The experiments described here involved the use of mice and included both male and female C57BL/6 mice (8 weeks of age; Sipra Labs, Hyderabad, India). All procedures were conducted according to guidelines and recommendations of the Guide for the Care and Use of Laboratory Animals by Institutional Animal Ethics Committee (IAEC), Sipra Labs, Hyderabad, India. The study was approved under protocol SLL-PCT-IAEC/20-21/B. The animals were also maintained according to institutional guidelines and the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research.

The clinical strains of MDR-PA (L-2051/18) and S-PA (L-2050/18) were isolated from the vitreous of patients clinically diagnosed with infectious endophthalmitis at our institute. Overnight cultures were grown to logarithmic phase in brain heart infusion broth and serially diluted in saline to approximately 10,000 colony-forming units (CFU)/μl for intravitreal injection, as described below.

C57BL/6 mice were anesthetized with a mixture of ketamine and xylazine, and topical anesthetic (0.5% proparacaine HCl) was instilled in each eye before injection. Intravitreal injections were carried out just posterior to the superior limbus, and 1 μl volume was injected directly into the midvitreous, using a surgical microscope (ZEISS Stemi 508 Stereo Microscope, Jena, Germany). The contralateral eyes of all mice were injected in a similar manner with 1 μl of phosphate-buffered saline. All eyes were assessed throughout the course of infection by slit lamp biomicroscopy, while immunohistochemical and histologic analyses, along with quantification of bacterial growth and microarray analysis, were performed after enucleation of the eye at 6 and 24 h postinfection (p.i.).

The clinical changes occurring during the induced experimental endophthalmitis were scored independently by an ophthalmologist who was blinded to the type of injection using a hand-held slit-lamp biomicroscope (PSLAIA-11, Appasamy Associates, Chennai, India). Ocular inflammation was scored based on anterior segment inflammation, presence/absence of red reflex, vitreous inflammation, and retinal clarity. Clinical changes were graded on a scale from 0 to 4+ based on the criteria described earlier (20). Photographs of mouse eyes were taken for visualization of progression of disease severity.

MDR-PA- and S-PA-infected eyes were harvested at 6 and 24 h p.i. The eyeballs were placed in 400 µl of sterile phosphate-buffered saline on ice and homogenized (EzLyser Genetix Biotech Asia, New Delhi, India) with 1.0-mm glass beads for 90 s at maximum speed. The homogenates were serially diluted and plated on BHI agar plates. Values represent the means SEM for 3 eyeballs at each time point.

For histopathological scoring of endophthalmitis, eyeballs were enucleated and fixed in 10% neutral buffered formalin, then embedded in paraffin for sectioning at indicated time points. Tissue sections were cut through the pupillary-optic nerve axis. The sections were then stained with hematoxylin and eosin (H&E). Disease severity was assessed by a pathologist who was blinded to the tissue, slide identity, and strain used, using defined grading criteria ranging from 0 (no disease) to 4 (maximum disease) based on the extent of inflammation in the cornea, anterior chamber, vitreous, and integrity of retinal architecture. Eyes that went into panopthalmitis were given a score of 4. Further sections were additionally assessed by gram stain, CD45 (leukocyte marker), MPO (neutrophil marker), and GFAP (astrocyte and muller glia marker) staining and examined using an Olympus light microscope (BX51). Briefly, the procedures included acetone fixation and nonspecific rabbit serum block, along with anti-CD45, anti-MPO, and anti-GFAP antibody (Dako, Agilent, Santa Clara, CA, USA) labeling followed by chromogenic staining using a DAB substrate (Dako, Agilent) and hematoxylin counterstain. Sections of CD45 and MPO were analyzed by counting the total positive cells in the posterior chamber of ten ×400 random microscope fields per case and averaging the number. The GFAP intensity was evaluated based on different force of IHC expression in investigated groups and include: “negative” (−), “weak” (+), “moderate” (++), “strong” (+++), and their variations.

