Main

Glucocorticoids have well-documented maturational effects on the fetal lung(1), although the mechanisms by which they promote maturation are still uncertain. One of the primary functional changes seen in experimental animals is an increase in lung compliance(2,3). The compliance of the lung parenchyma is primarily determined by the surface tension of the surfactant film at the air liquid interface and by the mechanical properties of the continuous fiber network, which forms a structural "skeleton" within the fluid continuum of the pulmonary interstitium(4). The primary fibers of this structural skeleton are collagen and elastin. The functional status of these fibers is known to impact on the mechanical properties of the lungs, as demonstrated by a number of in vitro and in vivo studies in which proteases have been used to selectively degrade collagen or elastin(5–7).

Information on the effect of glucocorticoids on interstitial collagen and elastin during lung development is scant. The reported effects of glucocorticoid treatment on fetal lung elastin synthesis are varied; some studies suggest a reduction(8) and others an increase(9,10). Lung collagen synthesis is generally inhibited in the presence of glucocorticoids(11–13). In addition to the effect on collagen and elastin metabolism, glucocorticoids have been shown to inhibit in vitro activity of the enzyme lysyl oxidase, which catalyses the cross-linking of mature collagen and elastin fibers(14).

This study was one of a series in preterm sheep investigating feasibility and optimal treatment strategies for direct fetal hormone treatment(15–17). Developmental changes in lung function, with or without antenatal hormone exposure, have previously been reported in this group of animals(18). Fetuses received a single intramuscular injection of betamethasone (0.5 mg/kg) 48 h before delivery. Thyroid hormones have been shown to enhance the maturational response to betamethasone(19,20); therefore, T4 (15 mg/kg) was administered in conjunction with betamethasone. The improvement in lung function with hormone treatment in these animals coincided with significant morphometric changes in the lung parenchyma(21). Because collagen and elastin are important determinants of tissue mechanics(4), we hypothesized that changes in lung collagen and elastin content may also contribute to the improvement in lung mechanics. The aims of the present study were first to examine changes in lung collagen and elastin and second to extend our preliminary morphometric examination to provide a more detailed assessment of the alveolar region and to include an evaluation of changes in the pleura and interlobular septa. Proliferative activity is greatest in newly forming secondary alveolar septa during alveolarization(22); therefore, we were particularly interested in this region of the lung parenchyma. Because the connective tissue framework of the alveolar interstitium is continous with that of the pleura and interlobular septa(4), and these structures may contribute to mechanical behavior of the lung(23), we reasoned that structural changes in these regions might also be important.

METHODS

Animal preparation. Protocols were approved by the Animal Ethics Committees at the Harbour-UCLA Medical Center and the Western Australian Department of Agriculture. Fetal sheep received a single ultrasound-guided intramuscular injection of betamethasone (0.5 mg/kg) and T4 (15 µg/kg) or vehicle control (saline) 48 h before delivery at 121, 128, or 135 d (d 121, d 128, d 135) gestational age (term approximately 148 d). Betamethasone dose and treatment to delivery interval in the present study were chosen on the basis of results of a previous study(15). T4 dose was based on the replacement dose of T4 used in congenital hypothyroidism(24). Fetuses were delivered by cesarean section and mechanically ventilated for 40 min to assess lung function, using a ventilation strategy designed to optimize ventilation and minimize the risk of ventilator-induced injury(16,18).

Preparation and sampling for morphometry and image analysis. Because lung maturation is known to vary between regions of the lung(25,26), all morphometric assessments were performed on the right cranial lobe. Other regions were processed for biochemical measurements and were therefore not available for morphometric examination(27). The right cranial lobe was fixed overnight via bronchial instillation of Karnovsky's fixative. A pressure of 30 cm H2O was chosen to ensure adequate inflation of these immature lungs.

Five animals per group were randomly chosen for morphometric assessment. They were representative of the group as a whole for all physiologic variables recorded (data not shown). Morphometric assessment of changes in the gas exchange region of these lungs has previously been reported(21). However, because proliferative activity is greatest in the secondary septa during alveolarization(22), hormone exposure might be expected to preferentially induce changes in this region; therefore, we extended our morphometric examination to include a separate assessment of secondary septa. In addition, we made morphometric measurements on the pleura and interlobular septa, because structural changes in these regions might also impact on lung function. Collagen and elastin content were measured in the parenchyma (as a whole and in secondary septa separately), pleura, and interlobular septa. FLV was measured by volume displacement(28). Each lobe was cut into 5-mm serial slices, and three slices were randomly chosen for morphometric examination(29).

