Specific elastin degradation products are associated with poor outcome in the ECLIPSE COPD cohort

Chronic obstructive pulmonary disease (COPD) is characterized by a slow heterogeneous progression. Therefore, improved biomarkers that can accurately identify patients with the highest likelihood of progression and therefore the ability to benefit from a given treatment, are needed. Elastin is an essential structural protein of the lungs. In this study, we investigated whether elastin degradation products generated by the enzymes proteinase 3, cathepsin G, neutrophil elastase, MMP7 or MMP9/12 were prognostic biomarkers for COPD-related outcomes. The elastin degradome was assessed in a subpopulation (n = 1307) of the Evaluation of COPD Longitudinally to Identify Predictive Surrogate End-points (ECLIPSE) cohort with 3 years of clinical follow-up. Elastin degraded by proteinase 3 could distinguish between COPD participants and non-smoking controls (p = 0.0006). A total of 30 participants (3%) died over the 3 years of observation. After adjusting for confounders, plasma levels of elastin degraded by proteinase 3 and cathepsin G were independently associated with mortality outcome with a hazard ratio per 1 SD of 1.49 (95%CI 1.24–1.80, p < 0.0001) and 1.31 (95%CI 1.10–1.57, p = 0.0029), respectively. Assessing the elastin degradome demonstrated that specific elastin degradation fragments have potential utility as biomarkers identifying subtypes of COPD patients at risk of poor prognosis and supports further exploration in confirmatory studies.

Disease progression of chronic obstructive pulmonary disease (COPD) is slow and very heterogeneous, this is most likely consequent to different phenotypes with different disease trajectories that should be treated individually 1,2 . Phase III studies in COPD are long and costly, and consequently, there is a medical need to develop new and improved biomarkers that accurately identify COPD patients who progress within a short time period, consequent to a given disease phenotype which may be pharmaceutically attenuated. This is essential for the execution of improved phase II clinical studies that will allow confident phase III decision based on actual effects on forced expiratory volume in the first second (FEV 1 ) 3,4 .
Elastin is an essential structural protein of the lungs and is responsible for tissue elasticity 5,6 . Loss of the elasticity and elastin content during pathological situations is reported in inflammatory diseases including COPD with co-existing emphysema [7][8][9][10][11] . Tropoelastin, the monomeric form of elastin, has a unique structure that is composed of highly cross-linked and extremely hydrophobic domains, which renders it resistant to proteolytic degradation in healthy conditions 12,13 . Under pathological conditions such as COPD increased numbers of inflammatory cells and fibroblasts leads to an up-regulation of proteases including serine proteinases and matrix metalloproteinases (MMPs) 14 . Both excessive serine proteinase and MMP activity are associated with the destruction of elastin, resulting in specific pathological protein fragments and loss of lung elasticity 11,15 . These proteolytically processed fragments also referred to as neoepitopes are released into the circulation and may be assessed as simple 1
Baseline characteristics of smoking controls, non-smoking controls and a subgroup of age, BMI and gender-matched COPD participants are listed in Table 1. As expected, COPD subjects had lower FEV 1 and FEV 1 / FVC ratio compared to both control groups. Plasma ELP-3 level was significantly up-regulated in COPD when compared to levels of non-smoking controls (p = 0.0004), but not when compared to smoking controls (Fig. 2). No significant difference was observed for EL-CG, EL-NE, ELM7 and ELM12. Furthermore, no correlation  www.nature.com/scientificreports www.nature.com/scientificreports/ between the elastin fragments and FEV 1 or degree of emphysema defined as more than 10% of lung volume with a density of −950 Hounsfield units on inspiratory computed tomography was observed (data not shown).

Elastin degradation by proteinase 3 and cathepsin G is related to mortality. Baseline character-
istics for survivors and non-survivors are listed in Table 2. Non-survivors were significantly older, had increased mMRC dyspnea score, a higher number of previous exacerbations, a lower use of inhaled corticosteroids, and fewer subjects were current smokers ( Table 2). A total of 30 (3%) out of the 1000 COPD participants assessed died over the three years of observation. ELP-3 was significantly increased in non-survivors compared to survivors (p = 0.0202). No significant increase in EL-CG, EL-NE, ELM7 and ELM12 was observed (Fig. 3). After adjusting for relevant covariates, plasma levels of ELP-3 and EL-CG were independently associated with mortality with a hazard ratio per 1 SD increase in biomarker level of 1.

