Quantitative proteomic analysis after neuroprotective MyD88 inhibition in the retinal degeneration 10 mouse

Abstract Progressive photoreceptor death occurs in blinding diseases such as retinitis pigmentosa. Myeloid differentiation primary response protein 88 (MyD88) is a central adaptor protein for innate immune system Toll‐like receptors (TLR) and induces cytokine secretion during retinal disease. We recently demonstrated that inhibiting MyD88 in mouse models of retinal degeneration led to increased photoreceptor survival, which was associated with altered cytokines and increased neuroprotective microglia. However, the identity of additional molecular changes associated with MyD88 inhibitor‐induced neuroprotection is not known. In this study, we used isobaric tags for relative and absolute quantification (iTRAQ) labelling followed by LC‐MS/MS for quantitative proteomic analysis on the rd10 mouse model of retinal degeneration to identify protein pathways changed by MyD88 inhibition. Quantitative proteomics using iTRAQ LC‐MS/MS is a high‐throughput method ideal for providing insight into molecular pathways during disease and experimental treatments. Forty‐two proteins were differentially expressed in retinas from mice treated with MyD88 inhibitor compared with control. Notably, increased expression of multiple crystallins and chaperones that respond to cellular stress and have anti‐apoptotic properties was identified in the MyD88‐inhibited mice. These data suggest that inhibiting MyD88 enhances chaperone‐mediated retinal protection pathways. Therefore, this study provides insight into molecular events contributing to photoreceptor protection from modulating inflammation.

demonstrated that TLRs contribute to neuronal death in inherited and induced retinal degenerations. [2][3][4][5] The Myeloid differentiation primary response protein 88 (MyD88) protein is a central adaptor molecule for most TLRs and mediates TLR-induced signalling in inflammatory and non-inflammatory cells. Therefore, MyD88 has been considered as a potential target to block aberrant activation of TLR during neurodegeneration. [5][6][7][8][9][10][11][12][13] We recently demonstrated that inhibiting MyD88 by gene knockout 14 and pharmacologic inhibitor 2 led to higher photoreceptor survival in two mouse models of retinal degeneration. Furthermore, neuroprotection from inhibiting MyD88 in the rd10 mouse model of retinal degeneration was associated with increased microglia/macrophage expressing the Arg1 neuroprotective marker and altered levels of several anti-inflammatory cytokines. 2 However, additional molecular changes associated with MyD88 inhibition are currently unknown. In this study, we were interested in identifying early molecular events after MyD88 inhibition as an important first step to understanding how it induces photoreceptor protection.
High-performance liquid chromatography tandem mass spectrometry combined with isobaric tag labelling with iTRAQ is a sensitive, highly accurate and high-throughput method to identify differentially expressed proteins. The iTRAQ method has not been reported for the analysis of the retinal proteome in photoreceptor degeneration, although it has been used to profile protein changes after retinal detachment, 15 myopia 16 and optic nerve transection. 17 In this study, we performed iTRAQ quantitative proteomic analysis on rd10 mouse retinas to characterize protein changes that may contribute to the protective effects of inhibiting MyD88. We identified altered expression of 42 proteins, including anti-apoptotic crystallins, chaperones and regulators of protein biosynthesis. These findings suggest a link between modulating inflammatory pathways and chaperone expression and suggest that inhibiting MyD88-mediated signalling may enhance intrinsic tissue-protective pathways. Therefore, this study provides new insight into molecular changes that may contribute to photoreceptor protection and provide a foundation for understanding molecular mechanisms contributing to retinal homeostasis in rd10 mice.

| Animal studies
All experiments using mice were approved by the Animal Care and Laboratory. The rd10 line is homozygous for a mutation in the rodspecific visual transduction Pde6b gene and is commonly used to model retinitis pigmentosa and test experimental therapies. [18][19][20] The retinal degeneration phenotype in rd10 is specific to the retina due to the causative gene, Pde6b, being exclusively expressed in rod photoreceptors. Mice of both sexes were used, and they were housed under 12-h light-dark cycle with ad libitum access to food and water.
Littermates were used for control and inhibitor injections, and all animals were housed at an equivalent distance from the overhead light.
Mice at age post-natal day 18 were intraperitoneally (IP) injected with MyD88 inhibitor peptide (IMG2005, Novus Biologicals) following the procedures of our recent study. 2 The MyD88 inhibitor peptide contains a MyD88 homodimerization sequence (underlined, DRQIKIWFQNRRMKWKKRDVLPGT) and a protein transduction (PTD) sequence derived from antennapedia for cellular permeability.
The control peptide contains only the PTD sequence. Both peptides were solubilized in sterile PBS according to the manufacturer's directions, aliquoted and stored frozen, then freshly diluted immediately prior to use. Mice (n = 4 MyD88 inhibitor peptide, n = 4 control peptide) were injected with MyD88 inhibitor peptide or control peptide at a concentration of 2 mg/kg body weight, which is the dose that induced maximum photoreceptor rescue in our previous study. 2 The animals were sacrificed 3 days after injection to collect tissue for proteomics analysis. After enucleation, the lens was removed, and the retina was carefully separated from the eye cup. Investigators were masked to the identity of the injected compound for all analyses.