Terminal deoxynucleotidyl transferase-mediated dUTP nick-end labeling (TUNEL) assay-based protocol (Click-iT^® TUNEL Alexa Fluor^® 488, Thermo Fisher Scientific, Waltham, MA, USA) was used to assess extent of DNA fragmentation as an indicator of apoptotic death within the entire retinal thickness. DNase I enzyme was used to generate strand breaks in the DNA to provide a positive TUNEL reaction. The coverslips were washed with molecular biology-grade water. Followed by this, reaction mixture was prepared using deionized water, DNase I buffer, and DNase I. A total of 100 µl of the above solution was added to coverslip and was incubated for 30 min at room temperature. After the incubation, the coverslips were then washed once with deionized water and was subjected to TdT reaction. Images were taken using an upright AXIO Imager.M2 Zeiss fluorescence microscope under a ×20/0.8 lens. For each eye, 3–4 serial sections were taken. Random images were also counted by pathologist who was blinded to the tissue, slide identity, and strain used while TUNEL⁺ cell density was calculated as the ratio of TUNEL⁺ cells/mm of the length of the region of interest using an automated ImageJ macro as previously described (21).

Total RNA from mice eye ball (n = 3) from each group was isolated by Qiagen RNA extraction followed by DNase I treatment. Quality and quantity of isolated RNA was evaluated by determining RNA integrity number (RIN) using an Agilent 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA, USA) with nanochip systems. Total RNA with RINs ≥7 was selected for control and infected eyes at each time point for gene expression microarray analysis.

The mouse tissue gene expression profile was assessed at two time points (6 and 24 h) postinfection with MDR-PA and S-PA using SurePrint G3 Mouse Gene Expression Microarrays (Agilent, Santa Clara, CA, USA) containing 60,000 probes, allowing for the analysis of >30,000 well-characterized mouse genes and long noncoding RNAs. To identify statistically significant differences in gene expression, the microarray analysis was performed on 3 separate biological replicates at each time point. Briefly, total RNA from each sample was amplified and transcribed into fluorescent cRNA with using Agilent Quick Amp Labelling protocol (version 5.7; Agilent Technologies). RNA samples from MDR-PA-infected mice were labeled with Cy5 (red), and Cy3 (green) was used for S-PA-infected mice eye. The labeled cRNAs were hybridized onto SurePrint G3 Mouse Gene Expression Microarrays (Agilent, Santa Clara, CA, USA). After hybridization at 65°C for 17 h, subsequent to washing of the slides with washing buffer, the arrays were scanned and analyzed using the Agilent Scanner G2505C (Agilent Technologies, Inc. Agilent Feature Extraction software (version 11.0.1.1; Agilent Technologies, Inc.).

The probe sets of two dye arrays were annotated with Enterz Identification (ID) using the Bioconductor and the DAVID gene-ID conversation tool. The differential gene expression analysis was performed with limma package in R (v4.0.0.)/Bioconductor (v3.11). After applying background correction within the arrays and loess normalization, a linear model was fitted across the samples of two groups (3 biological replicates each) to estimate the differential expression of each gene. Bonferroni correction was applied to p-value, and Benjamini Hochberg correction was used for false discovery rate A gene is said to be expressed if it has an absolute log2 fold change of more than or equal to 1 and an FDR of less than 0.1. The fold change was calculated over 3 biological replicates to reduce false-positive rates.

In order to study the similarity between the biological replicate of the MDR-PA- and S-PA-infected groups, we perform principal component analysis on the expression data using R statistical package. The scoring and loading plots for the first two components were created. The gene expression patterns across the two groups were studies by performing hierarchical clustering in R to create heatmaps. These heatmaps also revealed the extent of the separation between the two groups and samples.

The differential expression of genes between the two groups were examined for the enrichment of the pathways compiled in Kyoto Encyclopedia of Genes and Genomes (KEGG) database using online DAVID tool. For a given set of DEGs from our analysis, an enrichment analysis was performed using the official gene symbol as an identifier for the species Mus musculus to obtain a functional annotation chart consisting of enriched pathways in the KEGG database. Venn diagrams were also created using Venny (1.0) to illustrate overlap between data points in these gene sets.

