Introduction

Chronic obstructive pulmonary disease (COPD) is a widespread, multifactorial airway disease characterized by persistent respiratory symptoms, airflow obstruction, and progressive, irreversible damage to the lungs, often resulting from a combination of chronic inflammation, oxidative stress, and structural changes in both the small airways and lung parenchyma1. COPD is primarily caused by long-term exposure to tobacco smoke or environmental pollutants, with genetic factors and respiratory infections also contributing to chronic lung inflammation and damage. Acute exacerbations of COPD (AECOPD) are defined as sudden worsening of lung function and symptoms, mainly caused by inflammatory responses to bacterial or viral infections1. The primary aims of AECOPD treatment are to alleviate symptoms, to reduce the severity and duration of exacerbations, and to prevent rapid loss of lung function by using, for example, bronchodilators, corticosteroids or antibiotics1. However, treatment options are very limited and exacerbations remain a major disease burden in COPD. Therefore, better strategies for prediction, early detection, and prevention of AECOPD as well as new specific anti-inflammatory and immune-modulating drugs are urgently needed.

The translation of findings from preclinical studies to humans is a crucial step in early clinical drug development. Preclinical studies, often conducted in vitro or in animal models, provide essential initial data on the efficacy, toxicity, and pharmacokinetics of new drug candidates. However, significant genetic, anatomical and physiological differences between animal and human lungs restrict translation, emphasizing the importance of early human proof-of-concept studies2,3. The mentioned differences can result in variations in drug absorption, distribution, metabolism, and excretion, leading to discrepancies in drug efficacy and safety profiles observed in preclinical studies versus clinical trials. The translation of findings from preclinical studies to humans usually involves initial studies with healthy volunteers. However, one significant limitation of non-diseased organs in healthy volunteers is the absence of specific pathological features including pathway activation, relevant cell migration, and inflammation3. Human challenge models with e.g., endotoxin, allergen or infectious compounds offer the possibility of inducing specific inflammatory responses in healthy volunteers that resemble aspects of diseases such as AECOPD and are therefore valuable for testing the efficacy of investigational new drugs or vaccines4,5,6.

Segmental lung challenge with lipopolysaccharide (LPS), an endotoxin from the outer membrane of gram-negative bacteria, is a well-established method to induce transient airway inflammation in healthy participants6. This model involves the direct instillation of LPS into a specific lung segment via bronchoscopy, which leads to localized inflammation. The inflammatory response induced by LPS can be thoroughly assessed through bronchoalveolar lavage (BAL) fluid collected from the challenged lung segment. Notably, there is a significant infiltration of inflammatory cells into the lung following LPS challenge, which is dominated by neutrophil granulocytes7. Therefore, the LPS challenge model mimics some aspects of COPD and exacerbations thereof8,9. In line, transcriptomic analysis of BAL-derived cells from the segmental LPS challenge model revealed overlapping features with respiratory diseases in general and infection-triggered respiratory insults such as AECOPD or community-acquired pneumonia in particular10. Furthermore, pro-inflammatory cytokines such as interleukin 6 (IL-6), IL-8, albumin, and myeloperoxidase (MPO) are increased in BAL and sputum after LPS challenge7. However, to our knowledge, there is no comprehensive analysis of the human proteome in the LPS challenge model and its overlap with AECOPD to date. While investigation of individual selected proteins allows only limited interpretation, proteomic profiling offers the opportunity to identify new biomarkers, protein-interactions and signalling pathways that are involved in infection-triggered inflammatory responses. Such insights are crucial for understanding the molecular basis of airway inflammation in general and endotoxin-induced inflammation in particular.

The aim of this study was to characterize the proteomic profile in BAL fluid following LPS challenge and to contextualize the data with AECOPD processes.

