INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The walls of pulmonary capillaries must be extremely thin for gas exchange. For example, in the human lung, half of the area of the blood-gas barrier has a thickness of only 0.2 to 0.4 µm (1). An inevitable consequence of this extraordinary thinness is that the wall stresses become extremely high when capillary transmural pressure rises. For example, during maximal exercise, direct measurements of mean pulmonary artery and pulmonary artery wedge pressure imply that the capillary pressure in some parts of the lung exceeds 30 mm Hg, and the consequent wall stresses approach the breaking stress of collagen (2).

In experimental animal preparations, a capillary transmural pressure of this magnitude causes ultrastructural changes in the blood-gas barrier, including disruptions of the alveolar epithelium and capillary endothelium, and sometimes of all layers of the capillary wall, leading to alveolar edema and hemorrhage (3). There is evidence that this "stress failure" is the mechanism of high-altitude pulmonary edema (4), exercise-induced hemorrhage in racehorses (5), and the increased capillary permeability seen at abnormally high states of lung inflation (6). Even 5 to 10 min of maximal exercise in elite human athletes apparently causes changes in the integrity of the blood-gas barrier, because bronchoalveolar lavage (BAL) has shown increased concentrations of red blood cells, total protein, and leukotriene B₄ (7). However these changes were not seen following 1 h of submaximal exercise in a similar group of athletes (8).

These findings highlight the dilemma of the blood-gas barrier, which needs to be extremely thin for efficient gas exchange but at the same time must be strong enough to withstand the very high wall stresses that develop when capillary pressure rises. Indeed, it appears that the blood-gas barrier is just strong enough to withstand the maximal stresses to which it is exposed, such as during severe exercise.

How this delicate balance is maintained is a central issue for the lung. Our hypothesis is that the structure of the blood- gas barrier is continuously regulated in such a way as to keep it extremely thin but just strong enough to withstand the maximal mechanical stresses to which it is subjected. Presumably, some element in the capillary wall senses wall stress and regulates wall structure in response to this. There is evidence that most of the strength of the blood-gas barrier, at least on the thin side of the capillary, is attributable to the type IV collagen in the basement membranes, and it is therefore of particular interest to know whether extracellular matrix (ECM) proteins are being regulated in response to wall stress. Evidence that this occurs is that patients with mitral stenosis or pulmonary venoocclusive disease, both of which increase capillary pressure, develop marked thickening of the basement membranes in the capillary wall (9, 10).

To test this hypothesis, we have designed experiments in which capillary wall stress is increased and in which we can then determine whether changes in ECM proteins have occurred. Capillary wall stress can be increased in two obvious ways: (1) by increasing capillary transmural pressure; and (2) by greatly increasing lung volume. The latter method is effective because part of the increased tension in the alveolar wall at high states of lung inflation is transmitted to the capillary walls. A difficulty with increasing capillary transmural pressure is that it tends to cause severe edema. Therefore, our first experimental design was to greatly increase lung volume in one lung of anesthetized, open-chest rabbits while ventilating the other lung at a normal volume (11). Additional control animals had both lungs at normal states of lung inflation. These experiments showed that high states of lung inflation over a period of 4 h resulted in increased gene expression for 1(III) and 2(IV) procollagen, fibronectin, basic fibroblast growth factor (bFGF), and transforming growth factor-1 (TGF-₁). By contrast, mRNA levels for 1(I) procollagen and vascular endothelial growth factor (VEGF) were unchanged (11).

In a second set of experiments, capillary transmural pressure was increased, but this was done intermittently to increase wall stress while avoiding pulmonary edema. Isolated perfused rat lungs had their pulmonary venous pressure increased cyclically to 28 cm H₂O for 15 s every minute for 4 h (12). Controls were similar lungs perfused at low pressure, and also unperfused lungs. This study showed significant increases in 1(I) and 1(III) procollagen, fibronectin, laminin, and TGF-₁ (12). The differences in results of the two experiments might be partly explained by the fact that neither increasing lung inflation or increasing capillary transmural pressure specifically changes capillary wall stress alone.