Morphometry. Parenchymal volume per lobe (PV) was derived from FLV as follows: PV = PF × FLV, where PF is the volume fraction of lung parenchyma and is estimated by superimposing a linear point counting grid (464 lines/928 points) onto photographic prints of 5 mm hematoxylin and eosin (H&E) stained sections (magnification × 16). PF is equal to Pi/Pt, where Pi is the number of test points hitting the structure of interest (parenchyma), and Pt is the total points hitting the reference space (i.e. parenchyma + other structures). Pleural and interlobular septal volume were similarly estimated. A substantial proportion of lung parenchyma in d 121 animals appeared to be uninflated. This was less evident in animals of later gestations. It was assumed that those regions of parenchyma that filled with fixative were likely to have been aerated during mechanical ventilation, whereas those that remained closed and collapsed were poorly or nonaerated. The volume fraction of aerated (inflated airspaces, PFaer) and nonaerated parenchyma (collapsed airspaces, PFcol) was also estimated from these photographic prints.

The volume fraction of alveolar septa (ASF) and alveolar airspace (AAF), MLI, an index of alveolar size, and mean alveolar septal thickness (TD) were estimated by superimposing a linear point counting grid (21 lines/42 points) onto gray-scale images of lung parenchymal fields, captured as described below. A total of 30 images per animal were captured (magnification × 1800). MLI was calculated according to the formula MLI = 2Lr/I0, where Lr is the length of the test line within the reference space, and I0 is the number of intercepts with the air tissue interface. Alveolar wall thickness was determined as volume per unit area of alveolar surface according to the formula TD = (ASF × Lr)/2I0, where ASF is the volume fraction of alveolar septa. Total ASV per lobe was estimated as follows: (equation 1) where PF is total parenchymal volume fraction, ASFaer and ASFcol are volume fraction of alveolar wall for aerated and collapsed parenchyma, and PFaer and PFcol are volume fraction of aerated and collapsed parenchyma, respectively. AAV was similarly estimated substituting AAFaer and AAFcol into equation 1.

Secondary alveolar septa (septal crests) are initially visible as ridges along the walls (primary septa) of primitive airspaces, lengthening progressively to subdivide the airspaces into alveoli(30). Secondary septa were manually isolated on stored images by drawing a straight line along the base of the secondary alveolar septum at the "junction" of the primary and secondary septa, parallel with the luminal surface of the primary septum. The number and volume fraction of septal crests were measured only in regions of aerated parenchyma. Septal crest number was expressed as number per "adjusted" field area. Adjusted field area was obtained by multiplying by a factor proportional to volume of parenchymal tissue per lobe, to account for variability in parenchymal volume between animals.

Average pleural thickness was estimated from a total of 150 measurements per animal (5 measurements per image × 5 images × 6 sections). Measurements were made at five equidistant points along the length of each pleural image captured from Picrosirius red and Miller's elastic stained sections (see below). Interlobular septal thickness was similarly estimated on images captured from Picrosirius red stained sections.

