The BRD4 selective BD1 competitive inhibitor, ZL0591, was synthesized as previously described and determined to be >99% pure by HPLC (Liu et al., 2018; Zhou et al., 2021). The compound was initially dissolved in DMSO and diluted into 10% hydroxypropyl β-cyclodextrin (Sigma-Aldrich, St. Louis, MO) in PBS for intraperitoneal (IP) administration.

Mouse protocols were reviewed and approved by the University of Wisconsin-Madison School of Medicine and Public Health Institutional Animal Care and Use Committee (IACUC) and adhered to guidelines set by National Institutes of Health (NIH) and Association for Assessment and Accreditation of Laboratory Animal Care (AAALAC).

Eleven-week-old C57BL/6J female mice (Jackson Laboratories, Bar Harbor, ME) were housed in groups of five or less, in static cages with Diamond Dri paper pellet bedding (Inovive, San Diego, CA) and unlimited access to laboratory chow and water. The specific pathogen-free level 3, animal biosafety level 1 facility was under controlled temperature (∼74°F), humidity (∼30%) and illumination (12 h light/12 h dark cycle). Animals received daily veterinary health checks and weekly cage changes.

For bleomycin instillation, mice were anesthetized with IP administration of ketamine (100 mg/kg, Zoetis Inc., Parsippany-Troy Hills, NJ) and xylazine (15 mg/kg, Akorn Pharmaceuticals, Lake Forest, IL) prior to delivery of a single intratracheal dose of bleomycin (1 U/kg, Teva Pharmaceutical Industries, Israel) dissolved in 50 µL 0.9% normal saline (NS, B. Braun Medical Inc., Bethlehem, PA). Control animals were treated with an intratracheal dose of 50 µL of NS and were kept in separate cages. Mouse weight and survival after treatment were documented daily until endpoint on days 14, 21 or 28 post-bleomycin. Cages carrying five mice/cage were allocated to each group [one cage for NS-treated group and two cages for each of the bleomycin-treated groups to account for previously observed mortality rates post-bleomycin in C57Bl/6J females (Bernau et al., 2015; Bernau et al., 2021)]. Daily IP injections of ZL0591 (10 mg/kg body weight), or vehicle control, were administered to surviving mice from day 14 until 21 and every other day thereafter until endpoint at day 28. At the experimental endpoint (day 14, 21 or 28 post-bleomycin), mice were anesthetized by IP injection of ketamine (200 mg/kg) and xylazine (30 mg/kg) and exsanguinated prior to collection of biological samples.

Three days before each endpoint (day 11, 18 or 25 post-bleomycin), murine pulse oximetry was assessed using the MOUSEOX® PLUS (Starr Life Sciences Corp., Oakmont, PA). A small-adult mouse collar-based sensor was placed on awake mice according to the manufacturer’s instruction. In preparation for this, the mice were acclimated to the sensor daily as the animal was weighed.

Mice were initially anesthetized with 4% isoflurane gas and maintained with 2% isoflurane during whole-body micro-computed tomography (µCT) on a Siemens Inveon (Siemens Healthineers, Knoxville, TN) at the University of Wisconsin-Madison Small Animal Imaging and Radiotherapy Facility (SAIRF). Imaging was performed with the following parameters: 360 rotation degrees and rotation steps, binning 4, 250 ms exposure time, 80 kVp, 1 mA and ∼105 µm resolution. Reconstructions were performed with no down-sampling using a Shepp-Logan filter and Hounsfield Unit (HU) calibration.

Image analysis assessing µCT lung tissue density was performed using Image Research Workplace similarly as before (Tian et al., 2019a), by selecting and outlining 16 axial lung slices, starting with one immediately superior to the liver in n = 4–10 mice/condition. HU values associated with each voxel in the outlined volume were binned into indicated groups with lowest density voxels being in the < −600 HU group and highest density being between 0 and 100 HU. Percent voxels in each group of HU values were calculated based on the total number of voxels in the mouse lung. Individual axial slices were reconstructed in MATLAB and HU represented with a heat map for ease of visualization (Higham and Higham, 2016).

