Assessment of Left Lung Remodeling With Magnetic Resonance Imaging in a Murine Model Following Exposure to Douglas Fir Smoke

Abstract

Wildland firefighters (WLFFs) experience lung function decline due to occupational exposure to fire smoke. WLFFs typically do not wear respiratory personal protective equipment, and if they do, it is a simple bandana, which is not effective at filtering smoke. To pinpoint the biological underpinnings of abnormal respiratory function following 3–7 years of WLFF service, we exposed mice to Douglas fir smoke (DFS) over 8 weeks. Following exposure, we assessed changes in lung structure through Magnetic Resonance Imaging (MRI) and histological analysis, which was supported by immunohistochemistry staining. With MRI, we found that the signal decay time, T₂^*, from ultrashort echo time (UTE) images was significantly shorter in mice exposed to DFS compared to air controls. In addition, the variation in T₂^* was more heterogeneously distributed throughout the left lung in DFS-exposed mice, compared to air controls. As confirmed by histological analysis, shorter T₂^* was caused by larger parenchyma airspace sizes and not fibrotic remodeling. Destruction of the alveolar spaces was likely due to inflammation, as measured by an influx of CD68+ macrophages and destruction due to enhanced neutrophil elastase. In addition, measurements of airspace dimensions from histology were more heterogeneously distributed throughout the lung, corroborating the enhanced relative dispersion of T₂^*. Findings from this study suggest that the decline in lung function observed in WLFFs may be due to emphysema-like changes in the lung, which can be quantified with MRI.

Introduction

A growing population lives in the wildland-urban interface regions, resulting in an increasing number of people being exposed to wildland fire smoke (WFS). Moreover, with a continuous accumulation of organic materials in fire-prone regions, the extent, severity, and frequency of destructive wildland fires have increased in recent years. This is expected to continue to worsen in the forthcoming years [1–3]. Exposure to wildland fire smoke enhances the risk of respiratory and cardiovascular mortality, as evidenced by epidemiological data [4–6]. WLFFs are at an elevated risk as they work long hours at sites with high smoke concentrations. In addition, WLFFs rarely employ respiratory personal protective equipment (PPE). Some WLFFs wear bandanas to try and mitigate exposure, however, bandanas provide no protection from the smoke [7]. This amplified risk experienced by WLFFs has been supported by data highlighting a decline in lung function, inflammation of the airways, increased systemic inflammation, and arterial stiffening [8,9].

Rodent models of WFS exposure can provide a detailed understanding of the biological underpinnings of smoke exposure. Immediate respiratory and cardiovascular abnormalities, including changes in inhalation patterns [10] and blood pressure [11], as well as increased inflammation [12], have been reported. Recently, we have outlined biomechanical changes of the respiratory and cardiovascular system in a mouse model representing 3–7 years of wildland firefighting service [13]. Notably, we observed that airway flow resistance was larger, the aorta was structurally stiffer, and ejection fraction was reduced in Apoe^-/- mice exposed to DFS for 8 weeks [13]. This work expands upon these previous findings [13] to include more detailed insights into the structural changes of the lung, as quantified with MRI and supported by histological and immunohistochemistry data.

Magnetic resonance imaging of the respiratory system is challenging due to the many air–tissue interfaces in the lungs. These interfaces cause a rapid decline in signal intensity, which is characterized by the T₂^* relaxation time. T₂^* is extremely short (typically <1 ms, depending on the magnetic field strength), resulting in little to no signal in typical imaging scenarios (e.g., spin-echo imaging). Furthermore, lung movement during respiration can make it difficult to acquire high-resolution images that are free from motion artifacts. To overcome these challenges, UTE imaging techniques may be utilized [14–16]. In UTE sequences, data are collected from the center of k-space outward, often in a radial sampling pattern, with the acquisition occurring immediately following the RF excitation phase. This strategy overcomes the very short T₂^* relaxation time in the lungs.

