https://orcid.org/0000-0002-7441-8560
Abstract
Inhalation of environmental toxicants such as cigarette smoke, metal or wood dust, silica, or asbestos is associated with increased risk for idiopathic pulmonary fibrosis (IPF). IPF involves progressive scarring of lung tissue, which interferes with normal respiration and is ultimately fatal; however, the complex cellular mechanisms of IPF pathogenesis remain unclear. Fibroblast apoptosis is essential in normal wound healing but is dysregulated in IPF. Recent studies suggest that Toll-like receptor 4 (TLR4) is key in the onset of IPF. Here, radiation-induced PF was used as a model for IPF because it very closely mimics the progressive and intractable nature of IPF. Female C57BL/6J (C57) and C57BL/6J TLR4−/− mice were exposed to a single dose of 13 Gy whole-thorax ionizing radiation. Although both strains showed similar levels of immediate radiation-induced damage, C57 mice exhibited more extensive fibrosis at 22-week postirradiation (PI) than TLR4−/− mice. Isolated C57 primary 1° MLFs showed decreased apoptosis susceptibility as early as 8-week postirradiation, a phenotype that persisted for the remainder of the radiation response. TLR4−/− 1° mouse lung fibroblasts did not exhibit significant apoptosis resistance at any point. Systemic release of high mobility group box 1, a TLR4 agonist, during the pneumonitis phase of the radiation response may act through TLR4 to contribute to fibroblast apoptosis resistance and thus interfere with wound resolution. These findings demonstrate that apoptosis resistance occurs earlier in pulmonary fibrosis pathogenesis than previously assumed, and that TLR4 signaling is a key mediator in this process.
Pulmonary fibrosis (PF) is a progressive and intractable interstitial lung disease involving the accumulation of lung scarring, resulting in a decline in pulmonary functional capacity and ultimately early death (Spagnolo et al., 2015). Patients with PF depend on lung transplantation for primary treatment, as pharmacological therapies currently approved for PF have limited effectiveness (Raghu et al., 2011). PF has a diverse array of origins and risk factors. Idiopathic PF (IPF) is associated with inhalation of environmental toxicants, such as cigarette smoke, metal or wood dust, microbial agents, as well as with certain polymorphisms in genes for surfactant protein C, surfactant protein A2, and mucin 5B (Nogee et al., 2001; Seibold et al., 2011; Spagnolo et al., 2015; Wang et al., 2009). Silicosis and asbestosis are other forms of PF and are usually associated with occupational exposures (Lazarus and Philip, 2011; Pollard, 2016). Cancer patients are also at increased risk of developing radiation-induced PF (RIPF) as a consequence of thoracic radiotherapy (Mehta, 2005). Although the aforementioned causes and risk factors for PF are varied, they all have one after-effect in common: chronic epithelial damage (Korfei et al., 2008; Mittal et al., 2014; Mossman and Churg, 1998; Oberdörster, 1995; Riley, 1994). Though it is now widely accepted that epithelial damage is a common thread in the causative agents and risk factors for PF (Zoz et al., 2011), the mechanism linking sustained epithelial damage and subsequent chronic lung inflammation to the progressive accumulation of scar tissue remains unknown.
In the context of normal wound healing, an acute epithelial injury results in healthy inflammation, then subsequent fibroblast recruitment to begin the process of reepithelialization. Once fibroblasts are recruited to the wound from the surrounding tissue, they transform into myofibroblasts in response to a transforming growth factor-β (TGF-β) signal. Myofibroblasts are more proliferative and produce more extracellular matrix (ECM) components to aid reepithelialization (Darby et al., 2014). Once the area is prepared for epithelial regeneration, myofibroblasts undergo apoptosis, thus allowing normal wound healing to proceed (Greenhalgh, 1998). In contrast, PF is now viewed as a disease of aberrant wound healing, as fibrotic myofibroblasts do not respond appropriately to apoptotic signals (Gauldie et al., 2002; Moodley et al., 2004; Selman et al., 2001; Thannickal and Horowitz, 2006). Apoptosis resistant myofibroblasts then continue to proliferate and lay down ECM, and the lack of response to the extrinsic apoptotic signal results in the progressive nature of scarring in PF.
