Repetitive diacetyl vapor exposure promotes ubiquitin proteasome stress and precedes bronchiolitis obliterans pathology

Abstract

Bronchiolitis obliterans (BO) is a devastating lung disease seen commonly after lung transplant, following severe respiratory tract infection or chemical inhalation exposure. Diacetyl (DA; 2,3-butanedione) is a highly reactive alpha-diketone known to cause BO when inhaled, however, the mechanisms of how inhalation exposure leads to BO development remains poorly understood. In the current work, we combined two clinically relevant models for studying the pathogenesis of DA-induced BO: (1) an in vivo rat model of repetitive DA vapor exposures with recovery and (2) an in vitro model of primary human airway epithelial cells exposed to pure DA vapors. Rats exposed to 5 consecutive days 200 parts-per-million DA 6 h per day had worsening survival, persistent hypoxemia, poor weight gain, and histologic evidence of BO 14 days after DA exposure cessation. At the end of exposure, increased expression of the ubiquitin stress protein ubiquitin-C accumulated within DA-exposed rat lung homogenates and localized primarily to the airway epithelium, the primary site of BO development. Lung proteasome activity increased concurrently with ubiquitin-C expression after DA exposure, supportive of significant proteasome stress. In primary human airway cultures, global proteomics identified 519 significantly modified proteins in DA-exposed samples relative to controls with common pathways of the ubiquitin proteasome system, endosomal reticulum transport, and response to unfolded protein pathways being upregulated and cell–cell adhesion and oxidation–reduction pathways being downregulated. Collectively, these two models suggest that diacetyl inhalation exposure causes abundant protein damage and subsequent ubiquitin proteasome stress prior to the development of chemical-induced BO pathology.

Introduction

Bronchiolitis obliterans (BO) is a fibrotic lung disease characterized by progressive luminal narrowing with subsequent obliteration of the small airways, or bronchioles (Barker et al. 2014). This debilitating lung disease occurs commonly after lung transplantation with nearly 50% of all lung transplant recipients being affected by BO at 5 years post-transplant (Goldfarb et al. 2016; Lynch et al. 2012; Moonnumakal and Fan 2008; Yusen et al. 2016). BO is also the leading cause of mortality at 1 and 5 years in lung transplant recipients for both children and adults (Goldfarb et al. 2016; Lund et al. 2016; Rossano et al. 2016; Yusen et al. 2016). In the non-transplant population, inhalation exposures to certain chemicals and severe respiratory tract infections can also result in BO (King et al. 2011; Lynch et al. 2012). Respiratory insults associated with BO development include: (a) chemical inhalation exposures to diacetyl (DA; 2,3-butanedione) (Kreiss et al. 2002), sulfur mustard (Ghazanfari et al. 2009), and chlorine (O’Koren et al. 2013), (b) autoimmune disorders like rheumatoid arthritis (Ippolito et al. 1993; Kakazu et al. 1990; van Thiel et al. 1991) and graft-versus-host disease (Dudek et al. 2003; Epler 1988; Matsumoto et al. 1996; Philit et al. 1995), and (c) severe respiratory tract infections such as adenovirus (Colom et al. 2006; Moonnumakal and Fan 2008). More recently, BO has received considerably greater attention in veterans exposed to open-air burn pits, sulfur mine fires, and dust storms (King et al. 2011; Szema et al. 2012). Collectively, these inhalation exposures highlight the more common occurrence of BO outside of organ transplantation alone.

One of the most well-known chemicals associated with BO is diacetyl (Boylstein et al. 2006; Fedan et al. 2006; Hubbs et al. 2008; Kreiss et al. 2002; Palmer et al. 2011; Pierce et al. 2014). Diacetyl (DA; 2,3-butanedione) is a highly reactive alpha di-ketone naturally occurring in many foods but also added for flavoring to popcorn and e-cigarettes for its buttery aroma (Allen et al. 2016; Clark and Winter 2015; Klager et al. 2017). Though previously classified as generally safe for consumption by the Food and Drug Administration (FDA), DA is now known to cause BO when inhaled at occupationally relevant concentrations (Bailey et al. 2015; Kreiss et al. 2002). One of the first published cohorts of works with lung disease associated with DA inhalation exposure was that of microwave popcorn manufacturing facility workers (Kreiss et al. 2002). Workers exposed to high concentrations of DA developed symptoms associated with BO such as severe shortness of breath and exertional dyspnea. Ultimately, many of these workers developed debilitating and persistent lung disease that infrequently progressed to lung transplantation (Kreiss et al. 2002). Over the past 20 years, inhalation exposure in other food processing occupations such as coffee bean roasting has also been associated with DA inhalation exposure and the associated airways pathology identified (Bailey et al. 2015; Clark and Winter 2015; Duling et al. 2016). Equally concerning, many of the most popular e-cigarette liquids contain DA, placing workers in e-cigarette factories potentially at risk for BO development (Allen et al. 2016). Despite its common use as a chemical flavoring, our understanding of how chemical inhalation exposure to DA leads to BO development remains in its infancy.

