Influence of airway wall stiffness and parenchymal tethering on the dynamics of bronchoconstriction

Abstract

Understanding how tissue remodeling affects airway responsiveness is of key importance, but experimental data bearing on this issue remain scant. We used lung explants to investigate the effects of enzymatic digestion on the rate and magnitude of airway narrowing induced by acetylcholine. To link the observed changes in narrowing dynamics to the degree of alteration in tissue mechanics, we compared our experimental results with predictions made by a computational model of a dynamically contracting elastic airway embedded in elastic parenchyma. We found that treatment of explanted airways with two different proteases (elastase and collagenase) resulted in differential effects on the dynamics of airway narrowing following application of ACh. Histological corroboration of these different effects is manifest in different patterns of elimination of collagen and elastin from within the airway wall and the surrounding parenchyma. Simulations with a computational model of a dynamically contracting airway embedded in elastic parenchyma suggest that elastase exerts its functional effects predominately through a reduction in parenchymal tethering, while the effects of collagenase are more related to a reduction in airway wall stiffness. We conclude that airway and parenchymal remodeling as a result of protease activity can have varied effects on the loads opposing ASM shortening, with corresponding consequences for airway responsiveness.

when the smooth muscle (ASM) surrounding an airway is activated, the resulting active force generated within the airway wall acts so as to narrow the airway lumen at a rate determined by the force-velocity behavior of the muscle and the loads against which it has to contract. These forces are not constant, however. As soon as narrowing begins to occur, opposing elastic forces begin to increase as a result of distortions induced in the local parenchyma and in the airway wall (1, 6). Eventually, these forces will come to match the maximum (isometric) force that the muscle is capable of generating, at which point the airway cannot narrow further. Accordingly, changes in the elastic properties of both the airway wall and its parenchymal attachments, such as might result from pathological remodeling processes, have been implicated as important modulators of airway responsiveness (1, 6, 13, 15, 32). Furthermore, it has been hypothesized that airway remodeling could potentially change the mechanical properties of these structures, with commensurate effects on responsiveness.

Understanding how tissue remodeling affects airway responsiveness is of key importance for research into airway diseases such as chronic obstructive pulmonary disorder (COPD). Nevertheless, experimental data bearing on this issue, which is still debated (27), remain scant. This is probably because addressing the matter in vivo is complicated by the difficulty of producing a steady level of ASM activation in animal models (24). Also, the forces of parenchymal tethering vary substantially throughout the breathing cycle (32), so assessing their net effect in a ventilated animal is not straightforward. On the other hand, both these factors can be readily controlled in vitro in a slice of explanted lung (30) by exposing it to a steady concentration of smooth muscle agonist at a fixed degree of distension. This affords the possibility of determining how the rate and degree of contraction of an airway section in the explant is affected by interventions that affect the structural integrity of the parenchyma and the airway wall.

In the present study, therefore, we used lung explants to investigate the effects of enzymatic digestion on the rate and magnitude of airway narrowing induced by acetylcholine. To link the observed changes in narrowing dynamics to the degree of alteration in tissue mechanics, we compared our experimental results to predictions made by a computational model of a dynamically contracting elastic airway embedded in elastic parenchyma.

METHODS

Animal handling.

All experiments were approved by the Animal Research Ethics Board of McMaster University and carried out according to the guidelines of the Canadian Council on Animal Care.

Female Balb/c mice (6–8 wk old) were purchased from Charles River Laboratories (Montreal, PQ, Canada) and maintained under specific pathogen-free conditions in an access-restricted area, on a 12-h light-dark cycle, with food and water provided ad libitum. Animals were killed by CO₂ inhalation in a closed chamber for 1 min, and lungs were removed for use in histological or physiological studies as described below.

Preparation of reagents.

Cell culture reagents were obtained from Invitrogen Life Technologies - GIBCO (Carlsbad, CA). DMEM for slice incubation was supplemented with penicillin (10,000 U/ml), streptomycin (10,000 μg/ml), amphotericin B (125 μg/ml), l-ascorbic acid (35 μg/ml), transferrin (5 μg/ml), selenium (3.25 ng/ml), and insulin (2.85 μg/ml). HBSS in most cases was supplemented with HEPES buffer (0.2 M, pH 7.4). Type IV porcine pancreatic elastase and type II collagenase were dissolved in HBSS (1×) for a stock concentration of 5 and 10 mg/ml, respectively, divided into 25-μl aliquots, and stored at −20°C for further use. All reagents were obtained from Sigma Chemical (St. Louis, MO) unless specified otherwise.

Lung slicing.