Cell lysate samples were generated using RIPA buffer supplemented with phosphatase and protease inhibitors (50 mM Tris-HCl pH 7.4, 1% (v/v), 150 mM, NaCl, 0.5% (w/v) sodium deoxycholate, 1 mM EDTA, 0.1% (w/v) SDS, 1 mM Na₃VO₄, 1 mM PMSF) and protease inhibitor cocktail (Sigma). Protein samples were denatured using sample buffer containing reducing agent (BioRad, hercules, CA, USA) and heated at 70°C for 10 min. Samples were separated on a 10% polyacrylamide gel and subsequently transferred to PVDF membrane (BioRad). The membrane was then blocked in 5% nonfat milk in TBS-T (PBS with 0.1% Tween20) for 1 h at room temperature followed by incubation with primary antibodies of MMP-9 (Abcam, Cambridge, MA, USA), phospho-MAPKAPK2 (CST), NF-кβ-p65 (CST), and β-actin (Santa Cruz, Dallas, TX, USA), overnight at 4°C. The membrane was then washed three times in TBS-T followed by a 1-h room-temperature incubation with IRDye 680RD goat antimouse IgG (LI-COR, Lincoln, NE, USA) secondary antibody. Followed by the incubation, membrane was washed three more times with TBST. Images were captured using Licor Odyssey CLx Imaging System (LI-COR Biotechnology, Lincoln, NE, USA). Abundance was determined by standard densitometry analysis, using ImageJ software (NIH, Bethesda, MD, USA) with β-actin as normalizing protein.

First-strand cDNA was synthesized from equal RNA amounts using the Verso cDNA Synthesis Kit (Thermo Scientific). cDNA was further amplified using primers for mouse caspase-1, caspase-4, caspase-8, NLRP3, and β-actin (internal control) using SYBR Green Master Mix (Thermo scientific).

Mice cytokine/chemokine magnetic bead panel (MCYTOMAG-70K, Merck Millipore, Darmstadt, Germany) was used for the Milliplex ELISA. The concentrations of 3 human mediators (IL-1β, IL-6, and TNF-α) were measured using a multiplex assay instrument (MAGPIX; Bio-Rad Laboratories, Hercules, CA, USA) according to the manufacturer’s instructions. Standard curves of known concentrations of recombinant human cytokines (R&D Systems, Minneapolis, MN) were used to convert fluorescence units to cytokine concentration (picograms/milliliter). The protein concentration of each sample was determined using BCA method (G-Biosciences, St. Louis, MO, USA).

An unpaired, two-tailed Student’s t-test was used to determine statistical significance for data from the cytokine ELISA and bacterial count. A nonparametric Mann-Whitney U test was performed for clinical score, histological score, and TUNEL-positive cells. A p-value <0.05 was considered statistically significant. Data are presented as means ± SEM or SD as indicated in the figure legends. Statistical analyses were performed using GraphPad Prism 9 (version 9; GraphPad Software Inc., La Jolla, CA, USA). R scripts were used to construct principal component analysis (PCA) plots, volcano plots, heatmap, and the pathway enrichment plot. All error bars are defined in the figure legends. Statistical methods were used to predetermine sample size.

Intravitreal injections of 10⁴ CFU/eye resulted in reproducible clinical endophthalmitis as assessed and graded clinically and histologically with mice having corneal haze, vitreous haze, and intraocular inflammation ( Figure 1A ). Disease severity measured by clinical scoring, showed a time-dependent progression which also correlated with the histopathological grading. Although MDR-PA showed early clinical signs of endophthalmitis, the average clinical score of MDR-PA did not differ significantly at 6 h from S-PA mice (p = 0.15); however at 24 h, the difference in mean clinical score was statistically significant (p = 0.01) ( Figure 1B ). At 6 h p.i., minimal flare in the anterior chamber was observed with slight diminution in red reflex. Inflammatory symptoms progressed at 24 h p.i. with intense flare, vitreous haze, and a significant decrease in red reflex. Inflammatory symptoms were more severe in MDR-PA-infected mice, with haze in the anterior chamber, along with severe iritis and vitreous and retinal clarity was completely obscured by infiltrates. In contrast, control animals had a clear cornea and anterior chamber with no visible signs of inflammation.