Collagen and elastin content. Parenchymal collagen and elastin content were estimated from Picrosirius red- and Miller's elastic stained sections, which selectively stain collagen(31) and elastin(32), respectively (Fig. 1). A Scion 50Hz LG3 frame grabber card and a Sony 3CCD color video camera connected to a Leitz Dialux 20 microscope were used to capture gray-scale images. Both collagen and elastin were able to be distinguished from surrounding tissue on the basis of gray-scale pixel density. Background light conditions were adjusted to ensure the average gray-scale pixel density of a standard captured image was constant at each sitting. From each 5-mm Miller's elastic stained section, images from five nonoverlapping, aerated parenchymal regions were captured using National Institutes of Health Image (Version 1.59; National Institutes of Health, Bethesda, MD) at a magnification of ×40 (final on-screen magnification ×1800) and stored on a Macintosh Power 80AV/8100. Each captured image was digitally enhanced first by performing a simple arithmetic transformation (i.e. multiplying each pixel value by 1.5) and then by sharpening (3 × 3 kernel configuration), which accentuated the difference between elastin and surrounding tissue, the elastin having a higher pixel density (Fig. 2). To create binary images, we used a carefully chosen threshold pixel density that rendered the elastin black (gray-scale level 255) and the remainder of the image (both alveolar septa and airspace) white (gray-scale level 0). The area of elastin was then estimated by applying a spatial calibration and measuring the number of black pixels. This procedure was repeated on regions of collapsed parenchyma. Elastin content as a fraction of alveolar septa was calculated according to the formula EF = (area of elastin per field × 100)/(ASF × field area). The EF was estimated in both aerated and collapsed parenchymal regions. Total PElobe was estimated as follows: (equation 2) where EFaer and EFcol are parenchymal elastin fraction in aerated and collapsed parenchymal regions, respectively. Collagen volume fraction and PClobe were similarly estimated on 5-mm Picrosirius red stained sections.

Figure 1
figure 1

Panel A: Miller's elastic stained 5-µm section. d 135 control animal (No. 93-123), original magnification ×230. Elastin (black) confined to outermost layer of pleura (P) and walls of blood vessels in interlobular septa (S). Inset: Alveolar septum, original magnification ×920. Elastin primarily confined to secondary septa (arrows). Panel B: Picrosirius red stained 5-µm section. d 135 control animal (No. 93-122), original magnification ×230. Collagen fibers (red) evenly distributed throughout pleura (P) and interlobular septa (S). Inset: Alveolar interstitium, original magnification ×920. Collagen evenly distributed throughout alveolar septa.

Full size image
Figure 2
figure 2

Panels A to C: Digitized gray-scale images of 5-µm Miller's elastic stained section. Positively stained material (elastin) differentiated from surrounding tissue on the basis of gray-scale density. A: Original gray-scale image. B: Arithmetic manipulation of original image. Each pixel multiplied by 1.25 to accentuate the difference between density of elastin and surrounding tissue. Panel C: Sharpening of multiplied image to increase contrast and accentuate detail. Image filtered by applying a 3 × 3 spatial convolution, where the value of each pixel is replaced by the weighted average of its 3 × 3 neighborhood. Original magnification × 1800. Panels D to F: Linear density profiles through secondary septum (region indicated by black line on images A to C). D: Density profile of septum in original image, showing two poorly delineated positively stained regions. E: Density profile of multiplied image. Pixel densities now extend over greater range, heightening difference between positively stained and negatively stained regions. F: Density profile of sharpened image. Enhanced contrast and sharper delineation of boundaries enables better separation of adjoining positively stained regions.

Full size image

Septal crest elastin content was estimated after isolating the septal crest regions within a given field, using the same threshold as previously determined for the whole field. EFcr was calculated according to the formula EFcr = crest elastin per field/crest area per field on Miller's elastic stained sections. CFcr was similarly estimated on Picrosirius red stained sections.

Pleural collagen and elastin fractions were relatively high in comparison with parenchymal fractions and were easily estimated stereologically (see Fig. 1). Five random, nonoverlapping images of pleura were captured from Miller's elastic stained sections. In most instances it was possible to capture images at ×40; however, where the pleura was relatively thick the magnification was reduced to fit the entire thickness of the pleura within the image frame. Because the pleura is anisotropic, a cycloid grid (45 lines/135 points) was used. EFpl was estimated by point counting. Total pleural elastin per lobe was derived as follows: PLElobe = EFpl × (PLF × FLV), where PLF is pleural volume fraction. CFpl and PLClobe were estimated on Picrosirius red stained sections.

Interlobular CFS was also relatively high in comparison with parenchymal collagen and could be easily estimated by point-counting as described above. In most instances images were captured at ×25; however, where the interlobular septum was relatively thick the magnification was reduced to ×10. Interlobular CFS and total SClobe were estimated as for pleural collagen. Apart from that surrounding blood vessels, interlobular septa were devoid of elastin in these preterm sheep (see Fig. 1); therefore, images were not captured on Miller's elastic stained sections.