Preclinical modeling of DA inhalation exposures has focused primarily on acute inhalation toxicity (Fedan et al. 2006; Hubbs et al. 2002, 2008; Morgan et al. 2008). Hubbs et al. developed a 6-hour DA vapor exposure at occupationally relevant exposures that resulted in significant respiratory cytotoxicity (Hubbs et al. 2002; Morgan et al. 2008, 2012, 2016). At DA concentrations greater than one hundred parts-per-million (ppm), injury to the lower respiratory tract and more specifically intrapulmonary airway epithelium, the primary location of BO, occurred (Morgan et al. 2012, 2016). Apoptotic cell death of bronchial epithelial cells noted by ubiquitin-positive punctate lesions co-localizing with certain airway epithelial cell markers such as acetylated tubulin and keratins was common after a single 6-hour DA exposure at concentrations greater than 150 ppm (Hubbs et al. 2016). In subsequent studies, Morgan et al. (2016) evaluated rats exposed to DA vapors for 2 weeks followed by a 2-week recovery period (Morgan et al. 2016). All rats surviving the initial exposure period developed histologic evidence of intrapulmonary airway fibrosis. While significant species differences exist between rodents and humans, DA-exposed rats develop airway lesions similar to human BO pathology following repeated DA vapor exposures. However, the underlying mechanisms of BO development in this repetitive DA exposure model remains limited (Morgan et al. 2016).

To complement these in vivo exposures, previous authors have exposed primary human airway epithelial cultures at air–liquid interface (ALI) to DA vapors to further interrogate the mechanism of DA cytotoxicity (Brass et al. 2017a; Foster et al. 2017; Zaccone et al. 2015). Global proteomics analysis of DA-exposed compared to PBS-exposed airway epithelial cells identified the loss of cilia and increased squamous differentiation in exposed cells suggestive of abnormal epithelial differentiation (Foster et al. 2017). Repetitive in vitro DA vapor exposures also resulted in abundant protein damage, most commonly seen in damaged keratin-associated proteins (Foster et al. 2017; Park et al. 2019). Keratins are commonly expressed in the airway epithelia and provide structural integrity as well as facilitate intracellular communication (Chandrakasan et al. 1991). Following DA vapor exposure, extensive non-reducible crosslinking of keratins occurred as seen with a shift in molecular weight (Foster et al. 2017; McGraw et al. 2020). With sufficient time after DA exposure, damage to airway epithelial keratins resolve primarily mediated through lysine (K)48-linked polyubiquitination and subsequent proteasome degradation (McGraw et al. 2020). Conversely, when proteasome function was inhibited with the non-selective proteasome inhibitor MG132, DA-induced protein damage and keratin crosslinking persisted despite cessation of DA exposure. These in vitro investigations suggest two conclusions: (1) significant protein damage occurs after DA vapor exposure in human airway epithelial cells and (2) airway epithelial recovery is mediated primarily through ubiquitination and proteasome degradation.

The primary goal of the following studies was to evaluate the ubiquitin proteasome stress response after repetitive DA vapor exposures using two clinically relevant models: an in vivo model of DA-induced BO and an in vitro model of primary human airway epithelial cells exposed to DA vapors. We hypothesized that repetitive DA vapor exposure results in abundant protein damage in the airway epithelium causing significant stress to the ubiquitin proteasome system and subsequently promoting the development of BO pathology.

Materials and methods

Animals

Studies were approved by the Institutional Animal Care and Use Committee of the University of Rochester Medical Center (URMC) and adhered to the National Institutes of Health Guidelines. Seven-week-old outbred male Sprague–Dawley (SD) rats weighing 200–250 g (Charles River Laboratory, Wilmington, MA) were maintained in an AAALAC-accredited animal care facility.

Aerosol generation, animal exposures, and recovery

Diacetyl (DA; 2,3-butanedione; Sigma-Aldrich, St. Louis, MO) inhalation exposures were performed at the University of Rochester’s Inhalation Exposure Facility (IEF). Additional information on DA vapor generation can be found in the Online Supplement ‘Methods’. Animals were exposed to 200-parts-per-million (ppm) DA using whole-body inhalation chambers for 6 h/day for 5 days. Two hundred ppm DA is comparable to 12 ppm in humans when corrected for nasal scrubbing due to larger surface area in rodents and DA concentration seen for occupational exposures (Boylstein et al. 2006; Kreiss et al. 2002). Immediately after the final exposure (DA Day 5), animals were returned to housing for daily weights and oxygen saturation (Starr Life Science Technologies, Oakmont, PA) for 14 days (DA Day 19; ‘recovery period’; Supplement Figure S1). Sample size for animals exposed to DA was estimated from previous studies (Flake and Morgan 2017), accounting for early death due to upper respiratory tract necrosis and airway obstruction, with the estimated combined animal sizes per group of DA (n = 43/group) and air control (n = 20/group).

Pulmonary function testing

A subset of rats surviving to DA Day 19 underwent pulmonary function testing (n = 6/group) before euthanasia and independent from those animals that underwent BALF collection and histologic evaluation. Further description of the procedure and parameters in the Online Supplement ‘Methods’.

Bronchoalveolar fluid (BALF) collection, tissue harvest and histopathology with grading

BALF cell counts and differentials were performed on cytospin-prepared slides stained with Diff-Quik (Siemens, DE) by two independent investigators blinded to exposure condition (> 200 cells/slide; cells/μl BALF). Left lung lobes were paraffin-embedded, sectioned (5 μm) and stained for hematoxylin and eosin (H&E) or Gomori Trichrome stain. All airways were considered bronchi when the luminal diameter was greater than 250 μm nor connected directly to the alveolar space. Bronchiolitis obliterans (BO) was defined as airway obstruction due to concentric sub-epithelial collagen deposition with luminal obliteration of intrapulmonary bronchi (Morgan et al. 2012, 2016). Two investigators blinded to exposure condition independently graded Trichrome-stained sections of the left lung lobe for: (1) the presence or absence of airway lesions in each rat, and (2) if present, the number of affected bronchial lesions per animal. At least three distinct regions were examined within the left lung lobe at 10× magnification per animal. Each fixed left lung lobe was also noted for the presence or absence of parenchymal fibrosis.