Gross histopathological examination of enucleated globes of infected eyes showed inflammation characterized by robust inflammatory infiltrate and disruption of retinal architecture along with retinal folds. The severity of disease was assessed by histological scoring in three major categories: inflammation in posterior and anterior segments and integrity of retinal architecture. Microscopic examinations of the control eyes were consistent with lowest histological scores with no signs of inflammation and intact retinal integrity ( Figure 1C ). At 6 h p.i histological analysis revealed mild inflammation with minimal retinal folds which was higher in MDR-PA-infected mice eye, though the difference in mean histological scores was not statistically significant (p > 0.05). On the contrary, sections from the eye of MDR-PA recipients at 24 h displayed showed greater accumulation of inflammatory cells, with higher retinal folds and indistinguishable retinal layers (p = 0.01). The anterior segment of the MDR-PA-infected mice eye showed significantly higher inflammatory cells (p = 0.008) and higher retinal layer disruption ( Figure 1D ). Additionally, severe periorbital and surrounding tissues indicated panopthalmitis. Taken together, these observations demonstrate that the disease severity was higher in MDR-PA-infected mice, which worsened with time as compared with the S-PA-induced endophthalmitis.

Next, we evaluated the intraocular bacterial growth as the ability to replicate and survive within host cells after infection, as critical determinants of disease pathogenesis. MDR-PA-infected mice eye had higher number of colony forming units (CFU) at both 6 and 24 h, though this difference was not statistically significant (p = 0.16 at 6 h, p = 0.08 at 24 h) ( Figure 1F ). Grossly, however, gram staining of the eye section of the mice confirmed the presence of rod-shaped bacilli in the vitreous cavity only at 24 h p.i. ( Figure 1E ), while it was not visible at 6 h p.i.

To quantify the inflammatory cell infiltration, the number of the CD45+ and MPO+ cells were counted in the posterior segment, then the mean value was calculated. We observed a significant increase in the number of CD45+ and MPO+ cells in MDR-PA-infected mice eye. Furthermore, the number of CD45+ cells were significantly lower in S-PA-infected mice eye at both the 6 h p.i. (p = 0.007) and 24 h p.i(p= 0.007) ( Figures 2A, B ). The CD45+ cells in MDR-PA-infected mice eye displayed progressive increase over time, whereas in the case of S-PA-infected mice eye, CD45+ cells showed no appreciable change over time. Higher MPO+ cells were found in MDR-PA-infected mice eye only at 24 h p.i (p = 0.01) ( Figures 2C, D ). To assess the extent of differential retinal glial (astrocytes and muller glia) stress following MDR-PA and S-PA infection, we used GFAP at both the time points. MDR-PA-infected mice retinas showed significant increase in immunoreactivity for GFAP, suggesting an increased retinal stress. As seen in Figures 2E, F , the intensity of GFAP was robustly increased in MDR-PA-infected mice eye at both 6 and 24 h compared with the susceptible (p = 0.02, p = 0.03, respectively; n = 4).

To correlate the retinal stress with retinal cell death, TUNEL assay was carried out at aforementioned time points. Early studies have shown that increased retinal cell death is associated with declined visual function in bacterial endophthalmitis. In the present study, MDR-PA induced robust cell death as compared with the S-PA infections, as seen in Figure 3A . Surprisingly, both S-PA and MDR-PA infections did not lead to retinal cell death in any layer of the retina at 6 h p.i. At 24 h, the retinal cell layers of MDR-PA-infected mice were indistinguishable followed by higher number of TUNEL+ cells in MDR-PA-infected group (9,285.43 ± 535.04) than that in the S-PA-infected group (6,363.16 ± 377.98, p = 0.008) ( Figure 3B ). Most apoptosis was observed in the inner nuclear layer and outer nuclear layer; however, further immunocytochemical analysis was not done.

In this study, we sought to explore the possibility of finding host genes collectively de-regulated by the MDR-PA strain. To this end, microarray was performed on RNA isolated from the mice eye infected with MDR-PA and S-PA over 6 and 24 h postinfection. The PCA for the 6 h time point indicated a strong correlation between biological replicates and highlights a rapid divergence in transcriptional profile between S-PA-infected group and MDR-PA-infected group ( Figure 4A ). At 24 h time, one of the biological replicates ( Figure 4B ) was clustered near to the S-PA group, while three of the biological replicates from the S-PA group clustered together similar to the pattern displayed at 6 h. Despite the differences, the MDR-PA group still managed to cluster differently from the S-PA-infected group. The analysis revealed distinctive clusters of samples consistent with the biological groups on the basis of the transcriptional profiles. Moreover, a volcano plot was constructed to visualize the DEGs between two groups at the two time points ( Figures 4C, D ). When compared with S-PA-infected mice eyes, a total of 1,088 genes at 6 h and 2,900 genes at 24 h were found to be significantly differentially expressed in MDR-PA-infected group (FC ≥2, p < 0.05). As this variance in number of DEGs has the potential to bias downstream comparisons, we further adjusted the p-value to reduce the false-discovery rate to take forward for comparative functional enrichment analysis. With an adjusted p-value, we found that a total of 923 genes at 6 h and 2,220 genes at 24h were significantly differentially regulated. Additionally, we also investigated the genes which were only expressed at 6 and 24 h along with the genes which showed temporal expression. In total, 259 genes were found to be expressed exclusively at 6 h while 1,699 genes were expressed at 24 h; 521 genes were found to be expressed at both 6 and 24 h ( Figure 4E ).