Discussion
There is a medical need for developing new and improved biomarkers to identify COPD subjects with a rapid disease progression, who potentially have a higher chance of benefitting from a symptom modifying COPD treatment. Here we investigated elastin degradation fragments, so-called neoepitopes, as blood-based biomarkers of disease progression for COPD in a non-invasive manner. We found that elastin degraded by protease 3 and cathepsin G, was associated with a higher risk of mortality in subjects with COPD.
Elastin is a unique signature protein of the lungs. Consequently, biomarkers generated from elastin could be associated with higher tissue accuracy and pathophysiological relevance; although these neoepitopes could potentially arrive from other organs than the lung. The importance of proteolytic elastin degradation in COPD is highlighted by the observation that exacerbations in COPD are associated with accelerated elastin turnover 4,10 . Desmosine and isodesmosine have long been proposed as biomarkers of lung tissue destruction as they are the crosslinking elements of elastin, released during lung tissue destruction 26 . It has been demonstrated that subjects with COPD have higher levels of plasma desmosine and that levels are able to predict mortality 27 which is in concordance with our findings. Moreover, a study with 349 subjects with one-third suffering from COPD showed that urinary desmosine was significantly correlated with all lung function measures 28 . This underlines the importance of elastin degradation as a potential biomarker in COPD, however, the desmosine technology identifies non-specific elastin fragments that are being generated through many different processes, both physiological and pathophysiological. The technique and biomarkers used in the present study allowed for investigation of elastin degradation predominantly produced by specific proteinases, known to be up-regulated in respiratory diseases and therefore more carefully quantify the aspects of the elastin degradome 10,[22][23][24] .
Plasma levels of ELP-3 and EL-CG were associated with all-cause mortality in COPD implying that proteinase 3 and cathepsin G play a significant role in COPD. During the progression of COPD, inflammatory cells infiltrate the lungs, which has been shown to release proteinase 3, cathepsin G and neutrophil elastase into the lungs 29 . These proteinases are known to efficiently degrade elastin 30 , which is in concordance with the results from this study where a higher degree of ELP-3 or EL-CG was significantly associated with a poor outcome. Moreover, proteinase 3 activity was present in the sputum of COPD subjects in a higher amount than the activity of neutrophil elastase implicating a bigger role for proteinase 3 in COPD than previously thought 31 . This may in part explain why ELP-3 demonstrated a better association with outcomes relative to EL-NE. Since plasma levels of ELP-3 and EL-CG were associated with all-cause mortality, a correlation with FEV 1 might have been expected, as this was previously observed for desmosine 28 . In general, the FEV 1 decline for the subpopulation of ECLIPSE studied here were associated with a considerable intra and inter-person variability which makes an evaluation of predictive www.nature.com/scientificreports www.nature.com/scientificreports/ biomarkers for lung function decline challenging, 31% improved or slightly decreased in FEV 1 while only 38% participants in the entire cohort demonstrated significant FEV 1 decline 32 .
The importance of unique elastin fragments has been emphasized by the fact that they can act as chemotactic peptide for different cell types, showing their ability to function as matrikines 33 . The hexapeptide VGVAPG within tropoelastin is well known for its chemotactic activity attracting monocytes and fibroblasts and its ability to regulate MMP expression and activity 34,35 . Likewise, other fragments of elastin corresponding to XGXXPG, where X is a generic hydrophobic residue, has also been shown to be active peptides [35][36][37] . In fact, the ELM12 fragment holds the sequence VGVAPG in its peptide, which indicates that it might be a matrikine, however, in the current study, no pathological relevance of this biomarker was observed.
The emphysema phenotype, such as the multi-organ loss of tissue (MOLT), is currently receiving increased attention, and high elastin turnover could be associated with it, which has been shown by others 23,38,39 . In accordance with our findings, another study measuring plasma desmosine in the ECLIPSE cohort was not able to show a relationship between emphysema and elastin degradation 27 . The lack of association could be explained by the fact that COPD participants from the ECLIPSE cohort had an established disease (GOLD II-IV) with a stable state during sampling. This could result in less elastin present in their lungs or a lower disease activity than during an exacerbation, which could explain why elastin degradation is not associated or increased in subjects with emphysema in this cohort.
The limitations of this study include the low number of deceased subjects, even though we were able to detect a significant association with all-cause mortality for the two biomarkers ELP-3 and EL-CG. To generalize these finding to a more general COPD population they have to be confirmed in secondary cohorts. Furthermore, using the subpopulation of the full ECLIPSE study comprising the study participant that progress the least and most in terms of FEV 1 decline during the study period might explain the low number and make a less optimal www.nature.com/scientificreports www.nature.com/scientificreports/ subpopulation to study mortality. In addition, the ELP-3, EL-CG, ELM7 and ELM12 was measured in heparin plasma at year 1, whereas EL-NE was measured in 6-month serum. This could create some difficulties in directly comparing the results.