Materials and methods

Study design and ethics approval statement

Healthy participants underwent bronchoscopy with segmental LPS challenge. Pre-challenge baseline BAL was collected followed by segmental instillation of LPS and saline in the contralateral lung as control. After 24 h, BAL was sampled from the challenged lung segments. Participant demographics and further details of the study design, including clinical monitoring procedures and outcomes, have been reported previously11. Here, a total of 1,500 proteins were assayed in cell-free BAL supernatants using the SomaLogic SomaScan platform. Protein data were analyzed and compared with a systematic review by Chen et al.12, where selected blood biomarkers of exacerbations in COPD were elaborated. The protocol was approved by the independent Ethics Committee of the Hannover Medical School (Approval No. 7193) and registered at clinicaltrials.gov (NCT03044327, 07/02/2017) prior to study start. The study was conducted at the Fraunhofer Institute for Toxicology and Experimental Medicine, Hannover, Germany in accordance with the Declaration of Helsinki and the International Council for Harmonisation (ICH) Harmonised Tripartite Guideline for Good Clinical Practice (GCP).

Participants and consent to participate

Ten healthy, non-smoking participants (7 males and 3 females, smoking history < 1 pack-year, mean age: 38 ± 10 years) with a body mass index (BMI) of 18.5–29.9 kg/m2, and normal lung function (forced expiratory volume in 1 s (FEV1): 101 ± 14% predicted) were included. Written informed consent was obtained from all participants after they were fully informed about all trial-related aspects before any study-related procedures.

Sample collection

Bronchoscopies were performed according to the guidelines for investigative bronchoscopies13,14. Participants underwent a first bronchoscopy to collect pre-challenge baseline BAL from a segment of the left lower lobe using 100 mL (5 aliquots of 20 mL) of pre-warmed saline, following segmental challenge with LPS (40 endotoxin units per kg body weight diluted in 10 mL of saline; Endotoxin from E. coli Type O113; List Biological Laboratories Inc., Campbell, California, USA) in the medial segment of the middle lobe and 10 ml saline (0.9%) in the medial segment of the lingula lobe as control. A second bronchoscopy was performed twenty-four hours later for collection of BAL (100 mL) from the saline- and LPS-challenged lung segments.

Processing and analysis of BAL samples

BAL samples were filtered (100 μm, BD Biosciences, Heidelberg, Germany) and centrifuged (300 g, 10 min, 4 °C). Cell pellet was resuspended in Dulbecco’s phosphate-buffered saline (DPBS), and total cell counts were determined by light microscopy using trypan blue staining and a Neubauer chamber. Absolute cell numbers were normalised to BAL volume recovery [106/mL]. Differential cell counts were determined by counting 800 cells microscopically on cytospins stained with Diff quick (RAL Diagnostics, Martillac, France). Cell-free BAL supernatants were stored at -80 °C until analysis. Total protein concentration in BAL supernatants were determined using the colorimetric Pierce™ 660 nm protein assay (Thermo Fisher Scientific, Rockford, Illinois, USA). Based on total protein content, samples were diluted with saline to a protein concentration of 50 µg/mL. Subsequently, 1,500 proteins were analysed by SomaLogic using the SomaScan assay platform (complete list of proteins is given in supplementary Table 1)15,16.

Statistical analysis

SomaScan data underwent quality control and transformation based on bioinformatic standards17. The lower limit of detection (LLOD) was defined as mean of eight blank samples plus three standard deviations (SD) of the blank. For the upper limit of detection (ULOD) a cutoff signal of > 100,000 was determined, as dilution linearity was not given for values above this cutoff (supplementary Fig. 1). Proteins with signal values below LLOD or above ULOD were excluded from further analysis. Signal values were then normalised to the original protein concentration of the respective sample. For analysis of expression differences in BAL following LPS compared to saline challenge, paired t-tests were performed, and p-values were adjusted (adj.) for multiple testing using the Benjamini-Hochberg procedure, which controls false discovery rate (FDR). Significance level was set to 0.05. Differential expressed proteins were defined by the criteria adj. p-value ≤ 0.05, and Log2 fold change (|Log2FC|) ≥ 0.58. Spearman correlations were applied to determine correlation of data obtained by the SomaLogic SomaScan assay compared to previously published data obtained by measurement of IL-6, IL-8, MPO and SP-D using the Meso Scale Discovery assay7 (data presented in Holz et al.7 originate from the same study, thereby allowing for correlation analysis).