To gain insight into the functional relevance of the DEGs, DAVID was used. Some of the DEGs involved in immune regulation were further represented by a heatmap, where not only the number of upregulated genes but also the similarity patterns between the three biological replicates were particularly striking. This suggests that MDR-PA infection initially triggers an increased immune response, manifesting as an early qualitative and quantitative changes in gene transcription, which grows rapidly and spreads exponentially over time compared with S-PA-infected mice eye. For instance, at 6 h ( Figure 4F ) genes like mitogen-activated protein kinase 1 (Map4k1 (fold change (fc), 2.3), C-type lectin domain family 7 member A (Clec7a, fc = 3.6), Mast cell-expressed membrane protein 1 (Mcemp1, fc = 5.5), macrophage-expressed gene 1 protein (Mpeg1, fc = 3.4), and B-cell leukemia/lymphoma 2-related protein A1d (Bcl2a1d, fc = 3.3) were differentially upregulated compared with the S-PA-infected mice group. In contrast, the downregulated genes involved calmodulin-1 (Calm1, fc = 2.2), Fas apoptotic inhibitory molecule 1 (Faim, fc = 2.1), Fas apoptotic inhibitory molecule 2 (Faim 2, fc = 2), G-protein coupled receptor-associated sorting protein 2 (Gprasp2, fc = 2.2), and regulator of G protein signaling (Rgs 7, fc = 2).

At 24 h ( Figure 4G ), the top upregulated genes compared with the S-PA-infected group involved lipocalin (LCN-2, fc = 3.5), interferon, alpha-inducible protein 27-like protein 2A (Ifi27l2a, fc = 2.2) which is suggested to be a novel regulator of neuroinflammation in microglia, LPS-induced TN factor (Litaf, fc = 2.4), CD-14 (fc = 14.6) which is a marker for monocytes along with TLR 2 (fc = 2.7), TLR 4 (fc = 5.1), TNF receptor-associated factor 1 (TRAF1, fc = 20.8) whose expression is largely limited to activated immune cells, including myeloid and lymphoid cells. TRAF1 is present at minimal levels in resting lymphocytes and monocytes, and its expression is increased upon activation through the nuclear factor kappa-light-chain enhancer of activated B-cell (NF-κB) pathway. Chemokines like C–C motif chemokine ligand (Ccl)3 (fc = 14.9), Ccl4 (fc = 42.2), Ccl6 (fc = 16.6), Ccl7 (fc = 2.23), and Ccl9 (fc = 6.91) were found to be upregulated. Few of these molecules are previously implicated in endophthalmitis pathogenesis. GFAP was also found to be upregulated at 24 h, which was also observed in the histopathological section. At the other extreme Fas apoptotic inhibitory molecule 1 (Faim 1, fc = 2.9), regulator of G-protein signaling 9 and 7 (fc = 2.73, 2 respectively), ubiquitin-specific peptidase 33 (Usp33, fc = 2.41), and ubiquitin carboxyl-terminal hydrolase Usp2 (fc = 3.81) were among the genes whose expression was downregulated. Reduced USP33 has been shown to decrease the stability of a suppressor molecule of major proinflammatory gene expression pathways of TNF-α, NF-κB, and IFN-β (22). This indicates that regulatory molecules related to apoptotic pathway, GPCR pathway, and suppressors of major proinflammatory pathways are possibly involved in the pathogenesis of drug-resistant endophthalmitis. The extent of differential expression of several genes further intrigued us to construct another heatmap to compare the expression level at 6 h post-MDR-PA infection with 24 h post-S-PA infection ( Figure 4H ). We observed that even at a 24-h time point, the amplitude of host immune response induced by S-PA was secondary to the immune response prompted by MDR-PA.

