2.1.

Animal Procedures

All procedures involving animals were approved by the Institutional Animal Care and Use Committee. All animals were treated in accordance with federal and state regulatory guidelines. Seventeen New Zealand White rabbits (nine controls and eight scrape-injured) weighing $3–5\mathrm{kg}$ (Western Oregon Rabbit Company, Philomath, Oregon) were anesthetized, intubated [see Fig. 1 ], and maintained under anesthesia as previously described²³ with the following differences: $0.2\mathrm{mL}$ maintenance anesthetic was administered by injection every $5\min$ , rather than continuously. The tracheal scrape procedure was performed as previously described.¹¹ Briefly, rabbits were scraped on three occasions—every other day for five days—and were then sacrificed three days after the final scrape. Some animals were sacrificed immediately after the first scrape in order to distinguish between immediate and long-term effects of the injuries. A $7\text{-}\mathrm{mm}$ -diam unsheathed cytology brush [Conmed, Utica, New York; see Fig. 1] was inserted through the endotracheal tube and scraped over the entire length of exposed tracheal lumen between the endotracheal tube opening and the carina $(\sim 3\mathrm{cm})$ . The brush was scraped against the tracheal mucosa for $\sim 15\mathrm{s}$ . Cells on the brush were shaken into $1\mathrm{mL}$ of phosphate buffered saline (PBS) containing 4% formaldehyde and were subsequently stained with Alexafluor 568 phalloidin (Invitrogen, Carlsbad, California) and DAPI (Sigma, St. Louis, Missouri). Phase contrast and epifluorescence microscopy confirmed the presence of clusters of columnar ciliated epithelial cells (data not shown), as well as erythrocytes in the brushings. Rigid bronchoscopy confirmed placement of the endotracheal tube and gross appearance of the tracheal lumen, which appeared ulcerated and edematous following the second and third scrape injuries. Optical coherence tomography (OCT) was also performed on the tracheal mucosa as previously described,²³ confirming development of mucosal edema $10\min$ following the brush injury. Following the scrape injuries and recovery periods (or similar time period for control animals), the rabbits were euthanized as described previously.²³

Fig. 1

Experimental schematic and low-magnification images of unwounded and scrape-wounded tracheas. (a) Diagram showing endotracheal intubation of an anesthetized rabbit that allowed nonsurgical access to the distal $3\mathrm{cm}$ of the tracheal mucosa and submucosa for scrape wounding. (b) Dissecting microscope view of an excised portion of a scraped trachea, cut along the trachealis to expose the lumen, plus the cytology brush used to create the scrape wound. H&E stained $5\text{-}\mu \mathrm{m}$ -thick circumferential sections of (c) control and (d) scrape-wounded tracheas. H&E stained $5\text{-}\mu \mathrm{m}$ sagittal sections (transverse to cartilage rings) of (e) control and (f) scrape-wounded tracheas. Mucosal thickness is indicated by solid black brackets. d, u, and h represent areas of denuded, ulcerated, and hyperplastic epithelium, respectively. (g) Quantification from circumferential histological sections of morphological changes after chronic scrape wounding. Scrape-wounded mucosa and wall cross-sectional area are significantly greater than control areas (t-test, $p$ value $<0.001$ and $p$ value $<0.01$ , respectively). Lumen cross-sectional area is significantly decreased ( $t$ test, $p$ value $<0.001$ ). ${}^{*}$ , †, and ‡ indicate statistical significance relative to corresponding control. Control $N=3$ rabbits; scrape $N=6$ rabbits.

2.2.