Western blot analysis of whole lung homogenates and staining for ubiquitin-C lung expression

To screen for ubiquitin stress, embedded left lung sections were stained for the ubiquitin stress protein ubiquitin-C (Ubq-C; 1:1000, Invitrogen). Rabbit IgG (1:1000, Agilent, Santa Clara, CA) was used as negative controls. Right lower lobe lungs were homogenized in RIPA lysis buffer (Abcam; Cambridge, MA) supplemented with a protease inhibitor cocktail (Roche; Indianapolis, IN). Thirty micrograms (μg) total protein were resolved in pre-casted 4–15% gradient Tris–Glycine gel (Bio-Rad, Hercules, CA), and immunoblotted for Ubq-C (1:5000; Biolegend). Total protein served as loading controls semi-quantitated from each corresponding stain-free gel (Mini-PROTEAN; Bio-Rad, Hercules, CA) with SuperSignal West Pico chemiluminescent substrates (Thermo Scientific, Waltham, MA). Semi-quantification and protein normalization were performed using Image Lab software (Bio-Rad, Hercules, CA).

Proteasome activity of total lung homogenates

To measure proteasome activity in rat lungs, the right middle lobe was collected and lysed in 50 mM HEPES (pH 7.5), 5 mM EDTA, 150 mM NaCl and 1% Triton X-100 supplemented with 2 mM ATP buffered solution. Proteasome activities were determined using the Proteasome-Glo 3 Substrate System (Promega, Madison WI), in accordance with the manufacturer’s instructions, with approximately 500 μg protein per sample. The luminescent signal was quantified in a SpectraMax M5 plate reader (Molecular Devices, San Jose, CA). Optimal timing of signal detection occurred 15 min after reaction initiation with the kit substrates.

Primary human airway epithelial cultures exposed to DA vapors

For in vitro DA vapor exposures, human EpiAirway™ (AIR-100) cultures generated from primary human bronchial epithelial cells from a healthy, non-smoking, 22-year-old donor (‘TBE-20’), were purchased from MatTek, Corporation (Ashland, MA). DA concentration was similar to previously published DA vapor cup exposures (Brass et al. 2017b; Foster et al. 2017; Kelly et al. 2014) and relevant to peak occupational exposure levels of DA identified in artificial butter flavoring factories (Boylstein et al. 2006; Kreiss et al. 2002) as well as similar in time-weighted average (TWA) to that seen with in vivo exposures. Cells exposed to phosphate-buffered solution (PBS) were used for exposure negative controls. ALI tissue cultures were exposed to DA or PBS vapors for 1 h/day for 1 or 2 consecutive days using vapor cups as described previously (Brass et al. 2017b; Foster et al. 2017; Kelly et al. 2014) as well as in the Online Supplement. Cultures were monitored for cytotoxicity via supernatant lactate dehydrogenase (LDH) (Thermo Scientific Pierce; Rockford, IL) and cellular permeability via trans-epithelial electrical resistance (TEER) (EVOM², World Precision Instruments, Sarasota, FL).

Proteomics analyses of primary human airway epithelial cell

Twenty-four hours after the end of the exposure, human bronchial epithelial cells were lysed in 1.5 ml mixture of 50 mM NH₄HCO₃, 8 M urea, 5 mM N-ethylmaleimide (ThermoFisher Scientific, Waltham, MA) and 0.5 mM Tris-(2-carboxyethyl)-phosphine hydrochloride (TCEP), which allowed protein denaturation, reduction, and alkylation. Approximately 400 μg protein per sample was subjected to trypsin digestion after diluting the samples eightfold with 50 mM NH₄HCO₃. Trypsin was added at the ratio of 1 to 50 (enzyme to protein) for digestion at 37 °C for 3 h. The digested peptides were further desalted using C18 solid phase extraction columns (Phenomenex, Torrance, CA). Finally, 5 μl of 0.1 μg/μl of peptides from each sample were analyzed by LC–MS/MS using a Waters nano ACQUITY UPLC system (Waters, Milford, MA) coupled with a Q-Exactive plus mass spectrometer (ThermoFisher Scientific, Waltham, MA).

Proteomics data analysis

A label-free relative quantification approach was applied for the global proteomics (Cox and Mann 2008; Piehowski et al. 2018) (Moghieb et al. 2018; Rindler et al. 2017). Raw MS data were searched using MaxQuant (version 1.6.10.43) with label free quantification (LFQ) method (Tyanova et al. 2016). In the statistical analysis, a principal component analysis (PCA) was performed after imputation. Student’s paired t-test p values were calculated to determine if DA exposed conditions had significant difference with PBS vehicle exposed conditions in the aspect of mean protein abundances. Gene ontology (GO) and KEGG pathways analysis of significant proteins were performed with DAVID (https://david.ncifcrf.gov/summary.jsp).

Additional statistical analysis

Prism 8.0 software (GraphPad Software) was used for statistical analysis with one-way ANOVA followed by Tukey’s post hoc analysis for multiple comparisons with naïve controls. A log-rank (Mantel–Cox) test was performed for survival analysis. Mean values are reported with standard deviation. A p value < 0.05 was considered significant.

Results

Bronchiolitis obliterans develops during the recovery period following repetitive diacetyl (DA) vapor exposures

An in vivo rodent model of repetitive diacetyl (DA) vapor exposures with a 14-day recovery period was developed for modeling of DA-induced BO development for these studies. The time-weighted average for DA vapor exposure for three distinct DA exposures was 200.2 ± 3.8 ppm (see Supplement Table 1). Survival was significantly worse in DA-exposed rats compared to air-exposed rats at study’s end (DA: 59% (n = 43) vs. air: 100% (n = 20), ***p = 0.0002, Log-rank Mantel–Cox test; Fig. 1a). Both early and late deaths occurred in DA-exposed rats; deaths occurring early (< 7 days) during or immediately following exposure were due to acute obstruction of the proximal trachea with airway epithelial necrosis and sloughing (Supplement Figure S2); while deaths during the recovery period were associated with poor weight gain, hypoxemia, and impaired lung function.

