Introduction

Top Abstract Introduction Materials & Methods Results Discussion References

Intrapulmonary airways are exposed to variable radial stresses generated by elastic elements contained within the lung and transmitted to the airway walls by their tensile attachments to the surrounding tissue (1). The extent to which such stresses are translated into airway strain or deformation depends on the contractile state of the airway smooth muscle, which is controlled primarily by the outflow of preganglionic vagal neurons. These neurons are functionally integrated in the control of breathing (2), which provides the central nervous system with a convenient regulatory mechanism for adjusting the mechanical coupling of airways and lung parenchyma according to the level of inspiratory activation.

Suppression of this mechanism by vagal denervation is likely to increase airway strain, especially when the transmural stresses applied to the airway walls by contraction of the respiratory muscles must rise to meet the demands of exercise or illness. Because repetitive strain can act as a stimulus for the synthesis of extracellular matrix proteins (5), lung denervation may lead to a fibroproliferative response in the airways. Such a response could explain, at least in part, the signs of airway obstruction observed after lung reimplantation in piglets (9, 10), and might contribute to the development of bronchiolitis obliterans after allogeneic lung transplantation in humans (11).

Based on these considerations, the present study was designed to characterize the effects of vagal denervation on the transcriptional activity of airway extracellular matrix proteins in rats. Specifically, we compared the levels of Type I procollagen and tropoelastin messenger ribonucleic acids (mRNAs) detected by in situ hybridization in the lungs after one of three experimental manipulations: unilateral lung denervation (causing increased airway strain with unaltered airway stress), unilateral vagal denervation and ipsilateral phrenectomy (decreasing airway strain and stress), and surgical manipulation without denervation (causing no change in strain and stress).

	Materials and Methods

Top Abstract Introduction Materials & Methods Results Discussion References

Surgical Preparation

Male Sprague-Dawley rats were assigned to one of three experimental groups (Table 1). The rats in the first group underwent unilateral lung denervation, consisting either of a cervical vagotomy (left or right, n = 21) or a left cervical vagotomy combined with an extensive destruction of pulmonary-bound vagal fibers at the pulmonary hilum (n = 19). The cervical vagotomy was performed through a paramedian cervical approach in which the vagus nerve was separated from the vascular package of the neck and resected for a distance of 1 cm. The hilar vagotomy was performed through a lateral thoracotomy. The lung was retracted forward to expose both the vagus nerve and the lung pedicle. The intrathoracic length of the nerve was dissected from the surrounding tissues, destroying in the process all fibers traveling toward the lung hilum. The surfaces of the bronchus and pulmonary vessels were then denuded from their serosal covers, in order to eliminate as many of the nerve fibers contained in these covers as possible. Both the cervical incision and the thoracotomy were closed by layers, the latter after removing all extrapulmonary air with a small tube left temporarily in the pleural space.

The rats in the second experimental group underwent cervical and intrathoracic vagotomies (as described earlier) and a resection of the pericardial segment of the ipsilateral phrenic nerve.

The rats in the third group underwent cervical dissection and a left thoracotomy without vagotomy. These rats served as controls for the potential effects of thoracotomy on procollagen mRNA production in the lung.

All surgical preparations were made with the aid of a dissecting microscope. The rats were anesthetized with halothane (0.5 to 2% in oxygen), administered initially into an induction box and, after cannulation of the trachea, through the inspiratory circuit of a Harvard 683 rodent ventilator (Harvard Apparatus, Cambridge, MA).

Removal and Preparation of the Lungs

Each rat was assigned randomly to a survival period of either 4 or 8 wk (except for nine rats in the first or unilateral lung denervation group, which, in order to obtain a more complete characterization of the time course of the airway changes produced by denervation, were killed 1 wk [n = 4] or 2 wk [n = 5] after the surgical preparation). At the end of the assigned period, the rat was anesthetized with pentobarbital and ventilated mechanically through a tracheostomy. The lungs and anterior mediastinum were exposed through an extended median sternotomy and the pulmonary artery was cannulated. Then, after the ventilator was replaced with a constant-flow circuit adjusted to keep transpulmonary pressure at 20 cm H₂O, and a venting hole was created in the left atrium, the lungs were sequentially perfused with heparinized saline and 10% formaldehyde (30 ml). Once fixed, the trachea and lungs were removed, stored in 10% formaldehyde for 72 h, and transferred through a series of ethanol washes of increasing concentration to 70% ethanol before being embedded in paraffin.