Lungs were sliced using the protocol described by Perez and Sanderson (30), with some modifications (11, 21). Briefly, following euthanization, the chest wall was removed, the trachea cannulated using an intravenous catheter (20G Intima; Becton Dickinson, Sandy, UT), and the lungs inflated with ∼1.2 ml of 2% agarose (type VII-A low gelling temperature; Sigma Aldrich, St. Louis, MO) warmed to 37°C. Approximately 0.15 ml of air was subsequently injected to flush the agarose-HBSS out of the airways of interest (those which would be visualized) so that they would be free to contract and not subject to intraluminal impedance. The agarose was gelled by cooling the lung preparation to 4°C for 5–10 min, after which the left lobe was removed and sliced (120-μm thickness) using an EMS-4000 tissue slicer (Electron Microscope Sciences, Fort Washington, PA) at 4°C. The slices were washed in HBSS and inspected under a phase-contrast microscope: those slices that were not torn or otherwise damaged were then incubated overnight in DMEM supplemented with antibiotics and antimycotics at 37°C and 5% CO₂.

Elastase and collagenase exposure.

Following overnight incubation, some lung slices were treated with porcine pancreatic elastase (2.5 μg/ml), collagenase (2.5 μg/ml), or both: however, the latter combination essentially disintegrated the tissues. Therefore, we opted instead for lower concentrations when combining the two enzymes: 1 μg/ml elastase and 0.25 μg/ml collagenase. Slices pretreated only in DMEM were taken as control. After overnight incubation, slices were examined under phase-contrast microscopy, and airways with intact epithelium were selected for assessment of cholinergic responsiveness.

Image acquisition and measurement of airway kinetics.

Lung slices were transferred to fresh HBSS for 30 min. We selected airways that: 1) nearly filled the charge-coupled device (CCD) camera imaging area (∼100- to 125-μm outer diameter); 2) displayed an intact epithelium; 3) exhibited active epithelial ciliary beating; 4) were not plugged with agarose; and 5) had a short-to-long axis ratio of 0.5–1.0. Slices that met these criteria were mounted on a glass cover slip (45 × 50 mm; Fisher Scientific, Suwanee, GA) and held in position by a piece of nylon mesh (250-μm mesh; CMN-0250D, Small Parts, Miami Lakes, FL) as well as a second glass cover slip (22 × 40 mm). DMEM was superfused between the two cover slips and over the tissue slice at a rate of ∼4.5 ml/min. ACh was added/removed via this perfusate and had excellent access to the airway of interest, as indicated by a latency for onset of the cholinergic response of less than 5 s. Phase-contrast images were acquired with a high-resolution CCD solid-state video camera (CV-252, Nikon, Japan) and recorded in time lapse (33-ms exposure, 30 frames/min) using image acquisition software (Video Savant; IO Industries, London, ON, Canada). We have previously shown these video rates to be optimal for capturing the dynamics of airway contraction (21). Video images were transformed to recordings of airway area vs. time using Scion image analysis software (Frederick, MD), which converted the eight-bit video images to binary and then measured cross-sectional area of the airway in each frame by pixel summing. Peak velocities of contraction or of relaxation were calculated using a simple mathematical algorithm, described in more detail in our earlier publication (21), which numerically differentiates the values of area with respect to time (SigmaPlot; Systat Software, Point Richmond, CA). Statistical comparisons were made using SigmaPlot and Minitab software. We also quantified the extent of parenchymal detachment by superimposing a disk over the image of each airway of interest and integrating the arc of that disk overlying regions of parenchymal detachment.

Processing and staining of lung slices.

Thin lung slices (120 μm) pretreated with enzymes or DMEM alone were first fixed in 10% formalin for 24 h, processed in microbiopsy cassettes, and embedded in paraffin wax. Embedded slices were cut into 3- to 5-μm sections using a microtome (Microm HM 550) and stained using PicroSirius red (20) and Miller's stain (28) for collagen and elastin, respectively. An area of interest was identified as the region between the inner border of parenchyma and the outer border of the epithelium. This region was overlaid with a grid (10 μm × 10 μm), and the matrix integrity was quantified as the number of grid intercepts that lay over stained tissue per mm² (thus complete connectivity of either stain throughout this region would yield a count of 1,000). Counts per square millimeter were performed using Northern Eclipse software, and results were expressed in intercepts per square millimeter area.

Analysis and statistics.

Airway area was standardized as a percent of lumen area measured at the beginning of the experiment (immediately before challenge with the first bolus dose of ACh). Student's t-test was used for comparing variables between treatment conditions. One-way ANOVA was performed among different treated groups. P < 0.05 was considered as statistically significant.

Computational modeling.

