A model of the aged lung epithelium in idiopathic pulmonary fibrosis

A seven-day injury protocol that induces cellular senescence in MLE12 cells

Cellular senescence is key characteristic of the alveolar epithelium in IPF. With this in mind, we first sought to develop an injury protocol that could reliably induce this pathological feature [12, 16–18]. After screening various concentrations [10, 25, 50 μg/ml), durations (24 or 48 hr), and frequencies (every 48 or 72hr) of bleomycin in MLE12 cells, we uncovered a protocol (Figure 1A) that reliably induced features of senescent cells. This included causing the characteristic flattening and enlarging of cell shapes as well as the suppression of cell proliferation (data not shown). We also detected an upregulation in β-galactosidase activity (Figure 1B), and higher expression of multiple cellular senescence markers, including the cell cycle inhibitors p21 and p53, the DNA damage marker γ-H2AX and the SASPs IL-6, MCP1, TNF-α, and IL-1α (Figure 1C–1E). Altogether, these findings indicate that our unique 7-day injury protocol induces multiple characteristic features of senescent IPF cells.

Figure 1. Cellular senescence markers are increased in MLE12 cells after our 7-day injury protocol. (A) Schematic showing the experimental design: On day 0, MLE-12 cells were treated with 10 μg/ml bleomycin for 24h. Day 2, bleomycin was removed, and the culture medium was refreshed. Day 4, cells were treated again with 10 μg/ml bleomycin. Day 5, bleomycin was subsequently removed, and cells were collected at day 7. (B) SA-β-gal activity (green fluorescence cytoplasmic staining) in control and chronically injured MLE12 cells. Number of SA-β-gal positive cells per 100 cells counted (right). (C) Western blot (WB) for p-p53^Ser21 and p21 in chronically injured lung epithelial cells (with β-actin loading control). Densitometry is shown on the right. (D) WB for γ-H2AX in control and bleomycin injured cells (with β-actin loading control). Densitometry is shown on the right. (E) Transcript levels for Il6, Mcp1, Tnf-α, and Il1α in controls vs. chronically injured lung epithelial cells. Statistical significance was assessed by unpaired Student’s t-test for two groups. * p< 0.05, **P < 0.01, ***P < 0.001 versus control group, n=6.

Mitochondrial function is impaired in MLE12 cells undergoing our seven-day injury protocol

Like cellular senescence, mitochondrial dysfunction has emerged as a key feature of the alveolar epithelium in IPF. To determine whether mitochondrial dysfunction was altered in our model system, we began by comparing mitochondrial oxygen consumption rates (OCR) between control and injured cells. Consistent with altered mitochondrial function, we found that OCR was significantly reduced in cells undergoing our injury protocol. This included a nearly 50% reduction in basal respiration as well as an even greater decline in ATP production (Figure 2A, 2B). Further, we found that DNA damage and redox balance (Figure 2C, 2D) were altered in these cells and we detected higher expression of ATF4, a transcriptional regulator of the mitochondrial unfolded protein response (UPR^mt) (Figure 3A). Consistent with the upregulation in ATF4, we observed an increase in several of its downstream targets, including the mitochondrial chaperone proteins Hspd1, Hspe1 and Trap1 and mitochondrial proteases CLPP and Lonp1 (Figure 3B, 3C). Lastly, we detected mitochondrial changes not yet described in the alveolar epithelium of IPF patients, such as an increase in levels of the deacetylase SIRT3, the transcription factor forkhead box O3a (FOXO3) and the antioxidant enzyme superoxide dismutase 2 (SOD2) (Figure 3D, 3E), suggesting our model can help to uncover new mechanisms in the pathogenesis of disease.