Data visualization

Data were visualized using GraphPad Prism 9.0.1. Principal component analysis (PCA)-Plots were created using SomaLogic DataDelve™ Statistics18. Gene set enrichment analysis was performed with upregulated differentially expressed proteins using g: Profiler (Version 2019) to determine enriched biological processes (GOTERM_BP_DIRECT)19. Adj. p-values were calculated for multiple testing by g: Profiler using the Benjamini-Hochberg procedure, which controls FDR. Protein network was visualized by using the STRING database (https://string-db.org/)20. STRING protein-interactions were filtered based on a confidence score threshold of 0.9 (highest confidence). Hierarchical protein clustering of the top 100 differentially expressed proteins with the highest Log2FC after LPS compared to saline challenge was performed with ClustVis using default settings (https://biit.cs.ut.ee/clustvis/)21.

Results

As expected, LPS challenge induced recruitment of high numbers of inflammatory cells, predominantly neutrophils, which was not observed following saline challenge (supplementary Table 2). After challenge with LPS compared with saline, four proteins (ALDH2, FGFR3, ACP5, ADGRE2) were downregulated and 599 proteins were significantly upregulated in BAL (Table 1; Fig. 1, supplementary Table 3). As expected, upregulated proteins were predominantly pro-inflammatory mediators, including IL-6, IL-8, MPO, LTB4, MMP9, VWF, SELL or G-CSF.

Fig. 1
figure 1

Differentially expressed proteins in BAL following LPS challenge compared with saline control challenge. Differentially expressed proteins are highlighted in red for upregulated proteins (adj. p-value ≤ 0.05, log2(FC) ≥ 0.58) and in blue for downregulated proteins (adj. p-value ≤ 0.05, log2(FC) ≤ − 0.58). Abbreviations: ACP5 = Tartrate-resistant acid phosphatase type 5, ADGRE2 = Adhesion G proteincoupled receptor E2, Adj. = adjusted, ALDH2 = Aldehyde dehydrogenase 2 (mitochondrial), APOA1 = Apolipoprotein A-I, APOL1 = Apolipoprotein L1, BAL = bronchoalveolar lavage, CAMP = Cathelicidin antimicrobial peptide, CD5L = CD5 antigen-like, CXCL8 = C-X-C Motif Chemokine Ligand 8, FGFR3 = Fibroblast Growth Factor Receptor 3, IL = Interleukin, ITIH4 = Inter-alpha-trypsin inhibitor heavy chain H4, KNG1 = Kininogen 1, LILRA3 = Leukocyte immunoglobulin-like receptor subfamily A member 3, Log2(FC) = Log2 fold change, LPS = lipopolysaccharide, MMP8 = Matrix Metallopeptidase 8, PI3 = Elafin, SAA4 = Serum amyloid A-4 protein, VNN2 = Vanin 2.

Full size image

In accordance with the differential protein expression analysis, principal component analysis as well as a heatmap display showing the top 100 proteins with highest Log2FC clearly separated BAL samples following LPS challenge from those collected at pre-challenge baseline or after saline control challenge (supplementary Fig. 2, supplementary Fig. 3). As expected, the proteomic profile in BAL collected at pre-challenge baseline is similar to that in BAL collected after saline challenge, resulting in a single cluster of these samples (supplementary Fig. 2). Notably, participant 7 exhibited a more pronounced and participant 10 a weaker immune response to LPS compared to other volunteers (supplementary Fig. 3). As neither participant showed any abnormalities during the procedure (e.g., adverse events or signs of acute infection with elevated leukocyte counts), their data were not excluded from further analysis. Hierarchical clustering using the top 100 differentially expressed proteins did not reveal distinct protein subgroups (supplementary Fig. 3).