Our computational model of a contracting airway embedded in elastic parenchyma has been described in detail previously (5, 6). Briefly, we model an airway in two dimensions as a circular ring of ASM wrapped around an elastic airway wall embedded in homogeneously elastic lung parenchyma. ASM contraction decreases airway radius by pulling against the parenchymal attachments to the outside of the airway wall. This outward pull comes from two sources: 1) the transpulmonary pressure (Ptp) that is transmitted across the parenchyma when it is undistorted (i.e., when it is uniform and isotropic), which is determined by lung volume under the assumption of a fixed tissue elastance, and 2) the local distortion of the parenchyma caused by narrowing of the airway, which is assumed to follow the relationship identified by Lai-Fook (23). The inward recoil of the airway wall is determined by its stiffness, which is assumed to arise from a fraction (1 − k) of the airway circumference that expands according to the one-third power of Ptp. The remaining fraction, k, of the circumference is assumed to be inextensible, where 0 < k < 1. Once activated, the ASM follows the classic Hill force-velocity relationship that is hyperbolic when active force (F_A) is less than isometric force (F₀), and linear when F_A ≥ F₀ with slopes matched at F₀ (18). Thus

$$\begin{array}{l}-2\pi \frac{dr}{dt}=\frac{b\left(F_0-F_A\right)}{a+F_A}\text{ when }{F}_A<F_0\\ =\frac{b{F}_0}{a+F_0}-\frac{b{F}_A}{a+F_0}\text{ when }{F}_A\ge F_0\end{array}$$ (1)

where r is airway radius, and a and b are constants. Following experimental findings reported in rats (7), we set a = F₀/4.

The balance of forces in the model is calculated using the Laplace law that assumes the wall to be thin. This is certainly not the case in reality, although it has recently been shown that even with a thick airway wall, the average pressure-radius behavior is quite well approximated by the thin-walled assumption (10). Using this assumption, we calculate the force, F_A, that adds to the outward recoil of the parenchyma and the inward recoil of the airway wall to give a net force difference of zero. The explicit expression for F_A that this produces is derived in Ref. 6, and is given by

$$\begin{array}{c}{F}_A=\left[r_{TLC}{\left(\frac{P_{tp}}{P_{tpTLC}}\right)}^{\frac{1}{3}}-0.3r\right]\frac{P_{tp}}{0.7}\\ -r{P}_{tpTLC}{\left[\frac{rk}{\left(1-k\right)r_{TLC}{\left(\frac{P_0}{P_{tpTLC}}\right)}^{\frac{1}{3}}}+\frac{r}{r_{TLC}}-\frac{k}{1-k}\right]}^3\end{array}$$ (2)

where r_TLC is the radius that the virtual hole occupied by the airway would have at total lung capacity (TLC) if it expanded like the rest of the parenchyma, P_tpTLC is P_tp at TLC, and P₀ is the value of P_tp at which the unconstricted airway induces no distortion in the parenchyma surrounding it. Note that the stiffness of the airway wall in this model does not include a contribution from the ASM itself, the stiffness of which has been shown to increase markedly during activation (25), because we are concerned here with the stiffness seen by the ASM due to the elastic structures upon which it acts.

We used the above equations to calculate how r varies with time (t) following activation of the ASM. Initially, the ASM was relaxed so that F_A = 0 and the relaxed airway radius, r₀, was determined by the force balance between the inward recoil of the airway wall and the outward recoil of the surrounding parenchyma. Activation of the ASM then gave rise to a finite value for F_A via Eq. 2. The result was substituted into Eq. 1 to provide dr/dt, which was then used to determine radius (r) at the next time step using first-order Euler integration. This new value of r was then used in Eq. 2 again to determine the next value for F_A, and so on, until a complete time profile of r was produced. The area of the airway normalized to its relaxed value was calculated as the ratio r/r₀.

The coupling of excitation to contraction in ASM occurs with a certain lag reflective of the sequence of events that are involved (33). We observed, however, that the explant started to respond within 5 s of application of agonist, demonstrating that the lag was short compared with the time-scale of the experiment itself. We therefore make the simplifying assumption in the model that the ASM becomes immediately activated to a fixed level as soon as the agonist is applied. The subsequent contraction dynamics are thus determined entirely by the force-velocity relationship of the ASM embodied in Eq. 1. The relaxation of the model following removal of stimulation is a somewhat different matter, however. The agonist was washed almost instantly from the bath in the experimental preparation when ACh was removed from the explant perfusate, but this does not mean that ASM activation would have disappeared with the same rapidity. The removal of ACh from its receptor on the ASM cell is linked to the cessation of active force through a series of intermediate steps that include Ca²⁺ resequestration, deactivation of myosin light chain kinase, and activation of myosin light chain phosphatase (36). Furthermore, the excitation-contraction coupling events involved in ASM relaxation have been shown to take longer than those involved in contraction (33). In addition, the parenchymal and airway wall tissues are viscoelastic (3), so that their return toward elastic equilibrium would not be instantaneous even if active ASM force were to suddenly become zero (33). We are not going to try to model all these various steps here because that would represent an inappropriate level of detail given the approximations made with regard to the rest of the model. We therefore approximate the dynamics of airway relaxation by assuming that F_A (Eq. 2) decays exponentially as soon as the agonist is washed from the bath. The time-constant (τ) of the force decay was chosen empirically to produce rates of airway relaxation similar to those observed experimentally. To find the value of r that balanced the forces in the model at each time step during relaxation, we increased r in small increments until Eq. 2 was satisfied using the current value of F_A.