Figure 2. 7-day bleomycin protocol disrupts mitochondrial homeostasis in alveolar epithelial cells. (A, B) Mitochondrial oxygen consumption in control and chronically injured MLE12 cells. Right, bar graph depicting basal oxygen consumption rate (OCR), ATP production and spare respiratory capacity. (C) mtDNA damage quantification in control and bleomycin injured cells. (D) Quantification of reactive oxygen species (ROS) by MitoSox staining in control and chronically injured MLE12 cells. Seahorse data are representative of 3 separate experiments. Statistical significance was assessed by unpaired Student’s t-test except for Seahorse analyses in which two-way analysis of variance (ANOVA) followed by Sidak post hoc analysis was used to adjust for multiple comparisons. Statistical significance was achieved when P < 0.05 at 95% confidence interval. Given multiple * p< 0.05, **P < 0.01, ***P < 0.001 versus control group, n=6.

Figure 3. 7-day bleomycin protocol activates the mitochondria unfolded protein response and induces expression of mitochondrial chaperones in alveolar epithelial cells. (A) WB for ATF4 in controls vs. chronically injured lung epithelial cells (with β-actin loading control). Densitometry is shown on the right. (B) Transcript levels for Hsp60, Hsp10 and Hps90 in controls vs. chronically injured MLE12 cells. (C) WB for Clpp and Lonp1 in control and bleomycin injured cells (with β-actin loading control). Densitometry is shown on the right. (D, E) WB for SIRT3, FOXO3a and SOD2 in control and bleomycin injured MLE12 cells (with β-actin loading control). Densitometry is shown on the right. Statistical significance was assessed by unpaired Student’s t-test for two groups. * p< 0.05, **P < 0.01, ***P < 0.001 versus control group, n=6.

Cellular metabolism is altered in our seven-day bleomycin injury model

Another recently described feature of IPF cells is dysregulated cellular metabolism [19–22]. To determine whether cellular metabolism was affected in our model, beyond the changes already described in OCR, we measured levels of different metabolites by LC/MS in control and injured cells. Consistent with findings described in IPF, we uncovered significantly higher levels of lactate in injured cells [23]. This also associated with elevated pyruvate levels and a trend toward higher levels of fructose 1,6-biphosphate (F1,6BP) and 1, 3-bisphosphate glycerate (1,3 BPG) (Figure 4A), suggesting glycolysis was upregulated in these cells. However, this did not correlate with higher extracellular acidification rates (levels were actually significantly lower), suggesting, at the very least, that lactate secretion was reduced in injured cells (Figure 4B). In addition to elevated levels of glycolytic intermediates, we also detected higher levels of several TCA cycle intermediates (citrate, isocitrate, α-ketoglutarate and succinate) (Figure 4B) and various amino acids, including the nitrogen-rich amino acids glutamine, asparagine and arginine (Figure 4C) [24, 25]. Consistent with altered nitrogen balance, multiple metabolites of the urea cycle were increased in our bleomycin-injured MLE12 cells (Figure 4E).

Figure 4. 7-day bleomycin protocol alters cellular metabolism in alveolar epithelial cells. (A) Levels of glycolytic metabolites in control and bleomycin injured cells. (B) Extracellular acidification rate in control and bleomycin injured cells. (C–E) Tricarboxylic acid (TCA) cycle intermediates, amino acid levels and urea cycle intermediates in control and bleomycin injured cells. Statistical comparisons were performed using two-way analysis of variance (ANOVA) followed by Sidak post hoc analysis in order to adjust for multiple comparisons. * p< 0.05, **P < 0.01, ***P < 0.001 versus control group, n=6.

Our seven-day injury model induces a loss of proteostasis in MLE12 cells

Another defining feature of IPF cells is a loss of proteostasis. For example, mutations causing the misfolding of surfactant proteins are linked to familial forms of IPF and misfolded and aggregated proteins have been shown to accumulate in the alveolar epithelium of patients with sporadic forms of disease [26, 27]. To assess whether proteostasis was altered in our model system, we first assessed levels of intracellular protein aggregates in injured vs uninjured cells. Consistent with impaired proteostasis, we found that intracellular protein aggregates were significantly increased in injured cells (Figure 5A, 5B). Further, this associated with homeostatic changes in the ubiquitin proteasome system (UPS), including marked reduction in chymotrypsin-, trypsin-, and caspase-like activities (Figure 5C, 5E) as well as significant accumulation of K48-linked ubiquitinated proteins (data not shown). Further, we detected reduced levels of various UPS proteins, such as the transcriptional regulator Nuclear E2-related factor 1 (NRF1) (Figure 6A), and several key components of the 26S proteosome (Psmd12, Psmd5 and PsmbB6/β1) (Figure 6B).