A third potential way of increasing capillary wall stress is to induce alveolar hypoxia. This is known to increase pulmonary artery pressure in rats within minutes, as a result of pulmonary vasoconstriction (13). This will only increase the transmural pressure of some capillaries if the vasoconstriction is uneven (14), but there is evidence to suggest that such unevenness is the case. For example, Dawson and colleagues (15) showed that alveolar hypoxia nearly doubled the dispersion of transit times through the pulmonary circulation of a lobe of dog lung. In addition, it has been shown that the distribution of India ink particles injected into the pulmonary circulation during alveolar hypoxia is more uneven than during normoxia (16). Furthermore, uneven pulmonary vasoconstriction during hypoxia at high altitude, and consequent stress failure of pulmonary capillaries, is a likely mechanism of high-altitude pulmonary edema (17). Madison strain Sprague-Dawley rats exposed to very low barometric pressures develop stress failure of pulmonary capillaries (4).

The only study to date of changes in ECM gene expression in lung following alveolar hypoxia was done by Vyas-Somani and associates (18) in rats. They found a biphasic response in 1(IV) procollagen mRNA, with decreases at Days 4 and 7 followed by an increase at Day 12. Immunohistochemical studies showed some changes in fibronectin and laminin in large and small pulmonary arteries, but not in lung parenchyma.

In the present study, Madison strain Sprague-Dawley rats were exposed to normobaric hypoxia (10% oxygen) for 6 h, 3 d, and 10 d, and evidence for gene expression of ECM proteins and growth factors was sought in peripheral lung tissue, which excluded the presence of large blood vessels and airways. A striking finding was a large increase in mRNA for 2(IV) procollagen after 6 h and 3 d, which then decreased after 10 d. There were also changes in other ECM proteins and in growth factors. The results suggest that alveolar hypoxia causes rapid remodeling in lung parenchyma.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Experimental Protocol

Fifty-three Madison strain Sprague-Dawley male rats were obtained from Harlan Laboratories (Indianapolis, IN). This strain of rats was selected because it exhibits an increased sensitivity to high-altitude pulmonary edema (19). Rats were divided into two main groups (a long-term group of 27 rats, which was studied first, and a short-term group of 26 rats, which was used later), and each of these groups were further divided into subgroups of five rats each, except for Day 3 (short-term group) and Day 10 (long-term) hypoxia subgroups, which began with six and seven rats, respectively. Long-term rats (body weight: 280 to 331 g) were exposed to air or hypoxia for 3 d or 10 d, whereas short-term rats (body weight: 328 to 399 g) were exposed to air or hypoxia for 6 h or 3 d. Prior to experimentation, all rats were housed in room air (two or three rats/cage) with free access to food and water. Rats also had free access to food and water during exposures.

At the start of an experiment, all rats in a group were weighed and the hypoxia subgroups were transferred to a separate room for exposure to normobaric hypoxia (15% O₂ during Day 1, followed by 10% O₂ for the remainder of the exposure period, at sea level barometric pressure). This procedure was adopted to provide a period for adjustment to hypoxia and to lessen the possibility of stress failure and edema formation in rats. During each exposure period, the other subgroups (air-breathing control rats) remained in a separate room to insure that they breathed uncontaminated (21% O₂) room air. This was necessary because the exposure chamber mixed pure nitrogen with room air at a self-adjusting rate to maintain O₂ at the prescribed concentration while venting excess nitrogen into the room. The amounts of oxygen and carbon dioxide in the chamber were monitored continuously throughout each experiment with a chart recorder and mass spectrometer calibrated with test gases.