In fibrosis research, in vivo mouse models are the cornerstone of studying the complex interplay of different cell types in disease pathogenesis. The C57BL/6J (C57) mouse is a popular fibrosis-prone strain, whereas the C3H/HeJ mouse is a commonly used fibrosis-resistant strain (Rubin et al., 1995; Sharplin and Franko, 1989). The C3H/HeJ strain possesses a loss-of-function mutation in the Tlr4 gene, resulting in transcription of nonfunctional Toll-like receptor 4 (TLR4) protein (Qureshi et al., 1999). In clinical research, TLR4 levels are higher in lung biopsies and serum samples from IPF patients than in samples from control patients (Ahmed et al., 2016). Additionally, a genome-wide association study found that a loss-of-function polymorphism in TOLLIP, a gene encoding the inhibitor of TLR4, was associated with increased risk for developing IPF (Noth et al., 2013). TLR4 is a pattern recognition receptor that binds bacterial lipopolysaccharide, to propagate a systemic innate immune response (Akira and Takeda, 2004; Chaudhuri et al., 2005). However, TLR4 is also activated by many endogenous damage-associated molecular patterns (DAMPs) present in the aftermath of cellular damage, such as the high mobility group box 1 protein (HMGB1) (Tsung et al., 2014). Additionally, TLR4 activation results in transcription of prosurvival signaling factors via NF-κB and PI3K-Akt activation (He et al., 2012; Wang et al., 1998).
Based on these insights, we hypothesized that HMGB1/TLR4 signaling precedes scar formation and contributes significantly to fibroblast apoptosis resistance in the onset of PF. We investigated the lung fibrosis susceptibility of TLR4−/− mice in comparison to fibrosis-prone C57 controls, and additionally characterized the pathology of both strains, including the appearance and possible origins of apoptosis resistant fibroblasts throughout the stages of fibrosis pathogenesis. We addressed these questions by utilizing a model of RIPF. RIPF not only has implications for cancer patients receiving radiotherapy, but it is also the model that most closely mimics the stages and progression of IPF, which concerns a far-reaching demographic including patients affected by occupational exposures, environmental hazards, and familial genetic predispositions.
MATERIALS AND METHODS
Materials
Cell culture medium (Dulbecco's Modified Eagle Medium (DMEM)/F12, Minimum Essential Medium (MEM), and DMEM; cat. no. 11330, 11095, and 11965), fetal bovine serum (FBS; cat. no. 25140), L-glutamine (cat. no 25030), penicillin/streptomycin (P/S; cat. no. 15070063), gentamicin (cat. no. 15750), nonessential amino acids (NEAA; cat. no. 11140), sodium pyruvate (cat. no. 11360), trypsin-EDTA, dispase (cat. no. 17105), phosphate-buffered saline (PBS) pH 7.2 (cat. no. 20012), and Hank’s Balanced Salt Solution (cat. no. 14175) were obtained from Gibco, Inc. (Gaithersburg, Maryland). IncuCyte Caspase-3/7 Green Reagent (cat. no. 4440) was obtained from Essen Bioscience (Ann Arbor, Michigan). Staurosporine (STS; cat. no. S6942) was obtained from Sigma-Aldrich (St Louis, Missouri). ELISA kit for HMGB1 (cat. no. SEA399Mu) was obtained from Cloud-Clone Corp (Katy, Texas). Rabbit monoclonal antibody (mAb) for γ-H2AX (cat. no. 9718) and Hoechst 33342 (cat. no. 4082) was obtained from Cell Signaling Technology (Danvers, Massachusetts). Rabbit polyclonal antibodies for alpha-smooth muscle actin (α-SMA; cat. no. ab5694) and secretoglobin 1A1 (SCGB1A1; also known as Club cell secretory protein or CCSP; cat. no. ab40873) were obtained from Abcam (Cambridge, Massachusetts). Rat mAb to F4/80 clone Cl: A3-1 (cat. no. MCA497) was obtained from Bio-Rad (formerly Serotec; Kidlington, Oxford, UK). Dako Target Retrieval Solution (cat. no. S1699) and Dako Tris-buffered NaCl solution with Tween 20 (TBS-T; cat. no. S3306) were obtained from Agilent Technologies, Inc. (Santa Clara, California). Peroxidased 1 (cat. no. PX968H), Rodent Block M (cat. no. RBM961L), Renaissance Diluent (cat. no. PD905L), Rabbit on Rodent HRP Polymer (cat. no. RMR622H), Rat Probe Kit (cat. no. RT517H), Betazoid DAB Chromogen (cat. no. BDB900F), 3,3′-Diaminobenzidine (DAB) Substrate Buffer (cat. no. DS900H), and CAT Hematoxylin (cat. no. CATHE-MM) were obtained from Biocare Medical (Pacheco, California).
Animals
Female C57 (C57BL/6J, No. 000664) and TLR4−/− (B6.B10ScN-Tlr4lps-del/JthJ, No. 007227) mice, 6 weeks of age, were obtained from Jackson Laboratory (Bar Harbor, Maine). Female mice are used in this model of RIPF due to the differences in radiation susceptibility between male and female mice, and also due to the relative ease of housing female mice for long-term studies. The TLR4−/− mice used here completely lack the TLR4 receptor as a result of a genetic deletion, which was originally discovered in C57BL/10ScN mice then backcrossed onto the C57BL/6 background. All mice were housed 5 animals per cage under pathogen-free conditions and acclimated for at least 2 weeks before experimentation. Animals were fed standard laboratory diet and water ad libitum. The University Committee on Animal Resources approved all animal protocols.