| Protein extraction and sample preparation for ITRAQ analysis
Isolated retinas were homogenized in 300 µl of T-PER buffer (Thermo Scientific # 78510) and then centrifuged at 1000 g for 10 min. The supernatant was collected in a separate tube, four volumes of cold acetone were added, and the samples were incubated overnight at −20°C. The samples were centrifuged at 10,000 g for 10 min and dried in a speed vacuum concentrator, then reconstituted in 0.5 M triethylammonium bicarbonate (TEAB). Protein concentrations were determined using a Bicinchoninic assay (BCA) (Thermo Fisher Scientific) according to the manufacturer's directions. The volume to obtain 40 µg of each sample was calculated, placed in a separate Eppendorf tube and dried in a speed vacuum concentrator. Each sample was reconstituted in 30 µl of 0.5 M TEAB. The proteins were denatured and reduced by adding 1 µl of 2% SDS and 1 µl of 110 mM tris-(2-carboxyethyl) phosphine (TCEP). The samples were vortexed, centrifuged at 12,000 x g for 5 min, incubated for 1 h at 60°C and then centrifuged again at 12,000 x g for 5 min. Proteins were alkylated by adding 1 µl of 84 mM iodoacetamide, vortexed and then centrifuged as in previous step. The samples were incubated at room temperature for 30 min in the dark then centrifuged at 12,000 x g for 5 min.
Proteins were digested using sequence-grade trypsin and incubated at 37°C overnight. ITRAQ Reagents (8-plex) (Sciex #4390733) were added to samples and incubated at room temperature for 2 h, following our previous study. 21 The reaction was quenched by adding 100 µl of LC-MS grade water and incubated for another 30 min.
All samples were combined into one tube, dried in a speed vacuum concentrator, washed three times with 100 µl of LC-MS grade water and then stored dry at 20°C until analysis by high-performance liquid chromatography tandem mass spectrometry (LC-MS/MS). The ITRAQ sample was reconstituted in 30 µl of 2% acetonitrile in water with 0.1% formic acid prior to LC-MS/MS analysis.

| High-performance liquid chromatographymass spectrometry
The trypsin-digested proteins were analysed using a Q Exactive mass spectrometer and an Easy-nLC 1000 equipped with a Nanospray The gradient ran from 2% solvent B to 30% solvent B over 57 min then to 80% solvent B over 6 min and was held at 80% solvent B for 12 min. The parameters of the ionization source were as follows: spray voltage was 1.8 kV, the capillary temperature was 250°C, the S-lens radio frequency (RF) level was 50, and all gases were set to 0. The full scan mass spectrometry settings included a resolution of

| Bioinformatics
The differentially expressed proteins were categorized with Gene ontology (GO) terms using http://geneo ntolo gy.org/ with PANTHER Version 15.0 (released 2020-02-14) 22 for enrichment analysis, using Fisher's exact test and Mus musculus as the reference list. Proteins with a p-value < 0.05 were considered significantly enriched.
Additionally, protein-protein interaction networks using physical and functional interactions were analysed using the STRING database (Search Tool for the Retrieval of Interacting Genes/Proteins) v.11 (http://strin g-db.org/) 23 with a minimum interaction score of 0.7 for high confidence.

| Statistical analysis
False discovery rate and enrichment analyses of the proteomics data and bioinformatics were performed by embedded software in the respective programmes.

