C57BL/6, Lyve1^cre/cre and Csf1r^flox/flox were purchased from Jackson Laboratory. Lyve1^cre/cre and Csf1r^flox/flox were cross-bred to generate Lyve1^wt/cre;Csf1r^flox/flox mice (14). All animals were maintained in specific pathogen-free animal facility (Comparative Medicine, National University of Singapore) and were handled in accordance to protocols approved by the institutional animal care and use committee (IACUC) of the National University of Singapore.

Female mice from 15 to 18 weeks of age were used for the experiments. Only female were used in this study because the depletion of Lyve-1⁺ macrophages in lung is mainly observed in female mice. To induce COPD, 15 weeks old female Lyve1^wt/cre;Csf1r^flox/flox and Csf1r ^flox/flox mice were exposed to 4% CS from nine 3R4F reference cigarettes (University of Kentucky, Lexington, KY, USA) daily at a frequency of 3 cigarettes every 2 h for 8 weeks at a frequency of 5 consecutive days per week as described previously (15).

Lung tissues were extracted, minced, and digested for 30 minutes with 0.2mg/ml of collagenase (Sigma). The digested materials were filtered through 160mm nylon mesh filter and pelleted by centrifugation at 1200rpm for 5 minutes at 4°C. Cells were stained at 4°C in FACS buffer containing 1% normal mouse serum (Jackson ImmunoResearch Laboratories).

The following fluorochrome- and biotin-conjugated mouse specific antibodies were used; CD64 (clone X54-5/7.1.1), CD45 (clone 30-F11), LYVE-1 (clone ALY7), CD11b (clone M1/70), CD115 (clone AFS98), F4/80 (clone BM8), MerTK (R&D System), MHCII (clone M5/114.15.2), Ly6C (clone HK1.4), SiglecF (clone E50-2440), and streptavidin PE-CF594 (BD Horizon). Alveolar macrophages were identified as CD45⁺ CD64⁺ MerTK⁺ SiglecF+. Interstitial macrophages were identified as CD45⁺CD64⁺MerTK⁺SiglecF^- (12). Lung monocytes were identified as CD45⁺ CD11b⁺ Ly6C⁺. Dead cells were excluded by using DAPI. The cells were acquired on Aurora Spectral Analyser (Cytek Biosciences) and analyzed with Flowjo Software (Tree Star, Inc).

Soluble collagen was determined by Sircol soluble collagen assay kit (Biocolor S1000 Kit). Lung tissues were mixed with digestion mix (1M Acetic Acid and 0.2mg/ml Pepsin) and incubated overnight on a rotating Ferris wheel in cold room. On the next day the tubes were centrifuged at 12,000X g for 10 mins. Supernatant was transferred into a new tube. The pellet was kept for insoluble collagen assay (Biocolor S2000 kit). To concentrate collagen in the test sample, 100μl of neutralizing reagent was added to the supernatant and 200 µl of cold isolation and concentration reagent was added to each tube. The content was mixed thoroughly by tube inversions and incubated on metal rack overnight on ice in cold room. Next day tubes were centrifuged at 12,000 rpm for 10 mins. The supernatant was removed, and the pellet was resuspended with 100μl of 0.5M Acetic Acid. Collagen standards with 6 different dilutions (0µl, 20µl, 40µl 60µl, 80 µl & 100 µl) were prepared and topped up by 0.5M acetic acid to a total volume of 100 µl. Equal amount of Sircol Dye Reagent (1ml) was mixed with 20ul test sample and 100µl of prepared collagen reference standard and incubated for 30 mins on vortex mixer. Then the tubes were centrifuged at 12,000rpm for 10 mins. The tubes were inverted carefully to drain the supernatant. Gently 750ul of ice-cold Acid-Salt Wash Reagent was added to the pellet by inverting 3 times. Then the content was centrifuged at 12,000rpm for 10 mins. The supernatant was removed slowly using a pipette then by using a cotton bud. Equal amount of Alkali Reagent (250µl) was added to blanks, standards, and tests samples. Then the tubes were vortexed and shacked on a mechanical shaker for 10 mins. Then 200ul of Sircol dye- alkali reagent from each tube was transferred to a 96 well micro well plate and absorbance was measured at 555nm.

