CELA1 Mediates Progressive Emphysema in Alpha-1 Antitrypsin Deficiency

Chymotrypsin-like elastase 1 (CELA1) is a serine protease that is neutralized by α1-antitrypsin (AAT) and prevents emphysema in a murine antisense oligonucleotide model of AAT-deficient emphysema. Mice with genetic ablation of AAT do not have emphysema at baseline but develop emphysema with injury and aging. We tested the role of CELA1 in emphysema development in this genetic model of AAT-deficiency following tracheal lipopolysacharide (LPS), 8 months of cigarette smoke (CS) exposure, aging, and a low-dose tracheal porcine pancreatic elastase (LD-PPE) model. In this last model, we performed proteomic analysis to understand differences in lung protein composition. We were unable to show that AAT−/− mice developed more emphysema than wild type with LPS. In the LD-PPE model, AAT−/− mice developed progressive emphysema from which Cela1−/−&AAT−/− mice were protected. In the CS model, Cela1−/−&AAT−/− mice had worse emphysema than AAT−/−, and in the aging model, 72–75 week-old Cela1−/− &AAT−/− mice had less emphysema than AAT−/− mice. Proteomic analysis of AAT−/− vs. wildtype lungs in the LD-PPE model showed reduced amounts of AAT proteins and increased amounts of proteins related to Rho and Rac1 GTPases and protein oxidation. Similar analysis of Cela1−/−&AAT−/− vs. AAT−/− lungs showed differences in neutrophil degranulation, elastin fiber synthesis, and glutathione metabolism. Thus, Cela1 prevents post-injury emphysema progression in AAT-deficiency, but it has no effect and potentially worsens emphysema in response to chronic inflammation and injury. Prior to developing anti-CELA1 therapies for AAT-deficient emphysema, an understanding of why and how CS exacerbates emphysema in Cela1 deficiency is needed.


Introduction
Alpha-1 antitrypsin (AAT) de ciency is a rare disease affecting approximately 1:2000 Caucasians with 10% of these individuals developing AAT-de cient lung disease (1,2). This lung disease is marked by progressive alveolar loss despite withdrawal of triggering agents such as cigarette smoke (3). Although the PiZZ genotype is the most common mutation, there are over 100 described variants making true population estimates di cult (4) and AAT-de ciency is typically diagnosed by reduced levels of AAT in the blood. Typically developing in the 4th and 5 th decades of life, AAT-de cient lung disease is marked by progressive emphysema and is the 4 th leading indication for lung transplantation (1). AAT augmentation therapy does not prevent disease progression (5) making the development of new therapeutic approaches critical.
Chymotrypsin-like elastase 1 (CELA1) is a serine protease synthesized and secreted by alveolar type 2 cells with a physiologic role in reducing postnatal lung elastance (6). At a molecular level, CELA1 binds and cleaves non-crosslinked, hydrophobic domains of tropoelastin, and its binding to lung elastin bers is increased with strain-similar to other pancreatic elastases (7). CELA1 is neutralized by covalent binding with AAT, and Cela1 -/mice were completely protected from emphysema in an antisense oligonucleotide model of AAT-de cient emphysema (6). This model however did not include any injury apart from administration of the antisense oligonucleotide with levels of emphysema exceeding that seen in mice with genetic ablation of ve Serpina1 paralogues and subjected to tracheal lipopolysacharide or cigarette smoke (8). Here, we use this murine genetic model of AAT-de ciency to test the role of CELA1 in AATde cient emphysema using multiple models to show that CELA1 has a role in progressive airspace enlargement in AAT-de ciency that is independent of in ammation.

Materials And Methods
Animal Care and Use Animal use was approved by the Cincinnati Children's Hospital Institutional Animal Care and Use Committee (#2020-0054) with food and water provided ad lib with 12-hour day/night cycles. For all experiments, approximately equivalent numbers of C57BL/6 male and female mice were use. Cela1 -/-(6) and Alpha-1 antitrypsin-de cient (AAT-de cient, Serpina1 em3Chmu /J, JAX Stock# 035015) mice were used and genotyped per published protocols (8).

