Lung Tissue Compliance Interacts with Smooth Muscle and Drives Hyperresponsiveness in Mice with Experimental Asthma

Abstract

Introduction

A recent study on BALB/c and C57BL/6 mice demonstrated a clear lack of association between the in vivo response to nebulized methacholine and the degree of airway narrowing ex vivo in a model of asthma induced by a daily exposure to house dust mite over 10 consecutive days. This finding raises the question of which factors determine the methacholine response in vivo.

Methods

Herein, multiple linear regression analyses were used to determine which baseline physiological characteristics are associated with the methacholine response.

Results

Among the 10 baseline characteristics studied, and depending on how the methacholine response was monitored during a concentration–response, lung tissue compliance was the most commonly and robustly associated with the methacholine response. Inspiratory capacity was the second most frequently associated.

Conclusion

These results suggest that lung tissue compliance and inspiratory capacity may be two important determinants of the methacholine response in BALB/c and C57BL/6 mice with and without experimental asthma.

Introduction

The mechanisms underlying hyperresponsiveness to inhaled methacholine in asthma remain incompletely understood. An increased contractile capacity of the airway smooth muscle, leading to excessive airway narrowing, has been proposed as a contributing factor. Yet, the science-based evidence for that is rather weak [1].

Comparative physiology across mouse strains can provide enlightening insights about the root causes of methacholine hyperresponsiveness in asthma. The invasive techniques to assess respiratory mechanics in mice are particularly useful, as they provide a level of mechanistic details that surpasses what is attainable in humans [2].

A recent study investigated two mouse strains with markedly different responses to methacholine [3], together with divergent susceptibilities for developing hyperresponsiveness in models of asthma [3–5]. More precisely, the C57BL/6 and BALB/c mice were studied with or without experimental asthma induced by a daily exposure to house dust mite (HDM) for 10 consecutive days [6]. Their in vivo response to incremental doses of nebulized methacholine were then analyzed on day 11, in conjunction with the assessment of ex vivo airway narrowing in lung slices and the histological quantification of airway smooth muscle remodeling. The results demonstrated that while the BALB/c mice with experimental asthma were clearly hyperresponsive to nebulized methacholine in vivo, they exhibited neither excessive airway narrowing ex vivo nor enlargement of the airway smooth muscle [7].

If methacholine responsiveness is not determined by the extent of airway narrowing, what other factors might influence the different response of these two mouse strains in this asthma model?

The present study seeks to identify the baseline physiological characteristics (i.e., prior to the methacholine challenge) associated with the in vivo response to methacholine observed in this previous study. Building on other previous findings showing that the hyporesponsive C57BL/6 mice exhibit stiffer lungs than the hyperresponsive BALB/c mice [8], we hypothesized that the lung tissue compliance is tightly associated with the methacholine response.

Methods

Twenty-four male BALB/c mice and 24 male C57BL/6 mice between eight and ten weeks of age were studied. All procedures were approved by the Committee of Animal Care of Université Laval (2020-652-4). Mice were exposed once daily to intranasal saline or HDM for 10 consecutive days to induce experimental asthma [9]. On day 11, before the methacholine challenge, ten baseline characteristics (variables) were measured. These included: (1) respiratory system resistance (R_rs), respiratory system elastance (E_rs), tissue elastance (H), tissue resistance (G), airway resistance (R_aw), and hysteresivity (η) measured by oscillometry; (2) quasi-static elastance (E_st), and lung tissue compliance (K) measured from a partial pressure–volume maneuver; (3) inspiratory capacity (IC) measured from a deep inflation maneuver; and (4) excised lungs weight. Detailed methodologies were recently described [7, 8]. Notably, mouse strain was also initially included as a covariate in our analyses, and it was, as expected, a strong predictor of the methacholine response. Yet, it was subsequently removed, as it provides no hints as to which physiological mechanisms are driving the methacholine response in these two mouse strains with and without experimental asthma.

Mice were then challenged with nebulized methacholine with increasing concentrations (0, 10, and 30 mg/mL). The changes in oscillometric readouts (∆R_rs, ∆E_rs, ∆H, ∆G, ∆R_aw & ∆η) from baseline values to peak following each concentration of methacholine were used to trace the concentration–response curves. Methacholine response was then quantified as the area under the curve (AUC) for each parameter. Parenthetically, the oscillometric response to methacholine in mice are typically greater than the one observed in humans [10], as the maximal concentration (30 mg/mL) used herein represents a near lethal concentration for HDM-exposed male BALB/c mice.