RNA Isolation and Northern Analysis

Following completion of the prescribed exposure period, animals were euthanized with sodium pentobarbital (50 mg/kg body weight, given intraperitoneally) and the lungs were rapidly removed and rinsed in cold phosphate-buffered saline (PBS). Peripheral lung tissue (1- to 2-mm wide strips from the edges of each lung) was then collected, weighed, and frozen in liquid nitrogen for subsequent isolation of RNA according to the method of Chomczynski and Sacchi (20). RNA preparations were quantitated through their absorbance at 260 nm, and intactness of the RNA was assessed by ethidium bromide staining following separation by electrophoresis in a 6.6% formaldehyde-1% agarose gel.

Fractionated RNA was then transferred by Northern blotting to a Zeta probe membrane (Bio-Rad, Hercules, CA) and crosslinked through UV exposure. Following transfer, the blots were probed according to the random primer method with complementary DNA (cDNA) probes labeled with (-³²P)deoxycytosine triphosphate ([-³²P]dCTP), with a specific activity of at least 1 × 10⁹ disintegrations · min¹ · µg¹ (Prime-It II kit; Stratagene, La Jolla, CA). The membrane was then prehybridized and hybridized in 50% formamide containing 5× standard saline-citrate (SSC) (3 M sodium chloride, 0.3 M sodium citrate), 10× Denhardt's solution (2% Ficoll, 2% polyvinyl pyrrolidone), 50 mM NaH₂PO₄ (pH 6.5), and 100 to 250 µg/ml salmon sperm DNA at 37° C or 42° C. Blots were washed with 2× SSC and 0.1% sodium dodecyl sulfate (SDS) at room temperature, and with 0.1 to 0.5× SSC and 0.1% SDS at 50° to 65° C, after which they were exposed to X-ray film with a Cronex Lightening Plus intensifier screen at 70° C. Autoradiographs were quantitated densitometrically within the linear range of signals, and the results were normalized to ribosomal 18S RNA levels.

Sources of Recombinant Plasmids

The rat fibronectin probe was a 0.875-kB EcoRI insert from the plasmid pRCabFN1 (21). The rat 1(I) procollagen probe was a 1.6-kB cDNA Pst I insert from the plasmid p1R1 (22). The mouse 1(III) procollagen probe was a 1.6-kB PstI insert from the plasmid pMSC.1 (23). The mouse 2(IV) procollagen probe was a 1.1-kB EcoRI/Sal I insert from the plasmid pE18 (24). The human PDGF- probe was a 2.1-kB BamHI insert from the plasmid PSM-1 purchased from American Type Culture Collection (ATCC). The rat VEGF probe was a 0.9-kB PstI/SmaI insert (25).

Determination of Collagen Content in Peripheral Lung Tissue

The level of 4-hydroxyproline was measured according to the method of Jamall and colleagues (26). Briefly, lung tissue was homogenized in deionized water (1:10, wt:vol) and hydrolized in 6 N HCl (20 h at 110° C). Aliquots of 25 µl of lung hydrolysate were then added to duplicate glass tubes and were evaporated to dryness with a Speed Vac (Savant Instruments, Farmingdale, NY). Duplicate tubes then received hydroxyproline standards dissolved in 50% isopropanol (0 to 1.6 µg hydroxyproline/1.2 ml final volume) and 0.2 ml of 0.56% buffered chloramine-T solution. After 10 min, 1.0 ml of Ehrlich's reagent was added to give a total volume of 2.4 ml. All samples were vortexed and incubated for 90 min at 50° C. Absorbance was then measured at 558 nm against a water blank, and the absorbance values were corrected for the reagent blank. In addition, total protein was measured with the Bio-Rad protein assay (Bio-Rad). Total DNA content was assayed fluorimetrically, using Hoechst 33258 reagent (27).

