Lung microenvironment proteomics identifies immune protein expression in interstitial pneumonia with autoimmune features

Extract

Several of the interstitial lung diseases (ILDs) present a conundrum for clinicians. No group of these conditions is more indicative of this challenge than interstitial pneumonia with autoimmune features (IPAF) [1]. The proposed research classification of IPAF identifies those patients who have serologic, clinical, or radiographic/histologic features of autoimmunity but fail to meet clinical criteria for an autoimmune disease [2]. Recent American College of Rheumatology (ACR) and American College of Chest Physicians (CHEST) ILD guidelines for the treatment of autoimmune ILD recommend first line therapy with immunomodulatory medications for all subsets of autoimmune ILD, though IPAF was notably excluded [3].

Short abstract

This study characterised differential protein expression using label-free proteomics of the lung microenvironment using bronchoalveolar lavage fluid and identified increased immune protein expression in the IPAF lung compared to IPF https://bit.ly/4edhp3B

To the Editor:

Several of the interstitial lung diseases (ILDs) present a conundrum for clinicians. No group of these conditions is more indicative of this challenge than interstitial pneumonia with autoimmune features (IPAF) [1]. The proposed research classification of IPAF identifies those patients who have serologic, clinical, or radiographic/histologic features of autoimmunity but fail to meet clinical criteria for an autoimmune disease [2]. Recent American College of Rheumatology (ACR) and American College of Chest Physicians (CHEST) ILD guidelines for the treatment of autoimmune ILD recommend first line therapy with immunomodulatory medications for all subsets of autoimmune ILD, though IPAF was notably excluded [3]. Meanwhile, the use of immunomodulatory medications in patients with idiopathic pulmonary fibrosis (IPF) was associated with worsened mortality in a seminal randomised trial [4]. Patients with IPAF do not clearly fit within either of these two poles in ILD management.

In the absence of validated markers of treatment prediction [5], clinicians treating patients with IPAF are left to guess if immunomodulatory therapy is likely to be effective or harmful. In the present study, we set out to leverage proteomics in bronchoalveolar lavage fluid (BALF) to compare and characterise differential protein expression across the IPF and IPAF lung microenvironment.

Patients with IPF and IPAF were identified from a biorepository of subjects from a prospective research study at the Department of Respiratory Medicine and Allergology at the Sapporo Medical University Hospital, Japan [6] via two pulmonologists who also assigned subjects as IPF or IPAF based on international criteria [2, 7]. Prior to receiving treatment during their diagnostic phase, patients (n=41) underwent standardised BALF protocol with 50 mL of 0.9% sterile saline being instilled into the right middle lobe or lingula via a wedged bronchoscope. Lavage fluid was then collected via gentle suction and repeated for a total of three lavages (150 mL of 0.9% in total). The collected BALF supernatant was combined, centrifuged to remove cells, and frozen at −80°C for future study. The Institutional Review Board of Sapporo Medical University Hospital approved this study (No. 342–201, approved on 9 February 2023).

As previously published [8], 100 µL BALF supernatant samples were reduced with TCEP (tris(2-carboxyethyl)phosphine), alkylated with iodoacetamide, and precipitated with acetonitrile. Protein pellets were digested with trypsin, and peptides analysed by liquid chromatography–mass spectrometry (LC-MS)/MS on an Orbitrap Ascend mass spectrometer (ThermoFisher Scientific, Waltham, MS, USA). Raw files were searched within Proteome Discoverer 3.0 against the UniProt human protein database. Peptide groups and protein abundances were normalised by the total abundance values for each channel across all files, equalising the total abundance between different BALF samples. The raw mass spectrometry files and an Excel results table of proteins identified and discussed below have been uploaded to MassIVE (MSV000097958).

Non-human and processing contaminant proteins were removed. Analysis included only proteins detected in at least 80% of the 41 samples. For proteins retained after filtering (≥80% detection), remaining missing values were imputed as one fifth of the global limit of detection, defined as the smallest non-zero detected value across the dataset, prior to log2 transformation and normalisation. Protein abundances were median-normalised, log2-transformed, and auto-scaled before performing principal component analysis (PCA), differential expression analysis with adjustment for clinical covariates (age, sex, smoking, baseline forced vital capacity percentage) and multiple testing, and pathway analysis. Significant differential expression proteins were defined as those with |log2 fold change| ≥1 and adjusted q≤0.1. Overrepresentation analysis was performed in QIAGEN IPA to identify enriched pathways. Pathways were considered significant at p≤0.05, with activation/inhibition determined by z-scores (≥2 activation, ≤ −2 inhibition). Baseline comparisons of clinical and demographics were carried out via t-test. Analyses were performed in R v4.3.0.

Our cohort (n=41) included IPF (n=27) and IPAF (n=14) subjects with most patients meeting IPAF criteria for serologic (92.9%) and morphologic (71.4%) domains. The IPF group had significantly more males (23 of 27 versus 4 of 14) and greater smoking history (43.9±32.4 versus 18.2±32.6 pack-years) than IPAF (both p<0.01). Age and baseline pulmonary function measures (forced vital capacity and diffusion capacity) were not significantly different between groups. No subjects were on treatment at the time of BALF collection, although a higher proportion of IPAF patients were subsequently treated with steroids (57% versus 25.9%) and had a shorter mean time to steroid use (108 days) than subjects with IPF (543 days).