Significant upregulated proteins (adj. p-value ≤ 0.05, Log2FC ≥ 0.58) were used as input for gene set enrichment analysis using g: Profiler19. The three most significantly enriched biological processes were “immune system process”, “response to external stimulus” and “response to chemical” (Fig. 2, supplementary Table 4). Other significantly increased biological processes were related to, for example, apoptotic or migration processes (Fig. 2). Due to the small number of downregulated proteins in BAL in the LPS challenge model, no gene set enrichment analysis was performed for this group.

Fig. 2
figure 2

Enriched biological processes that are related to upregulated proteins (adj. p-value ≤ 0.05, log2(FC) ≥ 0.58) in BAL following LPS challenge compared with saline challenge. The 35 most significant results are depicted (range of adj. pvalues: 3.4e-28 to 6.1e-53). Abbreviations: Adj. = adjusted, BAL = bronchoalveolar lavage, Log2(FC) = Log2 fold change, LPS = lipopolysaccharide

Full size image

In addition, a STRING protein-network analysis was performed to investigate the interactions among the significantly upregulated proteins identified in the LPS challenge model. The resulting network revealed different protein clusters of e.g., chemokines, matrix metalloproteinases or proteins that are implicated in immunoproteasome assembly (Fig. 3). The five key hub proteins within the network were IGHV3-43D (22 nodes), CD44 (16 nodes), EGFR (16 nodes), SYK (16 nodes) and FN1 (15 nodes), which all exhibited high connectivity to other proteins.

Fig. 3
figure 3

Network analysis, depicting protein interaction clusters of significantly upregulated proteins (adj. p-value ≤ 0.05, log2(FC) ≥ 0.58) in BAL following LPS challenge compared with saline challenge. Proteins that are related to the three most significant enriched biological processes „immune system process“ (red), „response to external stimulus“ (blue) and „response to chemical“ (green) (figure 2) were highlighted in the respective colour. Top five key hub genes were highlighted with black rectangles. Abbreviations: Adj. = adjusted, BAL = bronchoalveolar lavage, Log2(FC) = Log2 fold change, LPS = lipopolysaccharide. Abbreviations of proteins are explained in supplementary table 1

Full size image

Chen et al. has published a systematic review summarizing current blood biomarkers of exacerbations in COPD12. In Table 2 of this review, 24 different biomarkers of studies with receiver operating characteristics (ROC) area-under-the curve (AUC) statistics are listed that were found to be elevated in AECOPD. We have checked whether these proteins were also increased in BAL following LPS challenge within our data set. Three (procalcitonin, RBP4, serum Amyloid-A 4) out of the 24 proteins were not measured using the SomaLogic SomaScan assay, and four biomarkers (BNP, CCL18, IFN-γ, IL-5) showed values below LLOD and were therefore not evaluable. From the remaining 17 proteins, 16 were significantly upregulated in BAL following LPS challenge compared to saline control (Table 2). The AECOPD biomarker sTREM-1 did not show changes in BAL collected after LPS challenge.

The concentrations of IL-6, IL-8 and MPO measured by the SomaLogic SomaScan assay highly correlated with previously published cytokine levels of the same study obtained by using the Meso Scale Discovery assay (supplementary Fig. 4a-c)7. Despite this high correlation, the SomaScan assay showed an overestimation of IL-6, IL-8 and MPO at lower concentration levels (supplementary Fig. 5a-c). No correlation between these two systems was observed for SP-D (supplementary Figs. 4 and 5d).

