Gene ontology (GO) enrichment analysis of differentially expressed molecules (DEMs)
The raw transcriptome data of BALF (PXD026983) and serum (PXD017500) had been upload to PRIDE (https://www.ebi.ac.uk/pride/), and the raw transcriptome data of PMBCs had been upload to GEO (GSE179183). We first used UNIPORT (http://www.uniprot.org/uploadlists/) to convert the obtained protein identities from BALF and serum into their respective gene symbols. We defined differentially expressed genes (DEGs) and proteins (DEPs) as differentially expressed molecules (DEMs). FunRich software was used to assign the GO classification (Ashburner et al. 2000) and to analyze DEMs in PBMCs, serum, and BALF at the early infection stage (0–24 h). We found that cellular components including lysosomes, exosomes, and the cytoplasm are commonly expressed DEMs in all the three groups. Specifically, DEMs in BALF were significantly enriched in endosomes, endoplasmic reticulum and centrosomes; while in PBMC and BALF were plasma membrane and nucleus categories (Fig. 1A). There was almost no consistent DEMs in molecular function aspect in PBMCs, serum and BALF. Moreover, DEMs in PBMCs were mainly annotated as being involved in transcription regulator activity, and DEMs in serum were mainly related to ubiquitin-specific protease activity. In contrast, DEMs in BALF were involved in molecular functions such as ligase, hydrolase, catalytic, and transporter activities. Additionally, DEMs in oxidoreductase activity and TGPase activity changed consistently in both serum and BALF (Fig. 1B). In biological processes, DEMs in PBMCs were involved in regulation of nucleobase, nucleoside, nucleotide and nucleic acid metabolism; DEMs in both serum and BALF had similar changes in the cell growth and/or maintenance, protein metabolism, energy pathway and metabolism categories (Fig. 1C).
With the progression of infection (24–120 h), the significant cellular component DEMs included those in the cytoplasm, exosomes, lysosomes, Golgi and centrosomes. Specifically, DEMs in PBMCs were decreased in Golgi apparatus and nucleus, while DEMs in serum were increased in Golgi apparatus, centrosome, plasma membrane, but decreased in the mitochondrion. In BALF, there were no significant DEMs except the plasma membrane category (Fig. 1D). At 120 h post infection, the DEMs in PBMCs showed an increase in hydrolase activity, receptor signaling complex scaffold activity, and transcription regulator activity. DEMs in serum were enriched in cell adhesion molecule activity and, the changes were almost continuing from 24 to 120 h in BALF similarly. Only DEMs of cytoskeletal protein binding showed consistently up-regulation in PBMCs, serum and BALF, implying phagocytosis and cell migration for bacterial clearance is activated in both lung and periphery (Fig. 1E). Interestingly, biological process analysis showed that cell growth and maintenance, communication and signal transduction were highly active in serum, but not in PBMCs and BALF at 120 h; while material metabolism was consistently upregulated in PBMCs, serum and BALF, indicating material metabolism changes mainly during the late stage of APP infection. Moreover, regulation of nucleobase, nucleoside, nucleotide and nucleic acid metabolism was highest in PBMCs, while protein metabolism in BALF (Fig. 1F).