| Data sharing
The proteomics raw data have been deposited in the PRIDE

| Identification of differentially expressed proteins
To identify proteins that may stimulate neuroprotective pathways in rd10 mice, and to capture initial changes that result from MyD88 inhibition, we used an early timepoint of 3 days after MyD88 inhibition. This timepoint is prior to significant photoreceptor death. 19,20,24 We also reasoned that later timepoints would mostly reveal altered photoreceptor proteins, overwhelming the samples with degeneration-associated proteins and would be less likely to reveal early mechanistic changes leading to increased photoreceptor survival.
A quantitative proteomics analysis was performed on retinas obtained from mice treated with MyD88 inhibitor peptide or control peptide. To identify differentially expressed proteins, highly sensitive combination iTRAQ LC-MS/MS analysis was performed on total cellular proteins extracted from neural retina tissue. This analysis detected 5586 proteins in the mouse retinas (n = 4 MyD88 inhibitor peptide, n = 4 control peptide), of which 394 proteins were identified with high confidence (FDR p < 0.01) and 54 were medium confidence (FDR p < 0.05), for a total of 448 combined high and medium confidence proteins. The remaining 5138 proteins were detected at low confidence and were not analysed further. Between 1 to 26 unique peptides mapped to each high and medium confidence protein, 301 (out of 394) high confidence proteins had ≥2 unique matching peptides and 31 (out of 54) medium confidence proteins had ≥2 unique matching peptides. Furthermore, there was an average of 14% protein coverage for the high confidence proteins and average of 7% for the medium confidence proteins (Table S1).
The differentially expressed proteins were identified from the 448 high and medium confidence total proteins by selecting iTRAQ ratios of MyD88 inhibitor injected to control peptide injected mice that had fold changes >1.2 or <0.83 and with p < 0.05, as used in previous studies. 15,25 In total, 42 retinal proteins were differentially expressed: 20 proteins were upregulated in the MyD88 inhibitor injected mice and 22 proteins were downregulated ( Table 1). The fold changes are relatively modest with a maximum change of 2.7-fold, which is expected at this early timepoint prior to significant photoreceptor death. 19,20 Fifty-seven of the high confidence proteins and 15 medium confidence proteins were not labelled with the isobaric tag and relative quantification and their relative expression could not be determined.

| Functional categorization of differentially expressed proteins
Gene ontology analysis using PANTHER 22 was used to categorize the 42 differentially expressed proteins, into "Biological process," "Molecular function" and "Cellular component" categories using Mus musculus as the reference organism. The significance of the different GO terms in our dataset was determined using an enrichment analysis. The top four terms for enriched biological processes with the number of genes in each category included multicellular organism development, eye development, peptide metabolic process and response to hypoxia ( Table 2). The top enriched molecular functions were eye lens structural constituents, unfolded protein binding, ligase activity and small molecule binding. There were no significantly enriched cellular components categories. Enrichment analysis was also performed separately for the increased and decreased proteins. The 20 increased genes were significantly enriched in three molecular functions (structural constituent of eye lens, unfolded protein binding and structural molecular activity) and 13 biological processes (five distinct processes when redundancies were removed) ( Figure 1). There were no significantly enriched cellular compartments for the increased proteins.
In contrast, the 22 decreased proteins were categorized into two significantly enriched molecular functions (pyrophosphatase activity, small molecule binding), three distinct significantly enriched biological processes (peptide biosynthesis, organonitrogen compound biosynthesis and amide metabolic process) and three distinct significantly enriched cellular compartments (photoreceptor inner segment, outer segment cytoplasm) ( Figure 1).
Interestingly, the upregulated proteins were enriched for crystallin genes. Out of 20 upregulated proteins, seven of them belong to the crystallin family (Table 1). Numerous studies have reported expression of alpha, beta and gamma crystallin proteins outside the lens within the photoreceptor, inner nuclear and ganglion cell layers of the retina. [26][27][28] It is now understood that α-crystallins belong to the small heat shock protein superfamily and play neuroprotective and stress response roles, whereas β-and γ-crystallins are stress response and structural proteins. 27 However, to exclude the possibility that the retina samples were contaminated with lens tissue, we searched the total 5586 identified proteins for lens-specific and lens-enriched proteins. 26 The lens proteins GJA3, GJA8, GJA1, aquaporin 0, radixin and BFSP1 were not detected or were detected with only low confidence, and 10 crystallins that are the major structural components of lens were also not detected in the retina samples: CryβA2, CryβA4, CryβB2, CryγA, CryγB, CryγD, CryγF and CryγN.
Therefore, the absence of these lens proteins indicates that the detection of known extra-lenticular crystallins was not likely due to contamination of the retina tissue with lens during the tissue dissociation steps.