Fig. 1.

a Kaplan–Meier survival curve. Sprague–Dawley rats exposed to air (n = 20; blue) versus rats exposed to 6 h/day for 5 days of 200 ppm diacetyl (DA) vapor exposure (n = 43; red) followed by a 14-day recovery period. Comparison of survival curves via log-rank (Mantel–Cox) test at study end (Day 19; ****p < 0.0001). b Daily oxygen saturations measured by pulse oximetry in air (n = 20; blue) versus DA-exposed (n = 43; red) for 5 days of exposure followed by a 14-day recovery period. Two-way ANOVA with Sidak’s compensation for multiple comparisons with respect to treatment condition and time (days). Statistical difference significant when adjusted p value < 0.05. c Percent weight change with respect to time in air (n = 20; blue) versus DA-exposed (n = 43; red). Two-way ANOVA with Sidak’s compensation for multiple comparisons with respect to treatment condition and time (days). Statistical difference considered significant when adjusted p value < 0.05

Persistent hypoxemia and failed weight gain after repetitive DA exposures

Animal monitoring with oxygen saturations (Fig. 1b) and daily weights (Fig. 1c) occurred during DA exposures and recovery. Oxygen saturations in DA-exposed rats differed from air controls on Day 4 during exposure but did not fall below a clinically relative threshold of 90% in DA-exposed rats until 5 days after DA exposure ended (Day 10: 87.2 ± 9.2%). Oxygen saturations remained below 90% until study’s end (Day 19: 83.5 ± 13.1%) and differed significantly for air-exposed controls on Days 10–19 during the recovery period (Fig. 1b, two-way ANOVA with Sidak’s correction for multiple comparisons, ****p < 0.0001).

Percent weight change from pre-exposure weight also differed significantly in DA-exposed rats compared to air controls for all time points (Fig. 1c, ****p < 0.0001; two-way ANOVA with Sidak’s correction). DA-exposed rats (red) lost weight during exposure reaching a nadir of −17.9 ± 9.2% on Day 5, and continued to have impaired weight gain during the 14-day recovery period (Day 19 DA % wt.: + 39.2 ± 23.0% vs. Day 19 air % wt.: + 97.3 ± 8.6%). Collectively, these two clinical parameters of hypoxemia and poor weight gain suggest the persistence of significant lung disease following repetitive DA vapor exposures.

Pulmonary function testing was performed on a subset of animals who survived until study’s end (Day 19, Fig. 2). Significant changes occurred in multiple lung function parameters of DA-exposed animals (red, n = 6) compared to air controls (blue, n = 6) including: pressure–volume loops (Fig. 2a), total lung resistance (R, Fig. 2b), Newtonian resistance (R_N, Fig. 2C), tissue damping (G, Fig. 2d) inspiratory capacity (Fig. 2e), lung compliance (C_rs, Fig. 2f) and total system compliance (C_st, Fig. 2G). Total lung resistance (R) increased six-fold in DA-exposed animals relative to air controls (means of 0.52 ± 0.38 cm H₂O*s/ml vs. 0.09 ± 0.02 cm H₂O*s/ml; *p < 0.05) while Newtonian resistance (R_N), a measurement representative of central airways resistance, increased eightfold in DA-exposed compared to air controls (means of 0.31 ± 0.20 cm H ₂O*s/ml vs. 0.04 ± 0.02 cm H₂O*s/ml: **p < 0.01). Tissue damping (G), a surrogate measurement of parenchymal resistance, did not differ significantly between groups (means of 0.54 ± 0.29 cm H₂O/ml vs. 0.30 ± 0.06 cm H ₂O/ml; p = 0.08). Both lung and total system compliance were decreased in DA-exposed compared to naïve controls (means of 0.32 ± 0.11 ml/cm H₂O vs. 0.64 ± 0.10 ml/cm H₂O; ***p < 0.001 and means of 0.62 ± 0.16 ml/cm H₂O vs. 0.98 ± 0.15 ml/cm H₂O; **p < 0.01). A two-fold reduction in inspiratory capacity was seen in DA-exposed animals compared to naïve controls (means of 6.9 ± 1.3 ml vs. 11.7 ± 1.9 ml; ***p < 0.001). These pulmonary function findings support a mixed obstructive and restrictive lung disease pattern, however, the greater relative change in proximal airway resistance (measured by Newtonian resistance) in comparison to distal parenchymal resistance (seen with tissue damping) suggest greater airways than parenchymal lung injury and remodeling.

Fig. 2.

Lung function measurements in air-exposed (blue) and diacetyl-exposed (red) rats (n = 6/group) performed at study’s end (Day 19; 14 days after 5 consecutive days of diacetyl exposures). Individual lung function components include: a pressure–volume loops, b total lung resistance, c Newtonian resistance, d tissue damping, e inspiratory capacity, f lung compliance, and g system compliance. (Unpaired t test, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001)

Histology consistent with bronchiolitis obliterans pathology after 14-day recovery

Animals that survived until study’s end (DA Day 19) were evaluated for the presence or absence of BO lesions. Greater than 80% (13/16) of DA-exposed animals demonstrated histologic evidence of BO pathology with sub-epithelial collagen deposition, concentric luminal narrowing and obliteration of the airway lumen (Fig. 3). The mean number of affected airways per rat was 2.6 ± 2.1 in DA-exposed. As suggested by increases in tissue damping by pulmonary function testing, some DA-exposed rats also demonstrated histologic evidence of parenchyma changes in addition to the airway abnormalities identified in DA-exposed animals. These parenchymal changes included small fibrous masses, alveolar septal thickening, and collagen I deposition (Supplemental Figure S3). The number of animals with interstitial changes was less than those animals affected with airway lesions (10/16 surviving animals). Similar to the changes seen on pulmonary function testing, histology demonstrated mixed airway and parenchymal lung disease.