Insoluble collagen was determined using Sircol insoluble collagen assay kit (Biocolor S2000 Kit). The pellet collected from the sircol soluble collagen assay, after pepsin-acetic acid digestion was fragmented by adding Fragmentation Reagent to the pellet (50μl/mg of tissue) and incubating the mixture at 65°C for 2-3 hrs. Then the tubes were centrifuged at 12,000 rpm for 10 mins and the supernatant (test sample) was transferred into fresh tubes. Based on the weight of the tissue, from 5μl or 50μl of test sample was transferred to a new tube for each reaction and topped up to 100μl using dH₂0. Denatured collagen standards with 5 different dilutions (0µl, 20µl, 40µl 60µl & 100 µl) were prepared and topped up to a total volume of 100 µl using dH₂0. Then 20 µl of test sample was incubated with 100μl of prepared Collagen Reference Standards & 1ml of Sircol Dye Reagent. The content was mixed by inversion and incubated for 30 mins with mixing. The tubes were centrifuged at 12,000rpm for 10 mins. The supernatant was drained carefully by inverting tubes.

Gently 750μl of ice-cold Acid-Salt Wash Reagent was added to the pellet by inverting 3 times. Then the content was centrifuged at 12,000rpm for 10 mins. The supernatant was removed slowly using a pipette then by using a cotton bud. Equal amount of Alkali Reagent (1ml) was added to blanks, standards, and tests samples. Then the tubes were vortexed and shacked on a mechanical shaker for 10 mins. Then 200ul of Sircol dye- alkali reagent from each tube was transferred to a 96 well micro well plate and absorbance was measured at 555nm.

Lung tissues were fixed in 2% paraformaldehyde (PFA) and 30% sucrose and prepared for Cryo-sectioning (10µm). Picrosirus Red (PSR) staining was done using commercially available kit according to manufacturer instruction to estimate the amount/nature of collagen. Bright field microscope was used to take images. Linear polarized microscope was used to see nature of collagen. Cryo sections of the lung (10µm) were also stained with hematoxylin and eosin (H and E) for checking morphology of the lung. The condition of emphysema was assessed by observing the size of alveoli by using the mean linear intercept technique (MLI). MLI refers to the total length of hypothetical line on the lung sections divided by the numbers of alveolar walls intersecting the test lines as described (16). Fields that include larger airways and blood vessels were excluded from the analysis. The analysis was done using 30 different fields of view per lung using the 20X objective. ImageJ software was used to quantify the amount of collagen. Collagen quantification from images taken using bright field microscope was done by converting PSR stained image in to RGB image and by thresholding the collagen stained with only red. The collagen quantification from linear polarized microscope was done by setting 2 Hue values; magenta, red and orange for thick collagen fiber and yellow-green for thin collagen fibers followed by thresholding of each colour.

Cryo-sections of mouse lungs (10µm) were prepared for immunohistochemistry by using 2% paraformaldehyde (PFA) and 30% sucrose fixative. Primary antibodies used include anti-LYVE-1 (rabbit polyclonal; Abcam), anti-CD68 (rat clone FA-11; Bio-rad) anti-CD206 (rat clone MR5D3; Bio-rad), anti-CD31 (armenian hamster clone 2H8; Chemicon), anti-collagen type I (rabbit polyclonal; Millipore), and anti-collagen type III (rabbit polyclonal; Abcam).