Intraobserver variability. All morphometric and image analysis measurements were performed blind and by the same observer (K.E.W.). To investigate intraobserver variability of image analysis measurements, repeatability of elastin measurements was determined. Duplicate measurements (runs) were made 2-3 wk apart, the observer being blinded to initial measurements on the second occasion. Five parenchymal images per field were collected from a total of 30 tissue blocks (5 animals each of gestational ages d 121, d 128, and d 135; 2 blocks per animal). Variability was examined using multilevel modeling. Animal number (1-15), field number (1-5), section number (1-2), and run number were included as random effects in the model. Measurements on the same section, field, and animal were highly reproducible, with less than 3% of the total variability being attributed to repeated runs.

Statistical analyses. Analysis was based on a linear regression model analogous to analysis of variance, with treatment and gestational age as predictors. The model included a gestational age/treatment interactive term to allow for any systematic variation in the effect of hormone treatment with gestational age. For cases in which a significant (or borderline significant) interaction was found, the response to treatment at each gestational age was studied separately and observations were noted. Statistical significance was accepted at p < 0.05.

RESULTS

Parenchymal, nonparenchymal, and FLV. FLV increased by approximately 75% between d 121 and d 135, whereas parenchymal volume almost doubled (p < 0.0005 for both parameters, Table 1). This increase in parenchymal volume was primarily due to an increase in AAV, which increased almost 3-fold (p < 0.0005), whereas ASV did not change appreciably (Table 1). Total volume of nonparenchymal tissue per lobe (airways plus blood vessels) increased substantially between d 121 and d 128 (p < 0.05), but there was no further increase between d 128 and d 135.

Table 1 Lobe volumes
Full size table

Overall, neither FLV nor parenchymal volume was significantly altered by hormone treatment, although a slight increase was observed at d 135 (Table 1). AAV increased with hormone treatment at each gestational age, and this was most marked at d 135. In contrast, ASV decreased with hormone exposure, particularly at d 121 (p < 0.05). The volume of nonparenchymal tissue was unaffected by hormone treatment.

Alveolar size, wall thickness, and secondary septal characteristics. MLI was approximately 50 mm and did not vary with gestational age (data not shown). Alveolar septal thickness decreased by approximately 40% between d 121 and d 135 in control animals (p < 0.0005, Table 2). In line with the pattern of changes seen with normal maturation, alveolar septal thickness decreased significantly with hormone treatment at each gestational age (p < 0.0005), whereas MLI did not change.

Table 2 Alveolar septal, pleural and interlobular septal thickness
Full size table

The number and volume fraction of secondary alveolar septa provides an indication of septation of airspaces. Both volume fraction of secondary septa and the septal crest number per field increased with increasing gestational age (p < 0.05 and p < 0.0005, respectively), the greatest increase in both parameters occurring between d 121 and d 128 (Table 3). In contrast to the pattern seen with normal maturation in control animals, hormone treatment led to a significant increase in volume fraction of secondary septa but did not increase septal crest number.

Table 3 Secondary septal volume fraction and number and collagen and elastin content
Full size table

Pleura and interlobular septa. Pleural thickness varied quite considerably within each animal. For example, in one animal, values ranged from 23 to 177 µm, which was typical of the group. However, with a total of 150 measurements per animal, average pleural thickness was found to be strongly associated with gestational age, decreasing progressively between d 121 and d 135 (p < 0.0005). Pleural thinning coincided with decreasing pleural volume between d 121 and d 135 (p < 0.05, Table 2). Both pleural volume and pleural thickness also decreased significantly with hormone treatment at each gestational age (p < 0.05 and p < 0.0005, respectively).

Interlobular septa were very conspicuous in d 121 and d 128 animals, but less so in d 135 animals. As with pleural thickness, interlobular septal thickness varied considerably within each animal. Average interlobular septal thickness decreased significantly in control animals between d 128 and d 135 (p < 0.0005), although total interlobular SV was relatively constant (Table 2). Interlobular septal thickness also decreased significantly with hormone treatment at d 128 and d 135 (p < 0.05). Hormone treatment led to a reduction in SV at each gestational age (p < 0.05). This was most pronounced in d 135 animals, in which SV decreased by almost 40%.

