The effect of adrenalectomy on bleomycin-induced pulmonary fibrosis in mice

Abstract

Progressive lung fibrosis is often fatal and has limited treatment options. Though the mechanisms are poorly understood, fibrosis is increasingly linked with catecholamines such as adrenaline (AD) and noradrenaline (NA) and hormones such as aldosterone (ALD). The essential functions of the adrenal glands include the production of catecholamines and numerous hormones, but the contribution of adrenal glands to lung fibrosis remains less well studied. Here, we characterized the impact of surgical adrenal ablation in the bleomycin model of lung fibrosis. Wild-type mice underwent surgical adrenalectomy or sham surgery followed by bleomycin administration. We found that although bleomycin-induced collagen overdeposition in the lung was not affected by adrenalectomy, histologic indices of lung remodeling were ameliorated. These findings were accompanied by a decrease of lymphocytes in bronchoalveolar lavage (BAL) and macrophages in lung tissues, along with concomitant reductions in alpha-smooth muscle actin (αSMA) and fibronectin. Surgical adrenalectomy completely abrogated AD, not NA, detection in all compartments. Systemic ALD levels were reduced after adrenalectomy, whereas ALD levels in lung tissues remained unaffected. Taken together, these results support the presence of a pulmonary-adrenal axis in lung fibrosis and suggest that adrenalectomy is protective in this disease. Further investigation will be needed to better understand this observation and aid in the development of novel therapeutic strategies.

NEW & NOTEWORTHY The lung-adrenal axis plays a significant role in pulmonary fibrosis. Adrenalectomy provides protection against lung fibrotic ECM remodeling and lung inflammation by reducing the levels of lymphocytes in BAL and macrophages in lung of bleomycin-treated mice. Although compared with sham surgery, adrenalectomy raised collagen concentration in uninjured mice, there was no discernible difference in bleomycin-induced collagen accumulation. However, adrenalectomy significantly reversed the enhanced expression and colocalization of αSMA and fibronectin induced by bleomycin.

INTRODUCTION

Progressive pulmonary fibrosis (PPF) is characterized by progressive scarring of the lungs (1), remodeling of the lung’s extracellular matrix (ECM), and increased deposition of collagen and alpha-smooth muscle actin (αSMA) (2). This is thought to be driven by elevated levels of fibroblasts that have differentiated into activated myofibroblasts (3, 4). PPF is a devastating form of interstitial lung disease (ILD), has limited treatment options, and is often fatal, especially in idiopathic pulmonary fibrosis (IPF) with a life expectancy of 3–5 yr after diagnosis (5). Furthermore, the mechanisms behind these diseases are poorly understood, making it difficult to develop effective treatments (1). A better understanding of the drivers of fibroblast to myofibroblast differentiation, collagen deposition, ECM remodeling, and how these processes might be reversed is paramount to the development of more effective therapies.

The adrenal glands are important regulators of systemic homeostasis and respond to various stimuli via their production of catecholamines and hormones, such as aldosterone (ALD). Emerging studies suggest that fibrosis is associated with catecholamines such as adrenaline (AD) and noradrenaline (NA) (6–10). For example, in their study of sepsis-associated pulmonary fibrosis in mice, Su et al. (11) demonstrated that NA induces fibroblast differentiation via the alpha-adrenergic receptor 2A. NA has also been shown to increase collagen type I, matrix-metalloproteinase 2, and tissue inhibitor of metalloproteinases 2 via predominantly alpha-adrenergic mechanisms (9). In addition, work from our laboratory has shown that NA derived from lung-specific adrenergic nerves contributes to pulmonary fibrosis (7, 8, 12), which was ameliorated with alpha-adrenergic receptor 1D (ADRA1D) antagonism (8). ALD, as the final effector of the renin-angiotensin-aldosterone system (RAAS), contributes to the regulation of ECM deposition and the activity of fibrotic mediators such as TGFβ1 (13). Furthermore, ALD has been shown to be an important mediator of inflammatory lung diseases such as sarcoidosis and contributes to the release of proinflammatory cytokines (14). A better understanding of these associations has the potential to provide insights into the mechanism(s) by which the adrenal glands and catecholamines contribute to pulmonary fibrosis and unveil new opportunities for treatment.

This study sought to address these knowledge gaps by assessing whether adrenalectomy might mitigate lung inflammation, biochemical collagen accumulation, histologic ECM remodeling in the lung, and catecholamine and ALD levels in the bleomycin model of experimental pulmonary fibrosis.

MATERIALS AND METHODS

Animals

All experiments involving animals were approved by Yale School of Medicine IACUC in accordance with federal regulations (protocol #20292). All experiments used C57 Black 6 (C57BL/6) wild-type mice. The mice underwent either an adrenalectomy or sham surgery (Fig. 1) conducted by both our own laboratory personnel and the Jackson Laboratory. The mice were injected subcutaneously with buprenorphine proximate at the time of surgery. More than 10 min after the injection, the mice were administered one dose of ketamine and xylazine. A one-third re-dose was administered if proper sedation, assessed by the absence of response to toe-pinch, was not yet achieved. The anesthetized mice were placed in ventral recumbency. Isoflurane was administered nasally by a vaporizer, and an eye lubricant was applied to both eyes. The mice were shaved from the hip to the mid-thorax on the dorsal side, and then lidocaine was injected intradermally around the planned incisions. The skin was sterilely prepared with betadine and 70% ethanol. A 1–2 cm incision, adjusted to the size of the animal, was made in the mid-dorsal area, with its extreme cranial end at the level of the 13th rib. A window through the muscle fibers was made by entering the fibers and separating them bluntly with sharp (iris) scissors. The adrenal gland was removed intact with forceps without any cutting. The muscle layer was closed with absorbable suture, and the skin was closed with surgical clips. Immediately after surgery, the mice were subcutaneously injected with 500 µL of meloxicam in 0.9% NaCl. Meloxicam was injected every 24 h for 3 days, and 1% saline was added to the drinking water to counteract mineralocorticoid deficiency. After 7 days, the surgical clips were removed under anesthesia. After the mice were allowed to recover from the surgery for 3 wk, they were administered either a single dose of 1.5 U/kg pharmacologic grade bleomycin (Northstar Rx LLC, NDC 16714-8860-0) or sterile saline by orotracheal aspiration (15). Mice were anesthetized with isoflurane and suspended by their incisors on a standing rack. With the tongue held in gentle retraction, 50 µL bleomycin or saline was pipetted into the oropharynx and aspirated. All mice were humanely euthanized after 21 days for sample collection.

