KLF17 is an important regulatory component of the transcriptomic response of Atlantic salmon macrophages to Piscirickettsia salmonis infection

Abstract

Piscirickettsia salmonis is the most important health problem facing Chilean Aquaculture. Previous reports suggest that P. salmonis can survive in salmonid macrophages by interfering with the host immune response. However, the relevant aspects of the molecular pathogenesis of P. salmonis have been poorly characterized. In this work, we evaluated the transcriptomic changes in macrophage-like cell line SHK-1 infected with P. salmonis at 24- and 48-hours post-infection (hpi) and generated network models of the macrophage response to the infection using co-expression analysis and regulatory transcription factor-target gene information. Transcriptomic analysis showed that 635 genes were differentially expressed after 24- and/or 48-hpi. The pattern of expression of these genes was analyzed by weighted co-expression network analysis (WGCNA), which classified genes into 4 modules of expression, comprising early responses to the bacterium. Induced genes included genes involved in metabolism and cell differentiation, intracellular transportation, and cytoskeleton reorganization, while repressed genes included genes involved in extracellular matrix organization and RNA metabolism. To understand how these expression changes are orchestrated and to pinpoint relevant transcription factors (TFs) controlling the response, we established a curated database of TF-target gene regulatory interactions in Salmo salar, SalSaDB. Using this resource, together with co-expression module data, we generated infection context-specific networks that were analyzed to determine highly connected TF nodes. We found that the most connected TF of the 24- and 48-hpi response networks is KLF17, an ortholog of the KLF4 TF involved in the polarization of macrophages to an M2-phenotype in mammals. Interestingly, while KLF17 is induced by P. salmonis infection, other TFs, such as NOTCH3 and NFATC1, whose orthologs in mammals are related to M1-like macrophages, are repressed. In sum, our results suggest the induction of early regulatory events associated with an M2-like phenotype of macrophages that drives effectors related to the lysosome, RNA metabolism, cytoskeleton organization, and extracellular matrix remodeling. Moreover, the M1-like response seems delayed in generating an effective response, suggesting a polarization towards M2-like macrophages that allows the survival of P. salmonis. This work also contributes to SalSaDB, a curated database of TF-target gene interactions that is freely available for the Atlantic salmon community.

Keywords: Atlantic salmon; Piscirickettsia salmonis; gene regulatory network; host-pathogen interaction; macrophage polarization.

Figure 1

Experimental design and comparative analysis of transcriptomic response in Atlantic salmon challenged with P. salmonis. (A) Experimental strategy. (B) Quantification of DEGs at 24- and 48-hpi, shared and exclusively expressed in each experimental condition. (C) Comparison of our DEGs with other studies using GeneSectR (78, 79) where Atlantic salmon immune response was used (Ssa-org: Atlantic salmon tissue analysis; Psa: infection with P. salmonis; POMV: infection with Pilchard orthomyxovirus; ISAV: infection with Infectious Salmon Anaemia virus), obtaining significance of the overlap of genes shared between the different studies. A study about intestine Atlantic salmon without infection feed with different diets was used as an outlier. The upper value is the log2 fisher odds ratio, the middle value is the log10 p-adj, and the lower is the overlap size between datasets. *padj < 0.05, **padj < 0.01. Results >1 log2 transformed Fischer odds ratio along with a log10 adjusted p-value of -1.3 indicates an adjusted p-value of 0.05.

Figure 2

Co-expression analysis of all DEGs and GO terms enrichment analysis. (A) WGCNA analysis of all obtained DEGs, rows represent the different co-expression modules, and columns indicate the different infection times analyzed. (B) GO term enrichment analysis for each co-expression module, the size of each point is proportional to the gene ratio calculated between the obtained genes from a determinate GO term and the total genes that have that GO term in its annotation; Enrichment represents the -log10(adjusted p-value).

Figure 3

Gene regulatory network at 24-hpi. (A) Co-expression network along with regulatory interactions obtained from our reference GRN. Each module is color-coded, module 1: genes upregulated since 24-hpi; module 2: genes upregulated at 48-hpi; module 3: genes downregulated at 48-hpi; module 4: genes downregulated at 24-hpi. Purple, undirected purple dashed edges represent co-expression and directed green edges represent TF-target regulatory interactions. The transcription factors are represented as octagons. (B) GRN of all the DEGs at 24-hpi, the nodes are color-coded to match upregulation (red) and downregulation (blue). The target genes are represented as modules for ease of view of TF hierarchy and regulation. Purple, undirected purple dashed edges represent co-expression, while directed grey edges represent TF-target regulatory interactions. (C) Abundance of DEGs in percentage between 48-hpi DEGs and the DEGs found in our global analysis. (D) Functional analysis by GO term enrichment performed for each co-expression module (horizontal axis). The size of each point is proportional to the gene ratio calculated between the obtained genes from a determinate GO term and the total genes that have that GO term in its annotation; Enrichment represents the -log10(adjusted p-value).

Figure 4

Gene regulatory network at 48-hpi. (A) Co-expression network along with regulatory interactions obtained from our reference GRN. Each module is color-coded, module 1: genes upregulated since 24-hpi; module 2: genes upregulated at 48-hpi; module 3: genes downregulated at 48-hpi. Purple, undirected purple dashed edges represent co-expression and directed green edges represent TF-target regulatory interactions. The transcription factors are represented as octagons. (B) GRN of all the DEGs at 48-hpi, the nodes are color-coded to match upregulation (red) and downregulation (blue). The target genes are represented as modules for ease of view of TF hierarchy and regulation. Purple, undirected purple dashed edges represent co-expression, while directed grey edges represent TF-target regulatory interactions. (C) Abundance of DEGs in percentage between 48-hpi DEGs and all the DEGs found in our global analysis. (D) Functional analysis by GO term enrichment performed for each co-expression module (horizontal axis). The size of each point is proportional to the gene ratio calculated between the obtained genes from a determinate GO term and the total genes that have that GO term in its annotation; Enrichment represents the -log10(adjusted p-value).

Figure 5

Gene regulatory network for core genes. (A) Co-expression network along with regulatory interactions obtained from our reference GRN. Each module is color-coded, module 1: genes upregulated since 24-hpi; module 2: genes upregulated at 48-hpi. Purple, undirected purple dashed edges represent co-expression and directed green edges represent TF-target regulatory interactions. The transcription factors are represented as octagons. (B) GRN of all the core DEGs, the nodes are color-coded to match upregulation (red) and downregulation (blue). The target genes are represented as modules for ease of view of TF hierarchy and regulation. Purple, undirected purple dashed edges represent co-expression, while directed grey edges represent TF-target regulatory interactions. (C) Abundance of DEGs in percentage between core DEGs and all the DEGs found in our global analysis. (D) Functional analysis by GO term enrichment performed for each co-expression module (horizontal axis). The size of each point is proportional to the gene ratio calculated between the obtained genes from a determinate GO term and the total genes that have that GO term in its annotation; Enrichment represents the -log10(adjusted p-value).

Figure 6

Atlantic salmon macrophage response against P. salmonis. Summary of the regulatory interactions between TFs and the biological processes they regulate in Atlantic salmon macrophage-like cells SHK-1 infected by P. salmonis.