Systems medicine advances in interstitial lung disease

Genomics and transcriptomics

Advances in genomic techniques have allowed high-throughput analysis and discovery of gene deregulation in IPF [6]. In particular, genetic studies have contributed to a better understanding of IPF, e.g. the expression of MUC5B (mucin 5B) and TOLLIP (Toll-interacting protein) [7–9]. Polymorphism in the promoter region of MUC5B (rs35705950) is associated with a higher likelihood of IPF development, although patients carrying this allele present a milder disease course and improved survival [7]. MUC5B was found expressed in areas of microscopic honeycombing and honeycomb cysts [8]. To date, the precise role of MUC5B in the pathophysiology of IPF is unclear.

Furthermore, variants in TOLLIP have also been linked to susceptibility and treatment responses in IPF. Single nucleotide polymorphisms within TOLLIP (rs5743890/rs3750920) are associated with increased mortality risk (rs5743890) and better response to N-acetylcysteine treatment (rs3750920) in IPF patients [9, 10]. Similarly, DSP (desmoplakin) variance (rs2076295) is associated with increased risk of IPF. MUC5B and DSP expression in the lung, especially in the airway epithelium, indicates the contribution of the aberrant epithelium in IPF [11].

Most IPF cases are sporadic. However, genetic variations also include an autosomal dominance pattern of inheritance, leading to familial pulmonary fibrosis. Several studies showed a strong relationship between familial IPF and telomerase mutations and their shortening. Recently, a prospective study performed genetic evaluations of IPF patients and affected relatives, and confirmed a strong relationship between familial IPF and telomerase mutations and shortening [12, 13]. Several studies have reported that familial IPF is associated with variances of the genes TERT (telomerase reverse transcriptase), TERC (telomerase RNA component), DKC1 (dyskerin pseudouridine synthase 1), TINF2 (TERF1 interacting nuclear factor 2), RTEL1 (regulator of telomere elongation helicase 1) and PARN (poly(A)-specific RNase) [12, 13]. Familial IPF also presents mutations within surfactant protein-encoding genes SFTPA2 (surfactant protein A2) and SFTPC (surfactant protein C), and ABCA3 (ATP-binding cassette subfamily A member 3) [12, 14].

MicroRNAs (miRNAs) are small noncoding RNAs involved in the regulation of gene and protein expression, thus altering cellular phenotypes. IPF patients display downregulation of miRNA levels in members of the let-7, mir-29 and mir-30 families, and upregulation in members of the mir-155 and mir-21 families, which modulate biological pathways and modify the IPF phenotype [15]. Altered expression of let-7 family members leads to changes in epithelial–mesenchymal transition in lung epithelial cells and inhibition of mir-21 modulates fibrosis [15]. mir-29 is also being explored as a potential target for IPF therapy as it decreases collagen expression in fibrotic lungs [16].

Moreover, IPF shares features with other chronic lung diseases, such as chronic obstructive pulmonary disease (COPD), where the molecular mechanism of lung injury leads to airway fibrosis. Kusko et al. [17] revealed a shared mRNA–miRNA transcriptional network between IPF and COPD. Specifically, upregulation of mir-96 in both diseases may control part of the shared disease gene expression network. Kusko et al. [17] describe overexpression of mir-96 as a key novel regulator of p53 expression in both epithelial cells and fibroblasts, which modulates the expression of genes related to the “emphysema–IPF” gene network [17]. Currently, miRNA therapy is being explored as a therapeutic option in diverse fibrotic conditions.

Cell-based RNA genomics can also provide an insight into disease features and targetable molecules. In line with the study by Kusko et al. [17], single-cell RNA sequencing analysis of IPF and control lungs identified the cross-talk of four distinct lung epithelial cell subtypes (alveolar type 2, goblet, basal and indeterminate) [18]. Xu et al. [18] showed that cells isolated from IPF patients express genes associated with activation of canonical transforming growth factor (TGF)-β, HIPPO/YAP, PI3K/AKT, p53 and WNT signalling cascades, which are activated in an integrated network [18]. Myofibroblast differentiation and massive ECM deposition are both ideal targetable phenomena in fibrotic diseases [19]. Parker et al. [19] used fibroblasts to explore at the RNA level how proteins present in the ECM of acellular lungs can provoke a feedback loop to normal living cells exposed to an aberrant environment. The results suggest that characterisation of lung proteins, specifically the lung fibrotic ECM, will help to not only determine its composition, but also to define targetable molecules for advanced stages of fibrosis [19].

Advances in genomic technologies show an exciting time ahead, where analysis of the genetic makeup of IPF patients will dictate therapeutic approaches and predict disease outcome. Hence, implementing routine clinical genotyping and biomarker testing is essential for future patient stratification and personalised treatment.

Proteomics

As a substantial part of systems medicine, proteomics covers information on protein abundance, variation and modification, including their partners and networks, in order to explain cellular processes. Given the impact of aberrant biological processes in IPF, proteomics analysis provides global protein quantification as a critical tool to identify disease-driving molecules and potential targets. The most universal and powerful method for global protein measurement is mass spectrometry, which uses liquid chromatography together with high-resolution tandem mass spectrometry to identify and quantify peptides at a large scale. These techniques, together with bioinformatic tools, contribute to a better understanding of protein biochemistry, nature and interaction.