Figure 5. 7-day bleomycin protocol leads to loss of proteostasis in alveolar epithelial cells. (A, B) Proteostat staining (aggresome) in controls and chronically injured MLE12 cells. Quantification is shown on the right. (C–E) Chymotrypsin-like, trypsin-like and caspase-like activities in control and bleomycin injured MLE12 cells. Statistical significance was assessed by Student t-test **P < 0.01, ***P < 0.001 versus control group, n=6.

Figure 6. 7-day bleomycin protocol reduces NRF-1 expression in alveolar epithelial cells. (A) WB for NRF1 in control and bleomycin injured cells (with β-actin loading control). Densitometry is shown on the right. (B) Transcript levels for PSMD10, PSMD12, PSMD5, PSMD11, and PSMB6/β1 in controls vs. injured MLE12 cells. Statistical significance was assessed by unpaired Student’s t-test for two groups. * p< 0.05, **P < 0.01, ***P < 0.001 versus control group, n=6.

Having established that our model impairs proteostasis, we next sought to uncover other mechanisms that contribute to loss of proteostasis. Since chaperone proteins are critically involved in the trafficking and folding of proteins, we next sought to determine whether expression of these proteins was altered in injured cells. Consistent with the loss of proteostasis, we found that transcript levels for heat shock proteins (HSP)-70, -90, and -40 were significantly reduced in injured versus uninjured cells (Figure 7A). Further, this associated with various protein modification of heat shock transcription factor 1 (HSF1) (Figure 7B), which is a key regulator of many HSP genes. Although total protein levels were not significantly different, we detected a nearly 1.5 fold increase in levels of the inactivating modification at Ser307 (Figure 7B), and a 30% reduction in levels of the activating phosphorylation at Ser326. Taken together, these findings support the notion that our model could help to advance our understanding of why IPF causes a loss of proteostasis.

Figure 7. The HSF1 activity is impaired in MLE12 cells after our 7-day injury protocol. (A) Transcript levels for Hsp90, Hsp70, and Hsp40 in control and bleomycin injured MLE12 cells. (B) WB for HSF1, p-HSF1^ser307, p-HSF1^ser326 in control, and bleomycin injured MLE12 cells (with Laminin-B1 loading control). Densitometry is shown on the right. Results of densitometry analysis are depicted in bar graphs (n= 6, per group). Statistical significance was assessed by unpaired Student’s t-test for two groups. * p< 0.05, **P < 0.01, ***P < 0.001 versus control group, n=6.

Cell culture and reagents

Mouse MLE12 cells were obtained from the American Type Culture Collection (ATCC, Manassas, VA) and cultured according to recommended protocols. To induce cellular injury, cells were exposed to different bleomycin exposures, although ultimately we settled on a two-hit, 7-day injury protocol because of its ability to reliably induce features of senescent cells. In brief, this protocol involved exposing cells to two, 24h exposures of low-dose bleomycin (10 μg/ml) (Enzo life science, BML-AP302-0050, 1mg of Bleomycin is equivalent to 1500 IU) on days 1 and 4, followed by 48 hr post-exposure rest periods. Endpoint analyses were performed immediately after the second 48 hr rest period.

Mitochondrial ROS assessments

Mitochondrial ROS levels were measured using the Mito-Sox Red dye (Invitrogen, Carlsbad, CA). In brief, MLE12 cells were incubated with dye at a concentration of 5 mM for 10 min at 37° C and fluorescence was measured using a microplate reader.