Biological processes (BPs) enrichment analysis of DEMs
Metascape analysis (Zhou et al. 2019) was used for BPs enrichment analysis of DEMs in PBMCs, serum, and BALF post infection. The results showed that the metabolism of amino acids and derivatives, and metabolism of RNA were enriched in BALF, and the lysosome is the main way for BALF to eliminate pathogens (Fig. 2A). Furthermore, we found that adaptive immunity was significantly down-regulated in the early stages of APP infection. According to our previous findings, PBMCs were one of the important sites for host immune responses, such as response to wounding, wound healing, regulation of cell adhesion, and positive regulation of organelle organization (Jiang et al. 2018). These results indicate that serum also plays an important role in the activation of immune responses (Fig. 2A). We could further see that biological pathways including metabolism, lymphocyte activation, and regulated exocytosis were present throughout the entire process of infection (0–120 h); while the immune response in serum mainly occurs in the early stage of infection (0–24 h), including activation of immune response, regulated exocytosis and adaptive immune system (Fig. 2A). Leukocyte activation involved in immune response was significantly up-regulated in the early stage of infection (0–24 h), but was gradually down-regulated by 120 h in BALF, serum and PBMCs, indicating that this pathway plays an important role in the process of host resistance to APP. PBMCs showed the strongest activities of BPs in the early stage of infection (0–24 h), not only including up-regulated signaling pathways in interleukins, response to wounding and positive regulation of cell migration, but also down-regulated signaling pathways in negative regulation of cellular component organization, regulation of cytoskeleton organization, regulation of cell adhesion, cellular response to nitrogen compound, adaptive immune system, transcriptional regulation by TP53 and organelle localization. BALF had the more obvious up-regulated BPs in terms of protein localization to membrane, aromatic compound catabolic process, response to oxidative stress, transmembrane receptor protein tyrosine kinase signaling pathway, vesicle-mediated transport, leukocyte activation involved in immune response, and metabolism of RNA, while little BPs were shown in serum (Fig. 2B). During 24–120 h post infection, more BPs of BALF were down-regulated and few were up-regulated. In contrast to BALF, more BPs were up-regulated than down-regulated in PBMCs (Fig. 2C). Together, these data demonstrate that the innate immune responses in PBMCs and serum responded rapidly and were maintained compared to the lung where metabolism and cell adhesion activities were enriched upon APP infection.
KEGG enrichment analysis of DEMs
KEGG enrichment analyses of the host immune response against APP infection were performed using Database of Annotation, Visualization, and Integrated Discovery (DAVID) Bioinformatics Resources 6.8 (https://david.ncifcrf.gov/) (Huang et al. 2008, 2009). The results from DAVID analysis are consistent with the GO enrichment analysis. During infection, BALF responds to APP mainly by induction of natural immune pathways such as phagosome, endocytosis, lysosomes and various metabolisms, including metabolic pathways, oxidative phosphorylation and TCA cycle (Fig. 3A). Moreover, these pathways were significantly up-regulated in the early infection stage (0–24 h) (Fig. 3B), while subsequently down-regulated in the late infection (24–120 h) along with metabolic pathways (Fig. 3C).
Cytokine-cytokine receptor interaction, NOD-like receptor signaling pathway, and T cell receptor signaling pathway are the main signaling pathways in PBMCs upon APP infection (Fig. 3A, B), which were in consistent with previous results (Jiang et al. 2018). Similar to that in BALF, phagosome and other pathways were also significantly enriched in PBMCs (Fig. 3B). However, the T cell receptor signaling pathway in PBMCs showed a dramatic down-regulation in the early infection stages (Fig. 3B).
The response in serum to APP infection was mainly enriched in biosynthesis and metabolism, such as carbon metabolism and glycolysis/gluconeogenesis, which was both up-regulated and down-regulated in early infection (0–24 h) and late infection (24–120 h), respectively (Fig. 3B, C). However, unlike in BALF and PBMCs, the complement and coagulation cascades and phagosome in serum showed a significant down-regulation throughout the infection (Fig. 3B). In summary, metabolic activities in lungs (based on KEGG analysis) were significantly up-regulated in the early stage, but inhibited at the later stage of infection.
Visualization of common DEMs in serum, BALF and PBMCs during APP infection
We used Venn diagrams to identify common DEMs in serum, BALF and PBMCs. It was found that there were in total of 147 DEMs in at least two of serum, BALF and PBMC at 0–24 h, including 7 common DEMs. Twenty-three DEMs were identified in both BALF and serum at 24 h post infection, indicating that the number of DEMs in BALF (23/619 + 23 + 96) was lower than that in serum (23/133 + 28 + 23). At 120 h post infection, the number of common DEMs in both BALF and serum was similar, but the number of DEMs in BALF plus serum decreased significantly, while the number of DEMs in both PBMCs and lung were increased (Fig. 4A, B), which implied that the lung had less material exchange with serum but more with the PBMCs over the time of infection process, and immune cells in blood may participate in the immune response in lung.