T₂^* may also provide information about tissue characteristics, as there is evidence that T₂^* is shorter in lungs with alveolar air space enlargement or pulmonary fibrosis [17,18]. It can be advantageous to measure T₂^* by performing a UTE imaging sequence that implements multiple echoes. To tease out if remodeling is due to fibrosis or emphysema, histological analysis needs to be performed in conjunction with MRI, to quantify airspace size and collagen deposition. The objective of this study was to explore the utilization of multi-echo UTE imaging [16] for quantification of tissue remodeling following DFS exposure in a mouse model representing 3–7 of wildland firefighting service.

Methods

Douglas Fir Smoke Exposures.

Details of DFS exposures, including smoke characterization, are provided in Ref. [13]. Briefly, a custom exposure system designed based on DIN 53436 [19] was utilized to produce, characterize, and expose mice to the smoke. Douglas fir needles were sourced from Missoula, MT, and were dried prior to being placed in the exposure apparatus.

All experiments involving animals received approval from the Northeastern University Animal Care and Use Committee (IACUC) and complied with National Institute of Health (NIH) guidelines. Male Apoe^-/- mice on a C57BL/6 background were purchased from The Jackson Laboratory (Bar Harbor, ME) at 8 weeks of age. For exposures, mice were placed in a custom-designed whole-body exposure chamber, located downstream of the exposure apparatus [13], for 2 h/day, 5 days/week, over 8 weeks. The mice were exposed to either fresh air (n = 8) or DFS at 22 mg/m³ particulate matter and 139 ppm carbon monoxide concentrations (n = 9) [13]. These exposure conditions are expected to result in a total retained mass of 1.5 g of particulate matter in the lungs, modeling 3–7 years of wildland firefighting service [13].

Magnetic Resonance Imaging Data Collection.

Magnetic resonance imaging experiments were conducted on a 7.0 Tesla scanner with a 35 cm horizontal bore, and images were captured with Paravision 6.0.1 software (Biospec 70/30 & B-GA12S2 gradients, Bruker Biospin, Germany). A mouse body coil was utilized. Mice were anesthetized with inhaled isoflurane (2.5–3% for induction, 0.5–1.5% for imaging). Respiratory rate, temperature, and ECG were continuously monitored (SA Instruments Inc., NY).

An in-house developed radial three-dimensional-UTE MR sequence with a Golden Means trajectory ordering scheme, based on previously published work by Guo et al. [15], was used to collect images of the mouse lung parenchyma. The sequence consisted of a 4.3-μs RF excitation pulse to acquire 52,800 projections (64 points/projection) at five different echo times = [0.09, 0.15, 0.2, 0.35, 0.5] ms, repetition time = 6 ms, flip angle = 5.3°, field of view = 32 mm × 32 mm × 64 mm, voxel size = 0.25 mm × 0.25 mm × 0.50 mm, receiver bandwidth = 277 kHz.

Magnetic Resonance Imaging Data Reconstruction.

Offline image reconstruction and respiratory gating were performed in Matlab (MathWorks, MA) [15]. Theoretical trajectories obtained with a phantom (50 ml falcon tube with Gadolinium-doped agarose) were used for trajectory gradient delay compensation. End-expiration and end-inspiration data were generated using postacquisition respiratory gating following the methods proposed by Stecker et al. [16].

Magnetic Resonance Imaging Data Analysis.

Following image reconstruction, exhalation Signal Intensity (SI) images were converted to T₂^* maps by solving the following equation:

$$\mathrm{SI}=A_0{e}^{-\mathrm{TE}/T_2^{\mathrm{*}}}$$ (1)

where TE is a vector of the echo times, and A₀ is an unknown parameter that represents the signal intensity when TE is zero. T₂^* was calculated for each voxel by linear regression of the log of Eq. (1). Only signal intensities larger than three times the standard deviations of the background noise were utilized for the fits. Background noise was determined by taking the average SI of the voxels outside of the mouse. In addition, only voxels with a coefficient of determination greater than 0.75 were used for statistical analysis.

Regions of interest (ROI) encompassing the left lung were drawn on each transaxial slice in which the left lung was visible, resulting in about 20 slices analyzed. The right lung was not analyzed to avoid artifacts originating from lobe fissures. The average and standard deviation of T₂^* were calculated over the left lung, and the relative deviation (RD) was calculated as the standard deviation normalized by the average T₂^*. RD was chosen as the metric to assess heterogeneity because the means of T₂^* of the DFS exposed and air control groups are different, and therefore, solely examining standard deviation would not pinpoint heterogeneous differences [20].