Mitochondrial DNA damage assay

Mitochondrial (mt) DNA damage was assessed by qPCR as previously described. In brief, polymerase chain reaction (PCR) was performed to amplify short and long forms of mtDNA and genomic DNA (β-globin). Mitochondrial and genomic DNA were quantified using a microplate reader (PerkinElmer, Waltham, MA.) based on Pico-Green excitation and emission wavelengths of 485 and 530 nm, respectively. Results were normalized by calculating the ratio of small to long mtDNA fragments in each sample. Mitochondrial lesions were determined based on the following equation: D= (1-2 (−Δlong−Δshort) × 10,000 (bp)/size of the long fragment (bp).

Oxygen consumption measurements

Oxygen consumption rate (OCR) was measured using the Seahorse XFp Bioanalyzer (Seahorse Bioscience, Billerica, MA). In brief, MLE12 cells were seeded at a concentration of 30,000 cells/well on XFp cell plates 24 h prior to the initiation of studies. The concentration of FCCP (also known as trifluoromethoxy carbonylcyanide phenylhydrazone), antimycin A, and rotenone were 0.75 μM, 2μM, and 2μM respectively. Oligomycin was used at a concentration of 4 μM. Pyruvate (1 mM) and glucose (25 mM) served as the substrate for studies. Results were analyzed using wave 2.4 software.

Senescence-associated beta galactosidase (SA-β-gal) detection

β-galactosidase activity was assessed by performing SA-β-gal staining (Dojindo Molecular Technology, Rockville, MD), according to manufacturer’s instructions. To prevent false positives, staining was performed when cells were at or below 70% confluence.

Protein aggregation assay

The proteostat aggresome detection kit (Enzo Life Sciences, Farmingdale, NY) was used to quantify protein aggregates and aggresomes. In brief, MLE12 cells were washed with 1X PBS and fixed with 4% paraformaldehyde for 10 min at room temperature (RT) before exposure to permeabilization reagent and stained with ProteosStat dye (1:2000 for 30 m).

Proteasome activity

Chymotrypsin-, trypsin- and caspase-like activities were assayed at RT using 10 μg of protein in the presence of 100 μM Suc-LLVY-aminoluciferin (Succinyl-leucine-valine-tyrosine-aminoluciferin), Z-LRR-aminoluciferin (Z-leucine-arginine-arginine-aminoluciferin and Z-nLPnLD-aminoluciferin(Z-norleucine-proline-norleucine-aspartate-aminoluciferin), respectively (Promega Corporation, Madison WI). The reaction was monitored by performing fluorimetric measurements every 10 min for 1 h (excitation 350 nm, emission 460 nm) using the Synergy HT Multi-Detection microplate reader. Proteasome activity was determined as the difference between the total activity in the presence or absence of 20 μM MG132 (Millipore-Sigma).

Western blot analysis

Samples (20 μg) were solubilized and loaded on a 10% Tris-HCl-SDS-polyacrylamide gel and run for 1 h at 100 V before proteins were transferred to a nitrocellulose membrane (ThermoFisher Scientific). Membranes then underwent overnight incubation at 4° C with primary antibodies to p-p53^ser15, β-actin, Lamin B1(Cell Signaling), γ-H2AX, HSF1, NRF1, FOXO3A, SOD2, SIRT3 (Cell Signaling), ATF4 (Invitrogen), CLPP, LONP1, p-HSF1^ser307 (Abcam), p-HSF1^ser326 (Bioss antibodies) or p21 (Millipore), at dilutions of 1:1,000 in blocking buffer with 0.1% Tween-20. This was followed by incubation with secondary antibody which was either a Donkey anti-Rabbit (Li-Cor Biosciences) or a Rabbit anti-mouse (Li-Cor Biosciences) antibody at dilution of 1:8,000. After washing, protein bands were visualized using the Odyssey infrared imaging system (Li-Cor Biosciences).

RNA isolation and real-time quantitative PCR (qRT-PCR)

Gene transcript levels were quantified by real-time PCR. In brief, total RNA was isolated using RNeasy Mini-Kit (Qiagen), according to the manufacturer’s instructions. A summary of primer sets manufactured by Integrated DNA Technologies can be found in Supplementary Table 1. Relative gene expression levels were quantified with the Livak method (2-ΔΔCT) by normalizing expression of the gene of interest to Hprt, demonstrated to be unaffected by bleomycin, and calculating the relative fold differences between experimental and control samples. Assays were completed in technical triplicate.