Statistical Analysis

Statistical comparisons between groups were made by one-way analysis of variance (ANOVA) and the Student-Newman-Keuls test. Values are expressed as mean ± SEM, with a value of p < 0.05 considered significant.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Effects of Hypoxia on Survival and Body Weight

Of the total of 53 rats used, all survived exposure to air or hypoxia for the prescribed period except for two rats subjected to hypoxia: one of these animals, in the short-term group, died during Day 2, and one rat in the long-term group died during Day 7. Also, one rat in the Day-10 air-control subgroup died during anesthesia, and no tissue was collected from this rat. This gave a total of 25 rats for the short-term group and 25 rats for the long-term group. As expected, the body weight of air-breathing control rats increased over time, by an average of 2% over a 3-d period and by 13% over a period of 10 d. In contrast, the body weight of the hypoxic rats decreased by an average of 14% and 8% following 3 d and 10 d exposure, respectively, to hypoxia. Body weight was not measured in rats after 6 h exposure to hypoxia.

Gene Expression of ECM Components in Lung Parenchyma

In both the short-term and long-term groups, Northern blots (Figures 1A and 1B) and densitometric analyses (Figures 2-5) revealed increases in levels of mRNA for 1(I), 1(III), and 2(IV) procollagens and fibronectin in all rats in both groups following exposure to hypoxia for 3 d. The magnitude of the response was identical in the short-term and long-term groups for 2(IV) procollagen (a sevenfold increase) and fibronectin (an 11-fold increase) as compared with mRNA levels in lung parenchyma of age-matched air-breathing rats. The response for 1(I) and 1(III) procollagens was more variable, but there was a significant increase in both groups (short-term: 12- and eightfold increases, respectively; long-term: seven- and 1.6-fold increases, respectively). Unlike the other ECM component molecules, levels of mRNA for 2(IV) procollagen increased sixfold after only 6 h exposure to hypoxia (Figure 4A). After 10 d of hypoxia, levels of mRNA for all ECM component molecules were significantly lower than their Day-3 values (except 2(IV) procollagen), and not significantly different from their Day-0 levels (Figures 2-5).

View larger version (76K): [in this window] [in a new window]   View larger version (56K): [in this window] [in a new window]   Figure 1.   Northern blot analysis of total cellular RNA isolated from peripheral lung tissue of rats following exposure to air or hypoxia (H; 10% oxygen at sea level barometric pressure) for up to 3 d (A) or for 3 d and 10 d (B). Each lane represents the response of one rat following hybridization of samples to 32P-oligolabeled cDNAs specific for the ECM components 1(I), 1(III), and 2(IV) procollagens and fibronectin (FN); and for PDGF-B and VEGF 18S (ribosomal) RNA.

View larger version (10K): [in this window] [in a new window]   Figure 2.   Densitometric analysis of Northern blots hybridized to 32P-oligolabeled cDNA specific for 1(I) procollagen mRNA in peripheral lung tissue of rats exposed to air (open bar) or hypoxia (solid bar) for up to 3 d (A) or for 3 d and 10 d (B). mRNA levels are expressed in arbitrary densitometric units normalized for loading. Values represent means ± SE (n = 5 in each subgroup, except for Day 10 air [n = 4] and hypoxia [n = 6]). *Significantly greater (p < 0.05) than in each of the other four subgroups.

View larger version (10K): [in this window] [in a new window]   Figure 3.   Densitometric analysis of Northern blots hybridized to 32P-oligolabeled cDNA specific for 1(III) procollagen mRNA in peripheral lung tissue of rats exposed to air (open bar) or hypoxia (solid bar) for up to 3 d (A) or for 3 d and 10 d (B). mRNA levels are expressed in arbitrary densitometric units normalized for loading. Values represent means ± SE (n = 5 in each subgroup, except for Day 10 air [n = 4] and hypoxia [n = 6]). *Significantly greater (p < 0.05) than in each of the other four subgroups.