Assessment of Denervation

To define the extent to which the vagotomized lung was still supplied by preganglionic fibers from the contralateral vagus nerve, we measured the responses of each lung, right or left, to stimulation of the remaining vagal trunk. These measurements were made before removal of the lungs, on a sample that included 17 of the 40 rats that underwent unilateral vagotomy and 11 of the rats that underwent both unilateral vagotomy and phrenectomy. Similar measurements were made during unilateral stimulation of both vagal trunks in seven of the control rats.

After exposing the lungs as described earlier, a small ensemble, consisting of a Fleisch No. 000 pneumotachograph, in line with a small piece of tubing equipped with a side port for pressure measurement, was interposed between the tracheostomy cannula and the connector to the ventilator. Both the pneumotachograph and the side port of the cannula were attached to differential pressure transducers (MAP45, ± 2.3 and ± 56 cm H₂O, respectively; Validyne Engineering, Northridge, CA). The length of the attachments was adjusted to assure that the outputs of the transducers were well matched dynamically and had frequency-response characteristics appropriate for the signals recorded during the experiments (12). Next, the vagus nerve contralateral to the vagotomy site was dissected and placed on a bipolar electrode. The effect of supramaximal vagal stimulation (13) on the transpulmonary pressure and gas flow to each lung was then measured, after advancing the tapered end of the tracheal cannula into the corresponding main stem bronchus under direct visualization. The increase in flow resistance produced by vagal stimulation was estimated for each lung by dividing the maximum increase in transpulmonary pressure by the amplitude of the flow signal when the increase was maximum.

In Situ Hybridization and Special Stains

Procollagen mRNA levels were ascertained by in situ hybridization. The paraffin-embedded lung tissue was cut in 5-µm sections. The sections were deparaffinized in xylene, hydrated in decreasing concentrations of ethanol, digested in a solution containing 1 mg/ml nuclease-free proteinase K in phosphate-buffered saline (PBS) for 30 min at 37°C, and washed in 0.1 mol/L triethanolamine buffer containing 0.25% acetic anhydride. An antisense ³⁵S-labeled human Type I procollagen complementary RNA (cRNA) probe complementing the C-terminal propeptide region of ₁(I)- procollagen mRNA (Hf677 [14, 15]; a gift of Dr. M. D. Botney of Washington University) was prepared with [-³⁵S]UTP as previously described (16). This probe has been shown to hybridize specifically to rat ₁(I)-procollagen mRNA in situ (17). The sections from each rat were incubated together with the probe (2.5 × 10⁶ cpm) at 55°C for 18 h in a humidity chamber, and were then washed in a solution of NaCl, Na citrate, and dithiotreitol and exposed to a buffered solution containing 20 mg/ml of ribonuclease A (RNase A) for 30 min at 37°C to remove nonhybridized probe. The sections were then dehydrated in ethanol and processed for autoradiography (18). A ³⁵S-labeled sense probe transcribed from bovine tropoelastin cDNA was used as a negative control for the stringency of the wash conditions; this transcript is 68% GC-rich and therefore should have a high propensity for nonspecific hybridization (19). All slides were counterstained with hematoxylin and eosin (H&E).

Tropoelastin mRNA levels were determined with a similar in situ hybridization technique in adjacent lung tissue sections from the three experimental groups. An antisense ³⁵S-labeled rat tropoelastin cRNA (pREL 124DM; a gift of R. A. Pierce of Washington University [20]) was used. A ³⁵S-labeled sense probe transcribed from rat tropoelastin served as a negative control.

Slides from each rat were stained with the Verhoeff- van Gieson trichromic stain to obtain a basic assessment of collagen deposition in the walls of the airways after denervation.

Data Analysis

Slide pairs (right and left lung from a rat) were compared for Type I procollagen and tropoelastin mRNA signals by two independent observers blinded to the origin of the tissues. Each observer evaluated the pair twice in random order, assigning in the process a score of 0, 1, or 2 to the pair, depending on the intensity of the hybridization signal. In order to make a final decision, the observer was allowed to review those slides for which there was a discrepancy between the two evaluations. The results provided by the two observers were then contrasted, and the value of the  statistic was calculated to measure their agreement rate (21). To increase the accuracy of the analysis, however, mRNA levels were considered different only if the two observers agreed on the difference. The resultant information was arranged in a contingency table and analyzed with a chi-square analysis.