Multiphoton Imaging

Rabbit tracheas were excised following euthanasia and stored in $50\mathrm{mL}$ of ice-cold PBS until removal for imaging $(\sim 2–3\mathrm{h})$ . In addition, six control tracheas from nonintubated, noninjured rabbits were cut into strips encompassing one-third of the tracheal circumference and subjected to $1\mathrm{mg}∕\mathrm{mL}$ CLSPA collagenase (Worthington, Biochemical Corporation, Lakewood, New Jersey) in PBS; $4.5\mathrm{mg}∕\mathrm{mL}$ ESL elastase in $0.2\mathrm{M}$ Tris-HCl buffer, pH 8.8; or no enzyme for a period of $24\mathrm{h}$ . Imaging and mechanical testing of these enzyme-treated and control tracheas commenced after several rinses in fresh PBS, and MPM images of these control tracheas were found to be indistinguishable from tracheas imaged at earlier time points. All tracheas were cut along the trachealis [for example, see Fig. 1] and then into $\sim 4\times 4\mathrm{mm}$ rectangular sections. Tracheal sections were sandwiched between two circular no. 1 borosilicate coverglasses (Fisher, Hampton, New Hampshire) and maintained in a hydrated state during imaging. All imaging was performed on a Zeiss 510 Meta multiphoton laser scanning microscope (Zeiss, Jena, Germany). SHG and TPF signals were collected in the epiconfiguration with an Achroplan $40\times ∕0.8\mathrm{NA}$ water-immersion objective, and $12\text{-bit}$ images contained $512\times 512\text{pixels}$ , with pixel size $0.44\mu \mathrm{m}$ . A circularly polarized Chameleon laser (Coherent Incorporated, Santa Clara, California) was focused on the trachea samples at wavelength $920\mathrm{nm}$ with incident power $\sim 60–175\mathrm{mW}$ (laser power was increased for more highly scattering samples). SHG signal was collected using a $390–465\mathrm{nm}$ infrared-blocking bandpass filter; TPF signal was collected simultaneously in a different channel using a $500–500\mathrm{nm}$ infrared-blocking bandpass filter. Two scans were averaged per frame with a pixel dwell time of $1.60\mu \mathrm{s}$ . Three to ten serial depth sections with $2–10\mu \mathrm{m}$ vertical spacing between individual frames (“z stacks”) were recorded per trachea section, with at least $225\mu \mathrm{m}$ between stacks. Additionally, 25 individual frames were tiled together laterally to form one large $1.125\times 1.125\mathrm{mm}$ tiled image for control and scrape-injured trachea sections.

2.3.

Mechanical Testing

Indentation testing was performed using a $1.3\text{-}\mathrm{mm}$ -diam cylindrical platen to indent samples at a rate of $0.05\mathrm{mm}∕\mathrm{s}$ up to $0.1\mathrm{mm}$ . After multiphoton imaging, the trachea sections were placed in PBS and transported to a Synergie 100 testing system (MTS Systems Corporation, Eden Prairie, Minnesota). Samples were placed on 600-grade ultrafine waterproof sandpaper (3M, St. Paul, Minnesota) during the test. A $1.3\text{-}\mathrm{mm}$ -diam platen was placed just touching the tracheal mucosa, and stress and strain were recorded for a $100\text{-}\mu \mathrm{m}$ indentation. The Young’s modulus in compression $\left(E\right)$ was calculated as the best-fit slope of the resulting stress-strain curve.

2.4.

Histological Procedures

Standard procedures were used to prepare histological stains of tracheal sections. Tracheas were paraffin embedded and sectioned in $5\text{-}\mu \mathrm{m}$ -thin sections. Sections were prepared parallel and perpendicular to the long axis of the trachea. Masson’s trichrome, Verhoeff-van Gieson (VVG), and hematoxylin and eosin (H&E) stains were performed on serial tissue sections in order to analyze collagen, elastin, and cell content of the tissues. Additionally, some tracheas were flash frozen in OCT medium (Ted Pella Incorporated, Redding, California), and then sectioned on a cryotome to $20\text{-}\mu \mathrm{m}$ sections. These sections were sufficiently thick to image with MPM after thawing in PBS. Following MPM imaging, these sections were stained in a fashion similar to the $5\text{-}\mu \mathrm{m}$ histology sections.

2.5.

Image Analyses

All image analysis was performed with ImageJ (Wayne Rasband, NIH, Bethesda, Maryland), and $10\times$ magnified brightfield images of histologically stained circumferential sections were manually thresholded to distinguish tissue from the background. Various tissue and layer edges were traced in order to determine average areas of the tracheal lumen, the tracheal wall, the entire wall plus lumen, and the mucosa only (i.e., tissue inward of the tracheal rings). Five control and six scrape-injured trachea sections from separate rabbits were analyzed in this way.

$200\times$ magnified bright-field images of Masson’s trichrome and VVG-stained trachea sections were analyzed to estimate image area occupied by collagen and elastin, respectively. RGB images of Masson’s trichrome images were split, and the blue channel, solely, was analyzed. The blue image, containing blue-stained collagen, was manually thresholded, and the entire mucosa from cartilage to lumen was traced. The fraction of blue-stained pixels was then determined. Raw RGB images of VVG-stained sections were inverted and thresholded in order to separate the elastin (originally stained black) from tissue stained with a lighter shade. Regions of tissue containing elastin were then traced, and the percent of elastin-containing pixels was measured.