Tracheal Porcine Pancreatic Elastase Model
In 8-10-week-old mice, 0.2 to 10 units of porcine pancreatic elastase (PPE, Sigma #E1250, diluted to 20 units or 2 units per mL in PBS) was administered to mice after anesthetization with 2% iso urane, suspension by the incisors on a rodent intubating board, tracheal cannulation with a 22-gauge angiocatheter and administration of PPE during inspiration. Mice were recovered and sacri ced at speci ed time intervals after intraperitoneal administration of ketamine/xylazine/acepromazine 70/5/2 mg/kg followed by exsanguination and vital organ harvest.

Cigarette Smoke Exposure Model
Beginning at 10-12 weeks of age, mice were exposed to 4 hours of whole-body cigarette smoke ve days per week for ten months using a Teague TE-10z smoking machine using smoke generated from 3R4F Kentucky Reference Cigarettes (University of Kentucky) at a concentration of 150 mg/m 3 total suspended particulates. This work was performed under approval from Institutional Animal Care and Use Committee at the University of Cincinnati Medical Center Protocol# AM06-20-10-28-01.

Aging Model
Male and female mice were allowed to age to 72-75 weeks and sacri ced as above.

Lung Tissue Processing and Analysis
After exsanguination, a neck incision was made and tracheal cannula secured with 3-0 silk suture around the trachea, left bronchus ligated with a surgical clip and left lung snap frozen, and right lung in ated with 2% paraformaldehyde in PBS at 20 cm water pressure. The trachea was ligated with removal of the cannula, xed overnight in 2% PFA, dehydrated to 100% methanol, para nized, and right lung lobes randomly oriented, embedded, sectioned at 5µm, and mounted. Sections were H&E stained and up to ve 10X images obtained per lung lobe. Mean linear intercepts of blinded sections were determined using the methods of Dunnill (9). Airspace diameters of tile-scanned 4X sections were determined using a previously described MatLab (MathWorks, Natick, MA) code (10).

Biochemical Assays
The upper half of snap frozen left lung lobes was homogenized in PBS and myeloperoxidase activity quanti ed using published protocols (11).

Proteomic and Bioinformatic Analysis
Frozen left lung lobes were cut in half, proteins were extracted using the MPLEx methods as previously described (12). For each sample, extracted proteins were denatured, alkylated, digested with trypsin, and desalted on a C18 solid-phase extraction (SPE) cartridge (Discovery C18, 1 ml, 50 mg, Sulpelco). The peptide concentration was measured by BCA assay (ThermoScienti c). Samples were analyzed using a Label free LC-MS/MS strategy. Brie y, 5 μl of samples at the concentration of 0.1 μg/μL were loaded on a SPE column and separated on a reverse phase C18 analytical column using a 180-min gradient.
MS data was recorded 60 min after injection. The e uents from the LC column were ionized by electrospray ionization by applying 2,200 V to the electrospray tip. Electrosprayed ions were introduced into the mass spectrometer (Orbitrap Q Exactive plus) via a heated capillary. The resulting ions were mass analyzed by the Orbitrap at a resolution of 70,000 covering the mass range from 300 to 1,800 Da with a maximum injection time of 20 ms and automated gain control (AGC) setting of 3e6 ions. Most abundant ions were subjected to MS2 analysis using a top12 strategy (top 12 ions are fragmented during between two precursor scans). For MS2, ions with charge states different than 1 were isolated by quadrupole mass lter in monoisotopic peak selection mode using isolation window of 1.5 Da, maximum injection time of 50 ms with AGC setting at 1e5 ions and fragmented by high-energy collision dissociation (HCD) with nitrogen at 30% normalized collision energy. Fragment ions were mass analyzed by the Orbitrap at a resolution of 17,500 covering the mass range from 200 to 2,000 Da, and spectra were recorded in the centroid mode. Ions once selected for MS2 were dynamically excluded for the next 30 seconds.

Statistical Analysis
Comparisons were made using Welch's t-test or ANOVA with Holm-Sidak post hoc comparison as appropriate using rstatix (17) using R-4.1.0 (18). Plots were created using ggpubr (19) with central lines indicating mean and whiskers one standard deviation. p-values of <0.05 were considered statistically signi cant.