| Protein-protein interaction network analysis
To explore potential signalling pathways among the differentially expressed proteins, protein-protein interaction networks were investigated in silico using STRING v11 analysis. 23

| DISCUSS ION
The purpose of this study was to characterize and quantify early protein changes after MyD88 inhibition in order to gain insight into how blocking MyD88 signalling may lead to photoreceptor protection in the rd10 mouse model of retinal degeneration. Using iTRAQ quantitative proteomic analysis, we demonstrated differential expression of 42 proteins in the treated retinas. A major class of proteins that was changed was crystallins, a family of proteins that function as TA B L E 1 Differentially expressed genes (20 upregulated, 22 downregulated) were identified in the 448 high and medium confidence total proteins using iTRAQ ratios of MyD88 inhibitor injected to control peptide injected that were >1.2 or <0.83 stress response and anti-apoptotic proteins. Additional chaperones and cell stress response proteins were also upregulated. This study is the first quantitative proteomics analysis of rd10 mice treated with an experimental therapy and identifies a link between modulating inflammatory pathways and crystallin expression. These data suggest that inhibiting MyD88-mediated signalling may enhance intrinsic tissue-protective pathways.   Overexpression of βB2-crystallin also increased RGC survival after ON injury. 43 βA3/A1 crystallin was highly upregulated in our study, and this protein is known to mediate lysosome function and autophagy in RPE, 46  Out of the 20 upregulated proteins, many are known to act as chaperones or to reduce cell stress. In addition to the seven crystallins already mentioned, we observed upregulation of chaperonin 10 and serine/threonine protein phosphatase PP1-beta (Ppp1cbP), a nucleophosmin phosphatase that regulates cellular responses to genotoxic stress. 47 Adenomatous polyposis coli protein 2 (APC2), which regulates the response to ER stress, 48  Also, cluster analysis indicated enrichment of proteins involved in peptide metabolism (Rps9, Eif4a2, Xpnpep1, Npepps) and pyrophosphatase activity (for example, Gnb1, Atp2b1, mt-Atp6), suggesting changes in protein synthesis. It is currently unclear how these groups of proteins are related to MyD88 inhibition. Two photoreceptor proteins, rhodopsin and Gnb1, were also reduced which was unexpected due to the early timepoint prior to degeneration, but their decrease may be related to reduced protein synthesis proteins.

| Crystallins and other chaperone proteins
We did not detect changes in inflammatory proteins in the retinal samples at the timepoint tested despite previous studies from our group and others that showed cytokines and other in- Future studies using direct experimental validation will determine whether these proteins contribute to protection from MyD88 inhibition.

| Limitations and future directions
Several limitations are noted. First, as mentioned above, 72 proteins that were identified by MS were not labelled with the isobaric tags and differential expression could not be assessed. However, this number represents only 1.2% of the total medium and high confidence proteins identified, which is unlikely to affect the overall conclusions of the study. Second, the study is focussed only on protein expression because post-translational modifications in response to MyD88 signalling inhibition, such as phosphorylation or acetylation, could not be detected with our methods. Third, although iTRAQ is highly efficient and sensitive, standard confirmation analyses, such as Western blotting or ELISAs, would provide further validation. However, mass spectrometric sequencing of proteins provides high confidence identification compared to other methods that rely on antibody specificity. Finally, we deliberately chose an earlier timepoint to detect changes in the retina prior to extensive photoreceptor death, and future experiments would be performed to characterize expression of the differentially expressed proteins at additional timepoints to determine whether they show sustained expression changes.

| CON CLUS IONS
This study presents the first iTRAQ analysis to quantify protein differences in degenerating retinas from mice treated with an experimental neuroprotectant molecule. In summary, rd10 mice treated with the MyD88 inhibitor peptide showed a significant increase in chaperone and crystallins proteins and a decrease in peptide metabolism proteins. These findings indicate that neuroprotection by MyD88 inhibitor treatment is associated with induction of signalling pathways that reduce cellular stress and protein misfolding, suggesting enhancement of an intrinsic tissue-protective response.
Therefore, these findings identify candidate pathways that may contribute to neuroprotection in MyD88 inhibitor treated rd10 mice.

CO N FLI C T O F I NTE R E S T
The authors declare that they have no competing interests. Project administration (lead); Supervision (lead); Visualization (lead);

AUTH O R CO NTR I B UTI
Writing-original draft (lead); Writing-review & editing (lead).

E TH I C A L A PPROVA L
Approval for the use of the rd10 mouse model in these experiments was obtained by the University of Miami IACUC committee.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that supports the findings of this study are available in the manuscript and its supplementary material. Raw