Assessment of total lung collagen was performed by utilizing a colorimetric hydroxyproline assay, as we have previously done (Bernau et al., 2021). Flash frozen superior, inferior and post-caval lobes of the right lung were homogenized in double distilled H₂O (ddH₂O, Sigma Aldrich, St. Louis, MO) using a dounce homogenizer. Lysates were then hydrolyzed in 6 M HCl for 3 h at 120°C. The samples were cooled to room temperature (RT), centrifuged to remove debris, aliquoted in triplicate into 96-well optical plate (Greiner bio-one, 655101) and dried for 45 min at 65°C. Subsequently, samples were incubated in chloramine T solution for 20 min at RT, followed by a 15 min incubation in Ehrlich’s solution at 65°C. Samples were then cooled to RT for 20 min. Absorbance was measured using a colorimetric plate reader at 550 nm. Hydroxyproline content was calculated by utilizing trans-4-Hydroxy-L-proline-generated standard curve (Sigma Aldrich).

Mouse left lungs were inflated with 4% neutral buffered formalin (Fisherbrand, Pittsburg, PA) and fixed for 24–48 h before being submerged in 70% ethanol, paraffin embedded and sectioned. Serial lung sections were then subjected to either Masson’s trichrome staining or immunostaining against smooth muscle α-actin (αSMA; rabbit polyclonal antibody from Abcam (ab694), Cambridge, United Kingdom). Negative control stains using the corresponding IgG isotype along with matching secondary antibodies were utilized. Stained slides were scanned using Aperio Digital Pathology Slide Scanner System. Digital snapshots were taken in Aperio ImageScope (Leica Biosystems, Wetzlar, Germany). To quantify αSMA immunostaining, ∼ 11 snapshots were taken of the lung parenchyma at ×40 magnification per mouse in n = 5–6 mice/condition. The ImageJ immunohistochemistry Toolbox was used to isolate the brown stain, after which images were inverted and their integrated density measured (Schneider et al., 2012; Bernau et al., 2015). High-level magnification (×40) permitted analysis of parenchymal regions that excluded blood vessels and airways so that αSMA + signal from these regions did not bias the results.

Modified Ashcroft scoring to quantify lung fibrosis severity (Hübner et al., 2008) was performed by blinded observers utilizing lung sections that had been subjected to Masson’s trichrome stains, as we have previously done (Hübner et al., 2008). Observers visualized and scored ∼10 field of view (FOV) at ×20 magnification per mouse in n = 4–10 mice/condition.

Flash frozen middle lobes of the right lungs were homogenized using a bullet blender with RNA STAT-60 and 0.9–2.0 mm stainless steel RNase-free bullet blender beads before being subjected to the previously described RNA isolation protocol (Bernau et al., 2015; Bernau et al., 2021). RNA was reverse transcribed using SuperScript IV (ThermoFisher Scientific), according to manufacturer’s instructions. cDNA was amplified using TaqMan™ Universal PCR Master Mix (ThermoFisher Scientific) and gene-specific TaqMan™ primers and probes against Collagen 1α1 (Col1α1), Vimentin (Vim), as well as the housekeeping gene, Peptidylprolyl isomerase A (Ppia). The reaction mixtures were subjected to 1 cycle of 2 m at 50°C; 1 cycle of 10 m at 95°C; 45 cycles of 15 s at 95°C, 1 m at 60°C using an AriaMx Realtime PCR System (Agilent Technologies, Santa Clara, CA). For assessment of Tenascin (Tnc) and Surfactant protein C (Sftpc), Glyceraldehyde 3-phosphate dehydrogenase (Gapdh) was utilized as a housekeeping gene, RNA was reverse transcribed using iScript cDNA synthesis kit (Bio-Rad, Hercules, CA) and RT-qPCR analysis was performed using iTaq SYBR Green supermix (Bio-Rad) in an Applied Biosystems 7,500 multicolor real-time PCR detection system (Applied Biosystems), as before (Bernau et al., 2021). The following SYBR Green primer sequences were used: Tnc forward: ACC​ATG​CTG​AGA​TAG​ATG​TTC​CAA​A, Tnc reverse: CTT​GAC​AGC​AGA​AAC​ACC​AAT​CC; Sftpc forward: GTA​GCA​AAG​AGG​TCC​TGA​TG, Sftpc reverse: CCT​ACA​ATC​ACC​ACG​ACA​A; Gapdh forward: GCC​CAA​GAT​GCC​CTT​CAG​TG, Gapdh reverse: CAT​CCA​CTG​GTG​CTG​CCA​AG. Prior to experimental use, standard curves and primer efficiencies were determined for the primer pairs above. Following all experimental RT-qPCR, relative changes in gene expression were quantified compared to control housekeeping gene using the ΔΔCt method.