Table 1 Top 30 differentially expressed proteins (adj. p-value ≤ 0.05, |Log2FC| ≥ 0.58) in BAL following LPS challenge compared with saline challenge.
Full size table
Table 2 Expression of AECOPD biomarkers, according to table 2 in Chen et al.12, in BAL following LPS challenge compared with saline control challenge. Abbreviations: Adj. = adjusted, ID = identification, LLOD = lower limit of detection, Log2FC = Log2 fold change.
Full size table

Discussion

In this study, we performed a comprehensive proteomic analysis of BAL fluid following segmental LPS challenge in healthy participants, revealing significant insights into the inflammatory response induced by endotoxin exposure.

As expected, principal component analysis and hierarchical clustering allowed differentiation of BAL following LPS challenge from pre-challenge and post-saline challenge BAL, reinforcing the specificity and effectiveness of the LPS model in inducing localized inflammation in the human lung. Baseline BAL and BAL collected after saline control challenge clustered into a single group, which highlights the fact that the bronchoscopic procedure itself did not elicit an inflammatory response within the challenged lung segments. As expected based on the highly increased cell numbers and related changes within the transcriptomic profile in response to LPS7,10, endotoxin challenge resulted in upregulation of 40% of the analyzed proteins in BAL. Since primarily immune response-relevant proteins were selected for analysis, this high percentage of upregulated proteins was expected. Although two of the ten participants (S7 and S10) showed differences in the intensity of their immune response to LPS challenge, the overall protein expression patterns were consistent across all patients. As a result, hierarchical cluster analysis of the top 100 differentially expressed proteins did not identify distinct responder groups. Among the upregulated proteins were e.g., the cytokines IL-6 and IL-8, which play a crucial role in the immune response. Both cytokines were reported to be potential serum biomarkers for describing the severity and course of COPD exacerbations, and for predicting which patients are susceptible for future exacerbations22,23,24. Other significant upregulated proteins within the LPS challenge model were MPO and LTB4. Levels of both proteins were found to be increased in sputum of patients with bacterial versus non-bacterial AECOPD25 and therefore are potential biomarkers to guide antibiotic treatment during exacerbations. However, several proteins were below LLOD, which would have been expected to be increased after challenge with LPS (e.g., IL-2, IL-4, IL-13, IL-23, TSLP, IFN- γ, IL-1β, MCP-2, TNF-α, MIP-1α, MMP-3, MMP-8, WNT16, WNT5B, WNT7a). This likely reflects either an absence of these proteins in the samples or concentrations below the assay’s detection limit. It should be noted that the analyzed BAL samples were collected 24 h post LPS challenge, a timepoint where the kinetic peak of cytokine secretion is known to be almost back to baseline for many proinflammatory proteins26.