Fig. 3.

Representative Trichrome-stained lung sections highlighting intrapulmonary airways of air-exposed animals at 4× (a; bar: 500 μm) and 20× (b; bar: 200 μm) compared to DA-exposed animals at 4× (c; bar: 500 μm) and 20× (d; bar: 500 μm) at study’s end (Day 19). Solid arrows identify intrapulmonary bronchial airways with airway lumen diameter > 250 μm) while hollow arrows identify adjacent bronchial vessels

BALF total protein, cell count and cell differentials

Next, bronchoalveolar lavage fluid (BALF) was characterized for total protein, cell count and cell differentials as surrogate markers of lung permeability and cellular influx, respectively, to two time points: immediately following the last DA vapor exposure (DA D5) and at the end of the recovery period (DA D19). BALF total protein increased significantly at Day 19 compared to air controls (0.24 ± 0.04 mg/ml (air) and 0.30 ± 0.14 mg/ml (DA D5) vs. 0.50 ± 0.22 mg/ml (DA D19), ANOVA ***p < 0.001, Fig. 4a), suggestive of a persistent increase in lung permeability during the recovery period after repetitive DA exposures. Similar to BALF total protein, BALF cell counts also increased at Day 19 compared to air controls (320 ± 129 cells/μl (air) and 559 ± 340 cells/μl (DA D5) vs. 884 ± 508 cells/μl (DA D19), ANOVA **p < 0.01, Fig. 4b). Percent macrophages in BALF decreased both at D5 and D19 (means of 94.8 ± 2.3% (air) vs. 83.4 ± 7.4% (DA D5) vs. 79.8 ± 10.2% (DA D19), ANOVA ****p < 0.0001, Fig. 4c) while percent neutrophils increased in BALF at D5 and D19 compared to air controls (2.0 ± 1.5% (air) vs. 9.5 ± 5.1% (DA D5) vs. 12.1 ± 9.8% (DA D19), ANOVA **p < 0.01, Fig. 4d). No significant changes in BALF percent lymphocytes, eosinophils or basophils were seen immediately after exposure or during the recovery period compared to air controls (data not shown). Collectively, these BALF results suggest increased lung permeability with a persistence of lung neutrophilia after repetitive DA vapor exposures.

Fig. 4.

Bronchoalveolar lavage fluid (BALF) total protein (a), total cell count (b), macrophage (c), and neutrophil percentage (d) in air controls (n = 12, blue hollow bars), diacetyl (DA)-exposed rats immediately after 5 days’ exposure (DA D5) (n = 12, red striped bars) and 14 days post-recovery (DA D19) (n = 12, red checkered bars). (ANOVA, *p < 0.05. **p < 0.01. ***p < 0.001. ****p < 0.0001)

Increased ubiquitin-C immediately after DA vapor exposure localizing to the airway epithelium

In previous in vitro experiments using human airway epithelial cultures, our group as well as others have shown significant protein damage to airway keratins exposed to 1 hour of DA vapor (Foster et al. 2017). Recovery from this acute protein damage was mediated through ubiquitination and subsequent proteasome-mediated degradation. Hence, we hypothesized that repetitive DA vapor exposures result in overwhelming protein damage and abundant polyubiquitinated protein accumulation within the lungs of exposed rats. To interrogate this hypothesis, we first evaluated for signs of ubiquitin stress using the polyubiquitinated protein stress marker ubiquitin-C (Ubq-C) in whole lung tissue homogenates as well as in histologic lung sections at both Day 5 and Day 19 after DA exposure.

Immediately after DA exposure (DA D5), whole lung Ubq-C expression increased significantly in DA-exposed animals compared to air controls (1.58 ± 0.36 vs. 1.00 ± 0.12, ANOVA ***p < 0.001, Fig. 5a–c). Conversely, no significant difference occurred in whole lung Ubq-C expression of DA-exposed lungs compared to air control lungs during the recovery period (0.95 ± 0.08 vs. 1.00 ± 0.12, ANOVA p > 0.05). Spatially, increased Ubq-C protein expression localized to the intrapulmonary airways, the primary location of BO lesions (Fig. 5e), that was not seen in air controls (Fig. 5d). Ubq-C expression localizing at intrapulmonary airways normalized in DA-exposed animals during the recovery period with equivalent staining to that seen in air controls (Fig. 5f). Hence, accumulation of polyubiquitinated proteins occurred early after DA exposure and resolved during the recovery period in DA-exposed animals.

Fig. 5.

