Temporal retinal transcriptome and systems biology analysis identifies key pathways and hub genes in Staphylococcus aureus endophthalmitis

Abstract

Bacterial endophthalmitis remains a devastating inflammatory condition associated with permanent vision loss. Hence, assessing the host response in this disease may provide new targets for intervention. Using a mouse model of Staphylococcus aureus (SA) endophthalmitis and performing retinal transcriptome analysis, we discovered progressive changes in the expression of 1,234 genes. Gene ontology (GO) and pathway analyses revealed the major pathways impacted in endophthalmitis includes: metabolism, inflammatory/immune, antimicrobial, cell trafficking, and lipid biosynthesis. Among the immune/inflammation pathways, JAK/Stat and IL-17A signaling were the most significantly affected. Interactive network-based analyses identified 13 focus hub genes (IL-6, IL-1β, CXCL2, STAT3, NUPR1, Jun, CSF1, CYR61, CEBPB, IGF-1, EGFR1, SPP1, and TGM2) within these important pathways. The expression of hub genes confirmed by qRT-PCR, ELISA (IL-6, IL-1β, and CXCL2), and Western blot or immunostaining (CEBP, STAT3, NUPR1, and IGF1) showed strong correlation with transcriptome data. Since TLR2 plays an important role in SA endophthalmitis, counter regulation analysis of TLR2 ligand pretreated retina or the use of retinas from TLR2 knockout mice showed the down-regulation of inflammatory regulatory genes. Collectively, our study provides, for the first time, a comprehensive analysis of the transcriptomic response and identifies key pathways regulating retinal innate responses in staphylococcal endophthalmitis.

Figure 1. Principal Component Analysis (PCA) of normalized expression data obtained from controls and SA infected samples.

The first component with the highest variance (65.7%) is on the X-axis separating 12 and 24 h post-infection samples from control samples and 3h post- infection samples. The second component, with the second highest variance (13.8%), is on the Y-axis, depicting a maximum variation of the 12 h post-infection samples from rest of samples. The biological replicates from control, 3 h, 12 h, and 24 h samples formed separate clusters on the PCA plot, indicating transcriptional differences among different groups.

Figure 2. Significant transcriptional changes induced due to infection.

(A) Self-organizing map (SOM) analysis of 1,234 genes commonly differentially expressed among different post-infection time points compared to controls. The SOM maps were clustered into ‘constitutive’ (similar expression changes at 3 h, 12 h, and 24 h) and ‘temporal’ (progressive expression changes at 3 h, 12 h, and 24 h) expression changes on the basis of structure of expression pattern. The figure shows violin plots of selected 3 maps depicting constitutive down-regulation (I), temporal down-regulation (II) and temporal up-regulation (III) patterns. Each cluster represents a set of genes that depict similar expression patterns and are biologically linked to a specific function. The X-axis represents different time points and groups, and the Y-axis represents gene expression on pseudoscale from −3 to +3. (B) Heatmap of constitutive genes depicting similar expression changes at 3 h, 12 h, and 24 h post-infection. (C) Heatmap of temporal genes depicting progressive expression changes from 3 h, 12 h, and 24 h post-infection. Columns represent samples, and rows represent genes. Gene expression levels are shown on a pseudocolor scale (−1 to 1), with red denoting a high expression level and green denoting a low expression level.

Figure 3. Comparison of affected pathways between constitutively and temporally altered genes post-infection.

Pathways significantly enriched in constitutive versus temporal post-infection altered genes. Pathways affected by constitutively altered genes are shown as black bars, while those affected by temporally altered genes are shown as grey bars.

Figure 4. Interactive network of the regulatory molecules and their down-stream targets that are significantly activated or inhibited during the temporal post-infection phase.

Top regulatory genes were selected on the basis of enrichment score, as derived from the dysregulation of target molecules and their fold changes due to infection. In this network, activated and inhibited regulatory molecules are shown with orange and blue color, respectively. The regulatory molecules include the activation of multiple key inflammation related molecules, including NFKB2, IL-1β, CXCL2, and JUN. Also, in this network, each node represents a gene and each edge represents an interaction. The up-regulated and down-regulated target genes are shown with shades of red and green colors, respectively.

Figure 5. Counter regulation of gene expression by PAM3 pre-treatment, as determined by transcriptome profile comparison of temporal samples with and without PAM treatment prior to infection.

(A) SOM maps representing infection-related gene clusters that were significantly counter regulated by PAM3 pre-treatment. PAM3 pre-treatment counter regulated 280 genes out of the approximately 1,200 genes altered due to infection. (B) Heatmap of genes that are significantly counter regulated due to PAM3 pre-treatment. Columns represent samples, and rows represent genes. Gene expression levels are shown on a rainbow pseudocolor scale (−3 to 3), with red denoting high expression levels and blue denoting low expression levels.

Figure 6. Functional and Pathway enrichment analysis of genes significantly counter regulated by PAM3 pre-treatment.

(A) Functional enrichment analysis depicting the counter regulation of multiple functional categories related to the immune and inflammatory response, as well as cell death and survival. (B) Pathway enrichment analysis depicting the counter regulation of multiple inflammatory and cell adhesion response related pathways, including IL-10 signaling, IL-6 signaling, JAK/STAT signaling, TNFR2 signaling, and Toll-like receptor signaling. The significance of the effect of PAM3 pre-treatment on functional and pathway categories was determined using multiple test corrected Fisher’s exact test p values.

Figure 7. Master regulatory molecules counter regulated by PAM3 Pre-treatment to improve the outcome following infection.

Figure 8. qRT-PCR analysis of regulatory molecules for validation of gene expression data.

C57BL/6 or TLR2^−/− (B6 background) mice were intravitreally injected with S. aureus (SA) for the indicated time points. Total RNA was extracted, reverse transcribed, and subjected to qRT-PCR using specific primers for thirteen master regulator genes (IGF1, Jun, STAT3, NUPR1, CEBPB, CSF1, CyR61, EGFR1, SPP1, TGM2, IL-6, IL-1β, and CXCL2) with glyceraldehyde 3-phosphate dehydrogenase (GAPDH) as the control. Modulations of gene expression were expressed as relative fold changes with respect to the GAPDH control. Statistical analysis was performed using one-way ANOVA (*) and Student’s t-test (^§) for comparisons of control versus stimulated mice over time, and C57BL/6 vs. TLR2^−/− mice, respectively. Data points and error bars represent the mean ± SD of triplicates from three independent experiments. (*^§P < 0.05; **^§§P < 0.005; ***^§§§P < 0.0005).

Figure 9. Analysis of regulatory molecules at the protein level.

C57BL/6 mice were given intravitreal injections of PBS (control) or SA/Pam3. After 24 h, enucleated eyes were either embedded in OCT or fixed in paraffin. Cryosections were subjected to immunostaining (A), as described in material and methods. From a separate group, eyes were enucleated and lysates were made in PBS using a tissue lyser. Retinal lysates were subjected to Western blotting to validate protein expression (C, left panel). The band intensities were quantitated by densitometric analysis using Image J and presented as bar graph using β-actin as control (C right panel). Protein levels of inflammatory cytokines were assessed by ELISA (B). Paraffin embedded eyes were subjected to H&E staining for histological analysis (D). (*P < 0.05, **P < 0.005, ***P < 0.0005 ANOVA).