Irradiation
Mice were restrained in plastic jigs and received 13 Gy thorax-only irradiation with a 137Cs γ-ray source operating at a dose rate of 1.45 Gy/min. This dose of thoracic radiation has previously been shown to produce a fibrotic outcome in female C57BL/6J mice as a late effect of the radiation response (Johnston et al., 1995). Age-matched sham-irradiated (0 Gy) control mice received identical handling. Mice were weighed biweekly or weekly after irradiation to monitor morbidity (data not shown).
Sample collection, fibroblast isolation, and cell culture
Samples were collected at 24-h, 8-, 12-, 16-, and 22-week postirradiation (PI). Briefly, mice were euthanized, blood samples were collected, and lungs were perfused with sterile PBS. Right lungs were extracted for fibroblast isolation; then left lungs were inflated with zinc-buffered formalin (ZBF) and extracted. Left lungs for histology were incubated in ZBF with gentle agitation for 24 h at RT, rinsed with ddH2O twice, then stored in 70% ethanol at 4°C until lungs were paraffin embedded and sectioned.
Primary mouse lung fibroblasts (MLFs) were isolated as described by Seluanov et al. (2010), with minor changes. All mouse work and subsequent tissue processing were carried out in culture hoods to avoid contamination of cell isolates. Right lungs were minced and digested in DMEM/F12 containing 1.8 U/ml dispase + 0.02% gentamicin at 37°C for 90 min, then rinsed and plated with expansion media (DMEM/F12 containing 15% FBS + 0.02% gentamicin) to facilitate the growth of primary lung fibroblasts from lung fragments. Lung fragment cells were allowed to expand in this media for 2 weeks, with media changes every 2–3 days. Then cells were trypsinized and media was switched to selection media (MEM containing 15% FBS + 1× P/S + 1× NEAA + 1× sodium pyruvate + 0.02% gentamicin) to select for only fibroblast proliferation. In preliminary studies, qualitative immunofluorescence for CD326 and CD31 showed no evidence of proliferation of epithelial or endothelial cells (data not shown). Isolated cells exhibited an elongated, spindle-shaped structure or prominent lamellipodia, and proliferated into a tightly packed monolayer when approaching confluence, both indicative of fibroblast identity. At this point, the population was considered p0. For experiments described here, MLFs were used at p1. MLFs were grown in a humidified incubator at 37°C in 5%–7% O2 and 5% CO2.
Immunohistochemistry and staining quantification
Left lungs were inflated with ZBF and 6 μm paraffin-embedded sections were prepared. Sections were deparaffinized with xylene and rehydrated through graded alcohols and antigen retrieval was performed with Dako Target Retrieval Solution at 95°C for 20 min. Sections were blocked with Biocare Medical Peroxidazed 1 for 15 min then Rodent Block M for 30 min at RT. Primary antibody incubations were performed at the following dilutions in Renaissance Diluent, duration, and temperatures: α-SMA 1:200 for 1 h at RT, SCGB1A1 1:2000 for 1 h at RT, F4/80 at 1:100 for 16 h at 4°C, and γ-H2AX at 1:500 for 18 h at 4°C. Sections were incubated with secondary HRP antibody, then developed with DAB chromogen until staining developed and quenched with water. Sections were counterstained with hematoxylin and TBS-T to stain nuclei, dehydrated and cleared through graded alcohols and xylene, and cover-slipped. Images were obtained with an Olympus BX40 Research Microscope and Nuance 3.0.2 acquisition software at magnifications indicated in figure legends.
Images obtained for γ-H2AX and F4/80 quantification focused on alveoli and did not include peribronchial or perivascular regions, for consistency across images. Quantification of γ-H2AX abundance was obtained by counting total positively stained nuclei per field of view in each group (n = 3–5 FOV per lung section; n = 2–5 mice per group) with ImageJ v1.50b software’s manual cell counter plugin. Quantification of F4/80 abundance was obtained by counting total positively stained cells per FOV and normalizing to number of alveoli per FOV (n = 3–6 FOV per lung section; n = 3–4 mice per group). SCGB1A1 abundance was quantified by tracing the perimeter of all similarly sized airways (approximate range of airway diameters 250–500 μm) in each section with ImageJ v1.50b software and comparing to negatively stained perimeter (n = 4–7 airways per lung section, and n = 5 mice per group).
Fibrosis quantification
Paraffin-embedded lung sections harvested at 22-week PI were Gomori trichrome-stained and imaged at 12.5× and 400×. Low-power images of whole left lung sections were used for percent fibrosis quantification. Using ImageJ v1.50b software, the cumulative area of all fibrotic foci in a lung section was quantified, and then normalized to total area of the lung section to obtain a percentage. The identification of fibrotic foci was aided with images taken at higher magnification to differentiate between fibrosis and other dense lung tissue, such as lung tissue with high inflammatory infiltrates or lung tissue that was not fully inflated. Areas of lung tissue with low-grade fibrosis, such as areas with slight septal thickening, were not included as being part of a fibrotic mass for this assessment.