Representative western blot for polyubiquitin-c (Ubq-C) a from total lung homogenates in air-exposed, DA-exposed immediately after exposure (Day 5) and DA-exposed at study’s end (14 days post-recovery) normalized with respect to ‘stain-free’ gel for total protein within each lane. b Quantification of polyubiquitination-C (Ubq-C) normalized for total protein relative to air controls. (ANOVA, **p < 0.01. ***p < 0.001). c Representative staining for Ubq-C in formalin-fixed intrapulmonary airways from air-exposed animals (d), DA-exposed immediately after exposure (DA D5) (e), and DA-exposed at study’s end (DA D19) (f) obtained 10× (bar: 200 μm). Solid arrows identify intrapulmonary airways and hollow arrows identify adjacent bronchial vessel

Lung proteasome activity increases acutely after DA exposure but returns to baseline levels during recovery

To further investigate whether the accumulation of polyubiquitinated proteins seen immediately after DA vapor exposures was secondary to abundant protein damage or impaired proteasome processing, we assessed for the three primary proteolytic activities of the 20S proteasome, chymotrypsin-like (CT-L), trypsin-like (T-L) and caspase-like (C-L), in total lung homogenates of exposed animals at Days 5 and 19 (Fig. 6a–c, respectively). These three major proteolytic activities are contained within the 20S proteasome core and account for the majority of the protein degradation required to maintain cellular homeostasis including damaged cellular proteins. All three proteasome-associated activities increased acutely after DA vapor exposures (DA D5) compared to air controls (CT-L: 82,400 ± 16,307 vs Air: 6590 ± 15,878 relative luminescent units (RLU), ANOVA ****p < 0.0001; T-L: 14,194 ± 7,602 vs. Air: 151 ± 212 RLU’s, ANOVA ***p < 0.001; C-L: 10,798 ± 8,784 vs. Air: 429 ± 391 RLU’s, ANOVA **p < 0.01). By Day 19, all three activities decreased acutely from immediately after exposure, and did not differ significantly from air control activity levels (CT-L: 11,836 ± 27,521; T-L: 1631 ± 3301; T-L: 732 ± 1170). These results support that the abundance of polyubiquitinated proteins seen in the lung immediately after DA vapor exposure are more likely due to an overwhelming production of damaged protein requiring degradation, and less likely due to impaired proteasome function, considering CT-L, T-L, and C-L activities were not decreased at the time of increased Ubq-C expression seen after DA exposure.

Fig. 6.

Proteasome-associated chymotrypsin-like (CT-L) (a), trypsin-like (T-L) (b), and caspase-like (C-L) (c) proteolytic activities in total lung homogenates in air-exposed (n = 6, blue hollow bars), DA-exposed animals at immediately after exposure (Day 5, DA D5, n = 5, red striped bars) and DA-exposed at study’s end (14 days post-recovery, DA D19, red checkered bars). (ANOVA, **p < 0.01. ***p < 0.001. ****p < 0.0001)

Ubiquitin stress protein co-localizes to the basal cell layer of human airway epithelial cells in vitro after repetitive DA vapor exposures

Considering Ubq-C increases in whole lung homogenates after repetitive DA exposures, we then assessed for Ubq-C in human airway epithelial cells cultured at air–liquid interface and exposed to DA vapors. Using the previously published DA ‘vapor cup’ exposure model (Kelly et al. 2014), primary human airway epithelial cultures were exposed to pure DA vapor (25 mM) or PBS for 1 h/day for 2 consecutive days and then assessed 24 h after the last exposure for cell death, permeability and total proteomics. No significant change in cellular necrosis, assessed by supernatant lactate dehydrogenase levels (LDH), occurred with repetitive daily DA exposure compared to PBS controls (Supplement Figure S4A, ANOVA p = 0.15). Trans-epithelial electrical resistance (TEER) decreased in DA-exposed samples compared to PBS controls (Supplement Figure S4B, ANOVA ** p < 0.01), supportive of increased airway epithelial permeability with repetitive DA exposures. While Ubq-C was expressed faintly throughout all layers of the epithelium after two consecutive days of 25 mM DA vapor exposure, staining for Ubq-C was seen more commonly in cells located adjacent to the basolateral membrane and co-expressing keratin 5 (Krt5), a common basal cell marker, than Ubq-C co-localizing with cells expressing acetylated tubulin (AT), a common ciliated cell protein marker (Supplement Figure S5D). These findings support the prior in vivo findings that repetitive DA vapor exposures causes increased airway epithelial permeability and the accumulation of poly-ubiquitinated proteins in airway epithelial cells.

Proteomics of human airway epithelial cells notable for increased ubiquitination-proteasome degradation and decreased for mitochondrial function and basal cell adhesion expression

To support our in vivo findings of DA exposure causing abundant protein damage, polyubiquitination and subsequent proteasome activation, global proteomics analysis was performed on primary human airway epithelial cells exposed to DA or PBS vapors for either 1 day or 2 consecutive days. As demonstrated in Fig. 7a, no clear signals were observed for airway epithelial cells exposed to 1-day DA exposure in the p value distribution for each comparative analysis. However, clear signal separation occurred for 2-day DA exposure compared to PBS-exposed controls. In addition, principal component analysis (PCA) analysis displayed clear separation between cells with 2-day DA exposure and controls. (Supplement Figure S6). Thus, all further proteomics analysis was performed on samples exposed to either PBS or DA for 2 consecutive days (Fig. 7b–d). Through MS analysis, 4160 proteins were precisely identified of which 2980 proteins were quantified and 519 were significantly different between conditions (p value < 0.05) (see Supplementary Data for information on protein quantification). Protein abundances in significantly differential proteins were categorized as 235 up-regulated (‘Cluster 1’, red Fig. 7) and 284 down-regulated (‘Cluster 2’, blue Fig. 7) and then were input for gene ontology/pathway enrichment analysis to examine biological processes (BP) and KEGG pathway. Notable biological processes (Fig. 7c) in upregulated proteins included protein folding and proteasome-mediated ubiquitin-dependent protein catabolic process. Numerous ubiquitin-associated proteins (UB2L3_HUMAN, PSB5_HUMAN, Fig. 7d) were observed with significantly abundant changes, suggesting DA exposure induced ubiquitination-proteasome degradation. Notable biological processes associated with the down-regulated proteins included oxidation–reduction process, mitochondrial translational elongation, cell–cell adhesion, as well as the ribosome by KEGG pathway identification.