Metabolite extraction

The organic solvent extraction protocol was modified and optimized for 10 cm dishes. Briefly, medium was removed from dishes and cellular metabolism was rapidly halted with 700 ul of chilled mixture of 70% methanol and water containing the internal standards, 25 μM d4-citrate, and d3-Leucine. Plates were immediately placed at -80° C for 10 min to ensure complete metabolic quenching, followed by lysis using a cell scraper. Lysates were vortexed and pelleted (10000 x g for 10 min, 30s mix 30s rest) and pellets were washed twice with a quenching mixture before undergoing additional high speed vortexing. Supernatants were collected and dried under a stream of N2 gas followed by resuspension in 250ul of water containing 0.1% formic acid and subsequent analysis by LC-MS.

Liquid chromatography-mass spectroscopy

Chromatographic analysis was performed in a Waters Acuity UPLC system (Waters Corp., Milford, MA) using Acuity UPLC HSS T3 1.8μm, 2.1×100mm column with a flow rate of 0.4 and 0.6 ml/min for TCA cycle/glycolytic intermediates and amino acids, respectively. The chromatographic separation was achieved using 9 min gradient elution. For TCA cycle-glycolytic intermediates with solvents of A: 0.2% formic acid water and B: 0.1% formic acid acetonitrile, the initial mobile phase composition was 5% B gradually increasing to 90% B in 2.0min, then held constant at 95% B for 0.5 min, and finally brought back to the initial condition of 5% B in 1 min followed by 2 min re-equilibration. For amino acids and urea cycle metabolites solvents of A included: 0.1% formic acid, 95% water/acetonitrile (pH=5.8) and B: 90% acetonitrile/water, the initial mobile phase composition was 0.1% B, gradually increasing to 40% B in 4.0min, gradually increasing to 99.9% B in 2.0min, and finally brought back to the initial condition of 0.1% B in 1 min followed by 1 min re-equilibration. The injection volume of all samples was 5 μL, but for the glycolytic intermediates, we increased volume to 10 μL.

Mass spectrometry detection was performed using a Xevo Triple Quadrupole MS (Waters Corp) equipped with an electrospray ionization source (ESI) operating simultaneously in positive and negative ionization mode for amino acids and TCA cycle-glycolytic intermediates, respectively. For TCA cycle, and glycolytic intermediates: the desolvation gas flow rate was set to 800 l/h at a temperature of 600° C, the cone gas flow rate was set at 150 l/hand the source temperature at 150° C. For amino acids and urea cycle metabolites: the desolvation gas flow rate was set to 550 l/h at a temperature of 450° C, the cone gas flow rate was set at 150 l/hand the source temperature at 130° C. The capillary voltage was set to 2500 volts for positive ion mode; 1000 volts for negative ion mode; the cone voltage was set depending upon each specific MRM for each metabolite. Data was collected in MRM mode by screening parent and daughter ions simultaneously. MRM transition and parameters for each metabolite are listed in Supplementary Table 2. Except for lactate, which was from 1-0.01mM, ten concentrations of mixed standards were prepared by diluting concentrated stock solutions down to 100-0.001μM for all metabolites. The data were validated by the linearity of the standard curves and recovered internal standards from the biological sample extractions. The MS raw data analysis and the calibration curve for each targeted compounds were calculated in the software TargetLynx (Waters Corp., Milford, MA, USA). The collected data were normalized by the protein mass and internal standard of each sample.

Statistical analysis

Statistics were performed using GraphPad Prism 8.0 software (GraphPad, San Diego, CA). Two-group comparisons were analyzed by unpaired Student’s t-test, and multiple comparisons were performed using two-way analysis of variance (ANOVA) followed by Sidak post hoc analysis. Statistical significance was achieved when P < 0.05 at 95% confidence interval.