A two-way ANOVA was used to assess the effect of HDM, the mouse strain, and their interaction on each measured baseline characteristic. When the interaction was significant, it was followed by a Tukey’s multiple comparisons test and the significant differences between saline- and HDM-exposed mice within each mouse strain, as well as the significant differences between mouse strains within each exposure group, are shown.

Multiple linear regression analyses were then performed to determine the baseline physiological characteristics associated with the methacholine response. Since the methacholine response was measured by tracking the changes in six different readouts (AUC ∆R_rs, AUC ∆E_rs, AUC ∆H, AUC ∆G, AUC ∆R_aw & AUC ∆η), six distinct models were generated. The following steps were thus repeated six times, once per readout used to monitor the methacholine response. The data from all mice from both strains were included in each model. First, given the strong interrelationships among several characteristics, the Least Absolute Shrinkage and Selection Operator (LASSO) regression was applied. This approach effectively identified the most relevant characteristics while addressing collinearity issues among them. The regularization parameter lambda (λ) was optimized for minimizing the squared mean of the prediction error using a tenfold cross-validation. The value of λ for each model is depicted in Table 1. We additionally tested an alternative LASSO model selection approach using the Bayesian Information Criterion, which yielded similar results to those obtained with cross-validation, further supporting the stability of the selected characteristics. The characteristics selected by the LASSO regression were then analyzed by multiple linear regression to assess their associations with the methacholine response. Variance Inflation Factors (VIF) were finally calculated in the final multivariable model, and a threshold of 5 was used to confirm the absence of collinearity among the characteristics chosen by LASSO. Statistical analyses were performed using STATA, version 14.0, and a p value under 0.05 was considered significant.

Table 1.

Table showing the p value for each independent variable selected by the LASSO regression in each of the six distinct multivariate linear regression models

	Model 1 AUC ∆Rrs	Model 2 AUC ∆Ers	Model 3 AUC ∆H	Model 4 AUC ∆G	Model 5 AUC ∆Raw	Model 6 AUC ∆η
LASSO λ value	2.2	73.2	35.0	4.1	3.1	0.21

Independent variables	p values of multivariate linear regression (β coefficient [95% CI])
Treatment	0.011 (41.1 [10.0,72.2])	0.001 (757.3 [313,1202])	0.003 (340.6 [119,562])	0.002 (173.9 [67.8,280.0])	0.059 (18.0 [−0.72,36.6])	0.091 (1.49 [−0.25,3.23])
Lung weight	0.09 (47.3 [−656,751])	0.44 (3863 [−6092,13819])	0.21 (3123 [−1830,8078])	0.99 (−6.2 [−2568,2556])
Rrs				0.135 (157.3 [−51.2,365.9])
H						0.40 (−0.28 [−0.94,0.38])
G	0.49 (−5.3 [−20.6,10.1])			0.286 (−32.1 [−92.1,27.9])
η					0.65 (3.74 [−12.8,20.3])	0.17 (1.39 [−0.61,3.38])
Est	0.16 (7.2 [−2.8,17.2])			0.077 (29.7 [−3.3,62.8])
K	0.014 (1349 [282,2415])	0.001 (19,173 [7809,30536])	0.02 (6769 [1115,12423])	0.004 (5385 [1832,8937])		0.23 (65.8 [−44.2,176])
IC	0.020 (272.5 [45,499])	0.041 (2371 [101,4641])	0.10 (940.2 [−189,2069])	0.008 (1079 [296,1861])	0.044 (123.1 [3.24,243])	0.89 (1.57 [−20.2,23.3])

Empty cell signifies that this independent variable was not selected by the LASSO regression. Independent variables not shown in the table, namely E_rs and R_aw, were not selected by the LASSO regression in any models

P values in bold are statistically significant

AUC Area under the curve CI confidence interval E_rs respiratory system elastance E_st quasi-static elastance G tissue resistance H tissue elastance IC inspiratory capacity LASSO least absolute shrinkage and selection operator K lung tissue compliance R_aw airway resistance R_rs respiratory system resistance η hysteresivity