LC-MS/MS analysis of BALF samples identified 965 proteins detected at least once across 41 samples. After removing contaminants and proteins with more than 20% missing values, 649 proteins remained for analysis.

PCA revealed separation between disease states, with most IPAF samples distinct from IPF along PC1 (26.7% variance) and PC2 (8.5% variance), and clustering patterns were driven primarily by disease state rather than sex, as samples within each disease group clustered together irrespective of sex (figure 1a). Differential expression analysis with adjustment for clinical covariates and multiple testing identified 26 significantly altered proteins (|log2 fold change| ≥1, adjusted q<0.1) between conditions, with 17 upregulated in IPAF and 9 in IPF (figure 1a). The upregulated proteins in IPAF included immune signalling proteins such as components of the complement system (C1QA, C1R, SERPINF2, SERPINA4) and apolipoproteins (APOE, APOA1, APOA4), while those upregulated in IPF had established roles in tissue remodelling and injury repair (GRB2, SLC44A4, ARPC4-TTLL3, FLNA).

FIGURE 1.

a) Principal component analysis biplot of bronchoalveolar lavage fluid (BALF) proteomics in interstitial pneumonia with autoimmune features (IPAF) and idiopathic pulmonary fibrosis (IPF) patients. Dots are samples (n=41), with red indicating IPAF (n=14) and blue IPF (n=27). Lines are loadings of significantly differentially expressed proteins in IPAF compared to IPF, after adjusting for covariates (age, sex, smoking, baseline forced vital capacity percentage) and multiple testing (|log2 fold change| ≥1, adjusted q≤0.1); red lines show higher expression in IPAF, and blue higher expression in IPF; selected proteins have been labelled for readability. The x- and y-axes show variance described by PC1 and PC2, respectively. Clustering patterns were primarily driven by disease state rather than sex, with samples grouping by diagnosis irrespective of sex. F: female; M: male. b) Top 10 most significant pathways predicted to be enriched based on the protein differential expression between IPAF and IPF. Pathways (bars) are ranked by p-value (x-axis). Colours represent z-scores. Red: activated in IPAF; blue: inhibited in IPAF; grey: inconclusive direction. Directions with |z-score| ≥2 are considered significant.

The top three pathways predicted to be altered between disease states with the strongest significance included: Complement Cascade (p=2.51×10⁻²⁶), Acute Phase Response Signalling (p=1.26×10⁻¹⁵), and LXR/RXR Activation (p=3.16×10⁻¹⁴) (figure 1b). The Complement Cascade pathway showed hyperactivation in IPAF (z-score=4.243). The significant pathways collectively suggested that IPAF displayed a stronger immune-inflammatory signature than IPF. Pathway analysis in IPA also predicted that methylprednisolone was likely to have an inhibitory effect on the protein expression in IPAF (p=3.19×10⁻⁷, z-score= −2.56) by targeting the highly expressed immune signalling proteins CD22, IGHM, IGLV3-25, IGLC7, IGKV1-17, IGLV2-11, IGLV7-46, and SERPINA6.

In this study, we characterised differential protein expression in the lung microenvironment in patients with IPAF compared to IPF. Notably, despite sex imbalances between groups, PCA clustering appeared to be driven primarily by disease state rather than sex, supporting the biological signal observed. We identified increased immune protein expression in the IPAF lung compared to IPF. This study used an unbiased proteomics approach without predefined targets, but notably, a more targeted analysis in this cohort previously observed increased cytokine expression in IPAF BALF (CXCL9, CXCL10, and CXCL11) compared to IPF [6]. Others have previously identified increased levels of Krebs von den Lungen-6 (KL-6), surfactant proteins A and D, and chemokine ligand 2 circulating in IPAF patients when compared to non-IPF fibrotic lung diseases [6, 9]. Our analysis expands on these findings by leveraging mass spectrometry-based proteomics to globally characterise differential protein expression and pathways across the IPAF and IPF lung microenvironment.

Several limitations warrant consideration. As a single-site cohort from Japan, generalisability may be limited due to ethnic and environmental factors and BALF collection methods. The sample size is small, and baseline differences in ILD pattern, gender and smoking could have influenced the results. Although sex was included as a covariate in regression models, the marked sex imbalance between groups and the modest sample size limits our ability to fully exclude residual confounding. Larger, prospective studies are required as next steps in addition to similar comparisons with autoimmune ILD cohorts. Because proteomic profiling was performed on cell-free BALF supernatant and differential cell counts were not available, we cannot distinguish whether observed differences reflect inflammatory cell recruitment or altered protein regulation; however, either mechanism supports a more immune-activated lung microenvironment in IPAF compared with IPF.

Patients with IPAF occupy a unique position in the spectrum of ILDs. Since the inception of the IPAF classification, there has been ongoing debate regarding both its validity as a distinct entity and its proper categorisation: whether these patients should be managed as IPF or as autoimmune ILD. The proteomic data presented here, while not definitive, demonstrate enhanced immune and complement pathway activation in the IPAF lung microenvironment compared with IPF, potentially supporting a more immune-inflammatory lung microenvironment in IPAF. As demonstrated in our previous work, BALF molecular profiling across aetiologies of idiopathic ILD can identify important autoimmune lung phenotypes and outperform radiographic patterns or prognostic models in predicting survival [8]. In total, these current data highlight the need for molecular markers that consider the underlying inflammatory and immune signature of each patient rather than relying solely on clinical classification.