Choosing the baseline value of Ptp for the model is not entirely straightforward. The explants were injected with a volume of agarose that would have inflated them close to total lung capacity. Prior to experimentation, however, the agarose in the lumen of the airway itself was removed. Also, it is not clear what effects slicing the lung would have on the state of inflation of the resulting slice, even with its alveolar regions still full of agarose. We therefore arbitrarily set the baseline Ptp value for the model to be 25 cmH₂O as a nominal representation of full lung inflation. To have the airway area traces generated by the model match those observed experimentally, we found by trial and error that suitable values for F₀ and b were 28.0 cm/cmH₂O⁻¹ and 0.20 cm/s⁻¹, respectively. On the basis of our previous studies with this computational model (4, 6, 13), the other baseline mechanical parameters were set to P₀ = 10 cmH₂O and k = 0.78. Finally, we chose the rate of model relaxation, again on purely empirical grounds, to be governed by τ = 300 s.

As the effective Ptp value for the experimental preparation was difficult to assess for the reasons given above, we performed additional simulations with the model Ptp set to 5 cmH₂O to explore its behavior at a state of inflation representative of normal functional residual capacity. For this lower Ptp, the values of F₀ and b had to be reduced to 5.0 cm/cmH₂O⁻¹ and 0.015 cm/s⁻¹, respectively. The other model parameters remained the same as before with the exception of τ which was set to 150 s.

RESULTS

Effects of enzyme treatments on resting airway lumen area.

In one set of tissues (n = 7), we assessed whether overnight enzymatic treatment was accompanied by luminal narrowing, which might reflect loss of tethering attachments and consequent wall retraction from the parenchyma. Images were taken before and after enzymatic digestion and were compared with those taken of tissues incubated with DMEM alone. Representative images are given in Fig. 1A; all data are summarized in Fig. 1B. There was a modest but statistically significant luminal narrowing in those tissues treated with porcine pancreatic elastase or Collagenase (PPE or Col, respectively; both 2.5 μg/ml). When both enzymes were applied together at these same concentrations, tissue integrity was completely lost. However, pretreatment with both enzymes at lower concentrations (1 μg/ml and 0.25 μg/ml, respectively, which allowed handling of these partially digested tissues) led to airway narrowing, which was significantly greater than that achieved with either enzyme alone.

Fig. 1.

A: changes in resting airway diameter before and after incubation in DMEM only (i), porcine pancreatic elastase (PPE) (ii), Collagenase (Col) (iii), and both PPE and Col (iv) (see text for concentrations). Arrows show detachment of airway wall from parenchyma. B: changes in airway lumen area post-enzyme treatments. Points (●) represent individual data; symbols (large ●) indicate mean data (± SE). *Significantly different from control; P < 0.05.

Effects of enzyme treatments on cholinergic responses.

Next, we examined the effect of enzymatic digestion on cholinergic-induced contractions (magnitude; kinetics), airway wall-parenchyma attachment, and wall connective tissue (content; distribution). Representative tracings from tissues preincubated with DMEM, PPE, or Col are given in Fig. 2; summary data are given in Fig. 3.

Fig. 2.

Raw tracings of airway narrowing induced by ACh (10⁻⁵ M; 5 min) as well as the relaxant response upon washout of ACh, in lung slices incubated overnight in DMEM, PPE, or Col (as indicated).

Fig. 3.

A: mean changes in airway area diameter in response to Ach (10⁻⁵ M; solid line) and after 5-min or 20- to 25-min washout (dotted lines). Mean velocities of contraction (B), relaxation (C), or the ratio of the two (D). *Significantly different from control; P < 0.05.

Contractile responses to a 5-min challenge with ACh (10⁻⁵ M) were significantly larger in tissues preincubated overnight with either PPE or Col (n = 5). Despite considerable tearing away of the airway wall from the parenchyma (described below), the airways nonetheless reopened towards their baseline diameters upon washout of ACh. However, 5 min was not sufficient time for the airways to fully dilate to their initial resting diameters. For this reason, we added a second group of tissues (n = 5) in which 25–30 min were given for relaxation following washout of ACh; over this longer duration, there was indeed further relaxation, although the airways still did not return to the diameters measured before cholinergic stimulation. For analysis of these two sets of data, we pooled the magnitudes and peak velocities of contraction (given that the experimental conditions for that part of the experiments were identical) as well as the peak velocities of relaxation (given that the peak velocity of relaxation occurred well within the first 5 min; typically within the first 1 min). The mean values of peak velocity of contraction (Vcon) were significantly greater in tissues pretreated with PPE or Col, and the mean value of peak velocity of relaxation (Vrlx) slower following treatment with PPE. Likewise, the pooled ratios of Vcon/Vrlx were dramatically and significantly greater in the proteolytically treated tissues compared with control. We have shown in our earlier publication (21) how this ratio is a particularly sensitive tool for revealing even subtle changes in wall stiffness, given that the latter changes are expected to produce opposite effects on Vcon and Vrel, e.g., a loss of wall matrix proteins will both remove a load which opposes contraction (thereby increasing Vcon) as well as decrease the restorative force that helps the wall spring back upon relaxation (decreasing Vrel).