CELA1 in Tracheal Lipopolysacharide Model AAT-de cient Emphysema
Borel, et al previously reported that AAT-de cient mice developed more emphysema at 14 days than wildtype mice after tracheal doses of 1 and 0.5 units of lipopolysacharide at days 1 and 10 (8). We intended to test whether Cela1 -/-&AAT-de cient mice were protected from emphysema in this model. However, we observed a similar degree of emphysema in wild-type and AAT-de cient mice ( Figure 1A), with male and female mice similarly affected. We then performed a dose titration experiment with 1&0.5, 2&1, 5&2.5, and 10&5 units of LPS with evaluation at 21 days instead of 14 and evaluated airspace sizes using a previously described method that generates descriptive statistics of airspace sizes in tile-scanned images of all lobes (6, 10). There was no clear dose-effect and wildtype mice experienced as much or more emphysema as AAT-de cient mice (Supplemental Figure 1). Since in our hands, a previously reported LPS-based model of AAT-de cient emphysema could not be reproduced, we developed a different model of AAT-de cient emphysema.

Low-Dose PPE Model of AAT-de cient Emphysema
Tracheal porcine pancreatic elastase (PPE) is a well-established model that causes signi cant injury and, in some studies, progressive emphysema (20,21). In most studies, 1-2 units of PPE is administered tracheally, and emphysema is assessed at or before 21 days. We administered 2 units of PPE to AATde cient mice, and 6 of 8 mice died with extensive lung hemorrhage on necropsy. We performed a dose titration experiment and discovered that 0.2 units of tracheal PPE caused substantial emphysema in AATde cient mice at 21 days while wild type mice had virtually no emphysema at this dose. Furthermore, emphysema progressed over time ( Figure 1B&C, Supplemental Figure 2). We therefore administered 0.2 units of tracheal PPE to AAT-de cient and Cela1 -/-&AAT-de cient mice to test for a role for CELA1 in emphysema in the context of AAT-de ciency.

Lung Proteomic Changes in Low-Dose PPE Model
We performed unbiased proteomics of wild type, AAT-de cient, Cela1 -/and Cela1 -/-&AAT-de cient lungs ( ve per genotype) 42 days after low-dose PPE to understand the differential response to injury. In comparing wildtype and AAT-de cient lungs, only one protein had an adjusted p-value<0.1 and at least 2fold change. This protein was the AAT isoform Serpina1d that is one of the ve AAT paralogues deleted in this mouse(8). Seventeen genes had protein levels at least 2-fold lower in AAT -/compared to wildtype lungs, and 71 genes were at least 2-fold higher (Supplemental Table 1). As expected, many of the most downregulated genes in this analysis were AAT paralogues (e.g., Serpina1a, Serpina1b, and Serpina1d). Genes of many of the upregulated proteins were related to Rho & Rac1 GTPases (Abr and Srgap2), thrombosis and complement (Klkb1, C2 and Mpi), and protein oxidation (Hacd2 and Ermp1) ( Figure 2A). Gene ontogeny enrichment for upregulated protein identi ed neutrophil degranulation, functions related to innate immunity, and nucleotide biosynthesis pathways as the most activated ( Figure 2B). The statistical signi cance of downregulated pathways was greater and largely involved processes related to mRNA translation ( Figure 2C). The protein-protein interaction network of both increased and decreased proteins showed many edges related to ribosomes and translation, and that many of the proteins related to neutrophil degranulation were either not adjacent in the network or not within the protein-protein interaction network at all ( Figure 2D, Supplemental Figure 3). These data are consistent with the expected pulmonary biology of AAT-de ciency with regards to AAT synthesis and innate immunity but also some unexpected changes with regards to protein synthesis.