Several studies have addressed the role of cytokines and chemokines (inflammatory mediators) in bacterial endophthalmitis. Because of their importance, we focused on the differentially expressed cytokines and chemokines in MDR-PA-infected mice eye at both study time points. At the early time point (6 h), cytokines like in interleukin (IL)-1β (fc = 5.9 ± 0.4), IL-20rb (fc = 2.3 ± 0.12), IL-1rn (fc = 3.9 ± 0.19), IL-10ra (fc = 7.15 ± 0.16), IL-2rb (fc = 5.65 ± 0.11), chemokine (C-X-C motif) ligand (CXCL)-9 (fc = 2.14 ± 0.2), CXCL-5 (fc = 4.6 ± 0.27), C-X-C chemokine receptor (CXCR)-2 (fc = 3.2 ± 0.10), CXCR-3 (fc = 2.3 ± 0.11), CXCR-4 (fc = 2.6 ± 0.12), CC chemokine ligand (CCL)-9 (fc = 2.4 ± 0.04), CXCL-10 (fc = 2.2 ± 0.3), and CXCR-6 (fc = 2.5 ± 0.07) were found to be upregulated. Similarly, inflammatory mediators like CXCL-1 (fc = 3.2 ± 0.01) and CCL-17 (fc = 2.1 ± 0.05) were found to be downregulated. However, during the late hour (24 h) time point, a whole range of inflammatory mediators were found to be upregulated. Table 1 enlists all the inflammatory mediators, associated proteins, and few regulatory molecules along with its average expression in the three biological replicates that were affected in this cytokine storm. In addition, we observed that components of the complement system were also significantly overexpressed at 24 h p.i. Those include the following, Cfb (fc = 3.4), C5ar1 (fc = 10.2), C1qa (fc = 2.1), C1qb (fc = 2.1), C1qc (fc = 2.4), C3 (fc = 8.1), C4b (fc = 2.7), while C1ql3, which stimulates glucose uptake in various cells, was found to be comparatively downregulated by 3.2-fold compared with the S-PA-infected group. Altogether, this suggests an explosive immune response at the later time point, which is likely to be responsible for collateral retinal damage and a good percentage of vision loss.

To validate the role of identified immune mediators that were overexpressed in the pathogenesis of MDR-PA endophthalmitis, we analyzed the level of IL-6 and TNF-α by multiplex ELISA. In accordance with microarray, IL-1β in MDR-PA-infected eyes was observed to be expressed higher than the S-PA-infected ones both at 6 h (192.75 ± 13.69 vs. 117 ± 11.23; p = 0.01) and 24 h (10,395.84 ± 86.67 vs. 385.33 ± 10.45; p = 0.007). While IL-6 was also observed to be higher at both 6 h (68.43 ± 8.27 vs. 46.74 ± 6.43; p = 0.03) and 24 h (119.4 ± 14.01 vs. 46.01 ± 2.93, p = 0.009), we could not validate the differential expression of TNF-α by ELISA ( Figure 5A ). Additionally, genes encoding pro- and antiapoptotic/pyroptosis molecules were found to be differentially regulated in MDR-PA-induced endophthalmitis and included upregulation of caspase-1, caspase-4, and caspase-8. NOD-, LRR-, and pyrin domain-containing protein 3 (NLRP3) which is known to be a critical component of the innate immune system that mediates caspase-1 activation and the secretion of proinflammatory cytokines was also found to be upregulated. In addition, caspase recruitment domain6 (CARD6), which was observed to be involved in assembly of signal complexes leading to activation of caspase family proteases and induction of NF-кβ was also differentially upregulated. Furthermore, BCL2L15 (B-cell lymphoma 2) which is an amplifier of the apoptotic signal (23) was also differentially expressed in MDR-PA induced endophthalmitis. Another important mediator of caspase-8 activation, known as FADD was found to be upregulated. TRAF6 which is believed to be triggered by LPS and interacts with caspase-4 was also upregulated. The interaction of these two molecules leads to NF-кβ-dependent transcriptional upregulation and secretion of cytokines and chemokines. On the contrary, Faim2, which is known as an antiapoptosis gene that directly interacts with Fas receptor and inhibits apoptosis by interfering with caspase 8 was found to be downregulated, along with Fas apoptosis inhibitory molecule (FAIM) which is an antiapoptotic protein. We speculate that the aberrant expression of IL-1β is due to pleiotropic functions of caspases mediated by NLRP3 and is involved in the disease pathogenesis. The above-mentioned observations prompted us to validate the expression of the caspases and NLRP3. To this end, our RT-PCR results were in agreement with the microarray data and we observed differential expression of caspase-1, caspase-4, and caspase-8 and NLRP3 in MDR-PA induced infection ( Figure 5B ).