Bronchoalveolar lavage (BAL) fluid was extracted from mouse lungs as before (Bernau et al., 2015; Bernau et al., 2021). Following anesthetization and exsanguination, mice were intubated by inserting a catheter into the exposed trachea. Lungs were lavaged five times with 0.8–1 ml of Dulbecco’s phosphate-buffered saline (DPBS, Corning Inc., Corning, NY), and BAL cells isolated, counted and cytospins prepared. Wright Giemsa staining (Fisher Scientific, Hampton, NH) was performed on the cytospins and differentials were counted by blinded observers to determine the ratio of neutrophils out of a total of 200 counted cells.

Logrank test was utilized for statistical analysis of mouse mortality. Two-way Analysis of Variance (ANOVA) with Šídák post-hoc test was utilized for analyses of mouse groups compared to voxel HU range or weight-loss. Pearson’s product-moment correlation analysis was utilized to establish a relationship between αSMA IHC quantification and hydroxyproline content, Ashcroft scores, µCT densitometry (voxels between −200 and 100 HU) or blood O₂ saturation, as well as the relationship between blood O₂ saturation and µCT densitometry (voxels between −200 and 100 HU and −500 to −400 HU). All other data were analyzed using Student’s t-test (to compare two conditions) or One-Way ANOVA with Holm-Šídák post-hoc test (to compare more than two conditions). For all analyses, p < 0.05 was considered statistically significant.

In order to monitor the effects of ZL0591 longitudinally in live mice, each week we collected serial blood O₂ saturations via pulse oximetry, as well as heart and respiratory rates. We first obtained these measurements on day 11 post-bleomycin injury, 3 days prior to commencing treatment with ZL0591, in order to determine the baseline differences between bleomycin-treated animals and controls. At this point, we found that blood O₂ saturation and heart rate were reduced in the bleomycin-treated group compared to uninjured controls, while the respiratory rate was unchanged between the two groups (Figures 1A–C). After initiation of ZL0591 administration, we continued to collect vital sign measurements weekly (on days 18 and 25). On day 18 (4 days after initiation of ZL0591 treatment), we did not observe any differences in the bleomycin/ZL0591 group compared to the bleomycin/vehicle controls (Figures 1D,E). In contrast, by day 25 (11 days after initiation of ZL0591 treatment and during the phase of mature fibrosis), we found a significant improvement in blood O₂ saturation in the bleomycin group treated with ZL0591 compared to bleomycin/vehicle controls (Figure 1F), consistent with an improvement in gas exchange. We did not observe a change in the heart rate as a result of ZL0591 treatment at this time point (Figure 1G). The observed improvement in blood O₂ saturation provided a real-time indication that treatment with ZL0591 may be modifying components of the fibrotic response.

The same treatment groups then underwent µCT imaging of the lungs to provide longitudinal observations of the fibrotic response (Scotton et al., 2013). At day 14 (prior to treatment with ZL0591), we observed an increase in radiodensity indicative of parenchymal fibrosis in bleomycin-treated lungs (Figure 2A), consistent with previous observations (Scotton et al., 2013; Ruscitti et al., 2017). Analysis of the distribution of voxel-based HU from volumetric analysis of lung µCT images is shown in Figure 2B. These data show a shift in the distribution of HU in the lung images from bleomycin-treated animals compared to controls, with a decrease in low-attenuation HU voxels (which correspond to air), and an increase in high-attenuation voxels corresponding to fluid/tissue. These results are consistent with an increase in soft-tissue density within the lung parenchyma from either edema or fibrosis, for example. µCT imaging of control- and ZL0591-treated groups was then performed on day 21 post-bleomycin (7 days into ZL0591) treatment. At this time point, we did not observe an overall change in radiodensity between the two bleomycin-treated groups (Figures 2C,D). However, µCT imaging at day 28 post-bleomycin (and 14 days into the ZL0591 treatment) showed a leftward shift in voxel HU distribution in the bleomycin/ZL0591-treated group compared to bleomycin-treated control animals, indicating a significant improvement in radiodensity, and thus diminished fibrosis, in this group (Figures 2E,F). Corroborating this finding, analysis of the highest attenuation HU, voxels from the µCT images (−200 to +100) that correspond with soft-tissue density (i.e., fibrosis), showed that bleomycin/ZL0591-treated mice had significantly fewer voxels in this range compared to bleomycin/control-treated animals (Figure 2G). It should be noted that we also observed that the percentage of high HU voxels inversely correlated with blood O₂ saturation (Supplementary Figure S1A). Furthermore, we found an increased voxel percentage in the−500 to −400 HU range under treatment with ZL0591 (Figure 2H), indicative of improvement in lung aeration. Notably, the percentage of voxels in this range positively correlated with blood O₂ saturation (Supplementary Figure S1B). Taken together, the µCT imaging data demonstrate that 14 days after treatment with ZL0591 (day 28 post-bleomycin injury), there was a decrease in µCT lung opacification (higher voxel HU), consistent with attenuation of overall fibrosis.