Fibrosis was also assessed by modified Ashcroft scoring, which is more sensitive to slight changes in histology. First, 400× original magnification images were obtained from 22-week PI trichrome-stained lung sections in a predetermined pattern for each lung section; since fibrosis begins at the periphery of the lung and progresses inwards, 3 FOV each were obtained at the superior and inferior ends of the lobes, as well as 2 FOVs each at both lateral edges, for a total of 10 images per lung. The identification numbers of all images were blinded by a colleague. Then, all images were scored on a scale of 0 through 8 based on standardized pathology guidelines detailed by Hübner et al. (2008). A score of 0 indicates no fibrotic burden; 1 indicates isolated septal thickening; 2 indicates isolated septal thickening but with knot-like formation; 3 indicates contiguous fibrotic alveolar walls. Images with any fibrotic masses are assigned a score of 4 or above; a score of 4 indicates single fibrotic masses in ≤ 10% of field; 5 indicates connected fibrotic masses in 10%–50% of field; 6 indicates contiguous fibrotic masses in >50% of field; 7 indicates alveolar structure is “nearly obliterated with fibrotic masses but still up to 5 air bubbles”; and 8 indicates complete obliteration of lung structure with fibrotic masses. Images were reidentified, and then the final modified Ashcroft score of each lung was obtained by averaging the scores of all 10 images.
Apoptosis susceptibility quantification
Fibroblast apoptosis susceptibility was quantified as previously described (Hanson and Finkelstein, 2019). Briefly, after isolation from pulmonary tissue via ex vivo culture, p1 MLFs were seeded into 96-well plates and allowed to establish for 2 days prior to treatment. Cells were treated with 1× Essen Bioscience IncuCyte Caspase-3/7 Green Reagent (final concentration 5 μM) and 1 μM STS or vehicle. Each treatment group was carried out with n = 4 wells.
Treated plates were analyzed with SpectraMax M5 ROM v3.0.22 multimode microplate reader and corresponding SoftMax Pro v6.4 Microplate Data Acquisition and Analysis software. Treated plates were read immediately after treatment (0 h) and again at 8, 12, 16, 20, and 24 h after treatment. At each timepoint, fluorescent signal was quantified with excitation and emission wavelengths of 500 and 530 nm, respectively. First, raw data were reduced by subtracting mean background signal of untreated (blank) wells. Linear regression was performed on timecourse data for each sample, yielding a slope value corresponding to the robustness of the apoptotic response over 24 h, as described previously (Hanson and Finkelstein, 2019). Then, data in each treatment group were normalized to the mean of the respective control group. Next, ranges of control groups were scaled to account for intertimepoint variability, and the same transformation was applied to the corresponding treatment group to present data as percentages relative to controls.
Enzyme-linked immunosorbent assay for HMGB1
HMGB1 content was measured by ELISA following the instructions of the manufacturer. Plasma samples were diluted 1:100 for the assay and final results accounting for dilution are expressed in μg/ml. The limit of detection for this assay is 18.29 pg/ml with a detection range of 46.88–3, 000 pg/ml. Data analysis by 2-way ANOVA revealed that mouse strain did not have a statistically significant effect on the variation observed in the data; therefore, Dunnett’s multiple comparisons post hoc test was conducted to compare sham-irradiated groups to each timepoint PI, regardless of strain.
Statistical analysis
Data were visualized and analyzed by GraphPad Prism v6.0 software. Data were analyzed by 2-way ANOVA, followed by Tukey’s, Sidak’s, or Dunnett’s multiple comparisons post hoc test, as detailed in figure legends.
RESULTS
C57 and TLR4−/− Mice Show Similar Levels of Acute Radiation Injury
To assess whether TLR4 affects the extent of acute radiation injury in vivo, lung sections were first immunostained for γ-H2AX, a common marker of DNA damage. Quantification of γ-H2AX staining indicated that while control lungs exhibited little to no γ-H2AX positive nuclei, significant staining was present in lung tissue of both strains at 24-h PI, and diminished by 8-week PI (Figure 1 ). Though the level of γ-H2AX staining observed at 8-week PI was not significantly different from sham-irradiated controls in this study, the presence of low levels DNA damage at this timepoint is consistent with previous literature. DNA damage has been previously shown to persist or recur throughout the stages of the pulmonary radiation response, and as late as 32-week PI (Beach et al., 2018). Control lungs exhibited little to no γ-H2AX positive nuclei. No significant difference was detected between strains in the sham-irradiated, 24-h PI, or 8-week PI treatment groups.