Parenchymal collagen. In control animals, CFaer increased progressively between d 121 and d 135 (p < 0.0005, Fig. 3A). CFcol was able to be estimated in all control animals at d 121 and d 128 but in only three of the five animals at d 135. Regions of collapsed parenchyma were too few and too small in the remaining two animals to enable measurements to be made. As with CFaer, CFcol increased slightly with gestational age, but this was not significant (Fig. 3B). Paralleling the increase in CFaer, PClobe increased progressively with advancing gestation (p < 0.0005, Table 4).

Figure 3
figure 3

Group mean values ± SEM are shown for control (white) and hormone-treated (black) animals. Panel A: CFaer. Significant increase in CFaer with gestational age and with hormone treatment (p < 0.0005, respectively). Panel B: CFcol. No measurements were made on d 128 or d 135 hormone-treated animals, as regions of nonaerated parenchyma were too few and too small. Panel C: EFaer. Significant increase with gestational age in hormone-treated animals (p < 0.05). Significant increase with hormone treatment at d 135 (p < 0.05). Panel D: EFcol. No measurements were made on d 128 or d 135 hormone-treated animals, because regions of nonaerated parenchyma were too few and too small.

Full size image
Table 4 Total collagen and elastin volumes
Full size table

CFaer was uniformly increased by hormone treatment at each gestational age (p < 0.0005, Fig. 3A). CFcol was able to be estimated in only four of the 15 hormone-treated animals, all of which were delivered at d 121. It was not possible to take measurements from d 128 or d 135 hormone-treated animals, because regions of nonaerated parenchyma were too few and too small. In all 17 animals (13 control, four hormone-treated) from which CFaer and CFcol measurements were taken, CFcol was consistently lower than CFaer (p < 0.0005, paired t test). In contrast to the increase in PClobe seen with increasing gestation, hormone treatment did not significantly increase PClobe at any gestation (Table 4).

Parenchymal elastin. Parenchymal EF was consistently approximately 4- to 6-fold lower than collagen fraction. In contrast to the increase in CFaer with gestational age, EFaer did not change in control animals (Fig. 3C). Similarly, there was no significant change in EFcol between d 121 and d 135 for control animals (Fig. 3D). The EFcol was higher than EFaer in 11 of 17 animals, but this was not statistically significant (p = 0.071, paired t test). Total PElobe varied significantly with gestational age (p < 0.01), increasing between d 128 and d 135 (Table 4).

In contrast to the pattern seen in control animals, EFaer increased progressively with advancing gestation in hormone-treated animals. The effect of hormone treatment on EFaer varied significantly with gestational age (p < 0.05), increasing at d 135 but not at the earlier gestations (Fig. 3C). As with CFcol, EFcol could not be estimated on d 128- and 135-treated animals. EFcol was not significantly different for control and hormone-treated animals at d 121 (Fig. 3D). In line with the increase in EFaer, PElobe also increased progressively with gestational age in treated animals (p < 0.005, Table 4). The effect of hormone treatment on PElobe varied significantly with gestational age (p < 0.05), decreasing markedly at d 121 and increasing slightly at d 135.

Collagen and elastin content of secondary alveolar septa. Septal CFcr was more than double that in the parenchyma as a whole and increased with advancing gestational age in control animals (p < 0.0005, Table 3). Approximately 15-20% of total parenchymal collagen was localized to secondary alveolar septa, and this proportion did not vary significantly between d 121 and d 135 (Table 3). Septal EFcr was 4- to 6-fold higher than in the parenchyma as a whole and varied significantly with gestational age (p < 0.05, Table 3), being lower at d 128 than at d 121 or d 135. Crest elastin accounted for 40-60% of total parenchymal elastin and was higher at d 135 than in the less mature animals (p < 0.0005, Table 4).

Neither CFcr nor the proportion of total collagen localized to secondary septa was affected by hormone treatment at any gestational age (Table 3). The effect of hormone treatment on EFcr varied systematically with gestational age (p < 0.05), causing a decrease at d 121 and an increase at d 135 (Table 3). Hormone treatment did not alter the proportion of total elastin found in secondary septa at any gestational age (Table 3).