CELA1 in Low-Dose PPE Model of AAT-De cient Emphysema
We tested for a role for CELA1 in progressive emphysema by quantifying mean linear intercepts at 21, 42, and 120 days after low-dose PPE in AAT-de cient and Cela1 -/-&AAT-de cient mice. Compared to AATde cient mice, Cela1 -/-&AAT-de cient mice had less emphysema at 21, 42, and 120 days after low dose PPE administration ( Figure 3A&B). Overtime, AAT-de cient but not Cela1 -/-&AAT-de cient mice experienced progressive emphysema; between days 42 and 120, AAT-de cient mice had a 31% increase in mean linear intercept (p=0.002) while the increases in this groups between 21 and 42 days or and in the Cela1 -/-&AAT-de cient mice over time were smaller and not signi cant. We observed that at 42 and 120 days, female Cela1 -/-&AAT-de cient but not female AAT-de cient mice had more emphysema than male mice (p=0.02 and p=0.01 respectively). However, when analyzing by sex, the 120-day MLI differences remained signi cant for males (p=0.003) and females (p=0.03). These data indicate that CELA1 plays a role in AAT-de cient emphysema and perhaps this role is more pronounced in males than females.

Impact of CELA1 on Lung Proteome in Low-Dose PPE Model
We then compared the lung proteomes of Cela1 -/-&AAT-de cient and AAT-de cient mice 42 days after tracheal administration of low-dose PPE. There were only two proteins with adjusted p-values of less than 0.1 and at least 2-fold change. The genes of these proteins were UBX domain-containing protein 7 and Ig alpha chain C region. There were 19 proteins that were at least 2-fold increased and 28 proteins that were at least 2-fold decreased in Cela1 -/-&AAT-de cient compared to AAT -/mice (Supplemental Table 2). Among the proteins with increased abundance, some of the most signi cant were from genes related to oxidative stress (Glo1, Msra, and Aldh1b1) and acute in ammation (Saa4 and C8a). Proteins that were most increased were from genes related to ubiquitination (Ubxn7 and Ube2o), elastin bers (Mmrn2), and neurotransmitters (Nostrin, Palmd, Agap3, and Vps33a) ( Figure 4A). Gene ontogeny enrichment for proteins that were increased showed abundance of Rho GTPase proteins, nitric oxide signaling, and muscle-related proteins ( Figure 4B). Downregulated processes included elastic ber formation and neutrophil degranulation ( Figure 4C). Protein-protein interaction analysis of increased and decreased proteins identi ed a number of small but connected networks related to elastin bers, neutrophil degranulation, and glutathione metabolism ( Figure 4D). These data demonstrate that Cela1 mediates protein-level differences in elastin micro bril associated proteins in this AAT-de cient model of emphysema and that there are also potential differences in neutrophil activity and oxidative state.

CELA1 in Cigarette Smoke Induced Emphysema
As cigarette smoke exposure is highly associated with the development of emphysema in AAT-de cient individuals, we performed whole body cigarette smoke exposure to AAT-de cient and Cela1 -/-&AATde cient mice ve days a week for 10 months. Contrary to our expectations, and Cela1 -/-&AAT-de cient mice were not protected from CS-induced emphysema and Cela1 -/-&AAT-de cient mice had more evidence of airspace destruction than AAT-de cient mice ( Figure 5A&B). Given the proteomic ndings of altered neutrophil associated and chemotaxis proteins in Cela1 -/-&AAT-de cient mice, we hypothesized neutrophil activity might be increased. However, myeloperoxidase activity was reduced by more than half in Cela1 -/-&AAT-de cient mice compared to AAT-de cient mice in response to CS and comparable to activity levels seen 42 days after PPE and PBS. (Figure 5C). While the mechanism is unclear, these data indicate the absence of Cela1 enhances airspace destruction in AAT-de cient mice response to sustained cigarette smoke exposure.

CELA1 in Age-Related Alveolar Simpli cation
In humans and mice, aging is associated with slowly progressive airspace simpli cation, and AATde cient mice were reported to have more rapid simpli cation that wildtype (8). We validated these ndings and discovered that Cela1 -/-&AAT-de cient mice experienced a level of age-related airspace simpli cation similar to that of wildtype mice (Figure 6).