For isotype controls, rat IgG (eBioscience), armenian hamster IgG (eBioscience), and rabbit IgG (Jackson Immunoresearch) were used. AF488-, Cy3-and AF647- conjugated antibodies from Jackson Immunoresearch were used for detection. FITC or Cy3-conjugated anti smooth muscle actin (SMA, mouse clone 1A4; Sigma-Aldrich) was used to identify smooth muscle cells. Sections were counterstained with 4,6-diamidino-2-phenylindole (DAPI) for cell nuclei visualization and mounted for analysis. Specimens were viewed with a fluorescence widefield (Axio Imager.Z1, Axioxam HRM camera; Carl Zeiss Micro Imaging, Inc., Jena, Germany).

Immediately after dissection, lung tissues were infused with RNA stabilizing reagent (RNA later, QIAGEN) for one hour at 4 °C degrees. Trizol reagent was used for tissue homogenization and total RNA was extracted from the tissues using an mRNA isolation kit (QIAGEN) according to the manufacturer’s instruction followed by cDNA synthesis from 500-600ng of RNA using Taqman reverse transcription kit (Applied Biosystems). All PCR primers were purchased from Integrated DNA Technologies (Coraville, IA, USA) (listed in Table 1 ). Real time quantitative PCR was performed in triplicates using 100 ng template cDNA in PCR mixture containing SYBR Green Master Mix (Applied Biosystems, Carlsbad, CA, USA). The mRNA expression levels of all samples were normalized with housekeeping gene r12s. Analysis was done using Abi Prism 7500 Detection System (Applied Biosystems, Warrington, UK).

The activity of gelatinase enzyme was determined by gelatin zymography assay. Lung tissues were homogenized using tissue homogenizing buffer and protein concentration of the sample was determined by BCA assay kit. Resolving gel with 10% polyacrylamide containing 1% gelatin and stacking gel (5% polyacrylamide) was prepared. Gel electrophoresis was done using 10ug protein without any reducing agent. Then, the gel was washed in renaturing solution. The gel was incubated in development buffer for 15 minutes with gentle agitation at room temperature then transferred to 37°C incubator for 16-20 hours incubation. The gel was stained in staining solution for 4 hours and destained in destaining solution until the bands were visible against blue background of the gel. Image was taken using gel imager (Bio-Rad) and quantification of the gel band intensity which is equivalent to the activity of the enzyme was done using ImageJ (version 1.43).

Mice were anesthetized with ketamine/medetomidine (75 mg/kg and 1 mg/kg, respectively) 24 h after the last CS exposure, and tracheotomy was performed as described (15). Mice were placed in a whole-body plethysmograph connected to a computer-controlled ventilator (Forced pulmonary maneuver system, Buxco Research System, Wilmington, NC, USA) (17). Functional residual capacity (FRC), static compliance (C_chord), forced expiratory volume at 100ms over forced vital capacity (FEV100/FVC), FEV at 50ms over FVC (FEV50/FVC) and FEV at 20ms over FVC (FEV20/FVC) and total lung capacity (TLC) were recorded using the FinePointe™ data acquisition and analysis software (Buxco). The volume of air remaining in the lung after passive expiration is called FRC. If the alveoli are abnormally expanded and unable to contract sufficiently to remove air, large volume of air will be trapped inside the alveoli which increases the value of FRC as observed in the case of emphysema. This may happen due to breakage of elastin (18) and/or presence of less collagen in the parenchyma (19). VC is the volume of air that can be blown out forcefully after maximum inspiration with slow manoeuvre. FVC is the same as VC except the manoeuvre during FVC is rapid and fast. Less VC and FVC values indicate poor lung performance. FEV100 is the volume of air that can forcibly be blown out in first 100ms after full inspiration.

Statistical analysis was performed with Graphpad Prism version 5.0 (GraphPad Software). Data were presented as mean ± SEM and were analysed by nonparametric Mann-Whitney U and significance levels were set at p<0.05.