Pleural collagen and elastin. In control animals, CFpl increased significantly from approximately 30% at d 121 to approximately 45% at d 135 (p < 0.0005, Fig. 4A). Although the pleura occupies only 3-4% of total lobe volume, PLClobe is roughly equal to total parenchymal collagen volume. There was no change in PLClobe with gestational age (Table 4). EFpl was approximately one tenth that of the CFpl. EFpl increased significantly with gestational age (p < 0.0005), doubling between d 128 and d 135 (Fig. 4B). Despite the large increase in EFpl between d 128 and d 135, PLElobe did not change significantly with gestational age, remaining relatively constant at approximately 20-30 × 10-3mm3 (Table 4). Total PLElobe was approximately half that of total parenchymal elastin volume.

Figure 4
figure 4

Group mean values ± SEM are shown for control (white) and hormone-treated (black) animals. Panel A: CFpl. Significant increase with gestational age (p < 0.0005). Panel B: EFpl. Significant increase with gestational age (p < 0.0005) and with hormone treatment at d 128 (p < 0.05). Panel C: Interlobular CFs. Significant increase with gestational age (p < 0.0005).

Full size image

Neither CFpl nor PLClobe was altered by hormone treatment at any gestational age (Table 4). Overall, the effect of hormone treatment on EFpl was not significant, although there was a significant gestational age/treatment interaction (p < 0.05); whereas there was no effect of hormone treatment at d 121 or d 135, elastin fraction doubled after hormone treatment at d 128 (Fig. 4B). Despite this large increase in EFpl at d 128, PLElobe did not change (Table 4).

Interlobular septal collagen. Interlobular septa, like pleura, are rich in collagen fibers. In control animals, interlobular CFs increased with gestational age (p < 0.0005, Fig. 4C). Total volume of SClobe was approximately 3-5 times that of either pleural or parenchymal collagen volume, representing 60-70% of total interstitial collagen and did not change appreciably with gestational age. Hormone treatment did not significantly alter CFs at any gestational age (Fig. 4C) but did reduce total interlobular septal collagen volume (p < 0.05, Table 4), particularly at d 121.

DISCUSSION

This study investigated changes in lung morphometry and in interstitial collagen and elastin content of the developing lung, both as a function of gestational age and in response to prenatal exposure to betamethasone and T4. To place the changes associated with hormone treatment in their proper context, it is important to first understand the changes that occur during normal development. Between d 121 and d 135, FLV increased by more than 70%. This was primarily due to expansion of lung parenchyma, the volume of this compartment increasing from 17 to 33 mL. The increase in parenchymal volume, in turn, resulted primarily from increasing alveolar airspace volume, which almost trebled. The volume of alveolar septal tissue, on the other hand, did not change appreciably during this period, accounting for approximately 7-8 mL of parenchymal volume at each gestation.

When adjusted for differences in parenchymal volume, the number of secondary alveolar septa increased markedly between d 121 and d 135, suggesting an increase in alveolar number through the process of septation. The volume fraction of secondary septa also increased during this period. This is partly a reflection of outgrowth of secondary septa into the airspaces and partly a reflection of the reduction in reference volume due to alveolar wall thinning. Alveolar size did not change with increasing gestational age, as MLI remained constant. These observations suggest that the increase in potential airspace volume with advancing gestation is due to an increase in alveolar number, but not size. This is consistent with the work of Docimo et al.(33), who reported that alveoli are evident in fetal sheep by d 114 and that the increase in lung volume during subsequent development is by formation of new alveoli.

Alveolar septal thickness decreased by approximately 40% between d 121 and d 135, but despite this marked thinning, alveolar septal tissue volume was preserved, suggesting a "redistribution" of tissue in the formation of secondary alveolar septa. Increasing gestation was also associated with thinning of pleura and interlobular septa, although it is not clear whether this represents a developmental change or whether it is due to more rapid clearance of fetal lung liquid from the interstitium during ventilation in animals of a later gestation. The volume fraction of interlobular septa in unventilated animals is approximately one third of that in animals that have been ventilated for 40 min (Willet, unpublished observation), suggesting that interstitial volume may be at least partially dependent on the volume of fetal lung fluid absorbed before lung fixation. Fetal lung liquid is cleared from the lungs first by absorption from the airspaces into the interstitium, followed by removal from the interstitium via the pulmonary circulation, and, to a lesser extent, lymphatics(34). Complete clearance of lung liquid requires approximately 6 h in term lambs(34) and slightly longer in preterm animals(35). The influence of prenatal glucocorticoids on the rate of clearance of fetal lung liquid during mechanical ventilation is not yet known.