Four parameters were quantified from SHG and TPF images of the tracheas: SHG and TPF signal area fractions, and the segmented and raw SHG signal intensity depth decay coefficients. Single images from z stacks containing the brightest SHG and TPF signal, respectively, were collected from each trachea sample and pooled for analysis. Each image was manually thresholded to determine the cutoff between noise and void-containing pixels and signal-containing pixels, and the area fraction of signal was determined. SHG signal intensity decay coefficients were calculated from an exponential best fit of z-stack image intensities plotted against depth from the luminal surface. All images were noise subtracted (determined from the image intensity of a z-stack frame positioned below the tissue surface). Images were further thresholded as before to isolate signal-containing pixels, and the segmented mean of these signal-containing pixels were also plotted as a decaying function of tissue depth. The SHG decay coefficient is defined as $b$ in the following best-fit equation: SHG $\text{intensity}=A{e}^{-bz}$ where $z$ is depth from the tissue surface in millimeters and $A$ and $b$ are constants determined by the best fit.

2.6.

Statistical Analysis

Student’s t-test was used to distinguish significant differences between measurements from control and scrape-injured histology sections. Mean $E$ of control, scrape-injured, collagenase-treated, and elastase-treated tracheas were compared to a single analysis of variance (ANOVA) test, followed by a post hoc Tukey test. A $p$ value of 0.05 or less was considered significant. Correlations of $E$ with TPF area fraction, SHG area fraction, raw SHG decay, and segmented SHG decay were performed with linear best fits, and $R^2$ values were determined. All error bars represent standard deviation, unless otherwise stated.

3.1.

Histological Comparison of Uninjured and Scrape-Injured Tracheas

Injured tracheas developed whitish granulation tissue and increased mucus, easily removed by gentle rinsing postsacrifice, compared to controls. H&E stained circumferential sections [Fig. 1] and sagittal sections [Fig. 1] of injured tracheas display marked fibrosis and mucosal thickening [Figs. 1 and 1, brackets] with regions of denuded epithelium [Fig. 1, labeled d] and ulcerated tissue [Fig. 1, labeled u] neighboring hyperplastic epithelium [Fig. 1, labeled h]. The airway luminal area is decreased, and the mucosal border is irregular [Fig. 1]. In comparison, circumferential and sagittal sections of uninjured tracheas [Figs. 1 and 1 respectively] display thin subepithelial connective tissue [Fig. 1 and 1, brackets]. Higher magnification images of VVG-stained sections of uninjured tracheas show unbroken and normal differentiated epithelium [best seen in Fig. 2 , e]. Quantified from circumferential sections, mucosal area increased $\sim 150\mathrm{\%}$ ( $6.9\pm 2.0{\mathrm{mm}}^2$ versus $2.5\pm 1.0{\mathrm{mm}}^2$ for injured and control, respectively, $p$ value $<0.001$ ); area of the tracheal wall increased $\sim 50\mathrm{\%}$ ( $13.1\pm 3.3{\mathrm{mm}}^2$ versus $8.7\pm 1.9{\mathrm{mm}}^2$ for injured and control, respectively, see $p$ value $<0.01$ ); and lumen cross-sectional area decreased $\sim 25\mathrm{\%}$ ( $13.0\pm 1.9{\mathrm{mm}}^2$ versus $17.0\pm 1.6{\mathrm{mm}}^2$ for injured and control, respectively, $p$ value $<0.001$ ); while total area of the cross sections were similar ( $26.1\pm 2.0{\mathrm{mm}}^2$ versus $25.7\pm 1.5{\mathrm{mm}}^2$ for injured and control, respectively, no significant difference, $p$ value $=0.68$ ).