We then analyzed mouse lungs from each of the experimental groups described above to assess alterations in bleomycin-induced collagen deposition after treatment with ZL0591. Mouse right lung lysates were subjected to hydroxyproline assay to determine collagen content. In mice treated with ZL0591 starting 14 days after bleomycin injury and euthanized at day 21 (7 days of ZL0591 treatment), we observed a significant decrease in lung collagen content in comparison to mice treated with the vehicle control at the same time point (Figure 3A). Left lungs from the same mice were fixed, sectioned, and subjected to Masson’s trichrome staining, followed by semi-quantitative histologic scoring for fibrosis, using a modified Ashcroft method (Hübner et al., 2008). At this time point, we did not observe a significant difference in Ashcroft scores, but a trend was present (p = 0.0548) (Figure 3B). Given the observed trend toward reduced fibrosis after a 7-day treatment with ZL0591, we hypothesized that increased treatment duration may result in attenuation of the fibrotic response.

Thus, on day 28 after bleomycin injury and 14 days after initiation of ZL0591 treatment, we again determined hydroxyproline levels from lung lysates. In this case, we observed a similar, significant decrease in hydroxyproline in lungs from bleomycin-injured mice treated with ZL0591 compared to vehicle control, consistent with a reduction in collagen content (Figure 3C). In addition, Ashcroft scoring of histologic sections from left lungs indicated a reduction in fibrosis (Figure 3D), consistent with the decrease in collagen content that we observed at this time point. Taken together, these results suggest a time-dependent improvement in the amount of total fibrosis in response to ZL0591 treatment.

To ascertain whether the decreased fibrotic response was related to ZL0591s effect on the formation of a myofibroblast population, we utilized αSMA IHC staining of sectioned mouse lungs. We observed a loss of the myofibroblast marker, αSMA, in bleomycin/ZL0591-treated mouse lungs compared to bleomycin/vehicle controls at day 28 post-bleomycin (14 days following the start of ZL0591 treatment) (Figures 4A,B). This was consistent with a reduction in the myofibroblast population. At the same time point, we also found a reduction in bulk transcript levels of the mesenchymal marker Vim, along with reduction in myofibroblast associated ECM genes Tnc and Col1α1 (Figures 4C–E). These data support global downregulation of myofibroblast-associated genes by ZL0591. On the other hand, we did not detect a significant change in the transcription of markers of other cell types, including alveolar epithelial type 2 cell (AEC2) marker, Sftpc, suggesting that this cell population may be preserved (Figure 4F). Importantly, there was a significant correlation between myofibroblast presence (αSMA-positive IHC signal) and increased fibrotic burden, based on regression analysis of hydroxyproline, Ashcroft scoring, µCT densitometry (voxels between −200 and 100 HU) and (low) blood O₂ (Figures 4G–J). Altogether, these data indicate that inhibition of BRD4 activity by ZL0591 diminishes bleomycin-induced pulmonary fibrosis, possibly due to inhibition of myofibroblast transdifferentiation.

BRD4 is a multifunctional enzyme that is also implicated in the inflammatory response. Thus, we sought to assess whether our treatment approach had an effect on this aspect of the bleomycin injury. The inflammatory response to a single-dose intratracheal bleomycin-induced lung injury is most pronounced 3 to 7 days later, with increased numbers of total BAL cells and neutrophilia. In our study, we observed elevated numbers of BAL cells 14 days after bleomycin administration (Figure 5A), but an absence of a neutrophilic response (Figure 5B). By days 21 and 28 post-bleomycin injury, the total number of BAL inflammatory cells in bleomycin and vehicle treated mice was no longer increased compared to uninjured NS vehicle controls, and ZL0591 treatment did not modify this response (Figures 5C,D). At these time points, neutrophilic inflammation was low and we did not detect any significant ZL0591-induced changes at this late stage of bleomycin-induced fibrosis (Figures 5E,F). Finally, we assessed weight loss and survival after bleomycin administration. We found that initiation of ZL0591 treatment 14 days after bleomycin injury did not impact animal weight loss from that time point (Figure 5G). Similarly, we did not observe an effect of ZL0591 treatment on mouse mortality after initiation of treatment (Figure 5H).