Figure 1.

Adrenalectomy and sham surgery were applied to the bleomycin model. WT mice underwent surgical adrenalectomy or sham surgery. After 21 days, mice were anesthetized with isoflurane and bleomycin (1.5 U/kg of body weight) dissolved in saline or vehicle control was administered orotracheally as a single dose. The samples were harvested at 42 days. WT, wild-type.

Bronchoalveolar Lavage Cell Quantification

At the time of euthanization, two aliquots of 0.8 mL PBS were slowly instilled into the lung and the lavage fluid was gently aspirated. The combined bronchoalveolar lavage (BAL) sample was assessed for white blood cell (WBC) counts using a Beckman Coulter Ac.T Diff instrument.

Soluble and Insoluble Collagen Quantification

The soluble and insoluble collagen concentrations of the three lobes (superior, middle, and inferior lobes) of the right lungs of the euthanized mice were quantified using the Sircol Soluble Collagen Assay Kit (CLS 1111, Biocolor Ltd) and the Sircol Insoluble Collagen Assay Kit (CLS 2000, Biocolor Ltd), following manufacturer’s instructions.

Trichrome Staining and Scoring

Whole left lungs and the post-caval lobes were formalin-fixed and paraffin-embedded (FFPE), sectioned, and stained for histologic analysis with Masson’s trichrome to assess collagen deposition. Six images per slide were taken at ×20 magnification and assigned a modified Ashcroft score (MAS), as previously described (16). The scores were then averaged for each slide to determine an overall score for each lung.

Immunofluorescent Staining and Analysis

The FFPE-sectioned left lungs were also used to assess for ECM changes resulting from adrenalectomy. The slides were de-paraffinized in xylene and rehydrated in decreasing concentrations of ethanol in water. Antigen retrieval was done using BD Pharmingen Retrievagen A (Cat# 550524), following manufacturer’s instructions. The sections were penetrated with PBS + 0.25% Triton and blocked for 30 min using 3% bovine serum albumin (BSA) in PBS. Tissues were stained to test alpha-smooth muscle actin (αSMA) (Mouse mAb, Abcam, Cat# ab7817), laminin (Rabbit pAb, Abcam, Cat# ab11575), fibronectin (Rabbit pAb, Abcam, Cat# ab23751), CD68 (Rabbit, pAb, Abcam, Cat# ab125212), and CD14 (Rat mAb, BioLegend, Cat# 150102). All antibodies were diluted 1:250 in 1% BSA in PBS. Slides were incubated overnight at 4°C and stained with secondary antibody [Alexa Fluor 555 donkey anti-rabbit IgG (H + L) Invitrogen, Cat# A31572; Alexa Fluor 488 chicken anti-mouse IgG (H + L) Invitrogen, Cat# A21200; Chicken Anti-Rat IgG (H + L), Alexa Fluor 488, Invitrogen, Cat# A21470], diluted 1:500 in 1% BSA in PBS. The samples were incubated for 1 h at 37°C and then mounted using VECTASHIELD Mounting Media with DAPI (Cat# H-1200) and covered. Images were taken at ×20 and ×40 magnification using a Nikon Eclipse Ti microscope and the NIS Elements Br software. Images were merged and analyzed using ImageJ v1.54f. A threshold was set manually to separate positive from negative signals. The threshold was applied to all images taken from the same channel. Mean fluorescent intensity (MFI) and the expression areas of αSMA, laminin, and fibronectin were individually calculated. The JACoP plugin was used to measure the overlapping area of laminin or fibronectin with αSMA (17).

Noradrenaline and Adrenaline Quantification by ELISA

NA and AD concentrations from plasma, BAL, and homogenized right lower lobes of the lung tissues were quantified by ELISA using the Noradrenaline Research ELISA kit (LDN, BA E-5200R) and the Adrenaline Research ELISA kit (LDN, BA E-5100R), following manufacturer’s instructions.

Aldosterone Quantification by ELISA

Plasma and BAL supernatant aldosterone concentration was measured using an Aldosterone ELISA kit (LDN, MS E-5200R), following manufacturer’s instructions. All plasma samples were diluted 1:10 with the Standard A provided with the kit (LDN, E-5201). The right lower lobes of the lung tissue were homogenized in 500 µL PBS. After centrifugation, the supernatant was separated and used for the ELISA. The aldosterone concentration was then measured using an Aldosterone ELISA Kit (Competitive EIA) (LS Bio, LS-F39251), following manufacturer’s instructions.

BAL Differential by Microscopy

Harvested BAL cells were attached to blank slides by Cytospin. They were then fixed and stained using a DiffQuik staining kit (Thermo, Cat# 9990700), following manufacturer’s instructions. Cell counts based on morphology were done by two observers. Data considered for the differential were the percentage of each cell type in the number of cells counted and the total number of each cell type in the mouse’s BAL, determined by multiplying the percentages by the total BAL cell counts.

Flow Cytometry for Whole Blood Immune Cell Differential

Flow cytometry was done on lysed whole blood to characterize the types of immune cells. The whole blood was taken from the mice and incubated with 0.5 M EDTA. The red blood cells were then lysed in 10 mL of ACK lysing buffer (Gibco, A10492-01), and the supernatant was removed to isolate the pelleted peripheral blood mononuclear cells (PBMCs). The cells were washed with PBS before being blocked in FACS buffer (1X PBS, 1% BSA, 0.01% NaN₃, 1 mM EDTA) with 5% normal goat serum on ice for 20 min. The cells were fixed using BD Cytofix/Cytoperm Fixation/Permeabilization solution (BD, 51-2090kz). The cells were then stained with primary antibodies diluted 1:100 in FACS buffer. The panel included: Alexa Fluor 700 anti-mouse CD45 Antibody (BioLegend, Cat# 147716), FITC anti-mouse CD170 (SiglecF) (BioLegend, Cat# 155503), APC anti-mouse CD19 Antibody (BioLegend, Cat# 115512), PE/Dazzle 594 anti-mouse CD3 Antibody (BioLegend, Cat# 100245), PE/Cyanine7 anti-mouse Ly6G Antibody (BioLegend, Cat# 127617), BD PharMingen PerCP-Cy5.5 Rat Anti-CD11b (BD PharMingen, Cat# 550993), CD11c Monoclonal Antibody (N418), APC eFluor 780 (eBioscience, Cat# 47-0114-82), CD14 Monoclonal Antibody (Sa2-8), PE, eBioscience (eBioscience, Cat# 12-0141-81), and F4/80 Monoclonal Antibody (BM8), eFluor 450, eBioscience (eBioscience, Cat# 48-4801-82). LIVE/DEAD Fixable Aqua Dead Cell Stain Kit, for 405 nm excitation (Thermo Fisher, Cat# L34966), was used at a 1:100 dilution in FACS buffer. Cells were incubated for 30 min at 4°C in the dark. Cells were washed using BD Perm/Wash Perm/Wash Buffer (BD, Cat# 554723), resuspended in PBS, and filtered through 35 µm strainers before they were analyzed. Data were acquired using an LSRII cytometer (BD Biosciences, Franklin Lakes, NJ) and analyzed with FlowJo software (BD Biosciences, Franklin Lakes, NJ).