Prevalence of γ-H2AX as a marker of DNA damage at 24-h and 8-week PI. Representative images are shown of lung sections immunostained for γ-H2AX from C57 (A–C) and TLR4−/− (D–F) mice unirradiated and at 24-h and 8-week PI. Original magnification 400×; scale bar indicates 50 μm. Insets in lower left corners of (A–F) show enlarged view of indicated area of image for clarity. γ-H2AX abundance was determined by counting total positively stained nuclei per FOV in each group (n = 3–5 FOV per section; n = 3–5 sections per group). Mean and standard deviation of counts shown in (G), where (a) indicates a significant difference from sham-irradiated of same strain, and (b) indicates a significant difference from 24-h PI (post-irradiation) of same strain. Statistical significance was determined by 2-way ANOVA and Tukey’s multiple comparisons across all groups.
Radiosensitivity of each strain was also assessed by immunostaining for SCGB1A1 or CCSP. The loss of SCGB1A1-positive cells in the airways at 12-week PI has previously been used as a marker of initial radiation-induced epithelial damage (Groves et al., 2018; Manning et al., 2013). Quantification of SCGB1A1 in airways of lungs harvested at 12-week PI showed a loss of SCGB1A1-positive cells lining the airways in irradiated mice, consistent with previous literature. However, there was no statistically significant difference in SCGB1A1 prevalence between strains (Figure 2).
Loss of SCGB1A1 abundance in airways as a measure of radiation-induced epithelial damage at 12-week PI. Representative images are shown of airways in lung sections immunostained for SCGB1A1 from C57 sham-irradiated (A), TLR4−/− sham-irradiated (B), C57 13 Gy (C), and TLR4−/− 13 Gy (D) treatment groups. Original magnification 200×; scale bar indicates 125 μm. SCGB1A1 abundance was quantified by tracing the perimeter of all similarly sized airways in each section and comparing to negatively stained perimeter (E). In (E), bars indicate median, interquartile range, and range of each group. *indicates a significant difference between control versus irradiated within each strain. Statistical significance was determined by 2-way ANOVA and Tukey’s post hoc test (n = 4–7 similarly sized airways per lung section, and n = 5 mice per group).
C57 and TLR4−/− Mice Demonstrate Sustained Pneumonitis During Intermediate Radiation Response
To further characterize the radiation response of TLR4−/− mice compared with C57 mice, immunostaining for F4/80 (a general macrophage marker) was completed. Subsequent quantification showed that while there was no increase in macrophage density at 24-h PI in irradiated lungs compared with control, both strains showed significant increases at 8- and 12-week PI (Figure 3). The sustained elevation in macrophage density indicates chronic lung inflammation, or radiation pneumonitis. TLR4−/− irradiated samples generally showed more macrophage density than C57 irradiated samples, which agrees with previous work showing that TLR4-defective or -deficient mice exhibit more pronounced sterile inflammatory responses than wild type mice with functioning TLR4 (Groves et al., 2015; Zhao et al., 2010).
Extent of inflammation in C57 versus TLR4−/− lungs at 24-h, 8-, and 12-week PI. Lung sections were immunostained for F4/80, a general macrophage marker. Representative images from each treatment group are shown in (A–H); original magnification 400×; scale bar indicates 50 μm. Arrowheads indicate every positive cell identified in each representative image. Images were assessed for macrophage abundance by counting number of positive cells per FOV and number of alveoli per FOV. Quantification of F4/80+ cells per alveolus, normalized to pooled sham-irradiated controls, is shown in (I). Mean and standard deviation are shown for each treatment group, with white bars indicating C57 samples and shaded bars indicating TLR4−/−samples. (a) indicates a significant difference from C57 sham-irradiated, (b) indicates a significant difference from TLR4−/− sham-irradiated, and (c) indicates a significant difference from C57 irradiated within same timepoint. Statistical significance was determined by 2-way ANOVA and Tukey’s multiple comparisons across all groups (n = 3–6 FOV per lung section; n = 3–4 mice per group).
TLR4−/− Mice Exhibit Less Severe Fibrosis Than C57 Mice During Late Effects of Thoracic Irradiation
At 22-week PI, lung sections were harvested, processed, and stained for Gomori trichrome to visualize collagen, as well as α-SMA, to assess myofibroblast prevalence (Figs. 4A–D). The extent of fibrosis was quantified by area of fibrotic foci as a percentage of total lung area, as well as modified Ashcroft scoring (Figs. 4E and 4G). Both methods of fibrosis quantification showed that, while irradiated C57 lungs had developed substantial fibrotic burden, irradiated TLR4−/− lungs exhibited significantly less fibrosis (p < .01 for both percent fibrosis and modified Ashcroft scoring). This was corroborated by qualitative α-SMA immunohistochemistry, which indicated presence of myofibroblasts in areas of fibrotic accumulations in irradiated lungs of both strains.