As there were too many genes to look individually at, gene set enrichment analysis (GSEA) analysis was performed. GSEA with the significantly upregulated proteins (adj. p-value ≤ 0.05, Log2FC ≥ 0.58) following LPS challenge revealed general pro-inflammatory processes, including the three top hits “immune system process”, “response to external stimulus”, and “response to chemical”. Additional upregulated processes were related to e.g., apoptosis and cell migration. All identified processes are crucial for inflammatory immune responses and align with the known pathophysiology of endotoxin-induced inflammation, further validating the model. The STRING protein-network analysis revealed key clusters, thereby providing a deeper insight into the complex interactions among the upregulated proteins in BAL following LPS challenge compared with saline. As expected, different chemokines (e.g., CCL2, CXCL8, CXCL10, CXCL11) were upregulated in response to LPS, which are known to play a major role in the recruitment of inflammatory cells, particularly neutrophil granulocytes, monocytes and T helper 1 cells27. Disbalance of the CCL2-CCR2, CXCL8-CXCR1/2, and CXCL9/10/11-CXCR3 axes play a role in the pathophysiology of COPD, both at a stable state and during exacerbations28. Another cluster consisted of different metalloproteinases such as MMP1, MMP2, MMP7, MMP9 and MMP10, which are critical for tissue homeostasis and development29. Disbalance of these proteolytic enzymes may contribute to airway remodeling and emphysema, which are major pathological features of COPD30. In addition to metalloproteinases, several other proteases such as neutrophil elastase (NE), cysteine proteases (CASP), or proteinase 3 (PR3) play key roles in COPD exacerbations by e.g., contributing to alveolar tissue destruction or regulating apoptosis31,32,33. PR3 was not included in the analysis panel. Signals for CASP8 and CASP10 were below LLOD, while CASP3 was significantly upregulated in BAL after LPS compared to saline challenge. As expected, values for neutrophil elastase (NE) were almost seven times higher in BAL after LPS compared to saline challenge (Saline: 194613.49 ± 81833.29, LPS: 1311714.93 ± 619835.68). Unfortunately, all signals in both groups were high above the upper limit of quantification (set to ≥ 100,000 as described in the method section) and therefore not evaluable. Furthermore, proteins that are related to immunoproteasome assembly (e.g., PSMB9, PSMB10, PSME1, PSME234,35 build another cluster within the STRING protein-network. As the immunoproteasome improves major histocompatibility complex class I (MHC I)–mediated antigen presentation to defend against infections, impairment of the immunoproteasome function by e.g., cigarette smoke could lead to prolonged infections and exacerbations in COPD and idiopathic pulmonary fibrosis (IPF)36. Moreover, the key hub network-proteins IGHV3-43D (variations in the IGH locus impact the antibody repertoire37, CD44 (loss disrupts lung lipid surfactant homeostasis and exacerbates lung inflammation38, EGFR (major role in regulation of cell proliferation39, SYK (critical for signal transduction and antiviral responses40, and FN1 (involved in cell adhesion/motility41 are also known to have central roles in modulation of immune responses. In summary, GSEA- and protein-network analysis provided a deeper insight into ongoing mechanisms and pathways during the LPS-induced inflammatory response. As shown by the elaborated examples of chemokines, metalloproteinases, and immunoproteasome assembly, the human segmental LPS challenge model reflects some aspects of COPD and AECOPD. Targeting these pathways could be a possible intervention strategy for the treatment of COPD and AECOPD28,42. Furthermore, segmental LPS challenge could be a suitable model for respective first-in-human drug efficacy studies. The LPS challenge model has already been used in several clinical trials to investigate the anti-inflammatory effects of new investigational drugs. For example, treatment with the phosphodiesterase 4 inhibitor roflumilast43, the IL-1 receptor antagonist anakinra44 or the CXCR2 antagonist AZD830945 have all attenuated pulmonary inflammation after segmental or inhalative LPS challenge due to decreased cell influx of in particular neutrophils, which are as already mentioned the predominant cell type in the airways of patients with COPD and AECOPD.