We had also intended to compare the effect of combining both PPE and Col on airway contraction dynamics, but found that combined treatment left the tissues far too fragile to handle. When proteolytic digestion was partially limited by using lower enzyme concentrations, we found the airways could now constrict completely and significantly faster (Fig. 4); that is, to a greater extent than with either enzyme alone. More importantly, these airways could still relax, although with yet a slower Vrlx than the control tissues or those incubated with either enzyme alone (Fig. 4).

Fig. 4.

Raw tracing (A) and mean values of contraction magnitude (B), Vcon (C), and Vrlx (D) in tissues treated with both PPE and Col (1 μg/ml and 0.25 μg/ml, respectively); the control responses given in Fig. 3 are reproduced here for comparison. *Significantly different from control (P < 0.05).

Finally, we assessed the impact of enzymatic preincubation upon repeated stimulations with ACh, to evaluate whether the changes in magnitude and kinetics noted above were related to the process of parenchymal tearing per se or were a property of the detached airways themselves. Another set of tissues (n = 5) were pretreated overnight with PPE ± Col as described above, but were now challenged on the next day three times with ACh (10⁻⁵ M; 5 min application), separated by washout periods of 5 min (Fig. 5). The enzyme-induced changes in magnitudes and kinetics of the first cholinergic response were qualitatively similar to those described above. More importantly, though, we found these enzyme-induced changes were also reflected in the second and third cholinergic responses. That is, within a given experimental condition, the airways constricted to the same degree and at the same rate, and reopened to the same extent, each time; although it appeared that the resting diameter in the Col-treated tissues decreased progressively with each cholinergic stimulation (which might otherwise suggest incremental increases in the net level of parenchymal tearing; see next section), this trend was not statistically significant.

Fig. 5.

A: mean changes in airway area diameter in response to 3 consecutive challenges with ACh (10⁻⁵ M); data are expressed as a % of airway area immediately before the first challenge with ACh (i.e., after enzyme treatment). Also shown are the mean values of Vcon (B), Vrlx (C), or Vcon/Vrlx (D). *Significantly different from control; P < 0.05.

Structural changes caused by enzyme treatments.

The video recordings made in the experiments summarized above were reviewed to quantify the extent of detachment of the airway from the parenchyma (see methods for details). None of the tissues incubated with DMEM alone showed any such detachment. However, as summarized in Fig. 6, either PPE or Col caused considerable (∼180°) and statistically equivalent detachment, whereas the combined treatment with both enzymes was associated with even greater detachment (∼240°). Despite this massive detachment, the airways nonetheless reopened substantially (albeit more slowly) upon washout of ACh, as summarized above.

Fig. 6.

A: images of the airway before (i) and during stimulation (ii) with ACh (10⁻⁵ M), postincubation in DMEM only [a], [PPE] [b], [Col] [c], and both [PPE]+[Col] [d]. B: angle of detachment upon ACh stimulation. C: measurement of angle of detachment (θ) upon ACh stimulation. **Significantly different from control; P < 0.01.

A different set of tissues was incubated overnight with PPE ± Col, as described above, and was then fixed for histological analysis. In control (DMEM-treated) tissues, collagen was found distributed throughout the parenchyma and around the airways, while elastin was found primarily around the airways. Collagen staining was still present in those slices that had been treated with PPE but was completely eliminated in those treated with 2.5 μg/ml Col (Fig. 7A); interestingly, in those tissues treated with 0.25 μg/ml Col (plus PPE), collagen was eliminated from the parenchyma but was still present around the airways. Likewise, elastin staining was lost in those tissues treated with either 2.5 or 1.0 μg/ml PPE (Fig. 7B). We also quantified changes in matrix network intercepts per square millimeter area of the region between the epithelium and surrounding parenchyma (Fig. 7C), finding these to be significantly decreased in all three enzyme treatment groups.

Fig. 7.

Histological localization of collagen (A) and elastin (B) networks around airways in tissues incubated with DMEM, PPE, Col, or both PPE+Col. Collagen stains orange/red bands against background while elastin colors in deep purple bands. Arrow points to respective fiber distributions. C: measurement of matrix density in respective treated groups in terms of intercepts per area. *Significantly different from control (P < 0.05).