Fig. 7.

Global proteomics analysis from primary human airway epithelial cells exposed to DA (n = 4/group) or PBS (n = 4/group) for two consecutive days. a p value distribution for 1- and 2-day exposure. b Heat map of proteins with significant abundance changes (student t-test *p < 0.05) between air exposed and DA-exposed human airway epithelium cells. Relative protein abundances in the heatmap were displayed in log2 abundance format with the median centering to zero. c DAVID enrichment analysis with enriched Biological Processes (BP) and KEGG pathway (KP) ranked by significance for both up- and down-regulated proteins. d Heat map for representative proteins from Cluster 1 (up-regulated in DA-exposed) and Cluster 2 (down-regulated in DA-exposed) relative to PBS

Discussion

Repetitive inhalation exposure to the highly reactive diketone diacetyl in Sprague–Dawley rats as well as in primary human airway epithelial cultures caused significant protein damage, accumulation of polyubiquitinated proteins, and subsequent stress to the ubiquitin proteasome system prior to the development of BO pathology. DA-exposed animals developed persistent hypoxemia, poor weight gain, and reduced survival during a 14-day recovery period with increased lung permeability, increased cellular influx and neutrophilia by BALF. Ubiquitin stress proteins increased in total lung homogenates and localized to the proximal airway epithelium of DA-exposed animals, however, these changes dissipated during the recovery period. Proteasome activity increased concurrently with increased polyubiquitinated protein exposures, more suggestive of the increased accumulation of damaged and ubiquitinated proteins contributing to lung dysfunction than impaired proteasome function. These findings were further supported by global proteomics analyses in human airway epithelial samples exposed to DA vapors where multiple proteins were upregulated and associated with the ubiquitin proteasome degradation following exposure.

Multiple previous authors have developed preclinical models of DA inhalation exposures. Using intratracheal instillation (ITI) of DA, Palmer et al. showed a single dose of DA (125 mg/kg) resulted in significant airway injury with the subsequent development of intraluminal airway changes of increased airway resistance and BALF neutrophilia 7 days after exposure. These authors evaluated airway epithelial repair using BrdU staining as a marker of proliferation and various epithelial cell markers including club cell secretory protein (CCSP), acetylated tubulin (AT), and β-catenin expression for differentiation after injury. Three days post-DA exposure, injured rat airways expressed increased staining for BrdU suggestive of appropriate epithelial proliferation. Conversely, DA-exposed airways showed reduced expression of CCSP, AT, and β-catenin suggestive of abnormal epithelial differentiation following DA exposure. In contrast to this model, our current model used vaporized DA for airway injury (over intratracheal instillation). An important difference in the model design exists with respect to the route of exposure; specifically, the resultant injury seen following DA vaporization is more proximal affecting larger diameter airways (termed intrapulmonary bronchi due to luminal size being > 250 μm) than that seen with ITI of DA. This subtle difference in location of injury may play a significant role in modeling of airway repair, considering the proximal airway of the rat is more commonly populated by basal cells than mice and replaced distally by variant club cells (Mercer et al. 1994). In humans, basal cells function as the primary cell type for airway epithelial repair in both proximal and distal airways, a distinct physiologic difference between rodents and mammals (Rock et al. 2009). Hence, we utilized the current model of vaporized DA exposure to model an occupational inhalation exposure as well as injure the more proximal intrapulmonary airway over the distal bronchiolar airway.

The current model of repetitive DA exposures is similar that of two other chemical inhalation exposures modeling BO induction, specifically, the sulfur mustard (SM) exposure model (McGraw et al. 2017, 2018) and the chlorine (Cl₂) gas exposure model (O’Koren et al. 2013). In the SM model, a single dose of SM resulted in both airway and parenchymal fibrosis at 14 days and persisted for 28 days after exposure. SM-exposed rats developed hypoxemia and weight loss weeks after the initial exposure, similar to the current DA exposure where persistent hypoxemia and delayed weight gain developed during the recovery period. These results suggest that it is not the initial injury alone, but the abnormal repair response (McGraw et al. 2017, 2018) that likely contributes to the subsequent development of airway pathology. In the model reported by O’Koren, BO-like lesions were seen in the distal trachea of chlorine-exposed mice (O’Koren et al. 2013). The distal trachea was used (over the intrapulmonary airways) to more closely model airway repair in humans where airway basal cells are the primary source of epithelial progenitor cells in mice (Rock et al. 2010). Similar to repetitive DA exposures or a single dose SM exposure, survival worsened in the week following Cl₂ exposure, again supportive of failed epithelial repair. With Cl₂ exposure, aberrant re-epithelialization occurred with the loss of keratin 5+ basal cells. Hence, a common theme to these chemical inhalation models and BO development is abnormal repair of the proximal airway epithelium in small rodents.