We further used Cytoscape ClueGO (Bindea et al. 2009) analysis to identify cross-talk (associated with two or more pathways) DEMs, and found that these molecules were mainly enriched in mononuclear cell migration, apoptotic cell clearance and platelet degranulation (Fig. 4C). In addition, we also found that some important DEMs (i.e., THBS1, MAPK14) were involved in the regulation of different KEGGs. These data clearly showed the central points of cross-talk involves multiple BPs, e.g., the vascular endothelial growth factor receptor signaling pathway, positive regulation of receptor-mediated endocytosis, negative regulation of sodium ion transport, platelet degranulation, positive regulation of nucleocytoplasmic transport, mononuclear cell migration and antioxidant activity, several relatively concentrated interaction networks (some molecules were connected between the networks) and the main function was to initiate an immune response (Fig. 4C). However, the pathways of extracellular matrix disassembly, negative regulation of kinase activity, cellular response to reactive oxygen species, cellular response to oxidative stress, cellular response to chemical stress, neutrophil mediated immunity, granulocyte activation and leukocyte cell-cell adhesion had become the central points at 120 h post infection, and many molecules, such as CST3, A2M, RAB6A, LCN2, TGFB1, CASP3, CXCL12 were connecting points among these important pathways, connecting multiple signaling pathways to each other and forming a complex and integrated network. Moreover, innate immunity of PMNs (polymorphonuclear neutrophils) related with the anti-infection activity, complement activation, immune cell migration, immune cell stress response, antibiotic metabolism was highly active (Fig. 4D).
Metascape analysis of the interactions between common DEMs involved in APP infection
The main interaction molecules in BALF, PBMCs and serum were analyzed systematically and a large number of interactional molecules were found. At 24 h post infection, 14 molecules (TRIP10, WASL, TFRC, ARPC5, GALK2, EIF3I, SNRPE, FN1, ARFGEF1, TUBA4A, DHX15, SRSF1, EIF4A1 and PPP2R1B) were up-regulated in BALF, 10 of which were also dramatically changed in PBMCs, but only 3 of them were up-regulated. Moreover, only six (APOB, IARS, XPOT, GART, CAD, SOD1) were down-regulated in BALF, and 4 (IARS, XPOT, GART, CAD) in PBMCs. At 120 h, the up-regulated molecules in BALF were significantly reduced, only two (SOD1 and XPO1) found; while 20 down-regulated molecules (including PRKAR2A, PPP2R1B, IARS, ALDH2, DHRS11, and SLC25A6) were found in BALF, 11 of which were also up-regulated in PBMCs, and nine showed the same trend as in BALF (Fig. 5A). These data indicate that the immune response in lung is different from that in blood.
The common interaction molecular analysis of BALF and serum showed that there were four up-regulated molecules (ARPC5, SAR1A, TUBA4A, EIF4A1) found in BALF at 24 h, three of which (ARPC5, SAR1A, EIF4A1) were also up-regulated in serum; but two down-regulated molecules (APOB and SOD1) in BALF were up-regulated in serum. At 120 h, there were two up-regulated molecules (SOD1 and XPO1) in BALF. XPO1 was also up-regulated in PBMCs, while SOD1 was down-regulated. There were two down-regulated molecules (EIF4A1, LMNA) both in BALF and serum, and no up-regulated molecules in serum at 120 h (Fig. 5A). Thus, there were in total of eight common interaction molecules (ARPC5, SAR1A, TUBA4A, EIF4A1, APOB, SOD1, EIF4A1, LMNA) found in both BALF and serum, with only three (ARPC5, SAR1A, EIF4A1) having the same uptrend. Collectively, the data showed that there were significant differences in immune response between lung tissue and serum.