Model simulations.

The top panel of Fig. 8 shows model airway area vs. time during application and washout of agonist when the baseline Ptp of the model was 25 cmH₂O. The bottom panel shows a corresponding plot for the lower Ptp of 5 cmH₂O (see methods for the remaining baseline model parameter values in each case). The baseline simulations in the two plots exhibit contraction and relaxation time courses similar in shape and relative magnitude to those found experimentally in the explants (Figs. 2 and 4). When airway wall stiffness was reduced (k was reduced to 0.60), the degree of airway narrowing was increased, but the rates of contraction and relaxation (scaled by amplitude) were not greatly affected. In contrast, when Ptp was reduced by 3 cmH₂O (to 22 cmH₂O in the top plot and 2 cmH₂O in the bottom plot), the rate of contraction was increased while the rate of relaxation was reduced. Finally, with a decrease in both Ptp and wall stiffness together, the responsiveness of the model was increased in both rate and magnitude. Note, however, that not all features of the model response were the same at the two Ptp levels. In particular, greater airway narrowing occurred at 25 cmH₂O when Ptp was reduced compared with when wall stiffness was reduced. The effect was reversed, however, at 5 cmH₂O.

Fig. 8.

Simulations of airway area (normalized to baseline) vs. time generated with the computational model. The top shows simulations at a transpulmonary pressure (Ptp) of 25 cmH₂O, whereas the bottom shows corresponding simulations at 5 cmH₂O. To match the experimental tracings shown in Figs. 2 and 4, the model airway was maintained at baseline for the first 300 s and then constricted maximally for the next 300 s. Thereafter, the level of ASM activation was reduced exponentially.

DISCUSSION

Tissue remodeling in the lung has the potential to affect the mechanical loads opposing ASM shortening. These mechanical loads arise principally from two distinct sources: the viscoelastic impedance of the airway wall itself, and the outward pull exerted by the parenchymal attachments to the outside border of the wall. Experimental evidence suggests that both loads are significant and exert measurable influences on the time course of ASM shortening in situ (1, 2, 6, 9, 12). Nevertheless, the role of remodeling on airway responsiveness continues to be debated (27). On the one hand, fibrotic changes in the airways might be expected to stiffen the wall, making it more difficult for ASM to narrow the lumen. This would constitute a protective effect of remodeling and is most readily invoked relative to excess collagen deposition. On the other hand, there is no guarantee that altered airway collagen is going to stiffen the airway; even if there is more collagen than normal, abnormal organization of the collagen fibers might make the airway less resistant to deformation than normal, resulting in enhanced responsiveness. A similar question arises as to the effects of remodeling on the mechanical nature of the parenchymal attachments that tether the airway wall. Remodeling also has potential geometric significance for airway responsiveness through its ability to change airway wall thickness; an inner thickening of the wall might, for example, lead to an accelerated rate of luminal narrowing even when the rate of ASM shortening is normal (35). The potential effects of airway remodeling on responsiveness are thus numerous and disparate, and greatly in need of elucidation.

The complex link between tissue remodeling and phenotype in the lung also makes it difficult to study in vivo. For this reason, we chose here to examine the effects of remodeling in vitro using a lung explant system in which the dynamic behavior of a single airway can be studied under highly controlled conditions. We further chose to impose a very specific kind of remodeling on this airway, that caused by enzymatic digestion, because this is readily recapitulated in the laboratory. Enzymatic digestion can also be argued to be a major feature of COPD (31, 34). Of course, airway remodeling is a major feature of a number of other chronic lung diseases including asthma (29) and pulmonary fibrosis (22), so our experimental preparation may have limited applicability to remodeling in general. On the other hand, the remodeling processes involved in these other diseases are still poorly understood and likely involve numerous factors, so by focusing on enzymatic digestion we are able to link functional changes in the airway to a known pathological process. Even so, we do not know exactly where in the airway the proteases had their effects. They may, for example, have affected the connective tissues associated with the smooth muscle cells themselves, but presumably this would still have manifest as a change in overall wall stiffness.

Our experimental system thus trades off simplicity and transparency against the fact that it does not fully recapitulate the in vivo situation. Of course, it still has a number of key factors that we cannot control for. In particular, there is the possibility that enzymatic treatment of tissue slices may have affected the number of cholinergic receptors in the explant, which would have then affected its responsiveness to ACh. However, as far as we are aware, this has not been reported to occur in the numerous studies that have used enzymes to liberate single cells for patch-clamp studies, fluorimetry, cell culturing, etc., likely due to the very large number of spare muscarinic receptors (only a small percentage of the cholinergic receptors are needed for full activation of the muscle). Another possible effect of enzymatic treatment is an alteration in the mechanical stiffness of the agarose within the lung slices. However, we previously showed no change in the magnitudes or rates of contractions when the concentration of agarose was halved (21). All things considered, then, we believe that the lung explant allows us to directly study the functional effects of remodeling on airways responsiveness.