Results

Body weight was not different between strains and between mice with and without experimental asthma. The lung weight was also similar between strains, but lungs from mice with experimental asthma were heavier. The other baseline characteristics are shown in Fig. 1. All readouts of lung stiffness (E_rs, H & E_st) or its inverse, compliance (K), differed significantly between strains (all p < 0.0001), confirming that lungs of C57BL/6 mice are stiffer than those of BALB/c mice. Mouse strain also affected inspiratory capacity (IC) and tissue resistance (G), with IC being lower and G being higher in C57BL/6 compared to BALB/c mice. Experimental asthma decreased tissue compliance (K), and concordantly increased respiratory system elastance (E_rs) and tissue elastance (H). Experimental asthma had no significant effect on the remaining baseline characteristics. There were also significant interactions between mouse strain and experimental asthma for R_rs, E_rs and H, indicating that the increases caused by experimental asthma were observed in the hyporesponsive C57BL/6 but not in the hyperresponsive BALB/c mice for each of these baseline characteristics. For the methacholine response, as expected, BALB/c were more responsive than C57BL/6 mice in terms of changes in all parameters (R_rs, E_rs, H, G, R_aw, and η). Also, within each strain, but especially in BALB/c, HDM-exposed mice were more responsive than saline-exposed mice in terms of changes in R_rs, E_rs, H, G, and R_aw. The results (not shown here) were recently published [6].

Fig. 1.

Parameters of respiratory mechanics and lung volumes in BALB/c (open circles) and C57BL/6 mice (solid squares) exposed to either saline or house dust mite (HDM). Displayed from left to right and from the upper to the lower row are respiratory system elastance (E_rs), tissue elastance (H), quasi-static elastance (E_st), tissue compliance (K), inspiratory capacity (IC), tissue resistance (G), respiratory system resistance (R_rs), airway resistance (R_aw), and hysteresivity (η). Results of two-way ANOVAs are underneath each graph. When the interaction was significant, a Tukey’s multiple comparisons test was conducted, and asterisks indicate significant differences between saline- and HDM-exposed mice within each mouse strain, and between mouse strains within each exposure group (*p < 0.05, **p < 0.01 & ****p < 0.0001). N = 12

The results of the multiple linear regression analyses are presented in Table 1. For four out of the six different readouts used to monitor the response to methacholine (AUC ∆R_rs, AUC ∆E_rs, AUC ∆H, and AUC ∆G), the treatment (saline vs. HDM) was independently associated with the methacholine response, implying that these readouts are sensitive to the effect of experimental asthma on the methacholine response. For the same four readouts, K was also independently associated with the methacholine response. Finally, IC was independently associated with the methacholine response when the response was monitored through changes in R_rs, E_rs, G, and R_aw.

Discussion

The results demonstrate that, among all baseline characteristics studied and depending on how the methacholine response was measured, lung tissue compliance and the inspiratory capacity were the most commonly and robustly associated with the methacholine response in BALB/c and C57BL/6 mice with and without experimental asthma. The results also demonstrate that all indicators of stiffness (E_rs, H & E_st) or its inverse, compliance (K), were markedly different between BALB/c and C57BL/6 mice (Fig. 1). This confirms our previous study in naïve mice, showing that lungs of C57BL/6 are stiffer than BALB/c mice not because of smaller lungs but truly due to a stiffer lung tissue associated with an increased collagen content [8]. Notably, the surface tension was not compared in that latter study, which will be required in future studies owing to its important impact on lung compliance [11]. The results presented herein still extend our previous findings [8] by showing that this striking interstrain difference in lung stiffness is conserved with experimental asthma.

The reason why IC was associated with the methacholine response is uncertain. It is possible that with bigger lungs, a greater amount of methacholine entered the lungs during the nebulized challenge [12]. The fact that IC was associated even when the response was monitored by measuring the changes in airway resistance (R_aw), an indicator of airway narrowing, would support this conjecture. This is because the biggest fraction of nebulized methacholine entering the lungs is initially deposited in large airways [12]. Therefore, with a greater amount of methacholine getting into the lungs, this would imply that greater concentrations of methacholine are achieved in large airways, which would then further airway narrowing and amplify the changes in R_aw. The link between IC in the methacholine response will obviously need to be further investigated.

