Pneumonectomy incurs collagen production without fibrosis
Fourteen days following either PNX or BLM challenge, lung tissue remodeling and collagen deposition were investigated by H&E staining and hydroxyproline assay, respectively. Expectedly, BLM caused marked thickening of the alveolar walls. In contrast, PNX did not cause any noticeable histologic changes (Fig. 1A-C). Hydroxyproline assay indicated a robust increase in whole-lung collagen content following either PNX or BLM (Fig. 1D). As fibroblasts are the primary cell type responsible for collagen deposition [3], we then aimed to gain a deeper understanding of how they respond differently to BLM or PNX by performing bulk RNA-sequencing on isolated lung fibroblasts.
Overlap Between Fibroblast Transcriptome Responses To Blm Or Pnx
RNA-sequencing identified several thousand protein-coding differentially expressed genes (DEGs) in Cola1-GFP + fibroblasts following BLM administration or PNX (Fig. 2A and B, Supplementary file 1). To functionally determine the biological significance of these differential gene expression programs, we performed pathway enrichment analysis using Qiagen’s Ingenuity Pathway Analysis (IPA) software. IPA runs a knowledge-based prediction algorithm that accounts for the magnitude and direction of each DEG. Figure 2B and C highlight the top ten most enriched biological processes among genes differentially regulated by either BLM or PNX, respectively. Notably, many processes were similar following either procedure, and were related to wound healing or fibrosis.
To identify differences between the responses to BLM and PNX, we also predicted upstream regulators based on DEGs using IPA. We identified considerable overlap between the most highly predicted regulators following BLM or PNX (Fig. 2F and G). Many of the predicted upstream regulators for both procedures were growth factors such as TGF-β1, or molecules related to immune signaling. Intrigued by this apparent similarity, we next asked which genes were linked to the top-predicted upstream regulators. Figure 2G shows Venn diagrams displaying the number of DEGs unique to or shared by fibroblasts responding to either BLM or PNX for the top two predicted upstream regulators (TNF and TGFB1). Notably, despite the overlap in predicted upstream regulators, distinct gene expression program subsets appeared to be engaged differentially in response to BLM and PNX. Further investigation of DEGs shared by both processes revealed considerable differences in their expression levels (Fig. 2I). These results suggest that there is an underlying difference between the responses to BLM or PNX, despite apparent similarity.
Itpkc is a key differentially expressed gene following BLM or PNX
We next aimed to gain a deeper understanding of key differences in the responses between the responses to BLM and PNX. Figure 3A shows a Venn diagram of the total number of common and unique genes that significantly increase or decrease in expression following BLM or PNX. In line with above observations that shared DEGs may change at noticeably different magnitudes, we first examined genes which significantly change in the same direction following both BLM and PNX. Expression heatmaps in Fig. 3B show an exaggerated effect of BLM on expression of genes that also change in the same direction following PNX. 510 of 595 genes (85.7%) that significantly increase during both processes increase more following BLM, whereas 200 of 267 (74.9%) common decreased genes decrease more following BLM. This suggests PNX promotes a more regulated lung repair mechanisms compared to BLM.
To further characterize differences in the responses to BLM and PNX, we performed enrichment analysis for Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways on genes that increased or decreased following only one procedure. The most significantly enriched KEGG pathways associated with genes only increased following BLM were mostly related to the increased proliferation, chemokine production, and inflammation (Fig. 3C). Top pathways for genes only increased by PNX were related to matrix digestion and absorption, TNF/NF-κB signaling, and calcium signaling (Fig. 3D). KEGG pathways enriched among genes that decreased only following BLM were linked closely to drug metabolism (Fig. 3E), indicting pathways associated with resolution are being repressed [18] Only one KEGG pathway, related to antiviral signaling was significantly enriched among genes that only decreased following PNX (not shown).
To further understand the responses to BLM and PNX, we next employed motif enrichment analysis using the software iRegulon [19]. This software takes a list of genes as input, and outputs a list of transcription factor motifs enriched according to the input gene set. Our input lists contained only up-regulated genes. Accordingly, the most enriched motifs are those that would be predicted to be the most accessible during the associated response. Figure 3F shows the top 5 most enriched motif clusters associated with genes up-regulated by BLM. Each cluster contains similar transcription factor binding motifs and is marked by the most enriched motif for that cluster. The response to BLM is strongly characterized by motifs associated with ETS1 and SPIB (Transcription factor Spi-B) family members. ETS1 and Spi-B are both members of the ETS superfamily of transcription factors [20]. Members of this family may have divergent roles in the context of fibrosis. For example, ETS1 itself has been associated with a matrix-degrading fibroblast phenotype in multiple tissues, including the lung [21]. Contrarily, PU.1 (Spi-1; another member of the ETS superfamily) has been shown to promote a switch from an inflammatory to a fibrotic fibroblast phenotype [22]. ETS1 and PU.1 expression in fibroblasts has also been found to be mutually exclusive across multiple datasets, suggesting there are separate populations of fibroblasts involved in fibrosis progression and resolution [21]. Inputting genes up-regulated by PNX revealed strong enrichment for motifs similar to RELA and JUN (Fig. 3G). RELA and JUN are members of the NF-κB and AP-1 families, respectively. Interestingly, the top-ranked target associated with the JUN motif cluster was Itpkc, a gene whose role in lung fibroblasts has yet to be described (Fig. 3H). Table S1 shows the top 10 targets for the ETS1 and SPIB motif clusters associated with BLM, and the RELA motif cluster associated with PNX. Supplementary file 3 shows a list of predicted motifs and their associated targets among genes significantly up-regulated following BLM or PNX.