Assessment of fibrosis severity in C57 versus TLR4−/− lungs at 22-week PI. Representative images of Gomori trichrome staining and α-SMA immunostaining of serial sections are shown at indicated original magnifications for C57 sham-irradiated (A), TLR4−/− sham-irradiated (B), C57 13 Gy (C), and TLR4−/− 13 Gy groups (D) Blue staining in trichrome images indicates collagen (please refer to digital version for color images). Black arrow in (C) low-power trichrome image indicates a large fibrotic focus. Black arrows in α-SMA images indicate areas of myofibroblast accumulation in fibrotic lesions. Percent fibrotic area (E) indicates the total area of each lung comprised of fibrotic lesions, as determined from low-power (12.5×) images of trichrome-stained lungs. Fibrosis was also assessed by blinded modified Ashcroft scoring of high-power (400×) images of trichrome-stained lungs. High-power images were obtained at 10 predetermined locations of left lungs (F) to accurately capture fibrotic lesions, which develop starting at the periphery of the lung. Modified Ashcroft scores for each mouse (G) were determined by the average score of all images (n = 10 images per lung). In (E) and (G), bars indicate median, interquartile range, and range of each group, and **indicates p < .01 between strains. Statistical significance was determined by 2-way ANOVA and Sidak’s post hoc test; n = 4–5 mice per group for (E) and (G). “n.d.” in (E) and (G) indicates fibrosis was not detectable. Scale bars in (D) indicate 2 mm, 50, and 125 μm for 12.5× trichrome, 400× trichrome, and α-SMA images, respectively.
Apoptosis Resistance Occurs Early During the Radiation Response in C57 Lung Fibroblasts, and Is Less Pronounced in TLR4−/− Lung Fibroblasts
The apoptosis susceptibility of fibroblasts of both strains at 24-h PI did not differ from controls. However, as early as 8-week PI, irradiated C57 fibroblast samples exhibited apoptosis resistance, which persisted throughout the remainder of the radiation response, as indicated by significantly lower apoptosis susceptibility compared with respective controls. Although the trend in C57 irradiated samples did not reach statistical significance at 16-week PI, this may have been resolved with a larger sample size. When taken in context among the 8-, 12-, and 22-week PI timepoints, the data suggest that this apoptosis resistant phenotype occurred relatively early in the radiation response and persisted up to and including the accumulation of fibrotic scarring. Furthermore, TLR4−/− fibroblasts did not exhibit a significant decrease in apoptosis susceptibility at any time during the radiation response (Figure 5).
Fibroblast apoptosis susceptibility in C57 versus TLR4−/− lungs throughout radiation response. Fibroblasts were isolated from C57 and TLR4−/− lungs at the indicated timepoints PI and assessed for apoptosis susceptibility in response to 1 μM STS. Apoptosis susceptibility of each sample represents the rate of population apoptosis, quantified by the slope of cleaved caspase 3/7 signal accumulation over 24-h STS exposure. Each bar shown indicates apoptosis susceptibility of fibroblasts from irradiated lungs and is presented as a percentage of respective control group. Dotted lines indicate range of control values, scaled to account for intertimepoint variability. Each bar indicates median, interquartile range, and range of each group, and *indicates group is significantly different from its respective unirradiated control group. Statistical significance was determined by 2-way ANOVA with Sidak’s post hoc test at each timepoint (n = 3–4 wells per mouse and n = 4–5 mice per group).
HMGB1 Release Coincides With Chronic Pneumonitis and Onset of C57 Fibroblast Apoptosis Resistance
Another indicator of cellular damage is systemic release of HMGB1, a DNA chaperone protein that can also act as a TLR4 agonist when actively secreted by activated macrophages or passively released due to disruption of cellular integrity (Kang et al., 2014; Lange and Vasquez, 2009; Tsung et al., 2014). C57 and TLR4−/− mice were assessed for HMGB1 release into systemic circulation by ELISA on plasma samples (Figure 6). HMGB1 levels were elevated in plasma samples in both strains at 24-h and 8-week PI, and slightly but not significantly at 12-week PI. There were no statistically significant differences in plasma HMGB1 release between strains.
Release of nuclear HMGB1 as a measure of cellular damage throughout the radiation response. HMGB1 levels were assessed by ELISA from plasma samples obtained from C57 and TLR4−/− mice at the indicated timepoints PI. * indicates statistical significance, as determined by 2-way ANOVA and Tukey’s post hoc multiple comparisons test comparing each pooled timepoint to sham-irradiated controls (n = 1 well per mouse; n = 3–5 mice per group).
DISCUSSION
In this study, we demonstrated that fibroblast apoptosis resistance precedes scar formation in fibrosis-prone mice, a finding that contradicts the current assumption that fibroblast apoptosis resistance and scarring occur simultaneously in PF pathogenesis. Additionally, we observed that the severity of fibroblast apoptosis resistance and subsequent fibrosis was ameliorated in mice lacking the TLR4 receptor, suggesting that TLR4 plays a significant role in these biological mechanisms.