Statistical Analyses

Statistical analysis was performed using GraphPad Prism version 10.0.2. We used multiple unpaired, nonparametric Mann–Whitney U test (test) followed by the Bonferroni method for post hoc test according to the formula α′ = α/m, where α′ is the corrected P value, α is the original significance value 0.05, m is the number of tests 4, and P_corrected = 0.05/4 = 0.0125. Hence, we considered a P value ≤ 0.0125 to be significantly different.

RESULTS

Adrenalectomy Is Protective against Bleomycin-Induced Lung Inflammation

Adrenally derived hormones including NA, AD, and ALD are associated with inflammation (18–21). Reasoning that adrenalectomy would lead to a decrease in these hormone levels and therefore improved markers of inflammation, we evaluated BAL white blood cell (WBC) counts. Consistent with former studies (22), we found that bleomycin results in significantly increased lung inflammation by BAL WBC count and that this increase was ameliorated by adrenalectomy, though the change was not statistically significant (P = 0.0133, Fig. 2A). Furthermore, a differential BAL cell count was performed on a cytocentrifuge smear using a Diff-Quik stain kit. We found that bleomycin stimulation significantly affected the BAL differentiation compared with the mice that received saline. The percentage of macrophages significantly decreased, whereas their total number significantly increased (Tables 1 and 2 and Fig. 2, B and C). The total number and percentage of monocytes both increased, though only the total number was a statistically significant difference (Tables 1 and 2 and Fig. 2, D and E). Both the percentage and total number of lymphocytes, neutrophils, eosinophils, basophils, and mast cells were significantly increased in mice stimulated by bleomycin (Tables 1 and 2 and Fig. 2, D–O). Adrenalectomy significantly inhibited the decrease in the percentage of macrophages and the increase of the total number of lymphocytes (Tables 1 and 2 and Fig. 2, B and G). In addition, adrenalectomy inhibited the increase in both percentage and total number of eosinophils, basophils, and mast cells, though these changes were not statistically significant (Tables 1 and 2 and Fig. 2, J–O).

Figure 2.

The effect of adrenalectomy on bronchoalveolar lavage (BAL) cells in the bleomycin model. The total BAL cell counts and a differential BAL cell count were conducted to understand the changes in immune cell types that make up the BAL in the mice underwent sham surgery or adrenalectomy with and without bleomycin stimulation. In this assay, the total number of cells in the BAL (A), the percentage (B) and the total number (C) of macrophages, the percentage (D) and the total number (E) of monocytes, the percentage (F) and the total number (G) of lymphocytes, the percentage (H) and the total number (I) of neutrophils, the percentage (J) and the total number (K) of eosinophils, the percentage (L) and the total number (M) of basophils, and the percentage (N) and the total number (O) of mast cells in BAL were analyzed. Data are shown as means ± SD. Multiple unpaired nonparametric Mann–Whitney U tests followed by Bonferroni post hoc test were performed to determine significance differences between groups, after correction, P ≤ 0.0125 is considered as significantly different. Female mice are represented by open shapes, and male mice are represented by closed shapes (n = 15 female/15 male mice in sham saline group; 16 female/13 male mice in sham bleomycin group; 14 female/13 male mice in adrenalectomy saline group; 14 female/14 male mice in adrenalectomy bleomycin group).

Table 1.

The composition of BAL based on percentage of immune cell types

	Sham Saline (n = 15F/15M)	Sham Bleomycin (n = 16F/13M)	Adrenalectomy Saline (n = 14F/13M)	Adrenalectomy Bleomycin (n = 14F/14M)
%Macrophages	95.507 ± 3.323 (83.223–98.644)bbbb	77.336 ± 12.099 (43.04–96.249)aaaa,d	94.469 ± 4.818 (74.005–98.287)dddd	84.82 ± 6.95 (65.224–94.554)b,cccc
%Monocytes	0.149 ± 0.199 (0–0.658)b	0.443 ± 0.597 (0–2.452)a	0.178 ± 0.201 (0–0.65)	0.231 ± 0.283 (0–0.898)
%Lymphocytes	2.727 ± 2.304 (0.537–10.817)bbbb	19.53 ± 11.392 (2.928–50.88)aaaa	4.011 ± 4.611 (0.623–24.668)dddd	12.709 ± 7.284 (0.623–24.668)cccc
%Neutrophils	0.442 ± 0.856 (0–4.415)bb	1.062 ± 1.217 (0–5.213)aa	0.209 ± 0.259 (0–0.814)dddd	0.782 ± 0.692 (0–3.052)cccc
%Eosinophils	0.017 ± 0.07 (0–0.348)bbb	0.316 ± 0.595 (0–2.78)aaa	0.099 ± 0.199 (0–0.68)	0.225 ± 0.578 (0–2.976)
%Basophils	0.068 ± 0.191 (0–0.883)bbbb	0.289 ± 0.302 (0–1.075)aaaa	0.019 ± 0.049 (0–0.181)dd	0.141 ± 0.201 (0–0.73)cc
%Mast cells	0 ± 0 (0–0)bbbb	0.154 ± 0.209 (0–0.613)aaaa	0 ± 0 (0–0)	0.065 ± 0.197 (0–0.957)
%Other	1.09 ± 1.054 (0–3.929)	0.87 ± 1.302 (0–4.36)	1.015 ± 0.799 (0–2.85)	1.026 ± 0.976 (0–3.268)

Cells in the other category are cells that did not appear to match the criteria used to determine the cell type. Data are shown as means ± SD. The numbers in parentheses represent the range of values observed in the differential count. Multiple unpaired nonparametric Mann–Whitney U tests followed by the Bonferroni post hoc test were performed to determine significance differences between groups, after correction, P ≤ 0.0125 is considered as significantly different. a = significance compared with sham/saline, b = significance compared with sham/bleomycin, c = significance compared with adrenalectomy/saline, d = significance compared with adrenalectomy/bleomycin. One letter = P ≤ 0.0125. Two letters = P ≤ 0.005. Three letters = P ≤ 0.001. Four letters = P ≤ 0.0001. BAL, bronchoalveolar lavage.