Histology and Immunohistochemistry of Lung Tissue.

Following imaging, the lungs were perfusion fixed at a tracheal pressure of 20 cm H₂O with 3% glutaraldehyde in Ringer's solution (120 mM NaCl, 5.0 mM KCl, 0.15 mM CaCl₂) for 40 min or until the lungs were firm to the touch. Prior to being transferred to 70% ethanol for long-term storage, the lungs were stored in the glutaraldehyde solution for at least 24 h at 4 °C. For histology, the left lung and right apical lobe were dehydrated and sectioned above, at, and below the main stem of the bronchi. These sections were paraffin-embedded and serially sliced into 5 μm thick sections. For histology, slides were stained with MOVAT or Picrosirius Red (PSR) stains. Immunohistochemistry was performed with the following antibodies: neutrophil elastase (1:500, Abcam ab68672), or CD68 (1:500, Cell Signaling Technology, 97778) and counterstained with H&E. MOVAT and immunohistochemistry slides were imaged at 40× magnification with a EasyScan Infinity microscope (Motic, Kowloon Bay, Hong Kong). PSR slides were imaged at 40× objective with an upright microscope (DM4 B, Leica microsystems) with both brightfield and polarized light.

Parenchyma Structural Analysis.

Custom-built MATLAB codes were utilized to quantify parenchyma airspace dimensions and microstructure composition [21]. Airspace diameters (D₀) and their standard deviation (D₁) were calculated with the regionprops function in MATALB from black and white images. D₀ was determined from the average of the major and minor axis length for each identified centroid.

Microstructural constituents were automatically analyzed by setting the pixel thresholds for Hue, Saturation, and Lightness by utilization of custom MATLAB scripts (modified from Ref. [22]). Elastin and cell cytoplasm area fractions were identified from the MOVAT-stained slides. Collagen fractions were determined by normalizing the total number of birefringent pixels from polarized light images by the total tissue area in the brightfield images. High-density and low-density collagen were specified based on the red and orange or yellow and green pixels, respectively, from the polarized images. Image analysis was performed on the airways and parenchyma separately.

Immunohistochemistry Data Analysis.

Similar to the microstructural composition analysis, the area fractions of neutrophil elastase area were calculated by thresholding for pixel Hue, Saturation, and Lightness pixel thresholds. The analysis was performed on parenchyma tissue for neutrophil elastase. CD68+ macrophages were counted in ImageJ utilizing the Color Deconverter 2 plugin [23] to separate the stains into separate channels and to count CD68+ macrophages by thresholding. The number of CD68+ macrophages was counted via the analyze particles tool in ImageJ and normalized by the image area to provide a fraction of macrophage content.

Statistical Analysis.

An unpaired two-tail t-test was performed to compare values for D₀, D₁, neutrophil elastase area fraction, and CD68+ macrophages. An unpaired Wilcoxon test was used to assess statistical differences for T₂^* and RD because the sample size per group was 8–9 data points. Prism 10 was utilized to create box plots, and R version 4.3.2 was used for the statistical analysis. Data for T₂^* and RD are presented as averaged values across the left lobe for each mouse. For histology and immunohistochemistry, data for each region analyzed (apical, middle, base of the lobe) were treated as separate data points.

Results

Magnetic Resonance Imaging Signal Decay Behavior.

The MRI signal decay time constant, T₂^*, was significantly shorter in DFS-exposed mice compared to air control mice (Fig. 1(a)). Representative T₂^* maps of air control (Fig. 1(c)) and DFS-exposed lobes (Fig. 1(d)) highlight the shorter T₂^* observed in the smoke-exposed mice. Furthermore, the heterogeneity of T₂^*, as quantified by RD, was significantly larger in the DFS exposed, compared to air control, mice (Fig. 1(b)).

Fig. 1.

MRI signal decay time constant, T₂^*, (a) and relative dispersion (RD), (b) of the mouse left lungs following 8 weeks air (control, gray squares) and DFS exposure (red triangles). Representative T₂^* maps of a transaxial slice of the left lung from air control (c) and DFS-exposed (d) mouse. The same color bar is used for both (c) and (d).*p < 0.05 comparison for DFS exposed and air controls using unpaired Wilcoxon tests.