View larger version (11K): [in this window] [in a new window]   Figure 4.   Densitometric analysis of Northern blots hybridized to 32P-oligolabeled cDNA specific for 2(IV) procollagen mRNA in peripheral lung tissue of rats exposed to air (open bar) or hypoxia (solid bar) for up to 3 d (A) or for 3 d and 10 d (B). mRNA levels are expressed in arbitrary densitometric units normalized for loading. Values represent means ± SE (n = 5 in each subgroup, except for Day 10 air [n = 4] and hypoxia [n = 6]). aSignificantly greater (p < 0.05) than at time zero; bsignificantly greater than in each of the other four subgroups, except for the 6 h hypoxia subgroup; csignificantly greater than in each of the other four subgroups, except for the Day 10 hypoxia subgroup.

View larger version (10K): [in this window] [in a new window]   Figure 5.   Densitometric analysis of Northern blots hybridized to 32P-oligolabeled cDNA specific for fibronectin mNRA in peripheral lung tissue of rats exposed to air (open bar) or hypoxia (solid bar) for up to 3 d (A) or for 3 d and 10 d (B). mRNA levels are expressed in arbitrary densitometric units normalized for loading. Values represent means ± SE (n = 5 in each subgroup, except for Day 10 air [n = 4] and hypoxia [n = 6]). *Significantly greater (p < 0.05) than in each of the other four subgroups.

Gene Expression of Growth Factors

We also determined mRNA levels for platelet-derived growth factor- (PDGF-B) and VEGF in lung parenchyma, since levels of these growth factors are known to change during hypoxia. The level of mRNA for PDGF-B increased twofold following 6 h exposure to hypoxia, and returned to control values after 3 d (Figures 1A and 6). In contrast, levels of VEGF mRNA did not change significantly in peripheral lung tissue of rats during exposure to hypoxia (Figures 1A and 7).

Collagen Content in Peripheral Lung Tissue

The collagen content of lung parenchymal tissue was determined by measuring the level of hydroxyproline in tissue samples. Total protein and DNA content were also measured in these samples, and the collagen content was expressed as µg hydroxyproline/mg protein (Figure 8A) or µg hydroxyproline/ µg DNA (Figure 8B). Exposure to hypoxia did not measurably affect the collagen content in lung parenchyma during the 3-d or 10-d periods of exposure. Collagen content was not determined in rats following 6 h of hypoxia.

View larger version (21K): [in this window] [in a new window]   Figure 8.   Total collagen (hydroxyproline) content in peripheral lung tissue of rats exposed to air (open bar) or hypoxia (solid bar) for 3 d and 10 days. Values (means ± SE, n = 5 in each subgroup, except for Day 10 air [n = 4] and hypoxia [n = 6]) are expressed as µg hydroxyproline/mg protein (A) or µg hydroxyproline/µg DNA (B).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Evidence That Alveolar Hypoxia Increases the Pressure in Pulmonary Capillaries

Studies have shown that capillary pressure increases during alveolar hypoxia. For example, Nagasaka and colleagues (28) used micropuncture techniques to determine capillary pressure in isolated perfused cat lungs following exposure to 2% O₂. Calculations from their data (28 [Figure 3]) show that mean capillary pressure increased by approximately 25% (from 13.5 to 17 cm H₂O) during hypoxia. Similarly, Siegel and associates (29) used pulmonary artery occlusion techniques to measure pulmonary capillary pressure in anesthetized sheep, and found increases of 17 to 49% during hypoxia (14% O₂ for 10 to 15 min).

Previous Studies Suggesting That Increased Wall Stress Causes Pulmonary Vascular Remodeling

Vascular remodeling is a dynamic process that involves both synthesis and degradation of collagen molecules. Functionally, types I, III, and IV collagen provide tensile strength to the vessel wall. Types I and III collagen are synthesized by fibroblasts, myofibroblasts, and smooth-muscle cells, and are distributed diffusely throughout the media and adventitia of small vessels and on the thick side of the capillary wall. In contrast, type IV collagen is synthesized by endothelial cells, epithelial cells, smooth-muscle cells, pericytes, and myofibroblasts, and is associated with the basement membrane, where it provides major support for the thin side of the blood-gas barrier (2).