Table 2.

The composition of BAL based on total numbers of immune cell types

	Sham Saline (n = 15F/15M)	Sham Bleomycin (n = 16F/13M)	Adrenalectomy Saline (n = 14F/13M)	Adrenalectomy Bleomycin (n = 14F/14M)
Total BAL cell count (×106)	3.04 ± 0.0964 (2–5)bbbb	5.467 ± 1.613 (2–10)aaaa	3.464 ± 1.5 (2–10)dddd	4.586 ± 1.18 (3–7) cccc
Macrophages (×106)	3.043 ± 1.16 (1.811–6.801)bb	4.173 ± 1.389 (1.761–9.192)aa	3.207 ± 1.375 (1.785–9.316)cc	4.173 ± 1.697 (2.468–11.086)dd
Monocytes (×106)	0.004 ± 0.006 (0–0.022)b	0.024 ± 0.032 (0–0.122)a	0.006 ± 0.006 (0–0.021)	0.013 ± 0.024 (0–0.117)
Lymphocytes (×106)	0.089 ± 0.042 (0.017–0.541)bbbb	1.121 ± 0.819 (0.108–4.07)aaaa,d	0.154 ± 0.249 (0.019–1.233)dddd	0.633 ± 0.444 (0.177–2.025)b,cccc
Neutrophils (×106)	0.017 ± 0.042 (0–0.221)b	0.059 ± 0.073 (0–0.261)a	0.007 ± 0.008 (0–0.024)dddd	0.037 ± 0.032 (0–0.122)cccc
Eosinophils (×106)	0.006 ± 0.002 (0–0.01)bbbb	0.018 ± 0.032 (0–0.139)aaaa	0.003 ± 0.006 (0–0.02)	0.009 ± 0.023 (0–0.119)
Basophils (×106)	0.002 ± 0.009 (0–0.044)bbbb	0.018 ± 0.021 (0–0.077)aaaa	0.0006 ± 0.002 (0–0.006)dd	0.008 ± 0.013 (0–0.047)cc
Mast cells (×106)	0 ± 0 (0–0)bbbb	0.009 ± 0.013 (0–0.04)aaaa	0 ± 0 (0–0)dddd	0.004 ± 0.012 (0–0.057)cccc
Other (×106)	0.029 ± 0.023 (0–0.08)	0.044 ± 0.068 (0–0.237)	0.031 ± 0.025 (0–0.086)	0.051 ± 0.054 (0–0.212)

Cells in the other category are cells that did not appear to match the criteria used to determine the cell type. Data are shown as means ± SD. The numbers in parentheses represent the range of values observed in the differential count. Multiple unpaired nonparametric Mann–Whitney U tests followed by Bonferroni post hoc test were performed to determine significance differences between groups, after correction, P ≤ 0.0125 is considered as significantly different. a = significance compared with sham/saline, b = significance compared with sham/bleomycin, c = significance compared with adrenalectomy/saline, d = significance compared with adrenalectomy/bleomycin. One letter = P ≤ 0.0125. Two letters = P ≤ 0.005. Three letters = P ≤ 0.001. Four letters = P ≤ 0.0001. BAL, bronchoalveolar lavage.

In addition, immunofluorescence staining of lung tissues for monocytes and macrophages (Fig. 3A) showed that the percentage of CD68⁺ cells significantly increased after bleomycin stimulation, and adrenalectomy markedly reversed this increase (P = 0.0016, Fig. 3B). The percentage of CD14⁺ cells significantly decreased, whereas CD68⁺CD14⁺ cells significantly increased after bleomycin stimulation; however, adrenalectomy had no statistically significant effect on monocyte accumulation in lung issues (Fig. 3, C and D). To explore the effect of circulating immune cells on bleomycin-induced lung inflammation in adrenalectomy mice, peripheral blood mononuclear cells (PBMCs) immunophenotyping was analyzed by FACS staining. The gating strategy for the immune cells in PBMCs is shown in Fig. 4. Cell populations were defined, as shown in Table 3. As shown in Fig. 5, no significant changes were observed in macrophages, monocytes, lymphocytes, or eosinophils in PBMCs. Interestingly, the percentage of neutrophils was significantly reduced by adrenalectomy compared with sham surgery in bleomycin groups (Fig. 5).

Figure 3.

Differential counts of macrophages and monocytes in lung tissue as determined by immunofluorescent staining. Lung tissue was double-stained with antibodies of CD68 and CD14 to determine how macrophage (CD68⁺) and monocyte (CD14⁺) cell accumulation changed as a result of adrenalectomy surgery and bleomycin treatment. A: immunostaining of CD68 and CD14 in the lung tissues of mice that underwent adrenalectomy or sham surgery in the bleomycin model of fibrosis. The representative images used were taken at the same time and under the same conditions. The percentage of cells in the images that are CD68⁺ (B), the percentage of cells in the images that are CD14⁺ (C), and the percentage of cells in the images that are CD14⁺ and CD68⁺ (D). In all images, CD14 is in the green FITC channel, CD68 is in the Texas Red channel, and nuclei are counterstained with DAPI. Scale bar = 100 microns. Data are shown as means ± SD. Multiple unpaired nonparametric Mann–Whitney U tests followed by Bonferroni post hoc test were performed to determine significance differences between groups, after correction, P ≤ 0.0125 is considered as significantly different. Female mice are represented by open shapes, and male mice are represented by closed shapes (n = 4 female/8 male mice in sham saline group; 5 female/6 male mice in sham bleomycin group; 3 female/6 male mice in adrenalectomy saline group; 3 female/5 male mice in adrenalectomy bleomycin group).

Figure 4.