Fig. 2

Comparison of histology and MPM images of trachea sections. Masson’s trichrome stain and Verhoeff-van Gieson (VVG) stain of control trachea (a, b, respectively) and scrape-wounded trachea (e, F, respectively) $5\text{-}\mu \mathrm{m}$ -thick sections. Coregistered SHG and TPF images of control trachea (c, d, respectively) and scrape-wounded trachea (g, h, respectively) from $20\mu \mathrm{m}$ cryosections. bv, lr, e, f, n, L, and u represent blood vessel, lamina reticularis, epithelium, fibrosis, neutrophil, lymphocyte, and ulcerated epithelium, respectively. Black and yellow arrows represent damage to the elastic fiber network in the lamina reticularis and a blood vessel, respectively. (i) Quantification of area covered by collagen (assessed from trichrome stain and cryosection SHG) and by elastic fibers (assessed from VVG stain and cryosection TPF). Collagen cross-sectional area is significantly increased in the scrape-wounded tracheas (t-test, $p$ value $<0.0001$ and $p$ value $<0.01$ , from trichrome and SHG images, respectively). ${}^{*}$ and † indicate statistical significance relative to corresponding control). Control $N=3$ cryosections from three rabbits; nine histological sections from three rabbits; scrape $N=5$ cryosections from five rabbits; and 12 histological sections from four rabbits. Scale bars are $50\mu \mathrm{m}$ .

Damaged blood vessels, fibrin, and primarily lymphocytic infiltration are visible at high magnification [Fig 2]. Scrape-injured sections contained more than a three-fold increase in immune cells—mostly polymorphonuclear neutrophils [Fig. 2 inset, labeled n], and lymphocytes characterized by round heterochromatic nuclei [Fig. 2 inset, labeled L; $47\pm 9$ and $162\pm 52$ cells/ $400\times$ field of view for control and scrape-injured tracheas, respectively, two-tailed t-test, $p$ value $<0.001$ ]. Masson’s trichrome and VVG stains of serial sections revealed the nature of mucosal collagen and elastin [Figs. 2, 2, 2, 2]. In uninjured tracheas the lamina reticularis (labeled lr) forms a thin $(30–80\mu \mathrm{m})$ layer of dense, interwoven collagen [Fig. 2] and elastic fiber [Fig. 2] networks. Beneath the lamina reticularis, sparse subepithelial collagen surrounds small blood vessels [Figs. 2 and 2 labeled bv]. In injured tracheas the organization of the lamina reticularis is lost, with extensive fibrotic collagen (labeled f) deposited apically near the site of injury [Fig. 2] concurrent with distortion of the collagen network and damage to the elastic fiber network in the lamina reticularis [Fig. 2, damage indicated by arrow].

Multiphoton images of $20\text{-}\mu \mathrm{m}$ -thick frozen [Figs. 2, 2, 2, 2] sections revealed similar information as histology sections, but the uniqueness, in these conditions, of SHG signal to collagen and TPF signal to elastin and cells allows for clear discrimination of those tissue constituents. The origins of SHG signal from collagen and TPF signal from elastin, immune cells, epithelium, and fibrin exudates are confirmed by comparison of MPM images of cryosections with Masson’s trichrome and VVG-stained tracheal sections taken from serial sections of the same tracheal tissue. The lamina reticularis of uninjured tracheal sections contains a uniform and dense network of SHG-producing collagen [Fig. 2] and TPF-producing elastin [Fig. 2] basal to a uniform, columnar epithelium (e) which also emits a TPF signal [Fig. 2]. Blood vessels are visible basal to and within the lamina reticularis as void space with little SHG signal [Fig. 2, yellow arrow]. Phalloidin-conjugated fluorophore staining of tracheal sections demonstrate the presence of blood vessels within these regions devoid of SHG and TPF signal [Fig. 2]. In contrast, the scrape-wounded tracheal mucosa shows a thicker, less connected collagen and elastin network within a disrupted, nonuniform lamina reticularis, evidenced by SHG [Fig. 2] and TPF [Fig. 2] signal from these layers. The thickened epithelium is weakly autofluorescent, and a fibrin exudate/ulcer covers the jagged apical surface of the lamina reticularis [Fig. 2, labeled u]. From MPM images of the same trachea cryosections that span the whole tracheal cross section, the thickness of the uninjured epithelium is $35–60\mu \mathrm{m}$ ; the lamina reticularis is $45–90\mu \mathrm{m}$ ; and the whole tracheal wall is $\sim 0.9\mathrm{mm}$ . In contrast, MPM images of the scrape-injured trachea cryosection reveal a fibrin layer thickness of $90–150\mu \mathrm{m}$ , a lamina reticularis of $100–250\mu \mathrm{m}$ , and a tracheal wall thickness of $1–1.5\mathrm{mm}$ .

