Latent inter-organ mechanism of idiopathic pulmonary fibrosis unveiled by a generative computational approach

Abstract

Idiopathic pulmonary fibrosis (IPF) is a chronic and progressive disease characterized by complex lung pathogenesis affecting approximately three million people worldwide. While the molecular and cellular details of the IPF mechanism is emerging, our current understanding is centered around the lung itself. On the other hand, many human diseases are the products of complex multi-organ interactions. Hence, we postulate that a dysfunctional crosstalk of the lung with other organs plays a causative role in the onset, progression and/or complications of IPF. In this study, we employed a generative computational approach to identify such inter-organ mechanism of IPF. This approach found unexpected molecular relatedness of IPF to neoplasm, diabetes, Alzheimer's disease, obesity, atherosclerosis, and arteriosclerosis. Furthermore, as a potential mechanism underlying this relatedness, we uncovered a putative molecular crosstalk system across the lung and the liver. In this inter-organ system, a secreted protein, kininogen 1, from hepatocytes in the liver interacts with its receptor, bradykinin receptor B1 in the lung. This ligand-receptor interaction across the liver and the lung leads to the activation of calmodulin pathways in the lung, leading to the activation of interleukin 6 and phosphoenolpyruvate carboxykinase 1 pathway across these organs. Importantly, we retrospectively identified several pre-clinical and clinical evidence supporting this inter-organ mechanism of IPF. In conclusion, such feedforward and feedback loop system across the lung and the liver provides a unique opportunity for the development of the treatment and/or diagnosis of IPF. Furthermore, the result illustrates a generative computational framework for machine-mediated synthesis of mechanisms that facilitates and complements the traditional experimental approaches in biomedical sciences.

Figure 1

General overview of the multi-modal generative topic modeling approach for IPF. The previously developed method is adapted to IPF.

Figure 2

The organ and cell-enrichment analyses of the latent IPF-features. (A) The organ enrichment. (B) The cell-type enrichment. The enrichment level of the 83 IPF-features in each organ and each cell-type is shown as bar-graph of −log10(q-values) in the descending order. The q-value (qvalue) = 0.05 (the threshold for the statistical significance) is indicated as a red line in each graph. The raw data are available as Supplementary Table S2.

Figure 3

The latent diseases to which IPF is molecularly related. The frequency of the appearance of the 83 IPF-features in each disease is indicated as “count”. Shown are the diseases of which counts are above 20 in the descending order. The long disease names are cut short and indicated as “...” at their ends. The raw data are available as Supplementary Table S3.

Figure 4

General overview of the computational framework to generate an inter-organ mechanism of IPF. See the “Methods” section for the detailed step-by-step description. The 83 latent IPF-features and 112 lung DEgenes (IPF vs. non-IPF) are found in Supplementary Tables S1 and S4, respectively.

Figure 5

The hepatic expression of the ligands and receptors for the IPF pulmonary receptors and ligands. The level of each ligand and receptor in each cell-type in the liver is shown as dot. The size and the heat-intensity represent the ratio of cells expressing the gene in each cell-type cluster and the mean expression level of log-transformed counts [i.e., log(1 + count per 10,000)], respectively, as shown on the right side of the panel. The raw data are available as Supplementary Table S5. nk cell: natural killer cell.

Figure 6

The expression of IL6 and BDKRB1 in the lung. (A) The level of each ligand and receptor (including IL6 and BDKRB1) in each cell-type in the lung of the healthy subjects (Tabula Sapiens) is shown as dot. The size and the heat-intensity represent the ratio of cells expressing the gene in each cell-type cluster and the mean expression level of log-transformed counts [i.e., log(1 + count per 10,000)], respectively, as shown on the right side of the panel. The raw data are available as Supplementary Table S6. nk cell: natural killer cell. (B) The differential expression of IL6 and BDKRB1 in each cell-type in the IPF-lung is shown as dot. The cell-types are indicated on the left. The differential expression of IPF vs. non-IPF is indicated as log₂fold change (“log2FoldChange”). The dot size indicates the statistical significance of the differential expression as − log₁₀p-adj (“− log10padj”)—the larger size indicating more significant (i.e., less padj values). The blue and gray colors indicate padj $<$ 0.05 and padj $\ge$ 0.05, respectively. The raw data are available as Supplementary Table S7. padj adjusted p-value, AT1 cells alveolar type I cells, AT2 cells alveolar type II cells.

Figure 7

The expression of the signaling targets in the liver and the lung. (A) The differential expression of CALM1/CALM2/CALM3 in each cell-type in the IPF-lung is shown as dot. The cell-types are indicated on the left. The differential expression of IPF vs. non-IPF is indicated as log₂fold change (“log2FoldChange”). The dot size indicates the statistical significance of the differential expression as − log₁₀p-adj (“− log10padj”)—the larger size indicating more significant (i.e., less padj values). The blue and gray colors indicate padj $<$ 0.05 and padj $\ge$ 0.05, respectively. The raw data are available as Supplementary Table S7. padj: adjusted p-value; AT1 cells: alveolar type I cells; AT2 cells: alveolar type II cells. T/NKT cells: T/natural killer T cells. (B) The level of PCK1 in each cell-type in the liver is shown as dot. The size and the heat-intensity represent the ratio of cells expressing the gene in each cell-type cluster and the mean expression level of log-transformed counts [i.e., log(1 + count per 10,000)], respectively, as shown on the right side of the panel. The raw data are available as Supplementary Table S5. nk cell natural killer cell.

Figure 8

The predicted inter-organ mechanism of IPF. The solid arrows indicate the direct ligand–receptor interactions. The pathway connection (edge) is shown as dashed-arrows indicating the presence of one or more nodes (proteins: ligands, receptors, signaling targets) in between. The corresponding KEGG human pathways for each edge are indicated as hsa numbers. “?” indicates the lack of KEGG pathway connecting the nodes. BDKRB1 bradykinin receptor B1, CALM1/2/3 calmodulin 1/2/3, IL6 interleukin 6, IL6R interleukin 6 receptor, KNG1 kininogen 1, PCK1 phosphoenolpyruvate carboxykinase 1. KEGG pathways: hsa04151: PI3K-Akt signaling pathway; hsa05163: Human cytomegalovirus infection pathway; hsa05200: Pathways in cancer.