The likely link between K, an indicator of lung tissue compliance [13, 14], and the response to nebulized methacholine seems more obvious. In our opinion, it is due to a reduced load impeding the shortening of the airway smooth muscle [15]. Indeed, with a more compliant lung tissue, it should be easier to stretch the lung parenchyma attached on the outer edge of the airway wall during bronchoconstriction, which should then lead to greater airway narrowing for any given smooth muscle force [15]. It is also likely that this extra narrowing would be potentiated in an asthmatic environment. This is because the constriction of the airway lumen is amplified by oedematous airway wall and the luminal accumulation of inflammatory, exudative, and mucosal fluids [16]. Finally, and most importantly, a stiffer tissue may force the lungs to work more homogeneously and thereby prevents hyperresponsiveness by protecting against airway narrowing heterogeneity and closure, which are the main causes of hyperresponsiveness in asthma and experimental asthma [17–26].

Importantly, among all variables that are highly linearly correlated, LASSO typically selects only one, effectively simplifying the model by eliminating some variables. K was highly, and expectedly, correlated with E_rs, H, and E_st. Indeed, K being a volume-independent indicator of lung tissue elastance [13, 14], the lung feature it represents is somewhat embedded in any indicator of lung stiffness (E_rs, H, and E_st). The fact that LASSO selected K instead of these other colinear characteristics is interpreted herein as a sign that K better defines the feature of the lungs that is best associated with the methacholine response. Yet, it is acknowledged that K can be substituted by E_rs, H, or E_st in our analyses without losing much of the models predictability. This essentially means that although K was selected in our analyses, lung stiffness in general was strongly associated with the methacholine response, irrespective on how it was measured or the specific parameter that was used for appraising it (K, E_rs, H, or E_st).

Another important limitation in our multivariate analyses is that the significant effect of stiffness was driven by mouse strains. This group effect is consistent with our two previous studies [6, 8], showing high lung elastance and low methacholine responsiveness in C57BL/6 mice and, inversely, low lung elastance and high methacholine responsiveness in BALB/c mice. However, it also implies that what we are reporting herein are mere associations. Any other features than lung stiffness that are strikingly different between these two mouse strains may potentially explained, partially or totally, their markedly different levels of response to methacholine.

We thus proposed that the greater lung tissue compliance of BALB/c mice predisposes to excessive airway narrowing because it reduces the load impeding smooth muscle shortening, which may then yield a catastrophic response to methacholine when combined with other ‘asthmatic’ lung alterations (Fig. 2). Inversely, the stiffer lung tissue of C57BL/6 mice should protect against airway narrowing by increasing the load impeding smooth muscle shortening, thereby mitigating the methacholine response, especially in the context of experimental asthma.

Fig. 2.

The catastrophic response to nebulized methacholine in BALB/c mice with experimental asthma stems from a more compliant lung tissue, conspiring with inflammatory obstruction during smooth muscle shortening to promote excessive luminal airway narrowing. Spring thickness represents parenchymal lung stiffness. Mucus/inflammation, lamina propria and muscle are yellow, grey, and black, respectively

It is worth mentioning that human asthma is not typically associated with altered lung tissue compliance. Exceptions exist though. More persistent and severe asthma, often in the elderly, is sometimes associated with increased lung compliance [27–32]. Perhaps the BALB/c would be a good choice of mouse strains for studying hyperresponsiveness in this subgroup of asthmatics, or in subgroups presenting emphysema, such as in patients with an asthma-COPD overlap syndrome. Nonetheless, an important cause of hyperresponsiveness in BALB/c mice with experimental asthma may be irrelevant to hyperresponsiveness in most asthmatic individuals. This also suggests that even if a specific trait, such as hyperresponsiveness, resembles human asthma in a mouse strain, the underlying causes are not necessarily the same.

Author Contributions

All authors edited the manuscript, and read and approved the final manuscript. ARR, MB, CH, and YB contributed to the development of the experimental design. ARR, MB, and CH performed laboratory experiments. SH conducted the multiple linear regression analyses.YB wrote the first version of the manuscript.

Funding

Natural Sciences and Engineering Research Council of Canada (NSERC, RGPIN-2020-06355 and ALLRP-570485-2021), Canadian Institutes of Health Research (CIHR, 508356-202209PJT), and the Fondation de l’IUCPQ (Institut Universitaire de Cardiologie et de Pneumologie de Québec).

Data Availability

No datasets were generated or analysed during the current study.

Declarations

Competing Interests

The authors declare no competing interests.