the pre-mRNA, have been identified (36). The larger isoforms bind heparan sulfate and may accumulate on cell surface or in basement membrane. Consistent with this hypothesis, we found that VEGF189 mRNA, which codes for a cell-binding isoform, was the most abundant splice variant.

Dexamethasone treatment resulted in a small but statistically significant increase in VEGF and Flk-1 mRNA abundance in postnatal lung. There was no significant effect of dexamethasone on VEGF protein, cell-specific distribution of the mRNA, or splice variant expression. We did not perform morphometry on the DEX or CON lung, so we do not know if capillary density, the number of endothelial cells, or alveolarization were altered. Using a similar dose and timing of dexamethasone treatment, Ohtsu et al. (37) found no effect on the volume density of lung capillaries. Our findings on the effects of dexamethasone on postnatal lung are particularly interesting because several in vitro studies have shown that dexamethasone down-regulates the baseline VEGF mRNA level or inhibits induction of VEGF by different factors in several cell types. A recent study of fetal lung explants, however, found that dexamethasone treatment resulted in a small increase in VEGF mRNA levels (40). It is possible that a mechanism of the dexamethasone effect on the postnatal lung VEGF mRNA level may be by increased HLF expression. Transient DNA transfection experiments showed that HLF activated VEGF gene transcription under normoxic conditions (38). In our experiment, dexamethasone increased HLF mRNA levels, which in turn could induce VEGF gene transcription. Both HLF mRNA and VEGF mRNA have been located in alveolar epithelial cells in the postnatal mouse. Despite the increase in VEGF mRNA, we found no significant change in VEGF protein in the DEX group. There are no data suggesting dexamethasone affects VEGF translational efficiency. The lack of significant increase in VEGF protein in our studies is probably due to diminished sensitivity of the ELISA to detect small changes in whole lung VEGF protein concentration.

Certain limitations are important in interpreting our data. In situ hybridization may more likely detect mRNA in a compact cell such as cuboidal epithelial cells rather than spread flattened cells such as type I cells. Thus, our data do not exclude the possibility that other lung cells express VEGF, which may account for finding VEGF peptides in cells with minimal mRNA. In addition, our data pertain only to VEGF (also known as VEGF-A) and not to other members of the VEGF family.

In summary, the VEGF/Flk-1 transduction system is expressed during late fetal and postnatal mouse lung development. Abundance of VEGF mRNA increases in parallel to Flk-1 mRNA abundance, suggesting a role of VEGF/Flk-1 transduction system in lung microvascular development. The relatively high expression of VEGF by alveolar epithelial cells suggests that these cells regulate alveolar capillary development and location in developing lung. Besides a strong growth inhibitory effect, dexamethasone increased VEGF, Flk-1, and HLF mRNA. However, there was no major effect on VEGF protein, splice variants, or cell-specific expression of VEGF in postnatal mouse lung. These data suggest that although dexamethasone inhibits VEGF expression in vitro, it does not have a similar effect on VEGF expression in postnatal lung in vivo.