SHG images of scraped tracheas show a approximate twofold greater area of mucosal collagen [Fig. 2 versus 2, $p$ value $<0.01$ ), while the area of elastic fibers assessed from the TPF image does not significantly change ( $p$ $\text{value}=0.24$ ), though scrape injury clearly disrupts the elastic fiber network (see Figs. 3 and 4 ). A loss of epithelial-derived TPF signal was observed [Fig. 2 versus 2], corresponding to ulcerated tissue replacing the denuded epithelium and visible in the histology sections [Figs. 2 and 2 versus 2 and 2]. Figure 2 shows that changes to collagen and elastic fiber cross-sectional area determined from tracheal sections imaged with MPM or stained for histology display similar trends of increased collagen area within the mucosal tissue, with relatively unchanged cross-sectional areas of elastic fibers.

Fig. 3

En face MPM images of (a) control and (b) scrape-wounded tracheal submucosa showing SHG signal (blue) from collagen, and TPF signal (green) from elastin, immune cells, and epithelium. (c) En face MPM image of the lamina propria from a control trachea, displaying endogenous SHG (blue) and TPF (green) signals, as well as Alexafluor 568-conjugated phalloidin staining (red), revealing blood vessels and actin-rich cells. bv, fc, el, f, and e represent blood vessel, fibrin clot, elastin, collagen fiber, and epithelium, respectively. Arrows point to TPF signal consistent with immune cells. Images are $\sim 30\mu \mathrm{m}$ below the surface and tiled together to display regional heterogeneity in the submucosa. The scale bars are $200\mu \mathrm{m}$ (a, b) and $50\mu \mathrm{m}$ (c).

Fig. 4

MPM images characterize normal and altered submucosal extracellular matrix. Coregistered en face SHG and TPF images of tracheal submucosa (taken at the plane of the dotted red line) reveal changes in the SHG producing elements [i.e., collagen (a–e)] and TPF producing elements (f–j) of the superficial submucosa, consistent with elastin and immune cells. Z-stack projections of merged SHG and TPF images (k–o) reveal epithelial and subepithelial morphological changes in the apical-basal plane of the trachea. MPM images are from a control trachea (a, f, k), a single scrape-wounded trachea (b, g, l), a triple scrape-wounded trachea allowed to heal for three days (c, h, m), an elastase-treated trachea (d, i, n), and a collagenase-treated trachea (e, j, o). The dashed red line represents the approximate location of the corresponding en face images in (a)–(n). d represents damaged epithelium. The scale bar is $50\mu \mathrm{m}$ .

3.2.

Microstructure Assessed by MPM

Taking advantage of the inherent depth-resolved nature of SHG and TPF signals, we were able to resolve the microstructural details of the uninjured and scrape-injured subepithelial matrices by imaging a plane of tiled images beneath the epithelium and fibrin clot, respectively, in whole, unfixed tracheas (Fig. 3). Beneath the uninjured epithelium lies an interspersed network of SHG-producing collagen fibers and TPF-producing elastic fibers [Fig. 3]. The collagen fibers are thicker ( $2.3\pm 1.2\mu \mathrm{m}$ versus $1.7\pm 0.7\mu \mathrm{m}$ , $p$ value $<0.05$ , two-tailed t-test, $n=29$ collagen, and elastic fibers) and more sinuous, and both fibers are oriented preferentially in one direction (parallel to the tracheal cylindrical axis). The combined networks occupy most of the image area relatively homogeneously except for gaps that stain positive for f-actin [Fig. 3; red channel, labeled bv], and which histology reveals to be small blood vessels [Fig. 2, labeled bv].

In contrast, the scrape-injured trachea reveals an inhomogeneous network of finer collagen fibers (many not clearly resolvable as individual structures) deposited in a dense meshwork with random orientation [Fig. 3, labeled f]. Strongly fluorescing immune cells [Fig. 3, yellow arrows] are interspersed in a faintly autofluorescent fibrin clot [Fig. 3, labeled fc], with rare occurrence of areas of damaged elastin [Fig. 3, labeled el] and epithelium [Fig. 3, labeled e].