Throughout the course of the radiation response, both strains exhibited similar markers of immediate injury, as characterized by SCGB1A1 loss and γ-H2AX prevalence, demonstrating no difference in strain radiosensitivity. Both strains also exhibited increased macrophage density in lung tissue at intermediate stages of the radiation response, which is characteristic of radiation pneumonitis. The less extensive fibrosis in TLR4−/− mice compared with C57 mice 22 weeks after 13 Gy thorax-only irradiation helps to clarify the existing literature comparing TLR4-deficient mice with wild-type controls. Although the C3H/HeJ mouse has long been used as a fibrosis-resistant counterpart to the C57BL/6J, few studies thus far have directly compared RIPF susceptibilities of 2 mouse strains with functional TLR4 presence or absence as the sole genetic difference. Paun et al. (2010) compared the RIPF severity of C57BL/6J versus C57BL/6J TLR4-/- mice after 18 Gy thorax-only ionizing radiation exposure, and found no significant difference between the extent of fibrosis in the 2 strains. This is in contrast to our present findings, but differences in radiation dose may explain this discrepancy. Considering that radiation-effect dose-response curves tend to be very steep (Williams et al., 2010), the dose-response curve for TLR4−/− mice may be right-shifted compared with C57; therefore, a higher dose of radiation in the TLR4−/− mice would be required to produce the same fibrotic response as the C57. If so, 18 Gy may saturate the sigmoidal dose-response curve and thus make it impossible to discern the true difference in fibrosis susceptibilities of these 2 strains, which is evidently detectable with 13 Gy. Due to the prosurvival nature of TLR4 signaling, we hypothesized that TLR4 may contribute to fibrotic scarring by skewing the apoptosis-survival balance in fibroblasts throughout the radiation response.
The method used here for isolating 1° MLFs and assessing apoptosis susceptibility ex vivo has substantial advantages. First, this assay is fibroblast-specific, which is difficult to achieve with methods such as double-immunofluorescence or flow cytometry due to the current lack of a reliable fibroblast surface marker. This advantage enabled us to isolate the effects of TLR4 on fibroblasts specifically, which cannot be determined when assessing whole-tissue lung responses. Second, the method utilized here is a functional assay that measures the delicate balance of the complex apoptotic machinery, which involves a vast multitude of cooperating proteins that cannot be accurately represented by, eg, immunoblotting for abundance of a few chosen apoptotic mediators.
Primary MLFs isolated and assayed for apoptosis susceptibility revealed interesting trends about the 2 strains’ pathology development. No significant difference in apoptosis susceptibility was detected between irradiated and control fibroblasts at 24-h PI. This suggests that fibroblast apoptosis resistance does not occur as a direct effect of radiation exposure, but rather due to downstream events throughout the radiation response. However, apoptosis resistance in C57-irradiated fibroblasts was detected as early as 8-week PI, as demonstrated by a weakened response to an apoptotic signal, showing that this phenotype precedes the appearance of fibrotic scar tissue. Prior studies have extensively described the existence of apoptosis resistant fibroblasts in fibrotic tissue, but until now this phenotype has not been described as occurring separately from and prior to scar tissue accumulation (Im et al., 2016; Maher et al., 2010; Moodley et al., 2004; Pardo and Selman, 2016; Sisson et al., 2012; Thannickal and Horowitz, 2006). The lack of detectable apoptosis resistance in TLR4−/− fibroblasts, coupled with the attenuated fibrosis at 22-week PI, also demonstrates that TLR4 plays an instrumental role in the induction of apoptosis resistance and subsequent fibrosis. Furthermore, because TLR4−/− mice exhibited attenuated fibrosis despite the more severe pneumonitis at 8-week PI, this implicates TLR4 as a target of pneumonitis-associated inflammatory signaling prior to the accumulation of fibrotic foci.
HMGB1 is a possible mediator of TLR4-associated apoptosis resistance in this model. Because extracellular HMGB1 can act as a DAMP and activate TLR4, systemic release of HMGB1 not only indicates cellular damage or inflammation but also suggests TLR4 signaling. It is here that disease progression may differ between the 2 strains in this study; HMGB1 release may initiate prosurvival or proinflammatory signaling downstream of TLR4 in C57 mice, but this component of the radiation response would be absent in TLR4−/− mice.
Immunohistochemistry for γ-H2AX and F4/80 collectively suggests that release of HMGB1 from nuclei into systemic circulation is caused by immediate radiation damage at 24-h PI, and by radiation pneumonitis at 8-week PI. The first stage of the radiation response is the immediate biological reaction to widespread free radical formation and DNA damage provoked by ionizing radiation (Riley, 1994). This can involve necrosis, in which membrane integrity is lost and cellular components, including a wide variety of DAMPs, can be passively released into the extracellular space, thereby triggering damage responses from surrounding cells. This immediate phase of radiation-induced lung injury evolves into radiation pneumonitis at approximately 8-week PI in this mouse model (Williams et al., 2010). A vast array of inflammatory cytokines are elevated in pneumonitis (Johnston et al., 1996; Wirsdörfer and Jendrossek, 2016), which can then signal to recruited immune cells to propagate the inflammatory response, which partially involves active secretion of HMGB1 (Tang et al., 2011).