	Results

Top Abstract Introduction Materials & Methods Results Discussion References

Procollagen and Tropoelastin mRNA and Special Stains

In situ hybridization revealed an increase in airway and vascular ₁(I)-procollagen mRNA in the denervated as compared with the contralateral lung (Figure 1) in 11 of the 40 rats subjected to unilateral lung denervation (Group I), and in five of the 12 rats subjected to unilateral lung denervation and ipsilateral phrenectomy (Group II). No differences were detected between lungs in the control group. Chi-square analysis confirmed that unilateral denervation increased procollagen mRNA, but the increase was significant (P < 0.05) only when all the rats in the unilateral denervation and unilateral denervation and phrenectomy groups were pooled together (Table 2). The interobserver rate of agreement for differences in hybridization signal between lungs was 85% (rate of chance agreement: 61%), yielding a  value of 0.62. All disagreements (n = 11) resulted from one of the observers concluding that ₁(I)-procollagen mRNA was upregulated in the denervated lung and the other observer finding no difference between the denervated and nondenervated lungs. As stipulated earlier, all disagreements were resolved by considering that there was no difference between the pair.

View larger version (43K): [in this window] [in a new window]   Figure 1.   In situ hybridization (1-wk exposure) of denervated (right) and contralateral (left) rat lungs to a 35S-labeled 1(I)-procollagen cRNA 4 wk after cervical vagotomy in a 225-g rat. The various views illustrate the increase in transcript density affecting airways (b), pulmonary arteries (a), and, in some areas like the one shown in the right bottom panel, the interstitium of the denervated lung. There is virtually no riboprobe hybridization in the nondenervated lung.

The increased procollagen transcripts were located only in the subepithelial and serosal matrix of the airways (Figure 2) and in the vascular adventitia. No hybridization signal was noted in airway epithelium or vascular endothelium, or in airway or vascular smooth-muscle cells. However, in some cases, procollagen mRNAs were detectable in the parenchymal interstitium (Figure 1, right bottom panel ). The increase in procollagen mRNA was discernible as early as 2 wk after denervation, and extended from segmental bronchi to terminal bronchioles, with some variation between rats in terms of the size of the airways that were most prominently affected.

View larger version (149K): [in this window] [in a new window]   Figure 2.   Brightfield magnification of the wall of a denervated bronchus (see Figure 1, right top panel) after in situ hybridization to a 35S-labeled 1(I)-procollagen cRNA probe and counterstaining with H&E. The distribution of the developed photographic granules indicates that increased procollagen transcripts are present only in the airway's submucosal and serosal layers.

Neither the weight of the rats at the time of denervation nor the time elapsed after denervation had a discernible influence (as determined by an analysis of variance [ANOVA]) on whether procollagen mRNA was increased in the denervated lungs. The average weight of the rats that exhibited increased procollagen transcripts was 282 ± 71 g in the unilateral denervation group, as compared with 313 ± 76 g for the remainder of the rats in this group; and was 357 ± 37 g in the unilateral denervation and ipsilateral phrenectomy group, as compared with 376 ± 24 g for the remaining rats in this group. Increased transcripts were detected 2 wk after denervation in one rat, 4 wk after denervation in seven rats, and 8 wk after denervation in eight rats.

Verhoeff-van Gieson trichromic stains demonstrated variable increases in collagen deposition affecting both the serosal and submucosal layers and extending into the lung tissue surrounding denervated airways (Figure 3). There were no inflammatory cells in the walls of airways or vessels. However, an increase in cell density was noticeable in some airways and vessels, leading in extreme cases to replacement of the airway and vascular lumen by a cellular matrix with increased Type I procollagen expression (Figures 4 and 5).

View larger version (95K): [in this window] [in a new window]   Figure 3.   Verhoeff-van Gieson stains (collagen appears pink) of denervated (right) and contralateral lung (left) 4 wk after unilateral cervical vagotomy in a rat. The denervated bronchial wall (detail in inset) is thickened by collagen deposition on both sides of the smooth-muscle layer.

View larger version (171K): [in this window] [in a new window]   Figure 4.   Airway and vascular fibroproliferative changes observed in a 345-g rat 8 wk after unilateral lung denervation (combined cervical and hilar vagotomy) and ipsilateral phrenectomy. The lumina of many airways and vessels on the denervated (right) but not on the contralateral side (left) are occluded by a matrix-rich cellular infiltrate which is reminiscent of that found in patients who develop bronchiolitis obliterans after lung transplantation. Even though both lungs were fixed at the same airway pressure, the air spaces are more dilated in the denervated than in the contralateral lung, probably reflecting gas trapping in areas of airway obstruction.