Consistent with our previous study (12), SiglecF⁺ AMs and SiglecF^- IMs were identified in female mouse lungs by flow cytometry analysis at steady state ( Supplementary Figure 1A ). Among the SiglecF^- IMs, a sub-population expressed Lyve-1 and their presence in the lungs was further confirmed by immunohistochemistry ( Supplementary Figure 1B ; Figure 1A ). Since the lung is unique concerning Lyve-1 expression whereby all the reticuloendothelial systems including lymphatic and blood vessels of the lung are positive for Lyve-1, it makes difficult to distinguish different vessel types based on CD31, smooth muscle actin (SMA) and Lyve-1 expression ( Figure 1A ) (12). Thus, we focused our analysis on the larger bronchioles known to exhibit an artery and vein adjacent to it. Blood vessels expressed CD31, a pan-endothelial marker, SMA, and Lyve-1 whereas bronchioles are CD31^-SMA⁺Lyve-1^- ( Figure 1A ). Macrophage were identified by CD68, a pan-macrophage marker. Our immunofluorescent staining showed that Lyve-1⁺ macrophages were observed around the bronchiole and blood vessel ( Figures 1B, C ) and also expressed CD169 ( Figure 1D ). In line with our previous findings showing the M2-like phenotype of tissue Lyve-1⁺ macrophages (12, 14), we found in this study that lung Lyve-1⁺ macrophages were also positive for the M2 macrophage marker CD206 ( Figure 1E ). Next, we analyzed the frequency and number of Lyve-1⁺ macrophages in lung from control Csf1r^flox/flox mice and Lyve1^wt/cre; Csf1r^flox/flox mice in which Csf1r was specifically depleted in all Lyve-1-expressing cells. Because Lyve-1⁺ macrophages depend on the CSF-1R pathway, Lyve-1⁺ macrophages are depleted in Lyve1^wt/cre; Csf1r^flox/flox mice. Similar to the aortas (14), depletion of Lyve-1⁺ macrophages was observed in the lungs of Lyve1^wt/cre; Csf1r^flox/flox mice whereas lung Lyve-1^- macrophages and monocytes were not affected ( Figure 1F ).

Using Lyve1^wt/cre; Csf1r^flox/flox mice, we investigated how the depletion of Lyve-1⁺ macrophages affects the distribution, content and types of collagen in the lungs at steady state.

Analysis of lung sections stained with picrosirus red under bright field microscope revealed that collagen content increased both around the alveoli and the bronchioles in Lyve1^wt/cre ; Csf1r^flox/flox mice compared to control Csf1r^flox/flox mice ( Figures 2A, B ). To determine which type of collagen contributed the most to this overall increase in lung collagen, we specifically immunostained for collagen type I and type III, the predominant types of collagen in the lung. Notably, collagen type I expression was significantly higher in Lyve1^wt/cre ; Csf1r^flox/flox mice compared to control Csf1r^flox/flox mice whereas no difference in the amount of collagen type III was detected ( Figure 2C ).

Under polarized light microscope, the architecture including thickness and organization (orientation and spatial distribution) of collagen fibers stained by picrosirius red is readily identified by the changes of colors and packing density (20). The immature thin collagen fibers have weak birefringence and appear green while the matured thick collagen fibers become orange or red (20). Moreover, immature or less crosslinked collagen is weaker and more susceptible to rupture unlike the mature crosslinked collagen which is relatively stiffer (21, 22). Our polarized light microscopic analysis confirmed the accumulation of alveolar and bronchiolar collagen and revealed a predominant red orange (RO) birefringence, i.e., thick collagen fibers around both the alveoli and the bronchioles of Lyve1 ^wt/cre ; Csf1r^flox/flox mice and thinner collagen fibers with yellow green (YG) birefringence in control mice ( Figures 2D, E ). Color quantification of the image confirmed the increased proportion of red/orange collagen in Lyve1^wt/cre; Csf1r^flox/flox mice compared to control Csf1r^flox/flox mice ( Figure 2F ). This result suggests that the collagen accumulating in Lyve1^wt/cre; Csf1r^flox/flox mice may be more crosslinked.