Discussion
This is the rst study to show in a novel mouse model of injury-induced emphysema in AAT-de ciency, that Cela1 is required for progressive airspace enlargement, and that this gene is important in age related alveolar simpli cation. Interestingly, Cela1 did not play a role in cigarette smoke-induced emphysema suggesting that the mechanism by which chronic in ammation and injury causes emphysema is independent of the post-injury remodeling program. When considered in the context of our previous work showing that Cela1 binds to lung elastin in the context of strain (6), these ndings support the concept that altered local lung mechanics in emphysema promote CELA1-mediated remodeling leading to progressive emphysema and that the lack of neutralization of Cela1 protein by AAT exacerbates this process.
The four models used in our studies help elucidate the role of CELA1 in AAT-de cient emphysema. In our rst experiment, we were unable to recapitulate the emphysema seen with LPS administration in AATde cient mice (8). Potential reasons for this could be differences in microbiome, reduced LPS potency, or differences in administration technique. The nding that increasing doses of LPS caused more airspace injury in wildtype than AAT-de cient mice supports the conclusion that some intrinsic difference between facilities is responsible-a hypothesis that could lead to improved understanding of emphysema pathogenesis if followed up upon. For our second experiment, we modi ed the tracheal PPE model for use in AAT-de cient mice. The observed pulmonary hemorrhage and lethality with standard dose of PPE in AAT-de cient mice is probably because after tracheal PPE administration, there is leakage of serum into the airspace allowing AAT to neutralize of the proteases in PPE. In AAT-de cient mice, the proteases in PPE remain active for a longer period leading to more extensive damage and death. Low-dose PPE administration to AAT-de cient mice produced robust emphysema that was progressive-an important clinical feature of AAT-de cient emphysema. The cigarette smoke exposure model is more clinically relevant than the PPE model, but it is dissimilar to human AAT-related lung disease because while smoking cessation is a mainstay of treatment for human AAT-related lung disease, in the mouse model, exposure is continuous. Given the relatively small increase in MLI observed, cessation of cigarette smoke exposure experiments would need a large number of mice and are probably inappropriate for Cela1 studies since some consequence of its absence seems to worsen CS-induced emphysema. Notably, Steams et al recently showed that AAT knockdown using antisense oligonucleotides increased Cela1 expression but did not worsen lung injury in a cigarette smoke and in uenza model (22). Lastly, we validated previously reported ndings regarding age-related alveolar simpli cation in AAT-de cient mice (8) and showed that CELA1 plays an important role in this process. In considering these ndings in the context of the physiologic role of CELA1-namely the reduction of postnatal lung elastance (6)-data from these four models demonstrate that in the context of AAT-de ciency, CELA1 does not appear to be involved in the lung's response to injury but is important for continued airspace destruction once emphysema is established.
This study adds to our understanding on the role of sex on AAT-de cient emphysema. We previously reported that in the antisense oligonucleotide model, AAT-de cient mice had more signi cant emphysema (23), but we did not see these same changes in the genetic model. In the current study, while Cela1 ablation reduced emphysema in both male and female AAT-de cient mice in the LD-PPE model, the protection was greater in male mice. Previous studies did not show any effect of sex on lung Cela1 expression (6). These ndings suggest that CELA1 could play a role in the increased sensitivity of AATde cient female patients compared to male ones (24).
The progressive nature of emphysema in mouse models controversial. In mice, rats, and rabbits studies show that emphysema improves (25), is stable (26-28), or progresses (20,21,29) over time. While comparison between studies of different species and doses of PPE are di cult, it does appear that studies with stable or improved MLI have lesser degrees of initial injury than those that show progression. The role of localized lung tissue mechanics in emphysema progression was convincingly demonstrated by Toumpanakis et al. Mice treated with PPE were followed to 21 days with subsequent reduction of tracheal cross-sectional area by 50% for 24 or 72 hours. Emphysema was substantially greater with 72 hours of air ow restriction (30). Perhaps there is threshold of alveolar injury beyond which localized tissue mechanics become too deranged to make re-alveolarization possible; a hypothesis that warrants future study as understanding key processes of repair-regeneration in emphysema could lead to novel therapies to restore lung function in emphysema. There is likely an age-related component to repair potential as well since older mice have less compensatory lung growth than younger ones (31). While in human emphysema, alveolar loss is considered permanent, adult human lung has the potential to reinitiate alveolarization post-pneumonectomy (32) lending hope to the idea that if emphysema progression could be halted, other therapies could be employed to regenerate damaged lung tissue.
