*3.1. Comparison of Glaucoma-Relevant Transcriptomic Datasets Identifies Common Genes for Therapeutic Targeting*

Publicly available transcriptomic datasets from three different animal models relevant to glaucoma (mouse controlled optic nerve crush [3], DBA/2J mouse model of glaucoma [8], and rat optic nerve transection [4]) were analyzed to identify common pathways and shared gene expression profiles. Comparison of the DE genes revealed commonality between models with ~30–40 DE genes shared between individual comparisons (comparisons are displayed in Figure 1 and DE gene lists are detailed in Supplementary Table S1). Threeway comparison of differentially expressed genes identified 12 shared DE genes between

all three models (Figure 1 and Supplementary Table S1). The identity and role of these genes are detailed in Supplementary Table S2. Pathway analysis revealed a role of these shared DE genes predominantly in the immune system, neuroinflammatory signaling, and amino acid biosynthesis pathways (Table 2). However, the majority of pathways had a low number of gene hits (pathway hit %), demonstrating that these identified DE genes do not collectively belong to a single or conserved pathway of RGC degeneration.

**Figure 1. Comparison of RNA-sequencing data identifies gene changes common to models of retinal ganglion cell injury.** DE gene lists were compiled from publicly available RNA-sequencing datasets from the DBA/2J mouse model of glaucoma (Williams et al., 2017 [8]), mouse controlled optic nerve crush model (CONC; Yasuda et al., 2014 [3]), and rat optic nerve transection model (ONT; Yasuda et al., 2016 [4]). The results are displayed as a Venn diagram showing the total number of DE genes by dataset, and overlap demonstrating shared DE genes. A three-way comparison identified 12 common genes (listed to right) that may represent gene changes conserved to RGC injury, and thus useful therapeutic targets.

We then used the CTD to screen for compounds that interact with these shared DE genes in order to identify potential novel therapeutics. Screening revealed 40 compounds that interact with >50% of these shared genes (Supplementary Table S3). A number of these are chemical by-products or inorganic compounds tested in toxicity assays (e.g., for carcinogenic effects) and as such are not suitable therapeutics. Other compounds had known neurodegenerative or anti-neuroprotective properties/responses (e.g., LPS and genistein). We identified eight compounds that may be suitable therapeutics based on a literature search (Supplementary Table S3), as they are either hormonal compounds, simple dietary compounds, or compounds that have already been tested in neurodegenerative contexts. These eight compounds were valproic acid, SB-431542 (an inhibitor of TGF-Beta Type I Receptor/ALK5, ALK4 & ALK7), progesterone, estradiol, choline, folic acid, WY-14643 (pirinixic acid, a peroxisome proliferator-activated receptor alpha (PPARα) agonist), and rosiglitazone.



**Table 2.** Pathway analysis of shared differentially expressed genes between RNA-sequencing experiments.



150