Parenchyma Structural Changes.

Parenchyma airspace diameters (D₀) were significantly larger in DFS-exposed mice, compared to air control mice (Fig. 2(a)). Enlarged airspaces can be observed in the representative CD68+ (Figs. 3(b) and 3(c)) and neutrophil elastase (Figs. 3(e) and 3(f)) images. In addition, the distribution of D₀ was more heterogeneous in the DFS exposed, compared to the air control mice (Fig. 2(b)). While we did analyze the apical, middle, and base of the lung, we did not observe any regional differences in D₀. Elastin and cytoplasm tissue fractions were not significantly different for either the airways or the parenchyma between DFS exposed and air control mice. Similarly, red/orange (R/O) collagen (e.g., high-density collagen) was not different between the air control and DFS-exposed mice for both the airways and the parenchyma. In contrast, area fractions of yellow/green (Y/G) collagen for the airways were not statistically significantly smaller in DFS exposed (0.049 ± 0.013) compared to air control (0.062 ± 0.020, p = 0.066). In the parenchyma, Y/G tissue fractions were significantly smaller in the DFS exposed (0.004 ± 0.004) compared to the air control (0.007 ± 0.006, p < 0.05) exposed mice.

Fig. 2.

Equivalent airspace diameter (D₀, panel (a)) and standard deviation of airspaces (D₁, panel (b)) for air control and DFS-exposed mice. Note the differences in axis range in panels (a) and (b). *p < 0.05 for comparison of DFS exposed and air controls using unpaired t-tests.

Fig. 3.

Immunohistochemistry of CD68+ stained cells per image area ((a)–(c): representative images) and neutrophil elastase fraction ((d)–(f): representative images). *p < 0.05 for comparison of DFS exposed and air controls using unpaired t-tests.

Immunohistochemistry.

There were significantly more CD68+ macrophages, per unit area of the image, in DFS exposed compared to air control mice (Fig. 3(a); representative images from air control: Fig. 3(b); and DFS exposed: Fig. 3(c) mice). The area fraction of neutrophil elastase was significantly larger in smoke-exposed mice (Fig. 3(d)), as can also be observed in the air control and DFS exposed representative images (Figs. 3(e) and 3(f), respectively).

Discussion

Wildland fire smoke exposure has been shown to result in lung function decline and increased inflammation in wildland firefighters [8,9]. To uncover the biological underpinnings of changes in respiratory function, we performed in vivo mouse exposures that were designed to model 3–7 years of wildland firefighting service.

T₂^*, considered an imaging biomarker for assessment of lung emphysema and parenchyma airspace size [24–26], as well as fibrotic remodeling, was chosen to evaluate the effects of DFS in the lungs. The shorter left lung T₂^* observed in DFS-exposed mice compared to air control mice suggests an increase in parenchyma airspace size, as confirmed by analysis of parenchyma airspace size (Fig. 2(a)). Larger alveolar airspaces decreased the parenchymal density and presented larger air-to-tissue ratios, therefore, resulting in increased susceptibility gradients and shorter T₂^* within the alveoli. We further observed that there was an increase in lung T₂^* heterogeneity, assessed by the RD metric (Fig. 1(b)) in the DFS exposed group in comparison to the air control group, which is corroborated by the enhanced gradients due to larger airspaces and by the heterogeneous distribution of this remodeling [25,26]. This observation is supported by measurements of airspace size standard deviations obtained from histological images (Fig. 2(b)). Changes in airspace size and heterogeneity can be qualitatively viewed on the representative CD68+ (Figs. 3(b) and 3(c)) and neutrophil elastase (Figs. 3(e) and 3(f)) images. Enlarged airspaces were likely caused by inflammation, as evidenced by an influx of CD68+ macrophages (Fig. 3(a)), and elevation in neutrophil elastase (Fig. 3(d)), which is known to cause destruction of alveolar walls [27]. Note that changes in T₂^* were likely not attributed to fibrosis, as collagen fractions were not higher in the DFS exposed, compared to the air control mice.