View larger version (92K): [in this window] [in a new window]   Figure 5.   H&E stains (left) and in situ hybridization (1-wk exposure) to 35S-labeled rat tropoelastin (right, top) and human 1(I)-procollagen (right, bottom) cRNA probes in the denervated lung shown in Figure 4, showing increased procollagen I (but not tropoelastin) mRNA transcripts in the cells that occlude the lumina.

A cell-specific signal for tropoelastin mRNA was frequently present in the airway adventitia, pulmonary vessel walls, and ends and bends of the interalveolar septa. This signal was more intense in smaller rats, probably reflecting growth-related synthesis of elastin. However, we found no differences between denervated and nondenervated lungs in the intensity or distribution of the tropoelastin mRNA signal (Figure 6), independent of the weight of the rats.

View larger version (37K): [in this window] [in a new window]   Figure 6.   In situ hybridization (2-wk exposure) of denervated (right) and contralateral (left) rat lungs to a 35S-labeled tropoelastin cRNA probe in the rat shown in Figures 1 and 2. No differences in tropoelastin mRNA transcript density were detected between the denervated and nondenervated lung, despite the presence of specific hybridization localizing to the bronchial (b), arterial (a), and alveolar walls.

No specific hybridization signal was ever detected with the sense probes (data not shown).

Extent of Denervation

In most of the rats, the denervated lung responded to stimulation of the contralateral vagus with an increase in flow resistance (Table 3). Accordingly, and despite our efforts to eradicate vagal fibers, a substantial proportion of the parasympathetic outflow still reached the airways through the contralateral vagus. The responses were highly variable, however, as shown by the dispersion of the data.

Of the 28 rats that underwent resistance measurements during vagal stimulation after vagotomy, eight had increases in ₁(I)-procollagen mRNA. Four of these rats had undergone a unilateral vagotomy, and the remaining four had undergone a unilateral vagotomy and ipsilateral phrenectomy. Despite the inclusion of a disproportionate number of rats from the vagotomy-and-phrenectomy group in the sample selected for the resistance measurements, the proportion of rats that exhibited procollagen upregulation in the sample obtained from each group was similar to that of the group in general (chi-square analysis). An ANOVA showed no relationship between the magnitude of the response to contralateral vagal stimulation and the presence of increased procollagen mRNA in the denervated lung.

	Discussion

Top Abstract Introduction Materials & Methods Results Discussion References

Confirming our first hypothesis, the results presented here show that vagal denervation can indeed increase Type I procollagen (but not tropoelastin) gene expression in the submucosa and serosa of the airways and, surprisingly, in the adventitia of the pulmonary vessels. Even incomplete removal of pulmonary-bound vagal fibers produced detectable increases in procollagen mRNA and collagen deposition in these areas. Less frequently, lung denervation was also followed by a cell-proliferative response that occluded the airway and vascular lumina in a fashion resembling the pathologic findings in bronchiolitis obliterans after lung transplantation in humans (11). Contrary to the premises outlined at the beginning of this report, however, persistent increases in Type I procollagen mRNA transcripts were found after reducing the ventilation of the denervated lung with an ipsilateral phrenectomy. This finding undermines the notion that increased procollagen expression after lung denervation is initiated by increased airway strain (or, in light of the present results, by increased vascular strain).

Increased Procollagen Message after Denervation

The increase in ₁(I)-procollagen mRNA demonstrated by in situ hybridization in the denervated lungs in our study is probably the result either of increases in the rate of transcription or increased transcript stability, rather than of de novo expression of the Type I procollagen gene. Procollagen mRNAs were found in the same areas and, most likely, in the same population of cells in denervated and nondenervated lungs, even though in the latter the hybridization signal became very weak toward the periphery of the lung (as in the case of other investigators [17], we found no Type I procollagen mRNA in the interstitium of the non-denervated lungs). Although we could not conclusively identify the cells expressing the increased message, the distribution of these cells throughout the adventitia of airways and pulmonary vessels, the submucosa of the airways, and the interstitium of the lung parenchyma suggests that they were fibroblasts. Airway and vascular-wall fibroblasts have been implicated in other experimental and naturally occurring forms of local pulmonary fibrosis, such as those caused by silica inhalation in airways and pulmonary vessels of rats (17), or by alveolar hypoxia in pulmonary arteries of calves exposed to high altitude (6). The participation of fibroblasts in such processes is consistent with the idea that these cells have a highly unstable phenotype, which allows them to respond rapidly to chemical or mechanical signals (22).