We next explored the expressional variation of matrix metalloproteinases (MMPs) in the MDR-PA-infected mice eyeballs which were found to be higher compared with S-PA-infected mice eyeballs. These included upregulation of MMP-25, MMP-8, MMP-3, MMP-9, and MMP-19, while the inhibitory molecules like TIMP-2 and TIMP-3 were found to be downregulated at 24 h p.i. Interestingly, at the early time point of 6 h p.i., only MMP-3 and MMP-19 were upregulated while MMP-13 and TIMP-3 were downregulated ( Figure 6A ). Collectively, these findings highlight elevated expression of these peptidases that may contribute to the pathogenesis of MDR-PA endophthalmitis by ECM remodeling. Western blot analysis of MMP-9 confirmed its overexpression in MDR-PA, while comparatively, lower level of MMP-9 was observed in S-PA-infected mice eye at both the time points. These data suggested that the expression of MMP-9 increases over the course of infection caused by P. aeruginosa but is significantly higher in the MDR strains ( Figure 6B ). The Western blot band intensities were measured with ImageJ software. MMP-9 protein levels were normalized to respective β-actin levels and were plotted. The difference in band intensity was calculated between S-PA vs. MDR-PA at both 6 and 24 h ( Figure 6C ) and was found to be higher in MDR-PA-infected eyes.

While the top DEGs provided some interesting insights, it is important to take a systemic approach to the data. Therefore, overrepresented pathways were discerned among DE genes using KEGG pathway, in which all genes were mapped to the reference canonical pathways of KEGG. When the core upregulated genes were sorted into gene ontology categories, the cytokine- and chemokine-mediated signaling pathway emerged as the biological processes with the most significant changes in gene expression at 6 and at 24 h. Accordingly, at 6 h, cytokine-cytokine receptor interaction, chemokine signaling pathway followed by phagosome as well as NF-кβ pathways were the pathways to which upregulated genes could be assigned ( Figure 7A ). Accordingly, several genes that mediate activation of the abovementioned pathways at 6 h are listed in Supplementary Table 1 . Genes known to have a role in G-Protein-coupled receptor pathway were found to be negatively regulated in MDR-PA endophthalmitis and included Rgs7 and Gprasp2. At 24 h p.i., most of the genes upregulated by MDR-PA were found to be involved in cytokine-cytokine receptor signaling, chemokine signaling, MAPK signaling, NF-кβ, Toll-like receptor signaling, NOD-like receptor pathway, TNF signaling, and complement system ( Figure 7B ). Several genes that mediate activation of the abovementioned pathways at 24 h are listed in Supplementary Table 2 . These responses are necessary to limit pathogen infections, but at the same time it needs to be optimally regulated to avoid immune-mediated host damage. When we analyzed the downregulated genes, the pathways involved included Rap1, cAMP, Ras, cGMP-PKG, inflammatory mediator regulation of TRP channels, phototransduction, and phosphatidylinositol signaling system.

Out of all the enriched pathways by the differentially upregulated genes, we further validated NF-кβ and an important regulatory molecule of MAPK pathway MAPKAPK2(MK2) by Western blotting using phosphorylated antibodies ( Figure 7C ). While NF-кβ was expressed differentially at both the time points, MK2 was differentially expressed only at 24 h. The Western blot band intensities were measured with ImageJ software. MAPKAPK2 ( Figure 7D ) and NF-кβ ( Figure 7E ) protein levels were normalized to respective β-actin levels and were plotted. The difference in band intensity was calculated between S-PA vs. MDR-PA at both 6 and 24 h. These results indicated that Western blot data were consistent with the microarray analysis. Together, these results suggest that NF-кβ and MAPK may be facilitating the increased inflammation.