To further support the increase in collagen content and predominance of more crosslinked fibers in lungs from Lyve1^wt/cre; Csf1r^flox/flox mice, we used sircol collagen assay to quantify the amount of soluble and insoluble collagen. Because the solubility of collagen is determined in part by the degree and type of cross-linking between collagen molecules, the soluble collagen is in general less cross-linked, while the insoluble collagen is considered to be more extensively cross-linked (20, 22). The amount of total collagen (the sum of soluble and insoluble collagen), insoluble collagen and the ratio of insoluble to soluble collagen were significantly augmented in lungs of Lyve1^wt/cre; Csf1r^flox/flox mice than in control mice ( Figure 2G ). Altogether, these data revealed that Lyve-1⁺ macrophages are important in restraining collagen deposition in the lung parenchyma at steady state.

To understand the mechanism underlying the accumulation of collagen in lungs after depletion of Lyve-1⁺ macrophages, we examined the expression of genes controlling the synthesis and degradation of collagens by real-time PCR. Pro-collagen type 1 alpha 1 (Col1a1) gene expression was markedly upregulated in lungs of Lyve1^wt/cre; Csf1r^flox/flox mice. No significant difference in the mRNA expression levels of tropo-elastin (Eln), the precursor of elastin, was observed ( Figure 3A ). Having shown the presence of more thick and crosslinked type of collagen in the lungs of Lyve1^wt/cre; Csf1r^flox/flox mice, we examined the expression of genes involved in collagen crosslinking. The mRNA level of Lox was significantly upregulated in Lyve1^wt/cre; Csf1r^flox/flox mice while no difference in the expression level of Loxl2 was noted and Loxl1 expression was not detected ( Figure 3A ). The expression level of collagen-degrading matrix metalloproteinase 9 (MMP-9) was also significantly decreased in Lyve1^wt/cre; Csf1r^flox/flox mice ( Figure 3A ). In contrast, the pro-collagen form, MMP inhibitor Timp3 was upregulated in Lyve1^wt/cre; Csf1r^flox/flox mice ( Figure 3A ). In addition, gelatin-based zymography assay revealed a reduced MMP-9 proteolytic activity in lung homogenate from Lyve1^wt/cre; Csf1r^flox/flox mice compared to control mice ( Figure 3B ) which may contribute together with the increased Col1a1 expression to the accumulation of collagen observed in these mice.

Since collagen is an important structural protein which protects the lung from any mechanical stress, we investigated whether the accumulation of collagen observed in the lungs of Lyve1 ^wt/cre ; Csf1r ^flox/flox mice impact the homeostatic lung function by conducting lung functional tests at steady state. Lung function examination is a group of tests that enable us to evaluate lung competencies. It includes functional residual capacity (FRC), total lung capacity (TLC), static compliance (Cchord), Forced expiratory volume (FEV) at 100ms over forced vital capacity (FVC) ratio (FEV100/FVC), FEV at 20ms over FVC ratio and FEV at 50ms over FVC ratio. The lung function tests conducted in Lyve1^wt/cre; Csf1r^flox/flox mice and Csf1r^flox/flox control mice at steady state revealed no difference in all the lung function parameters tested among the two groups ( Figure 4 ). These results suggest that the deposition of collagen observed in the lung from mice depleted of Lyve-1⁺ macrophage was not sufficient to alter overall airway function at steady state.