The morphometric changes in the lung parenchyma after glucocorticoid exposure were not analogous with normal maturational changes. The reduction in alveolar septal thickness after hormone exposure was not associated with a "redistribution" of septal tissue into secondary septa, as was seen with normal maturation. In control animals, the 25% decrease in alveolar wall thickness between d 121 and d 128 was associated with a 50% increase in the number of secondary septa, and between d 128 and d 135 the 15% decrease in alveolar septal thickness coincided with a 30% increase in secondary septal number. In contrast, there was no change in secondary septal number with hormone treatment at d 121, despite a 25% decrease in septal thickness. Similarly, at d 128, secondary septal number again did not change, despite a 15% reduction in alveolar septal thickness. Instead, the reduction in alveolar septal thickness with hormone treatment coincided with a reduction in alveolar septal volume. Loss of alveolar septal tissue was particularly pronounced at d 121, and this coincided with a marked reduction in wet lung weight (data not shown).

Although lung function studies suggest that thyroid hormones enhance the functional maturation induced by glucocorticoids(19,20), their individual effects on alveolar structure may differ. The outgrowth of secondary septa is an active process, associated with heightened proliferation of fibroblasts and endothelial cells(22). In rats, corticosteroid concentration is low during the period of alveolar septation and increases as septation is ending and alveolar wall thinning begins(36). In this species, administration of dexamethasone during the period of septation results in rapid thinning of alveolar walls(37) and markedly impaired septation(38,39). The effect persists into adulthood, because compensatory alveolarization after cessation of treatment is insufficient to overcome the initial deficit(40,41). The formation of secondary septa is thought to be possible only when the primary septa are immature, and precocious maturation by glucocorticoids is believed to remove the lung's ability to septate(30). In contrast, serum concentration of thyroid hormones appears to increase at the onset of septation(42). Furthermore, exposure to triiodothyronine during alveolar formation accelerates septation, whereas exposure to propylthiouracil, which inhibits the synthesis of thyroid hormones, impairs septation(43). Intuitively then, inhibition of septation by glucocorticoids may be expected to be blocked, at least in part, when administered concurrently with thyroid hormones. In the present study, however, the effects of betamethasone appear to overshadow the effects of T4, the net effect being precocious wall thinning in association with arrested septation.

Although total parenchymal collagen and elastin volume increased in control animals over this gestational period, only collagen exhibited an increasing fraction or concentration. An increase in protein fraction may represent an increase in protein volume or a decrease in the reference volume of the compartment in which it is measured. For example, in the case of CFpl, the increase with gestational age does not reflect an increase in pleural collagen per se, but rather a decrease in pleural volume with increasing gestational age. In contrast, the increase in parenchymal collagen fraction with increasing gestational age cannot be accounted for by a decrease in reference volume. In this instance the reference volume (alveolar septal volume per lobe) does not change with gestational age, but total parenchymal collagen volume increases, as reflected by the increase in parenchymal collagen fraction. The increase in total parenchymal collagen and elastin volume with increasing gestational age appears to represent a net increase in deposition of these proteins in the alveolar interstitium. In contrast, the increase in pleural collagen and elastin fractions are more a reflection of reduced pleural volume with advancing gestational age than an increase in protein deposition per se.

After prenatal hormone exposure, the primary change in interstitial collagen was an increase in collagen volume fraction. Because PClobe did not change, this increase reflected a decrease in reference volume (i.e. alveolar septal volume) rather than an increase in collagen deposition. The observation that the effect of hormone treatment was not on collagen per se, but on other components of the lung parenchyma, highlights the importance of determining the reference volume from which measurements are taken. The present finding of an increase in collagen fraction after hormone treatments is at odds with biochemical studies in rats(8) and rabbits(9), in which prenatal glucocorticoids had no effect on postnatal lung hydroxyproline concentration. A number of other studies suggest that glucocorticoids have a potent inhibitory effect on collagen synthesis and deposition under control conditions(12) and in models of lung fibrosis(11,13).