Chen et al. has published a list of biomarkers with ROC analysis that were found to be elevated in blood of patients with AECOPD12. We examined whether these proteins were also elevated in BAL following LPS challenge within our dataset. Unfortunately, there is no collection of relevant biomarkers for AECOPD in BAL so far, which would have been more suitable to identify proteomic similarities between AECOPD and the LPS challenge model. However, comparison of blood biomarkers with BAL collected within the LPS challenge model may help to identify matrix-independent biomarkers. 16 out of 17 evaluable biomarkers were elevated in both blood of patients with AECOPD as well as in BAL following LPS challenge, while only one AECOPD biomarker (sTREM-1) did not show differences within the LPS challenge model. In the literature, the expression of sTREM-1 in BAL as a biomarker for infectious pneumonia is controversially discussed, with studies reporting inconsistent results regarding its diagnostic utility46,47. In our study, levels of sTREM-1 did not differ significantly between BAL samples from LPS- and saline-challenged lung segments, which supports the assumption that sTREM-1 is unlikely to serve as a reliable BAL marker for distinguishing between bacterial and non-bacterial causes of inflammation. The high number of matching proteins further underlines the similarities between AECOPD and the segmental LPS challenge model. Therefore, the LPS challenge model could be suitable for studying inflammatory processes that are related to AECOPD as well as for first-in-human drug efficacy studies for the treatment of acute exacerbations. For each of the biomarkers elaborated by Chen et al., there is a large body of literature related to AECOPD. Most of the listed AECOPD biomarkers were chemokines (CXCL8/10/11 and CCL11/17/18/23), which, as already mentioned above, play a key role in the pathophysiology of COPD and exacerbations thereof12,28. In line, CXCL8 (synonym: IL-8), the most potent human neutrophil-attracting chemokine, was found to be elevated in sputum and BAL of patients with AECOPD48,49. Another AECOPD biomarker is CRP. While healthy individuals and patients with stable COPD show CRP concentrations below 10 mg/mL, higher levels are associated with an increased hospitalization and mortality risk50. Furthermore, serum CRP levels are elevated during viral and bacterial AECOPD, usually higher in bacterial triggered exacerbations50, which may could be useful in the guidance of antibiotic therapy51,52. High levels of the AECOPD biomarker fibrinogen was shown to predict the rate and risk of exacerbations in COPD, but not mortality or treatment response in a clinical trial53. However, both CRP and fibrinogen are acute phase reactants, which are not specific to an etiologic or biologic pathway, emphasizing the need of more specific biomarkers for the prognosis and diagnosis of AECOPD.

This study carries limitations. Although we found that LPS challenge reflects some aspects of AECOPD, the model induces transient airway inflammation and therefore does only partially reflect chronic inflammatory conditions as they are found in patients with stable COPD. For this reason, it is essential to perform comprehensive analysis of the transcriptomic and proteomic profile of the LPS challenge model in order to elaborate which features of patients with respiratory diseases are reflected in this model at the cellular and molecular level. Furthermore, we have analysed the proteomic profile of healthy non-smoking participants in this study. Tobacco smoke, which contains LPS54,55, is the major cause of COPD1. Therefore, smokers with chronic exposure to cigarette smoke may have been closer to the COPD phenotype than non-smoking participants. However, the inflammatory response to LPS challenge is similar or only slightly increased (e.g., IL-1β) in smoking compared to non-smoking participants56. Moreover, it should be mentioned that we have used the SomaScan assay platform in this study. To avoid interference with the assay system due to excessively high protein concentrations, we followed the manufacturer’s advice to standardize the total protein concentrations of all samples to 50 µg/mL. Total protein concentrations in BAL were 72.59 ± 20.56 µg/mL at baseline, 70.63 ± 17.60 µg/mL after saline, and 273.55 ± 114.87 µg/mL after LPS challenge. However, it cannot be ruled out that the dilution of particular BAL samples collected after LPS challenge led to signals below LLOD for some AECOPD relevant proteins. While concentrations of IL-6, IL-8 and MPO measured by the SomaLogic SomaScan assay highly correlated with previously published data obtained by using the Meso Scale Discovery (MSD) assay, levels of SP-D did not correlate between these two assay systems. Since SP-D is susceptible for conformational changes, in particular under inflammatory conditions57, and it is not known which exact epitope of SP-D is recognized by the respective antibody (MSD) or aptamer (SomaScan), it remains unclear whether the lack of correlation could be due to the molecule itself or due to method-specific reasons. In the SomaLogic SomaScan assay up to 11,000 proteins can be measured simultaneously within a samples58. This enables a comprehensive multi-proteomic analysis, which cannot be realized by conventional enzyme-linked immunosorbent assays, thereby opening new possibilities for future research and drug target screening.

Conclusions

This study provides a very comprehensive analysis of the proteomic profile in BAL collected following LPS challenge. The identification of differentially expressed proteins, and the GSEA- as well as protein-network analysis allowed a deeper insight about ongoing mechanisms during LPS-induced inflammation. In summary, our data expanded the knowledge about the human segmental LPS challenge model and its similarities with infection-triggered respiratory inflammations such as AECOPD in particular.