The gating strategy for the immune cells in peripheral blood mononuclear cells (PBMCs) isolated from whole blood. Live/dead separation was done with Live/Dead fixable Aqua (shown as AmCyan). CD45 was tagged with Alexa Fluor 700. Ly6G was tagged with Pe-Cy7. F4/80 was tagged with eFluor 450 (shown as Pacific Blue). CD11B was tagged with PerCP-Cy5.5. CD14 was tagged with PE. CD3 was tagged with PE/Dazzle 594 (shown as Alexa Fluor 594). CD19 was tagged with APC. CD11C was tagged with APC-eFluor 780 (shown as APC-Cy7). SiglecF was tagged with FITC.

Table 3.

Definitions of Cell Types by FACS

Immune Cell Type	Markers Used to Define
All immune cells	CD45+
Macrophages	CD45+, Ly6G−, F4/80+, CD11B+
Monocytes	CD45+, Ly6G−, CD11B+, CD14+
B cells	CD45+, Ly6G−, CD19+
T cells	CD45+, Ly6G−, CD3+
Neutrophils	CD45+, Ly6G+, SiglecF−
Eosinophils	CD45+, Ly6G+, SiglecF+

Figure 5.

The analysis of immunophenotyping of PBMC cells isolated from whole blood. The quantities of different cell types were counted by FACS. A: the total number of macrophages, defined by the markers CD45⁺, Ly6G⁻, F4/80⁺, and CD11B⁺, was determined from 10,000 events. B: the percentage of macrophages in immune cells, which were defined by the marker CD45⁺. C: the total number of monocytes, defined by markers CD45⁺, Ly6G⁻, CD11B⁺, and CD14⁺, was determined from 10,000 events. D: the percentage of monocytes in immune cells. E: the total number of B cells, defined by the markers CD45⁺, Ly6G⁻, and CD19⁺, was determined from 10,000 events. F: the percentage of B cells in immune cells. G: the total number of T cells, defined by the markers CD45⁺, Ly6G⁻, and CD3⁺, was determined from 10,000 events. H: the percentage of T cells in immune cells. I: the total number of neutrophils, defined by the markers CD45⁺, Ly6G⁺, and SiglecF⁻, was determined from 10,000 events. J: the percentage of neutrophils in immune cells. K: the total number of eosinophils, defined by the markers CD45⁺, Ly6G⁺, and SiglecF⁺, were determined from 10,000 events. L: the percentage of eosinophils in the immune cells. Data are shown as means ± SD. Multiple unpaired nonparametric Mann–Whitney U tests followed by Bonferroni post hoc test were performed to determine significance differences between groups, after correction, P ≤ 0.0125 is considered as significantly different. Female mice are represented by open shapes, and male mice are represented by closed shapes (n = 3 female/3 male in each group).

These data indicate that adrenalectomy inhibits bleomycin-induced lung inflammation by regulating the accumulation of macrophages and lymphocytes. The lower level of circulating neutrophils induced by adrenalectomy may be associated with the protection from lung fibrosis. These hypotheses need further investigations.

Adrenalectomy Mitigates Bleomycin-Induced Fibrotic Remodeling

Work by our laboratory and others has shown that catecholamines and aldosterone contribute to the development of tissue fibrosis (7, 8, 23). We therefore reasoned that adrenalectomy may have an effect on tissue remodeling and fibrosis. To investigate this question, we examined Masson’s trichrome-stained FFPE lung tissue sections obtained from bleomycin-treated mice. We found that mice subjected to adrenalectomy had a significant decrease in MAS when compared with the sham surgery group (Fig. 6). This suggests that adrenalectomy may have a protective effect on mice with bleomycin-induced lung fibrosis.

Figure 6.

Adrenalectomy impacts bleomycin-induced lung remodeling. A: in the bleomycin model, trichrome staining was used to evaluate the lungs of mice that underwent adrenalectomy vs. sham surgery. The representative images used were taken at the same time and under the same conditions. B: after treatment with bleomycin, a significant increase in the Modified Ashcroft Score (MAS) was observed in lung tissues of mice from both the adrenalectomy and sham surgery groups. Notably, adrenalectomy resulted in a significant reduction of MAS when compared with the sham-operated counterparts in mice that received bleomycin. Scale bar = 100 microns. Data are shown as means ± SD. Multiple unpaired nonparametric Mann–Whitney U tests followed by Bonferroni post hoc test were performed to determine significance differences between groups, after correction, P ≤ 0.0125 is considered as significantly different. Female mice are represented by open shapes, and male mice are represented by closed shapes (n = 11 female/9 male mice in sham saline group; 9 female/12 male mice in sham bleomycin group; 9 female/9 male mice in adrenalectomy saline group; 10 female/9 male mice in adrenalectomy bleomycin group).

Adrenalectomy Does Not Reduce Collagen Deposition

To further examine mechanisms underlying adrenalectomy’s protective effects in the bleomycin model, we next decided to quantify the soluble and insoluble collagen content of lung tissues, as these measurements provide a biochemical surrogate for the severity of lung fibrosis. Using the Sircol assay, we found that there was no change in either the soluble or insoluble collagen between the sham and adrenalectomy groups in bleomycin-treated mice (Fig. 7). Interestingly, there was an increase in insoluble collagen in control mice after adrenalectomy. Together, these results indicate that adrenalectomy does not reduce collagen deposition in the bleomycin model of pulmonary fibrosis and may actually lead to an increase in overall collagen content. Thus, the improvements in fibrosis observed by histopathologic examination are likely to be the result of changes in the noncollagen components of the ECM.

Figure 7.

The effect of Adrenalectomy on lung collagen accumulation. Both soluble (A) and insoluble (B) collagen were measured in the right lung (RL) tissues of mice from both the adrenalectomy and sham surgery groups. Notably, adrenalectomy resulted in significant reductions of both soluble and insoluble collagens when compared to the sham surgery counterparts in the mice that received vehicle control (saline), but not bleomycin. Data are shown as means ± SD. Multiple unpaired nonparametric Mann–Whitney U tests followed by Bonferroni post hoc test were performed to determine significance differences between groups, after correction, P ≤ 0.0125 is considered as significantly different. Female mice are represented by open shapes, and male mice are represented by closed shapes (n = 8 female/8 male mice in sham saline group; 9 female/6 male mice in sham bleomycin group; 6 female/6 male mice in adrenalectomy saline group; 6 female/6 male mice in adrenalectomy bleomycin group).