In a previous study [13], we explored the functional changes in the lung following 8 weeks of exposure to DFS. With these global measurements, we observed increases in airway resistance with no change in tissue elastance or resistance. In an elastase model of emphysema, increased airspace dimensions resulted in softer lungs [28]. However, with DFS exposure, mice presented with thickened parenchyma septa, which likely compensated for the increase in airspace dimensions, resulting in no change in global lung stiffness. While not measured here, localized changes in lung strain as measured with digital image correlation [29] or computed tomography imaging [30] may enable for better correlation of structural and functional changes.

Ultrashort echo time MRI has been previously utilized to uncover changes in lung structure due to fibrotic [15,16] and emphysema [18,31] remodeling. Klotho knockout in mice and emphysema caused by elastase induction results in a severe pulmonary emphysema phenotype [18,31]. The klotho knockout emphysema resulted in approximately 25% reduction in T₂^* in a 3 T clinical MR scanner with mice that are expected to have 25% larger linear mean intercepts [32]. In the elastase model, a 53% increase in linear mean intercept resulted in a 24% decrease in T₂^* in a 4.7 T MR scanner [31]. The changes in airspace morphometry observed here were relatively minor (Fig. 2(a)), resulting in a 10% reduction in T₂^* in a 7 T small animal MR scanner. As previously reported, mice exposed to DFS for 8 weeks had 14.5% larger linear mean intercepts compared to air controls [13]. Therefore, in this study, we further highlight the sensitivity of using UTE MRI at high tesla strengths to uncover structural changes that occur in mild forms of emphysema that result from prolonged smoke exposure.

As we generated the smoke by burning Douglas fir needles, we were unable to study the individual contributions of the smoke components. As we previously reported [13], the particulate size count median diameter was 110 ± 20 nm, which is well within the respirable range. Polyols and phenols were the abundant materials in the particulates, while CO and CO₂ were the dominant gaseous species. In addition, we measured unburned hydrocarbons and alcohols due to incomplete combustion of the Douglas fir needles [13]. How each of these components contributes to the observed remodeling may be a topic of future studies. In addition, follow-up experiments may focus on uncovering the relationship between remodeling and dose by either changing the concentration of DFS or the duration of exposure.

While this study provides insight into structural changes that occur following smoke exposure, there are some MRI-related limitations. Coil homogeneity was not characterized in every imaging session, which might result in heterogeneity of the magnetic field. We attempted to reduce the susceptibility of UTE trajectories to gradient nonlinearities and field inhomogeneities with an ellipsoidal field of view, a Golden Means Ordering scheme, and an inert phantom for accurate trajectory measurements; however, it was not possible to fully eliminate their impact in the measurements. When segmenting the lung, we did not avoid the blood vessels, which might have influenced the calculated T₂^* values depending on the blood volume contained within each ROI.

In conclusion, mice exposed to DFS at concentrations and durations representative of 3–7 years of wildland firefighting service experienced structural remodeling of the lung parenchyma, as quantified by MRI analysis, histology, and immunohistochemistry. The elevation of neutrophil elastase and the influx of CD68+ cells likely contributed to the destruction of the parenchyma. These findings are supported by the decline in lung function observed in wildland firefighters [9], which may be attributed to emphysema-like remodeling of the lungs.

Funding Data

• Department of Homeland Security (DHS) Federal Emergency Management Agency (FEMA) (Grant No. EMW-2017-FP-00446; Funder ID: 10.13039/100008464).

• National Institute of Environmental Health Sciences at the National Institute of Health (Grant No. R01E5033792; Funder ID: 10.13039/100000066).

Data Availability Statement

The datasets generated and supporting the findings of this article are obtainable from the corresponding author upon reasonable request.

Nomenclature

DFS =

Douglas fir smoke

D₀ =

parenchyma airspace diameters

D₁ =

parenchyma airspace diameter standard deviation

MRI =

magnetic resonance imaging

PSR =

Picrosirius red

RD =

relative dispersion

ROI =

region of interest

SI =

signal intensity

T₂* =

transverse relaxation time

UTE =

ultrashort echo time

WFS =

wildland fire smoke

WLFF =

wildland firefighter