We hypothesized that enzymatic digestion would weaken the integrity of structures opposing ASM shortening, thereby increasing the responsiveness of an explanted airway to challenge with methacholine. Others have shown digestion using collagenase increased the magnitudes of cholinergic contractions in human tissue strips from 2nd to 4th order bronchi (9), but the airways used in that study do not have parenchymal attachments, nor did they look at the effect of digestion on the time course of contraction. In our study, we did indeed find that either PPE or Col increased the degree of luminal narrowing induced by ACh compared with control, while the two proteases in combination had a substantially greater effect than either enzyme individually (Figs. 1 and 3). This constitutes direct evidence that the mechanical loads opposing ASM shortening in the explant were reduced. By further focusing our attention on the time course as well as the degree of airway narrowing, we found additional evidence of the effects of PPE and Col on the mechanical loads opposing ASM shortening. In particular, the velocity of contraction was significantly increased following treatment with either collagenase or with elastase (Fig. 3B), and the velocity of relaxation slowed following treatment with elastase (Fig. 3C); the ratio of these two changes was significantly increased for both conditions (Fig. 3D). Treatment with both proteases together resulted in an even more precipitous decline in area in response to cholinergic stimulation (Fig. 4).

The reproducibility of airway narrowing was somewhat different following treatment with the two proteases. Repetitive exposure to ACh of the control explants and the explants treated with PPE produced quite consistent patterns of airway narrowing and dilation (Fig. 5), even though there was some detachment of the airway from the surrounding parenchyma in the latter case. By contrast, while the collagenase-treated explants also exhibited cyclic narrowing and dilatation, this was superimposed on a progressive move toward smaller areas (Fig. 5) suggestive of a plastic deformation of the airway that did not fully recover when the ASM was deactivated. Our computational model does not behave in this manner, so the explanation for these findings must lie in a mechanism not included in the model. Our interpretation is that collagenase caused the tissue to become fragile, so that when the airway wall contracted inward it caused some of the parenchymal attachments to break. This would then have reduced the outward elastic forces on the airway wall, rendering them unable to fully dilate the airway when the muscle relaxed. The two proteases thus seem to have exerted different effects on the mechanical loads opposing ASM shortening, which is perhaps not surprising considering that the PPE-treated explants retained collagen but eliminated elastin, while in the collagenase-treated explants the situation was reversed (Fig. 7). The two proteases are thus both qualitatively and quantitatively different in the way they affect the dynamics of airway narrowing.

Others have shown stimulation of protease-activated receptors by serine proteinases (such as trypsin and thrombin, but can also include elastase and collagenase) leads to direct relaxation of airways due to release of PGE₂ from the epithelium (9, 12). In contrast, we found enzymatic digestion enhanced constrictor responses (the latter were larger and/or faster). We point out, however, that the enzyme activities that we used in this study are orders of magnitude lower than those reported to exert a direct relaxant effect. That is, the previous publication (12) reported no direct relaxant effect of trypsin (a nonspecific protease) or thrombin at concentrations less than 0.01 U/ml, whereas we used highly specific proteases (specific for collagen or elastin, with only trace amounts of tryptic, clostripain, or neutral protease activities) at concentrations below that level (0.1 or 2.5 μg/ml, at 1–5 U/mg).

PPE and collagenase both reduced the forces of parenchymal tethering, as evidenced by the visible retraction of the airway following overnight incubation with either PPE or collagenase in the absence of cholinergic stimulation (Fig. 1). Subsequent challenge with ACh caused further airway narrowing of the airway that, in protease-pretreated tissues, proceeded to complete closure in some cases (Figs. 3–5). This was accompanied by extensive tearing of the remaining parenchymal attachments compared with control tissue (Fig. 6). These observations give clear indication of parenchymal tissue damage and confirm the functional importance of these forces, in accordance with in vivo studies showing that transpulmonary pressure is a major modulator of airway responsiveness (2, 4, 6, 15).

Our data also provide clear evidence of the presence of elastic recoil forces within the airway wall itself. In particular, even those airways that lost most of their parenchymal attachments were able to reopen to a substantial degree following removal of agonist (Figs. 2–6). Moreover, regardless of which proteolytic pretreatment regimen was used, the extent of airway reopening was quite reproducible (Fig. 5). This implies that the relaxed airway has some inherent resting diameter to which it will return in the absence of extraneous forces, in agreement with previous in vitro (26) and computational modeling (1, 6) studies. There are numerous structures within the airway wall that could potentially contribute to the maintenance of its equilibrium diameter. One obvious possibility is the network of structural proteins that give the wall its mechanical stability, but even the ASM cell itself may contribute. We have previously demonstrated complete reelongation of individual enzymatically dissociated cells challenged repeatedly with ACh and other bronchoconstrictor stimuli (19), perhaps as a result of the structural rigidity of the microtubule network that makes up the cytoskeleton.