Conclusion
In conclusion, we have demonstrated that five cleavage-specific fragments of elastin generated by five different proteinases reflect different pathological processes in COPD. ELP-3 and EL-CG demonstrated promise as prognostic biomarkers for all-cause mortality reflecting an increase of elastin remodeling by proteinase 3 and cathepsin G. This study demonstrated the importance of evaluating elastin turnover in the pathology and natural history of COPD.  Data are shown as mean (95% CI) hazard ratio for 1 log SD increase in biomarker for all-cause mortality and adjusted for age, smoking status, BODE index, mMRC dyspnea score, inhaled corticosteroids and number of exacerbations the previous year of study start. Asterisks indicate statistical significance *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001.  www.nature.com/scientificreports www.nature.com/scientificreports/

Methods
Study design and participants. The analysis was based on the three-year observational longitudinal study ECLIPSE (ClinicalTrials.gov. number, NCT00292552), described previously 32,40 . The full ECLIPSE study included 2163 participants with COPD. The enrollment criteria included an FEV 1 of less than 80% of the predicted value and a FEV 1 /forced vital capacity (FVC) ratio of 0.7 or less assessed after the use of bronchodilators. COPD participants had a smoking history of 10 or more pack-years. 343 smoking controls with a smoking history of 10 or more pack-years and 223 nonsmoking controls were also included in the study. The controls had an FEV 1 of more than 85% of the predicted value and an FEV 1 /FVC ratio of 0.7 or more after the use of bronchodilators. Control participants had to be free of significant comorbidities, determined from screening investigation, physical examination and medical history. Eight study visits were conducted at baseline, month three, six and subsequently every six months over three years. The current analysis was performed on 1307 participants consisting of 1000 COPD, 207 smoking controls and 100 non-smoking controls. All-cause mortality was recorded until year 3. The study was conducted according to the Declaration of Helsinki and Good Clinical Practice guidelines and was approved by relevant ethics and review boards (Supplementary Table 1). Participants provided informed consent.

Quantification of serological biochemical biomarkers.
Whole blood was collected from fasting participants and transferred to vacutainers containing sodium heparin. Plasma was obtained by centrifugation of vacutainer tubes at 2,000 g for 10-15 minutes and stored at −80 °C until analysis. Elastin degradation by proteinase 3 (ELP-3), cathepsin G (EL-CG) and MMP9/12 (ELM12) was measured in heparin plasma obtained from 1307 study participants at the year 1 visit using well-validated ELISAs each utilizing monoclonal antibodies targeting a specific neoepitope (Nordic Bioscience, Herlev, Denmark), see specifications in Table 3. Measurements were performed in a blinded manner according to the instruction of the manufactures. Previously, elastin degradation by neutrophil elastase (EL-NE) was measured in month six serum samples whereas elastin degraded by MMP7 (ELM7) was measured in year 1 heparin plasma 41 . The five unique elastin neoepitope biomarkers investigated in this study ELP-3, EL-CG, EL-NE, ELM7 and ELM12 originates from locations throughout the tropoelastin (Fig. 5) and are very different in terms of activation, inhibition, substrate specificity and source (   www.nature.com/scientificreports www.nature.com/scientificreports/ deviation (SD) increase in biomarker level. Logistic regression was used to find the odds ratio for all-cause mortality belonging to the upper quartile versus the lower quartile. The risk of death was assessed adjusted for confounders that were significantly different between survivors and non-survivors. The covariates adjusted for were age, smoking status, mMRC dyspnea score, use of inhaled corticosteroids and number of exacerbations in the year prior to blood sampling.
The software MedCalc (MedCalc version 14.8.1, MedCalc software bvba, Ostend, Belgium) was used to perform all statistical analysis.

Data Availability
The dataset generated during and analyzed during the current study are available from the corresponding author on reasonable request.