Adrenalectomy Reduces αSMA Expression

To better understand the mechanisms behind our observation that surgical ablation of the adrenal glands improves histopathologic markers of fibrosis, we next investigated whether other fibrotic markers were also affected by adrenalectomy. αSMA expressing myofibroblasts are well known to accumulate in fibrotic regions of the lung and contribute to the development and progression of pulmonary fibrosis (24). Hence, we first evaluated the expression of αSMA in lung tissues by IF staining. Using this method, we observed a significant reduction in both the expression area and mean fluorescent intensity (MFI) of αSMA in the adrenalectomy/bleomycin group compared with the sham/bleomycin group, indicating that adrenalectomy is protective against the increased αSMA expression seen in bleomycin-induced lung fibrosis (Fig. 8).

Figure 8.

The effect of Adrenalectomy on αSMA expression in bleomycin challenged mice lung. A: immunostaining of α-smooth muscle actin (αSMA) in the lung tissue of mice that underwent adrenalectomy or sham surgery in the bleomycin model of lung fibrosis. For all images, αSMA is in the green FITC channel, and nuclei are counterstained with DAPI. The representative images used were taken at the same time and under the same conditions. Bleomycin treatment significantly increased both the αSMA positive area relative to sham surgery (B) and the expression intensity (mean fluorescence intensity, MFI) of αSMA relative to sham surgery (C). This was significantly reversed by adrenalectomy. Scale bar = 100 microns. Data are shown as means ± SD. Multiple unpaired nonparametric Mann–Whitney U tests followed by Bonferroni post hoc test were performed to determine significance differences between groups, after correction, P ≤ 0.0125 is considered as significantly different. Female mice are represented by open shapes, and male mice are represented by closed shapes (n = 8 female/8 male mice in sham saline group; 9 female/6 male mice in sham bleomycin group; 6 female/6 male mice in adrenalectomy saline group; 6 female/6 male mice in adrenalectomy bleomycin group).

Adrenalectomy Is Protective against Fibrotic ECM Remodeling

Besides collagen, there are other core ECM proteins associated with fibrotic disease, such as laminin and fibronectin (25–28). To test the hypothesis that adrenalectomy improves remodeling in the lung, we also evaluated these ECM components using IF staining. We first measured the expression of laminin and observed a small but statistically significant increase in laminin expression by MFI in sham mice that received bleomycin (Fig. 9, A and B). Adrenalectomy eliminated this bleomycin-induced augmentation in laminin MFI though did not otherwise decrease laminin expression (Fig. 9, B and C). Thus, although adrenalectomy does not decrease overall laminin expression, it does appear to offer some protection against laminin-associated fibrotic remodeling in the bleomycin model of lung fibrosis. To further verify this observation, we measured the coexpression of laminin and αSMA and found that there was nonsignificantly less overlap in the expression of laminin and αSMA in bleomycin-treated mice subjected to adrenalectomy as opposed to sham surgery (Fig. 9, A and D). Turning to fibronectin, we discovered that adrenalectomy significantly mitigates increases in fibronectin expression caused by bleomycin as determined by MFI and area of expression (Fig. 10, A–C). As with laminin, we also assessed the coexpression of fibronectin and αSMA and found that adrenalectomy effectively protects against increases in fibronectin and αSMA coexpression that are induced by bleomycin (Fig. 10D). Taken together, these results indicate that adrenalectomy protects against fibrotic remodeling of the pulmonary ECM in the bleomycin model by mitigating increases in laminin and fibronectin expression and by reducing the coexpression of fibronectin with αSMA.

Figure 9.

Adrenalectomy impacted the correlation between αSMA and laminin. Through immunofluorescent imaging, we investigated the alterations in laminin expression and its coexpression with α-smooth muscle actin (αSMA). A: immunostaining of laminin (red) and αSMA (green) in the lung tissue of mice that underwent adrenalectomy or sham surgery in the bleomycin model of fibrosis. The representative images used were taken at the same time and under the same conditions. B: mean fluorescence intensity (MFI) of laminin was significantly increased by bleomycin in sham but not adrenalectomy groups. C: there was no significant change of laminin positive area among groups. D: adrenalectomy nonsignificantly reduced the double positive area of αSMA and laminin relative to sham surgery. E: the overall area of laminin expression is unchanged between groups. Scale bar = 100 microns. 1 pixel = 0.32 μm². Data are shown as means ± SD. Multiple unpaired nonparametric Mann–Whitney U tests followed by Bonferroni post hoc test were performed to determine significance differences between groups, after correction, P ≤ 0.0125 is considered as significantly different. Female mice are represented by open shapes, and male mice are represented by closed shapes (n = 8 female/8 mice in male sham saline group; 9 female/6 male mice in sham bleomycin group; 6 female/6 male mice adrenalectomy saline group; 6 female/6 male mice adrenalectomy bleomycin group).

Figure 10.

Adrenalectomy reduced fibronectin expression and impacted the correlation between αSMA and fibronectin. Through immunofluorescent imaging, we investigated the alterations in fibronectin expression and its coexpression with α-smooth muscle actin (αSMA). A: representative images of the immunostaining of fibronectin (red) and αSMA (green) are shown. The representative images used were taken at the same time and under the same conditions. B: the mean fluorescence intensity (MFI) of fibronectin increased in the sham-bleomycin group and was significantly less after adrenalectomy. C: the average area of fibronectin significantly increased following bleomycin stimulation in the mice undergoing sham surgery. However, adrenalectomy effectively reversed this increase. D: the average area where fibronectin and αSMA overlapped significantly increased following bleomycin stimulation in the mice undergoing sham surgery, while adrenalectomy effectively nullified this increase. For all images αSMA is in the green, laminin is in the red, and nuclei are counterstained with DAPI. Scale bar = 100 microns. Data are shown as means ± SD. Multiple unpaired nonparametric Mann–Whitney U tests followed by Bonferroni post hoc test were performed to determine significance differences between groups, after correction, P ≤ 0.0125 is considered as significantly different. Female mice are represented by open shapes, and male mice are represented by closed shapes (n = 8 female/8 male mice in sham saline group; 9 female/6 male mice in sham bleomycin group; 6 female/6 male mice adrenalectomy saline group; 6 female/6 male mice adrenalectomy bleomycin group).