There were few DEMs in PBMCs. Three DEMs (TUBB1, FN1, PRDX5) were up-regulated in PBMCs at 24 h, of which PRDX5 was up-regulated, while TUBB1 and FN1 were down-regulated in serum. Only IGF2R was down-regulated both in PBMCs and serum. At 120 h, both PRKAR1A and YWHAG were up-regulated in PBMCs and down-regulated in serum. Six molecules (TUBA4A, SOD1, EEF1G, TUBB1, ENO1, ENO3) were down-regulated in both PBMCs and serum (Fig. 5A). Overall, there are only 12 coexisting molecules, with eight of them have a similar change trend.
The proportion of the main DEMs in BALF, PBMC and serum was analyzed. It was found that there were less interactive DEMs in the early stage of infection (24 h), with 22 in BALF (red), 14 in PBMC (green), and eight in serum (orange), among which 10 molecules (red + green) were common in both BALF and PBMC. In the late stage of infection (120 h), complex networks were formed, and the number of related molecules increased. There were 22 (blue) in BALF, 21 in PBMCs (purple), and eight in serum (yellow), of which 16 (blue + purple) were common in BALF and PBMCs. Notably, SOD1 is an important interaction molecule that exists both in BALF, PBMC and serum, and participates in early and late stages of infection, followed by PRKAR1A, IARS, SNRNP200, and DHX15 (Fig. 5B). These molecules interact with other DEMs during APP infection, which may play an important role in the host’s immune response.
Key pathway node molecule detection and confirmation in the network of serum, BALF and PBMCs during APP infection
According to the data integration analysis of serum, BALF and PBMCs, many signaling pathways involved in innate immunity were closely correlated. In order to verify the results of omics network regulation, we selected DEMs including MAPK14, ALDH3B1 and CST3 (Fig. 4C, D) to detect their mRNA levels in PBMCs by qPCR, and TUBA4A, SOD1 and EIF4A (Fig. 5A, B) to detect their protein contents in serum and BALF by ELISA. The results showed that the mRNA level of MAPK14, namely mitogen-activated protein kinase 14, was significantly increased at 24 h and decreased at 120 h after APP infection compared to the control group. This is consistent with the results of the omics data. From the network analysis of the omics, it can be seen that MAPK14 connects eight pathways at 24 h, including vascular endothelial growth factor receptor (VEGFR) signaling pathway, positive regulation of nucleocytoplasmic transport, regulation of leukocyte chemotaxis, positive regulation of leukocyte chemotaxis, mononuclear cell migration, regulation of mononuclear cell migration, and macrophage chemotaxis. Positive regulation of chemotaxis is an important pathway. However, the difference was not significant at 120 h and was not screened out in the pathway association analysis, which is consistent with the gradual remission of systematic inflammation in the late stage of APP infection (Fig. 6).
The mRNA level of ALDH3B1 (aldehyde dehydrogenase family 3 member B1) was significantly increased in PBMCs at 24 h, but not at 120 h compared to control, which was consistent with the results of PBMC transcriptome analysis. CST3, also known as Cystatin-C, is an inhibitor of cysteine protease that belongs to cystatin superfamily. qPCR found no significant difference in gene expression at 24 h, but there was a significant decrease at 120 h post infection compared to the control, which was consistent with the results of transcriptome analysis (Fig. 6).
For the ELISA detection of TUBA4A, SOD1 and EIF4A in serum and BALF, we only analyzed samples from 12 to 120 h post infection due to the shortage of the same batch of samples for omics detection. We found that the level of TUBA4A, SOD1 and EIF4A in BALF were consistent with the results of omics analysis (12 h/0 and 120 h/12 h) (Fig. 5A, B). The level trends of TUBA4A, SOD1 and EIF4A in serum were also consistent with the omics data only at 12 h, no difference or increase trend at 120 h/12 h in serum by ELISA assay. It was not similar with that in serum at 120 h/24 h, which might be due to the different sampling time points for ELISA (12 h) and for the omics analysis (24 h), that is to say, these proteins levels may still be rising at 12 h, higher at 24 h, so 120/24 h decreased than 120 h/12 h (Fig. 7).
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.