Poiani and coworkers observed that the rate of collagen synthesis and the amount of 1(I) procollagen mRNA in the pulmonary artery of rats reached maximal values following exposure to hypoxia for 3 d, then returned to control levels after 10 d (30). In addition, the amount of collagen (as reflected by hydroxyproline content) began to increase in large pulmonary arteries on the fifth day of hypoxia, with the major response occurring on Days 7 to 10 (31). Interestingly, changes in collagen content were not observed at any time in vessels smaller than 250 µm diameter (31).

Comparison of Our Results with Those Obtained by Other Investigators

In the present study, rats were exposed to normobaric hypoxia (10% O₂ at sea-level barometric pressure) for 3 d or 10 d. We reasoned that hypoxia would increase transmural pressure in some small vessels and capillaries of the lung, and would thus increase mechanical stress in the vessel wall. The increased wall stress would then induce gene expression of ECM components to strengthen the vessel wall. We found that levels of mRNA for 1(I), 1(III), and 2(IV) procollagens and fibronectin were increased in peripheral lung tissue of rats after exposure to hypoxia for 3 d (Figures 1-4). In addition, increases in mRNA levels were observed for 2(IV) procollagen and PDGF-B after 6 h of hypoxia (Figures 1A, 4A, and 6). In contrast, the level of VEGF mRNA was unaltered during hypoxia (Figure 7).

View larger version (13K): [in this window] [in a new window]   Figure 7.   Densitometric analysis of Northern blots hybridized to 32P-oligolabeled cDNA specific for VEGF mRNA in peripheral lung tissue of rats exposed to air (open bar) or hypoxia (solid bar) for up to 3 d (A) or for 3 d and 10 d (B). mRNA levels are expressed in arbitrary densitometric units normalized for loading. Values represent means ± SE (n = 5 in each subgroup, except for Day 10 air [n = 4] and hypoxia [n = 6]).

These results are similar to those reported by Poiani and coworkers (30) for the pulmonary artery, and suggest that vascular remodeling also occurs in peripheral blood vessels of rats following hypoxia. Furthermore, expression of the gene for PDGF-B has been reported to increase in response to mechanical stress (32) and low oxygen tension (33) in isolated cell populations, and the observed changes suggest that PDGF-B may participate in vascular remodeling of peripheral vessels during hypoxia-induced hypertension. Because of the observations of Tuder and colleagues (34), the lack of a change in levels of VEGF mRNA during hypoxia were unexpected. However, the two- to threefold increase in levels of VEGF mRNA reported in their study (34 [Figure 2]) did not occur until 28 to 32 d after exposure of rats to hypoxia, and ventilation with 0% O₂ was used to cause the observed two-fold increase in VEGF mRNA at 2 h in isolated perfused rat lungs (34 [Figure 7]). Because of these differences in timing and experimental procedure, the results of Tuder and colleagues (34) do not conflict with the observations in the present study.

We also found that collagen levels were unchanged in peripheral lung tissue following exposure of rats to hypoxia for 3 d or 10 d. This observation is similar to that of Tozzi and coworkers (31), as discussed earlier. The observed lack of a change in the collagen content of lung samples in our study therefore supports our contention that extracapillary vessels in peripheral lung parenchyma are predominantly small (e.g., < 250 µm in size).