MPM z stacks of unmounted and unsectioned tracheal tissue show microstructures of uninjured rabbit tracheas and tracheas subjected to single scrape injury, triple scrape injury with recovery, elastase treatment, and collagenase treatment (Fig. 4). MPM images of uninjured tracheas display the previously described subepithelial collagen network [Figs. 4 en face and 4 axial view, blue signal, basal layer] and elastic fiber network [Figs. 4 en face and 4 axial view, green signal, basal layer] consistent with the lamina reticularis. MPM images of control tracheas also display normal epithelium [Fig. 4 axial view, green signal, apical layer] and lamina reticularis both $30–45\mu \mathrm{m}$ thick [Fig. 4]. A single scrape injury leaves the collagen network relatively intact [Fig. 4], but breaks many elastic fibers [Fig. 4, damage indicated with yellow arrow] and denudes the epithelium [Fig. 4]. Tracheas subjected to three successive scrapes and allowed to heal for three days formed a dense meshwork of fibrotic collagen [Fig. 4] interspersed with immune cells [Fig. 4, indicated by yellow arrows], and a partially restored, but damaged, epithelium mixed with fibrin clot apical to a thicker mucosal collagen layer [Fig. 4, labeled d]. These features are similar to the those seen in MPM images of tracheal sections [Figs. 2, 2, 2, 2 and are confirmed by histology of tracheal sections [Fig. 2, 2, 2, 2]. Elastase treatment allowed the collagen fibers to become more dispersed and relax from the sinuous, aligned network [Fig. 4] as the surrounding elastic fiber network is almost completely degraded, as assessed by lack of TPF signal from these samples [Fig. 4]. The collagen fiber layer expands and becomes sparser [Fig. 4]. Collagenase removes the epithelium [Fig. 4] but has little effect on the subepithelial collagen as assessed by SHG signal [Fig. 4], although individual elastic fibers display increased curvature suggestive of possible stress relaxation [Fig. 4].

3.3.

Relationship of $E$ with MPM Image Parameters

Figure 5 presents the mean and standard deviation of $E$ , the Young’s modulus in compression of the ex vivo tracheas. The mucosa of the scrape-injured tracheas allowed to heal for three days were the stiffest $(E=9.0\pm 3.2\mathrm{kPa})$ , followed by the uninjured tracheas $(E=4.1\pm 0.7\mathrm{kPa})$ . Collagenase-treated tracheas were somewhat less stiff than the control $(E=2.6\pm 0.4\mathrm{kPa})$ , and elastase-treated tracheas were the least stiff [ $E=0.8\pm 0.3\mathrm{kPa}$ , Fig. 5]. TPF and SHG signal area fractions largely failed to correlate with $E$ [Figs. 5 and 5, respectively], except the modest and negative correlation of $E$ and TPF signal area fraction from scrape-injured mucosa [ $R^2=0.56$ ; Fig. 5]. $E$ correlates highly with the SHG signal intensity decay coefficient from noise-subtracted images for the enzyme-treated and control tracheas only [ $R^2=0.77$ versus $R^2=0.16$ from scrape-injured mucosa, Fig. 5]. In contrast, $E$ correlates modestly with the SHG signal intensity depth decay coefficient from segmented SHG images (excluding void pixels in the calculation), including the scrape-injured tissues [ $R^2=0.44$ , Fig. 5].

Fig. 5

Mechanical testing of tracheas: (a) Young’s modulus in compression $\left(E\right)$ , measured in the low strain region by indentation testing to a depth of $100\mu \mathrm{m}$ on trachea sections varies significantly with experimental treatment of the tracheas. (ANOVA, $p$ value $<0.0001$ , ${}^{*}$ indicates statistical significance. $N=9$ , 8, 6, and 6 tracheas for the control, chronic scrape, collagenase, and elastase-treated groups, respectively, with $n=6–14$ indentation tests performed per trachea over the distal, injured region). (b) $E$ versus TPF signal area fraction from a representative z slice of the submucosa plotted for the control and three treatments. (c) $E$ versus SHG signal area fraction from a representative z slice of the submucosa plotted for the control and three treatments. (d) $E$ versus noise-subtracted SHG decay from MPM z stacks of the tracheal submucosa plotted for the control and three treatments. (e) $E$ versus noise-subtracted SHG decay from MPM z stacks of the tracheal submucosa, segmented to examine only signal-containing pixels, plotted for the control and three treatments. $N=8$ , 7, 6, and 6 tracheas for the control, chronic scrape, collagenase, and elastase-treated groups, respectively, with $n=3–10$ MPM z stacks imaged per trachea over varying locations on the distal portion. Linear best-fit lines to the data are shown with solid or dashed lines.