As previously mentioned, activation of TLR4 leads to prosurvival signaling. Although HMGB1 was increased systemically here at 24-h PI, it is likely that this is a relatively transient increase in this DAMP, as a previous study showed that this initial radiation-induced HMGB1 abundance in plasma only persists from 6- to 48-h PI (Wang et al., 2016). In contrast, data shown here suggest that systemic HMGB1 release persists for many weeks during radiation pneumonitis. During this timeframe, HMGB1 may activate TLR4 on lung fibroblasts after the initial radiation-induced epithelial damage, causing prosurvival signaling in these cells through NF-κB, PI3K/Akt, or both. Upregulation of prosurvival proteins due to low-grade, sustained lung inflammation could skew the apoptosis-survival balance in these cells towards survival, thus causing apoptosis resistance. When TGF-β induces fibroblast-to-myofibroblast differentiation at later timepoints (Wynn, 2008), it is possible that the cells are already apoptosis resistant due to the effects of endogenous HMGB1 signaling during earlier pneumonitis. Therefore, based on these findings, we propose a 2-hit mechanism for the genesis of apoptosis resistant myofibroblasts: first, TLR4 activation causes apoptosis resistance, then TGF-β induces the myofibroblast phenotype.
These data have significant implications for future fibrosis research. First, due to the nature of isolating 1° MLFs by expansion, these cells necessarily undergo multiple replications before daughter cells are assayed for apoptosis susceptibility. Since apoptosis resistance was observed in C57 13 Gy samples at 8-week PI and beyond, this indicates that the apoptosis resistant phenotype in fibrosis is cell autonomous and, once acquired, does not need additional stimuli to be inherited. Therefore, this phenotype likely has an epigenetic origin. This idea has previously been explored by Huang et al. (2013), who found that increased histone deacetylase expression was partially responsible for the downregulation of the Fas death receptor in fibrotic fibroblasts, another aspect of the apoptosis resistant phenotype. Second, the TLR4−/− mice in this study did develop fibrosis, although less so than C57 mice. This indicates that there are other factors contributing to the apoptosis resistant phenotype, which may include any number of altered microenvironmental cues or dysregulated intercellular signals. Additionally, if TLR4 inhibition were to be used clinically as a preventative treatment during the radiation response, it would likely be most effective if coupled with secondary treatments to shorten the duration of pneumonitis, alter the lung microenvironment to favor normal wound healing, or a multitude of other possible therapeutic strategies.
It is important to note a few limitations of this study. First, several findings are based on limited sample sizes, and these conclusions would be bolstered by larger sample sizes and greater statistical power. Limited sample size in this study also prevented characterization of changing fibroblast attributes throughout the radiation response beyond apoptotic susceptibility, and investigation into this aspect of RIPF could lead to further insight into disease pathogenesis. Second, the implication of HMGB1 specifically as a main agonist for TLR4 in this context is intriguing; however, future studies inhibiting HMGB1 throughout the radiation response are required to start to establish a causal link between HMGB1 and TLR4-mediated fibroblast apoptosis resistance. Third, future in vitro or ex vivo studies may also clarify the molecular signaling pathway that links TLR4 activation to apoptosis resistance in lung fibroblasts.
In conclusion, we present evidence that fibroblast apoptosis resistance occurs much earlier in PF pathogenesis than currently believed, and that TLR4 is a key mediator in the development of this aspect of the disease. Additional studies are necessary not only to further explain the molecular origins of fibroblast apoptosis resistance, but also to incorporate this knowledge into a cohesive understanding of the various manifestations of PF pathogenesis. With a better appreciation of these complex mechanisms, viable methods of PF prevention and treatment will be developed to help those adversely affected by this disease.
DECLARATION OF CONFLICTING INTERESTS
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
FUNDING
National Institutes of Health (NIH) National Institute of Environmental Health Sciences (T32-ES007026-40, T32-HL066988, and P30-ES01247); NIH National Institute of Allergy and Infectious Diseases (U19- AI091036, U19-AI067773, and R01-AI101732-01); and University of Rochester Medical Center (Lung Biology and Disease Program Pilot Grant 2017).
ACKNOWLEDGMENTS
We thank all members of our lab and collaborating labs, including Tyler A. Beach, PhD and Ravi S. Misra, PhD, for their insights and helpful discussions regarding this work. Additional thanks go to Matthew D. McGraw, MD and Elaine M. Smolock, PhD for article feedback.
Comments