We sought to investigate the impact of collagen deposition resulting from the depletion of Lyve-1⁺ macrophages on COPD. COPD was induced in Lyve1^wt/cre; Csf1r^flox/flox and Csf1r^flox/flox control mice through chronic exposure to CS which is a well-established animal model mimicking most of the COPD symptoms seen in human patients (15, 23). No significant differences in mouse weights were detected between Lyve1^wt/cre; Csf1r^flox/flox and Csf1r^flox/flox control mice after exposure to CS ( Supplementary Figure 2 ). As observed at steady state ( Figure 2 ), collagen accumulated more in the parenchyma and small airways of Lyve1^wt/cre; Csf1r^flox/flox mice after CS compared to CS-exposed Csf1r^flox/flox mice ( Supplementary Figure 3 ). CS inhalation induces various key features of COPD pathophysiology including emphysema, thickening of bronchial wall, and bronchitis (23). Emphysema, which is characterized by enlargement of the alveoli and, overtime, the rupture of the alveolar wall which ultimately leads to shortness of breath, can be assessed by estimating the size of the alveoli using mean linear intercept method (MLI). The smaller the MLI value is, the larger the size of alveoli is, which designates emphysema (16). Assessment of MLI from lung sections stained with Haematoxylin and Eosin showed emphysema in lungs of Csf1r^flox/flox mice after CS exposure compared to non-CS exposed Csf1r^flox/flox mice ( Figures 5A, B ). In contrast, emphysema was significantly reduced in Lyve1^wt/cre; Csf1r^flox/flox mice after CS exposure ( Figures 5A, B ). Next, we evaluated the impact of Lyve-1⁺ macrophage depletion on bronchial thickening, another pathophysiological feature of COPD. Haematoxylin and Eosin staining of lung sections revealed that thickening of small airways in Lyve1^wt/cre; Csf1r^flox/flox mice after CS exposure was also decreased compared to control Csf1r^flox/flox mice ( Figures 5C, D ). COPD is associated with an increase in airway smooth muscle cell mass (23) and bronchial thickening in CS exposed mice is accompanied by thickening of smooth muscle layer in the bronchiole (16). Immunofluorescence staining of SMA that identifies bronchiole smooth muscle cells revealed a marked decreased in SMA expression in bronchioles from Lyve1^wt/cre; Csf1r^flox/flox mice after CS exposure compared to control Csf1r^flox/flox mice ( Figure 5E ). Finally, we analysed inflammation after CS exposure since it is also associated with COPD. However, the expression of inflammatory genes including Tnfa, Il1b, Il10 and Il4 was not different in Lyve1^wt/cre; Csf1r^flox/flox mice exposed to CS compared to control Csf1r^flox/flox mice ( Supplementary Figure 4A ). Similarly, no differences were observed for the infiltration of monocytes ( Supplementary Figure 4B ) or macrophages ( Supplementary Figures 4C, D ) between Lyve1^wt/cre; Csf1r^flox/flox mice and Csf1r^flox/flox mice exposed to CS. These results suggest that depletion of Lyve-1⁺ macrophage prior COPD induction which is accompanied by collagen accumulation mainly restrained emphysema and bronchial thickening without affecting inflammation in response to chronic exposure to CS.

We investigated whether the reduction in CS-induced emphysema and small airway thickening observed in mice after depletion in Lyve-1⁺ macrophage was sufficient to improve lung function. CS-induced COPD mouse model is characterized by alterations in some lung function parameters including an increase in FRC, TLC and c_chord and a decrease in FEV100/FVC, FEV50/FVC and FEV20/FVC. As expected FRC, TLC, and C_chord were increased in Csf1r^flox/flox mice after chronic exposure to CS compared to non-CS exposed Csf1r^flox/flox mice ( Figures 6A–C ). However, these lung parameters were improved with depletion of Lyve-1⁺ macrophage ( Figures 6A–C ). Similarly, FEV100/FVC, FEV50/FVC and FEV20/FVC values were decreased in Csf1r^flox/flox mice after CS inhalation, but this decrease was attenuated in Lyve-1^wt/cre; Csf1r^flox/flox mice ( Figure 6D ). Altogether, lung function tests revealed that depletion in Lyve-1⁺ macrophage overall prevents loss of airway flow in response to chronic exposure to CS.