Transcriptome analysis of long non-coding RNAs in Mycobacterium avium complex-infected macrophages

Abstract

Mycobacterium avium complex (MAC) is a non-tuberculous mycobacterium widely distributed in the environment. Even though MAC infection is increasing in older women and immunocompromised patients, to our knowledge there has been no comprehensive analysis of the MAC-infected host-cell transcriptome-and particularly of long non-coding RNAs (lncRNAs). By using in vitro-cultured primary mouse bone-marrow-derived macrophages (BMDMs) and Cap analysis of gene expression, we analyzed the transcriptional and kinetic landscape of macrophage genes, with a focus on lncRNAs, during MAC infection. MAC infection of macrophages induced the expression of immune/inflammatory response genes and other genes similar to those involved in M1 macrophage activation, consistent with previous reports, although Nos2 (M1 activation) and Arg1 (M2 activation) had distinct expression profiles. We identified 31 upregulated and 30 downregulated lncRNA promoters corresponding respectively to 18 and 26 lncRNAs. Upregulated lncRNAs were clustered into two groups-early and late upregulated-predicted to be associated with immune activation and the immune response to infection, respectively. Furthermore, an Ingenuity Pathway Analysis revealed canonical pathways and upstream transcription regulators associated with differentially expressed lncRNAs. Several differentially expressed lncRNAs reported elsewhere underwent expressional changes upon M1 or M2 preactivation and subsequent MAC infection. Finally, we showed that expressional change of lncRNAs in MAC-infected BMDMs was mediated by toll-like receptor 2, although there may be other mechanisms that sense MAC infection. We identified differentially expressed lncRNAs in MAC-infected BMDMs, revealing diverse features that imply the distinct roles of these lncRNAs in MAC infection and macrophage polarization.

Keywords: Ingenuity Pathway Analysis (IPA); M1 or M2 macrophage; bone-marrow-derived macrophage (BMDM); cap analysis of gene expression (CAGE); long non-coding RNA (lncRNA); non-tuberculous mycobacterium (NTM).

Figure 1

Analysis of genes upregulated in macrophages infected with Mycobacterium avium complex. CAGE-seq transcriptome analysis was performed at 0, 4, 12, or 24 h post-infection. Average expression values of quadruplicate data were plotted in each figure. The standard deviation was omitted except for (C). (A) Induction of representative cytokine gene expression. (B) Induction of representative chemokine gene expression. (C) Expressional change of Nos2 and Arg1. (D) Induction of membrane protein genes, Cd14, Cd83, and Clec4e. (E) Transient induction of Tlr2 among the Tlrs.

Figure 2

Expressional change of M1- or M2-associated transcription factor (TF) genes in MAC-infected macrophage. Up- or down-regulated TF genes in IFNγ- or IL4/IL13-stimulated macrophages (M1 or M2 activation, respectively) were taken from previous work (24), and their expressional change in MAC-infected macrophages was shown by heat map; IFNγ upregulated (A) and downregulated (B) and IL4/IL13 upregulated (C) and downregulated (D) TF genes. Degree of the change was shown by color of log2.

Figure 3

Expressional clustering analysis of long non-coding RNAs (lncRNAs) differentially expressed in Mycobacterium avium complex-infected macrophages. Hierarchical clustering of the expression profiles of (A) upregulated and (B) downregulated lncRNAs. Upregulated lncRNAs peaking at the early (4 h) stage were termed cluster 1 (red), and those peaking late (24 h) were termed cluster 2 (blue).

Figure 4

Gene ontology (GO) enrichment analysis of protein-coding transcripts associated with long non-coding RNAs. We extracted protein-coding transcripts with Pearson correlation coefficients of both more than 0.8 and less than -0.8 with each differentially expressed lncRNA, and we then subjected them to GO analysis. The top 10 most enriched GO biological processes of (A) upregulated and (B) downregulated lncRNA-associated protein-coding transcripts generated by using the ToppCluster tool (https://toppcluster.cchmc.org/).

Figure 5

Prediction of canonical pathways associated with differentially expressed long non-coding RNAs (lncRNAs). We extracted protein-coding transcripts with Pearson correlation coefficients of both more than 0.9 and less than -0.9 with each differentially expressed lncRNA, and we then subjected them to Ingenuity Pathway Analysis (IPA) to predict canonical pathways associated with differentially expressed lncRNAs. The top most enriched canonical pathways of (A) upregulated and (B) downregulated lncRNA-associated protein-coding transcripts with z-scores greater than 4 generated by using the IPA platform.

Figure 6

Prediction of upstream transcription regulators associated with differentially expressed long non-coding RNAs (lncRNAs). We extracted protein-coding transcripts with Pearson correlation coefficients of both more than 0.9 and less than -0.9 with each differentially expressed lncRNA, and we then subjected them to Ingenuity Pathway Analysis (IPA) to predict upstream transcription regulators associated with differentially expressed lncRNAs. The top most enriched upstream transcription regulators of (A) upregulated and (B) downregulated lncRNA-associated protein-coding transcripts with z-scores greater than 5 generated by using the IPA platform.

Figure 7

Expression profiles of representative long non-coding RNAs in macrophages infected with Mycobacterium avium complex. CAGE-seq transcriptome analysis was performed at 0, 4, 12, or 24 h post-infection. Average expression values and their standard deviation were shown in each figure. (A) Induction of representative lncRNAs, U90926, Mir155hg, and AW112010. (B) Expressional change of another upregulated lncRNA Morrbid and its associated protein-coding gene Bcl2l11. (C) Suppression of representative lncRNAs, AI662270, AU020206, and Snhg15.

Figure 8

Expressional change of representative long non-coding RNAs during Mycobacterium avium complex infection under no preactivation (CTL) or under M1 (IFNγ) or M2 (IL4/IL13) preactivation. IFNγ or IL4/IL13 stimulation was carried out 24 h prior to Mycobacterium avium complex infection (M1 or M2 preactivation). After 0 and 24 h of the infection, total RNA was extracted from macrophages and subjected to CAGE-seq transcriptome analysis. Average expression values and their standard deviation were shown in each figure. *, #, $, +, and & denote statistical significance between 0 and 24 h, between no- and M2-preactivation at 24 h, between no- and M1-preactivation at 0 h, between no- and M1-preactivation at 24 h, and between no- and M2-preactivation at 0 h, respectively (P value less than 0.01 with Student’s unpaired T test).

Figure 9

Expressional fold change of representative long non-coding RNAs in Mycobacterium avium complex-infected wild-type (WT) and Tlr2 knockout (Tlr2KO) macrophages. Expressional fold change of representative long non-coding RNAs between 0 and 24 h of Mycobacterium avium complex-infected WT and Tlr2KO macrophages was analyzed by RT-qPCR. Average expressional fold change and its standard deviation were shown in each figure. Asterisks denote P value less than 0.01 with Student’s unpaired T test between WT and Tlr2KO macrophages.