Alveolar elastin content increased slightly with treatment at d 135, an effect that is similar to that which would be expected with normal maturation. However, at d 121, prenatal hormone treatment led to a marked reduction in alveolar elastin content, in association with a proportionate loss of alveolar septal volume. This effect appeared to be specific to alveolar elastin, because no comparable change was seen in the pleura. Neither a reduction in alveolar elastin volume nor in alveolar tissue volume is consistent with normal lung maturation, because these volume do not change significantly between d 121 and d 128 in control animals. The inhibition of septation, coupled with a decrease in both elastin and alveolar tissue volume with hormone treatment at d 121, is consistent with studies that demonstrate an association between impaired elastic fiber formation and retarded alveolar septal development(44,45), suggesting that elastin may play a vital role in septal formation. The only previous study to examine the effect of glucocorticoids on lung elastin content in fetal sheep reported an increase in desmosine concentration after infusion of cortisol plus triiodothyronine and/or prolactin in fetuses of 128 d gestation(46). However, in this study, the increase in desmosine concentration was almost certainly a reflection of a reduction in reference volume, because wet lung weight decreased quite substantially.

Combined, pleural and interlobular septal collagen accounted for 80-90% of total interstitial collagen. If the conjecture that collagen fibers contribute significantly to the mechanical behavior of the lungs at high lung volume is correct(4), then pleural and interlobular septal collagen may be important determinants of maximal lung volume. By contrast, the major repository for elastin, an important determinant of lung mechanics at low to middle volume(4), was the alveolar interstitium. Hormone exposure at d 121 led to a reduction in both alveolar elastin and interlobular septal collagen, which may be important contributing factors to the increase in compliance with hormone treatment at this gestational age(16,18).

There is limited information on the effects of glucocorticoids and/or thyroid hormones specifically on interstitial collagen and elastin during lung development and their relationship to mechanical properties. There is a considerable body of literature that suggests that interference with collagen and elastin will alter the mechanical behavior of the lungs(5–7); however, the effects of prenatal hormone treatment on collagen and elastin metabolism are less clear. Beck et al.(47) found that intrauterine exposure of rhesus fetuses to betamethasone led to an increase in lung volume, which coincided with an increase in the ratio of parenchymal collagen to elastin. Schellenberg et al.(46) reported that both desmosine and hydroxyproline concentrations were significantly higher in hormone-treated fetal sheep that displayed distensible [volume of air at 40 cm H2O (V40) greater than 1.5 mL/g wet weight] and stable (volume of air at 5 cm H2O greater than 0.8 mL/g wet weight) lungs than in those with nondistensible, unstable lungs. However, in agreement with several of the findings in the present study, the increase in desmosine and hydroxyproline concentrations in Schellenberg et al.'s study appeared to be more a reflection of decreasing reference volume than an increase in net protein synthesis per se(46).

Direct fetal injection of a single, clinically relevant dose of betamethasone plus T4 48 h before delivery promoted significant changes in both morphometry and in collagen and elastin content in the lungs of fetal lambs delivered between 121 and 135 d gestation. Although most changes in collagen and elastin fraction (or concentration) could be attributed to changes in reference volume rather than to changes in collagen/elastin deposition per se, a marked reduction in parenchymal elastin at d 121 appeared to reflect a direct effect on elastin deposition. This reduction in elastin was paralleled by a proportionate loss in alveolar septal tissue. The loss of septal tissue in hormone-treated animals appears to reflect rapid thinning of alveolar septa without associated secondary septal formation, in contrast to the pattern seen during normal maturation in which alveolar septal tissue remained relatively constant, as septal thinning was counteracted by addition of secondary septa. These data suggest that prenatal hormone treatment may cause significant "dysmaturational" changes in the lung parenchyma, the most profound effects resulting from exposure early in gestation. Although the present study did not examine longterm effects, similar studies in rats suggest that precocious attenuation of alveolar septa may permanently impair the lungs' ability to septate(40,41).