Among genes regulated in different directions by BLM and PNX, Itpkc was one of the most highly up-regulated genes following PNX (Fig. 3I). Itpkc was not highly expressed under sham conditions and was further repressed by BLM (Fig. 3J). Interestingly, many of the genes regulated the most differently by BLM and PNX were involved in immune and inflammatory signaling (Fig. 3I). This is interesting because variants of ITPKC have previously been found to be associated with the vascular autoimmune syndrome Kawasaki disease [23], indicating a possible link to inflammatory signaling and prompting us to investigate this gene further.
Antifibrotic Role Of Itpkc
To investigate a potential antifibrotic role for ITPKC, we transiently transfected primary normal human lung fibroblasts (NHLFs) with recombinant human ITPKC (Fig. S1) and used exogenous TGF-β1 to model fibrotic changes. ITPKC overexpression reduced the effects of TGF-β1 on genes encoding collagen-1α1 (COL1A1), α-smooth muscle actin (ACTA2), and connective tissue growth factor (CTGF) (Fig. 4A). ITPKC overexpression also reduced TGF-β1-mediated α-SMA up-regulation at the protein level (Fig. 4B).v
ITPKC is an enzyme that catalyzes the conversion of IP3 to IP4. IP3 binds to and activates IP3 receptors (IP3Rs), which are calcium channels localized to the endoplasmic reticulum (ER). Upon activation, IP3Rs increase intracellular Ca2+ levels by allowing Ca2+ to leave the ER. To confirm the dependency of fibroblast activation by TGF-β1 on IP3-mediated Ca2+ release, we treated NHLFs with TGF-β1 in the presence or absence of the IP3R inhibitor 2-Aminoethoxydiphenyl borate (2-APB). As predicted, 2-APB reduced the effects of TGF-β1 on COL1A1, ACTA2, CTGF, and fibronectin 1 (FN1) as determined by qRT-PCR (Fig. S2A). Thapsigargin, which raises intracellular Ca2+ levels by preventing ER calcium uptake by inhibiting the sarco/endoplasmic reticulum ATPase (SERCA), had a similar effect to 2-APB (Fig. S2B). This is in agreement with reports that Ca2+ oscillations are an important signal in mediating the pro-fibrotic effects of TGF-β1 [24].
Encouraged by these findings, we aimed to determine the effects of ITPKC overexpression in more detail using RNA-seq. ITPKC overexpression in NHLFs resulted in up-regulation of 736 and down-regulation of 619 protein-coding genes (Fig. 4C and Supplementary file 2). IPA revealed enrichment for pathways mostly related to cytokine signaling, wound healing, and fibrosis (Fig. 4D). DEGs resulting from ITPKC overexpression were mostly linked to the plasma membrane and secreted cellular components, including the extracellular matrix (ECM) (Fig. 4E). Figure 4F shows down-regulation of several collagen genes, and the elastic fiber components fibulin-1 and − 2. Up-regulated genes include collagen-degrading matrix metalloproteinase 1 and vascular endothelial growth factor A, which has been shown to be protective against BLM-induced lung fibrosis [25].
Itpkc Overexpression Induces A Pnx-like Innate Immune Response In Fibroblasts
To better understand the signaling effects of ITPKC, we employed upstream regulator analysis following IPA in our ITPKC overexpression experiment. Biological upstream regulators predicted following ITPKC overexpression were closely linked to inflammatory pathways (Fig. 5A). The highest-scoring biological upstream regulator, TNF (encoding tumor necrosis factor-α), is well-described to activate canonical NF-κB signaling, as are IL-1α and -β. The NF-κB complex and the NF-κB family member RELA are also predicted to be highly activated upstream regulators. Genes linked to the TNF upstream regulator show marked differential expression following ITPKC overexpression, with the majority of genes increasing in expression (Fig. 5B).
Noticing a considerable similarity between the effects of PNX and ITPKC overexpression, we aimed to identify their overlapping genes and pathways. Among significantly up-regulated genes, 143 were common to ITPKC overexpression and PNX. Notably, nearly half of the KEGG pathways related to genes up-regulated by ITPKC were also enriched among genes up-regulated by PNX (Fig. 5C). These shared pathways were mostly linked to inflammatory signaling, and closely related to signaling by NF-κB (Fig. 5D). Finally, motif enrichment with iRegulon strongly predicted a motif cluster marked by the motif for the NF-κB family member RELA (Fig. 5E). Interestingly, this transcription factor also marked the most enriched predicted cluster following PNX (Fig. 3B). Predicted targets of the RELA cluster in the dataset include NFKBIA and NFKBIZ, encoding inhibitors of κB-α and -ζ, respectively (Fig. 5F). Other predicted motif clusters were marked by signal transducer and activator of transcription 2 (STAT2), the AP-1 family member FOSL1, forkhead box J1 (FOXJ1), and PU.1 (Fig. 5E). Supplementary file 4 shows a list of predicted motifs and their associated targets among genes significantly up-regulated following ITPKC overexpression. Taken together, these data predict that ITPKC overexpression induces an inflammatory response in fibroblasts that closely mimics changes that occur following PNX.