A recent study by Parker and colleagues (12) provides additional evidence that mechanical forces (increased wall stress) cause vascular remodeling in pulmonary capillaries. In that study, cyclic increases in venous pressure, or high airway pressure, were used to increase capillary wall stress in isolated perfused rat lungs. After 4 h, lung tissue was collected and mRNA levels for EMC components and TGF-₁ were measured. Pulmonary capillary filtration coefficients (Kfc) were also determined, to confirm that the selected pressure regimens were sufficient to cause increased microvascular fluid conductance. Increased lung levels of mRNA for types I and III procollagen were found in rats perfused at high venous pressure as compared with rats perfused at low venous pressure, and levels of mRNA for fibronectin were increased in the lungs of rats perfused at both high and low pressures as compared with unperfused controls. In contrast to our findings in the present study, no changes were observed in levels of 2(IV) procollagen mRNA following perfusion at high venous pressure as compared with values for rats perfused at low pressure (12). This difference may reflect differences in the experimental models used in the two studies (isolated perfused lung versus in vivo exposure), differences in anatomic site of pressure application to the capillary in the two models (e.g., from the distal end of the capillary by high venous pressure, as compared with the proximal capillary during in vivo hypoxia), or differences in the response times of intact animals and isolated perfused lung preparations (35).

The results we report here disagree in several respects with those reported by Vyas-Somani and coworkers (18). They exposed Sprague-Dawley rats in a hypobaric chamber in which the PO₂ was about 90 mm Hg. This means that the hypoxic stress in their animals was not as great as in our animals, for which, after Day 1, the PO₂ of the air was about 76 mm Hg. Vyas-Somani and coworkers (18) looked specifically at changes in basement membrane components, including type IV collagen, fibronectin, and laminin in vascular tissue, airways, and gas-exchange compartments of the lung. They found progressive increases in the deposition of fibronectin and laminin but not in that of type IV collagen in large and small pulmonary arteries, and did not find such increases in airways or lung parenchyma. This contrasts with our findings, all of which pertain to peripheral lung tissue. Another area of contrast was that when they did observe changes in mRNA for fibronectin, laminin, and type IV collagen, they found reductions at Days 4 and 7 followed by increases at Day 12. We do not have any explanation for these contrasts between their findings and ours except that our levels of hypoxia were more severe.

Does Hypoxia Have a Direct Effect on Vascular Remodeling?

Our hypothesis is that the increase in wall stress of some capillaries and other small pulmonary blood vessels is responsible for the increased gene expression for the connective tissue proteins described here. However, it is also possible that hypoxia per se plays a role in this expression. Changes have been observed in levels of PDGF-B mRNA in cultured human umbilical vein endothelial cells (HUVECs) under hypoxic conditions (33), and in levels of VEGF mRNA in lung tissue following perfusion of isolated lungs or exposure of rats to hypoxia (34). However, the levels of hypoxia at which these changes were observed were much more severe than in our preparation. For PDGF-B, the response of the cultured HUVECs to hypoxia varied inversely with oxygen partial pressure, with mRNA levels increasing eightfold at 1% O₂ and twofold at 3% O₂ over those of cells grown in 21% O₂. Isolated perfused rat lungs exhibited a twofold increase in VEGF mRNA levels after 2 h of perfusion with 0% O₂. These are much more severe levels of hypoxia than were used in our experiments. In the same study (33), chronically hypoxic rats at a simulated altitude of 5,000 m showed a twofold increase in VEGF mRNA, but this was after 32 d of exposure, a much longer duration than the 4 h of our experiment.

In summary, we found that mRNA levels for 1(I), 1(III), and 2(IV) procollagens and fibronectin increased significantly in peripheral lung tissue of rats following exposure to hypoxia for 3 d. Additionally, we observed increases in mRNA levels for PDGF-B after 6 h exposure, and it is possible that this growth factor participates in regulation of ECM remodeling during hypoxia. Increases also occurred in 2(IV) procollagen mRNA at 6 h, suggesting that this basement-membrane component responds more quickly to increased wall stress than do the other procollagens or fibronectin. In contrast, changes were not observed in levels of VEGF mRNA in hypoxic rats. These results were highly reproducible and suggest that some structures in peripheral lung tissue undergo remodeling in response to hypoxia-induced hypertension. Although we cannot conclude that the pulmonary capillaries are responsible for these changes, the results of our study, taken in conjunction with those reported by Berg and associates (11) and Parker and coworkers (12), are consistent with the hypothesis that increased capillary wall stress results in remodeling of the blood- gas barrier.