The increase in procollagen mRNA was present in only 31% of the denervated lungs. The absence of a more universal response is one of the most intriguing aspects of our results, but the reasons for it unfortunately cannot be clarified with the information that we now have available. There are, however, several potential explanations for the observed variability. For instance, the time allowed between denervation and removal of the lungs may have been either too short or too long for the response to be manifest in all the rats. The similar proportions of rats showing increased levels of ₁(I)-procollagen message at 4 and 8 wk and the presence of increased procollagen expression as early as 2 wk after denervation argue against both possibilities, but the relatively small number of rats and the limited time frame of the experiments prevent us from excluding them completely. In addition, variability in procollagen expression between rats may simply reflect individual differences in the extent to which the lungs were denervated. The lack of concordance between the changes in flow resistance produced by stimulation of the contralateral vagus nerve in the denervated lungs and the level of Type I procollagen expression limits the credibility of this interpretation. However, it does not rule it out, because resistance measurements cannot localize the site of the airway constrictive response and, particularly after hilar denervation, the regional distribution of the surviving fibers within the lung may have varied substantially from one rat to another, possibly resulting in heterogeneous patterns of denervation. Moreover, because we did not perform the experiments in a genetically uniform sample of animals, the observed differences in the changes in procollagen expression induced by denervation may simply have been due to genetic variations among rats.

Mechanism of Type I Procollagen Upregulation

Production of Type I collagen and other matrix proteins by vascular fibroblasts can be induced by mechanical stimuli, such as the pulsatile distention of the pulmonary arteries. For instance, ₁(I)-procollagen message is absent from high-resistance, low-flow pulmonary arteries but is highly expressed in high-resistance, high-flow pulmonary arteries of chronically hypoxic calves (7). Although the mechanism by which mechanical stress is translated into a fibroproliferative response is not fully delineated, it is now known that certain growth-stimulating factors capable of increasing matrix-protein gene expression (23) display shear stress-responsive elements in their gene promoters. Specifically, the promoters for both platelet-derived growth factor-B (PDGF-B) and transforming growth factor-1 (TGF-1) contain a GAGACC core binding sequence that binds selectively to transcription factors present in nuclear extracts of bovine aortic endothelial cells exposed to shear stress (8). It is particularly interesting that PDGF immunoreactivity is increased in lung lavage fluid, alveolar macrophages, and bronchial epithelial cells from patients who develop bronchiolitis obliterans after lung transplantation (27).

The existence of well-established mechanisms for the transduction of mechanical stimuli into growth signals was the basis for the hypothesis that increased strain or airway deformation in the absence of parasympathetic control of airway tone would cause a fibroproliferative response in denervated airways. The absence of a decrease in the rate of upregulation of Type I procollagen expression after phrenectomy, however, suggests that increased airway strain is not the only factor in upregulating the expression of this protein after denervation. In our study, the rats subjected to phrenectomy displayed obvious asymmetries in ventilation, and a gross inspection of these animals' lungs before their removal frequently revealed areas of collapse. Consequently, and although it may not have completely abolished lung inflation, there is little doubt that phrenectomy accomplished its experimental purpose of reducing intrapulmonary stress and airway strain. Yet five of 12 rats subjected to phrenectomy (including the one shown in Figures 4 and 5) displayed increases in procollagen expression. This finding suggests that the neural supply to the lungs exerts an independent trophic effect on airways and pulmonary vessels. This suggestion is not without precedent. Specifically, neurotrophic factors produced by denervated tissue targets can cause hyperinnervation of surviving parasympathetic ganglia, and may even have paracrine effects leading to gene activation in non-neuronal cells. A superabundance of such factors has been implicated in the development of vascular hyperplasia in spontaneously hypertensive rats (28) and in normotensive rats treated chronically with nerve growth factor (29).

Summary

The observations presented here demonstrate for the first time that vagal denervation can influence extracellular- matrix-protein gene expression in airway and pulmonary vessels. Although the responsible signals remain unknown, this observation indicates that the neural supply to the lungs is important for the maintenance of airway and vascular-wall integrity. Deposition of increased amounts of collagen in denervated lungs is in itself likely to impair airway and vascular adaptability under conditions that require increased ventilation or pulmonary blood flow, respectively. Further progression of the fibrosis and, especially, the development of cell proliferation, can ultimately lead to airway and vascular obstruction. In this regard, the similarities between the alterations found in some of the denervated rat lungs in our study and in the lungs of humans who develop bronchiolitis obliterans after lung grafting are provocative, raising the possibility that denervation plays a role in the development of this common and frequently lethal complication of lung transplantation.