Our experimental observations thus show that the mechanical loads opposing ASM shortening can be differentially regulated by attacking the integrity of collagen and elastin in the airway wall and surrounding parenchyma. Linking these observations to the actual mechanical properties of these structures, however, is not obvious merely from visual inspection of the data. In fact, the translation of ASM force into the dynamics of airway narrowing is a rather complicated function of geometry, force balance, and the ASM force-velocity relationship. Therefore, to help us relate these factors to our experimental observations, we made use of a previously developed computational model that combines relatively straightforward equations for each of these phenomena into a representation of a circular thin-walled airway contracting against the opposing loads of airway wall stiffness and parenchymal tethering. This model predicts the dynamics of airway narrowing in vivo rather accurately (6, 13) and allows us to investigate the consequences of reducing wall stiffness (one of the model parameters) on the rate and magnitude of airway narrowing. We can also examine how reduced parenchymal tethering affects narrowing by reducing the inflation pressure (Ptp) applied to the model. The simulated area-time plots in Fig. 8 show that reducing these two loads is predicted to have qualitatively different effects on the time course of narrowing, even though these effects vary with changes in Ptp. In particular, reducing parenchymal tethering by reducing Ptp increases the rate of airway narrowing but has only a modest effect on its overall magnitude. Conversely, reducing wall stiffness increases the degree of narrowing, particularly at 5 cmH₂O (Fig. 8), but has relatively little effect on the rate of narrowing adjusted for magnitude. Reducing both effects increases both the rate and magnitude of narrowing.

We also extended the computational model to include a mechanism for airway relaxation based simply around the assumption that washing ACh from the explant induces an exponential decay in ASM force. This is certainly a highly simplified account of the rather complicated set of events involved in the excitation-contraction coupling in ASM (10, 33). Also, although our model deals with the dynamic balance of forces during muscle contraction, it does not specifically account either for the way that cross-bridge attachment is affected by strain rate (2, 3, 16) or for the force-length behavior of airway smooth muscle (8, 17). Nevertheless, it produces airway dilation profiles reminiscent of the sigmoidal profiles observed experimentally (Fig. 2). It also predicts that the initial rate of dilation is more sensitive to loss of parenchymal tethering (reduced Ptp) than to reduction of airway wall stiffness, particularly at the lower Ptp. This prediction is borne out in the measurements made following PPE vs. collagenase treatment (Fig. 3C). The implications are therefore that PPE treatment predominately reduces parenchymal tethering, while collagenase has more of an effect on the stiffness of the airway wall. This fits with the notion that collagen is the main contributor to airway wall stiffness, while the elastic recoil provided by elastin in alveolar walls provides most of the parenchymal tethering force. Curiously, however, our histological observations indicate that collagen in the parenchyma surrounding explanted airways was eliminated by collagenase but not by PPE (Fig. 7), so the roles of collagen and elastin are not so readily separated.

These findings have important pathophysiological implications, particularly for COPD in which proteolytic enzyme activity in the lung is dysregulated and a mixed pattern of fibrosis and emphysema develops throughout the lung (14). These remodeling events have the potential to affect airways responsiveness in a variety of ways depending on their nature and distribution. Connective tissue proteins laid down in the fibrotic regions may cause stiffening of the airways and thus oppose airway constriction, while emphysematous changes predispose to airway collapse. This heterogeneity of functional abnormality might also be expected to lead to variable efficacy of bronchodilators in COPD. In more fibrotic regions of the lung where parenchymal tethering is increased, a decrease in airway stiffness following β-agonist administration would be expected to lead to airway opening and improved ventilation. By contrast, bronchodilation of emphysematous regions would likely make the airways tend to collapse, leading to gas trapping and decreased ventilation.

In summary, we have shown that treatment of explanted airways with two different proteases results in differential effects on the nature of airway narrowing dynamics following application of ACh. Histological corroboration of these different effects is manifest in different patterns of elimination of collagen and elastin from within the airway wall and the surrounding parenchyma. Simulations with a computational model of a dynamically contracting airway embedded in elastic parenchyma suggest that elastase exerts its functional effects predominately through a reduction in airway wall stiffness, while the effects of collagenase are more related to a reduction in parenchymal tethering. We conclude that airway and parenchymal remodeling as a result of protease activity can have varied effects on the loads opposing ASM shortening, with corresponding consequences for airway responsiveness.

GRANTS

We acknowledge financial support of the Canadian Institutes of Health Research (FRN 81134), AstraZeneca International, and the National Institutes of Health (Grants R33-HL-087788, HL-087401, and NCRR COBRE P20-RR-15557).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the author(s).