It is possible that many of the proteins dysregulated in IPF may not cross the lung endothelial barrier or become diluted out by more abundant constituents, thus being undetectable in plasma [20]. Based on this concept, initial studies with proteomics in IPF have been primarily performed in bronchoalveolar lavage fluid (BALF). Osteopontin has been a more strongly validated marker increased in the BALF of IPF patients. Other targets have also been associated with disease, e.g. the CC chemokine ligand CCL24, surfactant protein A2, and transcriptional factors NF-κB, peroxisome proliferator-activated receptor-γ and c-Myc [20–22]. Recently, it was reported from proteomics analysis of fibroblast surface fractions that platelet-derived growth factor receptor (PDGFR)-α expression was altered by pro-fibrotic TGF-β in lung fibroblasts from IPF patients [23]. The authors suggest a potential cross-talk between two critical signalling pathways in fibrosis, i.e. TGF-β/PDGFR-α, which affects myofibroblast differentiation in the context of IPF [23], currently insufficiently targeted by approved therapies.

In an iTRAQ (isobaric tag for relative and absolute quantitation)-based proteomics study, Niu et al. [24] identified a set of proteins in the serum of IPF patients. The identified targets were related to the protein activation cascade, regulation of response to wounding and extracellular components (see table 1). Unfortunately, no correlation with clinical parameters was performed. Recently, using a 1129-analyte SOMAmer (slow off-rate modified aptamer) array in IPF plasma, a six-analyte index predicted better progression-free survival in IPF. Specifically, high levels of ficolin-2, cathepsin-S, legumain and soluble vascular endothelial growth factor receptor-2, and low levels of inducible T-cell costimulator (ICOS) or trypsin-3 were associated with IPF progression [25]. The use of this index should be further validated in independent cohorts for general applicability.

Metabolomics

Metabolites are the second product of metabolic reactions catalysed by enzymes that naturally occur in the cells, complementing gene and protein expression [26]. Metabolic changes of the lung are involved in IPF pathogenesis, as reported by Kang et al. [27]. IPF presented 25 metabolite signatures, which indicated alteration in metabolic pathways of ATP degradation, glycolysis, glutathione biosynthesis and ornithine aminotransferase pathways [27]. Interestingly, ornithine has a negative correlation with forced vital capacity (FVC), suggesting the potential role of ornithine in IPF pathophysiology [27]. Moreover, lactic acid levels and lactate dehydrogenase (LDH)-5 levels were elevated in IPF lungs when compared with controls [28]. Increased levels of LDH-5 activate the TGF-β pathway, inducing myofibroblast differentiation [28].

Microbiome

The human microbiota consists of 10–100 trillion symbiotic microbial cells in a single individual; however, little is known about the contribution of the microbiome to health and disease. Recently, it was reported that IPF patients have higher bacterial load in BALF when compared with controls and the increased bacterial load identifies patients with more rapidly progressive disease [29]. Interestingly, Molyneaux et al. [29] found bacterial burden to be independent of MUC5B genotype, suggesting a direct connection between host immunity and bacterial load [29]. In addition, it has been described in the COMET cohort that microbial signatures of Staphylococcus and Streptococcus help predict IPF progression [30]. Disease progression was significantly associated with increased abundance of two specific strains of Staphylococcus and Streptococcus. The study determined two operational taxonomic units (OTUs) associated with disease progression: Staphylococcus OTU 1348 and Streptococcus OTU 1345 [30]. This altered microbiome persists over time, which may implicate bacterial communities localised in the lower airways as persistent stimuli for repetitive alveolar injury in IPF [31]. IPF patients from the COMET cohort showed interactions between the host microbiome and progression-free survival [32]. Downregulation of the immune response was associated with changes in the abundance of Staphylococcus OTU 1348, which correlated with alterations in circulating leukocyte phenotypes, expression of several Toll-like receptors and fibroblast responsiveness [32].

Peripheral blood phenotyping

In the past, the role of the immune system in IPF has been controversial. However, increasing contributions indicate multiple alterations in the immune compartment of IPF patients, raising interest in the field. The peripheral blood compartment provides an easily accessible liquid biopsy, highly practical to study molecules and cell types as potential biomarkers. IPF patients have an increase in circulating biomarkers, e.g. CCL18, metalloproteinase (MMP)-7 and soluble intracellular adhesion molecule-1. To date, MMP-7 is the most validated biomarker to adopt clinically for diagnosis and prognostic evaluation of IPF [33]. Recently, several studies have validated the immunosuppressive environment present in the peripheral blood of IPF patients, showing that the gene expression of CD28, ICOS, lymphocyte-specific protein tyrosine kinase and interleukin-2 inducible T-cell kinase can predict outcome in IPF [34, 35].

In PROFILE, the largest IPF biomarker cohort studied globally, Jenkins et al. [36] showed changes in serum concentrations of proteolytically cleaved protein fragments/neopitopes. Levels of fragmented proteins generated by MMP activity and collagen synthesis in the serum of IPF patients were associated with IPF progression, as well as survival rate [36]. Immune cell type dysfunction has also been implicated in IPF. Abnormal B-cells and B-cell stimulator factor are often present in IPF patients, and are highly associated with disease manifestations and patient outcome [37]. Furthermore, myeloid-derived suppressor cells were described as a potent biomarker for IPF, where increased numbers of myeloid-derived suppressor cells measured in the blood of IPF patients correlated with lung function in cross-sectional and longitudinal analysis [38].