Unique to DA inhalation exposures, prior authors have shown the development of significant protein damage through radiolabeling of DA and instilling into the rodent trachea (Fennell et al. 2015). Following DA instillation, significant arginine adducts form between radiolabeled DA and hemoglobin or BALF albumin. Hubbs et al. evaluated for ubiquitin in mouse lung tissue after an acute DA exposure (200 ppm for 6 h) and identified co-localization of total ubiquitin to bronchial epithelial cells after acute exposure (Hubbs et al. 2016). We identified similar results here with increased expression of ubiquitin-C, a marker of ubiquitin stress, localizing primarily to bronchial epithelia after repetitive DA exposure. Hubbs et al. identified co-localization of ubiquitin puncta with the lysosomal associated membrane protein (LAMP) and sequestosome-1 (SQSTM1/p62) supportive of autophagosome after DA exposure in mice (Hubbs et al. 2016). We have shown previously in primary human airway epithelial cultures exposed to DA vapors significant increases in expression of lysine(K)48-linked (often referred to as ‘proteasome-mediated’) ubiquitination that resolves with cessation of exposure and time (McGraw et al. 2020). Consistent with prior in vitro experiments, when Ubq-C was evaluated 2 weeks after DA exposures, increased Ubq-C expression resolved suggestive of appropriate ubiquitin clearance with time and cessation of DA exposure. These findings support abundant protein damage occurring after DA vapor exposure with subsequent polyubiquitination and proteasome stress. However, the airway epithelium may have multiple other pathways such as autophagy or lysosomal degradation, also activated after diacetyl inhalation exposure. Future work is underway to assess how these individual components, or failure of their usual function, contribute to abnormal epithelial repair, inflammation induction, and BO development.

To support our in vivo findings and further investigate the mechanisms of DA exposure toxicity contributing to BO development, LC–MS/MS global proteomic analysis was applied to the protein expression profiles of primary human airway epithelial cells exposed to DA and PBS for 2 consecutive days (Fig. 7b, c). DAVID enrichment analysis identified protein folding, the ubiquitin proteasome system, endosomal transport, and response to unfolded protein being upregulated in DA-exposed samples relative to PBS controls while cell–cell adhesion and oxidation–reduction pathways being downregulated. Under homeostasis, proteins are properly folded in the lumen of the endoplasmic reticulum (ER) to perform their prior functions (Diaz-Villanueva et al. 2015). This highly complex folding process can be stressed by a number of environmental stimuli, commonly researched after inhalation exposure to cigarette smoke or other combustible pollutants in the airway epithelium (Marciniak 2017). When significant stress persists, misfolded or damaged proteins accumulate, leading to loss of ER function. This aggregation of misfolded proteins induces the unfolded protein response (UPR), resulting primarily in protein degradation by ubiquitin–proteasome system (UPS) or secondarily via autophagy to maintain cellular proteostasis (Liu et al. 2016). In our DA vapor exposures, numerous ubiquitinated proteins (UBP14_HUMAN, UCHL5_HUMAN and UB2L3_HUMAN) and proteasome-associated proteins (PSB5_ HUMAN, PRS4 _HUMAN) were upregulated (Fig. 7d) supportive of protein damage occurring with DA exposure that require polyubiquitination and subsequent proteasome degradation to prevent potential ER stress. In addition, sequestosome-1 (SQSTM1/p62), a player with dual roles in the function of autophagy and in delivering ubiquitinated proteins to the proteasome for protein degradation (Liu et al. 2016), was more abundant in DA exposure (Fig. 7d).

DA exposure not only induced over-expression of the UPS to degrade an excess of polyubiquitinated proteins, but also led to mitochondrial dysfunction seen via downregulation of multiple proteins associated with the oxidation–reduction pathways (NNTM_HUMAN and AL7A1_HUMAN). Nicotinamide nucleotide transhydrogenase (NNT, NNTM_HUMAN) is an integral protein of the inner mitochondrial membrane (IMM) and involved in antioxidant defense in the mitochondria. Coupling with mitochondrial proton gradient, NNT enables the proton transfer from NADH to NADP⁺ across the IMM and facilitates forward reaction to produce NADPH, which is utilized to remove reactive oxygen species (ROS) in mitochondrial antioxidative system (Ho et al. 2017; Hoek and Rydstrom 1988). Meimaridou et al. illustrated that NNT knockdown in a human adrenocortical cell line resulted in impaired redox potential and increased ROS levels (Meimaridou et al. 2012). Aldehyde dehydrogenase (AL7A1_HUMAN or ALDH7A1) prevents cells from oxidative stress through metabolism of lipid peroxidation-derived aldehydes. Brocker et al. demonstrated mitochondrial ALDH7A1 mitigated oxidative stress induced cellular cytotoxicity (Brocker et al. 2011). Collectively, increased protein expression associated with the ubiquitin proteasome system and decreased protein expression of detoxifying mitochondrial-associated enzymes illustrate that repetitive DA inhalation exposures promotes the accumulation of damaged proteins and associated oxidative stress.

The current work is not without limitations. One of the primary limitations is the lack of assessment of an intermediate time point (such as 7 days after DA exposure). This time point was not assessed considering our primary interests were in the initial injury response (DA D5) and fibrotic induction (DA D19), and less the inflammatory response. Similarly, animals were not carried out for a longer recovery time period considering the current study’s purpose was to evaluate the initial injury and BO induction, and less the persistence or progression of BO disease. Third, previous animal exposures and predictive animal modeling have shown that the current concentration of 200 ppm in rats is equivalent to about 12 ppm in humans, a concentration within the range of exposure for humans exposed in popcorn and coffee grinding factories (Boylstein et al. 2006; Kreiss et al. 2002). When lower concentrations are used for exposures in rats, the majority of DA is metabolized or scrubbed in the nasal mucosa prior to reaching to the intrapulmonary airways (Morris and Hubbs 2009).

In conclusion, repetitive DA vapor exposures contribute to substantial stress to the ubiquitin proteasome system in both rat and human airway epithelia and prior to the development of chemical-induced BO. Dysfunction of the ubiquitin proteasome system after chemical inhalation exposure likely contributes to abnormal epithelial repair and BO development. Future modulations to the ubiquitin proteasome system after chemical inhalation exposures will provide greater insight into the mechanisms of abnormal airway repair and may provide therapeutic direction for the devastating disease of BO, currently with no FDA-approved therapies.