Our study has several limitations. First, we did not determine if Cela1 -/-&AAT-de cient mice had preservation of lung function or diffusing capacity, and we did not measure lung volume in a way that would have permitted calculation of alveolar surface area. Our aged mice were harvested at 15 months and not 18 or 24 months as in some studies. We decided on this time point after an unexpected mortality in an aged AAT-de cient mice at 15 months and concern for losing additional mice. To the best of our knowledge, there are no reports of early mortality in AAT-de cient mice. We previously identi ed CELA1 as being expressed in alveolar type 2 cells, but we did not provide any cell-speci c information here. Lastly, while we showed a speci c role for CELA1 in this process, we did not interrogate the role of other proteases or test whether AAT-replacement was su cient to protect mice from emphysema in the various models tested. As AAT has anti-in ammatory properties (33), it may be that CELA1 inhibition and AAT replacement therapy could be complementary.
Conclusions CELA1 mediates progressive airspace destruction in AAT-de cient emphysema and its inhibition may represent a novel therapy in AAT-related lung disease.   Lung Proteomics in LD-PPE model of AAT-de cient Emphysema. (A) A volcano plot of detected proteins 42 days after LD-PPE showed the expected low abundance of Serpina1a, Serpina1b, Serpina1c, and Serpina1d (AAT) paralogues and increased abundance of proteins related to Rho & Rac1 GTPases Abr and Srgap2, related to thrombosis and complement Klkb1, C2 and Mpi, and related to protein oxidation Hacd2 and Ermp1. Abundance of the matrix metalloproteinase Mmp9 was increased but serine proteases neutrophil elastase (Elane) and cathepsin G (Ctsg) were little changed. (B) Enrichment analysis of proteins with increased abundance showed that processes related to neutrophil function, innate immunity, nucleotide metabolism, or long chaing fatty acid metabolism were increased. Scaled log10(pvalues) are indicated in the right column with red indicating increased. (C) Enrichment of proteins with decreased abundance showed stronger statistical signi cance (blue bars) and were largely related to Rho GTPases and protein translation. (D) A protein-protein interaction network of proteins with signi cant (p<0.05) differences in abundance identi ed a number of interconnected proteins and processes with ribosomal protein synthesis, Rho-GTPases, and biological adhesion being some of the most abundant.  Cela1 -/-&AAT -/vs. AAT-de cient lung 42 days after low-dose PPE showed a reduction in proteins related to oxidative stress (Glo1, Msra, and Aldh1b1) and acute in ammation (Saa4 and C8a). Proteins related to protein homeostasis (Ubxn7 and Ube2o), elastin bers (Mmrn2) and related to nitric oxide and glutamic acid neuronal signaling (Nostrin, Palmd, Agap3, and Vps33a). (B) Enrichment analysis of proteins with increased abundance in Cela1 -/-&AAT -/lungs compared to AAT -/lungs demonstrated increased abundance of proteins related nitric oxide and Rho GTPase signaling and processes related to muscle function. The red bar represents scaled statistical signi cance. (C) Analysis of proteins with decreased abundance identi ed differences in elastin ber formation and neutrophil degranulation. Blue bars represent scaled statistical signi cance. (D) A protein-protein interaction network of proteins with signi cant (p<0.05) differences in abundance several smaller networks related to elastic bers, neutrophile degranulation, and glutathione metabolism. Cela1 and CS exposure in AAT-de ciency. (A) After 10 months of whole-body cigarette smoke exposure, Cela1 -/-&AAT-de cient mice had more airspace simpli cation than AAT-de cient mice. Sex is shown for comparison only and was not used in the shown comparison. ***p<0.001 by Welch's t-test. (B) representative H&E images of CS-exposed mouse lung. (C) Lung myeloperoxidase activity was greater in AAT-de cient mice exposed to CS compared to Cela1 -/-&AAT-de cient mice and at 42 days post-PBS or PPE in both genotypes *p<0.05 **p<0.05 by Dunn's post hoc test. Wilcoxon rank sum p<0.01.