Adrenalectomy Ablates Adrenaline but Not Noradrenaline Levels

The adrenal glands play a pivotal role in the production and secretion of AD and NA. In contrast to AD that is primarily synthesized in the adrenal medulla, lung NA is primarily derived from tissue-resident post-ganglionic neurons (29, 30). We hypothesized that the protective effects of adrenalectomy in lung fibrosis would be induced by marked reductions systemic and pulmonary levels of adrenally derived catecholamines. To validate this hypothesis, we first evaluated AD levels in plasma, BAL, and lung homogenate by ELISA and found that AD was significantly reduced after adrenalectomy (Fig. 11, A–C). These data confirm the effectiveness of the adrenalectomy in reducing circulating AD levels. We similarly evaluated NA concentrations in plasma, BAL, and lung homogenate. NA levels in lung homogenate and plasma were unaffected by adrenalectomy, consistent with prior data published by our laboratory (7) but were reduced in the BAL of saline-control mice subjected to adrenalectomy (Fig. 12, A–C). The impact of this reduction on lung fibrosis remains to be explored.

Figure 11.

The concentrations of AD in mice underwent sham surgery or adrenalectomy in the bleomycin model. The adrenaline (AD) concentrations in plasma (A), BAL (B), and lung tissues (C) are shown. Data are shown as means ± SD. Multiple unpaired nonparametric Mann–Whitney U tests followed by Bonferroni post hoc test were performed to determine significance differences between groups, after correction, P ≤ 0.0125 is considered as significantly different. Female mice are represented by open shapes, and male mice are represented by closed shapes (n = 5 female/5 male mice in each group). BAL, bronchoalveolar lavage.

Figure 12.

The concentrations of NA in mice underwent sham surgery or adrenalectomy in the bleomycin model. The noradrenaline (NA) concentrations in plasma (A), BAL (B), and lung tissues (C) are shown. Data are shown as means ± SD. Multiple unpaired nonparametric Mann–Whitney U tests followed by Bonferroni post hoc test were performed to determine significance differences between groups, after correction, P ≤ 0.0125 is considered as significantly different. Female mice are represented by open shapes, and male mice are represented by closed shapes (n = 5 female/5 male mice in each group). BAL, bronchoalveolar lavage.

Adrenalectomy Affects Aldosterone Production

Aldosterone, secreted by the adrenal glands, is the final common mediator of the RAAS axis (31), and this system has been shown to interact with profibrotic pathways such as TGFβ1 to significantly impact fibrosis in numerous tissues (32, 33). To gain a more robust understanding of how the processes mediated by the adrenal gland impact pulmonary fibrosis, we measured systemic and local aldosterone concentration in plasma, BAL, and lung tissues by ELISA (Fig. 13). As expected, we found that the aldosterone concentrations in plasma and BAL were mostly eliminated after adrenalectomy. However, aldosterone was detected in the lung tissues of mice that underwent adrenalectomy with a nonstatistically significant increasing trend observed following bleomycin stimulation. Hence, we speculate that locally derived ALD from lung tissues may contribute to the ECM overdeposition in the absence of systemic adrenal hormones in the bleomycin model. Nevertheless, further experiments will be needed to better understand the impact of aldosterone on pulmonary fibrosis.

Figure 13.

The concentration of aldosterone (ALD) in mice underwent sham surgery or adrenalectomy in the bleomycin model. The concentrations of aldosterone (ALD) in plasma, BAL, and right lung lower lobes (RLL) were tested by ELISA. A: plasma ALD concentrations (n = 8 female/8 male mice in sham saline group; 6 female/9 male mice in sham bleomycin group; 4 female/6 male mice in adrenalectomy saline group; 5 female/6 male mice in adrenalectomy bleomycin group). ALD concentrations of BAL supernatant (B) and lung tissues (C) (n = 3 female/3 male in each group). Data are shown as means ± SD. Multiple unpaired nonparametric Mann–Whitney U tests followed by Bonferroni post hoc test were performed to determine significance differences between groups, after correction, P ≤ 0.0125 is considered as significantly different. Female mice are represented by open shapes, and male mice are represented by closed shapes. BAL, bronchoalveolar lavage.

DISCUSSION

An increasing number of studies have demonstrated that catecholamines such as AD and NA are associated with fibrosis (7, 8, 34, 35). However, the contribution of systemically circulating adrenally produced catecholamines to pulmonary fibrosis is less well understood. Here, we investigated the role of surgical adrenalectomy in pulmonary fibrosis using the bleomycin model of experimental lung fibrosis. We found that relative to sham surgery, adrenalectomy significantly decreased lung inflammation by regulating the accumulation of lymphocytes and macrophages. Although adrenalectomy increased insoluble collagen concentration in uninjured mice, there was no discernible difference in bleomycin-induced collagen accumulation. Interestingly, histologic indices of bleomycin-induced lung remodeling were reduced by adrenalectomy based on trichrome-stained images and MAS data. These findings were accompanied by concomitant reductions in αSMA and fibronectin and mild amelioration of bleomycin-induced increases in laminin. Whereas adrenalectomy abrogated AD detection in all conditions and compartments, it only reduced BAL NA in uninjured mice, though this change was not statistically significant. Although the systemic aldosterone concentration was significantly decreased in the mice that underwent adrenalectomy, aldosterone was found in the lung tissues, even showing a nonstatistically significant increase after bleomycin treatment.

There are numerous proinflammatory mechanisms that are proposed to exist in the pathogenesis and progression of pulmonary fibrosis (36). In our model, bleomycin-induced lung injury is proinflammatory, which coupled with a disordered repair response leads to fibrosis (37). Based on the prior work showing that catecholamines and RAAS activation can lead to an increase in inflammatory mediators and lung inflammation (8, 13), we reasoned that the protective effects of adrenalectomy in the bleomycin model may be in part related to the reduction in inflammation. Indeed, lung inflammation in our model was significantly, though not completely, ameliorated by adrenalectomy, as further indicated by the few changes to BAL cell type composition found through our differential. Significant decreases in the number of lymphocytes along with an increase in the percentage of BAL that is macrophages demonstrate an altered immune response to bleomycin after adrenalectomy. The presence of an altered immune response after adrenalectomy is further supported by the decrease in CD68⁺ macrophages present in the lung tissue of adrenalectomy mice administered bleomycin. These data corroborate prior research that circulating adrenally derived substances impact lung fibrosis, at least in part, by altering inflammatory responses in the lung. Although corticosteroid monotherapy was an unsuccessful treatment to IPF (38), our data indicated that expanding and deepening the understanding of adrenal-immune communication pathways and their contribution to lung fibrosis opens promising avenues for potential targeted interventions, including the repurposing of existing receptor agonists and antagonists targeting proinflammatory adrenal pathways. Further work will be needed to resolve the exact mechanisms by which this occurs and which specific mediators are involved.

Catecholamines have also been associated with the development of fibrosis by histologic indices and by expression of collagen type 1 (9). We expand upon these results by showing that adrenalectomy improved but did not completely rescue the lung from histologically assessed fibrosis in the bleomycin model. Surprisingly, and in contrast to Rassler et al.’s (9) findings, adrenalectomy resulted in a significant increase in insoluble collagen in mice treated with saline control but there was no difference observed in mice that received bleomycin (Fig. 7B). This suggests that the protective effects of adrenalectomy are not related to collagen secretion and deposition and implicates adrenally derived substances as regulators of ECM remodeling. Indeed, while both NA and ALD have been reported to promote fibrosis (7, 8, 31–33), AD has been shown to exhibit antifibrotic effects (34, 35). Furthermore, our results show that there was almost no significant difference between the NA concentrations of the various groups tested, indicating that most of the NA seen is nerve-derived and thus not affected by adrenalectomy. This is consistent with previous findings and implies that targeting locally derived NA is more likely to benefit fibrotic endpoints (8). In these ways, adrenalectomy can result in opposing effects in the lung by improving histologic markers of fibrosis but contributing to increased collagen deposition. While further investigation is needed to better understand these relationships, and though total adrenal ablation may not be an effective treatment option, our data highlight the importance of a lung-adrenal axis in pulmonary fibrosis.

αSMA and αSMA-expressing myofibroblasts are well known to be important in the development of lung fibrosis (24, 39). Consistent with previous publications (7, 40), we noted that bleomycin significantly increased the expression of αSMA but that this increase was reversed by adrenalectomy. As αSMA is the principal biomarker of myofibroblasts, our observation suggests that adrenalectomy may result in reduced differentiation of cells into αSMA-expressing myofibroblasts thereby offering some protection against bleomycin-induced fibrosis. Because collagen content did not change between sham and adrenalectomy groups that received bleomycin, we reasoned that the histologic changes we observed must come from other components of the ECM. We thus tested two core ECM proteins, laminin and fibronectin, that are associated with fibrosis (25–28, 41) and found that laminin MFI significantly increased in bleomycin-treated mice that underwent sham surgery but not adrenalectomy. This indicates that laminin expression is increased in mice that received bleomycin, consistent with prior data (25), and suggests that adrenalectomy may be protective against increases in laminin expression. In addition, the expression of fibronectin and its coexpression with αSMA were increased in bleomycin-treated animals and was significantly abrogated by adrenalectomy. Taken together, our results indicate that adrenally derived substances are important mediators of fibrotic lung remodeling and show that adrenalectomy protects against the development of fibrosis by ameliorating increases in αSMA and critical ECM components. Further investigation into these mechanisms could yield more targeted treatments for pulmonary fibrosis.

We measured the concentration of aldosterone in plasma, BAL supernatant, and homogenized lung tissue to take a preliminary look at the activity of the RAAS axis and found that, as expected, there was a significant reduction of systemic aldosterone in the plasma and almost no ALD present in BAL supernatant after adrenalectomy. However, we found that there was no change in aldosterone concentration in homogenized lung tissue between the sham surgery and adrenalectomy mice that were administered bleomycin. This indicates that locally produced aldosterone is unchanged, potentially upregulated, when responding bleomycin treatment in adrenalectomy mice. Although aldosterone is primarily secreted by the adrenal glands, the potential of a RAAS pathway localized within the lungs could be implicated by these results since there is little aldosterone circulating through the blood after adrenalectomy, as determined by these results. However, further research is necessary to confirm the source of the aldosterone and its role in pulmonary fibrotic pathways. In addition to the ECM remodeling effects noted above, the expression of ALD may provide another explanation for the protective effects of adrenalectomy in the bleomycin model. Previous research has demonstrated that RAAS antagonists are protective against bleomycin-induced lung injury in patients receiving bleomycin chemotherapy (42) and angiotensin-converting enzyme inhibitors have been shown to be effective in the mitigation of fibrosis caused by the SARS-CoV-2 virus (43). These results combined with our findings demonstrate a link between the RAAS pathway and ECM remodeling observed in fibrosis. Although our results are promising, further research is needed to understand the interactions between RAAS activity and pulmonary fibrosis.

Perspectives and Significance

This study underscores the importance of the adrenal gland as a regulator of pulmonary fibrosis. Mice that underwent adrenalectomy were more resistant to the harmful effects of bleomycin and exhibited changes in ECM deposition and orientation. However, there are significant drawbacks. By increasing insoluble collagen content, total adrenalectomy clearly has opposing effects on fibrotic remodeling that will require additional study to clarify. Future studies should also evaluate other elements of the ECM such as elastin and the different collagens to gain a better understanding of the specific effects of adrenalectomy. Adrenalectomy is also not itself a viable treatment modality for patients with pulmonary fibrosis due to the extensive off-target effects and resulting adrenal insufficiency that such a surgery would produce. Nevertheless, the proinflammatory and profibrotic pathways that are associated with adrenally derived catecholamines and aldosterone represent exciting areas for potential targeted interventions, including the repurposing of already available receptor agonists and antagonists. A better understanding of the biology underlying adrenalectomy’s protective effects in pulmonary fibrosis will be fundamental to translating our promising results into effective therapeutic strategies for patients suffering from this devastating disease, or even other fibrosis diseases.

DATA AVAILABILITY

Data will be made available upon reasonable request.

GRANTS

H.S. was supported by US Department of Defense (DOD) Award W81XWH-20-1-0157, ATS Foundation and Scleroderma Foundation. A.G. was supported by 5T32HL007778-27/28 from the NIH. E.L.H. was supported by R01HL152677, and R01HL163984 from the NIH, the Gabriel and Alma Elias Research Fund, and the Greenfield Foundation.

DISCLAIMERS

The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health or the Department of Defense.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

J.M., C.P., E.L.H., and H.S. conceived and designed research; J.M., C.P., A.G., and H.S. performed experiments; J.M., C.P., A.G., E.L.H., S.S., and H.S. analyzed data; J.M., C.P., A.G., E.L.H., and H.S. interpreted results of experiments; J.M., C.P., and H.S. prepared figures; J.M., C.P., A.G., and H.S. drafted manuscript; J.M., C.P., A.G., E.L.H., S.S., and H.S. edited and revised manuscript; J.M., C.P., A.G., E.L.H., S.S., and H.S. approved final version of manuscript.